**Intra-Writer Variation in Historical Sociolinguistics**

Markus Schiegg and Judith Huber (eds)

9781800797031\_cvr\_eu.indd All Pages 17-Aug-23 13:45:56

www.peterlang.com

and language contact.

Intra-individual variation is an emerging research field in linguistics with a rapidly growing number of studies. In historical sociolinguistics, this trend has been slow, as it is still largely dominated by the macroscopic approaches of earlier sociolinguistics. Microscopic studies focusing on intra-individual variation in writing, i.e. *intra-writer variation*, however, are able to reveal how writers functionalize social or text-type variation for reasons such as audience design or persona creation. They may also provide insights into how ongoing changes were perceived by speakers and writers. In general, micro-approaches are able to uncover a wide array of possible factors influencing variation, which may not

Historical Sociolinguistics

This volume comprises twenty-two research articles on a wide range of languages and periods, all closely connected by their focus on intrawriter variation in historical texts and by their use of empirical and corpus-based approaches. The studies demonstrate that the challenges that historical material have for research on intra-individual variation can certainly be met and that the insights gleaned from analysing

Markus Schiegg works in German Linguistics at Friedrich-Alexander-Universität Erlangen-Nürnberg. He currently leads the junior research group 'Flexible Writers in Language History' that is compiling a corpus of historical patient texts from the nineteenth and early twentieth century. His research focuses on historical sociolinguistics, in particular

Judith Huber works in English Historical Linguistics at Ludwig-Maximilians-Universität München (LMU). She holds a PhD from LMU and worked in English Historical Linguistics at FAU Erlangen-Nürnberg and KU Eichstätt-Ingolstadt before returning to Munich. Her research focuses on variation and change in the history of English from a usage-based perspective, including syntax, lexicology, pragmatics

on language variation and change in the history of German.

always carry sociolinguistic functions.

variation in individual writers are considerable.

# Intra-Writer Variation in Historical Sociolinguistics

Markus Schiegg and Judith Huber (eds)

# Historical Sociolinguistics

Intra-individual variation is an emerging research field in linguistics with a rapidly growing number of studies. In historical sociolinguistics, this trend has been slow, as it is still largely dominated by the macroscopic approaches of earlier sociolinguistics. Microscopic studies focusing on intra-individual variation in writing, i.e. *intra-writer variation*, however, are able to reveal how writers functionalize social or text-type variation for reasons such as audience design or persona creation. They may also provide insights into how ongoing changes were perceived by speakers and writers. In general, micro-approaches are able to uncover a wide array of possible factors influencing variation, which may not always carry sociolinguistic functions.

**Intra-Writer Variation in Historical Sociolinguistics**

Markus Schiegg and Judith Huber (eds)

9781800797031\_cvr\_eu.indd All Pages 17-Aug-23 13:45:56

PETER LANG

Historical Sociolinguistics

Intra-Writer Variation

Markus Schiegg and Judith Huber (eds)

in Historical

Sociolinguistics

This volume comprises twenty-two research articles on a wide range of languages and periods, all closely connected by their focus on intrawriter variation in historical texts and by their use of empirical and corpus-based approaches. The studies demonstrate that the challenges that historical material have for research on intra-individual variation can certainly be met and that the insights gleaned from analysing variation in individual writers are considerable.

Markus Schiegg works in German Linguistics at Friedrich-Alexander-Universität Erlangen-Nürnberg. He currently leads the junior research group 'Flexible Writers in Language History' that is compiling a corpus of historical patient texts from the nineteenth and early twentieth century. His research focuses on historical sociolinguistics, in particular on language variation and change in the history of German.

Judith Huber works in English Historical Linguistics at Ludwig-Maximilians-Universität München (LMU). She holds a PhD from LMU and worked in English Historical Linguistics at FAU Erlangen-Nürnberg and KU Eichstätt-Ingolstadt before returning to Munich. Her research focuses on variation and change in the history of English from a usage-based perspective, including syntax, lexicology, pragmatics and language contact.

Intra-Writer Variation in Historical Sociolinguistics

# Historical Sociolinguistics

# Series editors

*Nils Langer Stephan Elspaß Joseph Salmons Wim Vandenbussche*

Vol. 5

# Intra-Writer Variation in Historical Sociolinguistics

Markus Schiegg and Judith Huber (eds)

PETER LANG Lausanne ∙ Berlin ∙ Bruxelles ∙ Chennai ∙ New York ∙ Oxford Bibliographic information published by the Deutsche Nationalbibliothek. The German National Library lists this publication in the German National Bibliography; detailed bibliographic data is available on the Internet at http://dnb.d-nb.de.

A catalogue record for this book is available from the British Library.

Library of Congress Cataloging-in-Publication Data


The publication received financial support from the Bavarian State Ministry of Science and the Arts (Elite Network of Bavaria) and the Friedrich-Alexander-Universität Erlangen-Nürnberg.

Cover image: *Flexible Schreiberin* by Vera Schiller. Reproduced with permission. Cover design by Peter Lang Group AG ISSN 2296-1909 ISBN 978-1-80079-703-1 (print) ISBN 978-1-80079-704-8 (ePDF) ISBN 978-1-80079-705-5 (ePub) DOI 10.3726/b19157

Open Access: This work is licensed under a Creative Commons Attribution CC-BY 4.0 license. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/

© 2023 Markus Schiegg and Judith Huber Published by Peter Lang Ltd, Oxford, United Kingdom info@peterlang.com – www.peterlang.com

Markus Schiegg and Judith Huber have asserted their right under the Copyright, Designs and Patents Act, 1988, to be identified as Editors of this Work.

This publication has been peer reviewed.

# Contents






#### *Contents* ix

#### Oliver Currie


## Figures


## i Intra-writer variation in letter writing



#### *Figures* xiii


## ii Intra-writer variation in contact and migration settings


#### *Figures* xv


## iii From intra-writer variation to variation beyond the individual



# Tables

## i Intra-writer variation in letter writing



## ii Intra-writer variation in contact and migration settings


## iii From intra-writer variation to variation beyond the individual



## Acknowledgements

newgenprepdf

The chapters of this thematic volume are based on a selection of papers presented at the tenth conference of the Historical Sociolinguistics Network, *Intra-Writer-Variation in Historical Sociolinguistics*, at the Friedrich-Alexander-Universität Erlangen-Nürnberg in 2021. We thank the series editors, particularly Nils Langer, for their encouragement and intellectual stimuli during the creation of this book. The two anonymous referees provided helpful advice on the draft volume. Kerstin Kazzazi's and Samantha Litty's linguistic expertise facilitated significant improvements. Carima Jekel, Julian Mader and Regina Meyer provided further important corrections and assisted the editing process of this book. We also thank Laurel Plapp and her colleagues at Peter Lang for their editorial advice. Finally, we are grateful for the generous financial support from the Bavarian State Ministry for Education and Culture (Elite Network of Bavaria) and the Friedrich-Alexander-Universität Erlangen-Nürnberg for the publication of this volume.

> Markus Schiegg and Judith Huber Erlangen and Munich, February 2023

#### Markus Schiegg & Judith Huber

# 1 Intra-writer variation in historical sociolinguistics: The emergence of a new research field

#### Abstract

Intra-individual variation is an emerging research field in linguistics with a rapidly growing number of studies. This chapter provides an overview of different linguistic approaches to intra-individual variation and links them to the notion of stylistic variation in sociolinguistics. It discusses the challenges and opportunities of applying these concepts to historical sociolinguistics, where intra-individual variation in writing, that is, intra-writer variation, has long been neglected. Presenting twenty-two empirical, corpus-based research articles, all focusing on intra-writer variation, but in a wide range of languages and periods, the present volume contributes to the extension of this growing research area in historical sociolinguistics.

## 1 Introduction

Intra-individual variation, that is, observable variation within individual behaviour, has long been an under-researched phenomenon in linguistics. In traditional dialectology, for example, the individual was considered a factor that obstructs the researcher's access to the intended object of study, the oldest layers of a regional dialect. This focus on regional variation led to traditional dialectology's exclusion of both social and individual dimensions of variation (Chambers & Trudgill 1998: 45). Similarly, macroscopic approaches in sociolinguistics have focused on linguistic patterns in speech communities and factored out the individual, as they considered 'intraspeaker variation to be evidence of inherent variability in a communal grammar' (Meyerhoff 2006: 37).

In recent years, this view has changed and intra-individual variation has attracted increased attention and interest in different areas of linguistics. Therefore, intra-individual variation has reached 'the centre of sociolinguistic theorization and method' (Bell 2014: 297) and is 'a key ingredient of variationist sociolinguistic research' (Hernández-Campoy 2016: 30). In historical sociolinguistics, however, this trend has been delayed. Historical sociolinguistics has predominantly followed the macroscopic approaches of present-day sociolinguistics, while intra-individual variation in writing, which we label *intra-writer variation*, has often been neglected.

This volume contributes to the emergence of this new and promising research field in historical sociolinguistics. Section 2 of this introductory chapter presents the concept of intra-individual variation in linguistics in general. Section 3 explores the relation of intra-individual variation and stylistic variation in sociolinguistics. Section 4 gives an overview of existing work on intra-writer variation in historical sociolinguistics and discusses the challenges and opportunities of applying this concept to historical data. Finally, Section 5 presents the structure and contents of this volume, which consists of three thematically arranged parts with twentytwo research articles on a wide range of languages and periods. They are all closely connected by their focus on intra-writer variation (and sometimes also variation beyond the individual) in historical texts and by their use of empirical and corpus-based approaches.

## 2 What is intra-individual variation?

Intra-individual variation is a universal phenomenon in language, as two utterances are never the same, both with regard to their production and perception (Ulbrich & Werth 2021: 10). This was already described in Hermann Paul's (1898: 51) metaphor of a shooter who never hits exactly the same point of the bull's eye, just as the configuration of the articulatory organs will typically not be exactly the same in two utterances of the same sequence. Paul related this observation with his theory of sound change that often results from the accumulation of such unnoticed variation by individual speakers.1

As individuals are always part of social groups, there is a connection between intra- and inter-individual variation (Labov 2001a: 33). Sociolinguistic studies following Labov's pioneering work have observed that 'quantitatively, the degree of style variation never exceeds the degree of social variation' and that 'the range of style shift is less than the range of social differentiation available' (Bell 1984: 152). Individual linguistic repertoires have shown to be smaller than those of their respective speech community as a whole (Biber & Conrad 2009: 24). At the same time, there is a symmetry between intra- and inter-speaker variation, as the same variables are used for their differentiation (Labov 2001b: 86): Sociolectally marked variants may function as diaphasically marked variants.2 Therefore, the more formal styles of lower social classes have shown to be similar to the more informal styles of the higher social classes (see e.g. Trudgill 1974: 92). Consequently, it is, according to Labov (1972: 240), difficult 'to distinguish, for example, a casual salesman from a careful pipefitter'.

Nevertheless, there are also fundamental differences between intraand inter-individual variation, as the former can only be observed over time. Two utterances cannot be produced or perceived simultaneously, which makes intra-individual variation a different, or even independent, research topic (Ulbrich & Werth 2021: 11).3 From this it follows that a diachronic perspective, albeit a short-term one, is an essential part of and needs to be integrated into the concept of intra-individual variation. This also applies to writing: Two texts of a writer are never produced under identical circumstances, and even within one and the same text, changes in the external context or individual conditions during the writing process,


for example, decreasing concentration (Schiegg & Gunkler-Frank 2021) and increasing tiredness (Havinga 2021: 335), can trigger intra-writer variation (see Section 4.1).

The complexity of the phenomenon and different research foci have led to inaccuracies and overlaps in the terminology used to describe intraindividual variation. As the quotations from Bell (1984) given above have illustrated, the term *style variation* is often used in sociolinguistics to describe 'the range of variation produced by individual speakers within their own speech' (Bell 2007: 90). Another term commonly found in this context is *register*, which Hallidayan linguists (cf. Halliday 1978) have applied to refer to linguistic variation in different social situations, while Labovian sociolinguists have restricted registers to occupational varieties (Hernández-Campoy 2016: 33). Particularly in Romance linguistics, we encounter the term *diaphasic variation*, originally a rather broad and heterogeneous dimension of Coseriu's (1980: 111) architecture of language, but adapted by Koch and Oesterreicher (2011: 15) to denote stylistic variation (cf. also Grübl et al. 2021: 16).

The status assigned to intra-individual variation in modern linguistics crucially depends on the role which variation plays in the respective frameworks. Rule-based accounts, on the one hand, neglect variation in general and intend to formulate universal grammatical rules of an ideal speaker-hearer competence, unaffected by 'memory limitations, distractions, shifts of attention and interest, and errors (random or characteristic) in applying his knowledge of the language in actual performance' (Chomsky 1965: 3). Therefore, approaches such as Optimality Theory (Prince & Smolensky 2004) 'only include linguistically-conditioned variation and propose redundancy-free representation' (Ulbrich & Werth 2021: 10) in their frameworks. Exemplar- or usage-based accounts, on the other hand, treat language as a dynamic and emergent system and consequently, variation is part and parcel of their theoretical and methodological considerations (cf. Bybee & Beckner 2015).

A multilayered and dynamic model of intra-individual variation has recently been proposed by Ulbrich and Werth (2021: 18). As we will use it as a reference and apply it to historical intra-writer variation in the following sections, we will explain it in more detail. Ulbrich & Werth's model encompasses three types of intra-individual variation (IAV):4


The first type, *non-conditioned IAV* or *free variation*, concerns variation that 'occurs under maximally similar conditions' (Ulbrich & Werth 2021: 18) and cannot be predicted as it does not depend on any extraor intra-linguistic factors. This notion of intra-individual variation has sometimes been criticized: 'free variation does not exist because linguistic variation is not free at all, but rather constrained by social and/or situational factors' (Hernández-Campoy 2016: 69 with reference to Labovian studies). Empirical studies, for example, in the area of second language acquisition, however, have proved the concept of free variation to be valid (for a short research overview see Bülow & Pfenninger 2021: 2). In ad dition, a neglect of free variation quickly leads to the assumption of homogeneous and coherent varieties and thus deterministic speakers (Milroy 1987: 131; Macha 1991), which is invalid both from theoretical and empirical accounts (Boyd & Fraurud 2010; Bülow et al. 2021: 158).

*Conditioned IAV* or *formal variation* assesses the influences of the linguistic environment of the variable under study, that is, 'phonological, morphological, and syntactic constraints which promote or inhibit the application of a variable rule' (Bell 1984: 145). Relevant examples are combinatory variants and phonotactic constraints in phonology, and morphologically conditioned allomorphy (Ulbrich & Werth 2021: 19). For example, in modern Standard German the palatal fricative [ç] follows high vowels

4 A fourth category are the so-called *mandatory forms* that do not allow for any kind of IAV and comprise linguistic units and structures that have a 'correspondence between mental representation and output, and the use of another form more or less violates predictions or expectations' (Ulbrich & Werth 2021: 20). This lack of IAV can be observed, for instance, in word stress or derivational affixes (Ulbrich & Werth 2021: 20).

such as [ɪ] (e.g. *ich* [I]), while the velar fricative [x] follows low vowels such as [a] (e.g. *Dach* [roof ]).

*Functionalized IAV* or *functional variation* finally encompasses both 'linguistically and extralinguistically meaningful variants' (Ulbrich & Werth 2021: 20). This third type of IAV is therefore broader than in variationist accounts of intra-individual variation: In addition to extralinguistic factors such as age, gender, region, etc., it also subsumes intralinguistic ones, particularly in the areas of semantics and pragmatics (Ulbrich & Werth 2021: 20), such as the role of Aktionsart in the choice of *be* vs *have* as perfect auxiliaries in Middle English (Huber 2019).

An advantage of this model is both its clear differentiation of several factors for variation and its consideration of non-predictable, nonconditioned intra-individual variation that has often been overlooked or interpreted as functionalized variation in linguistic research (Häcki Buhofer 1998: 66). This particularly applies to sociolinguistics, which usually follows more narrow concepts of intra-individual variation with a 'rather field-specific terminology and methodology' (Ulbrich & Werth 2021: 9), as will be assessed in the following section.

## 3 Approaches to stylistic variation in sociolinguistics

Sociolinguistics often equates intra-speaker variation with stylistic variation5 and assumes it to be functionalized (Ulbrich & Werth 2021: 9). Following Halliday (1978), Bell (1984: 145) differentiates between *social variation* that 'denotes differences between the speech of different speakers' (i.e. *inter*-speaker variation) and *stylistic variation* that 'denotes differences within the speech of a single speaker' (i.e. *intra*-speaker variation). While these two types of variation have been subsumed under *extra-linguistic* variation, that is, influenced by extra-linguistic factors,

<sup>5</sup> See, for example, Schilling (2013: 327): 'variation analysis encompasses the study of variation in the speech of individual speakers, or stylistic variation'.

Figure 1.1. Sociolinguistic variation (Hernández-Campoy 2016: 30).

*(intra-)linguistic* variation, that is, variation depending on the linguistic environment, is considered the third ingredient in this concept of sociolinguistic variation (see Figure 1.1; Bell 1984: 146; Hernández-Campoy 2016: 30).

Linguistic and social variation has been the core of variationist research since the 1960s. Empirical variationist studies combined work on intra-linguistic variation, that is, 'constraints on variable speech output, sound change and syntactic change, the mechanisms of vowel shift, and structural relations among regional dialects' (Rickford & Eckert 2001: 2) with social variation and its interest in the 'relation between variation and social parameters, including class, gender, ethnicity, social networks, identity, local categories, and ideology' (Rickford & Eckert 2001: 2).

In these early correlational studies, style was treated as another independent parameter: 'style is one dimension that has often been measured but seldom explained' (Bell 1984: 145). Style was 'understood in a narrow sense, focusing on context and topic and only cursorily on speaker and listener' (Hernández-Campoy 2016: 30). The prominent sociolinguistic studies by Labov (1966) in New York City and Trudgill (1974) in Norwich have de scribed not only a social stratification of forms but at the same time also a correlation between situational parameters and degrees of formality in the speech communities they analysed. Style shifting was thus explained as the result of the degree of *attention paid to speech*, also called *audio-monitoring*.

In the 1980s, sociolinguistic studies began to challenge this audiomonitoring model of stylistic variation with regard to both theoretical and methodological issues.6 Bell (1984: 150) objected to both the mechanistic approach of the model and its lack of explanatory force and pleaded 'to go behind the mechanistic attention variable to see what factors in the live situation are actually causing these differing amounts of attention'. Building on Giles' (1973)*communication accommodation theory*, Bell (1984) conceived style choices primarily as responses to the audience. Similar to Biber and Finegan's (1994) approach to register variation, Bell's (2001: 146) *audience design model* connects '[s]tyle-shifting according to topic or setting' with 'the underlying association of topics or settings with typical audience members'. However, he puts the dialogic character of language in the centre of his model when he emphasizes 'the active role of speakers' (Bell 2001: 143) in their responsiveness to the audience. Bell's broad approach to real-world interactions led to a large number of fruitful applications in different research contexts. Nevertheless, his model is still rather static and is not able to explain all cases of stylistic variation, as it does not account for the 'speakers' creative freedom' (Hernández-Campoy 2016: 128).7

In the 2000s, the focus of sociolinguistic research on stylistic variation 'has turned from the reactive to the creative and from aggregate patterns to local usages' (Schilling 2013: 328). This so-called 'third wave of variation studies' (Eckert 2012: 93) is based on a social constructionist approach and sees 'speakers as actively taking part in shaping and re-shaping interactional norms and social structures, rather than simply accommodating to them' (Hernández-Campoy 2016: 128f.). People's strategic use of language styles shapes the situative contexts and supports the performance of multiple social identities (Coupland 2007). From the perspective


of this *speaker design model*, variables do not have fixed meaning but, in Eckert's (2008: 453) view, rather 'constitute a field of potential meanings', a so-called *indexical field*, which is 'a constellation of ideologically related meanings, any of which can be activated in the situated use of the variable' (cf. Silverstein 2003). In this context, influential concepts such as *crossing* (Rampton 1995), *enregisterment* (Agha 2007) and *hypervernacularization* (Cutillas-Espinosa et al. 2010) have been developed to assess the performative character of linguistic stylizations.

The focus on the agency of the individual speakers reversed the traditional view on the relation between language and society (Eckert 2012: 97) by 'questioning the primacy of social structure over individual capacity' (Hernández-Campoy 2016: 157). The shift in perspective from macrosociological categories and speech communities to local *communities of practice* (Wenger 1998) and individual language use allows for a much more detailed picture of stylistic variation. This shift of focus to individual linguistic practices in local settings, however, has raised questions about the extent to which the results can be generalized to other speakers, larger groups and the speech community (Bell 2014: 305).8 In addition, we cannot deny the existence of linguistic conventions with 'pre-existing associations between linguistic usages and social meanings' (Schilling 2013: 342). There are limitations on creative stylistic usages, as individual linguistic repertoires are closely connected to and part of the repertoires of the speech community (Schilling 2013: 342).

As a consequence, research needs to find a balance between responsive and initiative dimensions of style, that is, between the role of structure and agency in sociolinguistics (Bell 2014: 326–28).9 Style can be considered a 'multidimensional phenomenon that cannot be modeled in a single unidimensional theory' (Hernández-Campoy 2016: 189). Even proponents of the 'third wave' of stylistic variation now agree that '[t]he serious study


of variation calls for the unification of the macro- and micro-perspectives' (Eckert 2016: 82), as 'the three waves are part of the same ocean' (Schilling 2013: 343).

A study of linguistic variation thus needs to consider diverse dimensions of variation. Apart from stylistic variation, observed in the individual speaker, variation can also be conditioned by social and linguistic factors. Particularly the speaker design approaches made us aware of the fact that the 'division between social and stylistic constraints is a fine and highly permeable one' (Rickford & Eckert 2001: 6).10 Stylistic variation seen as an agentive phenomenon has the potential of being functionalized by the speakers to achieve their communicative goals, manage their persona and create an identity by the choice of indexicalized variants.

To include free variation, which has often been disregarded in sociolinguistic research but has empirically been proved valid in other areas of linguistics (see Section 2), we return to the three types of intra-individual variation proposed by Ulbrich and Werth (2021: 18): non-conditioned variation, conditioned variation, and functionalized variation (see Section 2), illustrated by three circles in Figure 1.2.

The figure also emphasizes that these three dimensions apply not only to individuals (*intra*-individual variation, IAV, indicated by the three circles in the minds of the individuals), but also between them (*inter*-individual variation, IEV). Social groups share patterns of variation that are conventionalized in their historical languages. Individuals rely on these patterns, while their linguistic repertoires are smaller than those of their respective speech community (cf. Section 2). At the same time, they can creatively re-shape interactional norms, which may lead to language change.

The research overview in the following section will show that all these three dimensions of variation can in principle be applied to historical, written data. There, however, we may observe further relevant factors for intra-writer variation that require some additions to the model. In particular, the possible factors for conditioned variation go beyond influences

<sup>10</sup> See also Ulbrich and Werth (2021: 33) for possible switches between the different categories of intra-individual variation and the non-teleological nature of these processes.

Figure 1.2. Dimensions of variation.

from the direct linguistic environment of a variable, but may on the one hand result from the writing process itself, which is susceptible to a writer's varying degrees of concentration or external events such as interruptions. On the other hand, we need to take into account contexts where the responsibility for a text goes beyond the individual, which is, for example, the case for delegate writing, copied texts or also for texts from highly formalized text types such as notary acts. They pose particular challenges for applying the concept of intra-writer variation to historical sociolinguistics (see Section 4.2).

## 4 Intra-writer variation in historical sociolinguistics

#### *4.1 Previous research*

Historical sociolinguistics evolved in the 1980s and draws on insights from present-day sociolinguistics to study language variation and change in the past (Auer et al. 2015: 4).11 Previous work applied, for example, Labov's methods such as the variable rule analysis to historical data (Romaine 1982). It showed a particular interest in the language use of lower social classes (see, for example, Mattheier 1989; Vandenbussche 1999), which gave rise to alternative language histories (Watts & Trudgill 2002) and a new perspective on histories 'from below' (Elspaß 2005).

In recent years, micro-approaches and the focus on intra-individual variation have also gained ground in historical sociolinguistics. Hernández-Campoy and García-Vidal (2018b: 412) observe that 'the analysis of linguistic patterns across styles is fortunately coming to be acknowledged as crucial for both the linguistic description of languages and for the development of crosslinguistic theories of use and change'. Applying both the audience design and the speaker design models (Hernández-Campoy & García-Vidal 2018a, 2018b) to historical data, namely the English Paston Letters (see also their two chapters (4 and 5) in this volume), the authors were able to obtain a 'wider and more accurate picture of speakers' sociolinguistic behaviour in earlier periods' (Hernández-Campoy & García-Vidal 2018b: 386).

Another historical text type that has been shown to be fruitful for the analysis of intra-writer variation are letters written by patients in psychiatric hospitals from the nineteenth and early twentieth centuries (see Gunkler-Frank's chapter (10) in this volume). Schiegg's (2022) analysis of the distribution of features of conceptual orality and distance in 191 letters by 28 different writers from Southern Germany, both patients and their acquaintances, has demonstrated that all of them vary their language depending on the different addressees. This quantitative audience design approach was complemented with qualitative analyses from the perspective of speaker design, which provided evidence for high degrees of linguistic creativity and flexibility, strategic uses of different varieties, and thus the performance of diverse identities, even among writers with a low degree of

11 For the development, goals and methods of historical sociolinguistics see Auer et al. (2015), Joseph (2012), Nevalainen and Raumolin-Brunberg (2017), Pickl and Elspaß (2019), Russi (2016), Säily et al. (2017) and the *Handbook of Historical Sociolinguistics* (Hernández-Campoy & Conde-Silvestre 2012).

schooling. Auer's (2015) analysis of six nineteenth-century letters by three female English writers from different social backgrounds came to a similar conclusion, as all three writers, despite having different linguistic repertoires, were able to shift between different social identities in their letters.

A diachronic perspective is found in Havinga's (2021) comparative ana lysis of three letters written by an Austrian maid to her sister in 1845. She noticed a connection between the writer's social advancement to a lady's maid and the language use in her last letter, in which she uses a more formal style to project her higher social status. This approach of integrating individual developments over time into intra-writer analyses has particularly been utilized in the work on the *Corpus of Early English Correspondence*. Nevalainen and Raumolin-Brunberg's (2017: 191) focus on individual *lifespan changes* allows for the differentiation of linguistically progressive and conservative writers and thus a localization of language change in individuals and their position in social networks (see also Nevalainen et al. 2011, 2018). When studying language use of individuals in different stages of life, we need to consider changes in their physical and mental condition resulting from old age or illness (Gerstenberg & Voeste 2015), which can also be observed in historical patient letters (Schiegg & Thorpe 2017; Schiegg 2022: 495–508).

Connections between intra-writer variation and diachrony have also been made in sociolinguistic analyses of historical migration contexts, where language accommodation and shift happen not only between the different generations but also in real-time changes of individuals (see, e.g. van Hattum 2018; Litty 2019; Stolberg 2019; and the chapters by Dörnbrack (13) and Stolberg (16) in this volume).

Detailed linguistic analyses of historical texts allow for an assessment of a variety of different motivations for intra-individual variation. In this context, historical *code-switching*, both between varieties of a language and between different languages, has shown to be a highly salient practice, where intra-writer variation appears on the micro-level and reflects the writer's conscious linguistic choices (see, e.g. Langer & Havinga 2015; Schendl & Wright 2011; Schiegg 2016; Schiegg & Foldenauer 2021; and the chapters by Brown (11), Girininkaitė (15), Krogull et al. (14) and van Eyndhoven (12) in this volume).

Varying degrees of concentration and attention that writers devote to their texts can lead to shifts in language use and may be connected to the attention to speech approach. The rushed and rather messy handwriting in one of the three letters analysed by Havinga (2021: 337) 'could indicate that less attention was paid to the writing process', which, in this case, resulted from the tiredness of the writer, as indicated by a metalinguistic remark. Attention to writing may also decrease during the writing process of single letters, as observed in a quantitative, orthographic analysis of fifteen patient letters by one writer (Schiegg & Gunkler-Frank 2021). Particularly at the beginning of letters, writers often follow the norms of the written standard, while highly marked variants are found more frequently later in the text when concentration is decreasing (Schiegg 2019: 178f.) and when less-experienced writers are no longer able to rely on opening formulae that they remember as fixed elements (Elspaß 2005: 170). Interruptions during the writing process could also lead to shifts in the language, as has been shown in an analysis of the handwriting of a medieval writer (the so-called *Tremulous Hand of Worcester*) suffering from different degrees of fatigue (Thorpe & Alty 2015: 3124). Also, merely turning the page or dipping the quill into the ink pot could condition a short pause, used for reflecting about one's language use. This can lead to, for instance, a request being repeated in a more formal phrasing at the top of the new page (Schiegg 2018: 107). Similarly, Schiegg and Freund (2019: 64) describe a case of self-correction where the first letters of a derogatory word are written with the ink supply in the quill fading, and are crossed out and replaced by a neutral word with the refilled ink (Schiegg & Freund 2019: 64).

These examples show that micro-analyses are essential to explain the large variety of possible factors influencing intra-writer variation. Nevertheless, not all instances of intra-writer variation can be explained as conditioned or functionalized variation. For example, Havinga's (2021: 336) analysis of three letters by an Austrian maid, as already referred to above, shows that 'no consistent conditioning factors can be determined for her use of punctuation marks or lack thereof '. Similarly, variation between standard German <g> spelling and <ch> as in *sagt* and *sacht* [he says], reflecting northern German spirantized pronunciation, in early twentieth-century letters by a locksmith cannot be predicted (Schiegg & Eichhorn-Hartmeyer 2020: 48). It appears under maximally similar conditions, for example, in two identical address formulae to the same person, and cannot be attributed to any conditioning factors. Consequently, such instances of intra-writer variation may be classified as non-conditioned or free variation.

Although free variation cannot be explained locally, it may occasionally be connected to Auer's (1986: 119) concept of *code fluctuations* that he uses to explain variation reflecting the writer's origin and writing experience. The lack of punctuation marks, for example, indicates less experience with orthographic norms (see also Auer, Gardner & Iten (6) in this volume). Some phenomena of free variation, however, simply 'reflect […] the status quo of language at the time' (Knüsli 2019: 15), as illustrated, for instance, by Knüsli's findings on variation between <th> and <t> spellings (as in *Rath* vs *Rat* [council]) in a diary from 1845 by a Swiss migrant to America, at a time when German orthography had not yet been codified and showed a considerable degree of variation (see also Uribe-Etxebarria (19) in this volume). Changes in frequency over time, however, may be connected to the linguistic progressiveness of individuals and their interest in following the changing linguistic norms. This was examined in a diachronic intra-writer analysis for the <th>/<t> variable in Schiegg (2022: 508–18), where a mill labourer and keen reader of newspapers gradually implemented the standardized distribution of these variants, which were codified around 1900, into his texts (see also Dörnbrack (13) in this volume). These examples illustrate that the boundaries between the different areas of intra-writer variation are fluid, and that the reconstruction of individual social, situational and historical contexts is essential for approaching possible factors of variation.

#### *4.2 Challenges and potentials*

Existing studies on historical intra-writer variation have shown that the concepts of stylistic variation developed in modern sociolinguistics can generally be applied to the analysis of historical data. This conforms to the *principle of uniformitarianism*: 'The linguistic forces which operate today and are observable around us are not unlike those which have operated in the past' (Romaine 1988: 1454).12 At the same time, however, historical sociolinguistics has to tackle the so-called *historical paradox*, that is, to find out and explain the differences between the past and present (Labov 1994: 21). As such, this requires careful reconstructions of the historical and social contexts of the language periods and linguistic communities under study (Auer et al. 2015: 5). The challenges here increase with the temporal distance of the material under study, for example, when working with ancient or medieval sources.

A crucial difference to modern sociolinguistics lies in the data, which cannot be elicited by the linguist in experiments but often survive 'by chance, not by design, and the selection that is available is the product of an unpredictable series of historical accidents' (Labov 1994: 11). In addition, the data transmitted to us usually have a bias favouring texts written by privileged and educated writers, which makes them hardly representative of the language of large parts of societies from the past and of orality in general (Pickl & Elspaß 2019: 6). Furthermore, when applying an intra-writer approach to historical data, we need to rely on requisites that are, however, not always available to us. To study audience design, for example, at least two texts to two different addressees by an individual are required, which already restricts the usable material to a certain degree. Macro-approaches involving quantitative analyses demand a considerable amount of data from an individual to obtain empirically valid conclusions. A micro-approach, on the other hand, has no fewer prerequisites. To interpret the large array of possible influencing factors for variation, we should follow the principle of *informational maximalism*, that is, 'the utilization of all reasonable means to extend our knowledge of what might have been going on in the past, even though it is not directly observable' (Janda & Joseph 2003: 37). Each source faces its own limitations: '[T]he data that are rich in so many ways are impoverished in others' (Labov 1994: 11).

To get 'a more holistic perspective' from the past, Nevalainen (2015: 247) pleads for 'more dialogue between the *micro* and the *macro*'. Applied, for example, to language change in communities, 'we need to

12 See Walkden (2019) on the 'many faces of uniformitarianism in linguistics'.

account for the micro-level agency of individuals within the community as well as macro-level diffusion across communities' (Säily et al. 2017: 3).

A further challenge to be considered is the lack of corpora designed for the analysis of intra-individual variation. Although phenomena of intraindividual variation can often be found in data that were compiled for other research interests – or even with the intention to avoid intra-individual variation at all, as in traditional dialectology13 – intra-individual variation in corpora 'is most likely accidental and depends on corpus size' (Ulbrich & Werth 2021: 16). Historical corpora, especially those of particular interest for historical sociolinguistics, that is, those comprising texts from people of diverse social backgrounds, are usually rather small in size, which limits the analysis of intra-writer variation considerably.

However, if historical sources are copious and allow for socio-historical contextualizations, it is possible to design a corpus with the main purpose of investigating intra-writer variation. This has, for example, been undertaken with the *Corpus of Patient Documents (CoPaDocs)*, where particularly such writers were included whose letters stretch over a longer period and are addressed to various addressees. In addition, the patient files in which the texts were found provide rich meta-information about the writers and their addressees (cf. Schiegg 2022: 92–97). Also, a historical corpus designed for changes across the lifespan, such as the *Early Modern Multiloquent Authors Corpus (EMMA)* (cf. Petré et al. 2019) with around 0.5 million tokens per individual, provides a fruitful basis for research on intra-writer variation. This allows for applying quantitative methods to examine changes in individuals' grammars (see e.g. Anthonissen 2021; Fonteyn & Petré 2022). In general, great potential is also held by corpora based on one or a few individuals with a large number of texts, such as *The Mary Hamilton Papers* (cf. Denison et al. 2019–22).

As the studies mentioned in Section 4.1 have illustrated, such corpora facilitate the analysis of intra-writer variation from a broad range of perspectives. In particular, highly frequent variables, for example, in the area of orthography, have proved to be suitable for quantitative studies, which can

<sup>13</sup> See Nickel's (2021) study on intra-individual variation in dialect morphology in the Bavarian Linguistic Atlas.

furthermore be backed up by statistical evaluation (see e.g. Schiegg 2018; Schiegg & Gunkler-Frank 2021). In rich sources, both quantitative and qualitative variationist analyses may be undertaken also with less frequent phenomena, for example, in the area of lexis and semantics (see Schiegg & Freund 2019; Schiegg 2021).

Nevertheless, large corpora are not a prerequisite for the study of intrawriter variation. Phenomena such as code switching or self-corrections may also be analysed if few or even only one text by a writer are transmitted (see Gardner (7) in this volume). Texts written periodically, such as diaries, logbooks and journals, allow for diachronic intra-individual analyses (see Dörnbrack (13), Girininkaitė (15), Linzmeier (18) and Stolberg (16) in this volume). Here, historical sociolinguistics has an advantage over modern sociolinguistics, where data for real-time studies are often difficult to obtain14 so that apparent-time approaches dominate. They are, however, not ideal for the analysis of language change, as '[a]dults' grammars are not diachronically crystallized' (Petré & Van de Velde 2018: 869) and are influenced both by ongoing community change and age grading.15

When applying the concept of intra-writer variation to historical texts, it is necessary to reflect about their authorship status. In several contexts, the responsibility for a text and its language goes beyond the individual. For example, a medieval text is usually transmitted as a copy so that its language may be a combination of that of the exemplar and individual preferences of the copyist, as shown both by Wallis (21) and Iyeiri (22) in this volume. Also, textual traditions, as can be observed in highly formalized texts, such as legal documents (see Markopoulos (17) in this volume), religious texts (see Uribe-Etxebarria (19) and Currie (23) in this volume) or mixed-genre navigational journals (see Linzmeier (18) in this volume), significantly influence the linguistic choices of the writers. In order to apply the concept


of intra-writer variation to sources as those just mentioned, it needs to be broadened beyond its general sociolinguistic sense.

As a consequence, analysing intra-individual variation in historical material poses challenges while at the same time providing ample potentials, as showcased by the individual studies of our volume, which we introduce in the next section.

## 5 Structure and contents of this volume

The volume is structured into three parts. Part I is the largest, containing nine chapters on intra-writer variation in letter writing. Letters are the prototypical material in which historical intra-writer variation has been observed (cf. the studies mentioned in Section 4). Part I of this book expands this body of research: The individual papers provide a wide breadth of perspectives, as they range from Antiquity to the twentieth century, comprise different languages (Accadian, Italian, English and German), and the letters they investigate are written both by prominent figures, such as Michelangelo or John Paston I, and by outsiders of society, such as British paupers and German patients in psychiatric hospitals. The six chapters in Part II are dedicated to intra-writer variation in language contact and migration settings, pursuing questions of language (or variety) choice and lifespan change. Again, the range of different sources analysed and of methods applied is considerable. Part III finally comprises seven chapters on texts that are less prototypical material for questions of intra-writer variation, due to their high formalization, textual tradition or complex textual transmission, which all leads to a lesser degree of individuality. Nevertheless, the papers in the final part of the book show that these less obvious sources also provide insightful material for the study of intra-writer variation.

#### *5.1 Intra-writer variation in letter writing*

The book's earliest texts for the analysis of intra-writer variation stem from a large corpus of Late Babylonian letters in the Eanna Temple Archive from the middle of the first millennium BCE: Martina Schmidl (Chapter 2) analyses two nearly duplicate autograph letters from this collection, written by a high-ranking temple official. One of the letters is addressed to a high priest (similar in rank to the letter writer), the other to a temple scribe lower down the hierarchy. By carefully dissecting the differences between the otherwise very similar letters, Schmidl's qualitative analysis shows how the writer negotiates his identity towards colleagues of different ranks and how he employs different persuasive strategies for the same requests depending on the addressee.

Eleonora Serra (Chapter 3) investigates variation in discourse-ending formulae (such as 'I have nothing else to tell') in the more than 400 letters of the Italian renaissance artist Michelangelo Buonarotti. She shows that Michelangelo's variation in these formulae is subject to lifespan changes and sensitive to the type of addressee. Her detailed analyses also elaborate the enregisterment and social meaning of particular formulae in this individual's writing.

Juan M. Hernández-Campoy (Chapter 4) focuses on the Late Middle English orthographic variants <th> and <þ> for [θ] and [ð], where <th> is the incoming form, replacing <þ> by the late fifteenth century. Zooming in on the letters of John Paston I (1421–66), Hernández-Campoy shows that the variation between the graphemes is a function of audience-design and register-based style-shifting, with John Paston I using more <th> the higher the rank of his addressee, and also more <th> to legal professionals. This demonstrates how intra-writer variation may give insights into the social meanings which variants in the process of change had for their users.

Audience design is also at the centre of Chapter 5: Tamara García-Vidal explores whether the choice between synthetic and analytic comparatives (*calmer* vs *more calm*) might depend on the addressee in sixteenth to eighteenth century English correspondence. She analyses letters by five writers from different social ranks to addressees of different ranks and relationships and finds that although the writers indeed tend to use more analytic comparatives when writing upwards, this cannot be seen independently of the more frequent use of long and Romance adjectives when writing upwards.

Moving closer to the present day, Anita Auer, Anne-Christine Gardner and Mark Iten (Chapter 6) investigate the autograph pauper petitions by two English women between 1818 and 1830. They analyse the women's use of the stigmatized variant h-dropping/h-insertion as well as their use of the orthographic variant long s (<ſ>), which was falling out of use in the early nineteenth century. They observe a high degree of inter- and intra-writer variation that does not seem to be functionalized or conditioned and can best be explained as free variation due to limited education possibilities.

Chapter 7 remains in the setting of poor relief in early nineteenthcentury England but looks at a more experienced writer. Anne-Christine Gardner undertakes a close reading of a draft letter written by a curate to the bishop of his diocese to ask for support of his parochial school. The letter contains numerous stylistic self-corrections which Gardner analyses in the framework of speaker design, carefully drawing on details of the sociocultural context.

In Chapter 8, Christine Elsweiler investigates the formulation of requests in letters by two Early Modern Scottish politicians. Similar to the approach in Krogull et al. (Chapter 14) and Gunkler-Frank (Chapter 10), Elsweiler first presents a quantitative macroscopic study of request strategies in a larger corpus of Early Modern Scottish letters, showing that requestive behaviour is relatively homogenous, mostly using performative requests, and adhering to contemporary epistolary conventions. On this backdrop, a microscopic analysis of the two writers reveals that within these conventions, they vary in their requestive styles (e.g. with mitigating devices) to accommodate to their addressees and auditors (Bell 1984: 172) and to modulate different and changing social roles.

Turning to German, epistolary conventions also play a role in Lucia Assenzi's analysis of the letters of Prince Ludwig von Anhalt-Köthen from the mid seventeenth century (Chapter 9). While Ludwig himself, as cofounder of the *Fruchtbringende Gesellschaft* [Fruit-bearing Society, the first German language academy], advocates a less formal and pompous writing style, the contemporary *Briefsteller* [letter writing guides] call for chancery style with highly formulaic language and complex syntax. Ludwig's letters are therefore a particularly rewarding object of study since they are produced between two conflicting priorities. Assenzi analyses Ludwig's letters to different members of the *Fruchtbringende Gesellschaft* for addressee-based variation in syntactic and pragmatic features. She discusses her results from the perspective of attention to speech, audience/referee design and speaker design, respectively, and concludes that Coupland's speaker design framework is suited best to account for Ludwig's stylistic variation.

While the writers investigated in the previous three chapters are all highly educated, Chapter 10 returns to largely inexperienced writers. Katharina Gunkler-Frank analyses the use of clitics in patient letters from two German psychiatric hospitals. A macroscopic analysis of clitics in thirty-nine writers reveals that while preposition article clitics (such as *im* [in dem]) are linguistically conditioned, the variation in pronominal clitics is sociolinguistic and stylistic. These are regionally restricted, used more frequently by inexperienced writers, and more frequently in private than in official letters, which shows, again, the connection between inter-writer and intra-writer variation. Taking a microscopic look at three individual writers, Gunkler-Frank then shows that stylistic variation in pronominal clitics is not only a matter of audience design, but that clitics are also employed to shape the relationship between writer and addressee, creating identification and marking the conversation as intimate or emotional.

#### *5.2 Intra-writer variation in contact and migration settings*

The six chapters in the second part all treat intra-writer variation in contact and migration settings. We start with two chapters which both deal with diatopically marked variants in letter writing that are used strategically for speaker design. Joshua Brown (Chapter 11) investigates the letters of a Milanese merchant from the early fourteenth century, addressed to members of a powerful Tuscan trading company. By analysing three variables quantitatively and qualitatively, Brown finds that the writer often uses distinctively Tuscan variants instead of his native Milanese ones, even in a hypercorrect way, as a strategic attempt to curry favour by accommodating to the variety used by his recipients.

Sarah van Eyndhoven (Chapter 12) presents a quantitative and qualitative analysis of the choice between Scots and English variants in letters by two Scottish politicians in the early eighteenth century. At this time, the Scottish elite is generally said to have all but abandoned Scots in favour of Standard English in their writing, and indeed, both politicians use very few lexical and orthographic variants from Scots. However, with the intense debates about the Union of Parliament with England at the time, Scots variants are likely to have been employed for persona creation and speaker design. Van Eyndhoven shows that Scots variants are used most often in letters to recipients in England, or with diverging political views, and argues that their function is to emphasize the social and political distance between correspondents.

The following chapter stays with Scotland and contact with Standard English but moves a century forward: Nora Dörnbrack (Chapter 13) analyses lifetime changes in the writing of Mary Ann Wodrow Archbald, a Scottish farmer who emigrated to America in her forties, and who left a large body of ego-documents from before and after emigration, spanning over 40 years. Dörnbrack focuses on variants in verbal inflection and finds that non-standard forms of preterites and participles tend to recede over time, except for verbal -*s* for all persons. Possible factors in this are general prescriptive tendencies and the new speech community after emigration.

The remaining chapters in Part II all focus on language choice in material from the long nineteenth century. Andreas Krogull, Jill Puttaert and Gijsbert Rutten (Chapter 14) investigate the choice between Dutch and French in family correspondence in the Northern Low Countries. Their article (like the ones by Elsweiler and Gunkler-Frank) showcases the benefits of combining the macroscopic and microscopic perspectives: They first present results from a large quantitative study of the correspondence of thirtysix families, finding that while Dutch predominates in most letters (c. 75 %), the use of French correlates with familial relationships, and increases in constellations with women, either as sender or as recipient. In microscopic studies of three families, they show that the use of French is not always a socio-cultural phenomenon, as suggested by the big picture, but may also more directly depend on a family member's association with France, which can influence the entire family's language choice in correspondence.

In Chapter 15, Veronika Girininkaitė presents the intriguingly multilingual diary of Vytautas Civinskis, written between 1904 and 1910 and consisting primarily of Polish, Lithuanian, Russian, and German. She shows how language choice in the diary is influenced by the writer's biography and changing linguistic attitudes, how code-switches are used in quotes, as euphemisms, and for expressive reasons, and how, in his letters, the author changes between official and unofficial spelling depending on addressee. She also demonstrates that while typically the author also switches the script when he switches between Russian (Cyrillic) and the other languages (Latin script), the script switches are not always simultaneous with the code switches, but may appear with a lag, even within a lexeme – probably due to reduced attention to writing (functionalized IAV).

Code-switches and corresponding script switches are also the topic of Chapter 16. Doris Stolberg investigates language choice in a diary from nineteenth-century Canada, produced in the setting of a German/English bilingual community with German as the heritage language and extending over almost 70 years of the writer's life. Language choice is often influenced by biographical events or by the topic of the text. The writer largely uses German cursive script for German and Roman script for English and other languages; he even uses self-corrections when he started a word with the 'wrong' script, so that script choice is a way to tell nonce-borrowings from established loans. However, Stolberg also finds what she calls 'script crossing', that is, the lack of a script switch to match the code switch. She argues that script variation, usually a conditioned type of IAV, dependent on code variation, can also be functionalized pragmatically to indicate that the referent of a word is (un)familiar or to distinguish new from given information.

#### *5.3 From intra-writer variation to variation beyond the individual*

All of the chapters so far have dealt with letters and diaries, the most obvious material to study intra-writer variation. The articles in the final part now move away from ego-documents to different text types that widen the perspectives on intra-writer variation. Theodore Markopoulos (Chapter 17) showcases that audience effects on intra-writer variation may be successfully studied even in legal texts. A challenge with notary acts is that we do not know to what extent the notaries acted as delegate scribes, writing down the words of their clients rather than their own (so that the observed variation would be *inter*-speaker rather than intrawriter). In his analysis of Early Modern Greek acts by a prolific notary from the island of Crete, Markopoulos solves this problem by concentrating on the formulaic passages, predominantly at the beginning or end of the text, as these are highly likely to have been written by the notary himself. He focuses on three morphosyntactic variables occurring in the formulaic parts, and shows in a quantitative analysis that there is a clear audience effect, for example, as to whether the act involves clients from the elite, or whether the main participant is a town-dweller rather than from the countryside.

Laura Linzmeier (Chapter 18) investigates French navigation journals and logbooks from the eighteenth century, another official text type used in professional settings. She starts by outlining the sociocultural context and the multiple functions of this largely unexplored text type, showing that navigation journals are essentially hybrid documents, fed by various discourse traditions. She then presents a qualitative analysis of a French navigation journal, demonstrating how the writer changes his styles depending on his changing communicative goals.

In Chapter 19, Oxel Uribe-Etxebarria looks at spelling variation in two manuscript texts of religious and moral instruction, produced by an early nineteenth-century Basque clergyman in his local dialect rather than in the prestigious variety. Basque orthography was not yet standardized at the time and writers adopted various spelling conventions from Spanish or French, or innovated new ones. Uribe-Etxebarria shows how in the few years' time that lies between the two texts, the writer becomes much more consistent in his spellings and adapts them to the phonemic system of Basque, thus also making the spelling more 'reader-friendly', considering the limited education of his audience.

The three chapters following examine medieval vernacular texts that pose particular challenges for the concept of intra-writer variation due to their complex textual transmission which involves not only one but several individuals, that is, author(s) and copyist(s). Christine Wallis (Chapter 20) and Yoko Iyeiri (Chapter 21) focus on spelling variation, a challenging topic for medieval texts, where the copyists may either (a) copy their exemplar *literatim*, even if this does not correspond to their own preferred usage, (b) replace the forms from the exemplar by their own preferred forms ('translator scribes'), or (c) use both strategies ('mixer scribes') (McIntosh 1973; Laing 2004). Copying from an exemplar therefore also constitutes a source for conditioned variation, which may become functionalized depending, for instance, on attention to writing (see Section 4.2).

These intricacies are investigated by Christine Wallis (Chapter 20), who takes a close look at spelling variants used by the eleventh-century English scribe Hemming of Worcester. By comparing the variants the scribe uses across various texts (scribe as control), and by comparing his variants to those in other surviving copies (text as control), she teases apart which of the variants must be due to the exemplar, and which a reflection of the scribe's own use.

Yoko Iyeiri (Chapter 21) deals with these different copying strategies as well. She shows how a scribe starts a text with a more *literatim* strategy and then increasingly turns to their preferred use, though sometimes employing the exemplar forms for emphasis. In this case, then, the conditioned variation is functionalized differently by the scribe. The shift to the scribe's preferred usage may, however, also take place abruptly (from strategy a to strategy b) as Iyeiri shows with the change from *womman* to *woman* in a fifteenth-century manuscript. Iyeiri argues that the text frequency of the lexical item and the 'value' of the exemplar text are factors in the retention or replacement of the exemplar variant.

Phil Beier, Gohar Schnelle and Silke Unverzagt (Chapter 22) conduct a register analysis to investigate so-called instructions from the Old High German Psalter by Notker and the Old Swedish Revelations of St Birgitta. They analyse how the use of imperatives, subjunctives, and modal verbs in instructions is affected by the social role the characters issuing the instruction occupy toward their addressee. The additional challenge here is that variation may be influenced by factors not only on the level of the author and the copyist, but also on what these believe to be the most adequate variant for the respective social role relationships of the characters in the text.

The final article by Oliver Currie (Chapter 23) broadens the perspective on intra-writer variation to literary style shifting by studying intra-writer variation in word order in two sixteenth-century Welsh Bible translations. Drawing on quantitative analyses, he shows that there is a striking difference in word order between the poetic books of the Old Testament, where positive declarative main clauses frequently have verb-initial word order, and the prose books of both Testaments, where this word order is rare. Currie demonstrates that word order variation in the translations is to some extent linguistically conditioned by the type of subject. Beyond this, however, he argues that the two translators innovatively functionalized the existing indexicality of word order – verb-initial order being associated with contemporary as well as earlier Middle Welsh poetry – to convey a poetic quality to their prose translations of Biblical Hebrew poetry.

The chapters of this volume show that historical sociolinguistics has a great deal to gain by empirical and corpus-based studies focusing on intraindividual variation. These refine the coarser results yielded by macroscopic studies, revealing how writers functionalize social or text-type variation for reasons such as audience design or persona creation. They may also provide insights into how ongoing changes were perceived by speakers and writers. In general, micro-approaches are able to uncover a wide array of possible factors influencing variation which may, however, not always carry sociolinguistic functions but may also be conditioned by linguistic or other, external factors or be classified as non-conditioned, free variation. By carefully taking into account the sociocultural context of the material – from the hierarchies of Babylonian temple officials (Chapter 2) to the changing social roles of a politician in Early Modern Scotland (Chapter 8), the stratification of society in Venetian-ruled Crete (Chapter 17), or the literary conventions of Middle Welsh (Chapter 23), to name but a few –, the studies collected in this volume demonstrate that the challenges that historical material holds for research on intra-individual variation can certainly be met and that the insights gleaned from intra-writer variation are considerable.

## References


D. Joseph & Richard D. Janda (eds), *The Handbook of Historical Linguistics*. Oxford: Wiley. 3–180.


part I

Intra-writer variation in letter writing

#### Martina Schmidl

2 A qualitative approach to intra-writer variation in late Babylonian letters: Two near-duplicate letters from the Eanna archive (528 BCE)

#### Abstract

This chapter examines a pair of near-duplicate letters sent by Nabû-aḫu-iddin, royal trustee of the Eanna temple in Uruk in southern Babylonia in 528 BCE. The letters address the high priest and a temple scribe of this temple, two officials from priestly, non-royal families. A qualitative approach focusing on the differences in the pragmatics of the two letters allows for an examination of Nabû-aḫu-iddin as an actor with a royal background within the temple hierarchy, and of the means he employed to position himself within this hierarchy, negotiating his relationship with two local officials of different hierarchical status.

## 1 Introduction1

Historical sociolinguistic approaches are rare in the field of Assyriology, with the laudable exception of Sallaberger (1999), who focused on Old Babylonian letters from the second millennium BCE. Work on letters

1 This chapter was written within the framework of the project REPAC, 'Repetition, Parallelism and Creativity: an Inquiry into the Construction of Meaning in Ancient Mesopotamian Literature and Erudition' (2019–24, University of Vienna), funded by the European Research Council within the Horizon 2020 research and innovation programme (Grant agreement no. 803060). It is based on material collected under the auspices of the Austrian National Research Network (NFN) 'Imperium and Officium', funded by the *Fonds zur Förderung der wissenschaftlichen Forschung* (FWF) (Vienna). I thank Yuval Levavi and Michael Jursa for discussing this chapter with me, and Frank Simons for correcting my English. Any remaining mistakes are, of course, my own. Conversions of Babylonian to Julian dates were done using from the first millennium BCE Babylonia is still in its infancy (e.g. Schmidl 2017). This chapter presents an in-depth case study of two nearduplicate letters, YOS 3, 17 and TCL 9, 129, dating to the reign of the Achaemenid king Cambyses (529–22 BCE). They were written on cuneiform tablets sent to the Eanna temple in Uruk in southern Babylonia in 528 BCE (Tolini 2011: 46f.; Kleber 2012a: 230).2 The pair is part of a corpus of around 700 letters which stem from the archive of the Eanna temple (Hackl 2007: 3).3

The two letters in question were sent by Nabû-aḫu-iddin, the 'royal trustee' (*ša rēš šarri bēl piqitti ayakki*) to two of his fellow officials at the temple, Nabû-mukīn-apli, the 'high priest' (*šatammu*), and Nādin, one of the temple scribes (*ṭupšarru ša ayakki*). These officials belong to the highest echelons of the temple hierarchy. In the period under review here, a 'royal resident' (*qīpu*) stood at the top of the temple, followed by the high priest and the royal trustee. These officials were supported by several temple scribes, in addition to other temple officials and regular scribes at the temple. Their hierarchy can be illustrated as follows:

the Babylonian Calendar Converter, <https://webspace.science.uu.nl/~gent0113/ babylon/babycal\_converter.htm> accessed 14 June 2021.


While the official hierarchy is clear from legal and administrative documents, where the high priest is always listed before the royal trustee in this period, their actual responsibilities were virtually the same, apart from the high priest participating in the cult.4 Both were involved in basically all aspects of the temple administration. The royal trustee was away from the temple more often to take care of issues relating to agriculture and building projects, but the two officials could also switch their places (Kleber 2008: 18ff.). This made their hierarchical relationship somewhat fluid, and in practice there was no clear relationship of superiority and subordination between them. A formal characteristic of Nabû-aḫuiddin's letter dossier reflects this relationship. He employs the appellation 'brother' (*aḫu*), used for equals or subordinates, both in letters to the temple scribes as well as to high priests (Schmidl 2017: 381).5

Nabû-aḫu-iddin is attested in office from 539 until 526 BCE (Kleber 2008: 36). His letter dossier consists of sixteen letters sent by him, thirteen letters addressed to him personally, and dozens of incoming letters to the temple addressed to both a high priest and the royal trustee. Nabûaḫu-iddin's dossier is especially well-suited for a sociolinguistic approach because epigraphy and writing idiosyncrasies make it highly likely that he wrote his letters himself (Jursa 2012: 380; Schmidl 2019: 138).


## 2 Methodology

The present analysis of the two near-duplicates is embedded in the Third Wave of Historical Sociolinguistic studies (Eckert 2012) and the roughly congruent turn towards Speaker Design (Schilling 2013: 328). A close reading will focus on how the sender, Nabû-aḫu-iddin, performs his roles and identities within the temple in his written communication with two other temple officials. In addition, the analysis builds on a classification system of rhetorical strategies devised by Jursa and Hackl (2015). They distinguish between five persuasion strategies employed to achieve the requests made in Babylonian letters from the Eanna and other contemporary archives (c. 626–484 BCE).

One is a bare statement of request without any supporting arguments, the most common way to express a request. The other four are rational elements, emotional or personal appeals and the invocation of higher authorities (Jursa & Hackl 2015: 106ff.). Rational argumentation works with 'objective arguments […] based on general practical, logical, legal, or […] social principles whose validity is presupposed *a priori* and which are not based on considerations *ad personam*' (p. 107). It is the strategy used most frequently in the entire letter corpus if there is any support provided for a request at all (pp. 110f.). Emotional appeals put 'verbal or nonverbal behaviour into writing to communicate emotion' (p. 108), with oaths – statements usually sworn by deities – as a special category among such emotional requests. Personal appeals use elements such as personal favours and attempt to obtain 'help voluntarily and out of goodwill, rather than on the basis of "professional" obligations' (p. 108). Invocations of higher authorities mostly refer to secular authorities, not to the divine realm. These invocations are twofold: on the one hand, they may phrase a request as a necessity to comply with an order of the said authority, on the other hand, they can be used as threats. These five strategies are distributed differently in letters from superiors to subordinates, between roughly equals, and from subordinates to superiors (pp. 110f.).

Figure 2.1. Cuneiform tablet YOS 3, 17. Courtesy of the Yale Babylonian Collection; photography by Klaus Wagensonner.

## 3 Case study

The two letters which form the basis of the present study are currently housed in the Louvre in Paris (TCL 9, 129) and in the Yale Babylonian collection in New Haven (YOS 3, 17), respectively (see Figure 2.1; see translations of the letters in the appendix). They were written in the context of construction work for which the temple was responsible. The Eanna temple was not only a place for cultic worship but it was also an important institution and had extensive landholdings and a considerable workforce. Apart from its cultic tasks, the temple was involved in agriculture, animal husbandry and trade to procure everything necessary for the upkeep of the cult and the institution itself as well as to fulfil its duties towards the crown. An important part of these duties was the provision of labour for construction sites. Labour was a sought-after resource at this time, and shortages of men or their supplies are a recurrent topic in the letter corpus in question. This is also the context our letters originate from. Nabû-aḫu-iddin wrote them to his colleagues at the temple because of his dire need for workers and their wages, as well as his demand for a scribe to handle their administration at a construction site. Judging from the content of the letters, there is no obvious reason why Nabû-aḫu-iddin sent two letters to his colleagues, as both could have taken care of his main demands.6 Although the letters run parallel for the most part, YOS 3, 17, addressed to the high priest, Nabû-mukīn-apli, contains some additional content, making it one of the longest letters found in the entire corpus from which this pair is taken. Most of these additions constitute rhetorical techniques rooted in Nabû-aḫu-iddin's different relationships with his two colleagues.

In the following, I first briefly discuss intra-writer variation in the formal parts of Nabû-aḫu-iddin's letters (Section 3.1). I then proceed to a detailed treatment of instances in which both letters contain identical sentences with only minor variation (Section 3.2). This is followed by a

<sup>6</sup> The main demands could have been fulfilled by both officials. See Section 3.3 on some minor differences in content.

treatment of the additional content present in the letter to the high priest, including an analysis of the evidence these variations provide for Nabûaḫu-iddin's different performances of his relationship with the high priest and with the temple scribe (Section 3.3).7

## *3.1 Intra-writer variation in the formal parts of the letters*

Formally, the letters differ in that Nādin, the temple scribe, is addressed by name, while Nabû-mukīn-apli, the high priest, is addressed by title only, as is the case in most contemporaneous letters in the Eanna corpus. In addition, the greeting formula differs. Nabû-aḫu-iddin addresses the high priest with a formula wishing for his well-being, while in the letter to the temple scribe he uses a formula which contains a wish for a divine blessing.8 In the letter heads, Nabû-aḫu-iddin refers to both addressees with the appellation 'brother' (*aḫu*), which is suitable for equals and subordinates.


## *3.2 Intra-writer variation in the letter body: Minor variation between parallel sections*

The three excerpts discussed in this section are of particular interest because of the slight differences between otherwise identically phrased passages in both letters. (1) concerns forms of address. In the letter to the temple scribe (TCL 9, 129: 5–7), (1a), direct address is used, that is, second person singular present tense (*tašakkanu*), and imperatives (*epuš*, *muḫuršunūti*, *šupraššunūti*). The letter to the high priest (YOS 3, 17: 4– 6), (1b), maintains the imperatives, but instead of the direct address in the second person singular, the third person singular, combined with the appellative 'lord', is used (*bēlu išakkanu*), that is, indirect address.


This is the only instance in the two letters in which a direct address in the letter to the temple scribe corresponds to an indirect address in the letter to the high priest. Otherwise, both letters use only direct forms. The position of this passage may be of importance. In both texts, this passage is the first request to contain a direct address to the recipients within the letter body. The use of indirect address, that is, a third-person verb and an appellation more respectful than that of the letter head ('lord' vs 'brother') at the beginning of the letter may have been used to frame the letter, setting the scene for the directly phrased remainder of the text.

(2) revolves around a request for workers (2a: TCL 9, 129: 40–42; 2b: YOS 3, 17: 43–45). Both letters request that certain rosters of workers be checked for available men. Although the grammatical form of the verbs is the same, the choice of words is informative for our research questions. In the letter to the temple scribe, Nabû-aḫu-iddin employs a well-attested away, send them all (to me)!]

stock phrase for a warning against negligence (*lā tašelli*), while in the letter to the high priest, he conveys the same message with a less conventional phrasing (*lā tumaššar*):


The phrasing in the letter to the high priest does not make any explicit reference to negligence. Instead, it relates to (trust in) the high priest's authority or sense of duty, presupposing that he will do what is necessary and stressing that he has the power to do this successfully, even if the people involved may not comply at first. This cannot simply be explained by the difference in rank of the two officials addressed. We know from other letters by Nabû-aḫu-iddin that admonitions for negligence were possible in letters to both officials (Schmidl 2017: 382). Equally, shortly before the quoted passage in the letters studied here, improper behaviour – going against the rules of the temple administration – is suspected both from the high priest and the temple scribe. In YOS 3, 17: 35ff. and TCL 9, 129: 33ff., we read 'If you say as follows, "Nabû-aḫu-iddin should (simply) go against procedure in this!" – as much as you, yourself, have gone against procedure, I certainly will not go against procedure!'.

(3) occurs in the description of a problem with workers. A man supposed to provide workers has sent (only) four men. In the letter to the temple scribe, this is stated plainly (3a). In the letter to the high priest, however, the statement is supported by an oath sworn by the gods, stressing the low number of workers standing available for work (3b).


This oath is noteworthy for two reasons. On the one hand, as stated above, oaths are a special category of emotional rhetorical strategy. As such, this difference in phrasing plays out on a different rhetorical level, drawing on emotions. On the other hand, Nabû-aḫu-iddin emphasizes the difficult circumstances at the construction site because of his need for workers only in the letter to the high priest. This can be connected to the particular hierarchical relationship between these two officials. In a discussion of rhetorical strategies in letters of high-ranking officials, Kleber (2012b: 228) has shown that appeals to empathy occur only in letters exchanged between officials of roughly equal status, remarking 'dignity seems to forbid writing about their dilemma to lower ranking officials'.9 They do not occur in letters to subordinates, as we will also see below. Based on these patterns, Nabû-aḫu-iddin positioned himself on a polite, though roughly equal footing with the high priest, and as a superior of the temple scribe.

## *3.3 Intra-writer variation in the letter body: Content addressed only to the high priest*

We now move on to content present in the letter to the high priest, but not in the letter to the temple scribe. One of these additions concerns a special type of worker only mentioned in YOS 3, 17: 47–49. Here, differences in administrative responsibilities could serve as an explanation – the temple scribe may not have possessed the authority to command these specific workers. Another addition, also concerning workers (YOS 3, 17: 23[ff.]), is unclear as it occurs in a broken section of the tablet. Other examples, however, are linked to the rhetorical means Nabû-aḫu-iddin employed in the letter to the high priest only. The additions in the letter to the high priest range from such small elements as a particle, *amur*, to the inclusion

<sup>9</sup> This relates specifically to elements which can be taken as personal shortcomings, while complaints about hard work are common in letters more generally (Schmidl 2019: 86f.).

of entire sentences not found in the letter to Nādin. These elements present a similar pattern to the examples discussed in Section 3.2.

The first addition occurs at the beginning of the letter body (4). It is the continuation of (1b) in Section 3.2. In (1), Nabû-aḫu-iddin urges his correspondents to send men back quickly, using direct address for the temple scribe but indirect address for the high priest. A reason for the request is only presented in the letter to the high priest:

(4) *kī ina muḫḫi dulli ušuzzū ṣābūšunu ul iḫalliqū* (YOS 3, 17: 7–8) [If they are present in the work(place), their men won't run away.]

Similar to the pattern we have seen above, the passage in the letter to the high priest relates to troubles with Nabû-aḫu-iddin's authority and performance in the workplace. The liability of his workers to run away may be seen as his own personal shortcoming. Such a reading is supported by the fact that this passage is not mentioned in the letter to the temple scribe. It likely aims at the high priest's pity or empathy. Since both the royal trustee and the high priest shared their responsibilities, it may also play on their joint administrative duties. Based on this assessment, I take this passage as an instance of the emotional rhetorical strategy.

Similarly, (5) relates to the theme of trouble with Nabû-aḫu-iddin's work and authority. Once again, it aims to persuade on the emotional plane with regard to its content, but it also does so in form: it uses an oath. The passage states that if Nabû-aḫu-iddin is not personally at the site, not only does work come to a halt, but his previous work is undone:

(5) *ilū lū īdû kī 3 ūmū ša ana Uruk alliku dullā ša araḫ ūmī lā baṭil* (YOS 3, 17: 38–40) [The gods know that in these three days that I went to Uruk, my work of a full month was ruined.]

Another passage at the end of the letter to the high priest draws on these motifs. In (6), Nabû-aḫu-iddin frames his troubles by a comparison of his work output with that of a district governor (*bēl pīḫāti*) working on the same site. The district governor, a royal official, is performing his work perfectly, while Nabû-aḫu-iddin cannot keep up the same pace. We are again dealing with an emotional strategy. It may additionally convey the nuance of a royal official noticing the temple's malperformance in a task for which both the royal trustee and the high priest could be responsible.

(6) *bēl pīḫāti ṣābū mādūtu ibbak mišḫu igammarū anāku ammerekku* [ak]*ī* ?  *ikkarū tapqidu l*[ūmurm]*a*? *ṣābūšunu ša perri lībukūnu* (YOS 3, 17: 51–53) [The district governor will take a lot of workers with him, they will complete (his) section (while) I am (lagging) behind. [*As soon as*] you have assigned the ploughmen, I want to [*inspe*]*ct* (them) and they should bring their excess men along.]

The authority of the high priest, and possibly also their joint administrative responsibilities, also play a role in (7): the use of the particle *amur* only in the letter to the high priest. This particle is a frozen imperative of the verb *amāru* 'to see', but retains its basic visual quality of checking information when used as a particle (Schmidl 2014: 29f.). Here, the particle hints at the authority of the high priest. Nabû-aḫu-iddin states facts, but he also indicates that the high priest can check his information himself if he wishes. This may aim at their shared responsibilities, or refer to the high priest's powerful status in the temple *per se*.

(7) *Ina-ṣilli-Nanāya u Aqria nikkassī ittišunu nītepuš rēḫšunu mīṣ amur akanna dullu ippušū* (YOS 3, 17: 15–17) [We balanced the accounts with Ina-ṣilli-Nanāya and Aqria. Their remainder (of debt) is small. As you can see, they are doing their work here.]

Based on this, the additional passage immediately before the use of the particle should be seen as highlighting the joint duties of the two officials. Nabû-aḫu-iddin remarks on an administrative process having taken place in the past, the balancing of accounts (*nikkassī ittišunu nītepuš*). The letter to the temple scribe does not contain this information. It is possible that the two officials, the high priest and the royal trustee, actually balanced these accounts together. It could also serve as a reference to their joint responsibility to keep the accounts in order, an extremely important task in the temple administration generally and especially in letters from Nabû-aḫu-iddin (Schmidl 2019: 164–68). These interpretations are not mutually exclusive. In any case, the passage adds another element linking the letter parties.

The clearest case of the use of emotional support to build a connection and to express reciprocity can be found in a vivid image Nabû-aḫu-iddin draws up in the letter to the high priest (8). In this image, the roles of the officials are reversed, putting the high priest in Nabû-aḫu-iddin's shoes and stressing that Nabû-aḫu-iddin would do everything that was needed in this case.

(8) *ṭupšarru u sēpiru ittia yānu kī ṭupšarru u sēpiru maṭû u dullu ina muḫḫikunu šupramma lušpurma ṭupšarru lūbuk* (YOS 3, 17: 32–34) [There is neither a scribe nor an alphabet scribe with me. If there were not enough scribes or alphabet scribes and you (pl.) were responsible for work, you'd (sg.) (just) write to me, and I would send (an order) and provide a scribe.]

Aside from emotional elements, the passage may additionally draw on a personal element. Nabû-mukīn-apli switches from the plural to the singular in his description of the problem with which he would help out. This adds a more personal component. On the one hand, Nabû-aḫuiddin first stresses that he would generally help temple officials if they were in his stead at a construction site, but on the other, he then breaks this general statement down to the personal level by using the singular, stressing that he would do this for the high priest without quibbling if their situations were reversed. Therefore, Nabû-aḫu-iddin presents himself not only as someone who would always perform the responsibilities of his position at the temple,10 he subtly adds a more specific, personal component to support his argument.

## 4 Conclusions

Nabû-aḫu-iddin relies on emotional elements in his letters to both the temple scribe and to the high priest, for instance, when he accuses both officials of improper behaviour. We have seen, however, that there are additional emotional elements in the letter to the high priest. These draw on several persuasive elements, among which are emotional elements and aspects of a sense of community, such as reciprocity and joint

<sup>10</sup> Proper administrative procedure in general, but also specifically Nabû-aḫu-iddin's proper handling of administrative procedures, is an important topic in his letters (Schmidl 2019: 161–68).

undertakings of the two officials. Elements referring to authority and performance are important as well within this emotional framework. These mostly refer to Nabû-aḫu-iddin's own shortcomings, but they also pertain to the power of the high priest. Such admissions of shortcomings are not suitable when addressing subordinates. They are therefore not present in the letter to the temple scribe, showing that Nabû-aḫu-iddin was clearly aware of his own higher status. The presence of these admissions in the letters to the high priest, on the other hand, shows Nabû-aḫu-iddin positioning himself roughly as the high priest's equal. The difference in hierarchy and status with regard to the temple scribe is further underlined by the fewer means of rhetorical support employed in Nabû-aḫu-iddin's letters to the temple scribe, which indicate that in some cases, he assumed that his requests would be carried out by the temple scribe even without the means used in his communication with the high priest. This is an important aspect not only for Nabû-aḫu-iddin personally, but also for the development of administrative structures and the bureaucratic mentalities attached to them.

The high priest, Nabû-mukīn-apli, and the royal trustee, Nabû-aḫuiddin, shared a considerable amount of time in office, and they must have had a personal relationship. The nature of this relationship, however, does not seem to have been close. Nabû-aḫu-iddin focuses primarily on emotional elements. Friendship or personal favours do not play a role in the persuasion strategies of Nabû-aḫu-iddin's letter to the high priest. Similarly, threats are not attested. Both personal favours and threats are rare in the contemporaneous letters in our corpus,11 but they occur with greater frequency in the letter dossiers of officials working on construction sites. These officials were of high and intermediate rank and possessed ties to the crown, and they threatened high and intermediate ranking officials of local extraction, among them the high priest and the temple scribe (Schmidl 2021). Although our pair of letters fits this setting, Nabû-aḫu-iddin does not choose this strategy, either here or in the rest of his letter dossier. This

11 Generally speaking, threats are restricted to letters from superiors to subordinates and to letters where the hierarchical relationship is somewhat fluid, as is the case here (Kleber 2012b). On favours, see preliminarily Schmidl (2020: 263f.).

means that he does not draw on his royal background in his choice of persuasion strategies, and that he does not stress this aspect of his identity. At present, the reason for his avoidance of threats cannot be ascertained. It could link to a personal preference, or to his office. The royal trustee was far more involved in everyday administration at the temple than the royal resident, who has only left us letters about construction sites. Further research is necessary to approach this question. Nonetheless, Nabû-aḫuiddin's choice of strategies is informative about the relationships of the royal trustee at the temple. A close reading of our near-duplicates provides not only important insights into temple hierarchies and what holding an office entailed, but also an expression of Nabû-aḫu-iddin's identity, which he negotiates and performs with regard to his position in the temple, his background, and towards his colleagues.

## References


## *Appendix: Full translation of both letters*

The following translations are based on Schmidl (2019: 434–36 and 439f., nos 37 and 38), with slight modifications. Line numbers are indicated in brackets at the beginning of each section. Round brackets are used for words added for a coherent or more easily accessible translation. Square brackets indicate broken or restored sections on the tablet, with italics marking uncertain restorations. Pointed brackets contain omitted elements.



(26–46) (The cost of ) 200 hired labourers is owed by me. I brought silver [and wool], but I couldn't give it to them without a scribe or an alphabet scribe. Scribe and list are there (with you). One talent of silver does not enter the Eanna in a whole year. Who am I that in a single month I have to spend two talents of silver as expenses?

If you say as follows, 'Nabû-aḫu-iddin should (simply) go against procedure in this!' – as much as you, yourself, have gone against procedure, I certainly will not go against procedure!

Later, they will tell us as follows, 'The temple serfs should have dug one half of the extent, and for the (other) half you should have given silver to hired labourers.' Check the ledger of temple serfs and whoever is in there, don't be negligent about this, send [them all] (to me)! The gentlemen who are in there should employ hired labourers'; otherwise, send (these men themselves) before me!

Have as much silver as has gone in(to the Eanna) delivered to me!

(27–51) (The cost of ) [200] hired labourers is owed by me. I brought silver and wool, but I couldn't give it to them without a scribe. Scribe and list are there (with you). One talent of silver does <not> enter the Eanna in a whole year. Who am I that in a single month I have to spend two talents of silver as expenses? There is neither a scribe nor an alphabet scribe with me. If there were not enough scribes or alphabet scribes and you (pl.) were responsible for work, you'd (sg.) (just) write to me, (and) I would send (an order) and provide a scribe. If you say as follows, 'Nabû-aḫu-iddin should (simply) go against procedure in this!' – as much as you, yourself, have gone against procedure, I certainly will not go against procedure! The gods know that in these three days that I went to Uruk, my work of a full month was ruined. Later, they will tell us as follows, 'The temple serfs should have dug one half of the extent, and for the (other) half you should have given silver to hired labourers'. Check the ledger of temple serfs and don't let anyone who is in there get away, send them all (to me)! The gentlemen who are in there should employ hired labourers; otherwise, send (these men themselves) before me. In addition, there are no workers of the cattle shed of Eanna and of the royal cattle shed from in there (i.e. the list) here. Have as much silver as has gone in(to the Eanna) delivered!


#### Eleonora Serra

# 3 The use of discourse-ending formulae: Exploring intra-writer variation in Michelangelo Buonarroti's correspondence

#### Abstract

The Florentine artist Michelangelo Buonarroti was a prolific letter writer: more than 400 autograph letters – most of which are private and everyday letters – have survived that address a range of individuals from different backgrounds and span 68 years (1496–1563). This chapter investigates intra-writer variation in Michelangelo's use of a set of discourseending formulae, charting their usage in relation to time and to the relationship with the addressee. This focus on intra-writer variation allows one to investigate the relative roles of writing experience and group practices in Michelangelo's language, suggesting that, for him, this set of formulae functioned primarily as social conventions.

## 1 Introduction

In light of the growing interest in letters as sources for studying variation and change in the past, the use of epistolary formulae has recently attracted considerable attention cross-linguistically. Studies on Germanic languages (Austin 2004; Elspaß 2005; Rutten & van der Wal 2012, 2014) and French (Große et al. 2016) have uncovered a correlation between a high use of epistolary formulae and low levels of writing experience, and have thus suggested that formulae functioned primarily as aids that helped less skilled writers to compose a text. However, other scholars have underlined that epistolary formulae could also function as markers of social identity, and could be used to signal in-group membership (Laitinen & Nordlund 2012; Conde-Silvestre 2016; Evans 2020). The interplay between the role of writing experience and group practices in the use of epistolary formulae remains unclear to date (Rutten & van der Wal 2012: 195).

In this chapter, I ask whether patterns of intra-writer variation within the language of a single individual may shed some light on the role that epistolary formulae performed in private and everyday letters from Renaissance Italy. Although the letter-writing manuals, epistolaries and letter anthologies that were flooding the printing market in sixteenth-century Italy have been the object of a number of in-depth studies (see, for instance, Quondam 1981; Matt 2005; Braida 2009), the use of epistolary formulae itself has re mained largely unexplored in the Italian context. When it has been investigated, it has usually been situated within a broader research encompassing lexis, style, topoi and discourse strategies, either with a focus on mostly literary letters (as in Barucci 2009) or on diplomatic correspondence (as in Felici 2018), whereas less attention has been paid to everyday, private letters. Even in this tradition, however, some studies have suggested a relative familiarity of little educated writers with letter-writing conventions (e.g. Palermo 1994; Telve 2019). This chapter aims to address this gap in the Italian context, in particular by investigating the correlation between the use of a set of epistolary formulae and the extra-linguistic factors of time and relationship to the addressee, and by evaluating whether patterns of intra-writer variation favour an interpretation of formulae as aids for formulation or as primarily social conventions.

The formulae investigated here are discourse-ending formulae: not to be confused with closing formulae, these optional formulae, as explained in detail in Section 5, signal the transition from the context-dependent information to the closing formula or from one piece of information to another.

The correspondence I analyse was written by Michelangelo Buonarroti (1475–1564), who, throughout his remarkably long life, was an incredibly prolific letter writer, mostly producing letters for practical purposes addressed to a wide and diverse range of people. He was an upwardly mobile individual whose writing experience increased considerably, and whose social networks drastically changed, thus enabling an assessment both of the role of writing experience and of social practices in his use of formulae. Further, the fact that many letters by his correspondents and family members have come down to us makes it possible to situate his usage within his broader network.

## 2 Michelangelo and his practice of letter-writing

This section gives a research overview on the Buonarroti letters, sketches a biographical and linguistic profile of Michelangelo, and provides some basic information on his correspondence and on the possible routes for his acquisition of letter-writing conventions.

## *2.1 The Buonarroti letters*

Michelangelo's letters and those of his correspondents were published between 1965 and 1983 (Barocchi & Ristori 1965–83), soon followed by the edition of other letters from the Buonarroti family archive (Barocchi et al. 1988–95).

Aside from being studied for their historical value, Michelangelo's letters have drawn the attention of linguists since at least the 1960s and, also thanks to the efforts of the Memofonte foundation, which has digitized this corpus and made it available online,1 the last few years have witnessed a flourishing of linguistic studies on this corpus. A number of studies have shed light on the linguistic features used by Michelangelo (Nencioni 1965; D'Onghia 2014; Valenti 2020), the evolution of his orthography over time (Bardeschi Ciulich 1989), the lexicon employed by Florentine artists and craftsmen (Barocchi & Maffei 1994), and the linguistic usage of Michelangelo's numerous correspondents (Serra 2020).

1 Cf. <https://www.memofonte.it/> accessed 1 March 2022.

#### *2.2 Michelangelo: A linguistic and letter-writing profile*

Before discussing my research hypotheses, it is worth surveying what we know, or might hypothesize, about the way Michelangelo learned to write letters, an issue which is tied to what we know about his life, education and writing training.

Michelangelo Buonarroti was born in 1475 to an impoverished Florentine family that belonged to the minor aristocracy. He grew up in the family villa in Settignano, a small village in the Florentine hills which primarily based its economy on stone carving, where he first approached sculpture (Wallace 2010: 51). However, since an artistic career was considered a step-down for a boy of his rank, he was initially sent to the grammar school of Francesco d'Urbino in Florence, but he never learnt much Latin here (Wallace 2010: 40). While passionate about drawing, he did not show interest for humanistic learning and his father reluctantly agreed to take him out of school and have him trained as an artist. There is general agreement that the formal education received by Michelangelo was modest (Hatfield 2002: 230; D'Onghia 2014: 93), since at the age of twelve he was already running errands for the workshop of the painter Domenico Ghirlandaio. Lacking in the kind of education that boys from the aristocracy usually received, he apparently suffered for this later in life (Hatfield 2002: 230), as he would eventually climb his way into patrician circles where knowledge of the classical languages was the norm.

In terms of the letter-writing training he might have received in his childhood, anecdotal evidence scattered across family books indicates that, in Renaissance Florence, a portion of elementary instruction might have been dedicated to the study of vernacular letter-writing (Witt 1995). Although his schooling was limited, Michelangelo would have doubtless received elementary instruction and it is therefore possible that he would have undergone some formal training in letter-writing. As noted by D'Onghia (2014: 93), however, he would not have had the kind of letterwriting training received by other boys from the Florentine aristocracy, who would usually undergo such training at an age when Michelangelo had already left school.

There were vernacular letter-writing manuals that circulated in Florence, such as Bartolomeo Miniatore's *Formulario*, first printed in Bologna in 1485 and reprinted several times in Florence since 1488. Hence, although we do not know that the Buonarroti family owned a copy, the use of manuals cannot be ruled out. On the other hand, it was extremely common in Florentine merchant families – and the Buonarroti were no exception – to keep a family archive where personal correspondence was preserved (Focarile 2022), so letters received by the household might have served as models for learning letter writing.

It has been hypothesized, however, that Michelangelo's first real 'exercise' in writing was indeed represented by his increasingly intense correspondence not only with colleagues and workers, but also with a range of high-ranking individuals (D'Onghia 2014: 93f.). The practical need to correspond – which increased as Michelangelo became an established artist receiving commissions from patricians and even popes – would have progressively familiarized Michelangelo with letter-writing conventions. A more secure writing style and mastery of certain stylistic structures in letters from the early sixteenth century, compared to Michelangelo's earliest letters from the 1490s, has been recognized (D'Onghia 2014).

In addition to this 'hands-on' training, Michelangelo also later made up for his lack of formal education by developing an interest for Tuscan vernacular literature and poetry. He had enjoyed writing verses as an amateur poet since the early 1500s, but his poetic endeavours became more serious from the 1530s, years that represent a sort of watershed in his life. During this time, he permanently settled in Rome and became friends with individuals from the aristocratic elites, with whom he would exchange sonnets and madrigals. This shift in his social circle has been interpreted as part of a wider effort on Michelangelo's part to raise his family's position, as he became more and more obsessed with status (Wallace 2010: 235).

The early 1540s had a significant effect also on Michelangelo's writing practices, which became less anchored to orality. For example, he abandoned the graphic rendering of the raddoppiamento fonosintattico, probably as a result of his contacts with two friends and literary 'advisors', Donato Giannotti and Luigi Del Riccio (Bardeschi Ciulich 1989: 14). Although a wealth of non-standard features are maintained even in his latest letters (D'Onghia 2014: 98), in my PhD thesis I showed that, after his relocation to Rome, at least some salient features, such as the masculine determiner, were subject to lifespan change in the direction of the archaizing, literary variety that was being promoted throughout Italy in those years, largely through the products of the printing press and, after 1516, also through the production of grammars (Serra 2020).2 Similar conclusions are reached by Valenti (2020), who shows the increase of several archaizing features in Michelangelo's language over time. Although a possible reading of Pietro Bembo's grammar *Le prose della volgar lingua*, first published in 1525, has been hypothesized (Valenti 2020), Michelangelo's re-orientation to standard features may simply be accounted for by his change in social networks and growing interest in literature. Such factors might also have triggered a change in his use of letter-writing conventions. This is an issue that the present chapter aims to explore.

In summary, a few points are worth underlining: first, after receiving little education in the early years of his life, Michelangelo's writing experience grew with time, also thanks to his prolific letter-writing. Second, the 1530s were years in which Michelangelo's social and linguistic practices significantly changed. Third, Michelangelo was an upwardly mobile individual, that is, somebody who could be termed a 'social aspirer'. These individuals are known to be particularly sensitive to prestige and stigma, and to show signs of linguistic insecurity (Nevalainen & Raumolin-Brunberg 2016: 133f.). Hints of the latter emerge in Michelangelo's correspondence and, whereas some of his anxieties specifically concern the drafting of formal business letters (e.g. n. 129),3 other times they express general feelings of inadequacy in addressing learned or powerful people (e.g. n. 273; 1225).

<sup>2</sup> For an account on language codification in Renaissance Italy see, for instance, Trabalza (1908), Patota (1993) and Vallance (2019).

<sup>3</sup> Here and elsewhere, I cite the Buonarroti letters by their identification number as it appears in the printed editions by Barocchi & Ristori (1965–83) and Barocchi et al. (1988–95).

## 3 Research hypotheses

As explained in the introduction, this chapter asks whether patterns of intra-writer variation in the language of Michelangelo may shed some light on the functions of epistolary formulae. Do these patterns favour a view of formulae as aids to compose a text, or do they favour an interpretation of formulae as social conventions, related to group practices?

I tackle this question by exploring two types of intra-writer variation: lifespan change (by charting Michelangelo's use of formulae across adjacent, subsequent time periods) and register variation (by charting Michelangelo's use of formulae on the basis of his relationship with the addressee).

Concerning lifespan change, an interpretation of formulae as a safe option for less experienced writers would lead us to expect a gradual decrease in Michelangelo's use of epistolary formulae, since his writing experience is known to have increased throughout his life. On the other hand, if formulae functioned as social conventions, signalling in-group membership, we would not necessarily expect a decrease in their use throughout his life. However, we might expect Michelangelo's use of formulae to change after the 1530s, in keeping with his relocation to Rome and change in social network.

Concerning register variation, if formulae served Michelangelo as a safe option to compose a text, they could be either expected to be used equally when writing to different addressees or, as was found by Große et al.'s (2016) study on correspondence by the semiliterate, the use of formulae could be expected to decrease with emotional proximity. An increase in the use of formulae with emotional proximity, instead, would not be compatible with this hypothesis, and would favour a view of formulae as conventions related to group practices.

## 4 The corpus used for analysis

In 2016, the Memofonte foundation kindly provided me with digital files of the Buonarroti corpus. After removing letters of non-autograph or uncertain status, I uploaded this corpus on the analysis tool Sketch Engine,4 tagging it with a range of metadata on the identities of each writer along with the name of the addressee.

In order to analyse Michelangelo's use of formulae, I have selected his autograph letters, excluding all unfinished letters, and letters that exhibited extensive damage. Because the formulaic frame was frequently added only on the final draft that was actually sent (e.g. cf. n. 1074 and n. 1075), I also excluded all of Michelangelo's own drafts and copies, unless they were ended by a conclusion, signature, or date.5 This yielded a total of 442 letters.

To allow for analysis of lifespan change, letters were then categorized on the basis of the time period in which they had been written. I selected four timespans of equal length (i.e. 17 years) from the date of Michelangelo's first autograph letter (1496) to the date of his last (1563).

Considering tenor – that is, the social relation between writer and addressee – as the primary factor behind register variation, as did Nevalainen and Raumolin-Brunberg (2016: 189f.) for the CEEC corpus, I classified letters on the basis of Michelangelo's relationship with the addressee, determined on the basis of secondary literature, and further confirmed by

4 Cf. <https://www.sketchengine.eu//> accessed 1 March 2022.

5 At times, a change in the use of discourse-ending formulae in different drafts of the same letter suggests that for Michelangelo, formulae were not interchangeable: a case in point is a very well-crafted letter addressed by Michelangelo in 1533 to his aristocratic friend (and possibly lover) Tommaso de' Cavalieri, to whom the artist also dedicated several sonnets. The first draft contains the only occurrence in this corpus of the formula *non altro che dirmi* [nothing else to say], which is changed to *non dirò altro* [I will not say anything else] in the second draft, and re-elaborated into a more complex and creative formulation, *per non vi tediare non scriverrò altro* [in order not to bore you I will not write anything else] in the third draft (see letters n. 893, 895, 896 respectively).

Michelangelo's use of forms of address, which tended to differ across the identified categories. I distinguished between:


The relevant divisions, and the number of letters that fall under each category, are shown in Table 3.1.


Table 3.1. Michelangelo's letters


## 5 Object of analysis: Discourse-ending formulae

Michelangelo's letters present formulae of different types, which seem to fit quite well the pragmatic categorization proposed by Rutten & van der Wal (2012), which distinguishes between text-type, text-structural, intersubjective and Christian-ritual formulae. I focus on a set of formulae that were used to end discourse, which can be categorized as a subtype of textstructural formulae, that is, formulae that 'mark the text structure by realizing the transition from one part of the discourse to another' (Rutten & Van der Wal 2014: 108) – in this case, they signal the end of the contextdependent information conveyed by the letter, realizing the transition from the body of the text to the closing formula, or (less frequently) from one topic to another. These formulae have not been systematically investigated in the Italian tradition to date, although they are identified as a specific type of epistolary formulae in Fabio Magro's analysis of the text typology of private letters (Magro 2014: 126, n. 50).8

These formulae are particularly frequent in Michelangelo's letters, and in the Buonarroti corpus in general. Unlike text-type formulae (i.e. formulae that mark the text as a letter, such as the address form, the signature, the date), discourse-ending formulae are optional. At the same time, if writers decide to use them, they have a range of options to choose from. This allows me to investigate two questions, that is, what factors influenced the choice to use or not to use a discourse-ending formula and, in case formulae were used, what factors influenced the choice of a particular type of discourse-ending formula.

The analysis yielded a total of 231 discourse-ending formulae in Michelangelo's letters. Allowing for some micro-variation in the order of

8 Magro lists *basta* [enough], *non ò da dirti altro intorno a questo* [I have nothing else to tell you about this] and *né altro per questa* [and nothing else through this] as examples of formulae used to signal closure of discourse in early modern letters (Magro 2014: 126). *Basta* is also discussed by D'Onghia (2014: 97) as a discourse marker in private letters that appears to be derived from speech.

constituents within a single type, the types of formulae used by Michelangelo may be identified as the following:9


It is possible that these formulae might have functioned as aids to reduce the writing effort since they could be retrieved as a whole from memory and could therefore have helped to speed up the writing process (see Rutten & van der Wal 2012). This pre-fabricated nature of formulae is suggested by their occasional occurrence in rapid succession, almost as if they were inserted absent-mindedly, perhaps triggered by adjacent formulaic expressions, as in the following by Michelangelo: *Altro non m'achade. Seguita e avisami. Altro non m'achade* [Nothing else happens. Follow up and answer me. Nothing else happens] (n. 1199). Moreover, some of these formulae are semantically non-transparent, a characteristic that is typical of holistic units (Wray 2002: 33f.): in particular, the formulae *non altro* [nothing else] or *altro non accade* [nothing else happens] are frequently reduced to *altro* and *altro accade* (which, literally, would translate into *something else* or *something else happens*!).

Most frequently, as mentioned earlier, discourse-ending formulae are used to signal the end of the context-dependent information conveyed in the letter and are followed by a closing formula, but they can occasionally be used to signal the end of one specific piece of information. However,

9 The instances of these formulae present considerable spelling variation as was typical of sixteenth-century private writings. When discussing a formula type, among the alternative spellings I use the one that conforms to the orthography of modern standard Italian (but that was in any case used by some writers in the corpus). Conversely, when quoting individual occurrences, I reproduce the original spelling.

just as was the case for the formulae studied by Rutten and van der Wal (2014: 112), it is not always the case that these formulae in fact close the discourse (e.g. n. 62, 67). This may be accounted for by a scant level of textual planning (Magro 2014: 133).

## 6 Analysis and discussion

My analysis consisted of two parts: first, I analysed Michelangelo's choice of whether to use a discourse-ending formula or not; second, I focused on the discourse-ending formulae he did use to examine the variation in his choice of different types.

## *6.1 Use of discourse-ending formulae*

As in Bijkerk (2004), in order to count formulae, I calculated the number of letters with at least one formula over the total of letters in each subgroup. This method was chosen over a normalized frequency count because it was only in very few cases (11 out of 442) that letters by Michelangelo contained more than one formula.10 The results of this analysis are shown in Table 3.2.

Within the subgroups that allowed for a quantitative exploration of lifespan change, that is, the group of letters sent to family members and the group of letters sent to high-status friends, the use of discourse-ending formulae remains relatively stable over time. Hence, in Michelangelo's usage, the quantity of formulae does not seem to decrease with an increase in writing experience, being highest in the latest letters sent to his family. Michelangelo's letters to patrons, on the other hand, may tell a different story: out of the thirteen letters, only the earliest (1), written at 21 years of

<sup>10</sup> Of these, nine contained two discourse-ending formulae (n. 2, 27, 62, 67, 130, 138, 981, 1196, 1199) and two (n. 20, 1136) contained three.


Table 3.2. Number of letters containing at least one discourse-ending formula/total number of letters in each category

age, contains a discourse-ending formula. Numbers are low in this group and discussion must necessarily remain speculative, but it is possible that in letters sent to patrons and superiors, Michelangelo's increase in writing experience progressively translated into an effort to be more creative and to distance himself from routine formulae. This is suggested by the existence of letters to superiors where the ending of the discourse is signalled by creative formulations: an example is Michelangelo's self-deprecating comment in a letter sent in 1547 to the humanist Benedetto Varchi, where the artist explains that being almost in the grave on account of his old age he does not have the time to write further (n. 1077).

On the other hand, the results provide evidence of register variation. The highest number of discourse-ending formulae appears in letters sent to family members: in this category, more than half of the letters contain at least one formula. In addition, the eleven letters that contain more than one formula were all sent to family members.

The number of formulae decreases in letters sent outside of the family circle. Concerning letters sent to low-status writers, quantitative considerations may be drawn only for the period 1513–29. A third of the letters contains discourse-ending formulae, compared to 60 % of letters sent to family members in the same time period. As for letters addressed to high-ranking friends, quantitative considerations may be made from the second time period onwards (i.e. from 1513). The percentage of formulae in these letters oscillates between 11 % and 29 % in the course of Michelangelo's life.

There is only one instance of discourse-ending formulae in letters to patrons, as previously mentioned, and no discourse-ending formulae are found in any of the eleven letters sent to distant business partners.11

The high frequency of formulae used in letters for family members, their moderate use in letters to low-status correspondents and close friends, and their virtual absence in letters to patrons and distant business partners that required a formalized correspondence suggest that these formulae were characteristic of an informal register.12 This finding does not support the idea that formulae, for this particular writer, served as a safe option for text composition, because we would otherwise expect the quantity of formulae to remain stable across letters to different addressees or to increase as the distance to the addressee increases.

The concept of enregisterment – recently adopted in the field of historical sociolinguistics to explore the process by which linguistic features


become historically associated with specific groups or practices (Beal 2019) – has already been used to shed light on the way epistolary formulae and other structures were attributed social meaning by letter writers (Pietsch 2015). In Michelangelo's language, discourse-ending formulae seem to have become linked with family practice, coming to constitute part of a register of informal letter-writing. Even assuming that these formulae had served the relatively unlearned, young Michelangelo as aids for formulation, they must have ceased to perform such function as his writing experience increased. The stable use of formulae within letters to his family even after the 1530s, and its higher rate compared to correspondence sent outside of the family, suggests that the originary link with writing experience, if present, had faded, with formulae assuming meaning as social conventions.13 This is further supported by the observation that the number of discourse-ending formulae is extremely high also in the letters Michelangelo received from his family members. For this writer, discourse-ending formulae must have become associated with a stable family practice that had its own standards and conventions.

In the following section, I shall explore the question of whether register variation and lifespan change may be detected in Michelangelo's use of different types of discourse-ending formulae, and the light this might shed on the function of these items.

#### *6.2 Choice of different types of discourse-ending formulae*

Since I aimed to investigate a possible social meaning behind Michelangelo's choice of particular discourse-ending formulae, I first analysed the distribution of these types in Michelangelo's broader social network. In previous research, I had reconstructed the backgrounds of the different writers in the Buonarroti corpus and, largely based on individuals' professions, I had adopted a bipartite social classification assigning

13 See also Rutten and van der Wal's (2014: 185–87) discussion of the possible role of social conventions for a Dutch writer, Kathelijne Haexwant, who despite her upper-class upbringing uses a great amount of formulae.

each writer to a 'higher' or a 'lower' rank (see Serra 2020). This distinction to an extent follows (although it does not completely coincide with) the distinction that existed in Florence between major and minor guilds. Individuals were classed in the higher rank if they were themselves, or were descended from a family of bankers, cloth merchants, notaries, lawyers, along with priests, secretaries, humanists, and politicians. The lower rank consisted instead of artisans, among whom artists were also included (unless, like Michelangelo, they came from an 'upper-class' background), stonemasons, assistants, along with a few other professional figures such as quarrymen and boatmen.

Table 3.3 and Figures 3.1 and 3.2 show the relative distribution of the different types of discourse-ending formulae in letters by other writers of the Buonarroti corpus, across the two social ranks and across two time periods of 34 years each (1496–1529; 1530–63). The years 1496 and 1563 were chosen so as to coincide with the first and last letter by Michelangelo.14 My analysis was restricted to the types of discourse-ending formulae that were also used by Michelangelo, but I further considered the type *altro (non) (mi) occorre* [(Nothing) else happens (to me)] because of its structural similarity and semantic equivalence with *altro (non) (mi) accade*, a formula frequently used by Michelangelo.

A few clear differences emerge in the use of formulae by the two ranks:



Table 3.3. Distribution among Michelangelo's correspondents of the discourse-ending formulae used by Michelangelo

Figure 3.1. Discourse-ending formulae across the two ranks from 1496 to 1529 (in per cent).

Figure 3.2. Discourse-ending formulae across the two ranks from 1530 to 1563 (in per cent).

The formula *e basta* also appears as more associated with the lower rank, since it is maintained by them for longer.


At this point, I was able to analyse Michelangelo's choices of discourseending formulae, situating them within his broader network. I analysed the distribution of the different types of formulae in his letters in relation to register and time. The results of this analysis are shown in Table 3.4.

A quantitative analysis of lifespan change in Michelangelo's use of different types of formulae is possible if we zoom in on the letters addressed to his family, as visualized in Figure 3.3.

In family letters, the formula *non altro* represents Michelangelo's favourite formula before 1530. Although further studies are required to assess whether this trend is true of the general Tuscan population, and not simply an idiosyncrasy of Michelangelo's correspondents, from the data shown above this formula appears to have been far more frequent in letters by lower-rank correspondents, and to have gradually decreased in usage in the course of the century. It seems significant, then, that after the 1530s Michelangelo virtually discontinues using it. This might be interpreted


Table 3.4. Distribution of types of discourse-ending formulae in Michelangelo's letters in relation to addressee and time

(*Continued*)


Table 3.4. Continued

Figure 3.3. Types of discourse-ending formulae over time in Michelangelo's letters to family members (in per cent).

as either a lifespan change reflecting change in the community, or as a reorientation towards linguistic forms that were less diastratically marked. In favour of the latter hypothesis, it should be noted that after the 1530s, Michelangelo also abandons *e basta*, suggesting that he was distancing himself from formulae that might have been considered popular and were associated with the lower rank.

On the other hand, the formula *altro non accade* – which, as previously observed, is used more or less equally, as a minority variant, by correspondents from the higher and the lower rank – sees a sharp increase in Michelangelo's family letters after the 1530s, becoming the majority variant. As Michelangelo's network changes, Michelangelo might have opted for what would have appeared to him as a more neutral and less popular formula. This choice might even have been driven by the influence of the structurally similar formula *altro non occorre*, strongly associated with the higher rank.

Comparing letters to the family with letters to other types of addressee, it is possible to detect register variation on a qualitative level. Michelangelo's correspondence contains only one instance of *non altro per questa*, a formula that, as we have seen, was far more frequent in letters by the higher rank. This only instance also represents the only formula used in letters addressed to patrons, that is, Michelangelo's early letter to Lorenzo di Pierfrancesco de' Medici written in July 1496 (n. 1). In this case, Michelangelo might have been consciously selecting a formula perceived as more prestigious and more suitable for the addressee's rank.

Furthermore, register variation may help understand Michelangelo's acquisition of the formula *altro non accade*, the usage of which, as we have seen, rises sharply after the 1530s. In his early letters, Michelangelo had produced a few instances of this formula, in its variant *non mi accade altro.* The first instances with the fronted constituent *altro*, which moves on to become the most frequent variant in his later correspondence, appear in letters addressed to individuals outside of his family circle. In particular, in the timespan 1513–29, when this formula is still a minority variant in letters to family members, the four formulae registered in letters for highranking friends are all instances of *altro non (mi) accade*. This suggests that this formula, subsequently employed as the majority variant even within the family, was at first adopted by Michelangelo as a more elegant alternative when corresponding with high-ranking friends.

In summary, while the use of discourse-ending formulae seems to be, for Michelangelo, typical of an informal register, and mostly of the family register, and the artist by no means dispenses with formulae as his writing experience increases, the types of formulae themselves are not stable, but change in the course of his life. Formulae that might have been perceived as too 'popular' or low-status were abandoned even in the most informal of registers in favour of a more neutral choice, originally adopted in letters exchanged outside of the family.

However, it should be noted that Michelangelo does not opt for formulae that exclusively characterize the highest echelons of society. The formula *non altro per questa* – mostly associated with the higher rank – occurs only once (cf. n. 1), and *altro non mi occorre*, so frequent in the letters of his high-ranking correspondents, is never used. Moreover, other high-ranking correspondents, in the latest time period, seem to develop a habit of inserting discourse-ending formulae within more complex syntactic structures, a practice that may have been influenced by the letter collections and anthologies that were progressively flooding the market. This is not the case for Michelangelo, who even in his latest letters keeps using these formulae as autonomous strings.

At any rate, Michelangelo's change in his use of formulae in the direction of less popular variants, along with his preference for less popular formulae in letters sent to high-ranking individuals, suggests that this writer used formulae, to an extent, to style his social identity. This is further supported by the observation that his usage changes most dramatically after the 1530s, coinciding with his relocation, change in social circle, and change in orthographic and linguistic practices. Returning to my earlier reflections on the way Michelangelo would have learnt to write letters, these findings also demonstrate that his early life was by no means the only time when the artist would have acquired his formulaic language.

## 7 Conclusions

This chapter analysed lifespan change and register variation in the use of discourse-ending formulae by focusing on the language of one single writer, whose usage was, however, situated in his broader social network. Future studies could explore whether similar patterns may be identified for other writers, or whether Michelangelo, as a social aspirer, was more sensitive to the social value of formulae than other individuals. Further, the analysis was based on a single set of text-structural formulae: it remains to be seen whether this pattern would be replicated for other formulae, and for formulae with different pragmatic functions, such as intersubjective formulae. It should also be pointed out that register categories were not equally represented, with a far higher number of family letters than letters of other types. The scarcity of letters outside of the family circle for the first time period, when Michelangelo was still an inexperienced writer, is particularly regrettable, and makes it difficult to establish whether his early use of a formula when addressing a patron was the result of chance, or provides, instead, evidence that when corresponding with high-ranking individuals, his formulaic usage decreased with growing levels of writing experience.

These results, however, suggest that, in Michelangelo's private and everyday correspondence, discourse-ending formulae were, to a large extent, conventions related to group practices. At an early stage, these formulae, as strings that could be retrieved as a whole from memory, might have served the relatively unlearned artist as aids to reduce the writing effort. However, the maintenance of these formulae in letters to his family throughout his life, and their high frequency compared to their absence in letters sent to patrons or business partners, suggests that the function of formulae as textcomposition aids was no longer prevalent as Michelangelo's writing experience grew. These formulae seem rather to have become enregistered within the practice of informal correspondence, especially within the family. At the same time, the analysis highlights a radical change in the types of discourseending formulae used after the 1530s, coinciding with a more general change in Michelangelo's social and linguistic practices: Michelangelo's abandonment of what were likely popular formulae in favour of less diastratically marked options further underlines the social meaning that formulae had, or had progressively acquired, for this writer.

## References


#### Juan M. Hernández-Campoy

4 Intra-writer variation and the real world of epistolary interaction in historical sociolinguistics: John Paston I's use of the orthographic variable (TH)

#### Abstract

The microscopic exploration of epistolary interaction and the observation of intra-writer variation is currently representing a shift in focus from the collectivity of the speech community to the individuality of specific users. Style constitutes an essential component for the social meaning of linguistic behaviour in interpersonal communication. The present study illustrates this microscopic approach through the observation of the orthographic variable (TH) in a core member of the Paston family ( John Paston I) both longitudinally throughout his lifespan and cross-sectionally, exploring his intra-writer variation when addressing recipients from different social ranks. When reconstructing the development of a change over time, how it is experienced by speakers individually becomes of paramount importance. Often, as our informant shows, style-shifting processes occur as the implementation of intra-speaker variation that feeds both the individual lifespan developments and the historical change in progress in the community.

## 1 Introduction: Historical sociolinguistics and intra-writer variation

As discussed in the introduction to this volume, until recently, intra-writer variation has not been given the same attention as inter-writer variation in historical sociolinguistics, especially when, given its ubiquity, style constitutes an essential component for the social meaning of speakers' sociolinguistic behaviour in interpersonal communication. Shared patterns of style-shifting are one of the defining characteristics of membership in a particular speech community (Rickford & Eckert 2001: 10). Speakers constitute the intersection between the speech community and the sociodemographic characterization of its diverse array of social groups. It is this intersection between the stylistic and the social dimensions that makes style an essential sociolinguistic concept.

In longitudinal and cross-sectional approaches to language variation and change within the speech community, speaker age and stylistic variation have become a key factor for the detection of change in progress, because they usually go together hand in hand (cf. Labov 1966, 1994; Wagner 2012; Bowie & Yaeger-Dror 2016; Beaman & Buchstaller 2021). The interrelationship between the individual and the community (i.e. collective vs individual instability) allows inter- and intra-writer variation to perform their respective roles in the development of a change. In fact, among the five related problems in the sociolinguistic model of language change (cf. Weinreich et al. 1968), together with the *actuation*, the advance in our understanding of how language change in the individual intersects with language change in the community is one of the central heuristic questions in Variationist Sociolinguistics: 'Subsequent research has not, however, directly addressed the question of the articulation between language change in the historical sense and language change as experienced by individual speakers' (Sankoff & Blondeau 2007: 560; cf. also Anthonissen 2021).

## 2 Objectives

The present chapter is a study using this microscopic approach on the use of an orthographic variable in Late Medieval English by a male member of the Paston family ( John Paston I) atomistically and cross-sectionally, exploring his intra-writer variation in the context of potential changes across the lifespan and complementing previous holistic-longitudinal studies.

Unlike traditional macroscopic historical sociolinguistic studies typically focusing on the speech community as a macrocosm, the treatment of language variation and change privileging microscopic approaches is complementarily enriching research through within-speaker designs both at theoretical and methodological levels. Consequently, in addition to approaching language variation and change holistically in a speech community as a macrocosm, atomistic observations within the community of practice as a microcosm may provide us with a wider and more accurate picture of speakers' sociolinguistic behaviours in earlier periods, as in the case of late medieval times. New studies are now focusing on the analysis and reconstruction of the sociolinguistic behaviour of individual speakers in social interaction microscopically (cf. Schiegg 2016; Voeste 2018; Hernández-Campoy & García-Vidal 2018a, 2018b; Hernández-Campoy 2021; Werth et al. 2021).

To this purpose, the development and diversification of digitalized archival data sources is allowing scholars to deal with ego-documents for sociolinguistic analysis (cf. Nevalainen & Tanskanen 2007; van der Wal & Rutten 2013; Auer et al. 2015; Evans 2020). In addition to tracing language variation and change throughout a speech community at a macro-level, private letters from historical corpora may also shed light onto the resources and driving forces for sociolinguistic variability and stylistic choice by individuals in past societies at a micro-level. The speakers' sociolinguistic behaviours observed in private correspondence allow the detection and reconstruction of community values from the past as reflected in the communicative conduct developed for language choice and use in style-shifting processes and for the transmission of linguistic as well as social meaning in communicative interaction (cf. Romaine 1998; Koch & Oesterreicher 1994; Biber & Finegan 1989; Eckert & McConnel-Ginet 1992; Palander-Collin et al. 2009).

## 3 Methodology

#### *3.1. Variable*

The adoption of the Latin alphabet introduced by Christians since the seventh century at the expense of the local runic practices used by Anglo-Saxons was part of the process of Christianization. Additional graphemes were added to represent the sounds of Old English for which no Latin grapheme was readily available (cf. Scragg 1974: 6–11; Hogg 1992: 73–81; Upward & Davidson 2011: 3), such as the introduction of symbols from the futhorc runic alphabet ('thorn' <Þ/þ> and 'wynn' <Ƿ/ ƿ>) under Germanic influence or variants introduced by Irish missionaries ('eth' <Đ/ð>, 'ash' <Æ/æ> or 'ethel' <Œ/œ> and even the variant 'yogh' <Ȝ/ȝ> for Latin <Y/y, J/j, G/g>). It was a rapid process for most graphemes – maintaining a one-to-one correspondence between letter and sound – except for a few letters that did not have an equivalent in the Latin alphabet and thus prevailed until the end of the Middle Ages, such as 'thorn' <Þ/þ> (> <th>) and 'wynn' <Ƿ/ƿ> (> <uu, w*>*). In this last case the replacement of the local runic practices by the Roman-based ones was a slow, gradual and often intermittent process in medieval times that reached into the early fifteenth century (cf. Scragg 1974: 10; Benskin 1977: 506f., 1982: 18f.; Stenroos 2004, 2006; Bergs 2005, 2007; Conde-Silvestre & Hernández-Campoy 2013).

The variable used for our present study is one of these instable graphemes: (TH) with the realizations <th> and <þ>, as in *þing* > *thing, broþer* > *brother,* or *comeþ* > *cometh*. Originally, the runic spelling 'thorn' <þ> and the Irish 'eth' <ð> were used for both the voiceless and voiced interdental fricative consonants [θ] and [ð] in Old English – despite some graphotactic conditioning in early texts, with <þ> used initially and <ð> medially (Stenroos 2006: 12).1 Later, in the Middle English period, both letters began to be replaced by the digraph <th>, which was a continental influence taken from the Latin alphabet. The grapheme <ð> was the first to disappear, remaining only sporadically in the fourteenth century (Lass 1992: 36). However, <þ> remained longer because 'confusion with similarly

1 The variant <y> is also a modified version of <þ> mainly used in abbreviations like *ye* and *yt* instead of *þe* and *þat* in the period mostly in northern areas (Lass 1992: 36). The origin of <y> lies in the confusion with <þ> by some scribes. According to Stenroos (2006: 20f.), the merger shows a wider distribution in the north with the form <y> (and sometimes <þ>) used initially for the voiced consonat [ð] and the Latin digraph <th> for voiceless [θ].

shaped graphs was less likely' in the southern dialects (Scragg 1974: 10; cf. also Hogg 1992: 76).

$$\text{ACE /<\S>} \to \text{ME /<\text{th}>} \to \text{EModE <\text{th}>}

$$

Phonotactically, according to Benskin (1982: 18 and 26), the digraph <th> began to affect [θ]-segments first (as in *think* or *myth*), medial [ð] segments later (as in *brother*), and then initial [ð]-segments, as in *they* or *than*. Bergs's (2007) analysis of <th>, <þ> and <ð> in *The Peterborough Chronicle* (c. 1154) detected an incipient situation of orderly heterogeneity where <þ> and <ð> were preferred for function words, while <th> was used for content words. Both Latin and Biblical influence point to the presence of external prestigious norms which certainly triggered the *actuation* of this spelling change. According to Benskin (1982: 18f.), the Roman digraph <th> occasionally appeared in early Old English writings, particularly in the spelling of vernacular names in Latin texts. It was reintroduced again in the twelfth century through Latin influence on Anglo-Norman scribes (see also Benskin 1977: 18f.), but it did not become popular until the mid-fourteenth century (Blake 1992: 10; Bergs 2013: 255); and it was not until the very late fifteenth century that the runic 'thorn' <þ> disappeared. These external written norms not only triggered this spelling change, but also contributed to its standardization, although often as a kind of misspelling:

In the aftermath of the Norman Conquest, English as a written language was displaced for nearly all administrative purposes by Latin and, to a lesser extent, by French. Immigrant Norman scribes, given the frequent task of spelling in their Latin and French texts English vernacular names having [θ], did merely what any number of English scribes from *before* the Conquest would have done: they wrote <th>. For some of these continental writers, <th> may perhaps already have been a familiar combination: those who had read or written much Biblical texts would certainly have been aware of it. Be it as it may, in adopting the Anglo-Latin <th> as the written correspondent of a sound which was not part of their own language, the imaginations of the Norman scribes were hardly being taxed: the separate elements of the combination, <t> and <h>, were thoroughly familiar symbols. (Benskin 1982: 19)

Thus, the diffusion of <th> can be considered as a part of a process of historical change operating above the level of social awareness, in connection with social and stylistic factors and very slowly diffusing along the social space in the literate, careful and conscious styles, acquiring overt prestige and becoming part of the accepted linguistic norm: a typical Labovian 'change from above'. In fact, the analysis of the text types favouring this variant confirms this hypothesis (see e.g. Stenroos 2004, 2006; Bergs 2007; Conde-Silvestre & Hernández-Campoy 2013; Hernández-Campoy & García-Vidal 2018a). Stenroos (2004, 2006) studied the behaviour of variable (TH) in correlation with extra-linguistic factors (region, chronology, and text type) in Middle English sources from the *Middle English Grammar Corpus*<sup>2</sup> (MEG-C), focusing on both documentary (legal and administrative writings, treatises, letters, etc.) and non-documentary (literary) texts. Her results show the gradual spread of <th> during the fifteenth century mostly in documentary texts without exhibiting any geographical anchoring (Stenroos 2004: 275). The preference for this spelling in official documents of the period was thus confirmed, with <th> being virtually universal, whereas, by contrast, literary texts still used to retain a majority of <þ> forms (see Stenroos 2004: 277f.; 2006: 20f.). This prac tice is also supported by Jensen's (2012) analysis of written materials from Yorkshire belonging to documentary and religious types. In this case, the spread of the innovating form <th> is chronologically clearer in the documentary than in religious texts, which also points to its prevalence in the most careful and self-conscious styles.

In our longitudinal study on the replacement of the old runic spelling <þ> with the Latin-based <th> in letters written in late medieval times by male members of the Paston family, Conde-Silvestre and Hernández-Campoy (2013) found a gradual but steady adoption of the new <th> form throughout the fifteenth century that spread through the literate ranks, operating above the level of social awareness and, as seen in Hernández-Campoy and García-Vidal (2018a), in connection with social and stylistic factors as a *marker*. The incoming <th> form expanded longitudinally, following an incremental monotonic pattern, first proceeding gradually, then cumulatively and consecutively, from generation to generation, as

<sup>2</sup> Cf. <https://www.uis.no/en/middle-english-grammar-corpus-meg-c-0> accessed 9 April 2022.

reflected from the first letter available in 1425 to the last one in 1503. The informants' sociolinguistic behaviour exhibited stability in the individual generationally but change within the community as the development across the lifespan for each member shows, as we will see below with the particular case of John Paston I.

The development from the categorical use of an old form *A* to the categorical use of a new form *B* goes through a set of intermediate stages of covariation between both variants at different levels until the propensity for the innovation becomes so high that covariation disappears and categoricity predominates. In fact, the situation of variability is present in our informants' verbal production not only between different documents but also within the same text, reflecting the intermediate period of instability, with the use of both forms:


Rather than paying attention to the results of the change under observation but to the processes undergone, it is these intermediate stages governed by both inter- and intra-writer variation that are explored in the present study of a male core member of the Paston family – given that linguistic change usually interacts with patterns of stylistic variation.

#### *3.2. Corpus and informant*

The Corpus used as linguistic material is the *Paston Letters*, a collection of 422 authored documents (246,353 words) written from 1425 to 1503 by four generations of this minor gentry Norfolk family, currently available online from the *Corpus of Middle English Prose and Verse*. 3 The informant observed was a male member of this family, about whom we have personal biographical information (cf. Davis 1971). John Paston I (1421–66) was the eldest son of William and Agnes Berry and was educated at Trinity Hall and Peterhouse in Cambridge and the Inner Temple in London. As a lawyer, he became Justice of the Peace for Norfolk (1447, 1456f. and 1460–66), Knight of the Shire (1455), and MP for Norfolk (1460–62). John I married Margaret Mautby and inherited her family estates and wealth. His multiplex social networks are reflected in the amount and social array of addressees found in his private correspondence (twentythree letters, written between 1440 and 1469), not only because he was clearly involved in a process of upward social mobility, but also because he was a core member of the family, as reconstructed by Bergs (2005: 69f.) on the evidence of the letters written and received (see Figures 4.1 and 4.2).

In order to carry out the microscopic sociolinguistic study all documents available were organized into five-year cohorts, and compared with groupings based on the recipients' social rank. The power relation (from John Paston I's perspective) to letter recipients were classified into three groups: John Paston I's social superiors (royalty and nobility), equals (his wife, relatives and minor gentry), and inferiors (legal professionals). For each rank, the number of tokens, percentage and total number of words is shown in Table 4.1.

<sup>3</sup> Cf. <http://quod.lib.umich.edu/c/cme/paston> accessed 25 March 2022.

Figure 4.1. The Paston network(s): internal links (Bergs 2005: 69).

Figure 4.2. The Paston network(s): external links. Adapted from Bergs (2005: 70).



newgenrtpdf

## 4 Evidence from John Paston I

#### *4.1. Sociolinguistic behaviour across the lifespan*

Figure 4.3 shows pool results of the use of the new variant <th> by John Paston I along the six time-periods of letters preserved in *The Paston Letters* collection: from 1444 in P1 (date of his earliest letter) to 1465 in P6 (date of his latest letter), spanning from when he was 23 up to 44 years old. If we focus specifically on his sociolinguistic behaviour as for (TH) across his lifespan, as Figure 4.3 allows, macroscopically the pattern of development is irregular. John I's overall rate of innovative <th> use4 of 80.1 % is aggregated on the following basis: 86.6 % in P1 (1440–44), 100 % in P2 (1445–49), 70.6 % in P3 (1450–54), 83.9 % in P4 (1455–59), 79.1 % in P5 (1460–64), and 79.9 % in P6 (1465–69). A chi-square test confirms that the differences in use of the innovating <th> between the six time-cohorts is highly significant (*p* ≤ 0.001, *χ*<sup>2</sup> = 67.086; df = 5). In light of these results, does his lifetime behaviour illustrate the process of change across the lifespan, age-grading, or generational change?

If John I's language practices in communicative interaction considering letter recipients are microscopically observed, the ups and downs in the graph showing how he was experiencing this change, can be easily described. As Table 4.1 and the bar-plot in Figure 4.3 show, during the different periods, John I's dialogic exchanges vary in terms of addressivity, recipiency and relationality, reflecting the multiplexity of social networks and personal communicative style of writers, which depend on their social relationship with their addressees, and the situation and purpose of the letter (context and topic).

There are, therefore, social asymmetries between writer and recipients as well as context types when addressing superiors (↑), equals (↔)

4 Assuming it was still an embryonic Standard English variety, or proto-standard, that was developed during the fifteenth century, and later fixed and codified between the sixteenth and eighteenth centuries (cf. Fisher 1996; Wright 2000).

Figure 4.3. John Paston I's sociolinguistic behaviour across the lifespan according to variant <th> and recipients.

and inferiors (↓) which occur due to accommodation, rather than changes across the lifespan, as shown in Table 4.2:

A chi-square test confirms that the differences in use of the innovating <th> between group recipients within each of the six time-cohorts is highly significant in 1450–54 (*p* ≤ 0.01; *χ*<sup>2</sup> = 35.878; df = 2) and 1460–64 (*p* ≤ 0.01; *χ*2 = 36.94; df = 1), though not in 1455–59 (*p* ≥ 0.05; *χ*<sup>2</sup> = 2.9685; df = 1). So rather than age-grading, changes across the lifespan or generational change, what we just have is reactive style-shifting, as the implementation of intra-writer variation. Sometimes, the technicality of professional text types might also condition the audience-based linguistic production (cf. Biber et al. 1998: 33; Palander-Collin et al. 2009: 12; Evans 2020).

#### *4.2. Audience design, referee design and register variation*

As we know, if inter-writer variation alludes to social differences reflected in the speech of groups of speakers, complementarily intra-writer variation refers to stylistic differences reflected in the speech of an individual speaker: 'the range of variation [for particular sociolinguistic variables] produced by individual speakers within their own speech' (Bell


Table 4.2. Symmetries and asymmetries between John I and his letter recipients

2007: 90). In this axiom, intra-speaker variation is largely a function of inter-speaker variation, in such a way that some individuals exhibit a much wider range of stylistic variation than others. Therefore, the different social class groups have different levels of usage of sociolinguistic variables drawing a perfect symmetry, where the most formal style of the lowest social group is similar to the most informal style of the highest social class. This means that there is a point along the symmetrical axis where, as Labov illustrated, it would be difficult to distinguish 'a casual salesman from a careful pipefitter' (1972: 240).

As found in Hernández-Campoy and García-Vidal (2018a), the obser vation of John I's behaviour according to addressees, text type and social networks shows an audience-design-based pattern of linguistic variation, organized through upward, symmetrical and downward accommodation attunements: despite his overall 80.1 % (2007/2,507 instances), he uses the innovating variant <th> 100 % when addressing royalty, 96.6 % with nobility, 82.4 % with his wife, 74.2 % with other minor gentry addressees, and 73.0 % with legal professionals (see Figure 4.4). There is a positive monotonic relationship for style-audienceship depending on the rank of letter recipients: the higher the social rank of the recipients, the higher the presence of the new form <th> and vice versa. Inferential statistics through Pearson's Chi-square test of significance confirms that the different sociolinguistic practices in John I's results when addressing different social-ranked recipients did not occur randomly: the relationship is significant at *p* < 0.01 (*χ*<sup>2</sup> = 116.98; df = 4). Individual comparisons between groups also suggest the existence of significant variation between them at *p* < 0.01, except between legal professionals (398/545 instances: 73 %) and minor gentry (594/801 instances: 74 %), where *p >* 0.05.

Twentieth-century tenets based on sociolinguistic competence and style-shifting through accommodative processes are thus found in these results of the late medieval times under the influence of the same external constraints and also in the spirit of the *Uniformitarian Principle*. John I's

Figure 4.4. John Paston I's use of innovative <th> for variable (TH) according to social status of his letter recipients (n = 2,507). Adapted from Hernández-Campoy and García-Vidal (2018a: 398; Figure 4).

sociolinguistic behaviour in social interaction demonstrates that the principles of *graded style-shifting*, *range of variability*, *socio-stylistic differentiation*, *sociolinguistic stratification*, or *stylistic variation* (cf. Hernández-Campoy 2016) characterizing Labov's (1966) Attention-To-Speech Model also governed intra-writer variation in English medieval society: no single speaker is mono-stylistic, though some will have and exhibit a wider verbal repertoire than others depending on external factors, such as social stratification. This quantitative evidence also justifies the attribution of stylistic variation to the effect of the addressee, as claimed by Bell (1984) in his *Style Axiom* developed for the Audience Design Theory: 'speakers have a fine-grained ability to design their style for a range of different addressees, as well as for other audience members' (Bell 2001: 146), with intra-writer variation being a function of inter-writer variation.

Yet both *style* and *register* have been indistinctly used to refer to diaphasic variation, implying particular ways of language use in particular socio-situational contexts. The term 'register' is used by Hallidayan linguistics to denote all types of socio-situational varieties (including technical/occupational ones), assuming 'styles' as an aesthetic option with no functional value in the communicative process (cf. Halliday 1975, 1978). Contrarily, the term 'style' is used by Labovian sociolinguistics to refer to socio-situational varieties, restricting 'registers' to technical or occupational ones (cf. Labov 1966, 1972; Trudgill 1974). All speakers, therefore, use dif ferent styles in different situations depending on the context of situation (topic, addressee and the medium), but not all speakers have the same proficiency in registers or verbal repertoire. Together with audience-based style-shifting processes through the use of the orthographic innovation <th>, there is also register-based variation when John I is addressing legal professionals (73 %), who are inferior in rank. This is probably due to the technicality of the official language of law, business and administration, as well as, more crucially, due to the distant nature of those trained interlocutors, as found by Stenroos (2004) in her exploration of the use of <th> and <ϸ> in legal documents and literary (prose) texts. These individual examples without adherence to inter-writer variation show this kind of register-based variation (Table 4.3):


Table 4.3. Symmetries and asymmetries between John I and his registered-based letter recipients

Unexpectedly, however, there is no symmetrical accommodation between John I and his wife (Margaret Paston), despite being equals in rank. On the contrary, he divergently uses more <th> forms to her (82 %) than to other addressees of the same social rank (74 %). Considering that his wife used <th> in only 68 % of (TH) contexts when addressing him, an account for the absence of reciprocity with her is not straightforward. Relevantly, given Margaret's illiteracy, usual in medieval times, her correspondence used to be written and read aloud by her scribes – usually family clerks and chaplains. Consequently, the presence of these auditors – as known and ratified third-person present interlocutors though not directly addressed – might have caused him some *cross-over* accommodation, tending towards hypercorrection. This could be understood as an example of Bell's outgroup referee design, where a speaker from a group A addresses a member of their own group A as if both were members of a group B (clergy), usually associated with prestige. Admittedly, as Bergs (2005) points out, this is an exercise of sociohistorical reconstruction where the non-existence of evidence often does not allow for conclusions about the non-existence of individual facts, and there might be other factors involved in this behaviour (politeness, power relationships, etc.).

## 5 Conclusion

Inter- and intra-writer variation perform their respective crucial roles in the development of a change through the intersection between the individual and the community, through the collective vs individual instability. As Säily et al. (2017: 3) suggest, individuals exhibit short-term accommodation in their everyday social interaction as well as long-term accommodation over their lifespans. For this reason, the study of intrawriter variation is also necessary in the reconstruction of language use in remote periods. What we can see in our informant John Paston I's lifetime behaviour is reactive style-shifting processes as the implementation of intra-speaker variation contributing to his individual development across his lifespan of the historical change in progress in the community. His insignificant monotonic increase with age is part of a generational change that exhibits apparent stability in the speaker but change in the community. As seen in this case study, descending into the real world of epistolary interaction and immersing into intra-writer variation is currently representing a shift from the collectivity of the speech community to the individuality of specific speakers. Given that linguistic variation and change inevitably interact with patterns of stylistic variation in complex ways, style constitutes an essential component for the social meaning of speakers' sociolinguistic behaviour in interpersonal communication.

## References


#### Tamara García-Vidal

5 Patterns of stylistic variation in the use of synthetic and analytic comparative adjectives: Evidence from private letters in sixteenth- to eighteenthcentury England

#### Abstract

This chapter aims at exploring socially based patterns of stylistic variation at the individual level through the study of synthetic and analytic mechanisms for the construction of comparative adjectives in English historical correspondence from the sixteenth to eighteenth centuries. To account for speakers' sociolinguistic behaviour in interpersonal communication, I investigate intra-speaker variation in letters written by five informants from diverse social ranks when addressing recipients ascribed to different social orders. By focusing primarily on number of syllables and etymology of the adjectives, this study shows that the distribution of comparative adjectives follows addressee-based accommodative patterns when writing upwards, by showing a preference for the analytic form with long and Romance adjectives, and for the synthetic form with short and Germanic/native adjectives when writing downwards.

## 1 Introduction

Patterns of stylistic variation are a sociolinguistic phenomenon denoting social change and differentiation which interacts in complex ways with linguistic variation and change in any given language (Ure 1982: 7). Hence, for the understanding of social linguistic variation in interpersonal communication, the concept of *style* has long been considered as a crucial factor in the field of sociolinguistic studies (Eckert & Rickford 2001; Coupland 2007; Hernández-Campoy 2016), allowing for the distinc tion between *inter*-speaker (social) and *intra*-speaker (stylistic) variation (Halliday 1978; Bell 1984). Within this framework, style-shifting, as a manifestation of intra-speaker variation, is considered as a key component of individuals' sociolinguistic behaviour in their communicative practices whose indexical nature has been explored by Third-Wave Sociolinguistics (Eckert 2018). While traditional approaches to stylistic variation focused on macro-sociological categories (context and topic within a speech community) for the explanation of style shifting, recent approaches consider stylistic choices to be conditioned by the construction of speakers' identities in the communicative interaction with other speakers in different social environments. Rooted in studies of speech accommodation (Giles 1980), recent socio-constructionist-based theories consider intra-speaker variation as a valuable sociolinguistic asset for the study of speakers' stylistic choices and the conveyance of social meaning, focusing on the proactive and reactive potential of style-shifting in discourse and on the individuality of speakers in reaction to the characteristics of a present or absent audience (Bell 1984: 182). In this sense, Bell conceives sociolinguistic variation as audience design since '[s]tyle derives its meaning from the association of linguistic features with particular social groups' (Bell 2001: 142). Thus, style-shifting emerges from the variability among different social groups where intra-speaker variation is seen as a function of inter-speaker variation.

Recently, historical sociolinguistic studies have applied theoretical assumptions and findings of current sociolinguistic models of intra-speaker variation to search for socially based variation patterns and stylistic variation in individuals' epistolary interaction (Palander-Collin et al. 2009; Auer 2015; Hernández-Campoy & García-Vidal 2018a, 2018b; Schiegg 2018; Hernández-Campoy et al. 2019; Havinga 2021). For the study of intra-speaker variation in remote speech communities, historical correspondence is a material worth examining for the detection of the personal communicative style of an informant in social interaction with their recipients, since it best represents the oral language (Romaine 1998: 18).

The current study investigates intra-speaker variation through the analysis of the behaviour of synthetic (*stronger*) and analytic comparatives (*more important*) in the letters of five informats from different social backgrounds in connection to the social status of their recipients from sixteenth to eighteenth-century England. The following section deals with the methodology used for intra-speaker analysis, presenting relevant information about the corpora and research design. Section 3 presents the analysis of the data, and conclusions are illustrated in Section 4.

## 2 Methodology

## *2.1 Corpora: Informants and audience*

The private letters used in this study come from both the Parsed Corpus of Early English Correspondence (PCEEC) (1410–1681) and the Corpus of Early English Correspondence Extension (CEECE) (1700–1800).1 Both corpora are socially stratified, containing personal letters written in England by informants from different social ranks. Such epistolary documents are central to the study of the sociolinguistic behaviour of individuals and communities present in style-shifting processes. With the aim of exploring stylistic variation as audience design in historical correspondence, this study focuses on five informants from the sixteenth to the eighteenth centuries ascribed to different social ranks, who were selected due to their extensive production and varied audience (see Table 5.1).


Table 5.1. Informants and social rank

1 I would like to thank Terttu Nevalainen for inviting me to have access to the CEECE database at the Research Unit for Variation, Contact and Change in English at the University of Helsinki.

To account for patterns of intra-speaker variation, the sociolinguistic behaviour of the informants is measured on the basis of the influence that the recipients, as well as the informants' attitude to recipients, have on style-shifting, as it is commonly found in accommodation processes of interpersonal social interaction. Following the conventional classification of historical social ranks (Nevalainen & Raumolin-Brunberg 2003: 36), the following typology based on the interpersonal relationship between the five informants and the social status of their addressees has been considered for the classification of audience: (i) nobility; (ii) clergy; (iii) gentry; (iv) professionals; (v) non-gentry; (vi) relatives; and (vii) friends.2 Consequently, the nature of the relationship that both informants and addressees could have maintained could be closer (kinship/friendship/social peers) or more distant (different social class) (Conde-Silvestre 2016).

#### *2.2 The linguistic variants and methodological considerations*

Previous research on the phenomenon of English adjective comparison has identified a range of factors conditioning comparative alternation between the synthetic (-*er*) and the analytic (*more*) variant.3 The majority of these studies have mainly focused on phonological, morpho-syntactic, semantic or pragmatic factors as the main precursors for comparative alternation in present-day English, such as syllable-length, stress patterns, presence of prepositional/infinitival complements, syntactic position or frequency of usage (Quirk et al. 1985; Leech & Culpeper 1997; Hilpert 2008; Mondorf 2009, amongst others).

In the diachronic front, the phenomenon of English adjective comparison has received a relatively sparse treatment. While some scholars outline that the analytic form emerged during the thirteenth century under


French/Latin influence as a consequence of language contact (Pound 1901: 3; Kytö 1996; Mustanoja 2016[1960]: 280), González-Díaz (2006) provides evidence that the analytic form was already present sporadically in Old English texts, emerging from the collocation of degree intensifiers (such as *ma* 'more') with participles.

Following Rohdenburg's *Complexity Principle* (1996), Mondorf developed the concept of *more-*support, whereby the analytic comparative favours contexts of cognitive complexity over the synthetic one in different domains, such as phonology, lexicon, morphology, syntax, pragmatics and semantics: '[s]imply by choosing the analytic variant as a signal, a language user can alert the addressee to the fact that a cognitively complex adjective phrase follows, so that some extra processing capacity can be allotted to that phrase' (2009: 7). Mondorf (2009: 135–64) also applies the concept of *more*-support to the analysis of historical data by focusing on the following degrees of complexity: (i) lexical complexity (mainly the factor syllable-length, measured by the number of syllables of an adjective: longer adjectives are more complex than shorter ones, e.g. *more important*) (see also Hilpert 2008: 412) and compound adjectives; (ii) morphological complexity (adjectives consisting of a base + a derivate suffix are considered as more complex, e.g. *more careful*); and (iii) syntactic complexity (the presence of infinitival/prepositional complements after the adjective indicates a more complex context, e.g. *more ready to go*).

Accounts of the sociolectal distribution of both comparative variants have also been found. Kytö (1996) shows that synthetic forms prevailed in matter-of-fact text types (handbooks), which reflect spoken or colloquial registers in Late Middle and Early Modern English, whereas the analytic form appears more recurrently in rhetorical texts (philosophical/religious treatises). Additionally, other studies on Middle and Modern English have revealed that the analytic pattern is preferred with adjectives of Romance origin (*more intelligent*), whereas the synthetic form is preferred with adjectives of native origin (*higher*) (Kytö & Romaine 1997: 346; González-Díaz 2008: 61–71; García-Vidal 2020a). Finally, García-Vidal (2020b) observes that different social groups differ in their use of the two comparative variants in Early and Late Modern English. More specifically, members belonging to upper orders present a more frequent use of the analytic variant in their letters from the late fifteenth century, which spread to lower orders from the seventeenth century onwards.

This chapter takes a step beyond linguistic factors and the speech community as a whole as it pursues the study of the distribution of comparative forms from an intra-speaker approach to assess the impact of audience and context types when using comparative strategies during Early and Late Modern English. For this purpose, synthetic and analytic forms are measured on the basis of the social rank of both informants and recipients and the possible relationship between addresser and addressee. In studying the socio-stylistic distribution of the comparative variants when addressing recipients from different social ranks, attention has been primarily paid to those linguistic factors that have already been accounted for in previous studies dealing with historical data: i) presence of complex complements that may condition the use of the analytic variant, including prepositional/ infinitival complements, bimorphemic adjectives and compounds;4 (ii) etymological origin5 of the adjective (Germanic/native and Romance adjectives);6 and (iii) number of syllables.

The data were collected in a three-staged process. First, the complete lists of occurrences of synthetic and analytic comparatives were retrieved by using the free concordance program *AntConc 3.5.9* (Anthony 2011). For the automatic retrieval of both synthetic and analytic instances, the following search elements were keyed in *AntConc*: \**ra*, *\*re*, *\*er(e)*, *\*er* '*more*', *mar(e)\**,


*mor\**, *moore*, *moare*, *moch*. 7 Then, the resulting occurrences were examined to exclude from the analysis: (i) quantifier uses of *more* (*many more great men*); (ii) defective comparative adjectives (*better*); and (iii) proportional clauses (*the more intelligent, the more nice*), since they may create parallelism with the adjective that follows. Subsequently, the final occurrences were classified according to the aforementioned linguistic factors to later crosscheck the ultimately selected occurrences with the social rank of each informant and recipient along with their type of interpersonal relationship.

With the aim of comparing the sociolinguistic behaviour of each informant when addressing their audience ascribed to different social ranks, raw frequencies were normalized to a common base of 10,000 words as the material obtained from the different writers is not balanced (see Table 5.2). Although raw data are rather limited, the chapter draws on quantitative and qualitative analyses to analyse intra-speaker variation in the informants' letters.


Table 5.2. Word count according to social-ranked addressees

7 For a more reliable search of instances, the text version of PCEEC has been favoured over the tagged one.

## 3 Analysis

#### *3.1 Synthetic and analytic comparison*

This section presents results obtained for both synthetic and analytic comparatives in connection with the audience of the informants. Results are divided according to social rank of informants (high vs low). Beginning with writers from higher social ranks, Figure 5.1 shows that Cecil (*d*value = 0.20)8 uses higher frequencies of synthetic forms when addressing gentry (7.08 per 10,000 words), but indicates a preference for the analytic form when interacting with nobility (5.55). The sociolinguistic behaviour of Carter (*d*-value = 0.23) also shows variability in the use of both comparative variants in correlation with the rank of recipients: the use of the analytic form amounts to 17.25 frequencies when addressing clergy, decreasing to 3.04 instances with professionals. Contrarily, the higher rates in the use of the synthetic variant are found in letters addressing gentry (7.14). Despite showing a narrower scope of social interaction with just professionals and relatives, Garrick (*d*-value = 0.25) also presents differing frequencies of comparative forms: the highest ratio of synthetic comparatives appears in letters sent to relatives (6.94), whereas the use of the analytic form is highly preferred in letters addressing professionals (13.17).

Figure 5.2 plots the distribution of comparative forms in informants with a lower social rank. As observed, Cowper (*d*-value = 0.24) shows a wider social array of addressees and exhibits different linguistic attunements in the use of comparative forms. Thus, an overwhelming preference for the synthetic form (17.23) is exhibited when addressing his equals

8 Inferential statistics through the Kolmogorov-Smirnov Test indicates that the data obtained in the informants' letters are normally distributed, exhibiting no significance skewness (Chakravarti et al. 1967). The meaning of *d*-value refers to the measurement of the divergence of the sample distribution from the normal distribution. Thus, the higher the *d*-value, the less likely it is that the data are normally distributed.

Figure 5.1. Distribution of comparative forms in informants with a lower social rank according to audience (normalized [axis] and absolute [bars] frequencies).

(professionals). Contrarily, analytic comparatives are more widespread in letters addressed to clergy (11.71), who receive lower scores of the synthetic form (7.19). This is also true in letters addressed to gentry inasmuch as they show higher occurrences of the analytic variant (11.1) than the synthetic one (8.33). Focusing now on Chamberlain's stylistic practices (*d*-value = 0.24), the results show nearly equal scores for both comparative variants when addressing different addressee-groups. Hence, Chamberlain appears to be the only informant who does not present a socio-stylistic use of both comparative forms in correlation to his audience. In his letters, the use of the analytic form seems to be slightly higher in letters addressed to gentry (5.16), followed by non-gentry (4.48) and friends (4.06). Closer examination of the latter group has revealed that all the letters were addressed to Dudley Carleton (1573–1632), ascribed to nobility, to whom he used to give advice on political matters. Since they were involved in political affairs, it would be safe to say that the letters may echo the formal type of relationship that they may have maintained.

The data obtained above reveal socio-stylistic attunements in the use of both comparative forms when addressing informants of different social ranks, which may indicate that style design is governed by the addressee's social position and/or the nature of the relationship between addresser

Figure 5.2. Distribution of comparative forms in lower social-ranked informants according to audience.

and addressee. In other words, the majority of the informants (except for Chamberlain) show intra-speaker variation in their sociolinguistic practices with their socially diversified audience, suggesting that their practices are based on addressee design. The use of the analytic variant is normally more frequent in letters addressed to higher status addressees at the expense of the synthetic one, which is more frequently found in letters addressed to relatives, social peers or recipients who are socially situated in lower positions. Despite generational and social differences, the informants show a similar patterning of audience-based language production by exhibiting a strong monotonic relationship between their styles and social rank of addressees.

Drawing on Giles' *Communication Accommodation Theory* (1980), stylistic variation is conceived as the effect that the norms normally associated with different addressees have on the speakers' attunements: individuals tend to accommodate to their recipients by making adjustments to their communicative behaviour according to the situations or roles assigned to them. In tune with this claim, these informants engage in style-shifting in response to their audience as a way of reaction to their audiences' roles. This entails that, when addressing different groups, the informants shifted to be more similar to the sociolinguistic characteristics of the recipient groups, which govern style design through accommodative style shifts. Therefore, they tend to converge with the prototypical recipient's social rank through *upward* and *downward* accommodation patterns (Hernández-Campoy 2016: 102) based on social class differentiation: the higher rates of analytic forms are mostly observed in letters addressed to higher social ranks (upward accommodation) whereas synthetic forms tend to appear more frequently in letters addressed to relatives, social peers or recipients situated below on the social scale (downward accommodation).

The observation of the differences between the highest and lowest scores, and average of the use of synthetic and analytic forms allows us to see how much range of variation the informants exhibit in the use of both comparative variants (see Figure 5.3). As observed, Carter and Garrick are the ones who show a greater range of variation with the analytic form, as opposed to Cowper, which is slightly wider with synthetic forms. Contrarily, Cecil and Chamberlain show little variation for both comparative forms. These differences reveal that some individuals exhibit a much wider range of intra-speaker variation than others, marking social positioning, which may be conditioned by the nature of the relationship with the audience and/or the recipients' social profile.

These results suggest that *The Principle of Graded Style-Shifting* along with *The Principle of Range of Variability*, operating in Labov's (1966) *Attention to Speech Model*, are also reflected in the sociolinguistic practices

Figure 5.3. Internal variation in the use of synthetic (S) and analytic (A) forms.

of the informants: their diaphasic versatility shown in the use of both comparative forms connected with the recipient's social profile reveals that these linguistic features may have been involved in stylistic variation, denoting social variation. In view of this, and following Bell's Style Axiom (1984), the style-shifting practices of the informants in the use of both comparative forms may reproduce the socio-stylistic variation existing at the community level.

## *3.2 Factors conditioning the socio-stylistic distribution of comparative forms in historical data: Audience and formality*

Section 3.1 indicated that the informants show variation in the proportions of synthetic and analytic forms when addressing their recipients of different social ranks. This section zooms in on linguistic factors (which have been historically associated with synthetic and analytic comparative forms) in connection with audience to show evidence for the writers' socio-stylistic choices as regards the above distribution of comparative forms. For this, I focus on (i) complex complements identified in historical data (see Section 2.2); (ii) etymology; and (iii) number of syllables of the adjective.

Figure 5.4 shows the distribution of both comparative variants in the presence of complex complements. For the sake of visual clarity, the different complex complements analysed are considered here jointly. As illustrated, the analytic variant predominates in the material. Notably, Carter (7.53), Garrick (4.95) and Cowper (8.61) exhibit an overwhelming preference for the analytic form in these contexts, mostly with derived forms (such as *charming*, *worthy*, *dreadful*, *favourable*, etc.) and in the presence of infinitival/prepositional complements. In line with previous literature, the emergent functional specialization of both comparative variants in the presence of complements and argument complexity developed after the eighteenth century (Mondorf 2009: 161–67). This finding is also reflected in this data, since the enhanced use of the analytic variant with complex complements is attested in those letters from these eighteenth-century informants. Despite showing a preference for analytic forms in the presence

Figure 5.4. Distribution of comparative forms in the presence of complex complements.

of a complex complement over synthetic ones, I did not find a well-marked addressee-based difference with this factor to merit breaking down these figures according to audience. Hence, for the purpose of this study, I focus on etymological origin and number of syllables of the adjectives to examine whether these factors had an influence on the selection of the comparative strategy to show stylistic variation.

## 3.2.1 Etymological origin of the adjective

This section offers an account of the influence that the etymological origin of the adjective could have exerted on the comparison strategy choice in connection with the social-ranked audience of the informants. Figures 5.5 and 5.6 display the results for both comparative forms with Germanic/native and Romance adjectives in informants of higher social ranks according to their audience. As illustrated, the default variant with Germanic/native adjectives is the synthetic comparative whereas the analytic form predominates with Romance adjectives. The sociolinguistic behaviour of these informants also reveals addressee-based attunements in their use of both comparative forms: the correlation of the synthetic variant with Germanic/native adjectives seems to prevail in Cecil's and Carter's letters addressing gentry (6.29 and 5.71, respectively), and in Garrick's when interacting with his relatives (6.94). A different tendency is observed with Romance adjectives, for they appear to be more frequently found when interacting with recipients of higher social ranks. In this respect, a strong preference for Romance adjectives with analytic forms is present in Carter's letters addressing clergy (15.1) and in Garrick's when interacting with professionals (11.53). In the case of Carter's letters, a larger share of French adjectives correlated with analytic forms is found, mostly when addressing clergy and gentry. This includes adjectives such as *riant* [cheerful], *verdant* [leafy, flourishing], *fatigued* [exhausted], etc., the latter being a past participle used in the presence of a prepositional complement, as illustrated in (1).

(1) […] for we wrangled & laugh'd about it most part of the way to Oxford where I am arrived not more fatigued with my Journey than you […] (CEECE, from Elizabeth Carter (lower clergy) to Hannah Underdown (gentry), 1739)

Although exhibiting lower numbers of occurrences, Cecil's preference for analytic forms with Romance adjectives is a bit higher when addressing nobility (4.16), as in (2):

(2) And so wishing yowe more hable than I am sure yowe are to take this Journeie […] (PCEEC, from William Cecil (nobility) to Robert Dudley (nobility), 1586)

Contrarily, synthetic forms with Romance adjectives and analytic forms with Germanic/native adjectives are quite scarce in these high-ranked social informants' writings.

Figures 5.7 and 5.8 show the distribution of both comparative forms with Germanic/native and Romance adjectives in informants of lower social ranks according to their audience. The results suggest that it is only with Cowper that relationship appears to play a role in the choice of the Germanic/native vs Romance adjective and comparative strategy: the highest rates of Germanic/native adjectives taking a synthetic comparative are more frequently found when addressing his equals (17.23), whereas Romance adjectives with analytic forms are used when interacting with

Figure 5.5. Distribution of comparative forms with Germanic/native adjectives in informants of higher social ranks according to audience.

Figure 5.6. Distribution of comparative forms with Romance adjectives in informants of higher social ranks according to audience.

recipients of higher social ranks, such as clergy (10.13) or gentry (8.33). However, Chamberlain shows a stronger tendency towards the use of analytic forms with Germanic/native adjectives than the rest of informants, as when addressing gentry (2.58). Among these types of adjectives, I find some ending in -*y* (such as *angry*, *weary* and *worthy*), considered as morphologically complex (Mondorf 2009: 140), as well as monosyllabic adjectives (such as *fit*, *coole*, *milde*, *glad*, *ripe*, *slacke*, etc.). See (3) and (4):


Interestingly, these comparative adjectives predominate in letters addressing Carleton, his friend, and gentry. What is more, they do not seem to appear in the presence of the aforementioned complex complements. More microscopically, Chamberlain's use of comparative forms shows vacillation for the same type of adjective (*fit*), as illustrated in (5) and (6):


This is particularly noteworthy in that the synthetic variant is the one that appears in the presence of a complex complement. Thus, Chamberlain shows a less stable correlation of Romance adjectives with analytic forms and Germanic/native adjectives with synthetic forms as the rest of informants. Although tentatively, it could be argued that these differences in his sociolinguistic practices may be explained by the fact that he was an informant from an earlier period (sixteenth century), ascribed to a lower social status, so it may be possible that these stylistic practices had not reached this social layer yet.

The data presented here shed some light on the results obtained in Section 3.1: the informants show a tendency to use more Romance adjectives correlated with analytic comparatives when writing to social superiors, and of Germanic/native adjectives with synthetic forms when writing to social inferiors. This might reflect the larger share of Romance vocabulary being frequently found in more formal and stylistically elevated writing,

Figure 5.7. Distribution of comparative forms with Germanic/native adjectives in informants of lower social ranks according to audience.

Figure 5.8. Distribution of comparative forms with Romance adjectives in informants of lower social ranks according to audience.

so the tendency for Romance adjectives to be compared analytically might explain the correlation of analytic comparatives when writing upwards.

Bybee (1985: 29) found a positive correlation between frequency and synthetic forms, assuming that frequently used adjectives are more prone to be compared synthetically. Therefore, and following Mondorf 's concept of *more*-support (2009), this fact suggests that infrequent adjectives may act as a complexity parameter which may trigger the use of analytic comparatives. That is, the less entrenched an adjective is, the more likely it is to take the analytic variant. Taking this into consideration, the analytic comparative may have previously been associated with more complex adjectives (preferably those of a Romance origin) which were initially compared analytically and used more frequently by high prestige groups, who, in turn, used more Romance loanwords. In this way, the stylistic variation existing within the written practices of the informants is revealed whenever they are observed in different contexts. By assuming that 'speakers associate classes of topics or settings with classes of persons' (Bell 1984: 181), the role of the addressee may foster stylistic variation in the epistolary interaction of the informants, who shift when writing on specific *topics* or *settings* as if they were writing to recipients with whom they associate the topic or setting.9 Hence, by applying these contemporary tenets on style-shifting to historical epistolary interaction, and also in accordance with the *Uniformitarian Principle*, it is observable that the way these informants write reflects an underlying association of topics/settings with certain audience members: the higher the frequency of Romance adjectives compared analytically, the higher the social rank of the recipient, and the higher the frequency of Germanic/native adjectives, the lower the social rank of the recipient. Consequently, what seems to emerge from the data is simultaneous accommodation to the rank of the addressee and the formality or informality of the letters, established in accordance with the relationship between addresser and addressee. Thus, the informants style-shift along this formal-informal continuum by self-monitoring their writing in response to their different audience groups. The high rates of synthetic forms with Germanic/native adjectives addressed to informants of lower social ranks may confirm that it is these types of settings, presenting less formal language, that may foster an overuse of the synthetic variant at the

9 Biber & Finegan's *Register Model* (1994), however, views setting and topic as more influential than audience: style is basically context-dependent and social class differentiation is just an echo of the different registers that are most commonly used in one's professional and personal life.

expense of the analytic one. Contrarily, the analytic form seems to be led by informants of higher social status since it is more salient and indicates a more formal choice of vocabulary, justifying the attribution of analytic forms to formal contexts.

## 3.2.2 Number of syllables of the adjective

This section presents comparative forms correlated with the number of syllables of the adjective according to audience. The results show that synthetic comparison is the most frequent strategy with monosyllabic adjectives (see Figure 5.9) in letters addressed to: (i) gentry: as in the case of Cecil (7.08), Carter (5), and Chamberlain (5.16); (ii) equals: in Cowper's letters when addressing professionals (14.77); and (iii) relatives: just in the case of Garrick (5.95). The majority of these monosyllabic adjectives compared synthetically are of a native origin (namely high-frequency adjectives, such as *greater*, *nearer*, *stronger* and *higher*), and few tokens are found with a Romance origin (*safer*, *larger* or *plainer*). Although being less frequent, monosyllabic adjectives compared analytically are preferred in letters addressed to higher social ranks. These are normally Romance adjectives (*calm*, *cleare*, *apt*, *grave*, *plaine*, etc.), as shown in Cecil's letters addressing nobility:

(7) And then hir majesty began to be more calm than befor. (PCEEC, from William Cecil (nobility) to Robert Dudley (nobility), 1586)

Surprisingly enough, Chamberlain uses the analytic variant more frequently with monosyllabic Germanic/native adjectives (such as *coole*, *fit*, *glad*, *quiet*, *ripe*, *slacke*, *straight*) than the rest of informants, as observed in example (8) when addressing gentry:

(8) […] the French shold not permit yt, nor the Spaniard neither to see him any way greater, unles he were of a more quiet humor. (PCEEC, from John Chamberlain (non-gentry) to Ralph Winwood (gentry), 1611)

A different picture emerges with disyllabic adjectives (Figure 5.10). Even though there is more variability in the use of synthetic and analytic comparatives, as previously considered by Kytö and Romaine (1997), disyllabic forms clearly favour the use of analytic comparison. Carter is the informant who

Nobility Clergy Gentry Prof. Non-Gentry Relatives Friends

Figure 5.9. Distribution of synthetic (S) and analytic (A) forms with monosyllabic adjectives according to audience.

shows higher frequencies of analytic forms when addressing clergy (5.39). Cecil's analytic forms are more widespread when addressing gentry (3.14), Garrick's when interacting with professionals (1.64) and Chamberlain's with his friend (2.65). However, Cowper shows a preference for the synthetic form when using disyllabic forms, for example, when addressing clergy (2.53). See (9):

(9) […] at least we have given proof of a Wisdom which abler Politicians than myself, would do well to imitate. (CEECE, from William Cowper (professional) to William Bull (clergy), 1783)

Contrarily, Cowper shows a preference for analytic forms with relatives (3.77), the majority correlating with Romance adjectives except for the adjective *worthy*, which appears in the presence of a complex complement. See (10):

(10) I wish his Company were more worthy of him. (CEECE, from William Cowper (professional) to Lady Hesketh (relative), 1767)

Finally, all instances of polysyllabic adjectives (e.g. *more convenient*, *more remarkeable*, *more eligible*) categorically take analytic comparison in this

Nobility Clergy Gentry Prof. Non-Gentry Relatives Friends

Figure 5.10. Distribution of synthetic (S) and analytic (A) forms with disyllabic adjectives according to audience.

dataset, with no variation at all. Despite not showing variation between synthetic and analytic forms, and as regards intra-speaker variation, Figure 5.11 shows that polysyllabic adjectives were far less frequent in Chamberlain's letters than in the rest of the informants, who exhibit a preference for this form with higher social-ranked addressees.

Nobility Clergy Gentry Prof. Non-Gentry Relatives Friends

Figure 5.11. Distribution of analytic forms with polysyllabic adjectives according to audience.

D'Arcy (2014: 228) illustrates how speakers of higher socio-economic status use significantly more trisyllabic adjectives than speakers of lower status in contemporary data. This implies that there is more analytic comparison in higher status speakers than lower status ones. This fact may have also been true for the results obtained here. Following this line of reasoning, social forces may have been operative on the choice of comparative strategies since speakers from higher social status tend to use more complex forms (probably polysyllabic Romance adjectives), leading to more analytic comparative forms, in letters addressed to higher social rank recipients, while less complex forms (short Germanic/native adjectives) will foster synthetic comparison.

## 4 Conclusion

The present study has investigated intra-speaker variation in the use of both synthetic and analytic comparatives in the letters written by five authors of different social ranks addressed to recipients from diverse social backgrounds during Early and Late Modern English. Particularly, the use of Bell's model of Audience Design (1984) applied to this historical data has allowed us to show addressee-based accommodative patterns of style-shifting in the communicative interaction of the informants, which reflect language choice for the transmission of social meaning in epistolary communication. Focusing on the distribution of the comparative adjectives, the sociolinguistic analysis of both linguistic variants has revealed that the factors etymological origin and number of syllables of the adjective have a great impact on the choice of comparative strategy. In this way, the informants design their styles to accommodate to their addressees by producing different levels of formality, normally associating short Germanic/native adjectives with synthetic forms when writing to social inferiors and long Romance adjectives with analytic forms when writing to social superiors. This fact may tentatively suggest that the correlation of the analytic comparative with formal texts (as attested in previous literature) and writing upwards may just be an artefact of the more frequent use in upwards writing of long and Romance adjectives. Moreover, a higher use of the analytic form is more frequently found in the linguistic production of Carter, ascribed to the clergy, suggesting that analytic forms with long Romance adjectives were led by higher social groups. In this way, it is possible to deduce that the particular social stratification of the informant(s) as well as the nature of the relationship to the addressee would determine their choice of the type of adjective employed along with the comparative strategy.

## References

#### *Primary sources*


#### *Secondary sources*


Schiegg (eds), *Intra-Individual Variation in Language*. Berlin: De Gruyter. 315–46.


# 6 Patterns of linguistic variation in Late Modern English pauper petitions from Berkshire and Dorset1

#### Abstract

The article focuses on patterns of linguistic variation during the Late Modern English period, that is, when English was codified in grammars, spelling books, dictionaries and pronunciation guides. It investigates inter- and intra-individual variation in two sets of pauper petitions written by Frances Soundy (Berkshire) and Charls Ann Green (Dorset) between 1818 and 1830. The letters are viewed in the contemporary socio-historical context where literacy was socially stratified. Patterns of variation in the petitions are examined regarding (1) H-dropping and H-insertion and (2) long s. While the first set of features was commented on in normative works at the time, long s was gradually disappearing across different text types. Our findings allow us to draw conclusions regarding social reasons for language variation and change, including the received writing training of the two petitioners.

## 1 Introduction

The study of patterns of linguistic variation, both on a community and on an individual level, allows us to shed light on social reasons for language variation and change, as well as attitudes to language variants (Labov 1972; Eckert 2008). Variationist concepts and paradigms, such as

1 This article was written in the context of the SNSF-funded research project *The Language of the Labouring Poor in Late Modern England* (2020–24; 100015\_ 188879). Many thanks to Judith Huber, Samantha Litty, Julian Mader, and Markus Schiegg for their valuable feedback on an earlier version of this chapter. All remaining shortcomings lie solely with us.

variation patterns, have also been applied in linguistic fields like historical sociolinguistics. In line with other historical sociolinguistic studies (cf. Auer 2015; Hernández-Campoy 2016; Schiegg 2018; Werth et al. 2021), this article tries to shed light on patterns of linguistic variation, notably inter- and intra-individual variation, in a specific historical text type with the aim to better understand whether factors that have been identified as explaining linguistic variation in synchronic studies are also relevant in the historical context, that is, *the uniformitarian principle* (see Labov 1972: 275; Bergs 2012).

The current study is thus concerned with the text type of pauper petitions during the Late Modern English period, which is strongly associated with the codification and prescription of English. More precisely, we focus on two sets of pauper petitions written by the women Frances Soundy (20 letters from 1818 to 1830, Berkshire) and Charls Ann Green (8 letters from 1820 to 1826, Dorset), which allow us to investigate their language use and patterns of variation over a period of time. While synchronic variationist studies, which are still largely based on oral data, take the perspective that the individual speaker has learnt and therefore reproduces the community language, the historical sociolinguistic approach relies on written data and therefore needs to take into consideration the acquisition of writing skills within the community. In fact, as compulsory elementary schooling in England was only introduced in 1880, the variation patterns displayed in the pauper petitions will reflect the level of education of the individual writers and provide insight into language learning processes. Considering that the petitions were written at a time when written English was largely codified, a comparison to linguistic norms allows us to shed light on the effect of normative rules on the language use of the labouring poor, which is linked to education opportunities. At the same time, the pauper petitions may contain reflections of actual speech.

Within this context, the linguistic variables under investigation are (1) H-dropping and H-insertion, which became stigmatized in normative works in the eighteenth century (Sheridan 1762: 34; cf. Milroy 1981), and (2) long s, which disappeared from print around 1800 (Nash 2001: 3), and gradually also in the language of letters of some educated contemporaries (for Lindley Murray's use, see Fens-de Zeeuw & Straaijer 2012). While the study of H-dropping and H-insertion in the pauper petitions is connected to phonetic variation, the long s case study is purely orthographic. Due to the link between H-dropping and H-insertion, the two variables will be discussed in one case study. The findings of the linguistic case studies, which are associated with different degrees of standardization pressures, allow us to make claims about social reasons for language variation and change in early nineteenth-century England, including the received writing training of the applicants.

This chapter is structured as follows: Section 2 provides background information about the pauper petition data, the context in which they were produced, and the education opportunities of the letter writers. Section 3 presents the data under investigation. Section 4 focuses on the linguistic case studies of (1) H-dropping and H-insertion and (2) long s. In this section, for each linguistic variable, relevant previous studies are presented, followed by the details regarding the method used, and then the findings. A general discussion and reflections regarding variation patterns in the investigated pauper petition samples are presented in Section 5.

## 2 The Old Poor Law and education opportunities for the labouring poor

The data under investigation in this study are retrieved from a corpus of c. 2,000 petitions of the labouring poor that has been created as part of the SNSF-funded research project *The Language of the Labouring Poor in Late Modern England* (2020–24), with the aim to better understand the role of social stratification in language variation and change in the period c. 1795–1834. The corpus, which is based on a collection of pauper petitions prepared by Tony Fairman from archives all over England, contains petitions written by labourers, artisans, people who had lost their farms and others (see for instance Auer & Fairman 2013). Background information about the text type of pauper petitions, as well as education opportunities for the labouring poor, are presented below.

#### *2.1 The Old Poor Law*

The Poor Relief Act of 1601, also known as the Elizabethan Poor Law or the Old Poor Law, aimed at creating a poor relief distribution system in England and Wales. The system's administrative unit was the parish, which means that individual parishes and appointed overseers of the poor were responsible for poor law legislation. During the period 1795–1834, the laws for poor relief legalized the payment and receipt of out-relief from parish funds, which was usually offered in the form of sending money or by removing the paupers from their current domicile and bringing them back to their parish of legal settlement. Citizens were entitled to apply for assistance from their parish of legal settlement if they had moved to a different parish and were 'in distress'. If the officials accepted the applicants' claims, money would be sent, or, less frequently, the applicant would be removed from the current domicile to the parish of legal settlement where they would then typically be placed in the workhouse (cf. Whyte 2004: 280; Auer & Fairman 2013). With often only limited schooling, the labouring poor had to write letters – petitions – to their parish of legal settlement in order to apply for out-relief. The existence of these petitions is thus a unique opportunity for historical sociolinguists to gain insights into the language use of the labouring poor.

#### *2.2 Education opportunities for the labouring poor*

For much of the nineteenth century, the task of providing elementary education for the population in England was left in private hands or the hands of the church (Stephens 1998: 5). It was only with the Education Act of 1870 that the English government committed to establishing staterun board schools, and it would take another ten years for schooling to become compulsory, yet only up to the age of 10 (Stephens 1998: 78). This lack of government involvement contributed to the great variety of school types we find in nineteenth-century England, for example, charity schools, national schools, part-time schools like Sunday schools, evening schools or factory, mine and work schools, private schools including dame and infant schools (Stephen 1998: 1–11; Chapter 7 by Gardner in this volume). Subjects typically taught at elementary level were the 3Rs (reading, writing and arithmetic), needlework for girls, as well as religion in church-run schools. Grammar, geography and other subjects were normally only introduced at a more advanced level (Stephens 1998: 2). In charity and industrial schools, pupils were also trained in a particular craft, for instance weaving or card-setting, preparing them for working life (Stephens 1998: 2; Iremonger 1813: 234–40).

Literacy rates in the nineteenth century, more specifically signature literacy as evidenced in marriage registers, varied across time, space and class. On average, by 1840 about 67 % of men and 50 % of women could read and write (Vincent 1989: 53). However, among unskilled labourers and miners, the literate proportion was as low as 27 and 21 %, respectively (Vincent 1989: 97). Female literacy was not as a rule lower than that of men, but was strongly affected by the occupational structure of their family and the area they lived in. Particularly in rural counties, among them Berkshire and Dorset, where male children and adults were predominantly employed as agricultural labourers, girls were often able to attend school for longer than boys (Stephens 1998: 37).

Many labouring poor were semi-literate. It was customary in many school types to teach reading before writing, and most children left school by the age of 8, consequently missing out on learning to write (Stephens 1998: 2). A notable exception, teaching both reading and writing at the same time, were the National Schools and British or Foreign Schools associated with two educational charities, which shaped the educational landscape in England roughly from the 1810s onwards (Vincent 1989: 75; Parker et al. 2020: 546f.; Chapter 7 by Gardner in this volume). However, even if children learned how to write, this did not imply that they would know how to compose a letter. This skill was not introduced into the curriculum until 1871 (Vincent 1989: 89).

Leaving school and taking up work did not always mean an end to education. Individuals continued their learning by themselves, for instance in mutual improvement groups, evening schools, regimental schools, even prison schools (Lawson & Silver 1973: 189–95, 238–50; Stephens 1998: 5; Crone 2018: 182).

## 3 The pauper petitions by Frances Soundy and Charls Ann Green

The current study focuses on two sets of pauper petitions written by the women Frances Soundy and Charls Ann Green. In the case of Frances Soundy, we base our analysis on twenty letters that were written during the period 1818–30 and that were sent to the parish of Pangbourne in Berkshire. The contents of the letters allowed us to find out that the writer of the letters Frances (F) Soundy was married to James ( J) Soundy and had at least three children. The oldest son Charles (C) was married to a woman called Mary (M) with whom he had two children. James Soundy and the children, like Charles, could often not work for reasons of illness or accidents, but were trying to find employment whenever possible. The family found itself in dire straits and required money for food, rent and clothing.

We also consider eight letters written by Charls Ann Green written during the period 1820–26, which were sent to the parish of Wimborne in Dorset. Charls Ann Green was married and had three children. As her husband was very ill and later hurt his hand, he was not able to work. She wrote in a letter that she has 'got a Litle shue binding but my housband Cante get eaney worke a torll' (1 February 1821). The family therefore required money for rent and food. Table 6.1 provides an overview of the twenty-eight letters under investigation.


Table 6.1. Overview of the Soundy and Green data2

2 See the references section for details about the archival records.


Table 6.1. Continued

Pauper petitions were not always written by the petitioners themselves due to limited literacy skills, and we therefore have to distinguish between the petitioner and the encoder/writer of the letter (for details regarding authenticity checks, see Gardner forthcoming). In the case of Frances Soundy and Charls Ann Green, we have however been able to determine that both sets are autograph letters, which in the case of Soundy also included petitions written on behalf of other family members. For instance, in a petition she clearly states that her husband is the petitioner and she signs with both their names: 'my husband surtinly is the Porpersist but I ham sorry to = say=[^INSERTION^] he will not undertake The jurney with out i could rase him somsurport' (24 March 1830). Another aspect that makes these datasets very valuable has to do with the fact that we have several letters written by the same petitioners (cf. the so-called *bad data problem*, see for instance Auer et al. 2015), which allows us to investigate variation patterns including intra-writer variation over a longer period of time.

To illustrate the language use of Late Modern English pauper petitions and how they reflect the education opportunities of the labouring poor, we present a philologically accurate transcription of one of the Soundy letters below (reused with permission of Berkshire Record Office; Pangbourne, D/P 91/18/4/1).

To the Honerable gentellman the church wardins and Oversears of the Parrish of Pangborn Gentellmen I humbly bag pardon for writing the Second letter gentillmen i humbly intreet you Would take our disſ trees in to considerration and asſ ist Us with a small trifell to inabell us to Surport our famely gentellmen we have not seen Our son sins he left is home a month last thersday We do not wish is wife and child to put you to {Exs}pence but unles that we are asſ isted a little we {\*\*}n no{t} keep her I ham very sorry to say my darter Is still in this world of sorrow but exspeting every Day that the lord will take her to his self

I have not gentellmen menchioned the perticklers in the Letter as i mencioned it in my letter last weeke I humbly hope gentellman that you will take it Into consederation and asſ ist a littell and your Perrishoner In duty bound will ever pray No 9 Sleeford Street F and J Soundy Battesea feelds Jinary 11 1829 To the Churchwardins or Oversearſ s of the Parrish of Pangborn Burks

## 4 Linguistic case studies

In this section, we discuss variation patterns related to the linguistic variables (1) H-dropping and H-insertion and (2) long s.

#### *4.1 H-dropping and H-insertion*

#### 4.1.1 Previous studies

H-dropping, that is, the non-realization of /h/ in initial position in stressed syllables before vowels, is considered a non-standard linguistic feature of British English that has become associated with 'uneducated' speech over time. These perceptions are for instance reflected in a metalinguistic comment by Alford (1864: 5, as quoted in Crowley 2003: 129):

First and foremost let me notice that worst of all faults, the leaving out of the aspirate where it ought to be, and putting it in where it ought not to be. This is a vulgarism not confined to this or that province of England, nor especially prevalent in one county or another, but common throughout England to persons of low breeding and inferior education, particularly to those among the inhabitants of towns.

This and other comments from the second half of the nineteenth century, like Sweet (1890: 195) noting that [h] is 'an almost infallible test of education and refinement', allow us to conclude that the shibboleth is in the first instance associated with social order and 'inferior education' rather than region.

As for the historical development of H-dropping, Milroy (1981) and Lass (1992: 62) argue that it started in the eleventh century, that it was common throughout Middle English and associated with Anglo-Norman scribes and their imperfect command of English (see also Mugglestone 1995: 111). Concerning its use in the Late Modern English period, Wells (1982: 255) claims that 'H Dropping has been known in popular London speech since at least the eighteenth century'. The stigmatization of Hdropping started in the second half of the eighteenth century, as the following quote by Sheridan (1762: 34) illustrates: 'There is one defect which more generally prevails in the counties than any other, and indeed is gaining ground among the politer part of the world, I mean the omission of the aspirate in many words by some, and in most by others'. As for H-dropping in the counties, Markus (2010: 182) notes that Wright (1905: 254) generally observed H-deletion in English dialects with the exception of Scotland, Ireland and some northern English counties. Based on *The Survey of English Dialects*, which contains data from the mid-twentieth century, Upton and Widdowson (1996: 46f.) found pockets of H-retention in Somerset, Dorset, Hampshire, Wiltshire, and East Anglia, in addition to the northern English counties of Northumberland, Durham and Cumberland.

H-insertion, that is, the insertion of <h> in initial position where it is historically not justified, is a hypercorrection that is 'introduced in order to avoid the greater stigma of h-dropping' (Beal 2004: 160). Like H-dropping, H-insertion was stigmatized in Late Modern England, as reflected in the works of Walker (1791) as well as Spence (1814), who con sidered it 'a vulgarism that could be corrected by teaching the poor to spell' (Beal 2004: 159). Markus (2010: 183) notes that H-insertion must have been less common at the time in comparison to H-dropping as Wright does not mention it in his dialect grammar, 'except for cases "when the dialect speaker wishes to express a strong emphasis"' (1905: 254). Both H-dropping and H-insertion were used for socio-stylistic purposes by contemporary authors like Charles Dickens (1812–70) (see Mugglestone 1995: 116).

Considering the clear stigmatization of H-dropping and H-insertion in Late Modern normative works and its association with the lower classes, it will be interesting to find out (a) to what extent the linguistic features are present in the pauper petitions under investigation and (b) whether we can determine certain usage patterns.

4.1.2 Method and results

As regards the method applied for H-dropping, first, we determined the variable context of words starting with an <h>, and then the words where the initial <h> was in fact dropped. In a next step, we considered internal factors that may explain patterns of variation such as word type, the number of syllables of the word where the <h> was dropped, and the preceding word (incl. final sound) (see for instance Mees 1990). Even though H-insertion differs from H-dropping in that it aims at avoiding the latter and introduces hypercorrection or is used for emphatic purposes, the same internal factors were considered to explore the use of the linguistic variable in the pauper petition sets. We present the H-dropping and H-insertion findings by writer, that is, Frances Soundy (Berkshire) and Charls Ann Green (Dorset), and then compare the results.

The Soundy H-dropping results reveal that within a variable context of 563 words, the H was dropped in 67 cases, that is, 11.9 %. H-dropping only occurs in one-syllable words (standard orthography) and is particularly associated with the verb HAVE. The Soundy data contain 201 forms of HAVE (including crossed-out and inserted forms)*,* notably 7 infinitive constructions with *have* having auxiliary (4 tokens) and lexical functions (3 tokens), 6 present participle examples, and 188 finite verbs. Within this variable context, H-dropping is found in 62 of the 201 HAVE tokens, thus c. 30.8 %. A closer look reveals that the form *have* appears 133 times (infinitive constructions and finite verbs; 88 auxiliary forms and 45 lexical forms) in the Soundy letters, but the H is never dropped in this verb form (i.e. neither in a lexical nor auxiliary function). The data contain 1 example of H-dropping in the present participle (*aving*). As regards the other forms of HAVE, the H in *had* (16 auxiliary and 17 lexical uses) is dropped in all cases (33 tokens of which 7 × 1sg, 1 × 2sg, 16 × 3sg, 8 × 1pl, 1 × 3pl) while *has* is dropped in 96.6 % of the cases (28 of 29 tokens of which 1 × 1sg, 27 × 3sg, 1 × 3pl). The findings suggest that the form *have* as a whole was restored by the writer on the basis of the written standard.

Table 6.2 provides an overview of all examples of H-dropping in the Soundy letters.


Table 6.2. H-dropping in the Soundy letters

Other examples of H-dropping are found in the possessive determiner *his*, thus *is* (3 tokens versus 53 tokens of 'standard' *his*), the adjective *whole*, as in *ole* (1 token), and the adverb *how*, spelled as *ough* (1 token). When considering the preceding words of *ad* (*had*), it is noteworthy that thirteen different word types precede *ad*, with the most frequent being *we* (8 tokens), *he* (7 tokens), and then *have* (4 tokens) and [*h*]*as* (4 tokens). In contrast, *as* (*has*) is mostly preceded by *he* (10 tokens) and *husband* (5 tokens). Considering all of the cases of H-dropping in the Soundy letters (67 in total), *he* precedes in many of these examples, followed by *husband* and *we*.

As regards H-insertion, which may be interpreted as hypercorrection, the Soundy letters contain sixty-nine instances in total. Of these, 26 (37.7 %) are found with the verb *am*, thus *ham*, and 24 (34.8 %) with the verb *is*, thus *his*. In fact, Soundy consistently inserted an H before *am* in all letters, while we can observe intra-writer variation with regard to the verb *is*, that is, 52 instances of *is* as opposed to 24 instances of *his*. Other H-insertion examples include *hone* (*own*, adj., 6 tokens), *hone* (*on*, prep., 2 tokens), *he*(*a*)*rn*(*t*) (*earn*(*ed*), 5 tokens of which 3 in the present and 2 in the past), as well as 1 token each of *has* (*as*), *himplore* (*implore*), *hill* (*ill*), *hoing* (*owing*), and *hafter* (*after*). It is noteworthy that mostly verbs are affected by H-insertion, notably consistently (*h*)*am* and (*h*)*earn* (in present tense) with the first-person pronoun. As regards the adverb *after*, the one token with H-insertion – *hafter* – occurs alongside 5 other examples of *after* with 'standard spelling'. The data also contain an interesting example of self-correction where the word *oferseers* 'overseers' was first spelt with an initial H, thus *hoferseers*, and later corrected. Concerning words preceding an inserted H, and therefore potentially triggering the insertion, the personal pronoun *I* occurs 24 times before *ham* (*am*; 26 tokens in total). As regards *his* (verb *is*), the third-person singular pronoun *he* occurs 8 times before *his* (*is*; 24 tokens in total) while *it* occurs only once and *she* not at all*.* In contrast, all 3 third-person singular pronouns occur with *is* (10 *it* tokens, 5 *she* tokens, 4 *he* tokens). This indicates that the pronoun *he* triggers H-insertion more than the other third-person singular pronouns.

In comparison to the Soundy letters (Berkshire), the Green letters show somewhat different results. Within a variable context of 53 words (tokens) starting with an H, H was dropped in 2 cases only, that is, 3.8 %. The two examples concern *is* for *his* and *as* for *has* (as an auxiliary verb), which appear in immediate vicinity and in the same clause as *harm* where H is retained. There are no other examples of *his* or *has* in the letters, which means that the H is dropped in all of the cases in these 2 word types. In *have* (8 tokens) and *had* (3 tokens), the H is consistently retained in the Green letters.

The Green letters contain 11 instances of H-insertion, of which 7 tokens of *ham* for *am*, 2 of *hoes* for *owes*, 1 of *hanſer* for *answer*, 1 of *harm* for *arm*. While *am*, *owes* and *arm* are subject to H-insertion throughout, the letters contain another H-less variant of *answer*, notably *anser*, as well.

The case study reveals that the stigmatized linguistic features of Hdropping and H-insertion are present in the pauper petitions. Despite varying in terms of sample size, it is possible to observe inter-writer variation in the Soundy and Green petitions. The Green sample contains fewer examples of H-dropping, but these 2 cases – *is* for *his* and *as* for *has* – are also found in the Soundy sample. It is noteworthy that both types do not categorically contain H-dropping in the latter petitions, that is, 28 of 29 tokens of *has* contain H-dropping in contrast to merely 3 of 56 tokens of *his*. Then again, Soundy appears to be rather consistent in her use of H-dropping, particularly with the verbs *had* and *has,* both in auxiliary and lexical verb function. Consistency can also be observed with regard to H-insertion in the case of *am –* thus *ham –* while a lot more intra-writer variation was found with the verb *is* (52 instances of *is* in contrast to 24 instances of *his*). In line with Soundy, Green also consistently inserted an H before the verb *am*. As regards the linguistic context, the latter type is mostly preceded by the personal pronoun *I* in both petition samples. While certain internal factors can be observed in relation to H-dropping and H-insertion such as word type, particularly verbs, and the pronouns preceding them, not all cases are as clear as the example of *I ham* which is consistently used in both petition samples. Even though H-dropping and H-insertion are primarily associated with the lower classes and lack of education, the petition writer's region of origin (here linked to the parish of legal settlement, see Gardner et al. (2023) for a discussion of this issue and related challenges) may also explain the differences. While Soundy comes from an H-dropping area, notably Berkshire, parts of Dorset, that is, Green's county of origin, are considered H-retaining, which could have affected Green's language use, as well as her possibly having received more writing training and being more aware of certain linguistic shibboleths. This may also explain the proportional difference between H-dropping and H-insertion in the petition samples.

#### *4.2 Long s*

#### 4.2.1 Previous studies

The history of long s needs to be viewed from two perspectives, namely the history of long s in print and its use in manuscript material. As regards print, Jones (1798: S, as quoted in Fens-de Zeeuw & Straaijer 2012: 322) made the following observation in his pronouncing dictionary:

In printing, the long ſ is generally uſed at the beginning and in the middle of words, but the ſ hort *s* always at the end. (A new mode of printing entirely without the long ſ has lately been brought into partial uſe.)

1800 is considered the year when the allograph disappeared from printed works. Fens-de Zeeuw and Straaijer (2012: 322) note that 'book antiquarians use the presence or absence of <ſ> to date books to either pre- or post-1800'. According to West (2011: 53), '[t]he death knell for *long s* was finally sounded on September 10th 1803 when […] *The Times* newspaper quietly switched to a modern typeface with no *long s* or old-fashioned ligatures'. While the use of long s in print had completely disappeared in the early nineteenth century, it has survived longer in handwritten documents (cf. Osselton 1984: 125). Fens-de Zeeuw and Straaijer (2012: 322) argue that printing practices would have likely had some influence on the disappearance of long s in handwritten documents and test this assumption by investigating the use of long s in the manuscript out-letters of the grammarians Joseph Priestley (1733–1804) and Lindley Murray (1745–1826).

Priestley did not comment on orthography in his *Rudiments of English Grammar* (1761) and therefore also did not provide a description of the use of long s. The Priestley corpus of 433 out-letters (1762–1804) contains 245 different words with the allograph <ſ> (excluding place names and abbreviations like *Meſ sr*). Long s never appears in initial position, but is consistently and almost exclusively used as part of double s, notably <ſ s>, in medial or final position. More generally, Priestley still used long s in the last letter he wrote before his death in 1804. It is noteworthy that Priestley died at the time when the shift from long s to short s took place in print.

Like Priestley, Murray did not provide any rules on the use of long s in his grammar editions. In contrast to Priestley, Murray's out-letters (1767– 1825) also contain words with long s in initial position but only few such as *ſeems* and *ſelfish*. It is striking that Murray's usage of long s changed in 1803, that is, he stopped using long s. Fens-de Zeeuw and Straaijer (2012: 334) argue that Murray must have been aware of the printers' practices and the move away from long s. The disappearance of long s in Murray's out-letters coincided with its disappearance in *The Times*.

Generally, it has been observed that '[l]ong into the 19th century it was still common practice in English handwriting to use long and short s for double s […]' (Mosley 2008; see also Mugglestone 2012 [2006]: 349). Within this context, we investigate the usage patterns of long s in the pauper petition samples and possible inter- and intra-individual variation. After all, it may be assumed that the labouring poor were not familiar with the usage rules of long s, as found in print and as described in selected grammars. It will therefore be interesting to find out whether their long s usage patterns are similar to those of the two highly educated grammarians (who display variation in their usage), and whether we can also determine a moment in time when long s is no longer used.

#### 4.2.2 Method and results

This case study describes and compares the long s occurrences and patterns in the Soundy and the Green pauper petitions. Special attention is paid to the position in the word and the word type.

The Soundy letters contain 143 long s in 20 letters, of which there are 3 single long s examples and 140 combinations of short and long. There is no single long s in word-initial position, but only in medial (2 tokens) and final (1 token) positions (see Table 6.3).

The short s/long s combination is mostly found in medial position (101 tokens) whereas the long s/short s combination is preferred in final position (32 tokens versus 7 short s/long s tokens), that is, it does not occur in medial position at all. As regards word type, we observed that long s is primarily used in content words, that is, in nouns (83 tokens), verbs (42 tokens), adjectives (13 tokens), and in conjunctions (5 tokens). All 5


Table 6.3. Long s occurrence in the Soundy letters

conjunction tokens are found in the conjunction 'unless'. Similarly, there are only 2 lexical types within the 13 adjective tokens, notably 12 times based on the lexeme 'distress' (*distressing* or *distressed*) and 1 example of *necessary*. The 42 verb tokens contain 40 times *assist*/*ed* including orthographic variations and single examples of other verbs (1 example of *induces* and 1 of *distressed*, notably *induceeſ*, *disſ treest*).

In comparison, we found slightly more variation in the lexical items within the 83 long s noun tokens, notably 15 different ones where *distress* (31 tokens), *goodness* (14 tokens), *overseers* (11 tokens), *assistance* (9 tokens), and *necessity* (6 tokens) are the most frequently used. The Soundy data also contain some intra-writer variation, for example, Soundy uses the noun token *goodness* 14 times, of which the combination of final long s/short s appears 12 times (85.7 %) and the combination of final short s/long s twice (14.3 %). Intra-writer variation is also found in the lexeme *Overseer*(*s*), which appears 45 times in 20 letters by Frances Soundy. In 34 cases (75.6 %), the word was written with short s, while in 9 cases (20 %), it was written as *Oversearſs*, and in 2 (4.4 %) as *Overſears*. These types of variants are not restricted to specific letters but variation is found within the letters. For instance, in a letter from 1829, long s was used 4 times in the type *assist*/ *ance* and it occurred once with a double short s. For all of these examples of intra-writer variation, we have not been able to find any common internal or external factors that would explain the variation. Instead, the restricted education and the lack of writing experience likely explains the orthographic variation regarding long s.

The Green letters contain 34 long s in 7 letters, of which there are 8 single long s in initial position, 18 in medial and 3 in final position (in contrast to 189 single short s, i.e. 52 in initial, 71 in medial, and 66 in final positions). Moreover, Green used long s/short s combinations in wordfinal position (5 tokens) (see Table 6.4).

As regards word type, 4 of the 5 long s/short s combinations concern the noun *Distress* (spelt as *Dreſs*) and 1 occurs in the pronoun *us*, spelt as *oſs*. As for the single long s occurrences in the Green petitions, they are found in nouns, verbs, adjectives, adverbs, determiners, and conjunctions, but there are few lexemes that are clearly predominant. One example is the proper noun *Dorsetshire* that contains a medial long s in 5 of 7 occurrences.


Table 6.4. Long s occurrence in the Green letters

Green used the following spelling variants for the proper noun: *Dorset ſhear* (1 token), *Doset ſhear* (2 tokens), *Dosetſher* (2 tokens), and *Doset Shear* (2 tokens). This example suggests that Green did not always apply long s according to specific rules, with the exception of *Distress* (*Dreſs*) where the final long s/short s combination was consistently found, indicating a lexical preference. As in the case of Soundy, the random use of long s may be explained by the lack of writing experience.

In line with Mosley's (2008) comment that 'it was still common practice in English handwriting to use long and short s for double s […]' in the nineteenth century (see also Mugglestone 2012 [2006]: 349), we can confirm this based on the observed patterns in the pauper petition samples under investigation, both for single and double s. A comparison of the Soundy and Green results regarding long s reveals several differences. While the Soundy data include very few single long s (3 tokens in medial and final positions; 140 combinations), they are dominant in the Green data (29 tokens in initial, medial and final positions; 5 combinations). While Soundy thus does not use long s in initial position, Green's use of initial long s agrees with Lindley Murray's usage. Soundy uses the short s/ long s combination mostly in medial position (101 tokens) and the long s/short s combination in final position. In contrast, Green used long s/ short s combinations in word-final position (5 tokens) and no short s/long s combinations. In both data sets, interesting patterns regarding lexical type were found, notably examples like *goodness, assist/ance*, and *overseers* in Soundy's petitions and *Distress* (*Dreſs*) in Green's petitions.3

It can generally be observed that the long s use in both pauper petition sets did not change over time, that is, it can be found in almost all letters under investigation. It may be assumed that both Soundy and Green were familiar with the long s orthography from the King James Bible and the Common Prayer Book, as well as other sources. They were clearly not aware of the disappearance of the allograph in printed works around 1800, or, in any case, they did not follow this development. As regards the inter- and intra-writer variation found in the data sets under investigation, as indicated above, this can best be explained by a lack of writing training and experience.

## 5 Patterns of linguistic variation in pauper petitions: Reflections and conclusions

In this chapter, we considered different types of variables (and related variants), notably (1) H-dropping and H-insertion, which had become stigmatized in normative works in the eighteenth century, and (2) long s, which was in the process of disappearing in printed texts around 1800 and in hand-written texts in the course of the following century. As regards the data source under investigation, we focused on two sets of pauper petitions written by two women during the early part of the nineteenth century. All of the letters under investigation (twenty letters from Soundy and eight from Green) were addressed to the overseers of

3 The distribution of s variants in Green's letter is as follows: <Dress/Dreſs>: 0/4; <Must(e)/Muſ t>: 4/2; <troubsom/troubſom>: 0/1; <wish/Wiſ h>: 1/1; <small/ ſmall>: 0/1; <case/caſe>: 0/2; <is (verb only)/iſ>: 10/1; <please/pleaſe>: 5/2; <anser/ (h)anſer>: 1/3; <as/aſ>: 12/1; <Lastorll/Laſstorll>: 0/1; <Grasen/Graſen>: 0/ 1; <Dorset Shear/Dorset ſ hear (all variants)>: 2/5; <send(e)/ſent/ſ sende>: 5/1/ 1; <assure/a ſure>: 0/1; <received/reaſeved>: 1/1; <h(o)usband/huſ band>: 6/1; <was/waſ>: 2/1; <several/ſevrell>: 0/1; <us/oſs>: 1/1; <short/ſhorte>: 0/1.

the writers' respective home parishes, that is, addressees who were higher up the social scale and who decided whether the requests for out-relief would be granted. Selected external factors like sex of writer, text type, and sex and status of addressee are therefore the same in both data sets. Moreover, the period of writing is roughly the same, that is, 1818–30 for Frances Soundy and 1820–26 for Charls Ann Green. Differences between the two writers can be found with regard to their parishes of legal settlement, notably Pangborn in Berkshire for Soundy and Wimborn in Dorset for Green, and their education and writing experience. As indicated in Section 2.2, compulsory elementary schooling did not exist when both writers were educated and they would have therefore received a different type of writing training, depending on the opportunities available in their home parishes and social networks. Considering this, reasons for linguistic variation can either be explained by language-internal factors or external factors linked to home parish and education possibilities.

As regards H-dropping, the fact that Soundy dropped H a lot more often than Green may suggest that (a) Soundy comes from an H-dropping region (community level), (b) that she dropped H frequently in speech (individual level), which is reflected in the petitions, or (c) that Green came from a county that had some H-retaining pockets and/or that she was more aware of the spelling norms (education and writing experience), which resulted in less H-dropping. The findings show in any case that certain words are more prone to H-dropping in both data sets, notably *has, his*, and *had* (in the Soundy data), while others are not, for example, *have* in the Soundy and the Green data. H-insertion, which may be interpreted as hypercorrection as a result of linguistic insecurity and can also shed light on a writer's intentions, is again found in both data sets, albeit differing proportionally. Here we observed that both writers consistently insert an H before the verb *am*. They do however display spelling variation (with and without H-insertion) regarding other words, that is, the verb *is* in the case of Soundy, and *answer* in the case of Green. These variation patterns are likely due to individual choices rather than being influenced by regional community speech patterns. Once the pauper petition corpus is complete, the latter interpretation can be tested on a larger scale.

The long s case study differs from the previous one in that we are dealing with a linguistic feature that is only used in writing and is therefore not influenced by spoken language. Rather, the allograph was gradually disappearing in printed and hand-written texts in the course of the Late Modern English period. While its use was not commented on in the works of the normative grammarians Priestley and Murray, the latter seemed to be sensitive to its use and abandoned it shortly after 1800 in his correspondence (see Fens-de Zeeuw & Straaijer 2012). The Soundy and Green findings suggest that both writers were not aware of the allograph's development and/ or any rules related to its use. The great amount of inter-writer variation, as well as intra-writer variation, can most likely be explained by different writing training that the writers have received and their writing experiences.

Generally, the patterns of linguistic variation found in the two pauper petition data sets, particularly the free/random variation found, can best be explained by education possibilities, and possibly regional differences in the case of H-dropping. In order to strengthen these interpretations, in future research, we will continue investigating pauper petition samples from other English counties.

## References

#### *Primary sources*

Green data: Dorset County Record Office; Wimborne, PE/WM/OV/13/3A, PE/ WM/OV/13/3B.

Soundy data: Berkshire Record Office; Pangbourne, D/P 91/18/4/1, D/P 91/18/4/ 2, D/P 91/18/6, D/P 91/18/10, D/P 91/18/11.

#### *Secondary sources*

Alford, Henry. 1864. *A Plea for the Queen's English: Stray Notes on Speaking and Spelling*. 2nd edn. London: W. Strahan.


#### Anne-Christine Gardner

# 7 Petitioning for the education of the poor: Self-corrections as stylistic choices in a Late Modern English draft letter

#### Abstract

Adopting a variationist perspective, this article illustrates the value of self-corrections as a data source for the study of intra-writer variation, focusing on a draft letter written in 1831 by the curate of Aymestrey to the bishop of Hereford in support of his parochial school. The numerous stylistic self-corrections, which are interpreted within the framework of Speaker Design, represent conscious linguistic choices through which the author is carefully constructing his identities and interpersonal relationships. With his use of in-group references on the local, diocesan and national levels, the clergyman's self-corrections can be seen as an attempt to position himself and his local school-building efforts in the institutionalized, nationwide movement to improve educational possibilities for the labouring poor.

## 1 Introduction1

While modern research in variationist sociolinguistics is mainly based on spoken data, we can explore the more distant past only through written texts. The most frequently consulted source type for diachronic studies is correspondence, since letters contain many oral features and thus range among the most speech-like texts (Biber & Finegan 1989: 493; Culpeper & Kytö 2010: 17). Letters also typically form the bedrock of historical studies on intra-writer variation, where the focus is typically placed on a set of letters by an individual, to illustrate how factors such as intended audience or communicative purpose influence the linguistic choices

1 The author would like to thank the reviewers for their useful comments.

made by the writer (e.g. Auer 2015; Schiegg 2018; Hernández-Campoy et al. 2019). The linguistic choices under investigation usually reflect stylistic variation in the language of an individual who employs stylistic resources actively and creatively for the 'presentation and recreation of speaker identity' (Schilling-Estes 2002: 388).

However, one aspect of letter-writing, or handwritten ego-documents in general, has not yet received much attention, namely self-corrections. These are textual alterations made by the author where


Self-corrections signal speaker choice, that is, conscious decisions made concerning linguistic items, giving us insights concerning level of schooling, intended readership, communicative purpose, stylistic repertoire, identity construction and more. Only few studies to date contain a systematic analysis of self-corrections in English autograph documents (Auer 2008; Fairman 2008; Tieken-Boon van Ostade 2008, 2014; Gardner et al. 2017; Gardner 2018, 2022, forthcoming), and the same seems to hold for other languages as well (but see Schiegg & Freund 2019; Freund forthcoming for German). Yet even in a single ego-document selfcorrections can offer valuable insights into the writer's mind, the linguistic options they perceive, which choices they make, and possible reasons why. Self-corrections concerning style, in particular, are indicative of an active and creative engagement of a writer with the linguistic resources at their disposal and can indicate identity work in action. Following the definition given above, such self-corrections thus exemplify intra-writer variation and are consequently placed at the centre of my analysis.

The focus of this article lies on an undated letter riddled with selfcorrections which was written in 1831 by the curate of Aymestrey, Thomas Taylor Lewis, to the bishop of Hereford, George Isaac Huntingford. The author is trying to establish a local school and hopes to persuade the bishop to provide funds for the building of a school room. This is encapsulated in (1), which contains the two most common methods of correction, that is, strike-throughs and insertions above the line, as illustrated in Figure 7.1. How these corrections are marked in the text deserves further comment. By adding diagonal lines to the horizontal lines, the author visually distinguishes the deletion of *the last four months* from *superintended*, which are two different instances of corrections. The former is inserted later-on in the sentence with the addition of *during*, while the verb is replaced by *endeavoured to form & establish*. In both cases the insertions appear above the line. The author often uses a caret (˰) below the line to signal the place of an insertion, sometimes surrounded by an opening and closing bracket, as was done to mark the insertion of *during the last four months* and of *is &*. The underlining of *upward of one hundred* and of *ninety eight* may indicate that these two elements belong together, but, in the absence of similar use of underlining elsewhere by this author, it seems more likely to be used to highlight the number of pupils (and presumably the school's success).

(1) *I have the endeavoured to form & establish here last four months superintended a School consisting of upwards of one hundred ninety eight Boys and Girls, & conducted it as far as =during the last four= months poſsible according to Doctor Bells' method .For want of a School Room I have been obliged to aſsemble the Children in the Church; which at the present & approaching Season of the year is & will be hardly practicable.*<sup>2</sup>

2 In examples the equal sign is used to represent text placed two lines above the main line.

Figure 7.1. Self-corrections illustrated.3

The aim of this article is to show that the purpose of many alterations in this letter can be understood and explained by approaching the material from a sociohistorical perspective, adopting the framework of Speaker Design (Schilling-Estes 2002; Kiesling 2013). A particular focus is placed on what identities the author constructs, which interpersonal relationships or social groups he is concerned with, and how the external situation shapes his writing.

After discussing the data source in more detail in Section 2, the sociohistorical context of the letter will be presented in Section 3, taking up the writer's mention of *Doctor Bells' method* and determining what kind of school the writer tries to establish. This will be contextualized by considering what schooling opportunities were prevalent in England at the time, and which role the Church of England played in providing education. The writer's self-corrections are analysed in Sections 4 and 5. Section 4 considers the frequency and types of emendations in the letter, and what we learn from them about the production circumstances and the main concerns of the writer. Section 5 examines the data from the perspective of Speaker Design, showing how the sociohistorical context shapes the curate's linguistic choices, which illustrate intra-writer variation during the writing process:

3 All images are reproduced with kind permission of the Herefordshire Archive Service.


The article concludes in Section 6 with a summary and outlook.

## 2 Data

The letter under investigation, which contains 807 words, appears to be a draft: there is no superscription, and it is neither signed nor dated.4 The Hereford Archive and Record Centre documents this as one item in the correspondence (F71/113) between the vicar of Aymestrey and the bishop of Hereford from the period 1839–42. However, two replies from the bishop contained in the same archive folder are dated 9 and 16 March 1831, and signed by 'G. I. Hereford', that is, George Isaac Huntingford (1748– 1832), who was bishop of Hereford from 1815 until his death (Robinson 2004). According to the Clergy of the Church of England Database (CCEd) and Foote (2004), there was only a curate at Aymestrey at the time in question, Thomas Taylor Lewis (1799/1800–1858). Graduating from St John's Cambridge with a BA and MA, Lewis became curate of Aymestrey in 1827 and in 1841 vicar of Bridstow. A comparison of the handwriting with two further draft letters in F71/113 written in response to the bishop's replies, signed 'T. T. Lewis' (undated) and 'T. T. L' (31 March 1831) respectively, confirms that the author of the draft letter was indeed the curate. Considering the date of the bishop's reply, Lewis must have sent his letter sometime before 9 March 1831. Owing to the detailed

4 The total word count of 807 includes not only the intended final version of the text, but also any words (including incomplete words containing at least one morpheme, e.g. *Establishm*) which were originally written and then deleted.

and fine-grained analysis of this letter, and for reasons of space, the two further draft letters can unfortunately not be discussed further in this article.

The curate's letter was sampled as part of the collection of handwritten documents used for the SNSF-funded project 'The Language of the Labouring Poor in Late Modern England' (LALP; see also Chapter 6 by Auer, Gardner & Iten in this volume).5 Documents about the poor are included as well because they offer intriguing insights into what kind of educational possibilities th labouring poor may have had. This will be explored further in the following section.

## 3 Church involvement in elementary schooling in nineteenth-century England

The curate of Aymestrey writes that he tries to conduct his newly founded school 'as far as poſsible according to Doctor Bell's method'. At the time this was an innovative teaching system which was developed by Reverend Dr Andrew Bell when superintendent of the Male Orphan Asylum in Madras (Blackie 2004) and published on his return to England in Bell (1797). The main principle of Bell's method, or the Madras system, was that advanced pupils taught younger ones. In this he was inspired by schoolchildren in India who drew in the sand on the seashore to teach the alphabet to younger children. Pupils were taught by tutors, who were in turn supervised by an assistant; both tutors and assistants were schoolchildren of advanced abilities. An adult teacher and a superintendent were at the highest levels of the hierarchy. The monitorial system was very cost-effective in that only one or two adult teachers were needed to preside over a group of several hundred pupils.

<sup>5</sup> This article was written within the context of this research project (SNSF grant 100015\_188879).

The Madras system combined the teaching of school topics (including reading, writing, arithmetic and geography) with physical exercise and practicing work-related skills and crafts like needlework and card-setting. Unusually for the time, writing was taught before reading, by first drawing letters in sand, and children were grouped by ability rather than age. Lessons were kept short at fifteen minutes at the lowest level, thirty minutes at more advanced stages. Having himself experienced severe beatings at school, Bell was expressly against corporal punishment and instead ensured discipline through a reward system. For instance, pupils who behaved and performed well were allowed to rise to the next class level or were selected to become tutors or assistants. The overall aim of Bell's method was to educate the labouring poor to become independent and find employment, as well as to become law-abiding and god-fearing citizens (Bell 1797; Iremonger 1813: 234f.).

Bell's method was promoted by the Church of England, which had been the most important provider of elementary schooling in England for centuries.6 In the eighteenth century, for instance, countless charity schools were established by the Society for the Promotion of Christian Knowledge (SPCK). In the early nineteenth century, parochial ministers were even required by law to teach 30 minutes a week (Bell 1808: 7f.). In 1811 the National Society for the Education of the Poor in the Principles of the Established Church was founded, which advocated Bell's method, and textbooks published by the SPCK were used as teaching material in the associated National Schools (Louden 2012: 17). A main aim of the National Society was 'to instruct and educate the poor in suitable learning, works of industry, and the principles of the Christian religion, according to the Established Church' (Dixon 2018: 191).

A rivalling monitorial system devised by the Quaker Joseph Lancaster was promoted by the British and Foreign School Society, established in 1808. Sharing many commonalities, a main difference between the two systems was that unlike in National Schools there was no religious instruction in British and Foreign Schools. The National Society was more

<sup>6</sup> For an overview of various types of religious education, also by other denominations, see Raftery (2012).

successful than its counterpart, not least because the Prince Regent, the future King George IV, became its patron. The British and Foreign School Society was supported by his father, George III, until his death in 1820 (Dixon 2018: 188f.; Parker et al. 2020: 546f.). Other main benefactors of National Schools were the SPCK and bishops, which also explains why the curate of Aymestrey turned to the bishop of Hereford for support in the building of a school room. His request was not unusual. Iremonger (1813: 203) had noted earlier that '[t]he expence of building new schoolrooms is indeed a very serious consideration, as well as one of the greatest impediments towards the furtherance of our *National* work'. By 1832, the National Society had increased its number of schools to c. 13,000, and the number of pupils to around one million. From 1833 onwards, government grants helped enable the spread of National Schools even further. However, private subscriptions and donations remained essential since all grants had to be matched in value by voluntary subscriptions. In the first five years 70 % of all state funding went to Anglican schools, dropping to c. 25 % by the late 1860s. The importance of Church-led schools diminished with the 1870 Education Act, when the state started providing (secular) elementary education. Only ten years later did schooling become compulsory at a national level (Dixon 2018: 195, 2019: 292; Crone 2018: 163; Parker et al. 2020: 547, 555).

Also from an educational perspective, Bell's method proved to be successful. In a nationwide survey conducted in 1836, pupils attending National Schools were described as 'useful members of Society', their 'good character', 'good conduct' and 'lack of criminality' were praised, and quite a few pupils were found to have become teachers themselves (Dixon 2019: 304f.). In a study of Suffolk prison registers, Crone (2018: 176f.) notes that between 1840 and 1870 only few offenders had been educated at a National School. Reports of more civilized behaviour and increased church attendance suggest that their children's education appears to have had a reforming effect on parents as well (Dixon 2018: 231).

By choosing to conduct his school according to Bell's method, the curate of Aymestrey was evidently following current educational trends supported by the Church of England and the government. In how far the external situation impacts the curate's identity construction and textual changes is discussed in Section 5. First, Section 4 provides an overview of the frequency and types of self-corrections in the curate's letter to better understand how the letter was produced and which aspects of the text the curate seems to focus on most.

## 4 Overview of self-corrections: Production circumstances and the writer's concerns

Lewis' letter contains sixty-five self-corrections, which in terms of normalized frequency amounts to 80.55 corrections per 1,000 words. Ensuring comparability of normalized frequencies with, for example, Auer (2008), any self-correction regardless of length (ranging from one character to a string of several words) which represents a coherent unit in the correction process is counted as one occurrence. This means that a change concerning only one character, for instance the overwriting of lower-case <p> with upper-case <P> in *[^P OVERWRITES p^]art*, is counted as one self-correction, as is, for example, a more extensive change affecting a verb phrase, such as the deletion of *superintended* and its replacement with *endeavoured to form & establish* (see Table 7.1).7

According to contemporary letter-writing manuals like *The Complete Letter-Writer*, self-corrections were seen as 'a rudeneſ s to the perſon to whom they are written' (1776: 12), and preferably only error-free letters should be sent. Auer (2008: 214f.) notes that as opposed to 'sent letters or copies of sent letters', draft letters 'are likely to contain most self-corrections'. As suggested in Section 2, it is likely that the curate's letter is a draft and a comparison with the frequency of corrections in letters by well-educated contemporaries lends further support to this hypothesis. In carefully written letters, like those by Lucy Whitaker and Jane Austen, the normalized frequency of self-corrections is 1.59 and 2.10 per 1,000 words, respectively

7 The editorial code [^a OVERWRITES b^] indicates that orthographic unit a (e.g. a character or a word) is written over unit b, thus replacing it.

(based on Auer 2008; Tieken-Boon van Ostade 2014), and Austen's most formal letters do not contain any alterations at all (Tieken-Boon van Ostade 2014: 85). In an example of unmonitored writing by Mary Hamilton, a travel journal addressed to her mother, the rate of self-corrections is 10.52 (Gardner accepted). In contrast, in two draft letters of hers there are significantly more emendations at 35.08 (to Queen Charlotte) and 39.90 (to Frances Burney) per 1,000 words, respectively (Gardner 2018: 91). The curate's text, with a rate of 80.55 corrections, contains about twice as many changes, which underlines the draft status of the curate's letter. Like the formal letters by Austen and the drafts by Hamilton, Lewis' letter to the bishop was produced in a formal context. Holding an MA from Cambridge, he is likely to have been aware of the letter-writing etiquette of the time and will not have sent off a letter full of corrections. Instead, the curate must have spent a considerable amount of time writing and re-working his letter, and afterwards produced a clean copy to send to the bishop.

While all types of self-corrections (such as 'slips of the pen', omissions, stylistic or grammatical changes) can be encountered in ego-documents from members of any rank in society, both level of schooling and communicative purpose are revealed through the predominance of specific types of self-corrections. Letters by paupers, for instance, tend to contain a significant rate of corrections pertaining to mechanical errors ('slips of the pen') and spelling, which betray the writers' limited schooling (Fairman 2008). In comparison, corrections in ego-documents by more highly educated individuals aim more strongly at changes in content or style (Auer 2008; Gardner 2018, 2022, accepted).

In the curate's case, fewer than a quarter of all self-corrections address mistakes or grammatical issues (see Table 7.1), which testifies to his university education. These include



Table 7.1. Frequency and types of self-corrections

As many as fifty alterations concern stylistic changes. Following the definition given at the beginning of this article, these are changes which generally reveal identity work in action and/or reflect on the communicative purpose of the letter. One example is the previously mentioned rephrasing of the verb *superintended* (see (1) and Table 7.1). Here the curate moves away from a verb which describes a more passive involvement to a more complex verb phrase which highlights his leading role in and commitment to building a school. Furthermore, additions can be made, or sentence elements moved, to affect a shift in emphasis. In (1) the curate

8 An asterisk within curly brackets indicates a character which cannot confidently be identified.

draws the reader's focus to the temporal dimension of his request for help by moving *the last four months* to the end of the sentence, and subsequently invokes urgency by adding the present tense (*is &*) to *will be hardly practicable*. Also, sentence elements can be removed, and a train of thought aborted (labelled false start), to ensure a certain focus or emphasis. This is illustrated by the deletion of *on account* in (9); rather than explain why nine of his pupils cannot read, Lewis focuses on the success of his school by merely describing the remaining pupils' skills. Stylistic corrections are thus not only purely lexical changes, for example, from Germanic to Latinate lexis (*give* to *add*), to elevate the style of writing, as in (6). The fact that over 75 % of all self-corrections constitute stylistic changes corroborates the hypothesis that Lewis was drafting and crafting his letter most carefully. Section 5 proceeds to illustrate in more detail how the curate's stylistic changes relate to identity construction in action.

## 5 Drafting the letter: Self-corrections and identity construction

#### *5.1 Thomas Lewis' identities*

For the discussion of the curate's construction of identities Kiesling's (2013: 450) definition is adopted, who states that '*identity is how individuals define, create, or think of themselves in terms of their relationships with other individuals and groups*' (original italics). Furthermore, Kiesling (2013: 450) points out that this definition highlights 'the dual individual and social nature of identity, because it is about how the individual relates to society, whether on a group or individual level'. Figure 7.2 represents the curate Thomas Lewis' professional relationships with other individuals and groups, as well as how he portrays himself in relation to them, based on his letter to the bishop of Hereford. The social groups he positions himself towards are arranged here on a scale from local to national. At the local level there is his parish congregation, together with pupils and staff from his school. His relationship at the diocese level concerns the bishop of Hereford, the addressee. Since the letter is a request for support, the curate primarily takes on the role of supplicant. At the national level, encompassed by the Church of England, he engages with the National Society, specifically their aims, ideals and methods concerning the education of the poor. The three adjectives *committed*, *successful* and *humble* represent the main character traits which Lewis projects in relation to the various social groups; these traits are alluded to in connection with (1) as well as in Sections 4 and 5.3–5.5.

As Schilling-Estes (2002: 388) states, speakers 'use their speech to help shape and re-shape the external situation (whether the immediate interactional context or wider societal forces), as well as their interpersonal relationships and, crucially, their personal identities'. In this vein, the following subsections illustrate how intra-writer variation is revealed through Lewis' stylistic self-corrections. Particular attention will be paid to how his linguistic choices mark the curate's concern with constructing his identities, his interpersonal relationships and his sense of belonging, that is, being associated with certain social groups, at the parochial, diocesan and national levels.

Figure 7.2. Thomas Lewis' identity construction: Character traits and interpersonal relationships.

#### *5.2 Belonging at parish level*

At the local level, the curate indexes group membership and belonging by adding the proximal deictic term *here* at the end in (2) (see Figure 7.1):

(2) *I have the last four months superintended endeavoured to form & establish a School here*

In (3), Lewis replaces *I* with inclusive *we* and subsequently deletes selfreferential *I have*:

(3) *We [^OVERWRITES I^] have provided the Cards for spelling – Spelling – Books – Testaments Sand trays & Slates –. And I havefound it absolutely neceſsary to clothe most of the Children*

It may simply be a matter of correcting facts, meaning that it was not just the curate who went to the expense of providing teaching materials and clothes to the pupils. However, the alteration makes it clear that to clothe the children was a group decision, and not one made by Lewis alone. The change of pronouns therefore signals that at parish level the curate is supported in his endeavours by members of staff at the school.

*5.3 Belonging at diocesan level: Interpersonal relationship with the bishop of Hereford*

At the very beginning of the letter (Figure 7.3), Lewis focuses on the dichotomy, and strengthens the link, between parish and diocese. In (4), he replaces the demonstrative pronoun *this* with the second-person personal pronoun *your*, stressing the fact his parish belongs to the bishop's diocese and therefore falls under the senior cleric's responsibility. The curate also reminds the reader that, by establishing a school conforming to the ideals of the National Society (see also (7)), he is acting in consequence of the bishop's own *Reccommendation.*

(4) *Having, in pursuance of your Lordships Reccommendation, given to the Clergy of this your Diocese early last Spring*

In (5), the curate further highlights the responsibilities of the bishop. However, by adding *profeſsedly* as a marker of epistemic stance, Lewis

Figure 7.3. The beginning of Lewis' draft letter to the bishop of Hereford.

suggests that he does not believe in the parishioners' waiting tactics and thereby reduces the pressure on the bishop to act. Nevertheless, he still signals that the bishop's support is needed to ensure donations from the inhabitants of Aymestrey.

(5) *Some Few of them* [i.e. 'The Habitants'] *would give a small Contribution, but these wait profeſ sedly for the authority of their Landlords Example & approbation in the Busineſs*

With respect to the addressee, Lewis portrays himself as a supplicant and the bishop as a potential benefactor. In (6) three sets of self-corrections highlight these social roles (see Figure 7.4):

(6) *It willgive add add materially to the Life & Vigour [of our humble Institution]*<sup>9</sup> *if your Lordship willgiv allow us the{\*} advantage of looking up to you as our <sup>P</sup> atron & [V OVERWRITES v]isitor –.*

Twice the Germanic verb *give* is replaced by a Latinate word, first *add* and then *allow*, elevating the style of writing. In the first instance, this change was not straightforward: Lewis struck out both *give* and *add*, before settling on *add materially* inserted above the line – the addition of the adverb

<sup>9</sup> This complex self-correction is discussed in Section 5.5 (12).

Figure 7.4. Example (6) illustrated.

emphasizes the significance of the bishop's support. The subsequent exchange of *giv* with *allow* underlines the bishop's power and freedom in choosing to act as benefactor. Finally, the insertion of *Patron &* before *Visitor* at the end of (6) lends further weight to the bishop's role as benefactor.10 The nouns *patron* and *visitor* have partially identical meanings in that they both refer to someone who supports an institution, typically financially, while in the context of the letter *visitor* has the additional sense of 'supervisor of a school' (*OED Online*). The coordinated binomial structure *Patron & Visitor* is an example of the rhetorical figure 'hendiadys' (Kopaczyk & Sauer 2017: 15), another means of elevating the style of the letter. Similar linguistic behaviour was found in a draft letter from Mary Hamilton to the admired author Frances Burney (Gardner 2018: 90–92). The careful crafting of a stylistically elevated text appears to be a means of expressing respect towards the addressee.

(6) also exemplifies at which stages Lewis made corrections. The change involving *allow* must have been made during the writing process since Lewis did not complete the discarded verb, stopping before the last letter (*giv*). Regarding *Patron &*, it is conceivable that the insertion was made at a later stage, for instance upon re-reading the draft, since the thinness of the ink stroke does not resemble the broader stroke found in its immediate vicinity, but rather the thinner stroke of the final passages (see Figure 7.5).

<sup>10</sup> The change of spelling from *v* to *V* in *Visitor* is addressed in Section 4 (Table 7.1).

Figure 7.5. End of Lewis' draft letter.

#### *5.4 Belonging at national level: The National Society*

One aspect of identity construction that Lewis pays particular attention to through self-corrections is to highlight that he closely follows the system of the National Society with its National Schools and the Madras method. The curate does so, for instance, by using in-group references and altering his text to include adjectives and nouns associated with the National Society. (7) and (8) are very similar, although word repetition is partially avoided by replacing *Plan* with *Method* in (8):


In both examples, *parochial* is added to allude to the local scale of the curate's endeavours. The addition of *national* in (8) allows Lewis to claim membership of the national educational movement. Perhaps the curate had hopes that his school would eventually be taken into union, that is, join the ranks of National Schools. It is certainly poignant that (7) and (8) frame the letter, appearing in the opening and closing passages, respectively (see Figures 7.3 and 7.5).

Furthermore, the curate projects the persona of a committed educator acting in line with the principles of the National Society, presenting his success at transforming his pupils from the lower classes into valuable members of Christian society (see Section 3). In (9) Lewis describes their advances in reading and spelling, and also notes how they have mastered, rather than are learning, the Catechism. Importantly, the pupils do not learn to merely reiterate religious teachings but are led to understand and enact Christian virtues. The change in (10) exemplifies this and suggests that the pupils are *learning the value of Christian Benefits*, rather than just *feeling thankfully for* them, meaning that the pupils are more likely to emulate this kind of behaviour. Lastly, the addition of *regular attendance* in (11) shows a further instance of the pupils' good behaviour in that they attend school regularly. This in turn has led to an increase of the curate's congregation, which will not only have encompassed his pupils but also their families. The curate thus evidences his success in promoting the aims of the National Society, reaching beyond the individual pupils into their family circles. Incidentally, regular church attendance by its pupils was a prerequisite for any school wanting to achieve the status of a National School (Louden 2012: 16).


#### *5.5 Interweaving identities through self-corrections*

Lewis' careful shaping of identities becomes particularly evident in (12), where he changes his mind several times and the self-corrections of the prepositional phrase encompass four lines (see Figure 7.6):

(12) of this our Establishm Institution parochial School humble =Sch= national Institution

Figure 7.6. Self-corrections across four lines.

The final version reads *of our humble Institution*. The phrase is part of (6), where the curate works to elevate his style and highlight the bishop's role of benefactor. Through his changes Lewis skilfully manages to interweave the three most important aspects of his identity as projected in the letter. By using *our* rather than *this*, he signals that at the parish level he is part of the school community. It is interesting to note that *our* was in fact his first choice and that he deleted *this* again after adding it above the line. Lewis must have considered *this* as an alternative only briefly since he did not even cross out his first choice *our*. The adjective *humble* presents the curate at diocese level as a supplicant in relation to the addressee, the bishop of Hereford. At the national level, in reference to the National Society and its National Schools, we see Lewis wondering which noun to use to describe his school, and whether to include a premodifier or not. He decides against an adjective like *parochial* or *national*, which may have taken away some of the reader's attention of *humble*. Concerning the noun, the author might have decided against *Establishm[ent]* since the meaning of 'school' was apparently only just emerging; the *OED Online* gives a first attestation for the year after the letter was written. Säily et al. (2018: 42) demonstrate that in eighteenth-century correspondence lexical innovations often appear in letters by professionals to family members and close friends. Using a word with a newly developing meaning in a formal setting may therefore have posed a social risk to Lewis. The curate may have preferred *Institution* to *School* for stylistic reasons. *School* appears twelve times in the text, either on its own or in correlation with *Room* or *House*, whereas *Institution* occurs only twice elsewhere, and repetition could consequently be avoided. In addition, as a morphologically more complex borrowing *Institution* was associated with a more learned register than monomorphemic *School*, which was already in use in Old English. The curate's deliberations in (12) signal once more his concern with using an elevated style to show respect towards the addressee and to further his cause, and with indexing in-group memberships at the parochial, diocesan, and national levels.

## 6 Summary and outlook

Self-corrections in handwritten ego-documents are a rich source of data, yielding significant insights into a writer's motivations and choices. Stylistic corrections, in particular, were seen to illustrate intra-writer variation. The focus in this article was placed on uncovering how the curate uses, and chooses between, linguistic resources at hand to construct his identities and interpersonal relationships on a scale from local to national. The letter by the curate to his bishop is clearly a draft, and changes Lewis makes can be interpreted from the perspective of Speaker Design as identity construction in action. Added and modified in-group references to the parish and the school, the diocese and the National Society strengthen and project a sense of belonging at the parochial, diocesan and national levels. The curate pays close attention to the relationship with the addressee, portraying himself as a supplicant and the bishop as benefactor. He also shows respect towards the addressee by stylistically elevating his letter, which he accomplishes by avoiding repetitions and a potential neologism, and by using the rhetorical device *hendiadys*. Through his references to the National Society, Lewis raises his local endeavours to a national level.

Was the curate successful in reaching his communicative goals? The bishop's reply to the curate's letter survives in the same archival holding, as mentioned in Section 2. From this brief letter (forty-seven words) and an annotation by Lewis we learn that the bishop, who had himself encouraged the clergy of his diocese to establish schools (see Section 5.3), donated the considerable sum of twenty pounds. He refused Lewis only one request, namely to become *Patron & Visitor*. In his second letter to the curate, the bishop states that *there is no need* for such an appointment owing to the presence of a board of trustees. The curate's reply, surviving in draft form, details his concern that this lack of support might lessen the chances of success for the school's *application for Union*. However, Tonkin (1967) suggests that from 1840 a school at Aymestrey did become affiliated with the National Society, so eventually the curate's identity work seems to have borne fruit on all levels.

The present analysis focused exclusively on self-corrections in the curate's letter, and what they reveal about the writer's intentions and identity construction. It would also be illuminating, for instance, to adopt the approaches of (critical) discourse analysis or relational work, to further elucidate the curate's motivations and strategies whilst drafting the letter. Little research has been done on self-corrections in ego-documents of the labouring poor, that is, the main target group of the curate's school, and the middling sorts. As part of my future research in the LALP project, I will investigate self-corrections in letters from these groups, considering issues such as social variation as well as the influence of schooling on the stylistic repertoire. Examining self-corrections in texts from a wider societal spectrum will hopefully lead us to a better understanding of language use in the past.

## References

#### *Primary source*

F71/113. Hereford Archive and Record Centre.

#### *Secondary sources*

Auer, Anita. 2008. 'The letter wihch that I wrote'. Self-corrections in Late Modern English Letters. In: Marina Dossena & Ingrid Tieken-Boon van Ostade (eds), *Studies in Late Modern English Correspondence. Methodology and Data*. Bern: Lang. 213–34.


#### Christine Elsweiler

# 8 Intra-writer variation in the requestive behaviour of two Early Modern Scottish letter-writers

#### Abstract

Early English correspondence has proved a valuable source for the study of intra-writer variation on different linguistic levels (see Hernández-Campoy et al. 2019). This study investigates the role of audience design (Bell 1984, 2001) and speaker design (e.g. Coupland 2001) in the variation of requestive styles displayed in the correspondence of Thomas Hamilton, 1st Earl of Haddington (1563–1637) and of John Erskine, 18th/2nd Earl of Mar (c. 1562–1634), two high-ranking Scottish politicians. It is shown that, while their requestive behaviour is governed by epistolary conventions, these nevertheless provide them with a repertoire of request strategies and mitigating devices allowing them to mould their requestive styles to accommodate to different addressees and to project changing social roles.

## 1 Introduction

Correspondence has been shown to be a rich source for the study of intrawriter variation, since 'letters reflect the personal communicative style of an author, who maintains and negotiates a particular social relationship with his/her addressees in a given situation and with a certain purpose' (Hernández-Campoy et al. 2019: 288). Accordingly, several studies (e.g. Nevala 2004a, 2004b; Auer 2015; Hernández-Campoy & García-Vidal 2018a, 2018b) have highlighted the role of accommodation towards the addressee (Bell 1984, 2001) and identity creation (e.g. Coupland 2011) for intra-writer variation in early English letters.

In the Early Modern period, the structure and language of letters were influenced by epistolary conventions, which provided for the due consideration of both the writer's and the addressee's social rank and power status (e.g. Nevalainen 2001). This adherence to conventions is reflected in the requestive behaviour of Early Modern Scottish correspondents, for example, in the manifest general preference for formulaic performative requests, a direct request strategy, in both non-private and private correspondence1 (Elsweiler 2022, 2023), exemplified in (1) and (2):


Despite the relative homogeneity regarding performative requests, (1) and (2) illustrate that variation is, for example, evident in letters reflecting different power relations between the writers and the addressees. Requests addressed at superiors tend to contain deferential mitigators such as the adverb *hwmillye* and the honorific address form *your grace* in (1), whereas in requests addressed at inferiors, see (2), such deferential modifiers are rarer.

This study will build on these findings by investigating intra-writer variation in the requestive behaviour of two leading figures in the Early Modern Scottish political scene: Thomas Hamilton, 1st Earl of Haddington (1563– 1637), and John Erskine, 18th/2nd Earl of Mar (c. 1562–1634). Specifically, I aim to explore the interplay between epistolary norms and individual requestive choices by analysing the effect different addressees as well as changing social roles have on the selection of request strategies. This qualitative analysis of intra-writer variation will be conducted against the background of quantitative findings for a corpus of Scottish private and non-private correspondence (1570–1700).

In Section 2, the corpus material and methodology of speech act annotation will be outlined. This is followed by a quantitative overview of request strategies in Scottish correspondence, which will be related to

<sup>1</sup> In this study, private letters are defined as correspondence exchanged between family members, with non-private correspondence comprising letters outside of the family circle, including administrative and other official letters (Meurman-Solin 1993: 122f.).


Table 8.1. Word count in the non-private and private correspondence sub-corpora

epistolary conventions in Early Modern correspondence (see Section 3). Section 4 will introduce the audience and speaker design approaches to stylistic variation, which provide the explanatory framework for the analysis of intra-writer variation in the correspondence of Thomas Hamilton and John Erskine in Section 5.

## 2 Corpus material and methodology

The quantitative data for this study, originally forming the basis for a larger-scale analysis (see Elsweiler 2019), are drawn from the *Helsinki Corpus of Older Scots* (*HCOS*) and from the *Helsinki Corpus of Scottish Correspondence* (*ScotsCorr*), covering two time periods, SC2 (1570–1640) and SC3 (1640–1700), as well as both non-private and private letters (see Table 8.1).

The letters written by Thomas Hamilton, whose requestive behaviour is analysed for stylistic variation, are partly comprised in this customdesigned corpus. For the purposes of this study, sixteen additional letters from *ScotsCorr* were added to the Hamilton sub-corpus at the basis of the qualitative analysis in Section 5. The correspondence of John Erskine, whose letters are also analysed in Section 5, is not contained in this corpus. Therefore, all twelve letters in the Erskine sub-corpus are an addition.

For the classification and analysis of request strategies, an adapted version of the annotation scheme developed for the *Cross-Cultural Speech Act Realisation Project* (*CCSARP*) (Blum-Kulka et al. 1989) was employed. In this scheme, requests are classified intro three broad categories of directness: direct requests, conventionally indirect requests and nonconventionally indirect requests. Direct requests, for example imperatives or performatives (see (1) and (2)), state the requested action explicitly. Although conventionally indirect requests, for example, the prediction statement hedged by the subjectivizer *I lippin* 'I expect' in (3) or the conditional clause in (4), do not explicitly state the illocutionary intent, due to their conventionality they can nevertheless be interpreted as requests.


Non-conventionally indirect requests, for example, hints such as (5), do not explicitly mention the requested action, which therefore must be inferred.

(5) I haue vreaten [written] offt to kno houe your man doeth plase you and to kip [keep] him and send me word bot I haue neuer hard [heard] of it as yet. (Charles Erskine of Cambuskenneth to his wife, 1643)

While indirect requests, which are further down the scale of explicitness, can generally be considered less binding, the force of a direct request may also be softened by the use of downtoners, for example, deferential adverbs such as *hwmillye* in (1) or honorific address terms such as *your grace* or *your lordship/ladyship*.

For this study, the requests were identified through close reading and then classified by realization strategy as well as within the three broad directness categories described above.

## 3 Request strategies in Early Modern Scottish correspondence and epistolary conventions

The analysis of all requests across the SC2 (1570–1640) and SC3 (1640– 1700) sub-periods reveals a generally relatively homogeneous pattern across both non-private and private letters (see Figure 8.1). Throughout the whole time period, letter-writers consistently prefer direct request strategies, which account for well over 60 % of all requests, over indirect strategies. This distribution confirms the results of previous studies on the development of directive speech acts in the history of English, which found that direct strategies prevail in the Early Modern English period (e.g. Culpeper & Archer 2008). Ability-oriented conventional requests, for example, *Can you?/Could you?*, the most common indirect realization strategy in present-day English (Aijmer 1996: 132), only gained in frequency in the course of the nineteenth century (Culpeper & Demmen 2011). Accordingly, ability-oriented requests are not found in the correspondence data, but other conventionally indirect strategies, for example, hedged prediction statements, see (6), are relatively common, albeit of much lower frequency than direct requests.

(6) My Lord, as to the cautionrie, I hoope yowr Lordship will doe all yow can to see me fred thereof as yow have writtine. (Donald McDonald of Moydart to Sir George MacKenzie, 1682)

Only in private correspondence do conventionally indirect strategies gain more currency in the course of the two periods with a rise from 25 % to 34 %, which is not reflected in non-private letters. On the contrary, indirect requests decline in frequency, whereas direct requests see an increase.

Among the direct request strategies in the correspondence data, performative requests are most widespread. This formulaic strategy proves particularly popular in non-private correspondence, where in the SC2 sub-period more than 70 % of all requests are realized as performatives and in the SC3 sub-period just under 50 %. In private letters, though not quite as pervasive, performatives account for 42 % and 46 % in the SC2 and SC3 sub-periods, respectively.

These distributional patterns suggest a degree of uniformity due to epistolary conventions, which were overall more closely observed in nonprivate than in private letters (Mack 2002: 114–16). Epistolary conventions regarding both the structure and language of letters were acquired in a variety of ways. In English and Scottish grammar schools, epistolary formulae were learned mainly through the medium of Latin by, for example, translating Latin phrases or whole letters into English and back (Mack 2002: 13, 24; Ewan 2015: 47). In private households, formal training was

Figure 8.1. Distribution of requests by directness in Scottish non-private and private letters, 1570–1640 and 1640–1700 (based on normalized frequencies/10,000 words). In mosaic plots, the width of the bars indicates the normalized frequency of the respective request strategies and the height of the bars their respective percentage share.

often provided by writing masters (Daybell 2012: 26). Beside formal tuition, writers moreover familiarized themselves with letter-writing conventions through exposure. In the Early Modern period, correspondence was only gradually becoming perceived as a private matter, so that, upon receipt, letters were regularly circulated among family members, read out aloud and copied into copybooks or manuscript miscellanies for future reference (Pallander-Collin 2010: 653; Daybell 2014: 61–63).

Yet another source of guidance for writers were formularies and letterwriting manuals, which became available in the vernacular in the course of the sixteenth century, for example, Angel Day's *The English Secretorie* (1586) or William Fulwood's *An Enemie of Idlenesse* (1568) (Austin 2007: 8f.). While formularies only included model letters, manuals contained both theoretical information and model letters covering a variety of communicative purposes. They therefore offer a good overview of epistolary norms for various communicative situations as well as writer-recipient dyads, reflecting, among others, differences in age, social status or relative power. In his introduction to the 'Epistles Petitorie', Angel Day in fact points out that petitory letters must 'be divers and variable' in view of 'mens conditions so divers, at whose handes or from whom the same are to be received' (Day 1586: 91). Factors influencing the choice of request strategy are, among others, the action or favour to be requested, the familiarity between requester and addressee and the addressee's social status. Accordingly, requesters may either be 'more bolder' or apply 'greater modestie' (Day 1586: 91). While the introductory text does not provide any concrete examples of requests, these can be found in the model letters. They suggest that requests may be realized through a variety of conventional direct and indirect strategies, for example, obligation statements with *must*, see (7), or impersonal requests with *please*, see (8).


However, in these sample letters, too, conventional performatives account for the bulk of requests. They feature in letters reflecting different contexts and social situations, for example, in an 'example of an Epistle Petitorie in a cause indifferent', see (9), or in an 'example Petitorie in the nature of reconciliatory, from a sonne to hys displeased father', see (10):


These examples moreover illustrate that the requestive force of conventional performatives may be mitigated, for example, by performative prefaces such as *I haue thought good to make thus bold to* placed in front of the performative verb in (9) or the appeal to the requestee's willingness by means of *will* in (10), which makes the request less binding. Such mitigated performatives can therefore be considered more polite (Kohnen 2002: 168–70).

The sample letters thus largely confirm the quantitative data illustrated in Figure 8.1. The variation of request and mitigation strategies across different contexts suggests, though, that Early Modern epistolary conventions allowed for individual requestive choices to fit different social situations as well as to accommodate to different audiences and to portray different social identities.

4 Audience design and speaker design

The audience design framework (Bell 1984) acknowledges the role of addressee(s) in intra-speaker variation over and above mere attention to speech on the part of the speaker. This framework elaborates on Communicative Accommodation Theory (CAT), according to which speakers accommodate their speech styles to their addressees and adapt their speech depending on the communicative situation (Giles & Powesland 1975). Speaker accommodation can take the form of convergence with the addressee's way of speaking but also of divergence from the addressee (Bell 1984: 162). Within Bell's audience design framework, convergence can, however, be conceived of in a larger sense including 'the speech (actually or believed to be) associated with audience demographic characteristics, as well as with other demographic groups, character traits, and interactional meanings upon which speakers may look favorably' (Schilling 2013: 338; see also Wolfram & Schilling-Estes 2016: 299f.).

The framework is applicable to quantifiable microvariables, for example, phonological variables, but also to 'less subtle levels of linguistic structure' (Bell 1984: 175), including discourse patterns, politeness strategies or pronoun usage (Bell 2001: 144f., 2014: 294f.; 299f.). Audience design seems particularly relevant in the context of requests, which are directive speech acts, that is, speech events in which speakers try to get hearers to perform a future act (Searle 1969: 66), and are thus inherently addresseerelated. Beside addressees, who are the direct participants in the interaction, three further audience roles are differentiated, whose effect on the speaker's stylistic variation lessens with further remove from the speaker, *viz.* auditors, overhearers, and eavesdroppers. While auditors are interlocutors who are not directly addressed but acknowledged by the speaker, overhearers are known to be present but not acknowledged, and eavesdroppers are not known to be present (Bell 1984: 159, 174). In the context of Early Modern correspondence, since letters were regularly read out aloud to family and friends, often passed on to other people or copied into manuscript miscellanies (see Section 3), all four audience types may have to be accounted for.

In its original form, audience design considered speakers to be primarily reactive to their audience, although it further comprised an initiative component, namely referee design. Referees are third persons who are not normally present in the interaction but who can cause speakers to redefine their identity through style shifts (Bell 2001: 147). Initiative stylistic variation was originally seen as secondary by Bell, but he has since acknowledged that responsive and initiative variation are 'two complementary and coexistent dimensions of style, which operate simultaneously in all speech events' (Bell 2001: 165). Moreover, newer approaches to audience design acknowledge that in order to gain the addressee's social approval, speakers may adopt a variety of linguistic strategies other than just linguistic convergence with the addressee.

The equal consideration of both responsive and initiative factors for intra-speaker variation is shared by the speaker design framework, a social constructionist approach to style-shifting. Social identity is viewed as local and dynamic, performed through stylistic choices, with 'speakers projecting different roles in different circumstances – since we are always displaying some particular type of identity' (Hernández-Campoy 2016: 149; see also Coupland 2011: 151–54). Thus, rather than passively accommodating to interactional norms, speakers are viewed as actively creating and moulding them. Yet, as Schilling cautions, 'we are all bound by structures and norms, and we cannot create meaningful styles out of nothing. We must draw on pre-existing associations between linguistic usages and social meanings' (Schilling 2013: 342; see also Bell 2014: 306f.). This is of particular interest in the context of Early Modern epistolary norms and the degree to which they determine the letter-writers' choice of request strategies. In the following section, I will explore this further by examining the influence of the addressees and of changing social roles on variation in the requestive behaviour of Thomas Hamilton, 1st Earl of Haddington and John Erskine, 18th/2nd Earl of Mar, both members of the Early Modern Scottish political elite.

## 5 Intra-writer variation in the requestive behaviour of Thomas Hamilton and John Erskine

#### *5.1 Thomas Hamilton and audience design*

Thomas Hamilton, Earl of Melrose and 1st Earl of Haddington, a lawyer by trade, was a leading figure in the Scottish political scene. He was admitted to the queen's financial council in 1593, and in 1596, he became a member of the Scottish privy council. He was part of a Scottish delegation in charge of negotiating the terms of the Anglo-Scottish Union with the English in 1604. From 1612, Hamilton held the office of Secretary of State for Scotland. After his appointment as president of the court of session in 1616, he held the two offices of state until 1626 (Goodare 2004).

The Hamilton correspondence sub-corpus considered for closer analysis comprises twenty-four non-private letters (totalling 7,882 words) written between 1600 and 1629 (see Table 8.2). Six letters, all dated to 1600, are addressed to King James VI. Seven letters were exchanged with unspecified addressees sometime after 1619.2 It is likely that at least some of them were addressed to the same person, as certain affairs are repeatedly referred to in the letters. His five letters to William Douglas, 7th Earl of Morton, a member of the prince's council and later Scottish treasurer, date

2 Although, in *ScotsCorr* these letters are dated to 1600, this date is doubtful. Thomas Hamilton signed them with *Melros* and he was only created Earl of Melrose in 1619 (Goodare 2004).


Table 8.2. Thomas Hamilton's correspondence

from 16253 and his six letters addressed to William Graham, 7th Earl of Menteith, president of the privy council, are dated to 1629.4

Most, if not all, the letters seem to have been written in Hamilton's official capacities. Accordingly, they mostly relate to state affairs but also occasionally to estate business.

Overall, across all letters, Hamilton employed direct and indirect request strategies almost equally, with a ratio of 17:19.5 While direct strategies are almost exclusively represented by performative requests,6 he drew on a handful of indirect request strategies, mainly prediction statements with *will*, conditional clauses and possibility statements with *may*. However, when differentiating between his letters to non-royal correspondents and


those addressed to King James VI, different stylistic preferences come to the fore.

## 5.1.1 Thomas Hamilton's letters to non-royal correspondents

In Hamilton's letters to the various non-royal addressees, he employs indirect request strategies for most of his requests. These total 14 instances. He varies these with performative requests, of which, however, there are only nine. Overall, Hamilton seemed to prefer conventionally indirect strategies for requests involving a comparatively high cost for the requestee (Leech 2014: 137f.), such as the return to Scotland to discuss and settle local business asked for in (11) by means of a conditional clause.

(11) If your lo [lordship's] affaires can permit your returne In so due time as I may haue sufficient and timelie securitie whic?h withowt your lo owne presence can not convenientlie be (Thomas Hamilton to the Earl of Morton, 1625)

By comparison, Hamilton's direct performative request in one of the letters to an unspecified addressee to pass on information regarding fur for his winter garments involves a lower cost for the requestee, see (12).

(12) And therfore must be so homelie [blunt] as to pray your lo to Informe sir Iames bailie what furres will be fittest for my neck and bodie. (Thomas Hamilton to an unspecified addressee, after 1619)

When distinguishing further among Hamilton's correspondence with non-royal addressees, slight stylistic differences become manifest in his letters to William Graham, 7th Earl of Menteith, a member of the king's close entourage at court in London. They display a relatively equal ratio of direct and indirect requests, namely four performatives compared to six indirect strategies. The performative requests seem to have been used interchangeably with indirect strategies in similar contexts, that is, when asking him to impart information regarding papists to the king, see (13) and (14).

(13) Since God and the king are vpon our side, they are not to be feared if we do our part which may be much confirmed by information to his Maiestie by your lordship (Thomas Hamilton to the Earl of Menteith, March 1629)

(14) as I wish that your lordship and other honorable and wise counsellours there may so informe his Maiestie of the circumstances (Thomas Hamilton to the Earl of Menteith, April 1629)

In (13) Hamilton resorted to a possibility statement by means of the modal auxiliary *may* in the passive voice. The 'action' character of the requested act is thus veiled through formal means, presenting it as an option and therefore as low cost. In (14), he chose a more explicit and therefore potentially more imposing performative strategy with *I wish*. The use of *may,* however, also softens this request by making it more tentative. Hamilton's more frequent choice of mitigated performative strategies in his correspondence with the Earl of Menteith may be due to the latter's role at the royal court, as will be explored in more detail in the following section.

## 5.1.2 Thomas Hamilton's correspondence with King James VI

Hamilton's requestive style in his letters to King James VI is markedly different from the style in his letters to non-royal addressees. He opts for direct performative strategies in ten of his 15 requests. These performatives, for example, (15), despite being explicit, are accompanied by a set of conventional formulaic deferential devices, among others the adverb *submislie,* the honorific address *your Maiestie* and the preface *to be graciouslie pleased* interposed between the performative verb *beseech* and the actual request.

(15) I submislie beseech your Maiestie to be graciouslie pleased at all efter following occasions to {del} be ple {del} make me happie by more particular direction of your most excellent wisdome (Thomas Hamilton to King James VI, 1600)

Performatives mitigated by deferential devices are a highly formal means of paying respect to the monarch. This respect also finds reflection in non-linguistic features of the letters, for example, the wide vertical space between the formal address and the body of the letter or the appropriate choice of handwriting (Daybell 2012: 85–95). Hamilton's choice of mostly deferential performatives in his letters to King James VI is therefore an accommodation to the high social status of the royal addressee and also to his style, because, in fact, James VI himself also almost exclusively issued his commands and requests using formulaic performative strategies, see (2), reproduced here as (16).

(16) quhairfoir we require and command yow that ye ressaue in your custodie and keping Johnne Grahame of Stobohill (King James VI to David Wemyss of Wemyss, 1587)

These are, however, largely unmitigated (Elsweiler 2019). Hamilton's preference for performatives in the royal letters may moreover explain why he chose performative strategies more often in his correspondence with the Earl of Menteith than in the letters to the other non-royal addressees. In view of Menteith's role as royal adviser, the king could be considered as an auditor for Hamilton's letters to him, which would also have triggered accommodation, albeit to a lesser degree (Bell 1984: 175).

#### *5.2 John Erskine and speaker design*

While writers evidently accommodate their requestive behaviour to different addressees, they may moreover vary their requestive style in their correspondence with the same addressee. This is manifest in the correspondence of John Erskine, 18th/2nd Earl of Mar, a high-ranking Scottish politician and courtier, with William Douglas, 7th Earl of Morton. John Erskine was educated alongside James VI, which led to a lasting connection with the king. His political career began in 1578, at the age of 16, when he became a privy councillor. After the Union of the Crowns in 1603, he accompanied the king to London. In 1616, he was appointed treasurer of Scotland and held this position until 1630, when he was succeeded by William Douglas, 7th Earl of Morton (Goodare 2006). John Erskine regularly corresponded with William Douglas, both during and after his treasurership (see Table 8.3).

Interestingly, John Erskine's letters, all in his own hand, manifest a change of requestive style after William Douglas had been appointed treasurer. In his letters written in his official capacity as treasurer, Erskine almost invariably employed direct request strategies, as is illustrated in (17), taken from a letter concerning a warrant signed by the king.


Table 8.3. Letters by John Erskine to William Douglas

(17) I haue only vryttin this letter to Zour lo [lordship] to aduertt vnto itt, (17a) and to desyr to see the letter sentt be the exchakker to his matie [majesty]. for thay haue many freinds thaer, (17b) keip this letter only to Zour self and mak Zour aun ouss of itt, for giue itt be nott menditt nou befoir this signator pass itt vil be harder to doo heirefter ( John Erskine to William Douglas, 1627)

In (17a), John Erskine chose a performative strategy to ask William Douglas to show the letter written by the exchequer commission to the king, which is followed by another direct request by means of two imperatives in (17b). Neither request is softened by deferential or distancing devices. Erskine does, however, provide justification for both requests, employing reason clauses with *for*. This direct requestive style is reflective of his office as treasurer.

After John Erskine's resignation from the treasurership, his letters to the new treasurer William Douglas manifest a different requestive style – a more equal balance between direct and indirect strategies. His six direct requests are all performative requests, which are mostly toned down by mitigating devices. His eight indirect requests are predominantly hedged prediction statements with *will*. The change in his requestive style is illustrated in (18), an excerpt from a letter in which John Erskine, who experienced financial difficulties in the early 1630s (Goodare 2006), begged William Douglas for debt relief.

(18) My honorable good lord I haue receued this fornoun this inclosed charge for a partt of theis detts quharof Zour lo and aine number of very honest and vorthie ('worthy') freinds ar bound to releiue me, (18a) I doutt nott bott Ze vill haue a caer that I may be putt in seurtie vithout troble […] thaer is ane number mor of this kynd to be att me shortlie and I leuk hoorelie for thaem Zour lo knaus thir detts vas takin on for his matis[majesty's] seruice (18b) I beseik Zou blaem me nott for sens itt goeth to this extremitie I man (altho sor aganst my vill,) sic my releiffe (18c) Thus assuring my self, Ze vill tak sum ordor vith this. ( John Erskine to William Douglas, 1633)

(18a) and (18c) are prediction statements hedged by the subjectivizers *I doutt not bott* and *assuring my self*, which at the same time appeal to the addressee's willingness to perform the requested action, thus making the requests less imposing. The less binding character of his requests is confirmed by the addressee-oriented, deferential semantics of the speech-act verb *beseik* 'beseech' utilized for the performative request in (18b). Thus, after resigning from his official role as treasurer, by employing more conventionally indirect and mitigated request strategies, John Erskine projected a different social role in these changed personal circumstances. His requestive behaviour across his letters indicates that he actively drew on the strategic options offered within the framework of epistolary conventions to shape both an official and a more submissive and less public persona.

## 6 Conclusion

This study explored the role of audience and speaker design in the varying styles evident in the requestive behaviour of individual Early Modern Scottish letter-writers. Since writers in the sixteenth and seventeenth century generally adhered to elaborate epistolary conventions providing for various writer-recipient dyads as well as situational differences, variation in the requestive style of individual writers was considered against the background of these formal conventions.

A quantitative analysis of the distribution of request strategies in a corpus of sixteenth and seventeenth-century Scottish non-private and private correspondence (c. 70,600 words) showed that performative requests are the most commonly employed strategy. This was corroborated by a survey of model letters in Angel Day's letter-writing manual *The English Secretorie*. Yet, Day's model letters further indicate that letter-writers had a variety of conventional request strategies at their disposal, which could be modified by a range of mitigating devices to suit various communicative and social situations. The qualitative analysis of intra-writer variation in the requestive styles of the two high-ranking Scottish politicians Thomas Hamilton, 1st Earl of Haddington and John Erskine, 18th/2nd Earl of Mar, in fact confirmed that writers drew on a repertoire of strategies to accommodate to their addressees and to shape different social roles.

Hamilton was shown to adapt his requestive style to his addressees. In his correspondence with non-royal addressees, he manifests a predilection for conventionally indirect strategies, in particular for requests incurring a high cost for the requestee. His preference for formal performative requests in his letters to King James VI, however, signals an accommodation to the royal addressee's high social status as well as his requestive style. James VI moreover may have played a role as auditor in Hamilton's correspondence with the Earl of Menteith, a close adviser to the king, in which he evinces a partial accommodation by employing performatives and indirect strategies on a relatively equal balance.

Beside audience design, the correspondence data further attest to writers taking an active part in projecting different social roles through their choice of request strategies. This is evident in Erskine's letters to the Earl of Morton both during and after his term as Scottish treasurer, in which a change of requestive style testifies to Erskine actively shaping an offical as well as a more submissive non-public persona. My analysis has thus shown that, although Early Modern letter-writers 'were bound by structures and norms' (Schilling 2013: 342), these offered a repository to draw on in order to fine-tune their requestive choices to different audiences and to actively portray themselves in changing social roles.

## References


#### Lucia Assenzi

9 Between societal constraints and linguistic selfawareness: Stylistic variation in the letters of Prince Ludwig von Anhalt-Köthen (1638–1646)

#### Abstract

The aim of the present chapter is to investigate the extent of stylistic variation in letters produced in a highly normative context, namely that of seventeenth-century German high aristocracy. The chapter will show how addressee-based intra-writer variation can occur even under the constraints of strict social and genre conventions because of – or in spite of – one's own linguistic self-awareness. To do so, the chapter analyses syntactic complexity and pragmatic elements pointing towards language formality in the correspondence of Prince Ludwig von Anhalt-Köthen from the years 1638–46. Different sociolinguistic models of stylistic variation will be scrutinized to assess whether Attention to Speech, Audience and Referee Design or Speaker Design best account for addressee-based stylistic variation in Ludwig's letters.

## 1 Introduction

## *1.1 Background and context: Letter writing in seventeenth-century German-speaking aristocracy*

Letters are concrete realizations of social practices and are thus always the product of a specific socio-cultural and historical context. It then goes without saying that letter writing may obey different rules in different cultural and historical settings. As a matter of fact, letters written by seventeenth-century German1 high aristocrats differ in many ways from our contemporary

1 As Germany did not exist as a political entity in the seventeenth century, in this chapter 'German' is used simply as a synonym for 'German-speaking'.

concept of a letter, which we normally associate with 'intimacy, spontaneity, and privacy' (Dossena & Del Lungo Camiciotti 2012a: 5).

Seventeenth-century German aristocrats were inherently public figures. For them, letters were 'hardly a medium of an individualized "I"' (Ruppel 2015: 264). Courtly letters reflected the identity of the writer as part of a dynasty or as a political actor rather than his or her personal bonds with the addressee (Furger 2010: 144). This also holds true for letters exchanged between close family members (Ruppel 2015: 262f.). Even letters discussing private matters typically lacked spontaneity, subjectivity and emotion. In fact, in seventeenth-century German aristocratic communication, the linguistic distinction between private and public letters tends to go unnoticed at first sight (Furger 2010: 137).

Since for seventeenth-century German nobility letters were a means of expression of the public rather than the private self, they had to obey the strict hierarchical rules of baroque society. *Briefsteller* [letter writing guides] were very popular in this context. They helped navigate the thorny rituals of aristocratic networks and provided orientation in the use of nobiliary titles and address forms, or in choosing the appropriate greeting and farewell formulas. As specific as *Briefsteller* were in discussing the appropriate use of titles and greetings, they did not give any explicit advice on how to formulate the letter content (see Harsdörffer 1661: part I). Instead, the aspiring letter writer had to learn by imitating the exemplary letters collected in the *Briefsteller*.

As a model for good language use ('guter Gebrauch', Harsdörffer 1661: part IV, 166), seventeenth-century *Briefsteller* listed letters written by chanceries (Furger 2010: 160). From that, we can infer the importance of the model of chancery style for letter writing in general, a style characterized by highly formulaic language and extremely complex syntax.

The lack of communicative immediacy (Koch & Oesterreicher 1985) and the expected adherence to the linguistic rituals of baroque estate-based society seem to significantly reduce the possibility of stylistic variation in the letters of seventeenth-century German high aristocrats. Still, it is inevitable that the different social positions of sender and addressee and the degree of personal closeness between the two should have produced some kind of stylistic variation (Furger 2010: 144).

#### *1.2 Prince Ludwig von Anhalt-Köthen as a letter writer*

No writer blindly complies with the prescribed norms of language use, and some of them explicitly position themselves against these norms by advocating a renovation of current linguistic practices. This was certainly the case with Prince Ludwig von Anhalt-Köthen (1579–1650).

Prince Ludwig was one of the central figures in seventeenthcentury German metalinguistic discussion. In 1617, he co-founded the *Fruchtbringende Gesellschaft* [Fruit-bearing Society], the first academy in the German-speaking area to have the improvement of the German language as one of its main goals (Ball 2020: 774). Until his death in 1650, Ludwig was at the head of the *Fruchtbringende Gesellschaft*. With and through him, the most important grammarians, translators and authors of the seventeenth century discussed the status of the German language, literature and poetics.

Prince Ludwig had a high linguistic self-awareness and actively promoted the creation of a new writing style, a more immediate, more comprehensible style that should take leave of the over-complications of chancery writing. Ludwig conceived this new style by reading and translating Italian and French literature, since there authors had already managed to bring the written language closer together with the spoken language (of the learned elite). Already in one of his first translations from 1619, Ludwig abandoned the chancery-like style of the first society letters and adopted a more immediate writing fashion that was unusual and modern for the time (Conermann 1992: \*21). From metalinguistic hints in his letters to other members of the *Fruchtbringende Gesellschaft*, we can gather that Ludwig stuck to this style later in life as well. For example, in one of his first letters to Ludwig, the baroque poet Martin Opitz used titles, salutations and formulaic expressions of pageantry as prescribed for commoners when addressing high aristocrats. The opening of Opitz's letter is reproduced in (1):

(1) Durchlauchter, hochgeborner, gnädiger fürst vndt Herr, Herr, daß E. Fürstl. Gn. die stralen ihrer gütigkeit auch hieher in diesen Mitternächtischen seehafen strecken, vndt mich dero alten diener ihres gnädigen handbrieffleins würdigen wollen, hievor habe ich mich in aller demut zue bedancken […] (Opitz to Ludwig; in DA Köthen, 4: 380402, 514) [Your highness, high-born, gracious Prince and Lord, Lord,

that Your Princely Grace resolved to extend the beams of Their goodness even to this midnight-dark seaport and to honour me, Their old servant, with a gracious letter written in Their own hand, for these things I have to thank in all humility]2

As can be seen in (2), in his reply Ludwig commented on the excessive formality of Opitz's letter and invited him to adopt a less pompous style:

(2) Jn dessen […] erinnert der Nehrende3 das hinfuro die schreiben an ihme nach der geselschaft-art, ohne sonderliche geprenge, möchten eingerichtet sein (Ludwig to Opitz; in DA Köthen, 4: 380504, 555) [Meanwhile, the Nourisher reminds that in the future the letters addressed to him should be written in the society style, without particular pomposity]

Ludwig's comment on Opitz's letter and the subsequent change in Opitz's way of addressing Ludwig confirm the existence of a society letter-writing style, namely a slightly less formal one, and one in which considerations for social hierarchy played a secondary role (Ball 2020: 776). This followed from Ludwig's general attitude towards language and social ranks. All members of the *Fruchtbringende Gesellschaft* were invited to use their society names instead of their birth names when exchanging letters or publishing books. The aim was to render invisible the hierarchical differences between members and free them from possible preoccupations or scruples deriving from their own social standing (Conermann 1985: 30).

Furthermore, Ludwig's invitation to use a less formal style anticipated the changes that would occur in the second half of the seventeenth century, in which even German *Briefsteller* started listing more immediate, less formal letters, in imitation of that French galant style that would serve as an impulse for the evolution of the genre in the eighteenth century (Furger 2010: 161). It is, however, to be noted that German translations of French *Briefsteller* did not reach the same level of immediacy as the French originals; the letters contained therein were less formal than the previously


exemplary chancery letters but still remained on the communicativedistance end of the continuum (Furger 2010: 162).

For these reasons, even if Ludwig promoted and used a less complex writing style, we should not expect him to have adopted a language of full immediacy. What we can legitimately expect is a letter writing style that is still on the distance end of the spectrum, but with a tendency to conceptional orality, understood as carefully monitored speech of the cultivated elite.

The question is now whether Ludwig used this more modern style consistently or whether – at least with some of his addressees – he still had to comply with social norms and follow the prescribed model of chancery writing. In other words, does his linguistic self-awareness always win over social norms? Or does the outcome of the conflict between these two tendencies vary depending on whom Ludwig is writing to?

#### *1.3 Stylistic variation in highly monitored written language*

Since Prince Ludwig's letters are written in a carefully monitored style, their investigation presents different challenges than the letters analysed in many historical sociolinguistic studies.

In the German-speaking context, nineteenth-century letters of semiliterates have received the most attention in recent years (Elspaß 2012; Voeste 2018). The temporal distance and the intervening evolution of the letter genre, as well as the social provenience of the writers from lower social classes, makes it difficult to compare the results of these studies with those of mine. For example, papers focusing on intra-writer variation in nineteenth-century letters of inexperienced writers, such as Schiegg (2015), show how lower-class writers used dialectal and colloquial varieties intentionally and consciously to accommodate their style, and not – as previously thought – because of their imperfect knowledge of the standard language. This is without a doubt a central finding, but intra-writer variation in Prince Ludwig's letters happens on a fully 'standard' level, in which neither dialectal nor non-standard varieties are present.4

4 It is problematic to speak of a standard German in the seventeenth century. Well into the eighteenth century, the discussion about which of the supra-regionally

Many recent studies on other languages also investigate letters from the eighteenth century onwards (see contributions in Dossena & Del Lungo Camiciotti 2012b), that is, from a period in which immediacy and traces of orality were common even in letters of the social elite (Auer 2015). No traces of informal orality are present, however, in Ludwig's letters.

The Middle English Paston letters have a similar degree of formality and had to obey similarly strict rules of social hierarchy. In their study of the Paston corpus, Hernández-Campoy and García-Vidal (2018) described the effects of responsive Addressee Design and initiative Referee Design (after Bell 2001) on the distribution of the graphemes <th> and <þ>. However, though German orthography and inflectional morphology were not yet standardized in the seventeenth century, Prince Ludwig used them very consistently. These two linguistic aspects were the most debated language topics at the time: Ludwig could not criticize other members of the *Fruchtbringende Gesellschaft* for their spelling and then be inconsistent in his own orthography.5

Since linguistic variety, register, spelling and morphology all remain quite stable in Ludwig's letters, I will look into syntactic and pragmatic elements to assess stylistic variation. Complex syntax and fixed pragmatic elements such as greeting formulas and address forms are, after all, the most salient features of chancery letters.

established written varieties should be considered as standard was ongoing, with the two main contestants being Catholic Upper German and Protestant East Central German (Havinga 2018: 24f.). In the case of Prince Ludwig, 'standard' has to be understood as the East Central German established written variety.

<sup>5</sup> For example, in a 1645 letter to Diederich von dem Werder, Ludwig criticizes the poet Johann Klaj for his spelling, which 'sich nicht fügen will' [refuses to conform] (DA Köthen, 7.1: 450505, 543f.). For this reason, Ludwig decided to send Klaj a copy of the newly published *Deutsche Rechtschreibung* [German Orthography] by Christian Gueintz.

## 2 Corpus and methodology

#### *2.1 Corpus: The letters of Prince Ludwig von Anhalt-Köthen*

In order to study the factors influencing stylistic variation in Ludwig's correspondence, a corpus of letters to eight different addressees was analysed. To limit the effects of other variables, the gender of the addressees and the topic of the letters were kept constant: all addressees were members of the (men-only) *Fruchtbringende Gesellschaft,* and all letters concern the literary, linguistic and in some case also administrative activities of this society. So that the effect of social hierarchy and familiarity on stylistic variation could be established, the eight addressees were chosen from different social classes, and had varying degrees of familiarity and closeness to Ludwig, as can be seen in Table 9.1. 6

All letters are taken from the edition of the letters and works of the *Fruchtbringende Gesellschaft* curated by Conermann's team (DA Köthen). The edition is very accurate in reporting corrections, additions and notes in the margin, and is thus reliable for linguistic analysis. However, the edition was not conceived for linguistic purposes. One first important limitation is that the edition does not comprise all letters ever written by Ludwig, but only those that provide any kind of useful information for the investigation of the *Fruchtbringende Gesellschaft*. Thus, the choice of addressees for my corpus is limited to those few society members to whom Ludwig sent five or more letters concerning more strictly the activities of the society during the years 1638–46.7 As some of Ludwig's addressees are



Table 9.1. Composition of the corpus and details of the addressees

better represented in the edition than others, the sub-corpora for each addressee differ in length.

Ludwig's nephew Christian II is the only family member whose letters were edited for this period. Ludwig was Christian's mentor, and the two had a long-lasting, close personal relationship.

Two of Ludwig's peers are also represented in the corpus: the Dukes August II and Christian Ludwig. Although their titles differed from Ludwig's, all three had ranks corresponding to that of an Imperial Prince (*Reichsfürst*). Whereas Ludwig had a friendly, though formal, relation with August II, one of the most learned noblemen of his time, he wrote to Christian Ludwig only to remind him of his promise to contribute to the printing costs of the society books.

Three of the addressees were either members of the lower nobility or commoners that were particularly active in the *Fruchtbringende Gesellschaft*. Diederich von dem Werder was one of Ludwig's confidants, a poet and a prolific translator from Italian. Carl Gustav von Hille was court master in Braunschweig. In 1647, he published *Der teutsche Palmbaum* [The German Palm Tree], the first society book of the *Fruchtbringende Gesellschaft*. Christian Gueintz, a commoner, authored the *Deutscher Sprachlehre Entwurf* [Outline for a German Grammar, 1641] and the *Deutsche Rechtschreibung* [German Ortho-graphy, 1645], the 'official' grammar and orthography of the *Fruchtbringende Gesellschaft*.

The last two addressees are commoners who discussed linguistic or poetic matters with Ludwig. Georg Philipp Harsdörffer was a member of Nuremberg's *Patrizier*, a kind of noble class of rich merchants. He was a prolific poet, translator and polymath. He corresponded and often argued with Ludwig on the subject of German grammar and metrics. The Poet Laureate Martin Opitz was a commoner, but was ennobled in 1638 for his poetic merits. His contact with Ludwig was very brief: it started in 1637 and was cut short by Opitz's death in 1639. They corresponded mostly on metrics, poetry and translation.

Through a linguistic analysis of Ludwig's letters, I will attempt to establish to what extent the social standing of the addressees and their (personal or professional) closeness to Ludwig may explain the stylistic variation in Ludwig's letters.

## *2.2 Method: Assessing stylistic variation and formality degree with syntax and pragmatics*

Though the notion of linguistic complexity has been much debated in recent years, particularly in studies on second language acquisition, there still is no standard way of measuring linguistic (or syntactic) complexity (Pallotti 2015: 117f.). However, some criteria are often mentioned as indicators of syntactic complexity, for example, the number and position of sentence components, the average sentence length and the number of subordinate clauses per sentence (Pallotti 2015: 123–25). Since Ludwig's letters are not annotated for linguistic analysis in any way, quantifying phrasal constituents would have been too time consuming. Therefore, for my analysis, I will remain on the clausal level and quantitatively assess: (i) the average number of words per sentence; (ii) the average number of clauses per sentence; (iii) the average number of main and subordinate clauses in hypotactic sentences; (iv) the degree of subordination of embedded clauses.

Sentence borders were identified by a combination of punctuation and syntax. I considered as a sentence a sequence of words ending with a full stop or a semicolon followed by a word with a capital letter. As was typical in the seventeenth century, full stops and semicolons are sometimes used to segment long text chunks, which at times leads to the separation of a syntactically non-independent element – such as a subordinate clause or an element in a list – from its matrix clause. Only in this case, the presence of full stops and semicolons was ignored in order to reunite the nonindependent element with its superordinate clause.

'Clause' is understood here as a group of words containing either a subject and a finite predicate, or a *zu*-infinitive. Groups of words introduced by a subordinating conjunction and showing an overt subject but lacking a finite auxiliary or copula verb are also considered as (subordinate) clauses (see Section 3.1: clause 6 in (4)). This phenomenon is commonly known as 'afinite construction' and was highly frequent in seventeenth-century German (Breitbarth 2005: 76–80).

The distinction between main and subordinate clauses is performed on the basis of their syntactic status. Independent clauses are main clauses, whereas subordinate clauses depend on a superordinate clause. For simplicity, the number of subordinate clauses includes finite and *zu*-infinitive clauses. Coordinated clauses sharing subject and finite verb were counted as one single clause.

After the quantitative analysis of these syntactic features, I will verify through a qualitative analysis whether the higher or lower syntactic complexity of Ludwig's letters to different addressees correlates with pragmatic factors pointing towards language formality, that is, the use of titles, address forms and of the more formal, invariable possessive pronoun *dero*.

## 3 Analysis

#### *3.1 Syntactic complexity*

From Table 9.2, we can gather the first differences in the way Ludwig writes to different groups of addressees. On average, Ludwig writes shorter sentences to his nephew Christian II and to one of his closest collaborators, von Hille. When addressing his peers, the Dukes August II and Christian Ludwig (and also when addressing the Poet Laureate Opitz), Ludwig writes longer sentences containing a higher number of clauses.

Figure 9.1 refines the analysis by only considering hypotactic sentences. The results confirm that the syntactic complexity of hypotactic sentences is higher when Ludwig addresses his two peers and lower when he writes to his other addressees.

The analysis of the subordination degree of embedded clauses in Figure 9.2 only strengthens the above argument. With his peers, Ludwig uses a greater quantity of embedded clauses of the third degree of subordination or higher. In the letters to all other addressees, more than 90 % of all dependent clauses are embedded directly in the main clause or in a clause of the first degree of subordination. Note, however, how letters to Opitz consistently come up on the high complexity end of the spectrum.


Table 9.2. Average sentence length

Figure 9.1. Length and composition of complex sentences.

Figure 9.2. Subordination degree of embedded clauses.

The next examples illustrate the difference in syntactic complexity between Ludwig's letters to his peers and those to other addressees. In (3), we have a specimen of the average complex sentence in Ludwig's letters to von Hille, that is, a sentence comprising a main clause and two embedded clauses of the first degree of subordination:

(3) <sup>1</sup> [Es wird auch der Unverdrosene gebeten,] 2-i [das er zu Hannover wolle zu wege bringen das Lenthische und Medingsche wappen,] 3-i [da sie bey dem Ertzschreine der wappen noch mangeln]. (Ludwig to von Hille; in DA Köthen, 6: 430407, 636)8

[The Indefatigable (i.e.: von Hille) is asked to procure himself Lenthe's and Meding's emblems in Hannover, since they are still missing from the emblem archive]

(4) represents the typical complex sentence in letters to August II, that is, a hypotactic sentence composed of six clauses, of which five are embedded clauses that reach the third degree of subordination.

(4) <sup>1</sup> [E.L. freundlichen begehren zufolge, schicke ich deroselben auch abschriftlich hiermitt zu, erstlich die eigene Nahmen der zweyhundert gesellschaffter, 2-i [deren gemählde, wort und acht Reime ihr jüngsten in Kupffer gestochen und gedrucket seind übersendet worden.] Wie dan auch der übrigen folgenden verzeichnus, ohne die gemählde und Reime, 3-i [welche die Jehnigen seind, 4-ii[die noch zu den ersten kommen müssen,] 5-ii[und man beyderley, wieder aufs neue, 6-iii[weill keine gedruckte mehr verhanden] wird zu verfertigen haben]]. (Ludwig to August II; in DA Köthen, 5: 391217, 380)

[Moreover, as per *Euer Liebden's* (Your Highness's) request, I enclose a copy of, firstly, the names of the two hundred society members whose emblems, mottos and poems, engraved in copper and printed, have been sent to You not long ago, and furthermore the list of the members who joined later, but without any emblems or poems, which are those who must be added to the first ones, and which both, since there are no more printed copies, will have to be published anew]

The distribution of simple sentences, paratactic and hypotactic sentences of various degrees of complexity only corroborates what has been stated above. In letters to his collaborators of lower social standing and to his nephew, Ludwig makes ample use of simple and paratactic sentences. As can be seen in Figure 9.3, nearly half of the sentences in Ludwig's letters to Christian II are either simple or paratactic sentences, or hypotactic sentences with only one main and one subordinate clause.9


Figure 9.3. Sentence distribution and complexity in Ludwig's letters to Christian II.

On the other hand, letters to peers contain nearly no paratactic sentences and only a few simple sentences, as can be gathered from the sentence distribution in letters to August II in Figure 9.4. In letters to his peers, Ludwig uses the whole range of syntactic complexity. It is no coincidence that the longest and most complex sentence in the whole corpus occurs in a letter to August II.

#### *3.2 Pragmatic elements pointing towards language formality*

The syntactic analysis in Section 3.1 showed that Ludwig consistently uses a more complex syntax when writing to his two aristocratic peers (and, to a lesser extent, to Martin Opitz) and shorter, less convoluted sentences when writing to his other addressees. For reasons of space, I will not be able to go into the stylistic variation of the letters Ludwig sent to each addressee in particular. Instead, I will now show how syntactic complexity in letters to the two main groups of addressees correlates with pragmatic features associated with a higher degree of linguistic formality.

Figure 9.4. Sentence distribution and complexity in Ludwig's letters to August II.

The two groups of addressees established on the basis of syntactic complexity in Section 3.1 are confirmed by the use of the more formal, invariable possessive pronoun *dero*. Ludwig uses *dero* exclusively when the possessor he is referring to is one of his two peers, August II or Christian Ludwig (5).10 For all other addressees, he uses the third person singular masculine possessive *sein* [his] (6).

(5) El. […] die ich nechst dero geliebten gemahlin und furstlichen jugend in den schutz gotlicher almacht befele (Ludwig to August II; in DA Köthen, 6: 421031, 500)

[*Euer Liebden* (Your Highness), whom I commend to the protection of God's almightiness together with Their beloved wife and princely children]

(6) Jm ubrigen wird der Spielende […] seine geschickligkeit ferner sehen lasen (Ludwig to Harsdörffer; in DA Köthen, 6: 420503, 433) [Moreover, the Player will show his adroitness further]

The use of titles, greetings and address forms also follows the composition of two different groups of addressees. (7) provides an example of the typical opening of letters Ludwig sent to commoners and members

<sup>10</sup> Note that Opitz also uses *dero* when referring to Ludwig in (1).

of the lower nobility. With them, Ludwig does not use titles or greeting formulas. Furthermore, Ludwig refers to these addressees by their society name and in the third person singular masculine *Er* [He], as can be seen in (8).


On the contrary, Ludwig refers to his peers in the third person plural *Sie* [They] and by *Euer Liebden*, a less formal version of 'Your Highness', which was used between aristocrats of comparable rank. He moreover opens the letters to the two Dukes with the appropriate titles, greetings and service offerings, as shown in (9):

(9) Hochgeborner furst, freundlicher vielgeliebter h. Vetter11, El. entbiete ich meine freundwillige dienste, und werden sie sich sonder Zweiffel fr. erinnern, das sie […] (Ludwig to Christian Ludwig; in DA Köthen, 7.2: 450923, 694) [High-born Prince, kind, much-loved Lord *Vetter*, to *Euer Liebden* I offer my eager services, and without a doubt They will kindly remember that They […]]

There are, however, two exceptions to this pattern. The first one is von Hille, from the first group of addressees. He is referred to in the third person singular and by his society name, but letters to him begin with the title 'Edler und Vester lieber besonder' [Noble and firm, especially dear] (in DA Köthen, 6: 430312, 623). The reason why Ludwig uses titles with von Hille and not with von dem Werder, the other member of the lower nobility from the first addressees' group, may reside in the fact that Ludwig had a closer relationship with von dem Werder, who had been an active member and collaborator of the *Fruchtbringende Gesellschaft* since 1620. Von Hille joined far later, in 1636, and his first epistolary exchanges

<sup>11</sup> In the seventeenth century, *Vetter* was used to indicate a male relative. Christian Ludwig was the great-grandson of Elisabeth von Anhalt, one of Ludwig's older half-sisters.

with Ludwig date back to 1642. This could indicate that the degree of closeness between Ludwig and his addressees was probably more significant than the mere observation of social hierarchy in determining the use of titles and other pragmatic elements pointing towards linguistic formality.

The second exception is Ludwig's nephew, Christian II. For reasons I was not able to determine and that do not seem to correlate with the topic of the letter, when writing to Christian II Ludwig sometimes follows the first pattern (no greeting, no titles, society name, third person singular), sometimes the second pattern reserved to peers (greeting, title, *Euer Liebden*, third person plural). Since the correspondence between Ludwig and Christian II spans from before 1617 up to Ludwig's death in 1650, it would be interesting to analyse it separately to better understand the dynamics of their relationship and the consequences it had on stylistic variation in Ludwig's letters.

#### *3.3 Explaining stylistic variation in Ludwig's letters*

Having assessed the extent of stylistic variation in Ludwig's letters to different addressees, it is now time to reflect on which sociolinguistic model of intra-writer variation best accounts for it.

One of the first explanations for intra-writer variation is Labov's Attention to Speech (Labov 1966). However, Labov himself later stated that his Attention to Speech model was never intended as a general description of how style-shifting works, but rather aimed at explaining the intra-speaker variation observed in his famous department store study (Labov 2001: 87). Labov's model is indeed not relevant for the analysed corpus. As was to be expected from the premises in Sections 1.1 and 1.2, attention to writing is equally high in all of Ludwig's letters. The fact that Ludwig wrote in a less complex fashion to some addressees does not imply that he wrote less carefully to them, as can be seen by the consistency of his spelling and morphology.

What we have observed time and again is that Ludwig writes differently to different addressees. He himself advocated for a more modern, less complex writing style, which he, however, used only with his nephew and with addressees from lower social classes. With his peers, he still complied with the complex rituals of chancery-style-based formulaic communication.

Addressee-based style-shifting can be explained in terms of Bell's Audience and Referee Design (Bell 2001). Following Bell's model, Audience Design would explain why Ludwig accommodates to the expectations and the language use of his peers and thus adheres to the chancery style that was common in communication amongst high aristocrats. With addressees of lower social classes and with relatives, he actuates Referee Design, that is, he distances himself from the prescriptions of aristocratic epistolary communication and initiates style-shifting towards the more modern, more immediate style he himself strove to actualize through his translations, metalinguistic reflection and influence on the works of other members of the *Fruchtbringende Gesellschaft*.

One of the criticisms that has often been directed at Bell's Audience Design, especially in its first formulation (Bell 1984), is that it may give the impression that accommodative style-shifting happens as an automatic response to the expectations of the addressees. Knowing the extent of Ludwig's linguistic self-awareness, however, it is far more plausible that he deliberately chose to accommodate to the linguistic use of his peers. For this reason, Coupland's Speaker Design seems to better explain Ludwig's addressee-based style shifting. Coupland's model emphasizes the role of style in projecting and defining social identities and relationships (Coupland 2001: 186). From this perspective, even accommodative style-shifting is a creative act in which speakers (or writers) actively 'opt to operate communicatively within normative bounds' (Coupland 2001: 200) and choose to project a self-identity which is consistent with the interlocutor's expectations (Coupland 2001: 201).

## 4 Conclusion

The above analysis of Prince Ludwig's letters to eight different addressees has shown the presence of two main addressee groups. The first group comprises commoners, lower-ranking nobles and one of Ludwig's relatives. With these addressees, Ludwig uses a less complex syntax and a less formal (and formulaic) language. He refers to this group of addressees in the third-person singular masculine and by their society name and does not use titles or greetings in the letter opening. Here, Ludwig can use the more modern, simpler style he advocated for members of the *Fruchtbringende Gesellschaft*. With the second group of addressees, comprising his two peers, the Dukes August II and Christian Ludwig, Ludwig conforms to the rules of chancery style prescribed by *Briefsteller* and uses a more complex syntax, titles and formulaic expressions.

Coupland's Speaker Design theory has proved to provide a convincing model for explaining accommodative and divergent style-shifting in Ludwig's letters as a self-aware and creative means of projecting different identities to different addressees: an aristocratic, princely persona with his aristocratic peers, and a linguistically highly self-aware, 'head of the *Fruchtbringende Gesellschaft*' persona with his other addressees.

## References

#### *Primary sources*

DA Köthen = Conermann, Klaus (ed.). 1992–2019. *Briefe der Fruchtbringenden Gesellschaft und Beilagen: Die Zeit Fürst Ludwigs von Anhalt-Köthen 1617– 1650. 9 vols*. Tübingen: Niemeyer/Berlin: De Gruyter.

Harsdörffer, Georg P. 1661. *Der teutsche Secretarius*. Nürnberg: Endter.

#### *Secondary sources*

Auer, Anita. 2015. Stylistic Variation. In: Anita Auer, Daniel Schreier & Richard J. Watts (eds), *Letter Writing and Language Change*. Cambridge: Cambridge University Press. 133–55.


#### Katharina Gunkler-Frank

# 10 Intra-writer variation in clitics in German patient letters from the nineteenth and the early twentieth century

#### Abstract

In this chapter, I will analyse the use of clitics in a corpus of 192 German letters by psychiatric patients from the nineteenth and twentieth century. The most frequent forms are preposition article clitics, but article clitics with non-prepositional hosts and pronominal clitics are also attested. Especially the use of these less frequent forms is influenced by writing experience, gender, region and time. Besides this inter-writer variation, there is intra-writer variation to be found as well. A pattern of audience design is observable, as some forms are more frequent in private than in official letters. A detailed analysis of the functions of clitics in selected texts reveals that they are used to perform aspects of the writer's identity and to form the relationship with the addressee.

## 1 Introduction

In German, clitics are primarily oral features that may also appear in written language, but their frequency and distribution differs in the two media. Werth (2020: 179) argues that this makes clitics suitable features for investigating the historical language of immediacy and distance (Koch & Oesterreicher 1985). Analysing intra-writer variation of clitics can therefore help to uncover how and why individuals vary between forms located in different places in this continuum. We can investigate if this variation is related to social factors and if influential theories of stylistic variation that were originally developed for present-day language are also applicable to clitics in historical data. In his audience design approach, Bell (1984) claims that speakers adapt their language to their audience. To test this approach for historical clitics, I will investigate if their use is influenced by the degree of immediacy and distance between writer and addressee. Another explanation for intra-writer variation that I will examine for clitics is that linguistic features are used to construct social identities (Coupland 2007). This analysis of intra-writer variation is complemented by one of inter-writer variation to gain a better understanding of the status of clitics in the speech community and the connection between the types of variation.

Section 2 provides a short overview of clitics in German, and on previous research on this topic, followed by a description of the corpus of patient letters and the methods used in this study in Section 3. After this, I will present the results of my analysis, starting with an overview of the types of clitics used in the corpus in Section 4 and the results on inter-writer variation in Section 5. Regarding intra-writer variation, I will start with a comparison of private and official letters (Section 6), and will then proceed to the use of clitic and full variants (Section 7), and the communicative functions of clitics in selected texts (Section 8). In Section 9, I will sum up the results.

## 2 Clitics in German

Following Nübling's (1992: 1, 5–7) definition, a clitic is an unstressed word leaning on another word, the host. Like words, at least some types of clitics are syntactically distributed, but like affixes they are phonologically dependent. In German, different types of clitics with different degrees of grammaticalization exist. Quite prominent in research on German clitics are article clitics with prepositional hosts (*preposition article clitics*). Since their degree of grammaticalization varies in present-day German, the different types of forms are often referred to as steps of a grammaticalization staircase, as shown in Figure 10.1.

Some preposition article clitics like *im* ('in dem' [in the-Dat-Sg-Masc/Neut]) have reached the stage of special clitics.1 This means that they are in complementary distributions with their full forms.2 For these special clitics, the clitic variant is the dominant one in the written standard language. Some forms are described as standing on the edge of this stage (e.g. *ins* 'in das' [in the-Acc-Sg-Neut]). Simple clitics like *fürs* ('für das' [for the-Acc-Sg-Neut]) are also used in written language, but they are always exchangeable with their full forms. Forms like *nachm* ('nach dem' [after the-Dat-Sg-Masc/Neut]) are exclusively used in spoken language, especially in rapid speech and dialects. Some forms, for example, *bein* ('bei den' [with/near the-Dat-Pl]), are ungrammatical in contemporary standard German3 (description of the grammaticalization stages cf. Christiansen 2012: 1–3, 2016: 2f.).

The assignment of forms to certain stages and the definition of the stages itself differ between authors (e.g. Nübling 1992: 188f.; Christiansen 2012: 2; Steffens 2010: 247). Nübling (2005: 123) has also suggested describing the differences between the forms as a continuum rather than a staircase with delimitable stages. Nevertheless, the existence of varying degrees of grammaticalization shows that preposition article clitics are subject to ongoing language change (Nübling 2005: 106).

Previous studies on these forms mainly focus on present-day German, especially on different grammaticalization stages and on the influence of phonological, morphological or semantic factors on the use of clitic or full


Figure 10.1. Grammaticalization stages of preposition article clitics (Christiansen 2016: 3).

form (e.g. Nübling 1992, 1998, 2005; Augustin 2018; Schaub 1979). There is also historical research (e.g. Christiansen 2012, 2016; Werth 2020; Steffens 2010; Waldenberger 2009: 56–66, 2020), but this primarily describes the development of preposition article clitics up to Early New High German, as the most frequent forms in present-day German are attested in this period and already used in similar contexts, although not with the same frequency as today (Christiansen 2016: 104–11).

Besides preposition article clitics, article clitics with non-prepositional hosts (e.g. with a noun as host: *S'Fenster* 'das Fenster' [the window]) and pronominal clitics (e.g. *gehts* 'geht es' [goes it]) also exist in German. They are mainly observed in dialects (cf. e.g. Weiß 2016; Kolmer 2012; Nübling 1992), especially in southern German dialects (Abraham & Wiegel 1993: 20). Studies on their historical development (Werth 2020; Somers Wicka 2009; Elspaß 2005: 449–51) or their use in supra-regional varieties (Salomonsson 2011) are, however, rare (cf. also Werth 2020: 185).

Given the current state of research, this study does not only use clitics as a test case to investigate intra-writer variation. The analysis may also help to fill some gaps in the research on clitics, especially concerning the use of clitics in the nineteenth and the early twentieth century, as well as the influence of social factors.

## 3 Corpus and methods

The data for this study was taken from the *Corpus of Patient Documents* (CoPaDocs) that is being compiled by a research group at Friedrich-Alexander-Universität Erlangen-Nürnberg. It consists of texts – mainly letters – written by patients of psychiatric hospitals and their relatives in the second half of the nineteenth and the early twentieth century in different German regions. Due to the practice of censorship, some of the patients' texts were not sent to the addressees but kept in the patient files so that they are still available in the archives today (Schiegg 2015). As several letters by individual writers have often survived together with detailed metadata, this material is useful for my research aims as it allows the investigation of intra- and inter-writer variation.

For my analysis, I compiled a corpus of 192 letters (1851–1957) from the CoPaDocs which were written by 39 writers. I only included letters written by patients. The intention was to create a corpus that is suitable to investigate the influence on the use of clitics of the factors region, time, writing experience and gender. Therefore, I selected letters written in Kaufbeuren in southern Germany and in Hamburg in the north of Germany, as well as letters from the second half of the nineteenth century (with writers born before 1850) and from the early twentieth century (with writers born after 1850). I included texts written by experienced and inexperienced writers4 and by women and men. While the amount of text for the two genders

4 I used information about the writers' profession from the patient files to determine their writing experience. For women this can be problematic when the files only contain information about their husbands' or fathers' professions. Nevertheless, this information seems to be an indicator of the education that a person received and her exposure to written language.

and the two time periods is approximately equal, there are more texts from Kaufbeuren and more written by inexperienced writers than for the complementary categories. The reason for this distribution is that texts by inexperienced writers and from the south are more promising for the use of clitics. The complementary categories are also less frequent in the CoPaDocs. For Hamburg only texts written by inexperienced writers in the twentieth century were available. Apart from that, I selected at least three writers for each combination of categories. Wherever possible, I included three official and three private letters from each writer to make the corpus useful for the investigation of intra-writer variation. Table 10.1 gives an overview of the structure of the corpus.


Table 10.1. Corpus of this study

I manually annotated all clitics in the material and checked their transcription against the facsimiles. Thus, I was able to gain an overview of the clitic forms used in the corpus, and their frequencies. After that, I compared the frequencies of the different types of clitics for different groups of writers, for example, women vs men, to investigate inter-writer variation based on the four writer variables introduced above. To check if the use of clitics is influenced by audience design (Bell 1984), I compared the frequencies of clitics in private and official letters. These two types of letters represent different groups of addressees who do not necessarily differ in social class, but rather in social role and relationship to the writer. 'Private' vs 'official' is a rather rough classification of addressees, as these groups contain addressees with relationships to the writer that differ in their degree of closeness. It is nevertheless suitable for a general impression of the addressee-dependent use of clitics, as it is more robust with regard to individual differences than finer differentiations.

In addition to these quantitative analyses based on the whole corpus, I selected a small number of writers for a more detailed analysis. First, I annotated full variants and compared their frequencies to those of the respective clitics. Then, I qualitatively analysed in which contexts clitics are used and which communicative functions they could serve there. In doing so, I investigated if these forms are used to construct social identities (Coupland 2007).

## 4 Overview: Clitics in the corpus

In total, there are 984 clitics in the corpus, which is a relative frequency of approximately 1.3 clitics per 100 tokens. With 868 attestations, the majority (88.2 %) of these forms are preposition article clitics like *im* [in the-Dat-Sg-Masc/Neut] (see Figure 10.2).

Their high frequency fits with the assumption in the research literature that some of the forms are already strongly grammaticalized in the nineteenth and twentieth centuries. Indeed, most of the preposition article

Figure 10.2. Absolute frequencies of different types of clitics in the corpus.

clitics in the corpus are special clitics in present-day German (see ranks 1–5 and rank 7 in Table 10.2).


Table 10.2. Forms and absolute frequencies of preposition article clitics in the corpus

Of the forms on the edge of the stage of a special clitic in presentday German, *ins* [in the-Acc-Sg-Neut] is the most frequent form in the corpus (6th frequency rank). The other forms of this type (*ans* [to the-Acc-Sg-Neut] and *aufs* [on/to/for the-Acc-Sg-Neut]) are as rare as the less grammaticalized forms *zun* [to the-Dat-Pl],5 *übern* [over the-Acc-Sg-Masc], *durchs* [through the-Acc-Sg-Neut], *nachm* [according to the-Dat-Sg-Masc] and *ums* [around the-Acc-Sg-Neut]. They are either used in colloquial contexts (see (1) in a private letter and together with the colloquial *hauen* [beat] and the regional feature of the use of the accusative case instead of the dative case in *den Peitschenstiehl*) or in idiomatic expressions (see (2)). The latter are typical contexts for forms otherwise not used in written language (Nübling 2005: 110).

(1) Meine Mutter konnte sich aber niht beruhigen und haut mih mit den Peitschenstiehl immer übern Körper, (Walter A. (ham-16524), carpenter, letter to friend, 1926)

[My mother could, however, not calm down and always beats me with the whip over the body (gloss.: over-the body)]6

	- [I was terribly heavy-hearted (gloss.: it was me terribly heavy around-the heart)]

Excluding *ins*, there are only seventeen attestations for forms below the status of a special clitic. This is only 2 % of all attestations of preposition article clitics so that observations made for preposition article clitics in the analysis hold for the grammaticalized forms.

With eight attestations (0.8 % of all clitics), article clitics with nonprepositional host are also rare in the corpus. This diverse group comprises enclitic (*ischt Pflegrin* [is the nurse]7 ) and proclitic (*d'Sach* [the thing])


definite articles as well as enclitic indefinite articles in the fixed form *son* [such a]. There are also two types of pronominal clitics in the corpus. With ninety-four attestations, clitics of the pronoun *es* [it] like in *gehts* [goes it] or *wenns* [if it] are the most frequent ones (9.6 % of all clitics), but thirteen clitic variants of *sie* [she-Nom/-Acc / they-Nom/-Acc / you-formal-Sg-Nom/-Acc/-Pl-Nom/-Acc] like in *könnens* [can you] can also be found (1.3 % of all clitics).8 Their comparatively low frequency suggests that these forms are not part of the standard language, which is reflected in the dialect focus of the literature on these forms. An analysis of interwriter variation will provide further insights on their usage.

## 5 Inter-writer variation

Figure 10.3 shows the frequencies of clitics per 100 tokens according to the different factors for inter-writer variation.

Preposition article clitics are the most frequent clitics in both the Hamburg and Kaufbeuren data (though even slightly more frequent in the latter), which supports the hypothesis of their advanced grammaticalization. For all other types of clitics, however, we can observe clear differences between the texts from Kaufbeuren and Hamburg. While there are no attestations of *sie*-clitics and only one attestation of an *es*-clitic in the texts from Hamburg, we find thirteen *sie*-clitics and even ninety-three *es*-clitics in Kaufbeuren. Hence, pronominal clitics are more common in the South, but the regional restriction seems less strong for *es*- than for *sie*-clitics. This is in line with Werth's (2020: 200) findings about the regional distribution of pronominal clitics in Early New High German. In his corpus, *es*-clitics are attested in texts from different regions of Germany, but more often in the South, whereas *sie*-clitics only appear in the South. Although there are four attestations of article clitics with non-prepositional hosts in each

8 As the clitics for *es* and *sie* are homophonous (/s/), for one pronominal clitic it could not be decided to which full form it belongs. Hence, it was excluded from the further analysis ('others' in Figure 10.2).

Figure 10.3. Inter-writer variation in the use of clitics (forms per 100 tokens).

regional group in my material, we can observe a difference for this clitic type, too. In Hamburg these are only indefinite article clitics in the form *son* [such a]9 whereas in Kaufbeuren we find different definite article clitics.10

There are also differences between the two time periods: *es*-clitics are more frequent in the twentieth century (68 attestations, 0.19 per 100 tokens, compared to 26 attestations, 0.07 per 100 tokens in the nineteenth century) and article clitics with non-prepositional hosts only appear in this later period. Therefore, there seems to be a tendency to use non-standard forms more often in the later letters. *Sie*-clitics are, however, not affected by this tendency, as they are equally frequent in both periods (0.02 per 100 tokens: 7 in the nineteenth, 6 in the twentieth century). Preposition article

9 In present-day colloquial German, at least singular forms are described as being used supraregionally (Eichhorn 2015: 383) although probably more so in northern Germany (Hole & Klumpp 2000: 235).

10 This fits with the description of Low German definite articles by Lindow et al. (1998: 151f.) that does not include clitic forms except in preposition article clitics and with the assumption that at least non-prepositional *die* [the-Nom-Sg-Fem/- Nom-Pl/-Acc-Sg-Fem/-Acc-Pl]-clitics are only used in Upper German (Paul 1916: 245).

clitics are the most frequent type of clitics in both periods, but their relative frequency increases over time (1 per 100 tokens in the nineteenth, 1.31 in the twentieth century). This could suggest that the grammaticalization process even of the most frequent forms was not yet fully completed by the nineteenth century.

Both experienced and inexperienced writers make frequent use of preposition article clitics. That the relative frequency of these forms is even slightly higher in the texts of experienced writers (1.17 per 100 tokens, compared to 1.13 for inexperienced writers) shows that the grammaticalization of these forms was already advanced in the period of investigation and the forms were already established in written language.11 Nevertheless the diachronic difference suggests that the grammaticalization process is still proceeding further by increasing the mandatoriness of these forms in certain contexts and/or their contexts of use. In contrast, pronominal clitics and article clitics with non-prepositional hosts are used more often by inexperienced writers. This underlines that they rather belong to orality than to the written language. The stronger use of these non-standard forms results in a larger inventory of clitic forms for inexperienced writers, which gives them more possibilities of variation. It is, however, possible that these forms are also part of experienced writers' spoken language so that their stylistic choices would differ more clearly between spoken and written language. As the spoken language of these writers is no longer accessible, this cannot be analysed. Nevertheless, the data shows that all types of clitics can be found in texts by experienced writers as well. Writing experience also influences the use of apostrophes with clitics. This graphic marker is rather rare in the corpus (2.54 % of all clitics). Figure 10.4 shows examples with and without apostrophe. It is more frequently used by experienced writers (for 5.08 % of the clitics, compared to 1.11 % clitics with apostrophes by inexperienced writers), which reflects their awareness of the respective full forms.

<sup>11</sup> The higher frequency of preposition article clitics in Kaufbeuren (see above) could also point in this direction, as experienced writers are only included in this regional group.

Figure 10.4. *wie's* [as they] (Georg B. (kfb-966), 29 April 1884) vs *wies* [as it] (Magdalena R. (kfb-2950), July 1936).

The greatest difference in the frequency of preposition article clitics is to be observed between the two genders: Men use 1.34 of these forms per 100 tokens, while women only use 0.96. In contrast, female writers use both types of pronominal clitics more often than men (60 *es*-clitics by women vs 34 by men, 9 *sie*-clitics by women vs 4 by men). Thus, women use non-standard forms more often, whereas men use more of the standard forms. This parallel between the distribution for writing experience and gender suggests that differences between women and men originate from a difference in writing experience between the two groups. This could be a result of less access to higher education and professions that comprise writing tasks for women at that time. It is also possible that gender roles influence the use of clitics more directly, for example, by allowing or expecting women to write in a more informal and emotional way. The number of article clitics with non-prepositional hosts is equally low for men and women, but the four attestations by men are all from one writer from Hamburg with the indefinite article clitic in *son* [such a]. In contrast, the attestations by women are different types of definite article clitics, which are only used in the South by three writers.

This analysis of inter-writer variation has shown that the factors region, time, writing experience and gender influence the frequencies of usage of the different types of clitics. While preposition article clitics are widespread in all groups, article clitics with non-prepositional hosts and pronominal clitics are more frequent among inexperienced writers and women as well as in letters written in the South and in the twentieth century. This indicates that these forms are part of the language of immediacy. Therefore, the next step of the analysis will be to investigate if clitics also show intrawriter variation.

## 6 Intra-writer variation I: Audience design

In this section, I will investigate whether writers differ in their use of clitics when writing to different addressees. Therefore, I will compare the frequencies of different types of clitics in private and official letters. Figure 10.5 illustrates the results.

Figure 10.5. Frequencies of different types of clitics in private and official letters (forms per 100 tokens).

The difference in the frequency of preposition article clitics between the two groups of addressees is small: 1.16 in private and 1.13 in official letters per 100 tokens. This suggests that these forms are rather unmarked and may be used irrespective of the communicative context, which supports the results for inter-writer variation that preposition article clitics are frequently used by writers with different sociodemographic characteristics. The comparison of official and private letters also shows that preposition article clitics are not preferred in more formal contexts, as could be suggested by their slightly stronger use by experienced writers and men.

The greatest difference between private and official letters is to be observed for *es*-clitics. With sixty-five attestations (0.15 per 100 tokens), they occur more frequently in private than in official letters (twenty-nine attestations, 0.09 per 100 tokens). Clitic *es* seems to be connected to informal contexts, which fits with its regional restrictions12 and stronger use by inexperienced writers. Hence, for this clitic type the writers in the corpus show the ability to mark the closer relationship to addressees of private letters by using the *es*-clitic as a feature of the language of immediacy more often. It is, however, important to note that *es*-clitics are not exclusively used in private letters but can be found in official letters as well. An explanation for this finding could be that not all the writers are able to make a consistent distinction between contexts in which the use of this form would be considered appropriate and in which it would not, that is, restrictions in the writers' ability to vary by addressee. Another reason for the use of *es*-clitics in official letters could be that due to the non-homogeneity of addressees of official letters, there are also comparatively informal official letters, in which *es*-clitics could well be expected. This shows that a more detailed analysis of the contexts in which clitics are used is necessary (see Section 8).

Unlike for *es*-clitics, the relative frequencies for article clitics with non-prepositional hosts (0.01 per 100 tokens) and *sie*-clitics (0.02 per 100 tokens) are equal in private and official letters. This is unexpected, as these types seem to belong to the language of immediacy given the regional restrictions, the stronger use by inexperienced writers and their generally low frequency in the corpus. This low number of attestations could be the reason why a tendency to use these forms more often in private letters is not visible here although it might be in a larger corpus. The absolute numbers are, however, slightly higher in private letters for both types (*sie*clitics: 8 vs 5, article clitics with non-prepositional hosts: 5 vs 3). Based on a different method (comparison of speech representations with passages not linked to orality), Werth (2020: 204f.) finds a clearer distribution for *sie*-clitics for Early New High German, as he only observes them in contexts of conceptual orality. Similar to my results (see preceding paragraph), *es*clitics show a tendency to oral contexts in his study as well. Therefore, he considers *sie*- and to some extent also *es*-clitics as characteristics of orality.

<sup>12</sup> Cf. Koch & Oesterreicher (2007: 356) on the connection of regional variation and variation along the continuum of immediacy and distance.

## 7 Intra-writer variation II: Clitic and full variants

After this general analysis of clitics in different text types, I will now turn to a detailed analysis of three writers' use of clitics: the servant Magdalena R. (kfb-2950), the accountant Georg B. (kfb-966) – both from the hospital in Kaufbeuren –, and the smith August A. (ham-19976) from the hospital in Hamburg.13 This allows us to compare their use of clitics with their use of the respective full form. I will focus on clitic and non-clitic *es* and *sie* only, because the use of preposition article clitics by the three writers seems to be conditioned predominantly by language internal factors, as most of the preposition article clitics occur in contexts typical for these forms (see Section 2), for example, in the time specification in (3).

(3) am Sontag mittag; erwartete ich Besuch; (Magdalena R. (kfb- ̄ 2950), servant, letter to relatives, 24 May 1936) [at (gloss: at-the) Sunday noon, I expected visitors]

Furthermore, non-clitic combinations of preposition and article can often be explained by phonological or morphological restrictions for clitics, for example, *mit* [with] is not a host for clitic articles in the corpus because of the low sonority of its coda, a clitic of *die* [the-Acc-Sg-Fem/-Acc-Pl] is not attested in the corpus as feminine articles and plural articles usually do not cliticize (Nübling 2005: 117–19). This again underlines the advanced grammaticalization of preposition article clitics so that writers cannot choose freely between clitic and full variant here. Articles that do not follow a preposition were not counted here because they are highly frequent and their clitic variant is rare. Figure 10.6 shows the frequencies of clitic and non-clitic *es* and *sie* in the analysed texts. Only full forms in syntactic positions where pronominal clitics are found in the corpus (i.e. close to the left bracket) were counted. Full forms in other syntactic

<sup>13</sup> These three writers were selected for the detailed analysis because they re-present the different factors of inter-writer variation (see Section 5) and use different types of clitics frequently.

Figure 10.6. Absolute frequencies of clitic and full form *es* and *sie* close to the left bracket in the texts of three writers.

positions – mainly in sentence-initial position – were excluded, as it is not clear whether the clitic could be used there.14

All three writers employ the clitic and the full form of *es*. The two writers from Kaufbeuren use *sie* in both the full and the clitic variant. The writer from Hamburg only uses the full form *sie*. The results for regional variation in Section 5 have shown that clitic *sie* is not used by any of the writers from Hamburg. Due to these regional restrictions, the form is not part of August A.'s linguistic repertoire.15 In all other cases, both the clitic and the full variant are part of the writers' linguistic competence. Sometimes writers even use both variants of the same form (e.g. Georg B. *wenn es* and *wenn's* [if it]). This confirms the results from Section 6 that there is intrawriter variation of clitics in the corpus. It also shows that the use of clitics by these three writers is not a result of an inability to produce the full, standard variant. To gain a better understanding of why they use clitics instead, we will have a closer look at the contexts of clitics in the next section.


## 8 Intra-writer variation III: Communicative functions

The comparison of private and official letters has already shown that writers' adjustment of their language to the addressee may be a reason for a stronger use of clitics. The analysis of the contexts of clitics confirms that one of the main functions of clitics in the analysed texts is constructing the relationship between writer and addressee. An example of this function is the only pronominal clitic in the texts from Hamburg (see (4)).

(4) Aber wie leicht, treten in farmsen, Veränderungen ein? Stimmt's? (August A. (ham-19976), smith, letter to nurse, 4 December 1933) [But how easily do changes occur in Farmsen, don't they (gloss.: Is true-it)?]

With *Stimmt's* the writer asks the addressee to identify what he said before as knowledge shared between them. This does not only aim for agreement. The writer also suggests that himself and the addressee have common knowledge and experience so that he constructs a connection between them and asks for identification. This function of the whole expression is supported by the clitic as a form of the language of immediacy. In her study on present-day online communication, Salomonsson (2011) observes similar functions for clitics and finds high frequencies of clitics in phrases with this function.

Another example of clitics used to build the relationship with the addressee can be found in another letter from Hamburg by Walter A. (see (5)).

(5) Aber keinen sagen, sonst Schluß mit uns. […] und son kleines bischen rund war ichr Geschlechtsteil (Walter A. (ham-16524), carpenter, letter to friend, 1926) [But don't tell anyone, otherwise our friendship is over. […] and her genitals were a little bit (gloss.: such-a little) round]

Here, he explicitly says that the information given should not be told to anyone. By using the indefinite article clitic in *son* as a colloquial form that is typically used in an intimate conversation, he marks this letter as intimate. Hence, the use of this intimate form strengthens his request for confidentiality.

Magdalena R. uses clitics to construct her relationship to the addressee as well. In her letter to the mayor of her hometown, she code-switches into the dialect, as in (6).

(6) Schuh do ischt Pflegrin schuld; da sies verkehrt vor Schlaftür stellt. (Magdalena R. (kfb-2950), servant, letter to mayor, August 1936) [The shoes are the nurse's fault; because she puts them in the wrong way in front of the bedroom door (gloss.: shoes there is-the nurse to-blame; because she-them in the wrong way […]]

Clitics like in *ischt* (see Footnote 7) and *sies* in this example are one means by which she creates this switch. She seems to use the dialect here – in an official letter – to point out that she and the addressee are both speakers of the same dialect and to create identification by their shared regional background. As the dialect is typically used in communication between closely related people, she also constructs a familiar relationship. Schiegg and Foldenauer (2021: 381f.) describe the creation of closeness as a function of codeswitching in her texts, too. They also note another possible function of Magdalena R.'s dialect use: the construction of a cunning identity.

The performance of aspects of the writers' identity is a recurrent function of clitics in my corpus as well. Magdalena R. uses these forms to exhibit her linguistic competence. In (6), the exhibition of her dialect competence is closely related to the construction of the relationship to the addressee. There are, however, other examples of Magdalena R. using language in a playful or poetic way with clitics that do not have a close connection to addressee-related functions.

(7) jetzt gehts los; aber nicht in die Hos. (Magdalena R. (kfb-2950), servant, letter to relatives, 24 May 1936) [Now it starts; but it won't be a complete flop (gloss.: now starts-it; but not into the trousers)]

In (7) the writer combines the *es*-clitic in *gehts* with other informal features such as the apocope in *Hos* (Havinga 2018: 2, 17). The phraseological expression modified by an ellipsis and the rhyme underline her ability to play with language.

Georg B. exhibits parts of his identity with clitics in a different way in contexts in which he expresses his emotions (see (8)).

(8) obgleich eben wieder sonst gut bei Leben aber bei Lumpen, wie's größer nicht denkbar sind – (Georg B. (kfb-966), accountant, letter to 'Richardinese' (probably private), 29 April 1884) [although in good health again, but with rascals, you can't imagine bigger ones (gloss.: as-they bigger not imaginable are)]

In the context of this passage, Georg B. complains about the hospital and his confinement there. His emotional affection by this topic and his anger about other people in the hospital are expressed by pejorative lexis, for example, *Lumpen* [rascals], to which the clitic in *wie's* refers. The clitic reinforces this effect as it is an untypically informal form for this experienced writer's text (cf. e.g. the contrast to the formal *obgleich* [although]) and a violation of conventions. Georg B. marks this inappropriateness of the form in written language by an apostrophe (Nübling 2014: 103). This underlines that the clitic is not caused by low attention to writing as a result of his emotionality, but that rather he uses this form to display his anger and aversion. Hence, he constructs himself as an emotional being and differentiates himself from the people criticized. Thus, clitics have acquired a meaning beyond their literal meaning so that they can be used to construct social identities.

## 9 Conclusion

The analysis of the use of clitics in a corpus of 192 private and official letters written by 39 patients of psychiatric hospitals in the nineteenth and early twentieth centuries has shown that the grammaticalization of preposition article clitics that have reached the stage of special clitics in present-day German is already advanced in the period of investigation, but that their number still increases over time. For article clitics with non-prepositional hosts and pronominal clitics, regional and diachronic differences and differences based on writing experience and gender were observable. This indicates that these forms are part of the language of immediacy.

In terms of intra-writer variation, I found that article clitics with nonprepositional hosts and pronominal clitics occur more often in private letters than in official ones, though not exclusively so. Thus, writers adapt their use of clitics to the addressee, that is, they show patterns of audience design. A more detailed analysis of selected writers has uncovered that they use clitics to perform aspects of their identity (e.g. linguistic competence and emotionality) and that besides the mere distribution over private and official letters there are also more complex ways of using clitics to construct the relationship to the addressee (e.g. asking for identification and marking confidentiality). This shows that audience design and the construction of social identities are related functions of stylistic variation. Both seem to result from clitics being a means to create a language of immediacy. Intra-writer variation with clitics is not only a consequence of different communicative conditions, but the social meaning of some clitics can be actively used for stylistic purposes. This could reinforce their use in written language. The forms that allow such functions are those with stronger regional restrictions (cf. also Werth 2020: 205) and are used more frequently by inexperienced writers. Hence, there is also a connection of intra-writer variation with patterns of inter-writer variation. Further investigations of this connection could be promising, for example, by comparing the audience design patterns exhibited by writers with different writing experience. Extending the analysis to other text types could also be helpful for gaining further insights into variational patterns of clitics.

## References

#### *Primary sources*


(Georg B.), 1005 (Elisabeth K.), 1145 (Ignaz L.), 1269 ( Josef B.), 1480 (Karl G.), 1614 (Karl D.), 1621 (Martin B.), 1630 (Xaver G.), 1634 (Lucie B.), 1945 (Max H.), 2108 (Cosmas R.), 2206 (Gustav H.), 2323 (Friedrich B.), 2325 (Anna S.), 2950 (Magdalena R.), 3085 (Katharina S.), 3138 (Philomena D.), 3349 ( Jakob F.), 3688 (Rudolf B.), 3842 ( Josef S.), 3914 (Karolina D.), 7021 (Emil W.), 8662 (Frieda W.), 10608 (Elisabeth H.).

## *Secondary sources*


part II

Intra-writer variation in contact and migration settings

#### Joshua Brown

11 Intra-writer variation and linguistic accommodation in the letters of the Milanese merchant Giovanni da Pessano to the Datini network (1397–1402)

#### Abstract

This chapter considers intra-writer variation in the letters (1397–1402) of the Milanese merchant Giovanni da Pessano to members of the Datini network, a large Tuscan trading company. Three variables are investigated: (1) voicing of devoiced intervocalic consonants; (2) rhotacism of intervocalic -L- and; (3) past participle endings of verbs deriving from Latin -ĀTU(M). The aim of the chapter is to investigate to what extent intra-writer variation can be seen to be a strategy in merchant writing. Linguistic accommodation is shown to be a significant factor in Giovanni's letters, owing to the unequal balance of power between his status and that of the Datini network.

## 1 Introduction1

Historical sociolinguistics has seen a revived interest in the question of intra-writer variation, particularly in the field of English historical (socio-) linguistics.2 Conversely, studies that consider intra-writer variation and linguistic accommodation are relatively rare in the research literature (but see Auer 2015; Schiegg 2018; Ulbrich & Werth 2021). Only recently have scholars begun to investigate the factors which may lead to such variation. Much work has been carried out since Bell (1984) first launched a serious

<sup>1</sup> The author is grateful to the editors for feedback on an early draft of this chapter.

<sup>2</sup> Cf. Auer (2018), Gardner (2018), Hernández-Campoy and García-Vidal (2018), Kerswill and Williams (2000), Anita Auer et al.'s project on *Emerging Standards*.

programme for the study of intra-speaker variation, in part taking inspiration from Giles et al.'s (1973) theory of accommodation. In contemporary studies, style is 'at the centre of sociolinguistic theorization and method' (Bell 2014: 297).3 The question of investigating intra-writer variation in historical texts, however, has thrown up a new series of methodological problems. Some of these relate to questions of reported speech, audience design, speaker design, and how best to apply the tools and concepts of modern sociolinguistics to historical data. Schiegg (2018: 102) has pointed to the need for 'the development of a detailed sociolinguistic framework for the analysis of stylistic variation in written and historical data'.

Studies on intra-writer variation which focus on Italian data are rare, and a brief review of the literature shows that most studies are still carried out in the domain of speech (e.g. Castellana et al. 2017). Similar comments are echoed by Anipa (2018) also with respect to French. More generally, the topic seems to have been overlooked in Romance linguistics as a whole. The term 'intra-writer' does not appear to be present in the major reference works on Romance sociolinguistics or histories,4 even though certain studies of historical variation at the societal or group level have been carried out in most varieties of standard Romance languages.

This chapter examines three variables (two phonological and one morphological), providing evidence of intra-writer variation (or the absence thereof ) in the sixty-eight letters written by the Milanese merchant Giovanni da Pessano between 1397 and 1402. All letters are addressed to members of the large trading company based in Tuscany controlled by Francesco di Marco Datini and referenced throughout the literature as the 'Datini network'. Using both quantitative and qualitative data, the aim of this chapter is to show empirically how stylistic choices can function as a linguistic resource for pragmatic purposes (cf. Hernández-Campoy & García-Vidal 2018; see Antenhofer 2005 for a qualitative, historical ap proach). I argue that a process of written accommodation was evolving in a situation where the balance of power between interlocutors was uneven,


and show how this difference is played out across several variables in writing with variants from distinct dialects. Section 2 introduces the context and corpus of the study, first describing the linguistic landscape of late medieval Milan to present the linguistic repertoire with which Giovanni and other merchants would have been familiar. This section then describes the biographical information available on Giovanni, before defining the corpus. Section 3 provides an overview of the methodology. Results and discussion are presented in Section 4. Section 5 discusses select phenomena in particular letters worthy of attention in the corpus. A brief conclusion is provided in Section 6.

## 2 Context and corpus: The letters (1397–1402) of Giovanni da Pessano

Renaissance Italy was characterized by a situation of multilingualism. Like all northern writers, Giovanni would have been in contact with, and had knowledge of, at least three linguistic varieties: his native Milanese vernacular, Latin, and also Tuscan, which had begun to circulate throughout the peninsula (Brown 2017a). The main tendency characterizing the evolution of the vernacular in northern Italy during the fourteenth and fifteenth centuries is the formation of a koiné, or supra-local variety. At the same time, histories of the vernacular in northern Italy have identified texts that have maintained a strong adherence to local forms of language whilst koineization was in progress (Morgana 2015; Sanga 1997; Stella 1994). Scholars have pointed to the 'learnèd' nature of the northern koiné, which found its most elaborate expression in the chanceries and the courts (Bongrani & Morgana 1992; Morgana 2012). This northern koiné is the main written variety which non-literary writers used, including merchants, for their everyday correspondence (Brown 2013). Although the vast majority of writing was still conducted in Latin at this time, most merchants were not in a position to use Latin with confidence. Indeed, there are very few letters available in Latin from Milan housed in the Datini Archive (only 4 out of 810), and the overwhelming preference was to write in vernacular (Brown 2017b). During this same time period, Tuscan forms of language (typologically distinct from Gallo-Italic varieties, to which Milanese belongs) began to circulate throughout Italy. Gallo-Italic, or what Maiden and Parry (1997: 3) call 'Gallo-Italian' (spoken in the regions of Piedmont, Lombardy, Liguria, Emilia-Romagna) are the main subvarieties of Romance found in north Italy. Gallo-Italic varieties are typologically distinct from Tuscan, and are divided from Tuscan by the La Spezia-Rimini isogloss as shown in Figure 11.1. 5

Figure 11.1. The distribution of Gallo-Italic in Italy.6


From this, it follows that Giovanni must have had a range of codes available to him and was in contact with different types of writing (commercial letters, tax invoices, inventories, etc.). His written language exhibited a mix of the Milanese, Latin, and Tuscan varieties to different degrees and in different ways. The varieties available to Giovanni mean that he was able to choose freely which variants to adopt in his writing. The extant correspondence from him, therefore, constitutes an excellent case-study for investigating intra-writer variation in a historical framework in the traditional sense of Bell (2007: 90), that is, 'the range of variation produced by individual speakers within their own speech'.

There is little information available about Giovanni da Pessano's life. He may be related to the *Pescina* family and to the three brothers (Damiano, Basciano and Francesco) mentioned in his correspondence and who also sent letters to the Datini network. Melis (1990) records his name in the index as 'Giovanni da Pessano (Giovanni da Pescina)', giving both variants. There is no clear reference to him that emerges from any study on the Datini Archive at this time, despite Barbieri (1961: 62–71) believing that he is the brother of Basciano and Damiano da Pescina. He uses a merchant symbol [*segno mercantile*] which is very different from other members of the Pescina family, as noted by Frangioni (1994: 71). She has described Giovanni's position as 'una eventuale parentela tutta da definire' [a possible relation still to be defined], and mentions that 'a lui non è riconducibile una stretta origine milanese' [a clear Milanese origin cannot be ascribed to him] (p. 71).7 All we know with certainty is that he was a 'caro amico' [dear friend] of another Milanese merchant, Giovanni da Dugnano, and that he was involved in the fustian trade.8 Frangioni's doubt about Giovanni da Pessano's provenance derives from one of his letters sent from Milan to Genoa in 1397, in which he writes:


Based on this assertion, Frangioni (1994: 73) describes Giovanni's provenance as being 'non proprio milanese' [not strictly Milanese]. However, his Milanese origin seems to be confirmed in a later letter in which he writes:

(2) Similli i' ò avixatto al ditto Bindo chomo dè essere una mia prochura in caxa di Zanobio di Tadeo chomo sono melanexe […] E che, anchora, è in Vinegia più merchadanti milanexe che me cognosono: se serà di bisognio dirano chomo sono milanexe e al ditto Bindo l'ò avixato de le nome di loro. (letter 774) [Similarly, I have advised aforementioned Bindo how there should be a proxy document at the house of Zanobio di Tadeo that I am Milanese […] And that, further, there are several Milanese merchants who know me: if it is necessary, they will say that I am Milanese and I have advised Bindo of their names.]

It is worth mentioning some details about Giovanni's main addressee and the relationship between these two merchants. Francesco di Marco Datini (c. 1335–1410), the 'merchant of Prato', moved to Avignon in the south of France at age 15 and soon after began trading in arms and armour, eventually founding trading warehouses (*fondachi*) in Prato, Avignon, Florence, Pisa, Genoa, Barcelona, Valencia, and the Balearic Islands. On his return to Prato from Avignon in 1382, he stopped for a week in Milan to gather supplies for his onward journey and to establish trade agreements with fellow merchants. The main trading partner Datini gained was the Pescina family, but Datini and his associates would eventually carry on direct correspondence with at least four other merchants from Milan itself or nearby, as well as from the main trading areas all over Lombardy (Brown 2017b). It was likely during this stay in Milan and in this context that the relationship between Giovanni da Pessano and Francesco di Marco Datini first began. Although there is no evidence that Giovanni and Francesco ever met in person again, employees from the Datini network made several trips to Milan to meet with fellow traders there, such as Tieri di Benci (in at least 1383, 1394 and 1386) as well as Tommaso di ser Giovanni from 1394 onwards (Brown 2017a: 50–52). In this sense, the corpus presented here mirrors Hendriks' study by adopting a speakerbased perspective, from which 'the potential for idiolectal change as a result of contact with speakers from outside an individual's dialect network is explored' (2018: 130).

The commercial letters sent by Giovanni da Pessano were written between 30 August 1397 and 17 December 1402. They are all sent from Milan. These letters have been published previously by Frangioni (1994), and a philological edition is available in Brown (2017b), on which this study is based. The analysis which follows below takes as its corpus a sample of sixty-eight letters by Giovanni out of the seventy-two letters included in Frangioni's corpus. In aiming to create the most homogeneous corpus possible, I have excluded four items from the analysis. These are two items denominated *estratti conto* (receipts) and two items which are not in Giovanni's hand.9

## 3 Methodology

The methodology chosen to investigate intra-writer variation in this chapter focuses on specific graphemic and phonological variables, which show contrasting features between Milanese and Tuscan. Given the lack of available documentation from Milan during the fourteenth century, I have had to use texts that fall outside this period for contrastive analysis. For verb morphology, I have made particular use of the thorough descriptions available of Bonvesin dra Riva's literature from the late 1200s (Domokos 2007). The varieties of language being considered here are in flux and so it would be erroneous to impose water-tight categories of either 'Tuscan' or 'Milanese'. Nevertheless, the language histories of both Tuscany and Milan ascribe unique, non-mutual features to both Tuscan and Milanese which I have used to verify whether Giovanni has used a Tuscan(ized) variant, or not.

Since certain linguistic phenomena appear common to both Tuscan and Milanese, I have focused on three variables which *do* contrast between

9 The two items not in Giovanni's hand but which appear in Frangioni's corpus are letter 697 (Frangioni 1994: 500f.) and letter 758 (p. 531). Letter 697 is written by one of Giovanni's cousins. Letter 758 is 'lettera non firmata di mano di Giovanni da Pessano' [not a letter signed in the hand of Giovanni da Pessano].

these two vernaculars.10 These variables therefore provide an excellent testing-bed to investigate whether Giovanni is being strategic in his use of intra-writer variation, and whether he is accommodating to his Tuscan interlocutors.11 The three variables investigated here are:


In the case of (c), hyper-correct occurrences in Giovanni's letters of the desinence provide evidence of imperfect acquisition of a second language variety. For example, in attempting to imitate a Tuscan ending, Giovanni reproduces instances with double consonants such as *receutto* for *ricevuto* [received]. This hyper-correction is likely due to the fact that single and double consonants are characteristic only of Tuscany, while varieties above the La Spezia-Rimini isogloss only have single consonants. For each of the variables listed above, I quantitatively assess the presence of the individual phenomena in the sixty-eight letters authored by Giovanni da Pessano. I then provide some qualitative comments on whether the item can be said to show evidence of a strategic decision to accommodate to his interlocutor.

<sup>10</sup> For example, 1sg. *-o* desinences appear common to the earliest documents in Tuscan (Rohlfs 1966: §527) and Milanese (Domokos 2007: 263) and are therefore not contrastive.

<sup>11</sup> Further research will be able to contrast the results presented here with the letters written by Giovanni to his non-Tuscan addressees.

## 4 Results and discussion

#### *4.1 Voicing of devoiced intervocalic consonants*

Voicing of intervocalic consonants is one of the nine features characteristic of all Gallo-Italic vernaculars, to which Milanese belongs. This goes for all intervocalic plosives in Milanese. In early language from Milan that is from the late thirteenth century consonants can be lenited to the point of disappearance. By the early sixteenth century, voicing was limited to only a few select lexemes, and intervocalic consonants were found to be one of the 'easiest' to be replaced by the unvoiced, Tuscan variant (Scotti Morgana 1983: 339). In general, one sees that the northern tendency towards voicing is weak in non-literary Milanese during the early sixteenth century, and 'forms with a voiced consonant in place of the unvoiced are very rare and are certainly due to the influence which the Latinizing tendency had on them as well as the adaptation to literary Tuscan' (Bonomi 1983: 258f.). In the corpus presented here, the following variants and tokens are present (see Table 11.1).12

This voicing occurs in seventeen different letters across the entirety of the corpus (1397–1402). While all letters in this subset were sent from Milan, those letters which contain voiced consonants were sent to a variety of geographical locations around north Italy (5 to Genoa, 4 to Prato, 4 to Florence, and 4 to Bologna). In other words, place is not a significant factor when it comes to this particular variable. Considering the particular lexemes in which such voicing occurs, a wide variety of forms are present, including technical terms (*charegato* [loaded]), verbs of saying (*digho* [I say]), and lexemes such as *merchadante* [merchant]*.* It is also present in forms of negation which are native to Milanese, such as *migha* [not]. An alternative way of viewing these data is that variants unique to Milanese,

12 I have excluded cases where /t/ and /d/ occur in past participles here (e.g. *nomerado* [numbered]), since these data are discussed in Section 4.3. The percentages included in Table 11.1 only refer to lexemes or to instances of voicing in the verb stem (e.g. *charegato* discussed above).


Table 11.1. Voiced vs unvoiced intervocalic consonants

such as *migha*, seem more resistant to change than others. Technical terms such as *merchadante* are also seen as part of the merchant's identity and are thus more likely to be voiced.

A patterning of voiced intervocalic consonants can be seen in Giovanni's writing more generally. For example, one also sees instances of voicing and a more relaxed register in other text types, specifically the socalled *carteggio specializzato*, that is, lists of merchandise sent and received and which are typologically distinct, therefore, from commercial letters. A case in point from Giovanni's writing can be seen in the four documents excluded from the corpus of seventy-two letters described above and which are part of the *carteggio specializzato*. These include item 755 in Frangioni (1994: 529), which contains *Gomo* 'Como' [Como, place] as well as *hendego* 'indaco' [indigo], *perdida* 'perdita' [loss]; item 804 has one occurrence of *fondego* 'fondaco' [warehouse] on p. 569 and three on p. 570, as well as *caregono* 'caricano' [(they) load, 3pl. verb] and *discharegono* 'discaricano' [(they) unload, 3pl. verb].

#### *4.2 Rhotacism of intervocalic -l-*

Rhotacism of intervocalic *-l-* is another distinguishing feature of western Lombard, to which the Milanese dialect belongs (e.g. *coror(e)* < Latin *colōrem* [colour], see Sanga 1997: 255, point 20). In the corpus, both Tuscan and Milanese variants are present in an uneven distribution. Bongrani and Morgana note that rhotacism 'has been described as uniquely Milanese' and that some have spoken of 'Milanese rhotacism'. They go on to say that rhotacism 'has never gone beyond the boundary of the Adda river, and it has represented an important and distinctive trait of western Lombard dialects' (1992: 91). Degli Innocenti remarks that cases of rhotacism are still 'notable' in Milanese during the fifteenth century, and that they 'often alternate with *l* that has been maintained' (1984: 50f.). In his study of documents from the Milanese chancery of the fifteenth century, Vitale notes 'only a few cases of intervocalic rhotacism, which is mainly limited to toponomy – a sign of an old diffusion of this phenomenon' and a 'more recent literary restitution of *l*' (1953: 72). The cases presented below thus foreshadow Vitale's results, and can be seen to be early evidence for tuscanization in non-literary documents, since Vitale's corpus fixes the *terminus post quem* for this phenomenon to the early fifteenth century. For ease of cross-referencing, I have included the Roman numeral in round brackets which refers to the number of the letter, while the Arabic numeral refers to the line number, as found in the critical edition in Brown (2017a).

In this corpus, there is one case of rhotacism, in *peroxe* 'pelose' [hairy]. Cases with intervocalic *-l-* number 34. In other words, the main variant shows the Tuscan outcome *-l-*, and can be seen as evidence of Giovanni's attempt to accommodate his language to his Tuscan interlocutors.13 When

13 These occurrences are*: cholore* (LX: 9, 10) (LXV: 22); *colore* (LVIIV: 2); *colory* (LVIIV: 1); *conseilio* (XXII: 9, 15); *dichonsolato* (XVII: 22); *diligentia* (XXIV: 29); *generale* (XXV: 14); *malanchonia* (XVII: 13) (LXIII: 11); *malanchonioxo* (LXI: 22); *palese* (XXIV: 19); *pelanda* (XVI: 8); *pelegrina* (LXXIV: 10); *pelegrino* (LXVIIIV: 43); *pericholo* (XIV: 4) (LXIX: 41) (LXX: 20); *saluti* (II: 7) (III: 5); *salutti* (I: 23); *soliva* (LXIII: 5); *valente* (XXIV: 21); *volentà* (XIX: 14) (XXX: 5) (XLII: 16) (XLIII: 10) (XLVII: 7, 25) (XLVIII: 6); *volere* (I: 10); *voleva* (VI: 25).

considered over the chronological period of the corpus, we find an unpatterned distribution.

The case of *peroxe* is significant, as it occurs right in the middle of Giovanni's correspondence and in a list of items for which he is providing the local prices of goods in Milan. Such lists often follow a formulaic pattern. In this case, we can observe the native Milanese language of Giovanni breaking through in his attempt to tuscanize his writing.14 Indeed, this occurrence of rhotacism occurs in letter 64, written in 1401, in which we see several attempts by Giovanni to introduce Tuscan variants other than the ones treated in this chapter. In doing so, he often produces hyper-correct forms in attempting to write a variety with which he is less familiar. For example, in the same lines of this letter, we see cases of *lanna* 'lana' [wool], *cottono* 'cotone' [cotton], and *cosse* 'cose' [things], characteristic of Lombard vernaculars, hence also of Milanese.15 Other instances in this letter show his northern provenance, such as *zò* for Tuscan *ciò* [that]; *chomo* for Tuscan *come* [how]; *serebeno* for Tuscan *sarebbero* [they would be] and others (see Brown 2013 for further examples).

## *4.3 Past participle endings of 1st conjugation verbs deriving from Latin -ātu(m)*

As mentioned above, the voicing of intervocalic consonants is characteristic of Milanese. In Tuscan, the unmarked outcome for past participle endings of 1st conjugation verbs deriving from Latin -ātu(m) is *-ato*


(e.g. *parlare* [to speak] *> parlato* [spoken]). For Milanese, in contrast, Domokos (2007) has identified three past participle endings from the late 1200s for *-are* verbs: *-ado, -adho, -ao*. Rohlfs (1966: §203) notes that, at an early stage, Latin -ātu(m) > -*ado* and then > *-aδo* (where *δ* has the phonetic value of a voiced interdental fricative). From this latter development, -*aδ* > *-a* 'in large areas in the north'. In eastern Lombardy, there was a return to a preceding form *-ado*. For other areas of Lombardy, including Milan, *δ* 'disappeared before the weaking of the final vowel: the result of such a development has been *-aδo* > *au* (*ao*)' (Rohlfs 1966: §203) which then turned into both *ou* and *o*. For fifteenth-century Milanese chancery documents, Vitale records *-ato* as the dominant ending which is only sometimes reduced to *ado > ao > à*. In non-literary Milanese documents from the early sixteenth century, Scotti Morgana's corpus (1983: 359) found only *-ato* to be present, which is 'never voiced or reduced'*.* Given that lenition of intervocalic obstruents is a fundamental characteristic of Gallo-Italic vernaculars (Bongrani & Morgana 1992: 86), the almost total lack of voicing in past participles in Scotti Morgana's corpus is surprising. In short, past participle outcomes previously recorded in the literature and uniquely ascribable to Milanese include *-ado, -adho, -agho, -ao.* Table 11.2 presents the different variants of past participle endings present in the corpus, the variety to which the variant belongs, and the number of tokens.

As can be seen in Table 11.2, the variants present in the corpus show an overwhelming preference for Tuscan *-ato* over voiced variants. Despite the one occurrence of *-agho*, native to Milanese, no other Gallo-Italic forms are present. Geminate consonants are not a feature of Gallo-Italic varieties.


Table 11.2. Past participle endings of 1st conjugation verbs < -ātu(m)

Tuscany is the only region in Italy where single and double consonants are phonemic. Giovanni appears unsure about when to use geminate consonants when writing to his Tuscan interlocutor. This over-correction can be seen in his practice of doubling the consonant in past participles deriving from Latin -ātu(m). Given that the outcome from the Latin participle in Tuscany for masculine singular participle is *-ato*, Giovanni's use of a geminate can be seen as a form of hyper-correction.16

What is surprising from these results is the hyper-correct doubling of the consonant *-t-* in forms such as *parlatto* [spoken], *possutto* [was able to], *giontta* [reached], etc., also in other participles other than < -ātu(m). The presence of Tuscan forms can be seen as evidence of Giovanni's desire to accommodate to his interlocutors by incorporating Tuscan forms that were spreading throughout the north (a similar process in English letters was found by Hernández-Campoy & Conde-Silvestre 2015). Giovanni's use of Tuscan may reflect the broader imbalance in power dynamics between himself and the large Datini network. At the same time, the presence of native Milanese endings with the voiced consonant suggests his incomplete, evolving knowledge of Tuscan and inability to reproduce Tuscan forms correctly. The frequency of all past participle variants can be visualized more easily in Figure 11.2.

These data reveal several interesting patterns throughout the ongoing relationship between Giovanni da Pessano and the Tuscan merchants with whom he corresponded. The immediate and continued presence of Tuscan *-ato* shows his ability and willingness to adopt forms that are not native to him. Given this attempt to accommodate his language to his economic superiors, the low number of tokens of Milanese variants is not surprising. A less obvious, but no less striking feature of the data, is his progressive use of hyper-correct forms. These are only present in a minimal way for the first few years of his correspondence, during 1397–99, but the last three years seem to show an increased confidence to adopt Tuscan forms and an increase in frequency. This confidence is so robust that the hyper-correct

<sup>16</sup> I point out that hyper-correction is not limited to past participles, but also to place names (in letter 60, *Pratto* is recorded for *Prato*) and in letter 68 one finds at least one instance of *merchatto* (for *mercato*).

Figure 11.2. Tuscan vs Milanese vs 'hyper-correct' past participle endings.

occurrences *-atto* increase dramatically from 1400 to 1402, and for the latter two years are almost in equal competition with Tuscan forms in terms of frequency.

These hyper-correct occurrences appear in a subset of twenty commonly used verbs throughout Giovanni's letters. In one sense, the semantic range of these verbs is unsurprising, since they are all verbs of some frequency and common to mercantile language, trade, and the exchange of goods and services.17 It *is* striking that the range of verbs used coincides with the period in which one also sees the greatest incidence of hyper-correct participles. That is, in the first fifty-four letters written by Giovanni (from 1397 to 1400), this range is limited to just six different verbs (*aparegiatto* [prepared], *datto* [gave], *adovratto* [employed], *avixatto* [advised], *'rivatto*

17 In terms of token frequency, there are twenty-four cases of *avixatto* [advised, informed] (with minor orthographical variation: one case of *avisatto*, and one of *'vixatto*). Verb participles which have five tokens include: *trovatto* [found], *ghuadegniatto* [earned], *statto* [was], *passatto* [spent]. Participles with two tokens include: *valichatto* [crossed], *usatto* [used], *schuxatto* [pardoned], *penatto* [struggled], *datto* [given], and *'rivatto* [arrived]. The remaining verbs only occur once: *achonziatto* [placed], *adovratto* [employed], *aparegiatto* [prepared], *chompratto* [bought], *'larghatto* [widened], *mandatto* [sent], *ordinatto* [ordered], *paghatto* [paid], *schorazatto* [discouraged].

[arrived], *trovatto* [found]). The occurrences of the other eighteen verbs occur in his later letters (letters 55–71; from 1401 to 1402). In short, the coincidence between an increased use of hypercorrect forms and an increased lexical range of verbs may be evidence for Giovanni's confidence in writing, but also for an increased attempt to reach correct Tuscan forms. Overall, the data reveal this writer's imperfect knowledge of a vernacular that was foreign to him, as well as his effort to reproduce Tuscan forms to the best of his ability.

## 5 Other Tuscan phenomena

Certain letters in the corpus contain other phenomena worthy of brief discussion. I note here the use of some forms of locative adverbs, which require particular knowledge about the spatio-temporal location of participants in the communicative act (Vanelli & Renzi 1997: 112f.). Specifically, Tuscan *costì* [there by you] and *costà* [there by you, further away] (Rohlfs 1966: §895) appear never to have developed as grammatical categories in the north (Ledgeway 2015; Prandi 2015).18 The ternary deictic system is a feature of Tuscany, central-southern vernaculars, and Piedmontese (Vanelli & Renzi 1997: 112; Da Milano 2015: 61), while Giovanni's native Milanese deictic system maintains a binary distinction. Nevertheless, these adverbs do appear in Giovanni's writing. Occurrences include *costà* (81 tokens), *costì* (43) and *chostì* (1). There is also one instance of demonstrative adjective *choteste* 'this'*.* Vanelli & Renzi explain that Tuscan and literary Italian *codesta* indicate a referent 'pertaining' to the addressee, as in Tuscan and literary Italian *codesta tua idea mi piace* [I like that idea of yours]. Giovanni's adoption of these forms is significant, and drives home

18 Both these terms can be translated into English as 'there by you', and generally speaking, do not differ in usage. *Costì* can indicate a place or person closer in proximity to the speaker when more than two people are present, while *costà* is used for the person who is further away in distance from the other two. For further information, see Treccani Vocabolario online.

even more his willingness to accommodate to his interlocutors. In other words, not only has he adopted a Tuscan lexeme, but he has also introduced a grammatical category which is foreign to his native variety. This variation could be interpreted as a strategic decision by him to appear more Tuscan in his linguistic behaviour.

## 6 Conclusion

The variation present in these letters can be seen as part of the broader power dynamic evolving between Giovanni and the Datini network, but also between Tuscany and Milan more generally. While studies of style, audience design, and stylistic variation have been popular ever since Bell (1984), the main focus of this research trajectory has been on spoken corpora, with little attention paid to writing or how this approach can be applied in historical perspective.

In some cases, the use of a particular variant in Giovanni's writing appears to be almost categorical. There is only one case of rhotacism, for instance, thus showing a process of 'upward' accommodation to his fellow merchants, and mirroring the results in Hernández-Campoy and García-Vidal (2018: 49), who found 100 % use of a particular graphic variable in a letter of 1475 to King Edward IV. The case of rhotacism in *peroxe*, appearing as it does in the middle of a list of items and their prices, offers further evidence of the way in which Giovanni mixes both Milanese and Tuscan variants freely throughout his writing. In simply listing a series of commodities to be traded, it is likely he pays less attention to the rhetorical and linguistic choices of the variants he uses, thus allowing for this case of rhotacism to emerge. In the other two variables discussed here, greater variation is present. In some cases, this variation is relatively constant over the period 1397–1402 such as for voicing of devoiced intervocalic consonants, while in others the patterning can be explained by the typological nature of Giovanni's writing. I have argued that hyper-correct occurrences can be seen as indicators of imperfect acquisition of a second language variety. Assessing hyper-corrections of past participle endings over the six years for which data are available show these to be distributed in a heterogeneous way, increasing in both type and token frequency later in the period, as Giovanni becomes more confident (indeed, overconfident) in his use of a vernacular which was foreign to him such as Tuscan. These letters provide evidence, therefore, of intra-writer variation and the linguistic outcomes of an author who freely mixes both Tuscan and Milanese variants.

Vitale's study of the Milanese chancery showed that Tuscan was already being used outside the sphere of literature in northern Italy during the fifteenth century. He noted that the first document in vernacular in the Milanese chancery appeared in 1426, and that use of Latin in chancery documents continued to decrease during the entire fifteenth century (Vitale 1953: 16f.). Before this, Tuscan was not a model for non-literary writing. Tuscan influence in orthography and morphology is little evident. The linguistic accommodation shown by Giovanni towards his Tuscan interlocutors opens up the question of language choice in merchant writing and whether what Maraschio has called the 'vertical' expansion of Tuscan in the fifteenth and sixteenth centuries might have begun earlier (1976: 37; Brown 2020). In other words, Tuscan's presence is felt not only in the highest strata of Milanese society but it was also, at the other end of the spectrum, known and used for successful communication between the less educated merchant class (Brown 2020). Giovanni's letters can be seen as a useful case-study of the way in which authors from northern Italy began to adapt and adopt Florentine forms into their writing, as a Tuscan variety began to spread throughout the peninsula. It was precisely this variety which ultimately formed the base for a national standard.

When taken together, these variables can be seen to show how intrawriter variation played out in a strategic way in order to obtain particular objectives by a merchant who was economically 'inferior' to his Tuscan counterparts. I have argued that Giovanni's efforts to reproduce Tuscan forms can be seen as a strategic decision to accommodate to his fellow merchants. This behaviour is likely carried out in order to ingratiate himself to members of the Datini network, in the hope of obtaining more favourable economic circumstances for the various trades and deals in which he was engaged. This chapter has quantitatively assessed the presence, frequency, and distribution of three variables in the sixty-eight letters written by Giovanni da Pessano. The variation inherent in these letters, in turn, can be seen as part of the broader power dynamic evolving between Tuscany and Milan in Renaissance Italy.

## References


#### Sarah van Eyndhoven

# 12 Eighteenth-century Scots in correspondence during the Union Debates: An intra-writer perspective

#### Abstract

Eighteenth-century Scottish correspondence is a rich source of historical sociolinguistic data, given the complexities of the socio-political climate facing the Scottish elite during the Union debates of 1689–1707. Yet this time period is underexplored in historical Scots research. To begin to address this gap in the research, I quantitatively and qualitatively analyse the correspondence of two Scottish politicians. After digitizing and tagging the correspondence for lexical and orthographic features characteristic of Scots, the authors are compared with one another in terms of their degree of 'Scottishness', after which intrawriter variability is explored. Results are suggestive of the strategic role Scots could play in emphasizing degrees of social distance between correspondents, indicating that the implementation of Scots features in correspondence remained indispensable even to the highest levels of Scottish society.

## 1 Background

#### *1.1 The rise and decline of written Scots*

Scots is a West Germanic language that has its origins in the Northumbrian variety of Old English, from which it diverged around the twelfth century, undergoing various linguistic and social changes. By the sixteenth century it had become a distinct national language that was on its way to standardization, but this process was interrupted by a number of historical developments, including the arrival of the printing press, the religious Reformation of 1560 and the Union of the Crowns in 1603 (Aitken 1979, 1997; Dossena 2011). These resulted in incremental increases in the prestige and influence of the emerging Southern English Standard, gradually displacing Scots within various higher state functions (Kopaczyk 2012). By 1700 written Scots had become much less distinct from English, and the Union of the Parliaments in 1707 (henceforth 'the Union'), which dissolved Scotland's formerly independent parliament, seemed to signify the final blow. As Scots had seen a dramatic decrease in status and context-suitability due to 150 years of Anglicization, its use now became increasingly associated with 'the domestic, the familiar, the sentimental [and] the comic' (Murison 1964: 37).

Previous scholarship has found very little Scots remaining in the writing of the upper elite by the eighteenth century (Aitken 1997; Cruickshank 2011), and Aitken (1979: 95) suggests it 'approximat[ed] to Standard English by about the time of the Revolution and of the Union of Parliaments, albeit […] occasionally marred by a lexical Scotticism or two'. Yet the lead up to the parliamentary union was marked by tense political debate on all sides, stimulating a heightened awareness of Scottish nationhood that was to find a linguistic outlet in the 'vernacular revival' half a century later. Already at its outset however, politically active figures faced conflicting demands as they sought to navigate a complex cultural dualism that interweaved the desire for social mobility and prestige, with loyalty to their country of birth (Whatley 2006). Politicians were constantly negotiating between their personal stance and various socio-political relationships within movements embracing linguistic and cultural uniformity or diversity. As a result, the overt prestige of the incoming standard and the covert value of the vernacular could mark or delineate certain relationships or identities.

#### *1.2 The role of correspondence*

Correspondence has often been argued to be the most 'speech-like' of historical genres (Biber 1995: 283–300) and thus liable to preserve the vernacular better than many other text-types. Unlike official, public or printed documents, self-monitoring tends to be reduced in personal writing (Biber & Finegan 1989), and correspondents are often on familiar terms, increasing the possibility to use variants of lower prestige or status (Dossena 2002: 109f.). Printed texts were prohibitively expensive in the eighteenth century, and thus frequently subject to anglicized alterations to reach the wider English readership (Dossena 2011). We might therefore expect that written Scots is best preserved in correspondence by this time.

Moreover, correspondence – as texts (usually) produced by a single individual – provides us with the opportunity to explore language variation at the micro-level. Various studies have already demonstrated the sociolinguistic insights that can be gleaned from examining the specific practices of historical individuals (Elspaß 2002; Hernández-Campoy & Conde-Silvestre 2015; Schiegg 2016). Given the scarcity of written Scots by 1700, its usage is likely to be highly sensitive to the individuals under examination and their network of addressees, shaped by particular author-centred and audience-based influences which in historical sociolinguistic analyses have increasingly been found to work in tandem (Auer 2015; Hernández-Campoy & García-Vidal 2018). Early eighteenth-century Scottish correspondence therefore provides a unique opportunity to observe whether the complex dynamics facing the ruling elite at this time intersected with their writing practices. By examining intra-writer variation within this context, we can explore some of the more subtle factors and sociolinguistic nuances at play during this time.

## 2 Aim

The debates around the Union of the Parliaments generated a coalescence of diametric interests; involving national concerns, local loyalties, sociopolitical ambitions and pragmatic decision-making that affected political players and how they negotiated their professional relationships. There is an intuitive sense in which Scots, as a facet distinct to lowland Scotsmen, might play a role in indexing this plethora of loyalties and interests. Patterns of intra-writer variability have the potential to elucidate the role of Scots as a sociolinguistic resource in persona-management (Eckert 2012) or audience-design (Bell 2001).

Thus, a series of letters by two prominent Scotsmen involved in the Union debates are analysed. Firstly, to identify any differences between the authors, which may suggest variability in the use of Scots and its possible indexing of political loyalties, and secondly, to identify differences within their collections of letters to uncover variation between specific recipients. Alongside quantitative measurements, snapshots of their correspondence are examined to assess possible motivations encouraging the use of Scots.

## 3 Methodology

#### *3.1 The authors*

The two individuals chosen for analysis are John Hay, the 2nd Marquess of Tweeddale (1645–1713) and George Lockhart of Carnwath (1681?– 1731). Tweeddale and Carnwath represent opposing political sentiments, but otherwise are similar in terms of their backgrounds and range of recipients. This allows us to make a more balanced comparison of their intra-writer practices.

John Hay, the 2nd Marquess of Tweeddale was a politician who became leader of the party known as the *Squadrone Volante*. This breakaway party split from the anti-Unionist Country party in 1704, and while initially opposed to the Union, came to support it after various amendments were made. Despite his prominent role as head of the new party, Tweeddale for the most part seems to have been largely devoid of political zeal, lacking the desire for political mobility and success (Young 2004). He was driven largely by his staunch Presbyterianism and loyalty to the Protestant cause (Patrick & Whatley 2007). His letters are mostly written to members of his own party, but also to correspondents from other political factions and various members of leading families.

George Lockhart of Carnwath was a Jacobite politician and writer, whose memoires of the Union paint a (somewhat skewed) picture of the debates as they took place. Despite his vehement opposition to the agreement, Lockhart was one of the commissioners chosen to negotiate the Union treaty before it reached the Scottish parliament, during which time he operated as a Jacobite spy. Lockhart saw himself as a quintessential Scottish patriot; his Jacobitism was built upon his belief that only the restoration of the exiled King James VII could secure Scotland's independence and ensure the nation's religious salvation (Szechi 2011). Lockhart was firmly aligned with the anti-Unionist Country party, though he also had strong social bonds with the Whig faction in the English parliament (Szechi 2011). He writes to various recipients, including political members of the Scottish parliament, various Jacobite correspondents, and the son of James VII.

#### *3.2 The letters*

A collection of Lockhart's letters were available online, having been digitized by the Scottish History Society (see Appendix for full details), and all original spellings were retained, making this data usable for this analysis. These letters were downloaded as individual text files, tagged for relevant metadata and given a unique identifier. The letters of Tweeddale were available within the *Tweeddale Papers* collection in the National Library of Scotland. These manuscripts were requested and photographed, before being uploaded to *Transkribus* (Kahle et al. 2017) for transcription. The first 100 images were transcribed manually, before a handwritten text recognition model was trained on this data using the tools available in *Transkribus*, with a high degree of success (Character Error Rate = 3.4 %). This model was run on the remaining 122 images, followed by manual corrections, before being exported as text files and tagged for metadata. This included information about the text (place of publication, date of production), the authors (their political affiliation, religious denomination, location of text production) and the recipient (their location and title).

These text files were uploaded to the purpose-built corpus, based within the platform LaBB-CAT (Fromont & Hay 2008). A Linguistic Enquiry and Word Count [LIWC] (Tausczik & Pennebaker 2010) layer was added to enable searches of variable Scots words and their English equivalents across the letters. LIWC was initially designed to assign words to psychologically meaningful categories, but in this instance, an untold number of Scots features can be compiled into a singular category 'Scots', and their equivalents in the corresponding 'English' group. This constitutes the variable – the choice (theoretically) of the writers to use words containing Scots orthographic spellings or lexis, *or* their anglicized equivalents, and allows the researcher to search for *all* Scots and English tokens at once, providing a more holistic approach to writer variability. The need to compile items in this way was partly induced by the scarcity of Early Modern written Scots, and highly individualized patterns of usage for single variants (van Eyndhoven forthcoming). Focusing on a single feature would risk missing a significant proportion of Scots present but scattered across the writings of the authors.

The wordlist of Scots and English features was accordingly uploaded to the LIWC layer manager in the corpus. A query was then run; this tags each instance of a token (in this case a variable feature) as it occurs in the letters as either Scots or English. Once tagging was complete, these 'hits' were extracted from the corpus and compiled along with the accompanying metadata, before being uploaded to R (R Core Team 2020) for quantitative analysis.

#### *3.3 Lexical and orthographic features*

The features included in the analysis consisted of Scots lexical items and words containing Scots orthographic practices. Given the variable nature of both Scots and Early Modern English at this time, there is no straightforward list of items and spellings that could be used. Instead, features must be assessed on a case-by-case basis and added to a custom-built feature list. These were identified by hand-checking the correspondence itself and referencing possible targets in the Oxford English Dictionary and Dictionary of the Scots Language for their etymology, history of usage and regional spread. They were included as Scots if they were labelled as *Chiefly Scottish* by both dictionaries or if they had fallen out of Middle English usage by the 1500s, whilst continuing in Scots. Orthographic variants represented particular spelling conventions that were established in Scotland (and often the far north of England), but not in the standardizing variety emerging in southern England by this time – which was the target variety for anglicization. For each Scots feature, the anglicized equivalent (as the southern English representation of the feature) was included. Table 12.1 demonstrates some examples.

Orthographic variants could consist of a single letter or cluster of characters within words, such as the use of <s> where English has <sh> in words such as *push* and *shall* (*pouss* and *sall* respectively OED 2022). This might also indicate a difference in pronunciation – in this case representing an alveolar rather than postalveolar fricative realization (Kniezsa 1997: 40), though it is difficult to know for certain whether Scots spellings kept pace with phonological developments. Indeed, Aitken (1971: 187) notes the tendency of late Middle Scots to retain established orthographic variants after the phonemic arrangements which they formerly reflected had ceased to exist. Given the uncertain status of phonological representation, all tokens differing by spelling were labelled as orthographic.

Other examples of orthographic variants may be suggestive largely of a difference in spelling conventions, such as the use of <ui> for <oo> in *book* (although its initial appearance may have been to distinguish words with Middle English /u:/ from those with /o:/, for which the *i*-digraph is a particularly Scottish innovation; Kniezsa 1997: 37). Similarly, the use of <quh> compared to English <wh> in relative and interrogative clause markers is a salient and recognizable Older Scots orthographic marker (Kniezsa 1997; Lass & Laing 2016 – though see Laing & Lass's 2019 ana lysis for possibly phonological links). The items were included in the list of variants as the words they occurred in (e.g. the variant <quh> for <wh> was included as *quhen*, *quhere* and *quhich* and corresponding *when*, *where* and *which* in the list).


Table 12.1. Levels of distinction for Scots variants with English equivalents

Lexical variants include words such as *mumping*, meaning 'grumbling', but also words such as *minch* which does not have a one-to-one translation in English, but instead various meanings including 'to minimize in reporting', 'to disparage' or 'to water down' (Dictionary of the Scots Language Online 2022). Lexical items are often unique words tied to a distinctly Scottish institution or way of life, which mean they do not always have a straightforward equivalent. However, they were still included to provide insight into whether 'untranslatable' Scots lexical items survived in this correspondence. Once completed, the feature list contained 796 Scots items, and 740 corresponding English equivalents.1

## 4 Results

#### *4.1 Inter-writer variation*

First the frequencies of Scots lexical and orthographic variants and their English equivalents were measured across Lockhart and Tweeddale, to determine 'how' Scots the authors were. Frequencies were calculated out of total variable count and normalized per 10,000 words. The raw counts are shown in Table 12.2.


Table 12.2. Counts of variants and variable in both authors

1 Fifty-six Scots lexical items had no equivalent, and accordingly no English option was given for these.

Using ggplot2 (Wickham 2016) in R, this data was then plotted on a mosaic plot (Figure 12.1), which is a form of stacked bar chart that allows for easy comparison between unevenly measured groups, representing output in percentages relative to the total observations for each factor level (Jeppson 2021). In the plot, the width of the columns is proportional to the number of observations across each speaker. The vertical length of the bars is proportional to the number of observations *within* each speaker. In this instance, we see that Lockhart has a larger volume of data than Tweeddale, but a proportionally similar level of Scots within his writings.

It is clear the authors are both significantly anglicized, with English variants composing 97 % of all variable tokens. This is not particularly unusual given their high social status and their connections with various high-ranking and English recipients – the social prestige of the English standard naturally lent itself well to social advancement, while English contacts similarly encouraged if not demanded use of the English standard. Yet it is noteworthy that Scots has not disappeared entirely in their writing, nor had it become fossilized to purely 'untranslatable' Scots lexical items

Figure 12.1. Mosaic plot of variants in Lockhart and Tweeddale's correspondence.

and phrases – such as those tied to the independent Scots legal system (Kopaczyk 2012).

Considering possible contributing factors, there is no obvious difference between political sentiment and frequency of Scots: Both Tweeddale and Lockhart exhibit similar frequencies, despite professing opposing political sentiments. In order to untangle other possible causes, a close-up analysis of these individuals and their specific writing practices is needed. Recent analyses have suggested the importance of audience design in analysing historical variation in correspondence (Conde-Silvestre 2016; Hernández-Campoy & García-Vidal 2018), and it is accordingly the role of the recipient that will be analysed here.

#### *4.2 Intra-writer Variation*

#### 4.2.1 John Hay, 2nd Marquess of Tweeddale

Firstly, the frequencies of Scots for Tweeddale across his various recipients were plotted using ggplot2, shown in Figure 12.2. The total number of variants (Scots and English) per recipient (T) are listed along the *y*axis, and counts of the Scots tokens are shown at the end of each bar. The frequencies shown are the proportions of Scots relative to T, normalized per 10,000 words.

Examining Figure 12.2, the number of Scots tokens within Tweeddale's correspondence is clearly low overall, peaking at just sixty-one to the Duke of Roxburgh – Tweeddale's most frequent correspondent. In terms of relative frequency, we see a favouring of Scots in letters addressed to the Duke of Marlborough, Queen Anne, and especially the Earl of Cromartie. At first glance, this patterning seems somewhat surprising. Both the Queen and Marlborough were of higher social status and based in London, and the latter was a member of the English parliament. They thus represent figures Tweeddale might want to impress to secure his fortunes post-Unification. The Earl of Cromartie was a leading incorporationist who professed a wholehearted commitment to a singular British identity, maintaining that

Figure 12.2. Normalized frequencies of Scots variants per 10,000 words across all recipients of Tweeddale.

Scottish nationality was a construct and ridiculing the 'misguided sentimentality and incoherent "sophisms" of false patriots' (Kidd 2008: 79).

Arguably, these factors would induce increased attention to language choices, thereby favouring the emergent English standard and its variants. Yet, that does not seem to be happening here. Instead, this is suggestive of linguistic divergence, in which the speaker, or in this case writer, makes use of linguistic resources to (consciously or subconsciously) move away from the register of their interlocutor (Babel 2010). Tweeddale might be marking out the social distance between himself and these three recipients through linguistic behaviour that diverges from the intended, expected or desired norm. For although Tweeddale eventually came to support unification with England, this did not suggest he wholeheartedly embraced English cultural and social standards, nor did he cast off his Scottish heritage. Those who supported the Union were not ipso facto antithetical to Scottish patriotism (Patrick & Whatley 2007). Indeed Tweeddale was initially opposed to the Union, and cared deeply about Scotland's future. There are unfortunately no meta-linguistic comments in Tweeddale's letters that could support the possibility that Tweeddale's Scots variants index a national, Scottish identity. Nonetheless, this increased use of vernacular, lower-prestige variants to high-status but socially distant recipients reflects a plausible and tangible connection to divergent linguistic practices.

However, frequencies alone cannot elucidate the intricacies of written communication. Instead, examining some extracts from Tweeddale's correspondence could give an indication of how Scots features cluster. The first of these is to the Earl of Cromartie, written 2 June 1704 (Scots features underlined).

I shall onlie mind yow yt there be a generall one about Religion and ye Church as is usuall […] to adjourn the parliament upon yt account seems not so verrie proper and would give ground of reflection, as it did last which I would avoid, and therfor desires a new order may be sent doun for it till ye 6th of July

Here, Tweeddale 'reminds' Cromartie, who had joined his breakaway party in 1704, of the upcoming session in parliament concerning the Kirk, using the Scots lexical feature *mind*. This was personally important to Tweeddale – his political goals were driven largely by the desire to protect the Presbyterian Kirk. Cromartie, on the other hand, argued the differences in church government between Scotland and England should not stand in the way of the Union, though he hoped its successful passing would ameliorate the position of his fellow Episcopalians (Kidd 2008). The two thus did not see eye-to-eye religiously, and it is doubtful whether use of Scots would have had any currency with Cromartie, who was in fact a leading advocate of a singular British language – which was to be English, not Scots (Kidd 2008). The occurrence of Scots might consequently reinforce the divergent linguistic, political and religious ideologies between them.

Turning now to Tweeddale's correspondence with the Queen, dated 30 May 1704, we can similarly observe use of Scots in a politically contentious context. Here, Tweeddale has proposed alterations to the Union agreement sanctioned by the Queen, and seeks her approval.

[…] I presume humbly to lay befor your Maj: yt there are sume alterations propos'd to be made in your present ministry previous to ye sitting doun of your parliament in pursuance of ye generall ones came doun with my Ld Chan.. preventing a great dale of heat and disorder in ye House that oyrwayes cannot be avoided. I trust they will be found verrie agreable and so hopes your Maj: will pardon the liberty I take to presse ym on this occasion as that which will contribute most to ye promoting your service.

Here Tweeddale explains his motivations, suggesting he has made 'sume alterations' which will prevent 'a great dale of heat and disorder in ye House', and accompanies this by Scots spellings for *some* and *deal* respectively. Despite the appeasing sentiments professed, such an act was socially and politically challenging, given it indirectly challenged the Queen's authority, and was likely to incur her displeasure. Tweeddale's behaviour does not align with the Queen's expectations, and again we see Scots features clustering around such an act. Furthermore, it is not the case that Tweeddale was wholly unaware of the corresponding English variants for these features (though some are dispreferred). Table 12.3 summarizes the frequencies of these variants across Tweeddale's collection of letters.

The fact that we observe higher proportions of Scots primarily to correspondents who were based in England or strongly in favour of unification is an unexpected pattern, but could be indicative of the increased social and linguistic distance characterizing these recipients, unlike many of Tweeddale's Scottish correspondents. This might reinforce the sociopolitical cause Tweeddale is writing on behalf of, to those most removed from it.


Table 12.3. English equivalents to the Scots features identified in Tweeddale's extracts and frequency across the Tweeddale collection

To explore this trend observed in Tweeddale's correspondence statistically, recipients were grouped together into the following categories – *Court* (the pro-Union party in the Scottish parliament), *Squadrone* (the party headed by Tweeddale), *Country* (the anti-Union party), *Other* (for members of the clergy or unknown) and *English* (for recipients based in the English parliament). From this, a conditional inference tree or *ctree* (Breiman et al. 1984) was grown on Tweeddale's variable usage and analysed in terms of political party. The algorithm searches for statistically significant clustering within the data across different factors, and represents this visually in tree form with binary branching. The output is represented in the terminal nodes, indicating the proportions of the variants. In this case Scots is represented by the dark grey shading and the English equivalents by pale grey. The ctree in Figure 12.3, grown using the partykit (Hothorn et al. 2020) package in R,2 shows Tweeddale's proportions of Scots across groups of recipients.

Although the proportions are low overall, the ctree still indicates a statistically significant (p < 0.01) preference for Scots to those who were

Figure 12.3. *ctree* for Tweeddale's use of Scots by participant political affiliation.

2 The minbucket was set to 50, the mincriterion to 0.95 and maxdepth to 3.

English. Regardless of whether Tweeddale's remaining Scottish recipients supported or opposed the Union, or formed part of his breakaway party, his use of Scots did not vary significantly among them. Instead, the biggest contributing factor seems to have been whether they were English or not, suggesting Scots was preferred towards correspondents who were most geographically distant (albeit by birth or upbringing), and therefore also most socio-politically and religiously distant to Tweeddale. Tweeddale's highest-ranking recipient Cromartie (as we observed earlier) *was* Scottish, but given his position as one of the leading and most advanced incorporationists of the Scottish parliament (Kidd 2008) and his self-identification as a Briton rather than a Scot (Whatley 2006), it is perhaps not surprising that Tweeddale's Scots usage towards him mirrors those of English recipients. This statistical patterning lends support to the possibility suggested in Tweeddale's extracts. Scots is frequently directed towards recipients most distant from Tweeddale and is perhaps acting as an identity-building marker, reinforcing his firm loyalty to Presbyterianism and his country of birth. Indeed, Tweeddale writes explicitly about such sentiments to Marlborough, in his letter dated 27 October 1703: 'we did what we thought for ye interest of our Countrey, which I am confident no true hearted Englishman can blame us for'. Tweeddale appears to be laying claim to nationalist motivations for his actions, and Scots may be part of this strategy.

#### 4.2.2 George Lockhart of Carnwath

The same analysis was then applied to George Lockhart, given he represents a different political and ideological persona from Tweeddale. Figure 12.4 shows Lockhart's proportions of Scots by recipient, with total count of the variable (T) along the *y*-axis and Scots counts at the end of each bar.

As Lockhart, like Tweeddale, used very little Scots overall, these results indicate slight tendencies rather than highly robust practices. Indeed, the highest-ranking recipients see less than ten Scots tokens each. The top ranking recipients include Lord Ormistoun, the Bishop of Rochester, Mungo Graham and the Duke of Montrose. Interestingly, all four recipients

Figure 12.4. Normalized frequencies of Scots variants per 10,000 words across all recipients of Lockhart.

are again socially and politically distant to Lockhart. Lord Ormistoun was a Court party member in support of Union. The Bishop of Rochester was an English clergyman based in Cambridge, and a favourite of William of Orange. Mungo Graham took the initiative in persuading the Squadrone to support the Union and was the personal assistant and right-hand man of the Duke of Montrose, who, contrary to Jacobite hopes, became a steady supporter of the Protestant succession (Sunter 2004). This bitterly disappointed Lockhart, who denigrated Montrose as being 'of an easy, mean-spirited temper […] and extremely covetous' (Lockhart 1817: 119). Montrose was also the political rival of Argyll, the close friend and ally of Lockhart.

These four recipients thus appear almost antithetical to Lockhart's political and personal allegiances: They represent pro-Unionist, incorporationist and anti-Jacobite standpoints. This might again suggest the role of Scots in reinforcing conflicting ideologies through divergent linguistic behaviour (Babel 2010), perhaps underlining Lockhart's staunchly patriotic and nationalistic agenda. Lockhart was openly Jacobite and the political ideologies underpinning his self-identification were heavily infused with nationalistic sentiments, support for the exiled king and a constructed 'Scottishness' – features that may have lent themselves well to use of written Scots, if it had become a marker of Scottish identity by this time.

Examination of an extract from one of Lockhart's letters to these topranking recipients could add further insight to his intentions behind his stylistic choices. The letter below is written to the Bishop of Rochester on the 4 October 1727:

This does not come through a canall that will allow me to put such ane interpretation upon it and I belive it will be expedient that your Lordship lose no time in forwarding this, or at least the substance of it, to him. I need not I belive take notice how necessary it is that this affair be keept in as few hands as possible […] I need not, I hope, make any appology for communicating this to your Lordship, since it proceeds from no other motive but a personall respect for yourself, as well as the good of the common cause. Tho I make use of a borrowd subscription, nothing can be more genuine than my profession of being in the most respectfull manner. [P.S.] I hope that you'l take care that the accounts you transmit come directlie into the King's own hands.

In this extract, Lockhart makes a demand upon Rochester, stressing the urgency of the matter and ordering the bishop to 'lose no time' in acting upon his request and to 'take care' that the accounts reach the desired recipient. Immediate action is required, and Lockhart is clearly in haste. While Lockhart's letter contains some of the expected politeness formulae, such as reassuring Rochester of his 'personal respect', the overall tone is rather importunate, and Lockhard suggests he 'need not […] make any apology' for his ordering. He also makes use of a borrowed subscription (the concluding clause or formula of a letter with the writer's signature, which served to acknowledge the recipient's status and relationship to the writer; see Nevala 2005), rather than providing his own, thus not


Table 12.4: English equivalents to the Scots features identified in Lockhart's extract and their frequency across the Lockhart collection

personally acknowledging the power and status relations between the two. The communicative style and format seems to mirror their divergent political attributes, marking out the correspondents' disparencies and concurrently this is accompanied by a clustering of Scots features. Lockhart, like Tweeddale, was variable in his use of these particular features as well (see Table 12.4).

Finally, to statistically analyse Lockhart's behaviour, recipients were grouped by political party (with the same categories as those used for Tweeddale), and a ctree was grown on his variable usage.3 There was no statistically significant separation in the data in this case. However, when recipients were grouped according to their political positioning a statistically significant split did emerge in the ctree. Factor levels included *Pro* – which includes both Court and Squadrone party members, *Anti* – which includes Country party members as well as Jacobites, *English* – for English nationals, *Unknown* – Scottish recipients who were not involved in the debates (including women), and *Neutral* – recipients too young at the time of the debates to have been politically involved, but with ties to politically involved families (see Figure 12.5).

The ctree indicates a slight preference in the use of Scots towards recipients who were English or were in support of the Union. The difference is minor, though the split is statistically significant (p < 0.02). This adds support to the idea that Lockhart is demonstrating divergent linguistic

<sup>3</sup> The minbucket was set to 200, the mincriterion to 0.95 and maxdepth to 3.

Figure 12.5. *ctree* of Lockhart's variable usage by the political positioning of his recipients.

behaviour to those most removed from him personally and politically, similar to what we observed earlier with Tweeddale.

#### *4.3 Summary of results*

Despite degrees of difference in their political, religious and ideological attributes, both authors reflect a tendency to use more Scots with recipients they were removed from socially, politically or professionally. While their political standpoint does not appear to have been the leading influence on their use of Scots (as Lockhart demonstrates most noticeably), political ideologies do play a non-trivial role when these authors are in correspondence with their social network. Indeed, examining these patterns it appears written Scots not only manifests, but might actively accentuate social relations with more distant recipients, in correspondence.

This is suggestive of the changing nature of Scots, in which it had gone from a fully fledged written medium to a selective linguistic resource, utilized (whether overtly or subconsciously) in combination with other rhetoric devices to delineate the distinctions between correspondents. In the process, such practices may have served to centre the authors within the political viewpoint they were writing from, perhaps juxtaposing themselves to the correspondents most removed from the local and nationalistic ideologies similarly guiding Tweeddale and Lockhart, despite their different voting behaviour.

We must also consider the possibility that these Scotsmen may have been staking their claim to a Scottish identity, constructing a persona grounded in local affairs when corresponding precisely with the recipients antithetical to this construct, and the ctrees add promising weight to this idea. Contemporary sociolinguistic studies have highlighted speaker agency in utilizing variation to distance oneself from ideologies that do not align with personally constructed identities (Babel 2010; Eckert 2012), and perhaps we can observe as much in this historic scenario. However, without metalinguistic comments from the authors themselves such possibilities are difficult to determine, thus this remains a tempting hypothetical. What is clear is that Scots has not disappeared by this time period, nor is it fossilized to set lexical bundles and phrases, as has been identified in other text-types (Kopaczyk 2012). Rather, it appears to have become intertwined into a linguistic repertoire that encompasses English and Scots options.

## 5 Conclusion

Although results are drawn from a micro-analysis of two authors, some perspectives on possible uses of written Scots in the early eighteenth century have surfaced nonetheless – interesting trends that are less visible from a macroscopic perspective that involves broader search queries and ignores intra-writer practices. These hypotheses can only be strengthened with a more detailed exploration within the intra-writer plane of variation combined with a multi-factor corpus analysis (van Eyndhoven in preparation). Nonetheless, the rich textual and personal histories bound within such correspondence are already appearing. Reflecting on early eighteenth-century Scots, such results suggest the complex cultural dualism taking shape transcended beyond purely legal and political decisions, to the sphere of language as well. What is presented here is a first step, and further investigation will surely uncover additional insights into factors conditioning the use of written Scots at this time.

## References


### *Appendix: Political authors and texts analysed*


#### Nora Dörnbrack

13 Variation in verbal inflection in the private writings of the Scottish emigrant Mary Ann Wodrow Archbald (1762–1841)

#### Abstract

This chapter discusses longitudinal variation in the private writings of the Scottish emigrant to America Mary Ann Wodrow Archbald (1762–1841). An investigation of syncretism of preterite and past participle forms, irregular preterites and verbal -s for all persons sheds light on the effects of normative rules at a time when written English was being codified. Possible contributing factors are contemporary prescriptive tendencies, the speech communities in Scotland and America as well as Archbald's lifelong interest in literature.

## 1 Introduction

This chapter contributes to the growing research on intra-writer variation within the framework of historical sociolinguistics and language history 'from below' (cf. Elspaß 2005), in particular to longitudinal variation over the course of an individual's lifespan in the Late Modern English period. Previous studies on historical writers have established that the idiolect of an individual can change not only in earlier phases of language acquisition, but throughout a person's life and into old age (see, e.g. Tieken-Boon van Ostade 1994; Raumolin-Brunberg 2005, 2009). Investigations seeking to analyse years or decades worth of writing by the same writer tend to favour texts by adult, educated, male writers, since material from children, minority speakers or women is less readily available (Kiełkiewicz-Janowiak 2003: 331; Raumolin-Brunberg 2009: 172). Kiełkiewicz-Janowiak (2012: 325) describes the situation pointedly:

[I]t is a major drawback of historical sources that they under-represent the less literate members of society, such as lower rank speakers and women. One way to circumvent this would be to look closely into the (often unpublished) texts authored by the socially backgrounded and read their stories. More generally, reading complete textual records of individual lifetimes is rewarding, and adds to our understanding of the (linguistic) developments of an individual over a lifespan, which so far have been under-appreciated.

Within this context, this chapter investigates the linguistic practices of just such an individual – a woman emigrant farmer who did not receive any formal education (see Section 2.2). Despite what her background might suggest, the Scottish emigrant to America Mary Ann Wodrow Archbald (1762–1841, henceforth 'Archbald') was an ardent reader of contemporary literature with a keen interest in writing letters and journals. She does therefore not fit into a one-dimensional societal category – she neither belonged to the barely literate, lower social classes nor to the higher ranks of society who might have granted their daughters a basic formal education. The fact that she is an emigrant woman sets her further apart and makes her writing the more appealing (cf. Hickey 2019). The case of Archbald is particularly interesting as she wrote at a time when written English was being codified in both Britain and North America. This raises the question if Archbald adopted any of the linguistic rules laid out in contemporary grammars and whether her doing so may explain any intra-writer variation.

Archbald's extensive writings were previously explored from a literary and historical perspective, first by Scott (1995) who wrote her dissertation on the reading life of Archbald, and later by Gerber (2006) who explored nineteenth-century emigrants' correspondence from a historical perspective with Archbald as one of four case studies. Neither analysed Archbald's linguistic practices.

The chapter is structured as follows: Section 2 presents background information on standardization processes in late eighteenth-century Scotland, as well as Archbald's life and the texts she produced over more than 45 years. Section 3 focuses on three non-standard variants concerning verbal inflection which can be found in Archbald's writings, namely syncretism in preterite and past participle forms, irregular preterites and verbal -*s* for all persons. In Section 4, I discuss the results against contemporary prescriptive tendencies, the speech communities which surrounded Archbald in both Scotland and America, as well as her lifelong interest in contemporary literature. The chapter concludes in Section 5.

## 2 Background

#### *2.1 Linguistic debates in late eighteenth-century Scotland*

Hickey (2010: 1) notes that '[t]he most prevalent wisdom about the eighteenth century' was the emergence of prescriptivism, that is, the idea that a prescribed norm shall apply to every speaker and writer alike. From 1750 onwards, grammars were printed in hitherto unseen numbers (Beal 2004: 90), lists of Scotticims which the reader should avoid were published (Dossena 2019: 29), and linguistic debates intensified especially in Scotland (Jones 1994: 71). Standardization debates in Scotland were additionally closely intertwined with efforts to anglicize Scottish varieties in both speaking and writing (Dossena 2003: 383).

Generally speaking, opinions ranged from the full extermination of Scottish features to a middle ground where only certain features should remain in usage, to the promotion of a Scottish Standard with its own alphabet and orthography. Among those who wished for a standard usage where little to no trace of Scots was to be found was David Hume, who, in 1752, published one of the first lists of expressions peculiar to Scotland which he advised his readers to avoid (Dossena 2019: 29). A number of similar lists were published throughout the latter half of the century, among others, James Beattie's *Scoticisms, arranged in Alphabetical Order, designed to Correct Improprieties of Speech and Writing* (1779), which was one of the first works to distinguish between covert and overt Scotticisms (Dossena 2019: 30), and John Sinclair's *Observations on the Scottish Dialect* (1782).

These works were countered by publications which argued for the use of what might be termed 'Standard Scots', while others still were 'far from being apologetic for any perceived "impropriety" or "impurity" of Scots, actively advocated its use, promoted its survival and pointed to its long and legitimate historical pedigree' (Jones 1994: 72). Alexander Geddes, for example, published an essay in 1792, which Dossena (2005: 85) describes as 'an aggressive attempt to justify the use of Scots', and in which he proposed a modified alphabet and orthography that would fit the Scottish tongue better than the English spelling did.

Those two opposing views – one in favour of using a southern English standard and moving away from Scottish forms, the other fiercely in favour of retaining a peculiarly Scottish form of expression in writing – were easily blurred, so that contrasting ideas about the usage of Scottish English might even be expressed by the same person. James Beattie, for example, published the above-mentioned guide on Scotticisms in 1779, which suggests that it was undesirable to speak and write a variety that was at odds with a southern standard. Interestingly, it was also Beattie (1788: 91f.), who, nine years later, issued the following statement in which he advocates for linguistic tolerance:

To speak with the English, or with the Scotch, accent, is no more praiseworthy, or blameable, than to be born in England, or Scotland: a circumstance, which, though the ringleaders of sedition, or narrow-minded bigots, may applaud or censure, no person of sense, or common honesty, will ever consider as imputable to any man.

The overall majority of contributions in this debate argued against the use of forms which would indicate a speaker's or writer's origin that was geographically and/or socially removed from centres of power – this was common to the degree that Scottish forms were only deemed acceptable in literature (Dossena 2019: 26).

## *2.2 The life of Mary Ann Wodrow Archbald*

Mary Ann Wodrow was born on 26 June 1762 as the youngest daughter of Reverend Robert Wodrow (1711–84) and his second wife Ann Ruthven (1719–1814). She spent her childhood, adolescence and early adulthood on Little Cumbrae, an island in the Firth of Clyde, off the west coast of Scotland, surrounded by her immediate family (Scott 1995: 29). Although there are no records pertaining directly to her childhood, it can be assumed that she was taught at home, since she commented twice on her lack of any formal education. On 4 August 1789, Mary Ann Wodrow married the tacksman and sheep farmer James Archbald (1763–1824) and together they had ten children of whom only four survived infancy. In 1804, the rent on their farm increased significantly, and the family was no longer able to sustain a living on the island. After fruitless attempts to find farmland on the mainland or on the adjoining islands, the family instead decided that emigration to America would promise better future prospects for their children and their place in society.

In late March 1807, Mary Ann Archbald, her husband and their four young children crossed the Atlantic, and, after resettling once, they moved permanently to a farm called *Riverbank*, near Auriesville, New York. Archbald described her life there as isolated, since her Dutch-speaking neighbours preferred to switch from English to their own first language on many occasions, thereby excluding her from conversations. In her letters to her cousin Margaret Wodrow, Archbald described a strong longing to return to Scotland, either for a visit or permanently. But even though James and Mary Ann Archbald were on the verge of returning several times in earnest, they never decided to cross the Atlantic once more. Archbald died on her farm on 3 January 1841, aged 78 (Scott 1995: 61).

#### *2.3 Data and methods*

The number of ego-documents which have been preserved from Archbald's private writings is unusually large. Smith College (Massachusetts) holds the largest collection of her writings. The collection includes, among others, her letter books (i.e. books into which Archbald copied outgoing letters), journals, commonplace books, watercolour drawings and poetry, all of which have been donated by Archbald's descendants. The present study includes only her letter books and journals in addition to six original letters from other collections.1 Letters in particular have proved to

1 Namely four letters from the James Archbald papers, Lackawanna Historical Society Library, New York State, as well as two letters from the Coolidge Collection of Thomas Jefferson Manuscripts, Massachusetts Historical Society.

be a viable source for longitudinal analyses of individual patterns of usage, even though Raumolin-Brunberg (2005: 46f.) found that intra-writer variation is greater between genres than within the same genre but with different recipients. The decision to include several genres to trace variation in verbal inflection in the current study is partly owing to the aim to cover the greatest possible time period and to ensure that the material is as comprehensive as possible.

Taken together this material amounts to roughly 440,000 words and spans over forty consecutive years in addition to five years from the last decade of her life. The material has been transcribed by Scott as part of her dissertation in 1995, and Archbald's journals and letter books are digitally available through the online database *Gale Primary Sources – Nineteenth Century Collections Online*. 2 The original letters have been digitized on request at the respective archives and have been transcribed by myself. Figure 13.1 shows how the material is distributed over the course of Archbald's lifetime.

Figure 13.1. Number of words per text type per year (N = 435,549).

2 Cf. <https://www.gale.com/intl/primary-sources/nineteenth-century-collecti ons-online> accessed 12 July 2022.

As can be seen from Figure 13.1, the material for this study is greatly skewed towards Archbald's earlier years of writing, in particular the time prior to the birth of her first son in 1790. Following the sudden death of this infant son, her entries become more sparse, so that the amount she wrote steadily decreased until the year of her migration (1807). Both text types that are available from this period, that is, journals and letter books, are affected by this decline in productivity.

Archbald ceased to keep a journal shortly before migrating, while she continued to use her letter books without interruption. The select excerpts of letters that she copied into her letter book in America do not exceed 12,000 words per year, but they give a picture of her everyday activities and the relationships she sustained with family members in Scotland. The year 1825 marks the start of a thirteen-year gap in the material that is only bridged by four original letters that Archbald addressed to her oldest son. The closing item of the present corpus is Archbald's final journal which she started in 1839, when debilitating illnesses and old age hindered her from making regular and extensive entries as had been her habit in her youth.

The uneven distribution of Archbald's writing over time is a characteristic of all naturally occurring written data. The amount that Archbald wrote might have been influenced by childbirth, the feeling of isolation after migration or infirmity in old age. A personal preference for one text type over another might lead to an imbalance between genres, and this imbalance might be even further exacerbated as not all of her texts have survived to this day, and of those which did, not all are available for research. The main focus of the present study is therefore qualitative and interpretive which enables me to reflect on the irregular distribution where it is necessary. In addition, I employ absolute frequencies in cases where the material allows for such an analysis. Individual occurrences of variant forms have been retrieved by using the corpus analysis tool AntConc. The results have been assessed manually.

## 3 Longitudinal variation in verbal inflections

In this section, I present three different features of northern, nonstandard verbal inflection which were discussed in contemporary grammatical works and which can be traced in Archbald's texts: syncretism in preterite and past participle forms (Section 3.1), irregular preterite forms (Section 3.2), and, finally, the use of verbal -*s* for all persons (Section 3.3). Each of these features shows a different development over the course of Archbald's life. Therefore, I discuss specific reasons that might have given rise to each emerging pattern in the respective subsection, while overarching factors that help explain the variation are explored in greater detail in Section 4.

#### *3.1 Syncretism in preterite and past participle forms*

The use of a preterite form in place of a past participle form in irregular verbs is well attested for eighteenth-century writers even beyond the Scottish border (Gustafsson 2002: 275). Contemporary prescriptive guides like that of James Beattie (1779), which pertained to Scottish speakers and writers more specifically, likewise note the syncretism of preterite and past participle forms in irregular verbs: 'The glass is *broke. –* Broken. – Broken, written, &c. are participles. Broke, wrote, &c. are preterites; […] This is not always attended to by English writers, but it deserves attention' (Beattie 1779: 5).

Archbald does not generally employ the preterite instead of the past participle in her private writings, though it does happen occasionally. Except for the verb *write*, the examples below occur only once in this form in her journals and letter books:


(3) I have wrote Wms Aſsignation on a stamp & wish it were signed ( Journal 7, 10 February 1798)

The verb *write* presents a special case in Archbald's texts. While most other preterite forms of irregular verbs occur only once in the place of a past participle, *wrote* occurs sixty-four times. More specifically, Archbald used the past participle form *wrote* both following auxiliary *be* in passive voice (4), following auxiliary *have* in perfect aspect (5) and adjectivally modifying a noun (6).


Figure 13.2 shows the distribution of *wrote* and *written* used as past participle forms.

Figure 13.2. Past participle forms of *write* including passive voice, following auxiliary *have* in perfect aspect and adjectivally modifying a noun (N = 97).

Figure 13.2 illustrates that Archbald used *wrote* as her past participle form almost exclusively prior to her immigration to America, while the form *written* – occurring thirty times in total – dominates thereafter. There seems to be a brief period shortly before migration, from 1804 to 1806, in which she used both forms alongside each other with no clear preference given to either of them. Both *wrote* prior to 1807 and *written* after 1807 are used in all the available text types from those periods, so that the switch is most likely not caused by the abandonment of her journals after migration.

This finding is in line with the results of Gustafsson (2002: 272), who found that *wrote* as a past participle form was prevalent in the private letters of educated writers during the latter half of the eighteenth century. Her finding holds true even across genres in Archbald's writing, as the use of *wrote* as a past participle in her journals clearly outnumbers the instances where she employed *written* instead. Gustafsson (2002: 272f.) concludes that the high number of occurrences of this feature might also be part of the reason why contemporary grammarians argued at length against this variant. It remains unknown whether Archbald ever saw a prescriptive guide that mentioned this particular usage as she left no comment in her journals or letter books that would indicate she did, nor is any prescriptive list or guide to be found in the index to her personal library (available in Scott 1995: Appendix I). That her use of *wrote* as a past participle form changes rather abruptly after more than 20 years of constant usage, and that this switch in addition conforms to contemporary practices, indicates that she experienced at least some impetus to conform to a southern standard variety.

It is furthermore interesting to note that Noah Webster included an entry on *wrote* in the first edition of his *American Dictionary of the English Language* (1828), which reads:

WROTE, *preterit tense* of write. He *wrote* a letter yesterday. Herodotus *wrote* his history more than two thousand years ago. [Note. *wrote* is now used as the participle.] (Webster 1828, s.v. *wrote,* italics in the original)

Webster's dictionary was published towards the end of Archbald's life, but she did not seem to be influenced by this as w*rote* does not re-appear as a participle in the (admittedly sparse) material from the time after 1828. Webster himself did not repeat this usage of *wrote* in his grammar *An Improved Grammar of the English Language* (1831: 78), where the past participle form of *write* is given as *writ, written* – which again aligns with Archbald's later usage. Other American grammars that might potentially have been available to Archbald likewise explicitly proscribe *wrote* used as a past participle form (e.g. Bingham 1791 [1785]: 35, Humphreys 1792: 64).

#### *3.2 Irregular preterite forms*

The pattern described in Section 3.1, where Archbald first used a variant form that does not conform to the standard variety before adopting the latter eventually, can be detected in her use of irregular preterite forms as well. Again, a look into contemporary prescriptive guides gives an indication which forms might be expected in her writings. John Sinclair wrote in *Observations on the Scottish Dialect* (1782: 69, italics in the original) that '*There are also many false formations in the* Scottish *dialect, which ought to be avoided; as* […] *Keept, sweept.* Kept, swept. *Keeped.* Kept. […] *Catched.* Caught'. Forms like *keept* and *sweept* follow the regular inflection in Scottish English forming their preterite with the bare stem and the regular -*(i)t* ending (Murray 1873: 205–08). Examples like *catched* refer to the practice – common in both non-standard usage and Scottish English – of using the regular -*ed* ending for the formation of otherwise irregular preterite forms (Beal 1997: 354).

Both practices can also be found in Archbald's writing. Some of the examples given in Sinclair's list, like *keept* and *catched* occur frequently in her texts, as (7) and (8) show. Other forms that Sinclair did not record, like *slipt* in (4), or *eat* used as a preterite (Beal 1997: 354) or *sleept/sleeped*, are nonetheless known non-standard preterite forms of the time.


The more frequent verb *keep* is suitable for a more specific analysis. Figure 13.3 depicts how Archbald initially varied between using *keept* and *kept*, before abandoning the former variant after 1788. Similar to the pattern that emerged for the use of *wrote* as past participle, she used the non-standard form only during the earlier period of her writing, and eventually ceased to use it in favour of the standard form. The main difference between those two patterns consists in the dominance of *kept* compared to the non-standard form *keept* which quickly disappeared after a few years of usage.

Yet another pattern becomes visible in Figure 13.4 which shows the occurrence of the preterite forms of *catch* over time. Again, her earlier

Figure 13.3. Preterite forms of *keep* (N = 71).

Figure 13.4. Preterite forms of *catch* (N = 23).

writings show both the standard and non-standard forms. It seems Archbald favoured *caught* after she had migrated, only for *catched* to make a sudden reappearance in 1840.

None of the variant forms discussed so far received attention in Webster's dictionary, but in his grammar, Webster (1831: 75) listed *catched* as a possible preterite and past participle form. This might indicate that Archbald encountered *catched* in America, potentially leading her to take up this variant form once more later in life.

The overall tendency when it comes to the use of preterite forms of irregular verbs seems to be that both non-standard and standard forms exist in the material prior to migration, and they might even co-occur for a certain period, before the standard form prevails in the end. However, this is only a tendency, and individual cases, like that of preterite forms of *catch*, might differ.

#### *3.3 Verbal -s for all persons*

The final feature to be discussed concerns the practice to apply the -*s* ending to verbs not only in the third person singular, but in fact in all persons and numbers. Beal (1997: 356) adds that this feature is prone to occur with third person plural subjects. In Scottish varieties of English, the Northern Subject Rule governs further that the verb ends in -*s* if it combines with any subject other than an adjacent personal pronoun (subject proximity constraint and NP/PRO constraint; McCafferty 2003: 109–10). (10) illustrates the subject proximity constraint where the first verb 'do' does not get verbal -*s* since it is directly adjacent to the personal pronoun subject 'they', in contrast to the following two, nonadjacent verbs 'causes' and 'makes'. (11) shows the NP/PRO constraint, as verbal -*s* can be applied to the verb 'talks' which is directly adjacent to the subject consisting of a noun phrase.

(10) they do not allways contribute to our happyneſs but often v causes us to feel an imaginary uneasineſs & makes us despise what gave us pleasure in the days of Youth & Simplicity. ( Journal 1, 1 January 1785)

(11) some People talks of indavouring to forget our Friends that are away (Letter book 1, Mary Ann Wodrow Archbald to Margaret Wodrow, November 1793)

Archbald used verbal -*s* with all persons – the only exception being the second person plural for which no occurrence is attested in the material. This feature, though, is difficult to trace in her writings, as the corpus is not parsed and the verb seldom follows the subject directly, as can be seen in (12). (12)–(16) show that Archbald used verbal -*s* for all singular persons and two of the plural ones, even in cases where a pronoun is directly adjacent to the verb as in (13) and (14).


Occurrences of this kind can be found both prior to and after migration. Two highly frequent verbs – present tense *have* (*has*) and past tense *be* (*was*) *–* exemplify the development of this feature. Figures 13.5 and 13.6 illustrate all occurrences of *has* and *was* with persons other than the third person singular as well as the first person singular in the case of *was*. Figures 13.5 and 13.6 show both lexical and auxiliary use to highlight that this distinction did not affect usage over time. In total, 22 out of 1,031 instances of *has* fall under this category, as do 72 out of 5,800 occurrences of *was*, compared to 2,880 instances of *have* and 1,360 occurrences of *were*.

It becomes clear from Figures 13.5 and 13.6 that Archbald used verbal -*s* in both *have* and *be* throughout her lifetime. However, those instances constitute rather the exception than the rule, as Archbald used *have* and *were* respectively with all remaining persons to a considerably larger degree throughout the period under investigation. The main difference is that *has* occurs more often as an auxiliary verb, whereas the majority of occurrences of *was* are as a lexical verb. While the overall tendency shown so

Figure 13.5. Verbal -s for all persons other than third person singular (*has*, N = 22).

Figure 13.6. Verbal -s for all persons other than first and third person singular (*was*, N = 72).

far indicates a decrease in the use of variant forms, verbal -*s* continues to occur up to the final years of her writing. Although the absolute number of occurrences decreases over time, this can be explained with the decrease in available material from the latter half of Archbald's life.

An explanation for Archbald's sustained use of the verbal -*s* despite the decline of other non-standard features might be found in contemporary grammarians' view of this particular usage. According to Beal (1997: 356f.), this feature is typically not included in prescriptive guides of Scotticisms from the eighteenth century, probably because it followed a detectable pattern and was thus deemed acceptable. This lesser degree of stigmatization might have guided Archbald's attention away to other, more fiercely prescribed features of her time. Several American grammars, though, comment on the expression *you was* (e.g. Mennye 1785: 79; Bingham 1791: 50; Alexander 1795: 53). However, Archbald did not change her use of verbal -*s* for all persons after migration, leaving it uncertain if she was aware of any of the American grammars mentioned here.

## 4 Discussion

In discussing the results presented in this chapter, it is not possible to make generalizations based on one individual writer for whom 'writing was a major interest and perhaps even daily occupation' (Tieken-Boon van Ostade 1994: 179). The aim of this study is therefore not to describe general processes that lead to intra-writer variation, but to highlight and discuss the actual occurrence of variation over the course of the lifespan of one emigrant writer.

The results of this study can be seen through the lens of Raumolin-Brunberg's observation that 'it is less easy to change one's language in qualitative terms than make shifts in the proportions of already acquired alternatives' (2009: 192). In other words, a change in frequency of existing alternative variants over time is more easily accomplished than a transformation of the writer's underlying grammar. As the examples in Sections 3.1–3.3 show, Archbald had in all cases two variant forms which occurred alongside each other at times. The changes that can be observed over her lifetime reflect rather changing frequencies of already established variant forms, not the introduction of an altogether new variant. In further analysing the present findings, I discuss broader, societal trends that might have influenced Archbald's writing practices, before zooming in on the speech community that surrounded her after migration, and finally I turn to the individual level when I discuss her literary interests.

The period in which Archbald learned to read and write and produced the majority of texts that are available today, is known as one where the pressure to conform to the standard variety increased. Archbald might very well have been sensitive to these prescriptive trends, as she wrote in 1785, upon receiving a letter from a suitor:

(17) the stile was simple if not vulgar & the spelling not good […] for a man to Write well is an accomplishment which I could not dispense with – I made no alowances for the disadvantages of Education or for want of Practice – ( Journal 1, 27 November 1785)

That Archbald was not unaffected by prescriptive tendencies can be seen in the overall trend which becomes visible if her usage of different verbal inflection patterns is viewed over time. Generally, usage of the nonstandard form decreases significantly in favour of the prescribed, standard form after an initial period where both forms occur side by side. The exception to this general pattern is the usage of verbal -*s* for all persons, but since this particular feature was not stigmatized to the same degree in Scotland, it might have escaped Archbald's attention.

Furthermore, Gustafsson (2002: 193), who investigated hand-written documents from the period 1760–1790, comes to the conclusion that 'in private writing [preterite and past participle] variants recede by the end of the eighteenth century essentially'. Taking Gustafsson's results into account, Archbald's transition from, for example, past participle *wrote* to *written* around the start of the nineteenth century, might be in line with the overarching trend towards less variation in preterite and past participle forms in private writing more generally.3 Overarching prescriptive trends might thus have left their mark on Archbald's use of verbal inflections over the course of her life.

3 A possible influence might therefore also have been incoming letters which show less variation. This, however, remains uncertain since no incoming letters have been studied as part of this investigation.

Another possible explanation for the receding use of variant forms can be found if we zoom in on the speech communities that surrounded Archbald prior to and after migration. The pressure mounted against speakers and writers from Scotland to conform to a southern standard has been discussed above. What remains is the speech community that surrounded Archbald after she departed for America. In 1810, Archbald explained her desire to return to Scotland partly with 'the cold hearted craft & unfriendly manners of our neighbours (who are all Dutch)'. Her letters back home over the following years give the impression that Archbald led a secluded life, with her social circle largely restricted to her immediate family, relatives from New York who visited her and a few select acquaintances. She does not seem to have had contact to other Scottish immigrants on a regular basis, since Archbald wrote in 1821 that her two youngest children, who had been 4 and 2 years old, respectively, when the family emigrated, did not retain a Scottish variety in speaking. She mentions in the same letter that her two oldest children made a conscious effort not to lose their Scottish dialect, so that the impression arises that the family primarily had contact with speakers of English from outside of Scotland. That Archbald over time sustained a lower level of overall Scottish verbal inflections in her writing, might be connected to the smaller number of Scottish speakers that she interacted with on a daily basis in America.

Finally, on the individual level, Archbald did not receive any formal education which could have familiarized her with spelling conventions of her time. Instead, she professed a lifelong interest in contemporary as well as Scottish literature (cf. Scott 1995). Gustafsson (2002: 259) asserts that the period between 1760 and 1790 was 'characterised by the absence of variation in prose works', which implies that the contemporary literature Archbald read while still in Scotland had a largely uniform spelling. Under the assumption that she was taught at home and continued to educate herself by means of contemporary literature and the occasional newspaper, the uniform spelling which she encountered in those works could have served as an incentive to streamline her own spelling practices accordingly. This is also where the circle comes to a close, as Archbald's reading practices on an individual level are closely tied to the wish for uniform spelling practices as expressed through prescriptive grammars and guides on the broader, societal level.

## 5 Conclusion

The analysis and discussion of verbal inflections in Mary Ann Wodrow Archbald's private writings have shown that no single factor can fully account for the variety of trajectories encountered in the material. Both non-standard features, like the syncretism of preterite and past participle forms, down to the level of individual verb forms of irregular preterites show a diverse range of possible developments over the course of Archbald's life. There is nonetheless the possibility to observe a general tendency to conform to standard practices over time even if the specific paths that lead there might vary significantly. The past participle form *wrote* occurs almost exclusively for over 20 years, only to be replaced by *written* in the remainder of the material. In contrast, Archbald used the preterite form *keept* sparingly during the first few years of her writing, but quickly abandoned it in favour of the standard form *kept* which had dominated in the material throughout. In both cases, the form that is adopted eventually is also preferred in a southern standard variety.

The variation found in Archbald's writing would sometimes parallel similar trends in society, as is the case with the syncretism of preterite and past participle forms in private letters during the latter half of the eighteenth century (cf. Gustafsson 2002: 272). Prescriptive tendencies during Archbald's lifetime constitute a possible force in the direction of using more standard forms over time, as does the language community that surrounded Archbald after her migration which included only few speakers of Scottish English. Lastly, Archbald's interest in contemporary literature likely called her attention to uniform spelling practices in printed works, possibly suggesting the idea that variant forms were less desirable in her own writing.

In analysing Archbald's writings, this study adds insight into the growing number of studies which investigate the changing linguistic practices of women and migrant writers over the course of their lifespans, and it exemplifies how a combination of qualitative and quantitative methods can be applied to highlight the changing patterns of usage of one Scottish woman emigrant. The results furthermore corroborate Raumolin-Brunberg's (2009: 192) finding that changes in frequency are more common than changes in underlying grammatical systems when it comes to longitudinal changes of individual writers.

While it is true that texts written by women writers, migrants or minority speakers have not been preserved to the same extent as those of male, educated writers, the effort to trace their material is worthwhile, as a combination of qualitative and quantitative analyses of their writings add further depth to the overall picture of longitudinal intra-writer variation.

## References


#### Andreas Krogull, Jill Puttaert & Gijsbert Rutten

14 Assessing Dutch-French language choice in nineteenth-century private family correspondence: From intra-writer variation to the bigger picture

#### Abstract

The present chapter examines Dutch-French language choice in the history of the Northern Low Countries, focusing on the private domain in the nineteenth century. Seeking to assess the phenomenon from a quantitative perspective, while meaningfully integrating the role of intra-writer variation, we present two complementary approaches. On the basis of a substantial dataset of private family correspondence, we first illustrate a quantitative methodology that allows us to systematically study the sociolinguistic dynamics that determine language choice. The variables under investigation include gender constellations and familial relationships. Secondly, we zoom in on intra-writer variation in three selected family archives, taking a more qualitative perspective in order to add valuable nuances to the 'bigger picture'.

## 1 Introduction1

Language choice in European settings of multilingualism has attracted a fair amount of interest in historical sociolinguistics and neighbouring disciplines (e.g. Rjéoutski & Frijhoff 2018). Many of these contact settings can be situated in the broader context of the phenomenon often referred to as 'European francophonie' (Rjéoutski et al. 2014), describing the

<sup>1</sup> The research was supported by the Dutch Research Council (NWO), project 'Pardon my French? Dutch-French language contact in the Netherlands, 1500–1900'.

practice of French in language communities outside France as a second or foreign language. In the Northern Low Countries, that is, the area roughly corresponding to the present-day Netherlands, Dutch-French contact goes back to the Middle Ages. Alongside the influx of French loans into Dutch and other contact-induced changes, the enduring contact setting also led to situations of language choice, where Dutch could potentially give way to French in various domains. Initially serving as a lingua franca for international trade and diplomacy, French also acquired a socio-cultural dimension as a 'language of distinction' among the upper ranks of Dutch society. Surprisingly, large-scale empirical studies of language choice, enabling us to test claims about the alleged *verfransing* 'Frenchification' (Frijhoff 1989), are still scarce (Rutten et al. 2015: 146).

When studying the sociolinguistic dynamics that determine language choice, the private domain appears to be of particular interest. Historical research on prominent individuals or specific families reveals that French was widely used in letters, diaries and other ego-documents, most notably in the eighteenth and nineteenth centuries (Ruberg 2011; van Strien-Chardonneau 2018). However, we argue that in order to understand '*who* speaks *what* language to *whom* and *when*', recalling Fishman's (1965) famous question, language choice needs to be assessed more systematically. In this chapter, we therefore investigate Dutch-French language choice from a quantitative perspective that can provide insights on the bigger picture, while fully acknowledging language choice as a facet of intra-writer variation.

First, we outline the sociohistorical context of Dutch-French language contact in Section 2. In Section 3, we describe our methodology and the dataset compiled for this study. Section 4 presents some quantitative results, before zooming in on intra-writer variation. Section 5 contains the discussion and conclusion.

## 2 Sociohistorical background and language contact

#### *2.1 French in the Dutch context*

French and Dutch share a long history from the early Middle Ages onwards, which continues into the period central to the present chapter. French was used in the Low Countries in the political and commercial domains: French was politically important, for example, during the Burgundian and Habsburg regimes in the late Middle Ages and the Earlymodern period. The two decades from 1795 to 1815 constitute the so-called French period, when the Low Countries formed a vassal state of France. French was also dominant in international diplomacy and trade in Early and Late Modern times (Frijhoff 2015: 116). Numerous social and cultural contacts brought French to the Low Countries. Subsequent waves of socalled Huguenot migration in the sixteenth and seventeenth centuries, comprising religious as well as economic refugees, led to the spread of French in domains such as religion and education: Walloon churches and French schools were established across the Low Countries (Frijhoff 2003; Dodde 2020). In addition, many printers, writers, and booksellers of French descent positioned the Netherlands, that is, the northern parts of the Low Countries, in the heart of the international Republic of Letters (Frijhoff 2003). French furthermore became a language of culture for aristocratic and learned circles, both in the Netherlands and internationally (Rjéoutski et al. 2014; Frijhoff 2015; Offord et al. 2018). While Dutch was the dominant language for many people by the eighteenth century, French books were typically still found in the aforementioned privileged circles (Keblusek 1997; Streng 2008; de Vries 2011). It has been shown that French was even used in private letters and diaries, particularly in the eighteenth and nineteenth centuries (Ruberg 2011; van Strien-Chardonneau & Kok Escalle 2017; van Strien-Chardonneau 2018).

The outcomes of the contact situation were manifold: a presence of French in various domains, increasing multilingualism among certain social groups and individuals, as well as contact-induced changes in the Dutch language. Both societal and individual multilingualism are connected to the issue of language choice, which is the topic of the present chapter. Contact-induced changes can be found in the lexicon and the morphology of Dutch, perhaps also in particular morphosyntactic patterns, which is however still in need of further investigation (van der Sijs 2002; Assendelft et al. forthcoming; de Vooys 1970: 135; van den Toorn et al. 1997: 405; van der Horst 2008: 1150).

An additional effect of the contact situation is the rise of a strong metalinguistic discourse against the alleged *verfransing* 'Frenchification'. This discourse has existed from at least the sixteenth century onwards, and has targeted both contact-induced changes such as borrowings as well as language choice, viz. when supposedly Dutch-speakers adopt French for certain purposes (Rutten et al. 2015: 148f.; Frijhoff 1989; Vogl 2015). The anti-French discourse, which peaked in the eighteenth century, did not only target linguistic issues. It also concerned wider cultural models and needs to be seen in the context of emergent Dutch cultural nationalism in the eighteenth century, which created an opposition of French monarchism and aristocratic values with Dutch republicanism and mercantilism (Kloek & Mijnhardt 2001: 76f.).

#### *2.2 French as a socio-cultural phenomenon*

As elsewhere in Europe (Rjéoutski et al. 2014; Offord et al. 2018), the use of French in the Northern Low Countries not only fulfilled functional needs, such as international communication, but was also a social and cultural phenomenon (Argent et al. 2014: 15) well into the nineteenth century. Associated with the upper ranks of society as a language of distinction, French became 'a means of raising one's status, because of the prestige of court culture and the behavioural model of the elite derived from it' (Frijhoff 2015: 129). This role of French as a form of cultural capital was often 'as important as strictly utilitarian considerations, if not more so' (van Strien-Chardonneau 2014: 154). While the practice of French thus functioned as 'a sign of recognition between people belonging to the same social group' (van Strien-Chardonneau 2014: 171), that is, the elite encompassing the nobility, aristocracy, patriciate and emerging bourgeoisie, it was hardly common among the middle and lower strata (cf. Böhm 2014: 206 on French in Prussia).

Focusing on the domain of private life, Kok Escalle and van Strien-Chardonneau (2017: 9f.) emphasize the role of French as 'a language of culture shared between people who are intimate', as found in ego-documents such as diaries and letters, widely preserved in Dutch family archives (cf. Ruberg 2011: 68–75). The function as language of intimacy has also been attested for other European settings of Francophonie. Offord (2014: 385f.), for instance, notes that for the Russian elite French was 'the preferred idiom […] for various kinds of ego-writing – the personal diary, the album, the travel account, autobiographic reminiscences'.

Studies on historical multilingualism in the Dutch context, often in relation to French, have predominantly explored the topic of language choice through case studies on well-known individuals or families (e.g. van Strien-Chardonneau 2018; Joby 2014) or specific cities (e.g. Kessels-van der Heijde 2015). Ruberg (2011), in addition to her qualitative observations on elite correspondence, also presents some quantitative findings on language choice, still based on a fairly small number of five families from the period 1770–1850. Partly due to the limited representativeness of her letter data, Ruberg (2011: 70) concludes that it is 'perhaps far more revealing to approach the question of language choice from a more qualitative perspective'.

In order to complement previous case studies, we argue that a more quantitative way of assessing language choice is needed to understand who, in the language community, wrote in Dutch and/or in French (to whom and when). Intra-writer variation is a crucial aspect of the phenomenon, and we therefore advocate a methodological approach that enables us to learn about the 'bigger picture' while meaningfully integrating the role of the individual writer.

## 3 Data and method

#### *3.1 Methodology*

Our principal aim is to systematically assess Dutch-French language choice in the history of the Northern Low Countries based on solid empirical evidence, thus going beyond the mostly qualitative observations of individuals' language choices. In order to gain a fuller understanding of 'the dynamics which determine language choice in circumstances where knowledge of more than one language makes choice possible' (Offord 2020: 14), we argue that the issue of historical language choice can and should be tackled quantitatively. At the same time, such an approach needs to acknowledge and incorporate the key role of the individual (identifiable) writer. Intra-individual variation with regard to language choice is, in fact, at the very heart of our methodology (cf. Fishman 1965: 76; Head 1995: 592).

We collected a considerable number of private letter data from Dutch family archives across the Netherlands, spanning the seventeenth to the nineteenth centuries. Based on digitized (though untranscribed) manuscript sources, a detailed inventory was compiled for each family, comprising an extensive set of metadata. These family databases include information about the letters (e.g. date and place of writing), biographical data of their writers (e.g. name, gender, date and place of birth), as well as sociolinguistic parameters describing the relationship between senders and addressees, that is, the communicative setting of letter writing in which a language choice is made. We operationalized metadata into variables like gender constellations (male-female, male-male, female-female, female-male) and familial relationships (e.g. parent-child, child-parent, siblings, spouses).

Most importantly, a language choice had to be assigned to each text. Informed by the process of inventorying and manually reading through all letters, we opted for five linguistic categories: (1) 'Dutch', (2) 'Dutch/ French', (3) 'French', (4), 'French/Dutch', and (5) '50/50'. Categories (1) and (3) refer to the two most monolingual language choices, either with Dutch or French as the primary language. Note that these categories may not be entirely monolingual, as they allow for loanwords from or brief switches into the other language. In categories (2) and (4), we can still assign one dominant language to a text, be it Dutch (in the case of 'Dutch/French') or French (in 'French/Dutch'). However, the amount of code-switching is more substantial, to the extent that multi-word switches or entire passages written in the other language become characteristic of these two categories. Finally, a fifty-fifty category was added to account for those cases where Dutch and French are used to roughly the same extent, making it impossible to identify a single primary language.

As regards the representative selection of historical correspondence data, a well-defined set of criteria seemed crucial. The scopes of family archives can range from copious to fragmentary, typically with a few (often male) family members being overrepresented in the preserved correspondence. Therefore, a careful selection needs to be made when establishing a balance in such an unevenly distributed mass of data, both within and across family archives. Taking into consideration those divergent scopes, we allowed for a maximum of forty letters per family, levelling the differences between larger and smaller archives. When possible, we selected texts by at least five different writers of each family. Furthermore, a maximum of four different addressees per sender was defined, as well as a limit of three letters per sender to the same addressee. No letter writer is thus represented by more than twelve texts, preventing the overrepresentation of language choices made by particularly prolific writers.

Importantly, language choice is represented at the level of the unique relationship between one letter writer (or sender) and one specific family member (or addressee), rather than the individual's outgoing family correspondence in its entirety.2 For the representative selection of (up to) forty letters per family, we draw on all inventoried letters. To illustrate this stage of our methodology, we give three simplified examples. If, for instance, all

2 The difference between these two approaches is also highlighted by Fishman, who argues that the approach considering relationships not only recognizes that 'interacting members of a family […] are *hearers* as well as *speakers* (i.e. that there may be a distinction between multilingual *comprehension* and multilingual *production*)', but also that 'their language behavior may be more than merely a matter of individual preference or facility but also a matter of *role-relations*' (1965: 76).

letters within a sender-addressee unit are written in the same linguistic category, say, 'Dutch', then three letters are selected to represent the language choice within this unit. If, in the case of more than one linguistic category within a sender-addressee unit, the majority of inventoried letters is 'Dutch', with some additional letters written in 'French', the selection of three letters comprises two texts representing the prevalent language choice 'Dutch', and one text representing the other language choice 'French'. A senderaddressee unit with three different language choice options (e.g. 'Dutch', 'French', 'French/Dutch') is represented by one letter for each category,3 irrespective of their exact proportion.

#### *3.2 Dataset*

The case study presented in this article focuses on private family correspondence in the nineteenth century, that is, letters written in the period 1800–99. The texts in this dataset were collected from thirty-six Dutch family archives, covering twelve cities from ten provinces across the language area.4 Three families were selected for each city. As summarized in Table 14.1, the dataset contains a representative selection of 1,329 private family letters (narrowed down from more than 7,000 inventoried letters in total), written by 371 individual letter writers. With regard to the level of representing language choice, the dataset comprises 563 unique senderaddressee relationships. Whenever possible, we also aimed for a balanced inclusion of genders (for both senders and addressees), familial relationships, and generations.

<sup>3</sup> The maximum of three selected letters had to be exceeded in only a handful of cases, namely when more than three different language choices (i.e. four or five) could be attested within the same sender-addressee unit.

<sup>4</sup> These cities are Amsterdam, Arnhem, Den Bosch, Groningen, Haarlem, Leeuwarden, Leiden, Maastricht, Middelburg, The Hague, Utrecht, and Zwolle (in alphabetical order).


Table 14.1. Dataset of nineteenth-century private family correspondence

## 4 Results

## *4.1 Quantitative results*

Looking at the overall distribution of language choice in the nineteenthcentury dataset (Table 14.2), the prevalence of Dutch immediately stands out. The two categories with Dutch as the primary language (i.e. 'Dutch' and 'Dutch/French') constitute over 75 %, while the two Frenchdominant categories combined (i.e. 'French' and 'French/Dutch') represent less than 23 %. Thus, the share of French in nineteenth-century family correspondence should not be overestimated, although French is no marginal language either. The fact that Dutch is the dominant language in our dataset should come as no surprise, since the heyday of French in the Northern Low Countries is traditionally located in the seventeenth and eighteenth centuries (Argent et al. 2014: 1; van der Wal & van Bree 2014: 254; Wright 2016: 134).

When we look at language choice across gender (of the letter writer), no major differences can be found. The two Dutch categories combined account for 77.5 % in letters from male writers and for 72.3 % in letters written by women. The share of the two French categories is 20.6 % and 26.4 %, respectively. These results may seem unexpected if we think of assumptions about French being a 'women's language' (cf. van Strien-Chardonneau

Table 14.2. Relative distribution of language choice in the nineteenth-century dataset (N = 1,329)


Figure 14.1. Language choice across gender constellations.

2018: 76; Ruberg 2011: 70). However, if we explore gender constellations, taking into consideration the role relations within the letters, a gender effect can indeed be attested (Figure 14.1).

Dutch occurs most frequently in letters written by and addressed to men, where the share of Dutch accounts for no less than 82 %, as opposed to 11.5 % for French. Strikingly, French is used more often in letters written to and by women: men writing to women choose French in 23.2 % of all cases, women writing to men in 22.4 %, and women writing to women in 24.1 %. This suggests that French is more frequently used when a woman is part of the communicative setting. In other words, if a woman is involved (either as sender or addressee), we find a higher proportion of French. Ruberg (2011: 70), in her study on Dutch elite correspondence from the eighteenth and nineteenth centuries, comes to a similar conclusion, emphasizing the role of women within the sender-addressee relation with regard to the choice of French.

Figure 14.2. Language choice across familial relationships.

Moreover, the analysis of language choice across familial relationships between close family members shows interesting patterns (Figure 14.2). In intergenerational correspondence particularly, Dutch occurs most often in letters from parents to children (82 %, as opposed to 64.6 % from children to parents), while the share of French is remarkably large in letters from children to their parents (25 %, as opposed to 10.9 % from parents to their children). These findings may suggest that the use of French (also) served as an educational exercise (cf. Ruberg 2011: 72), although we must emphasize that we did not account for the factor of age, which means that 'children' may also be adults. However, it is possible that the hierarchical relationship in child-to-parent correspondence triggers the choice of French. The results for siblings and spouses are less pronounced. Overall, though, the variation across familial relationships strongly supports the importance of role relations for (historical) sociolinguistic research on language choice.

#### *4.2 Intra-writer variation*

We now zoom in on intra-writer variation in the letters of selected writers from three families: (1) Van Hugenpoth tot Aerdt, an aristocratic family from Arnhem (in the eastern province of Gelderland), (2) Van Styrum, a well-off family from Haarlem (in the province of North Holland, near Amsterdam) that served in administrative roles, and (3) Van Haersolte, an aristocratic family from Zwolle (in the eastern province of Overijssel). These families display very different distribution patterns in terms of language choice. While Van Hugenpoth tot Aerdt and Van Haersolte are predominantly 'Dutch' families, the prevalence of French stands out in the correspondence of the Van Styrum family. This inter-familial variation is by no means exceptional, as we observe major differences with respect to language choice across the thirty-six families under scrutiny.

#### 4.2.1 *Van Hugenpoth tot Aerdt family*<sup>5</sup>

Language choice in the Van Hugenpoth tot Aerdt family correspondence is largely in line with the general distribution discussed in Section 4.1, with a strong prevalence of Dutch (72.5 %), whereas French is used in 20 %. The share of mixed-language letters written in 'Dutch/French' and 'French/Dutch' is 2.5 % and 5 %, respectively. If we look at the individual writers and their language choice, it becomes clear that the choice of French can mainly be associated with one family member: Caroline Rose Clotilde Flament (1802–35), wife of Joannes N. W. A. van Hugenpoth tot Aerdt (1789–1849), a lawyer and notary in Arnhem. Caroline, who was born on the island of Martinique (at the time a French colony), and her mother moved from France to the Netherlands after Caroline's father's death in 1804.6 When her mother passed away soon after, she moved to her uncle in The Hague, where she stayed until her marriage in 1824.


Taking a closer look at the sender-addressee relationships and language choice in Caroline's letters, we see that she mainly uses French in letters to her husband. Occasionally, she switches briefly to Dutch. The letters to her sister-in-law are also written in French. If we look at the letters *addressed* to Caroline, Dutch seems to play a slightly larger role, although French remains the dominant language. Her husband Joannes writes to her in French most of the time, occasionally switching briefly to Dutch. In contrast, he writes mainly in Dutch to his children, although we also find some letters to them in French. It is striking that Caroline and Joannes' son, Carolus Antonius Ludovicus (1825–1907), always uses Dutch when writing to both of his parents together and even when writing only to his mother. The same applies to Caroline's sisters-in-law and father-in-law, who choose Dutch in all their letters to Joannes and Caroline.

Caroline Flament certainly is the central figure in this family correspondence when it comes to the choice for French. Before she became a part of the Van Hugenpoth tot Aerdt family through her marriage with Joannes, only Dutch was used in the family correspondence. When she was not part of the communicative setting (either as sender or addressee), mainly Dutch was used. After her death in 1835, the share of French in the family correspondence drops considerably.

## 4.2.2 *Van Styrum family*<sup>7</sup>

Turning to the Van Styrum family and the overall distribution of language choice in their correspondence, we see a clear preference for French (77.5 %), while Dutch only occurs in 10 %. Mixed-language letters written in 'French/Dutch' and 'Dutch/French' account for 10 % and 2.5 %, respectively. The central figure in this family is Jan van Styrum (1757–1824), who held various administrative positions in the city of Haarlem as well as on a national level. He was appointed as a member of parliament of

in the early nineteenth-century Netherlands (cf. Rutten 2019), it seems unlikely, though, that she grew up in a monolingually French-speaking environment.

<sup>7</sup> The Van Styrum family archives are kept at the Noord-Hollands Archief (Haarlem), access no. 141.

the Batavian Republic (*de jure* a sister republic of France but *de facto* its puppet state) and was part of a special commission that was sent to Paris to negotiate with Napoleon about the kingship of Louis Bonaparte, king of Holland. In 1810, Jan van Styrum was transferred to France, where he was prefect of the department of Loire-lnférieure until 1813.

Looking at Jan's correspondence, we notice that French, unsurprisingly, plays a key role. The letters he writes to his wife, Johanna Anna van Vollenhoven (1767–1846), are always written in French and the same applies to the letter addressed to his daughter Anna Henriëtte Maria Wilhelmina (1786–1834). French is also the preferred language in Jan's letters to his sister Maria Jacoba (1763–1848), occasionally choosing Dutch (with some switches to French). When Jan writes to his brother, he sometimes uses French and sometimes Dutch. In one of his letters to his brother, Jan explains why he writes that particular letter as well as the previous letter in Dutch. Discussing his brother's problems with his estate, Jan clarifies that he deliberately writes in Dutch because his brother could then immediately transfer the content of his letters to the notary who is following up the case and defending his brother's interests. The fact that Jan is so explicit about the use of Dutch in his letters implies that Dutch must have been an exception, and that French was the default choice within this brother-brother unit. The implicit presence of the notary seems to be an intervention in the brother-brother/sender-addressee relationship, which may explain the use of the 'non-default' language, that is, Dutch. Jan's siblings also write to him mainly in French. His son, Adolf Jacob (1794–1816), invariably chooses French when writing to both of his parents together and when writing to them separately.

It is striking that almost all Van Styrum members primarily use French in their letter writing, many of them also opting for Dutch in some instances, or switching from French to Dutch within the same letter. It seems that the evident link with France, established by Jan's administrative functions, influences the choice of language within his entire close family.8

8 According to the family archives' description, Jan had little sympathy for the French authorities, as suggested in a report from the king's secret agent. This would imply that the prevalent choice of French was independent of any sympathies or antipathies towards the French political hegemony.

#### 4.2.3 *Van Haersolte family*<sup>9</sup>

For the previous two families, we thus established direct or indirect links with France that may explain their choice of French in their private correspondence. However, we must be aware that in other families using French in their letters, such a link with France is far less obvious. A case in point is the Van Haersolte family. The share of Dutch in this family correspondence is 77.5 %, compared to 12.5 % French and 7.5 % 'French/ Dutch'. Particularly among members of the second and third generations, interesting patterns with respect to language choice emerge. Geertruid Agnes de Vos van Steenwijk (1813–74), wife of mayor Johan Christiaan van Haersolte (1809–81), invariably writes to her husband in Dutch. However, in the letters to her daughter, Sophia Cornelia (1838–73), she mainly chooses French. Most of Sophia Cornelia's letters to her father are in Dutch, and although rarely writing to him in French, she proudly reports in one of her Dutch letters that she is learning a lot from her French teacher. Sophia opts for mixed-language letters (mostly 'French/Dutch') to her sister Louise Christine Egbertine Françoise (1840–1918).

In contrast to the female members, the men in this family predominantly choose Dutch. Johan Christiaan, for instance, exclusively writes in Dutch to his wife Agnes, both parents and his daughter Louise. His son, Coenraad Willem Antoni (1845–1925), also uses only Dutch when he writes to his sister Louise. It is evident that the women in this family show a preference for French in their correspondence. For these women, no link with France can be determined, neither on a personal nor on a professional level. This suggests that their use of French in the private sphere most likely served a socio-cultural function – as a language of distinction and/or intimacy (see Section 2.2).

<sup>9</sup> The Van Haersolte family archives are kept at the Historisch Centrum Overijssel (Zwolle), access no. 0237.1.

## 5 Discussion and conclusion

In this chapter we have addressed the phenomenon of language choice in the context of Dutch-French contacts in the history of the Northern Low Countries. Focusing on the private domain and on nineteenthcentury family correspondence in particular, we sought to assess the sociolinguistic dynamics that determine language choice in a quantitative manner. First, our analyses based on a dataset of more than 1,300 letters from 36 families (see Section 4.1) have shown that Dutch was the prevalent language choice for private letter writing in the nineteenth century (roughly 75 %). While the use of French cannot be considered marginal, our findings do not indicate that this specific social domain was overly 'Frenchified'.

Examining a number of sociolinguistic parameters incorporated into our methodology, gender (of the writer) did not appear to be a crucial variable. Patterns became more pronounced when looking at gender constellations, though. Women, both as senders and addressees, could be associated with higher proportions of French than men-to-men constellations particularly. Furthermore, the variable of familial relationships revealed differences between children-to-parents and parents-to-children, the former having a considerably higher share of French than the latter, which deserves to be discussed more closely in the future. These findings emphasize the importance of role relations or, more concretely, sender-addressee relationships when assessing the topic of language choice.

We then zoomed in on intra-individual variation in the letters from three families from our nineteenth-century dataset (see Section 4.2). The correspondence of the Van Hugenpoth and Van Haersolte families displayed a clear preference for Dutch, although French was also used in several letters. In contrast, French was the favoured language in the Van Styrum family correspondence. For the Van Hugenpoth and the Van Styrum families, we were able to establish links with France that provide possible explanations for the use of French. In the case of Van Hugenpoth, French was mainly linked to one family member of francophone descent. The use of French in the Van Styrum correspondence may be associated with the professional career of the central figure. However, we must bear in mind that such an evident link with France is not necessarily present in all families that use French in their correspondence. In many cases, French was rather a socio-cultural phenomenon and served as a language of distinction (see Section 2.2), as we could see in the women's correspondence of the Van Haersolte family.

We consider these two perspectives on historical language choice to be best treated as complementary. While previous studies on Dutch-French language choice (in different temporal and spatial settings) have largely focused on more qualitative micro-level accounts of individuals' language choices, their representativeness tends to be fairly limited, as the striking differences across families in Section 4.2 have demonstrated (cf. also Ruberg 2011: 69f.). It goes without saying that some qualitative interpretations about individuals and 'their personal situation and state of mind' (Ruberg 2011: 70), or political factors and identity awareness (van Strien-Chardonneau 2018: 77–81) can hardly be captured by a more macro-oriented framework. In this chapter, we have suggested a different perspective on language choice, showing that inter-writer variation can still be integrated in a meaningful way. With this approach, we aim to assess language choice on a larger scale, but incorporate the key role of intrawriter variation (i.e. on the level of unique sender-addressee relationships) in order to provide a well-balanced and representative dataset.

We try to advocate here for the study of (historical) language choice in the private domain, and recall Nevalainen and Raumolin-Brunberg's approach to the study of language change (2017: 244), namely that the 'various perspectives are not mutually exclusive' and that 'micro-level studies […] benefit from macro-level baseline data'. In other words, it may be risky to *depart* from qualitative observations only in order to make claims about the sociolinguistic dynamics that determine the distribution in a larger community – the 'bigger picture', if you will. However, departing from a substantial empirical dataset (or baseline evidence), which at the same time takes into account intra-writer variation, can help us see individuals' language choices in perspective. What is more, such an approach enables us to go back to the micro level of the individual and zoom in on intraand inter-writer variation (within a family), adding noteworthy nuances about the individuals' personal and professional biographies, their mobility across the lifespan, and so forth. Ideally, findings coming from these two approaches can complement each other.

The scope of this chapter only allowed us to share some insights (both quantitative and qualitative) on the intriguing phenomenon of Dutch-French language choice in the history of the Northern Low Countries. Further exploring our extensive dataset of private family correspondence, we will report on the examined variables (i.e. gender, familial relationships) in more detail as well as on the spatial and diachronic dimensions in future publications.

## References


#### Veronika Girininkaitė

## 15 Intra-writer variation in the multilingual *Diary* of Vytautas Civinskis (1887–1910)

#### Abstract

The article presents intra-writer variation in a multilingual idiolect, relying on examples from the *Diary* of Vytautas Civinskis (1887–1910). This manuscript, written over a timespan of six years, allows for tracing diachronic changes in the diarist's idiolect (L1 Polish), his ways of learning L2 (Lithuanian) and the synchronic variations he used. Moreover, it contains passages with Russian and German, as well as some phrases or words in French, Latin, and Yiddish. I measured the quantitative presence of each language and chose a qualitative approach to the materials, interpreting data in the context of the biographical and ideological circumstances under which the document was created. I interpreted seemingly deliberate instances of code-switching as quotations and as serving as rhetorical instruments of emphasis, specification, euphemism and wordplay.

## 1 Introduction and description of the corpus

## *1.1 The manuscript and the aims of the study*

The article aims to give a glimpse of the structure and multilingual aspects of a complex handwritten ego document and to interpret the functions of its code-switching instances. I will examine the manuscript, titled *Diary*, written from 1904 to 1910 by the Polish student of agricultural science and veterinary medicine, Vytautas Civinskis (born *Witold Cywiński*, he later adopted the Lithuanian version of his name). The *Diary* is held in part in the Manuscript Division of Vilnius University Library (VUB). Some portions are stored in the Wróblewski Library of the Lithuanian Academy of Sciences (LMAVB). The VUB manuscript consists of twenty-eight notebooks, most of them sized 21 × 17 cm, with different pagination, the total number of written pages exceeding 2,800 (see Figure 15.1). In most cases, pages were filled with writing only on one side, the other side usually being left blank for later notes, postcards and some other additional documents. The amount of text differs in every single year of the *Diary*. Also the document is not preserved in full: the notebook D1027 is apparently lost, and in some of the other parts, some pages are torn out or partly cut out by the author, who edited the manuscript in this way.

The *Diary* is a highly complex manuscript, more or less a scrapbook. It includes daily entries and reminiscences, multiple drafts of letters sent, letters received glued in, postcards, photographs, cut-outs from newspapers, used tickets, etc. Additional peculiarities of the *Diary* sometimes include extensive use of paragraphemic means (change of handwriting styles, unexpected layout of words on a page, etc.) and individual crypto-language, which is relatively rare. The languages used most in this document are Polish, Lithuanian, Russian, and German, with some phrases or words in French, English, Latin, and Yiddish. As the document's creation took six years, it offers an opportunity to study the author's idiolect and intra-writer variations, both in synchrony and diachrony.

Figure 15.1. The twenty-eight notebooks of the *Diary* of Vytautas Civinskis. Photography by Veronika Girininkaitė.

In the following Section 1.2, I start by introducing the circumstances of the author's life, which influenced his multilingualism and are essential for understanding the structure of this text. In Section 2, I discuss linguistic research on personal idiolect and the research methods applied in the present chapter. In Section 3, I show which languages, as well as which scripts and orthography styles the diarist used. I show examples where code switching is asynchronous on the level of language (Polish, Russian, etc.) and the level of script (Latin, Cyrillic). I also explain the possible motivations for code-switching in the *Diary*. Section 4 touches on diachronic changes in the diarist's use of his L2 Lithuanian. Section 5 concludes the article.

#### *1.2 Linguistic and social circumstances of the* Diary

The multilingualism of the manuscript is related to the diarist's biography. Biographical and other data have been collected from the *Diary* itself, from memoirs of contemporaries, as well as other publications (Jankauskas 2003, 2010; *Geni*) and encyclopaedia articles. This informa tion allowed me to interpret the use of languages in the manuscript.

Vytautas Civinskis (1887–1910) was born in Moscow and completed school in the Russian language, but was brought up in a Polish-speaking family of nobility, which had a manor house and land in Lithuania (part of the Russian Empire at that time). As the family was wealthy, the children had French and German governesses (serving also as language tutors). As a teenager, Vytautas became interested in the Lithuanian language, which was used by the servants in the family estate in *Latavėnai*, in contemporary Lithuania, and may have been encouraged by one of his grandmothers. She felt the patriotic urge to learn the language of the land they were living in, as mentioned in the memoirs of one of the younger members of the family (Trzebińska-Wróblewska 2002: 17). According to the same memoirist, some members of the Okulicz family, that is, relations on Vytautas' mother's side, at that time identified as Lithuanians, and some as Polish (Trzebińska-Wróblewska 2002: 22).

The father of the family, Hieronym Cywiński, worked as a high-ranking railroad engineer. As a result, the family spent most of their time in 'Russia proper', where Vytautas and his siblings studied in a Russian school. After finishing school in Tambov, Vytautas decided to study agricultural sciences at Leipzig University. On his way there, in 1904, at the age of 17, he began to write the *Diary*. At the time, the *Diary* seems to have been his companion, a remedy against loneliness, and a tool for articulating his thoughts and philosophical worldview and writing down any novel impressions. Vytautas also paid great attention to introspection regarding emotional phenomena and feelings, especially after attending the psychology lectures of the famous Wilhelm Wundt (Almonaitienė & Girininkaitė 2021).

In Leipzig, Vytautas also bought a grammar of the Lithuanian language and began to study it formally. The events of 1905, the uprising in Russia, coinciding with a renaissance of the Lithuanian language and of national identity, were significant for him. We can see a remarkable progress in his Lithuanian-language writing practice through the years that the *Diary* was created (more in Girininkaitė 2017b). After some years of study, Vytautas left Leipzig for the *Königliche Tierärztliche Hochschule* [royal veterinary school] in Berlin, where he began studying to be a veterinarian, a choice of profession rooted in his hope to achieve economic independence from his family. Later Civinskis continued his studies closer to Lithuania, in present-day Estonia in the Tartu Veterinarian Institute. Sadly, after finishing his studies, the young man committed suicide. The *Diary* which Civinskis left allows to reconstruct not only the outline of his life but also of his linguistic environment.

The Polish language which Civinskis used is the sociolect of the Polish gentry residing in Lithuania and his family's informal language, and his writing contains local dialect phenomena in pronunciation, morphology and lexis. By the nineteenth century, Polish was firmly established as the prestige language in the territory of modern Lithuania.

This sociolect of the Polish gentry in Lithuania would differ from the normative Polish language in many respects. It was utterly conservative in vocabulary, and also made use of specific local lexis and an abundance of diminutive forms. As it was acquired only informally in the families, there was often a lack of training to use it in writing (Čekmonienė & Čekmonas 2017: 324). That is why the orthography of this sociolect would often be irregular, with, for example, unsure use of diacritics (e.g. <z>/<ż>/<ź>/<ž>). On the level of pronunciation, this variant of Polish was characterized by a softer, palatal pronunciation of consonants, and on the level of morphosyntax, by particular usage of noun cases, which in some cases persisted well into the twentieth century (Karaś 2001). Civinskis, who perhaps was never trained to write in Polish correctly, has a very unstable, mutable orthography. He could, for example, interchangeably use the graphemes <ó>/<u> for [u] and <ż>/<ź>/<ž> for [ž]. Hence, the *Diary* may be considered an informative source of contemporary language usage. It is known that in 'tracing orality in written records, it appears to be worthwhile to focus on writing by semi-literate rather than highly literate or even professional writers' (Elspaß 2012: 158). The differences in orthographies used in the Cywiński family letters included in the *Diary* might be suitable material for a separate study.1

## 2 Research context and methodology

## *2.1 Approaching the idiolect of a multilingual*

Peter Koch and Wulf Oesterreicher introduced the notion of texts of 'immediacy' and 'distance' and the idea of a continuum between conceptual orality and conceptual literacy. The poles of this continuum coincide not with the medium that conveys the message but with the speaker's intentions, communicative situation and text genre (Koch & Oesterreicher 1985: 29). This concept allows an analysis of the traces of orality in written texts, broadening the possibilities for historical sociolinguistic research.

1 This sociolect in itself is also an inspiring topic of study, which has sparked the interest of many researchers from Poland and Lithuania. According to Irena Adomavičiūtė-Čekmonienė and Valerijus Čekmonas, in the late nineteenth and twentieth centuries the Polish language became part of the identity of local nobles (Čekmonienė & Čekmonas 2017: 326). Other researchers agree with the conception that this language variant became a phenomenon of cultural identity (Sawaniewska-Mochowa & Zielińska 2007: 216). In the circumstances of cultural and political oppression, using the Polish language became a symbol of resistance to Russian power.

More attention was subsequently paid to so-called ego documents, including letters, unedited notes, diaries and memoirs-documents which, for sociolinguistic research, are 'the next best thing' after oral speech.

A person's identity in general is fluid: 'identity is a matter of dynamic performance rather than inert, personal qualities' (Prior 2006: 104). A person's idiolect also changes through lifetime. The idiolect of a multilingual is an object doubly elusive and mutable. For a long time, idiolect was not recognized as a valid object of scientific study due to its fluidity and low level of predictability (Romaine 2009: 243; Oksaar 2000: 38). Peter Auer claims that proper research on the idiolect of a multilingual person was long unwelcome, as it was felt to call into question many convenient stereotypes about language usage (Auer 2006: 2f.).

Today bilingualism is a recognized and well-established topic of study. Lately, especially in historical sociolinguistics, 'there is growing recognition, that language change does occur, and can be captured and studied in the lifetime of an individual' (Evans 2013: 3), and that studies of unique idiolects may enrich our understanding of language changes that happen in a given society. Moreover, a researcher working with idiolects obtains material free from influencing factors of language choice such as age differences, gender, social position (Evans 2013: 23), enabling research on stylistic and interactive aspects of linguistic variation. Historical multilingualism and code-switching are emerging research fields in historical sociolingustics (Skaffari & Mäkilähde 2014: 259; Pavlenko 2005: 311).

#### *2.2 Qualitative and quantitative approaches to the manuscript*

For this study, a predominantly qualitative approach is implemented due to the current state of research on this document: it is a 'raw' manuscript because the text is not digitized nor transliterated, so the possibilities of performing quantitative analysis on it are limited. Therefore, this study mainly focuses on interpreting the cases of code-switching and orthographic variation found in the text. I looked at how code-switching happens: on which levels it happens as well as how it is usually expressed graphically, attempting to explain some of the causes that might have triggered those switches. Though I will provide a couple of examples from the letters, my main focus was on the language use in the entries in the *Diary*, where I will attempt to show that functions and causes for intraauthor variation were ample.

Nevertheless, the approximate quantitative changes in the use of languages through the years have also been calculated. I counted the filled pages of the *Diary* in order to evaluate at least approximately the part played by multiple languages used in entries. I considered only the pages which had at least half a page of text. Again, I measured the presence of the mentioned languages only in the *Diary* entries, omitting the letters which Civinskis received from other people.

## 3 Results of the analysis

#### *3.1 The languages in the* Diary

The way the languages were distributed in the text is shown in Figure 15.2. The *Diary* pages with entries were checked for this graph, marking them as written either in Polish, Lithuanian, Russian, German or French. Other languages do not reach a level of 1%. In the cases of several languages present on one page, which are rare in the first three years of the *Diary* but happened quite often later on, I marked them as written in the quantitatively dominating language. Each column represents a single notebook. The years in which the notebooks were created are marked at the bottom.

The chart shows that in the first two years the *Diary* was almost exclusively written in Polish, later becoming more varied. The Polish language decreased from 95 or 81 % (notebooks D1024 and D1030) to 28 % (D1033) but then rose again, Lithuanian rising from 3 % and 0 % (D1024 and D1028) to 40 % (D1033), then oscillating between this and 80 % and 75 % (D1043 and D1050). Such changes were related to changes in the biography and linguistic attitudes of the diarist (see Section 1.2). There are three notable increases in the use of Lithuanian: the first one (in D1033) is more or less related to 1905, the year of the Revolution in Russia, and the increase of

Figure 15.2. Distribution of languages in the twenty-eight notebooks of the *Diary* (1904–1910).

the movement for the independence of Lithuania. In that year, the diarist left his studies in Leipzig and spent more time in Lithuania. The later increase (D1043, D1050) is perhaps related to the studies in Tartu (1908–10), where Civinskis enrolled in some informal Lithuanian language courses and joined the local Society of Lithuanian students. He was appointed the secretary of this society, which involved a large amount of writing in this language and led to more progress in his writing skills in this language. The third increase may relate to the longer time spent inside Lithuania on student vacations.

The amount of Russian language is small, with 15 % at most (D1030, D1037), but 6.6 % on average. Use of French is occasional, amounting to less than 5 % (D1038), but usually not exceeding 1 % of the written text. Latin and English are present in the text only as words and sayings, Yiddish in the form of affixes and single lexemes.

The increase of the German language occurred during the time when Civinskis studied in Berlin (1904–07). Maximum use is observable at the end of 1905 and the beginning of 1906, with 39 % and 20 % (D1032 and D1033), respectively. While continuing studies at the Tartu Veterinarian Institute in 1907–10, Civinskis began to write more in the Lithuanian language, but this increase was not unidirectional.

#### *3.2 Orthographic variation and the linguistic attitudes of the diarist*

In this section, I shall look at orthograpic variation which the diarist exhibits in his letters while accommodating to the addressee's linguistic knowledge and attitudes. This accommodation could be expressed in switching between the languages or altering the writing norm in one language. For the latter case, while writing in Russian, Civinskis could choose between the unofficial, simplified way of writing when addressing his former classmate (1) or the official one, the traditional way in the letters to his grandmother (2) or a police officer.


The graphemes <ѣ>, <ъ>, <i> in the following words from (2) are typical of the older Russian writing style: въ августѣ, вслѣдствiе, Зрѣнie,

2 All translations of the examples into English are my own.

удовлетворительнымъ, съ. These graphemes, initially reflecting the specific Old Slavic vowels [ě], [e], [i:], remained in Russian standard orthography until officially cancelled by the spelling reform of 1918 (Cubberley 1996: 350). Still, because of their apparent redundancy, they were not used by more progressive authors already in the first half of the nineteenth century (Grigorjeva 2004: 38). Starting from the middle of the twelth century, the letter <ъ> was not pronounced in Russian. It was still placed after a word's final consonant, as a relic of the law of the open syllables, characteristic of the Old Slavic language (Dundaitė 2005: 16). The French word 'pince-nêz' [glasses held on a person's nose by a spring rather than by pieces that fit around the ears] is preserved in the original spelling as a realia lexeme. The following example (3) is not from a letter but a *Diary* entry.

(3) Митовъ въ самом дѣле милый уголокъ – если не перессорюсь с дядей и арендаторомъ – заживу припѣваючи. Впрочемъ арендаторъ кажется сравнительно интеллигентнымъ, с дядей ладим прекрасно (VUB RS F1- D1051, 83, year 1910)

[The estate of Mituva is indeed a lovely place – if I do not quarrel with the uncle and the tenant – I will live here happily. However, the tenant seems to be a comparatively intelligent man, and we get on with the uncle very well].

It seems that the traditional orthography of Russian was more comfortable to Civinskis, because he also used it in the *Diary* itself in this and other cases. In (3), we see words written with the afore-mentioned graphemes (Митовъ, дѣле, припѣваючи, etc.).

One more interesting thing about the *Diary* is the diarist's quite frequent metalinguistic notes, testifying to his linguistic attitudes. Civinskis must have been an observant listener, as he commented frequently about the pronunciations and accents of people he talked to. Sometimes, there are signs of metalinguistic awareness, even explicit explanations of language choice strategies.


(6) nuo šiuo laiko rašysiu letuviškai (VUB RS F1-D1036, 82, year 1906) [from now on, I will write in Lithuanian]

In (4), the German dialectal words are inserted into a Polish sentence as examples of the pronunciation of another person, which annoyed the listener because of their deviation from standard German. In sentence (5) in Polish, there is a note on the difficulty of breaking the habit of code-switching in the presence of a Lithuanian purist friend, and explicit declarations of which language he thinks it is better to use. The decision to write in Lithuanian (6), written in the Lithuanian language, was not final, as the use of the languages in the further text of the *Diary* was still diverse.

#### *3.3 Asynchronous switching: Language vs writing system*

Pieter Muysken differentiates three main types of code-switching: *insertion* (ABA), *alternation* (AB) and *congruent lexicalization* (ABABA, without clear borders between the two languages) (Muysken 2000: 8). The types of code-switching in the *Diary* differ: it occurs between entries, between sentences, and quite often inside a sentence or even a word. The languages used in the *Diary* are usually written in different scripts: Cyrillic for Russian, Latin for the other languages. However, it was unexpected to find that in some (relatively rare) instances the switching was asynchronous at different levels of the text: for example, switching to Russian language lexis was not always accompanied by switching to Cyrillic. Seemingly, this dissociation between language and writing system was more likely to occur while writing under stress or in a hurry. Perhaps, a certain amount of 'inertia' in the writing habit lingered in written graphemes, even when the writer had changed to another language of expression. This resulted in a number of examples of Polish/Lithuanian words written in Cyrillic letters and *vice versa*.


(9) Стравиński в Екатеринославе (VUB RS F1-D1032, 85, year 1905) [Stravinski is in Jekaterinoslavl]

The change of language or writing system may be triggered by a particular element (Auer 2006: 6). An *alternation* seems to have occurred in (7), where [after a portion of Russian text in Cyrillic] the Polish phrase is still in Cyrillic writing, and after a lag, turns to the Latin writing system within the sentence. This also applies to the *insertion* examples below. In (8), the Russian word *провалюсь* [to fail] was inserted into a Lithunian context. This triggered the use of Cyrillic writing in the last two words, although they switched back into the Lithuanian language. In (9), perhaps another example of *insertion*, inside the Russian phrase a typical Polish surname affix -*ński* manifested itself in the Polish name *Stravinski.* This triggered the change of writing system, but only for this affix. This last example shows that a switch of a code's graphic expression may happen even inside the lexeme. All these instances support the notion of Penelope Gardner-Chloros (2009: 11) that code-switching is a gradual and complicated process, not as discrete as suggested by the term.

## *3.4 Code-switching as quoting and as a means of stylistic expression*

It seems that in a considerable number of instances in this *Diary*, the choice to switch is deliberate and a tool of stylistic variation. The conceptual similarity of language-switching in one's speech to changing the style or an accent was discussed by Gardner-Chloros et al. (2000: 1307). My research shows that code-switching in this text is found in four primary contexts: when quoting, as a euphemism, for the precision of meaning and for expressive foregrounding.

In the first case, it is a way to preserve a precise *quote* in the original language. This phenomenon has been mentioned: 'as a writing person can create a dialogue not only with possible readers but also with earlier texts this is the place where the code-switching may arise' (Skaffari & Mäkilähde 2014: 262). In the *Diary*, a writer's name, a book's name, or a phrase heard earlier in a conversation could often be not translated (in case of the proper names – these were preserved in the original writing system).


In (10), within a Polish sentence we see the unchanged form of Толстой [Tolstoy], a Russian writer's surname, as well as a specialized Russian phrase for the 'collected works'. The Russian insertions are written in Cyrillic letters (10). In (11), the German word *Versammlung* [meeting] inserted into the middle of a Lithuanian sentence is perhaps quoted as it was used by the fellow students. However, the noun is not capitalized as would be the spelling norm in German; this is a steady individual peculiarity of Civinskis' way to write German nouns.

Secondly, a word of another language may have served as a *euphemism*. For example, it seems that when talking about money, the diarist would, as a rule, refer to this concept by a word from a different language. Hypothetically, this topic was perceived by the diarist as vulgar, inappropriate for writing, the feeling being alleviated by using terms from foreign languages with the same meaning. This hypothesis may be supported by data given by Aneta Pavlenko: in psychotherapy, switching to another language helped bilinguals to talk about topics that were emotionally not easy to mention in the L1 (Pavlenko 2005: 28).

In the examples below we see a word that seems to be a German noun with Yiddish diminutive suffix (*geldele*) (12), incorrect English (*monees*) (13), low style vernacular Polish (*fajgle*) (14).


Using words from another language was also a way to achieve the *precision* of expression, such as naming emotions in different languages. Despite the popular notion that there are basic emotions that are familiar to all humans, specific emotion names are in fact untranslatable. That is why bilinguals often use terms from two (or more) languages they know, to achieve precision, and even 'feel handicapped' when operating with words from only the second language (Pavlenko 2014: 261). Civinskis, who, as it seems, suffered from mild depression, when qualifying his emotions, took on a difficult task: he attempted to investigate and observe the change in his emotional state over several months. The resulting observations were included in the *Diary*. While attempting to create the scale of his own emotions experienced daily, Civinskis used words from Russian (телячiй восторг', [a calf 's joy, immense joy]), Polish (*ožywienie* [feeling moved]; *przygnębienie* [feeling depressed]), and German (*gemütlich* [pleasant, cosy]) (VUB RS F1-D1042, 28; more on this topic in Girininkaitė 2017a).

The fourth reason for the intended code-switching might have been to *foreground* some idea by iterating it in different languages, sometimes also playing with the sound and making a deliberately unusual combination of languages in one sentence. I propose as a term *montage of languages* (based on the concept of montage in cinematography) as a deliberate text editing technique involving multiple languages that are used to foreground some idea, giving it an especially salient place in the text.


[I dreamt [Polish] I dreamt [French] […] I dreamt [French] I dreamt [Lithuanian] because it just doesn't sound beautiful in Lithuanian [Lithuanian]. A dream [German]. A dream [French]]

In (15), in addition to Polish, the same phrase is iterated in Latin. This was perhaps meant to create a solemn effect due to its status as the sacred language. In (16), there is an interesting metalinguistic remark [it doesn't sound beautiful in the Lithuanian], which allows for the interpretation that in this instance the diarist, looking for an optimal means of expression, was guided by his aesthetic considerations.

## 4 Diachronic intra-writer variation in text: Learning a new language

The Lithuanian language in this text shows variations in language competence change over a lifetime. In this regard, the *Diary* is a chronicle of Civinskis' growing competence in Lithuanian. There is a shift in this idiolect from the 'naive' orthography, reflecting elements from the spoken language, to standardized orthography. The Lithuanian language for Civinskis was the L2, or, to be precise, L4, after Polish, Russian and German. Civinskis started by using isolated Lithuanian lexemes. As is usual for an inexperienced writer, he spelt them the way he heard them in the local dialect pronunciation: for example, *visuokių*, *ką tavi vielniai* [all kinds of; may devils get you] (D1037, 26), while the more conventional spelling would be *visokių*, *kad tave velniai*. After more studying, ample reading of Lithuanian in different sources, and joining the society of Lithuanian students, Civinskis shifted toward the more conventional spelling. This was one of the accepted orthographies at the time. Though the standard variety already emerged between 1883, with the beginning of the underground periodicals, and the beginning of the twentieth century, with Jonas Jablonskis' grammar published in 1901 (Senn 1944: 102; Zinkevičius 1998: 293), the language standard was still not stable by the end of this period.4 Civinskis practised language whenever possible by speaking, writing, and reading, and he even bought and used several grammars and dictionaries. That is how a noticeable diachronic change came about in his orthography.

## 5 Conclusions

Intra-writer variation may be found in autograph manuscripts, preferably ego documents like diaries, and is more easily observed in manuscripts written by one writer over a long time. This kind of material is quite rare. When a person has written over a sufficiently long time, it is sometimes possible both to see the linguistic alternation in synchrony and to trace diachronic changes in language use that occur diachronically during several years of the person's life. Of course, it is always important to keep in mind the relativeness of any conclusions we may draw from any historical corpora. The causes are the partial nature of the data we can observe and analyse, and the lack of texts that may not have survived, but that might have been crucial to understanding the linguistic situation of that period. We can often see only a part of the person's linguistic repertoire in idiolect data.

In historical sociolinguistics, the limits of the object of research usually are defined by the incident because researchers have to work with the limited material that has come down to them. The Civinskis *Diary* is a rare resource, providing comprehensive data on the linguistic usage of one person over six years. Analysing the linguistic use and habits of the diarist, I looked most at the text of the *Diary* itself, for the most part leaving the letters aside. The main reason for language and style choice in the letters

4 This process was so long due to the repressive law prohibiting education in the Lithuanian language and use of Latin letters for Lithuanian publications, issued in 1864 and valid until 1904, which was to ensure assimilation of this part of the Empire. All the periodicals in Lithuanian were published abroad and smuggled into the country (Zinkevičius 1998: 260f.).

seemed to be to adapt to the recipient's knowledge and attitudes. In the text of the *Diary*, Civinskis was not limited by this, so style and language switches were less constrained and can be considered rhetorical instruments. Usually the code-switching in *Diary* emerged as quote, as a euphemism, for the precision of meaning, for expressive foregrounding, and as a wordplay.

Though the Polish language constitutes the most significant part of the *Diary*, it is stylistically more or less homogenous and did not change much over time. The variant of Polish used by Civinskis is the sociolect typical for the nobility residing on the territory of contemporary Lithuania in the nineteenth–twentieth centuries. Its traits, described in specific studies, are the softer pronunciation of consonants, some characteristic morphological affixes, specific lexis and visible influence of the Lithuanian language.

No idiolect is isolated. It reflects language usage in the social groups with which a person seeks to identify and, in the case of migration, in the locations where the person has spent a long time. In the case of Civinskis, he showed a deliberate choice to learn and use the Lithuanian language. He was also influenced by the German language, which was due to the location of his studies. As the diarist spent his childhood and finished school in the Russian language and was actively reading the contemporary Russian fiction writers, this language never ceased to be used in the *Diary*. As shown, the French language was used only occasionally, mainly as *bon mots* or sayings. Yiddish was represented in the text by the use of a diminutive suffix and a small number of lexemes, which apparently were known to the diarist from his social circle.

The beginning of the twentieth century represents a peak in the national revival of Lithuanians, with language usage becoming a central part of the speakers' identity. In some cases this led to a conscious (but not easy to perform) breaking of an individual's linguistic habits: formerly speaking Polish, an individual could switch to the 'language of the ancestors' (Lithuanian) or the other way around. Bilingual people might decide to demonstratively use only one language to support it and express their social and political position in a changing world. Civinskis, in his *Diary*, did not follow one single strategy, instead creating a multilingual and complicated text, full of variations. During the person's lifespan, idiolect may change depending on the history of migration, the learning of new languages and the attrition of earlier ones, a person's deliberate preferences and poetic, expressive ambitions. This *Diary* shows the importance of an individual's linguistic attitudes and voluntary decisions for variant choice and the fluidity of the idiolect.

## References

#### *Primary sources*


#### *Secondary sources*


#### Doris Stolberg

# 16 *Picnick* and *Sauerkraut*: German–English intra-writer variation in script and language (1867–1900)

#### Abstract

Intra-writer variation is a wide-spread phenomenon that nevertheless has received only limited research attention so far. Different addressees, bi- and multilingualism, or changing life phases are among the factors that contribute to such variation. In a study of diary entries by one writer covering three decades (1867–1900), this chapter investigates patterns of intra-writer variation between German and English (language and script) in nineteenth-century Canada, with a special focus on single word borrowings, person reference and place names. The long-term perspective provides a unique insight into the dynamics of a bilingual writer's emerging sociolinguistic competence as reflected by the flexible yet structured use of his resources within the social space of a bilingual community.

## 1 Introduction

Variable use of language can lead to enregisterment (Agha 2007; Agha & Frog 2015; Anderwald & Hoekstra 2017) and hence can be employed as a strategy for positioning (Harré & Van Langenhove 1999; Beeching et al. 2018) and comparable activities to ground oneself with regard to a social group. It can be used to express membership or to distance oneself from others. Flexible language use is frequently linked to informal settings where norms are adapted, adjusted, and negotiated in interaction. Ego-documents, especially private papers, have been shown to allow for informal language production in writing; they are, therefore, particularly informative in this respect (cf. Elspaß 2012; Schiegg 2016; Van der Wal & Rutten 2013, among others).

The current chapter reports on a study investigating diaries from a nineteenth-century heritage setting in a German/English bilingual community in southern Canada. Variation in these data can be observed in two respects: (i) Changes in the forms and structures of the heritage language, German, in relation to the writer's age; and (ii) the choice of language (German/English) – and the corresponding script (German cursive script/ Roman script) – over time. The writer, a German-English bilingual, started keeping a diary at age 12 and continued to do so until late in his life (cf. Stolberg 2018, 2019a), affording an extraordinarily long-term perspective on intra-writer variation. In the current chapter, the earlier diaries (1867– 1900) are investigated. They cover a period marked by several changes in the writer's life and language use. While in the earliest entries, the writer is still in the process of acquiring a written register in German as well as in English, later entries exhibit skilled switches between German and English, several (though not all) of which can be linked to extra-linguistic events in the writer's life.

In his diary entries, the writer uses a low number of borrowings (partly accompanied by the appropriate switch in script), including some established loans for which independent evidence exists (e.g. in the *Berliner Journal*, a local newspaper). A noticeable feature of the diary entries is the alignment of names (place names, person reference) with the currently chosen language. This includes, for example, German and English versions of the names of his siblings (e.g. *Wilhelm*/*William*). Besides identifying the more general patterns of language mixing and switching in the diaries, the chapter investigates the usage patterns of proper names and different forms of person reference. It is argued that the choice of language (and script) and language-specific forms of person reference indicate social relationships and reflect the writer's increasing societal integration as he grows up and eventually establishes himself socioeconomically.

## 2 Research interest

Intra-writer variation is a phenomenon that has not received much research attention in the past. It can offer valuable insights into the style and register competence of individual writers and sheds light on the strategies with which writers adjust their language production to various settings and addressees (cf. Hernández-Campoy 2016). For in-depth studies of such variation, the availability of sufficient data, qualitatively and quantitatively, from a single individual is crucial. The current case study offers a rich data base in this respect, covering a total of almost 70 years of the writer's life. In addition, and crucial for the proper contextualization and understanding of the variation found, in-depth extra-linguistic information on the person and the community is available. While the diaries are not dialogic in form, they still reflect the writer's perception of social relations and adequate language choice through his choice of person reference. The study thus contributes to a better understanding of the individual, social and societal factors that interact to result in specific patterns of intra-writer variation in language use and language choice, and what functions such variation can fulfil.

From a wider perspective, this study of bilingual intra-writer variation across three decades sheds light on heritage language development (cf. Montrul 2016; Polinsky 2018) and preservation in the individual and in the community, on changes in language choice and language dominance across the lifetime, and on the flexible use of available linguistic resources to serve communicative and social needs.

## 3 Socio-historical setting

Starting in the late eighteenth century, (Pennsylvania) German-speaking Mennonites migrated from Pennsylvania to Ontario and established settlements in the area of modern-day Kitchener in southern Ontario (Bloomfield et al. 1993; Hayes 1999). The county was named Waterloo County, and in 1833, the town of Berlin was officially founded. The population consisted to a large part of Mennonites and of immigrants from the German-speaking parts of Europe (immigrated directly or via the USA). The use of German, side by side with English, was widespread and well established (cf. Lorenzkowski 2008, 2010). Around 1870, more than 50 % of the residents were of ethnic German origin, and German was reported as the dominant language of the area (Bloomfield et al. 1993).1

During the nineteenth and into the early twentieth century, several families of German origin played an important role in the economic and political life of Waterloo County and, in particular, in the town of Berlin/ Kitchener. Archival materials show that German was preserved over several generations in (some of ) these families.2 German was used as a family and a church language. It was also important in business and in education, and there were schools with German as the language of instruction, at least until the end of the nineteenth century (Grenke 2018; Lorenzkowski 2008; McKegney 1970). Coschi (2014: 315) points out that even 'despite dwin dling enrolment in German classes, the 1901 census reported that nearly 90 % of Berlin's residents of German origin claimed German as their mother tongue, suggesting that many learned German in the home as opposed to the formal setting of the classroom'.


## 4 Data base and methods

#### *4.1 The data base*

Louis Jacob Breithaupt (henceforth LJB), the writer whose language use is analysed here, belonged to one of the most influential German-origin families in Berlin. He was born in Buffalo, New York, in 1855, as the eldest of ten siblings. His maternal and paternal grandparents were first generation immigrants from German-speaking Europe, his father having immigrated as an adolescent.3 The family moved to Berlin, Ontario, in 1861 where LJB grew up and was rooted throughout his life. He died in 1939. Being the owner of a tannery and a successful businessman, he was actively involved in politics and in the local church community, held various high-level positions (e.g. mayor of Berlin, 1888–89), and played a decision-making role in his hometown.

LJB kept a diary from childhood until a few years before his death. The preserved diaries cover a total of 66 years, from 1867 to 1933. LJB presents himself in the diaries as bilingual (German and English), biscriptal (German cursive script and Roman script) and biliterate (attending school in German and English). The early diaries, starting when LJB turned twelve, show a childlike handwriting with uneven letter sizes and spacing. Over time, not only the handwriting matures but also the written language use changes in ways that can be attributed to language development and an increasing competence in the written registers of German and English. In terms of content, LJB reports on everyday occurrences like school, household chores, social interactions with relatives, friends and neighbours, later also on his higher education (college), business matters, business and private trips, and family matters.

The data base for the current study consists of LJB's handwritten diaries from 1867 to 1900. The earliest diaries (1867–71) contain an entry for every

<sup>3</sup> His father and paternal grandparents (Breithaupt) were from Hesse, his maternal grandparents (Hailer) came from Baden and the Alsace.

day. For subsequent years, there are smaller and larger gaps, with only few years, however, for which no entries exist.4

#### *4.2 Methods*

For all available diaries between 1867 and 1900, overall language choice (German, English) and script choice (German cursive script, Roman script) was determined. German cursive and Roman script are two different ways of writing the letters of the Latin alphabet. While they differ in many letters, there is also some overlap. A word is considered to be in German cursive whenever the differing letters are written in this script. A word containing no letters in German cursive is considered to be in Roman script.

The data analyses included the identification and categorization of borrowings, person reference (names and forms of address) and place names. Qualitative analyses were carried out for the completely transcribed diaries of the first five years (1867–71). In addition, selected diary entries from 1872, 1875, 1880, 1888 and 1900 were transcribed and analysed in detail.5 The analysed sections cover a total of 54,000 words (graphic units), of which c. 24,000 units are in English and 30,000 units in German.

## 5 Findings

The data exhibit intra-writer variation which, broadly speaking, falls into two types: developmentally determined variation, and age-independent variation, the latter often motivated by extra-linguistic factors.

<sup>4</sup> There are no diaries preserved for the years 1877, 1882, and 1886.

<sup>5</sup> The selection was based on the availability of contemporary data from other family members, viz. LJB's mother, wife and daughter, in order to contextualize LJB's written language use within the family and the community (cf. Stolberg 2019a).

Among the developmentally determined features, we found variation in handwriting, in spelling (to a limited degree), in sentence length, in vocabulary size, in style or register use (including features of orality), and in pragmatic competence, reflected in variable forms of self-reference.

Variation in language choice (German or English), on the other hand, is not linked to age or development. It seems often determined by extralinguistic factors. Language choice can depend on the topic (e.g. reporting on a family member's death, in German), on the current geographic and language environment (e.g. travels in Europe, in German), or on a changed social position within the community (e.g. starting to work after finishing school; attending college, in English).

For the two scripts LJB uses, functions are clearly divided in general: Overall, LJB uses German cursive when writing German, and Roman script when writing English. A script change can occur when other-language items are used, as in the case of (nonce) borrowings.6 In this way, script choice indicates sensitivity for the language affiliation of a lexical item.7

In this chapter we focus on variation in handwriting as a developmentally determined feature, and on language choice, script choice and borrowing as examples of age-independent variation. Person reference, in addition, shows variation in relation to age and social development as well as in terms of language choice (independent of age). The following sections serve to illustrate the different phenomena.

#### *5.1 Variation in language choice*

Across the diaries, LJB changes between German and English as the predominant language at different points in his life. Sometimes, the reason is easily conceivable, while in other cases, it remains hidden. The overall


distribution of German and English in LJB's earlier diaries is listed in Table 16.1.

The first overall shift from German to English, aside from single-word switches or (nonce) borrowings, occurs with the beginning of February 1870, after LJB had been keeping his diary in German for almost three years (cf. (1) & Figure 16.1). The reason is not mentioned explicitly but LJB reports in mid-January 1870 that he has started working in his father's leather store and no longer attends school. It is conceivable that this change triggers his decision to switch languages (and scripts) in his diary.8

(1) Montag 31ten Jan [1870] Vater war heute fort mit "*Lady* um "Lumber" zu kaufen9 *Tuesday Feby 1st/70* Hr *Mr Clemens + another gentleman are here*<sup>10</sup> [Monday 31st Jan Father was away with "*Lady*" today to buy "Lumber"]

#### *5.2 Variation in script*

Script varies in the diaries along two axes: for developmental reasons (maturation), that is, over time, and for pragmatic reasons, that is, corresponding to language choice and/or highlighting names and otherlanguage items. Variation for developmental reasons affects the size and



Table 16.1. Language choice, 1867–1900

(*Continued*)


Table 16.1. Continued

Figure 16.1. Language and script switch (31 January to 1 February 1870).

formal regularity of letters and includes spelling and punctuation (to a moderate degree). Variation with respect to pragmatic function relates to the parallel use of the German cursive and the Roman script and does not show a correlation with time or the writer's age. Rather, the most obvious connection of script variation is with language choice: German cursive is associated with German, and Roman script with English, in accordance with the practice commonly found in Europe (cf. Footnote 5) as well as in the contemporary local community (cf. Stolberg 2019a, 2019b). In the majority of cases, this distinction is also carried through with otherlanguage items, such as (nonce) borrowings or proper names. In addition, Roman script can be used within German sections to highlight names, especially at first mention.

## 5.2.1 Developmental variation in handwriting

Over the first decade, the diaries document LJB developing a trained and skilled handwriting, as a comparison of Figures 16.1 and 16.2 with Figure 16.3 illustrates.

With the increasing dominance of English over time, LJB's use of German cursive in his diaries becomes rare. Even in the later diaries, however, there is evidence for his using both scripts. In 1888, for example, he includes a short paragraph in German within an otherwise predominantly English diary (see Figure 16.3), attesting to his continued fluency in German cursive.

## 5.2.2 Script choice

Already in the earliest diaries, LJB has mastered both scripts and employs them according to the language he uses. The first example of the division of work among them is found in his very first diary entry, of 4 March 1867 (cf. 2), when he lists the presents he received on his twelfth birthday the day before. Here, he switches to Roman script for the English part (*marbles*) of the hybrid compound *glas marbles* (Figure 16.2, line 2).

Figure 16.2. Hybrid compound *glas marbles* (4 March 1867).

Figure 16.3. Letter of condolence (16 June 1888).

(2) Bleistifthalter u. 5 glas *marbles* auch für mein Geburtstag. [4 March 1867] [pencil holder and 5 glass *marbles* too for my birthday.]

The entry of 16 June 1888, as a late example within the investigated material, not only attests to LJB's continued use of German (including an appropriate register choice) but provides an example of a self-quotation accompanied by a switch in language and script (cf. (3) & Figure 16.3).

(3) *Much sorrow is felt in our Town with the "Fatherland" in its heavy bereavement. As Mayor I cabled to-day as follows:* Berlin Canada – Ihrer Majestät der verwittweten Kaiserin Viktoria. Berlin Deutschland: Die Bürger von Berlin, Canada, erlauben sich ihr tiefstes Beileid über den Tod des Kaisers, auszudrücken. Breithaupt, Bürgermeister. [16 June 1888] [Berlin, Canada – To her Majesty the widowed Empress Viktoria. Berlin, Germany: The citizens of Berlin, Canada, permit themselves to express their deeply felt condolences regarding the death of the Emperor. Breithaupt, Mayor]

#### *5.3 Borrowings*

The diaries contain various instances of single words from the respective other language. This is quite common in bilingual language use, and such elements have been variably referred to as nonce borrowings (e.g. Sankoff et al. 1990), single word insertions (e.g. Myers-Scotton 2002), or lone other-language items (Poplack 2012 and earlier; Stammers/Deuchar 2012). It is notoriously difficult, if not impossible, to distinguish nonce borrowings from single word code-switches, be it in written or spoken language (for a short discussion, see Poplack 2012). In this chapter, otherlanguage items are referred to as borrowings. They are distinguished, if possible, from established loans which are used regularly by the writer or can be shown to be in (established) use in other contemporary sources as well. In the latter case, it is assumed that they constitute an established part of the local variety of German or English. Scriptal integration can signal establishedness (at least at the idiolectal level), similar to the function of phonological integration in speech.

Other-language items are English items within the German text, and rarely vice versa. We suggest that this imbalance is a reflection of the dominant function of English in LJB's bilingual environment.

(4) illustrates the use of two hybrid compounds, *Lederstohr* ('leather store') and *Gram̅ar Schule* 'grammar school'. Both are written in German cursive, implying that LJB considers them as German. It should be noted that *Lederstohr* is also attested in a local German newspaper of the time (in the spelling *Leder-store*),11 providing evidence that it is an established (partial) loan in the German speaking community.

(4) Die Gram̅ar Schule fing heute an Ich gehe nicht mehr hinein sondern helfe in dem Lederstohr. [10 Jan 1870] [Grammar School started today. I do not go there anymore but help out in the leather store.]

The following examples show ways of handling (nonce) borrowings in terms of script. (5)–(7), and Figure 16.4, exhibit English items in Roman script within a German environment (*Tiles*, *Steam gauge*, *Picnick*). In (5) and (6) (*Tiles*, *Steam gauge*), the English words appear in Roman script but are not marked otherwise.12 In (6), the English item, *Steam* 

<sup>11</sup> In the *Berliner Journal*, for example, in 1862 (Uttley 1975: 100).

<sup>12</sup> The capitalization of these borrowings can be taken to be a concession to German as the orthographic matrix language.

*gauge*, follows the corresponding German item *Dampfmaß* (in German cursive), demonstrating LJB's bilingual competence.13


In (7) and (8), marking goes beyond a simple switch of script. The borrowings are additionally highlighted by double quotes, flagging their 'outsider status', and *Picnick* (Figure 16.4, bottom line & (7)), furthermore, is hyphenated (as if being a compound) and printed in larger letters that cross the writing line which LJB otherwise observes carefully. This expressive marking suggests that the word, and possibly the concept, is unfamiliar (at least in writing) to LJB at this time; here, scriptal marking serves the pragmatic function of emphasizing (in an emblematic way) that the item is considered irregular.

Figure 16.5 & (8) shows the rare case of a German item (in German cursive) within an English section (*Sauerkraut*, in line 2). Note that there is a self-correction in the first letter of the word *Sauerkraut*: Apparently, LJB started writing a Roman script capital *S* but changed it to a German cursive capital *S*, showing his deliberate decision to align script and language.


Figure 16.4. *Picnick* (1 June 1868).

Figure 16.5. *Sauerkraut* (25 October 1870).

#### (8) *We made some "S*Sauerkraut*" this evening*. […] [25 October 1870]

The examples show that it is not one specific script that carries pragmatic meaning but the switch in script in itself, similar as to what has been observed for oral code-switching in communication (e.g. Lattey & Tracy 2005). The switched item is set apart visually. While in the cases discussed here, this visibility coincides with a congruent script choice (matching language and script), in later diaries further patterns emerge, utilizing the visibility of a script switch to highlight new information and changing to the script of the predominant language for given information, as is illustrated for the place name *Paris* in (9) (in German cursive) and (10) (in Roman script). In such cases, the script switch fulfils the pragmatic function of indicating newness of an information.16

16 There has been limited research on the pragmatics of script in a setting like the current one where script is not a matter of language politics but is looked at through the lens of individual variation. See, for example, Spitzmüller and Bunčić (2016) on German biscriptality, Schiegg and Sowada (2019) and Choksi (2019) for a pragmatics


Spitzmüller and Bunčić (2016: 289, 300) relate the function of the Roman script (within German cursive texts) as marking 'foreignisms' (among other functions). While this was certainly true for (German-speaking) Europe, in the current case the concept of 'foreignism' cannot be applied in a straightforward manner. Both German and English are part of the German-Canadian identity of the community and its members (cf. e.g. Lorenzkowski 2008), and English is not perceived as foreign. Rather, the association is between the German language and German cursive, and between non-German language(s) and Roman script – not only for English but also, for example, for French names such as *Champs Elysées* in (10).

## *5.4 Script, person reference and place names*

## 5.4.1 Script and person reference

While LJB is largely consistent in using German cursive for German and Roman script for English, there are 'borderliners', and they behave in specific ways. Besides (nonce) borrowings, these are person reference, place names, and horses' (and dogs') names. As a category by themselves, LJB's siblings' names appear as doublets, that is, they are not only adjusted in script but their form is varied depending on language context: *Wilhelm*/

perspective on script choice and script switching, Sebba (2009) and Unseth (2005) for sociolinguistic perspectives on script choice, and Androutsopoulos (2020) on Greek/English trans-scripting. On the social and interactional meaning of written code-switching see, for example, Sebba et al. (2012) and Schiegg and Foldenauer (2021).

*William*, *Johann*/*John* and *Esra*/*Ezra*. 17 Person reference can thus shift with the language environment (within the diary). Thus, these items can and do appear in different forms: either integrated into the surrounding language and script, or marked as different by using the other language.

As person reference is language specific in LJB's diaries, it is therefore also subject to choice. The first diary entry after LJB switched to English in his diary in 1870 demonstrates his attention to the language specificity of address forms and his decision to match form of address, language and script. It is noticeable in the self-correction at the beginning of the first line (see Figure 16.1 & (1)).

A later example shows, however, that person reference can also occur in the other language and script. In (11), two persons bearing the same last name are introduced, distinguished by different forms of address. The overall entry is in German (language and script); the first instance of person reference (*Rev. Mr Hoare*) is in English and in Roman script, the second (*Hrn.* [= *Herrn*] *Hoare*) is in German and German cursive. The contextual information does not resolve this difference: LJB reports that he had visited Rev. Mr Hoare in a small town near Paris, France, and adds that he is the father of Mr Hoare, an acquaintance from his hometown, Berlin, Canada. The latter information, as well as the English last name, seems to favour the use of an English address form and of Roman script for both referents. This expectation is not met by the data. A potential parallel is the script variation with *Paris* in (9) and (10). We argued that pragmatic reasons could explain the variation, distinguishing between new and given information. In the current case, it seems conceivable that the English form signals less familiarity with the referent than the German form and can thus indicate a difference in personal relationship between LJB and the persons mentioned. On a more abstract level, then, familiarity (also in the

17 Especially LJB's brothers Wilhelm (William) and Johann ( John) are close in age to him and feature frequently in his early diaries, for example, regarding school, household chores, running errands or getting together with friends. Siblings who are much younger than LJB are not mentioned in the (earlier) German diary contexts; and some of the names do not have different written forms in German and English (e.g. *Albert*, *Daniel* or *Melvina*).

sense of something already being known, i.e. given information) emerges as a relevant factor for pragmatically motivated language/script choice.18

(11) Ich brachte den Vormittag mit Geschäften zu und war Nachmittags in *Lec*, einem kleinen Ort außerhalb der Stadt u. besuchte *Rev. Mr Hoare*, Vater des Hrn. Hoare (von der Merchants Bank) in Berlin19. [24 June 1878] [I spent the morning doing business and in the afternoon, I was in *Lec*, a small town outside of the city, and visited *Rev. Mr. Hoare*, father of Mr. Hoare (of the Merchants Bank) in Berlin.]

#### 5.4.2 Script and place names

Place names tend to be written in German cursive in the early diaries, but occur more often in Roman script in later diaries. This is particularly noticeable in the German entries covering LJB's journey to Europe in 1878 (e.g. (9)).

During the earlier period LJB seems to be more committed to creating homogeneous texts with regard to language and script choice. One strategy to achieve visual homogeneity is illustrated in (5) and (6), with English names written in German cursive (*Park Hill*, *New York, Toronto*). While in (5) it can be argued that the local place, Park Hill, is considered part of the German sphere by LJB, this explanation seems less likely for New York and Toronto (6).

Considering the overall evidence for treating other-language items, two conflicting priorities can be recognized: to achieve overall homogeneity in


script (accepting a 'mismatch' between language and script, e.g. in (5) and (6)), and, in contrast, to aim at congruence between language and script (accepting a visual difference, e.g. with *Paris* and *Hotel de Manchester* in (9) and *Champs Elysées* in (10)). The ranking of these two factors can change and appears to depend on the immediate textual and conceptual context, with additional factors, such as distinguishing new from given information or expressing distance/familiarity, adding to the range of options.

## 6 Discussion

The analyses of the data, with a special consideration of person reference and place names, showed that there is no period when one language is used to the exclusion of the other. There are English items in the (early) near-monolingual German sections, and German names or short phrases in the (later) near-monolingual English sections. Usually, there is one dominant language in the entries, with the other language playing a subordinate role, only surfacing in (inserted) borrowings or language specific items, such as names. Further, we observed back-and-forth shifts between German and English over the three decades (see Table 16.1), the successive steps of language choice reflecting the individual process of LJB's language (dominance) shift.

LJB's shift to English when he started to work in the leather store (in January 1870) suggests that English, notwithstanding the strong German mark of the community during the later nineteenth century, was felt to be the language of public communication and adult business life, from the perspective of an adolescent growing up in this community. The step from being a school boy to becoming an active member of the business community may have made it seem appropriate for him to stop using the family or private language also in his diaries as an indication to himself that he now belonged to the adult world. It also implies that German was understood to be a private, family and home language to some degree.

The analyses have shown that not only language choice but also script choice is highly relevant in these handwritten data (cf. Schiegg & Sowada 2019 for similar results on handwritten data from the same period). Two different patterns could be identified: (a) the use of German cursive with English items and Roman script with German items, which we propose to refer to as *script crossing*; and (b) a congruent script choice, matching language with script. Script crossing leads to the erasure of visual distinction by matching an other-language item visually with its scriptal environment (e.g. *Park Hill* in (6) and *Paris* in (10)), similar to the acoustic effect of phonetically integrating a borrowed item in speech. Congruence between script and language, in contrast, heightens the visibility of other-language items, sometimes additionally reinforced by quotation marks or larger-size letters (in (7) and (8)). For person reference within the bilingual and the formal/ informal space of variation, patterns of indicating familiarity/immediacy vs. distance emerged and were expressed by different forms of address.

Variation between the German and the English form of LJB's siblings' names (e.g. Wilhelm/William, Johann/John, cf. Section 5.4) indicates that also his family, just as the local community at large, employs both languages in their everyday lives. In his early diaries, LJB positions himself as German-writing, with only occasional insertions from English. German may not be his (only) dominant language, but it is the dominant language of these diaries. Over the next decade, this balance shifts to a (quantitatively) more even relationship between the two languages, in that some parts of the diaries are in German and others are in English. The data reflect LJB's high versatility in using script choice to highlight names and language switches and his ability to employ his script repertoire for pragmatic functions. Towards the end of the investigated period, English is the language predominantly used, with German playing no more than a marginal role in the diaries. Since LJB is largely consistent in aligning script with language, this shift in language use results in an ever-decreasing use of the German cursive script over time.

This study examined intra-writer variation with a focus on the role of developmental factors and pragmatic functions in language choice and script choice, discussing the nexus between script and language in terms of information structure and social relations. The results highlight the relevance of linguistic and scribal strategies for societal self-positioning within the larger context of a bilingual community.

## References

#### *Primary sources*

GA 24 Breithaupt Hewetson Clark Collection / Section 2.5.3 Diaries of Louis Jacob Breithaupt (1867–1900). Waterloo, ON: University of Waterloo Archive.

#### *Secondary sources*


Hayes, Geoffrey. 1999. From Berlin to the Trek of the Conestoga: A Revisionist Approach to Waterloo County's German Identity. In: *Ontario History* 91(2): 131–49.

Hernández-Campoy, Juan M. 2016. *Sociolinguistic Styles*. Malden: Blackwell.


#### part III

From intra-writer variation to variation beyond the individual

#### Theodore Markopoulos

# 17 Intra-writer variation in Early Modern Greek notary acts: Morphosyntactic patterns of accommodation

#### Abstract

In this chapter I address the issue of intra-writer variation in two notary books by Maras, a sixteenth-century notary from Crete. More specifically, I examine the morphosyntactic variation attested in three constructions: (a) the negative particle utilized in participial contexts, (b) the morphological forms used formulaically to refer to members of the elite and (c) the usage of the modal construction ήθελα [would] + infinitive. The investigation, carried out in both a quantitative and qualitative manner, reveals that patterns of intra-writer morphosyntactic variation can be accounted for as possible instances of accommodation/ audience design, in the sense of Bell (1984, 2001). This conclusion opens new dimensions in the examination of intra-writer variation in Early Modern legal texts, and even beyond.

## 1 Introduction

The notion of intra-speaker/stylistic variation has long been the subject of extensive research, especially in relation to the multiple factors that seem to govern the variability of speakers' linguistic choices (cf. e.g. Bell 1984, 2001; Eckert & Rickford 2001; Coupland 2007). The inherent dif ficulty in accounting for such variation only gets magnified when written texts replace speakers' utterances, that is, when intra-speaker becomes *intra-writer* variation, due to the well-known limitations of historical linguistic data, such as their fragmentary character, dubious dating, ambiguous/unknown communicative context, to name but a few. Still, it has been shown that such historical sociolinguistic investigations can be carried out (cf. recent discussions in Bülow & Pfenninger 2021; Werth et al. 2021) especially in specific genres/cases where at least some metalinguistic evidence is available, for instance in various private letter archives (cf. e.g. Auer 2015; Schiegg 2018; Hernández-Campoy & García-Vidal 2018). This chapter seeks to add to this growing literature on intra-writer variation by tackling a rather under-investigated and not quite promising – at least at first glance – genre, the early modern notary books written in Greek on the island of Crete in the sixteenth century.

Insofar as the main aim of historical (socio-)linguistic investigations is to unearth evidence in relation to the linguistic reality of the past, in all its inherent variability, the late medieval – early modern (c. fifteenth to seventeenth century) notary books do not constitute an 'ideal' linguistic material: They contain to a great extent legal documents/acts, and subsequently, they belong to a very 'formal' register, prone to contain formulaic/official linguistic traits and rarely to offer glimpses of different, either spoken or written registers (cf. the discussion on the involved-informational cline in Biber 2001). On the other hand, there are obvious advantages when examining such documents, based mainly on the fact that date and place of origin are usually known, and the texts have effectively survived as autographs, therefore providing us with direct access to the notary hand. Since this crucial metalinguistic evidence is a given in the case of notary books, potential instances of variation exhibited may be more amenable to a variationist linguistic analysis than traditionally considered. Still, the question remains whether this is also valid as far as *intra-writer* variation is concerned. Therefore, the chapter's aims are two-fold:


The chapter is structured as follows: Section 2 gives a brief historical background on the material consulted, sketching the overall sociolinguistic context of sixteenth-century Crete ruled by a Venetian/Italian elite, as well as providing what information can be found on the notary himself, Michele Mara(s), whose books are examined. Section 3 investigates three instances of stylistic morphosyntactic variation most prominent in the books of Maras and, finally, Section 4 discusses the findings of both the quantitative and qualitative analysis of variants and concludes the article.

## 2 Historical background

#### *2.1 The sociolinguistic situation in sixteenth-century Crete*

Since the early thirteenth century, the island of Crete was under Venetian rule, which would ultimately last until 1669. It is well-established that the long-standing cultural and linguistic contact between the Greek-speaking majority and the Venetian- (Italian-) speaking elite and community had led to a cultural osmosis and to the emergence of a Cretan 'mixed' identity, especially in the urban centres of the island (cf. Maltezou 1997; McKee 2000). This identity was effectively created and further strengthened by the noticeable multilingualism of many Cretans, as manifested in both historical and linguistic record (for more details cf. e.g. Markopoulos 2009, among others). To further complicate matters, literacy in the sixteenth-century Greek-speaking world (as in the whole history of Greek after the Hellenistic period, cf. Horrocks 2010) almost by definition implied the learning of an archaic variety of Greek, creating thus a situation where Greek-Italian bilingualism combined with diglossia between 'archaic' and 'vernacular' Greek. Obviously, in such a sociolinguistic context, the linguistic repertoire of Greek literate speakers, especially of those living in town quarters, must have been quite 'rich' and varied.

The main urban centre on Crete was Handakas/Kastron, the capital of the island, where multilingualism among both Greek- and Italianspeaking populations must have thrived. This is also the town where Michele Marà / Maras (Μιχαήλ Μαράς),1 our notary, lived and worked presumably throughout his life.

### *2.2 Who was Maras?*

Unfortunately, not much is known about Maras, the notary whose books are examined in this chapter. As a matter of fact, apart from his residence, almost nothing else is known about him, except that his son became a medical professional (Drakakis 2004). This by itself is not very enlightening, although it seems to indirectly suggest that Maras belonged to a 'middle' class of burgesses, who could afford to educate their children in an era when such an education was a privilege.

Maras' social status may be linked to his long-lasting career as a notary. Indeed, his notary books cover an impressive time span of 40 years (1538– 78), rendering him one of the most (if not the most) prolific notaries in the whole era of Venetian-ruled Crete, with an impressive eighteen long books (*catastica*) by his hand surviving today. The language of all his notary acts was exclusively Greek, a feature common enough for sixteenth-century Crete, when, apart from the originally Greek-speaking population, even an – either small or large – part of the originally Italian-speaking population must have shifted into Greek (cf. Markopoulos 2007, 2009).

## 3 Corpus and methodology

Only two of the surviving eighteen books by Maras have been published, and neither is included in any electronic corpus. Therefore, it is rather impossible, for the moment, to examine the totality of his writings, although

<sup>1</sup> The notary is listed in the Venetian archives as Michele Marà, while he himself used the Greek form of his name (Μιχαήλ Μαράς) in his documents. In this article, the form Maras, constituting a transliteration in Latin characters of the notary's surname, is used throughout.

that must constitute a clear *desideratum* for future research, given the possibility it offers to investigate the writings of an individual across large parts of his lifespan. For the purposes of this chapter, the two published notary books were partially examined: More specifically, the corpus of the chapter consists of the third and final part of book 148 (Mavromatis 2009),2 dating from 1538 to 1539 and comprising 681 notary acts (approxi mately 140,000 words), together with the first part of book 149 (Drakakis 2004), dating from 1549 and comprising 389 notary acts (approximately 90,000 words). The sheer bulk of the books is impressive, as Maras seems to have produced daily the same number of documents that most notaries in Crete produced in over a month (Drakakis 2004: xvii)! The chronological gap of ten years between the production of the two books adds an – admittedly limited, but still extant – diachronic dimension to the investigation.

Both notary books have similar contents, typical of their kind: a wide range of agreements and contracts (including wedding agreements), leases of land, wills and various other types of notarial acts. Maras' clientele resided mainly in Handakas/Kastron and nearby villages, but more rarely came from other areas of Crete and even from other islands, such as Santorini, Mikonos, Anafi, Rhodes, Zakynthos, etc. (Drakakis 2004: xxix). As far as the linguistic variety of the acts is concerned, it could be considered a 'typical' notarial language of the late medieval/early modern Greek world, exhibiting particular features such as archaizing and formulaic expressions, as well as various Italian loanwords and legal terms (cf. e.g. Manolessou 2008).

From a methodological point of view, the investigation of linguistic material with such features can hardly be straightforward, especially when the research objective is the study of variation. Given the formulaic character of large parts of the notarial acts, a combination of quantitative and qualitative examination is arguably necessary, to ensure that the quantitative data does not distort the linguistic picture due to the high token frequency of formulaic constructions.

2 The numbering of the notary books is the one followed in the Venetian archives, and has been used in their modern publication as well (Mavromatis 2009 for book 148; Drakakis 2004 for book 149).

A final methodological point concerns the important question whether notaries in general and Maras in particular are the real authors, the real 'egos' that produced these documents. The professional production of notarial acts could be regarded as an instance of delegated writing, since notaries supposedly wrote down what was demanded from them by their clients. Concerning especially the morphosyntactic level, it has been noted that medieval scribes apparently showed minimal involvement in personal letters dictated to them (cf. Bergs 2013 on the fifteenth-century English Paston letters). If notaries followed to some extent those scribal practices, one would be left to wonder whether instances of morphosyntactic variation observed in notary acts might be better envisioned as *inter-speaker* instead of intra-writer variation. To circumvent this potential pitfall, all three cases of morphosyntactic variation examined below occur in formulaic constructions and/or in formulaic parts of the documents, such as the beginning or the end, where presumably the notary himself is responsible for the linguistic features observed (cf. also Tuten & Torrens Álvarez 2021). The possible relationship between Maras' linguistic choices and intra-speaker variation will be addressed in more detail in Section 4.

## 4 Morphosyntactic variation in Maras

As already mentioned, this chapter focuses on three different morphosyntactic constructions attested in Maras' acts, namely:


Each construction is examined separately in both notary books, to determine whether a unified picture of intra-writer variation emerges, and if so, how this could be accounted for, given the formulaic contexts of use not readily associated with variation.

## *4.1 Patterns of negation*

One of the most conspicuous cases of variation observed in the books of Maras relates to the negative particle used to negate the indeclinable *-όντα* participle usually denoting adverbial meanings such as manner and time. Late Medieval/Early Modern Greek (similarly to Modern Greek) has two different clausal negative particles, *δεν* and *μην*, used in largely different semantic/syntactic contexts: *δεν* is used in realis, while *μην* predominantly in irrealis contexts (for instance subordinate clauses, commands, etc.). The situation is more complex as far as participial clauses are concerned: According to Holton et al. (2019: 1915f.), *μην* constitutes the 'traditional' variant since late antiquity in such syntactic contexts, whereas in Early Modern Greek *δεν* is the 'normal' option. In Maras' books, the negated *-όντα* participles are not very common, but still, the relevant variation is observed, as in the examples (1–2):


The quantitative analysis in both books is illustrated in Figure 17.1. Obviously, *δεν* constitutes the more frequent variant, as expected, although the distribution of the variants in the two sources is unequal: In the book of 1538, *δεν* is utilized in 88.1 % (37/42) of the attested cases, while in the book of 1549 in only 56 % (14/25). It could be argued that the striking difference in token frequency constitutes an indication of a diachronic development, in other words that we are facing a change in

<sup>3</sup> All translations are my own unless otherwise noted.

Figure 17.1. Token frequency of the negative particles *δεν* and *μην* in combination with the *-όντα* participle.

progress. However, the rather dramatic change observed in a short time span of a decade does not seem to support such an account, especially in the absence of major sociolinguistic developments that might have caused / facilitated sharp changes. Again, due to the rather idiosyncratic nature of the notary acts as linguistic material, a qualitative analysis is very much needed to understand the factors affecting linguistic choices.

Taking a closer look at all the attestations of the *μην* particle, we observe that in both books they all occur in the very same formulaic construction *μη έχοντα το εναντίον* [without having any objection] exemplified in (2). Not only that, but all such formulaic attestations are found in the very same type of notary act, namely parental transfer of property.4 It would be tempting, therefore, to assume that the use of the negative particle *μην* is exclusively limited to a very specific formulaic construction of a very specific type of notarial act. Although this might be a valid conclusion, it nevertheless does not tell the whole story.

To capture the factors responsible for the use of the *μην* particle, one needs to determine what this negative formulaic construction is in competition with in its context of occurrence. It is quite telling that in other documents of the same type (parental transfer of property), Maras uses a

<sup>4</sup> Book 148 (1538–39), acts 758, 760, 804, 833, 834; Book 149 (1549), acts 13, 14, 15, 56, 57, 108, 109, 110, 153, 213, 227.

different construction, exemplified in (3), in the very same context where the construction *μη έχοντα το εναντίον* was also used:

(3) Και τινάς να μην σου υπή το εναντίον And nobody that NEG you-GEN say the contrary [And (so that) nobody could / should say anything against you] (Maras 149, 9 / 1549)

Apparently, the formulaic construction in (3) must have constituted an alternative to the construction *μη έχοντα το εναντίον,* as they exhibited similar meaning and an identical context of occurrence:5 They were both used in the same type of act (albeit the construction in (3) could be used in other acts as well) and in the same context, namely in the part where it is stated that what is ordained in the act cannot be gainsaid by anyone. In this respect, the potential instance of morphosyntactic variation in the choice of the negative particle accompanying the *-όντα* participles (*δεν* or *μην*) should probably be examined in a different, more fruitful manner. More precisely, it could be argued that the -όντα participles were effectively negated by *δεν*, except in the case of the formulaic construction *μη έχοντα το εναντίον*. Furthermore, the latter was used as a variant in a specific context, where a different propositional formulaic construction *και τινάς να μην σου υπή το εναντίον* was also utilized. The question that naturally arises is what the main factor was – if there was indeed a clearly identifiable factor – determining the choice between the two variant constructions by Maras in such a highly formulaic context. A potential answer may be found in the 'audience effect' originally proposed by Bell (1984), that is, in the effect caused by the addressees and the auditors ('the third parties ratified in a conversation', Bell 1984: 172), in other words the parties present at the drawing up of the relevant notarial acts.

The moment the participants at each notarial act are taken into account, a clear pattern emerges: Crucially, in all occurrences of the negative participial construction *μη έχοντα το εναντίον*, the main party of the

5 As a matter of fact, there exists a third formulaic construction, *προμετέρω να φυλάγω από πάσα πίραξιν και εναντίον* [I promise to defend {you} against any objection], but, since it occurs only twice in the whole corpus (Maras 1549, 58, 152), it does not alter the overall picture in any significant way and will not be discussed further here.

act belongs to the elite (mostly of Italian origin) and/or lives inside the *burgo*, where all members of the higher social class dwelled. For instance, in 4/5 occurrences of the construction in the 1538 book, it is the very same person, *Ντομένεγος Φωσχαρής*, a lord (*άρχων*, see Section 4.2) and son of a lord, clearly of Italian origin and living in Handakas, who is the main person involved in the relevant acts. Interestingly enough, in the only occurrence of the construction where *Ντομένεγος* is not involved, his place is taken by *Αλωύζε Φωσχαρής*, probably his brother judging by his surname and the common father's name (*Νικολό*)! Obviously, more participants are mentioned in the acts of the 1549 book, but all, without exception, belong to the same upper echelon of Handakas society.

On the other hand, the textual distribution of the propositional construction exhibited in (3) is much more varied. In a total of fourteen occurrences across both notary books, six are found in acts drawn up in the burgo, another five in the outer city of Handakas (*Εξώπορτον*), mostly inhabited by the middle-class burgesses, while the final three attestations are found in acts written down in various villages in the countryside. Obviously, this formulaic construction was used by Maras across the board, without any indication of sociolinguistic – or for that matter 'strictly linguistic' – marking / specialization.

If we combine these facts with the formulaic context of the constructions, which effectively renders unlikely the active involvement of any other participant apart from the notary himself, we are led naturally to the conclusion that this instance of intra-writer variation may constitute another case of 'audience design' (Bell 1984, 2001), albeit a rather indirect one. In other words, even if Maras' clientele were present in the drawing up of the acts, this case of variation should be seen as 'stylistic' in nature, since the linguistic choices observed in the formulaic parts of the documents should be attributed to Maras himself. Apparently, Maras accommodated linguistically to his clientele by selecting the more 'formal', 'archaic' participial variant only when dealing with members of the social elite, presumably because such variants were known to both Maras and them and used mostly in 'high' registers. Admittedly, the effect of audience design on the morphosyntactic level is more subtle and perhaps less convincing as a factor determining the choice of variants, but still it becomes increasingly so when the evidence of the following sections is considered.

## *4.2 Morphological variation*

Another pattern of morphological variation involves the forms of the stereotypical designation *ο (ευγενής) άρχων* [the (noble) lord], which was very often used by Maras to refer to members of the social elite. Interesting facts arise when one looks into the declensional patterns of the noun *άρχων*, especially with regard to the genitive case, as exemplified in (4):


More specifically, the old, archaic genitive form *του άρχοντος* (4a), surviving presumably in written registers because of the Greek diglossia, alternated in Maras' documents with the indeclinable form *του άρχων* (4b)*,* a rather novel variant used presumably in low-registers where the ancient declensional patterns had become obsolete by the late medieval period (cf. Horrocks 2010: 286–88). The variation observed in the token frequency of the two variants can be seen in Figure 17.2.

As illustrated above, the number of occurrences of the two relevant variants is strikingly reversed across the two Maras' books: While the indeclinable form *άρχων* is the dominant form in the first book (100/136, 73.5 %), the same does not hold for the second book, where the more 'learned / archaic' form *άρχοντος* is much more frequent (60/73, 82.2 %). Again, as argued in Section 4.1, the very short time elapsed between the

Figure 17.2. Token frequency of genitive variants.

production of the two books disfavours any explanation appealing to diachronic developments. Obviously, one cannot rule out the possibility of a gradual change in the linguistic choices made by Maras himself, who might have steadily become more inclined to use 'archaic' forms in his acts, although much more evidence would be needed to substantiate such a claim. As will be seen below, it is rather more likely that 'audience design' is once more the major factor responsible for these frequency patterns.

The data in Figure 17.2 seems to partly correlate with the facts concerning the residence/provenance of the individuals mentioned in the whole body of the relevant documents. To be more precise, in the first book (1538) most participants are villagers (approximately 52 %), while those resident in Handakas amount to 42 % of the total; in the second book (1549) the situation is reversed, villagers constituting 36 % and town-dwellers 52 % of the total participants.6 However, even this reversal of numbers is arguably inadequate to account for the striking difference in the distribution of the two morphological variants. To achieve a better overall idea as to whether *there is* in fact a relationship between this instance of variation and the social identity of Maras' clientele, a more fine-grained examination was carried out. In particular, the residence / provenance of the individuals who ordained the specific acts containing all attestations of the relevant construction was examined. This was made possible due to the notarial

6 In both books, the remaining percentage of participants refers to individuals originating from outside Crete.

practice, adhered to strictly by Maras, to write down the provenance of the main auditor of the act, that is, of the individual who came to his office to ordain an act. So, each occurrence of the genitive nominal construction was matched to the residence of the main participant, mentioned in the beginning of each relevant act. As a result, a much clearer image emerged, depicted in Figures 17.3–17.4.

Figures 17.3–17.4 shed new light on this issue, as they manifest rather interesting patterns governing the choice of morphological variants in Maras' documents. First of all, concerning the 'archaizing' form *του άρχοντος* (Figure 17.3), it is obvious that it found its way in Maras' acts much more

Figure 17.3. Residence of main participants in acts exhibiting the form *του άρχοντος*.

Figure 17.4. Residence of main participants in acts exhibiting the form *του άρχων*.

frequently when the parties involved resided in Handakas, the main town of the island. As a matter of fact, this tendency is very pronounced in the data obtained from the second book of 1549, where this form is utilized almost exclusively when the participants are residents of Handakas. Concerning the indeclinable *άρχων* form (Figure 17.4), the data is not so clear-cut: Apparently, its use was not by any means related to the participants' identity in the first book, but in the second book of 1549, not only its frequency was demonstrably decreased, but it had also become clearly associated with the auditors' dwellings, as it occurs in acts related to villagers 61.5 % of the time (8/13).

In attempting to account for these patterns, one should initially determine the extent to which the morphological variation observed constitutes an instance of inter-speaker variation, intra-writer variation or a combination of the two. It is possible that Maras sometimes worked as a delegate scribe, writing down the – almost – exact words of his clients; consequently, this instance of variation could perhaps be seen as rather indirect evidence for social variation, leaving little room for a potential role played by Maras himself. Although this could tell at least part of the story, nevertheless, as already argued, it is rather unlikely that this professional approach extended to the formulaic parts of Maras' documents, which surely could have been prepared beforehand, to save time and effort. Moreover, the differences in the patterns in Figures 17.3–17.4 speak in favor of an active role for Maras, who must have been responsible – at least partly – for the very strong association between the variant selected and the social identity of the participants involved in each case. If this latter assumption is correct, then this intra-writer variation must at least partly be attributed to the notion of 'audience design', similarly to what has been observed in Section 4.1 for the negative participial construction.

But even a combination of delegated writing and audience design may not be enough to provide the complete picture, as, in some cases, the two variant forms are found in the very same document, almost next to each other. Consider for instance (5):

(5) Κομεσιόν κάμνωμεν ημείς Ντζουάν Γρηλιώνις του ποτέ Commission make we Djuan Grilionis the-GEN late

```
ευγενεστάτου άρχοντος μισερ-Μάρκο και Όρισα, θυγατέρα
noblest-GEN lord-GEN sir-Marko and Orissa, daughter
αυτού και γυνή του ευγενή άρχων μισερ-
his and wife the-GEN noble-GEN lord-GEN sir-
Μάρκο Σεγρέδο […]
Marko Segredo […]
[We make a commission, me Djuan Grilionis of the late noblest lord sir Marko and 
Orissa, his daughter and wife of the noble lord sir Marko Segredo […]] (Maras 
149, 141 / 1549)
```
The fact that the two variants co-occur in the very same document, next to each other, points towards an explanation based on Maras' active involvement. In this case, the variation might be simply the result of the notary's playfulness (cf. also Tuten & Torrens Álvarez 2021) or an attempt to differentiate the two individuals involved by using a varied formula.7 Such writing practices notwithstanding, the clear patterns illustrated in the figures above are too systematic not to convey a distinct social meaning, to a great extent because of 'audience design' from Maras' part.8

## *4.3 Analytic modal constructions*

The final instance of morphosyntactic variation to be investigated in the documents of Maras concerns the occurrence of a relatively novel modal analytic construction, involving the verbal form *ήθελα* [would] + infinitive, as in (6):


[And if something should be found against you and they should take the abovementioned vineyards from you […]] (Maras 149, 310 / 1549)

This construction, emerging and becoming very popular in Late Medieval Greek (cf. Markopoulos 2008) is used mainly in conditional contexts (both protasis and apodosis). It is also very common in the books of Maras, as it is used to build multiple formulaic expressions (e.g. *ήθελεν ήσθαι* [would be], *ήθελεν φανή* [would seem/appear (to you)] etc.).

Although morphosyntactic variation regarding mainly the exact verbal form (impersonal vs personal) and the type of the complement (finite vs non-finite) of the construction exists, it is really limited numbers-wise and does not alter the overall picture in any significant way.9 Therefore, the chapter focuses on the variation observed on the paradigmatic axis, where there is a multi-layered variation: (a) in the very same act, (b) in the same book and (c) among the two books by Maras between the occurrence of the analytic construction with *ήθελα* and the occurrence of other (usually synthetic verbal) forms utilized instead in identical morphosyntactic contexts, as in (7)–(8):


As can be seen in (7–8), the analytic construction (*ήθελα ατεντέρι*) was used in the very same context where a synthetic verbal form (*ατεντέρωμεν*) also occurs. The examples are taken from a rather formulaic part of their respective documents, and consequently, this instance of variation falls

<sup>9</sup> To be more precise, only seventeen instances of finite complementation and only a single case of an impersonal form of *ήθελε* are attested in a total of 609 occurrences of the construction.

Figure 17.5. Token frequency of the construction *ήθελα +* infinitive per act.

nicely in line with those discussed above (see Sections 4.1–4.2). To better understand the nature of the interplay between the analytic and the synthetic constructions, an examination of the token frequency of the analytic construction per act was carried out. To this purpose, all acts in the two books of Maras (681 acts in the 1538 book, 389 acts in the 1549 book) were examined manually, and all occurrences of the relevant construction in all personal forms of the verb *ήθελα* were isolated (349 in the 1538 book, 260 in the 1549 book). The relevant frequencies are shown in Figures 17.5–17.6.

A significant increase in token frequency of the construction is evidenced in Figures 17.5 and 17.6, which illustrate that ήθελα + infinitive was used more frequently in each act and in general in the whole 1549 book in comparison to the 1538 book. Given that the documents contained in each book are very similar in nature, this increase in the frequency of use of the

Figure 17.6. Token frequency of the construction *ήθελα* + infinitive per 1,000 words.

construction cannot be attributed to any differentiation in the type of the material. Moreover, it should be noted that the alternative of a synthetic subjunctive verbal form exemplified in (8) remained the most popular form in these specific contexts, exhibiting thousands of attestations in the acts which, as legal texts, abound with conditional contexts. In addition, from a diachronic perspective, the synthetic form outlasted the analytic construction, as it remains the 'standard' form in such contexts in Modern Greek, whereas *ήθελα* + infinitive has not survived. Therefore, the increase in the token frequency of the analytic construction does not correspond to any diachronic tendency in the community which could possibly lead to the demise of the synthetic form. Apparently, once again, this instance of morphosyntactic variation should be seen as stylistic variation.

The question remains if this instance of stylistic variation can be accounted for by appealing to the notion of 'audience design', similarly to what has been observed in Sections 4.1–4.2. From a methodological standpoint, though, it is rather infeasible to examine whether the analytic *ήθελα* construction is mainly used when the participants belong to an urban elite, due to the very high number of token attestations (609 in total). Only indirectly would it be possible to surmise such a connection, by looking at the identities of the individuals mentioned in the two books *en masse.* As already mentioned (see Section 4.2), the percentage of those belonging to the Handakas bourgeoisie rises from 42 % in the first book of 1538 to 52 % in the second book of 1549. The co-relation between the rise in the token frequency of the construction and the rise in the participation of towndwellers is quite evident, even though the nature of the linguistic material is not very susceptible to a proper quantitative analysis.

However, there are independent reasons that render the co-relation more plausible. It is certainly likely that the construction *ήθελα* + infinitive was popular especially among the nobility of Italian origin, dwelling mainly in Handakas. Let us recall that an intense language contact situation manifested in the town during the sixteenth century, resulting in extensive bilingualism and, ultimately, in language shift from most Italian speakers to Greek. It is well-known that, in shifting situations, many bilinguals are prone to morphosyntactic influence from their L1 to the L2 they shift to (cf. e.g. Thomason 2001). In this respect, it is important to note that a similar construction involving the equivalent verbal form *volesse* + infinitive existed in Venetian at the time (cf. Markopoulos 2008: 214 and references therein), presumably the first language / variety of the majority of the 'Italian-speaking' population on Crete and, consequently, it could well have strengthened the frequency of use of the *ήθελα* + infinitive construction.

## 5 Discussion and conclusion

From a modern perspective, notary acts could be considered as one of the least likely registers to exhibit intra-writer variation: Not only do they constitute legal texts and, as such, are prone to be carefully edited to achieve their final form, but they also contain numerous formulaic parts which, by definition, are not expected to feature variation. However, this assumption is not valid as far as the Early Modern Greek notary acts are concerned, as has been shown in the previous sections through the examination of three instances of morphosyntactic variation in the acts by Maras. In all three cases, the variation is observed mainly – if not exclusively – in formulaic constructions, a fact that, albeit slightly surprising, nevertheless constitutes part of the explanation.

To be more precise, in all three case studies the intra-writer variation was found to be associated with Bell's notion of 'audience design', as the notary attempted to linguistically accommodate to the variety presumed to be more 'acceptable' by the auditors of his acts. The active interference by Maras himself is arguably more pronounced in the formulaic constructions he used repeatedly throughout his books, as it is in precisely those parts that the linguistic varieties/choices of the participants must have played a minimal role. Therefore, the three case-studies of intra-writer variation speak volumes for the ability of Maras to navigate through a multitude of linguistic varieties of Greek utilized in written registers in this period, taking advantage of both the diglossic and the multilingual situation of sixteenth-century Handakas. Obviously, this is quite reminiscent of the observation that, all things being equal, stylistic variation is more likely to occur in the writings of educated individuals, who commonly possess and exploit a wide linguistic repertoire (cf. e.g. Auer 2015).

In this respect, the systematic investigation of intra-writer variation in early modern times appears to highlight the rich linguistic repertoire and the high degree of linguistic creativity from the part of authors, and especially notaries, who were very much linguistically engaged in the production of their acts. Even if not all instances of intra-writer variation are accounted for through 'audience design', notions such as the playfulness or the aesthetic choices of the writer involved only enhance the image of the creation of a legal text on sixteenth-century Crete as a dynamic process.

Nevalainen (2015: 250) has argued that, in the historical study of variation, we need to embrace 'a broader view of evidence than is customary in present-day studies based on spoken interaction'. The vast number of notary acts surviving from the period has the potential to unlock more secrets, and to shed new light on the issue of stylistic variation in such texts (and beyond). This type of legal document, perhaps rather unexpectedly, enters the central stage in the study of intra-writer variation.

## References


#### Laura Linzmeier

18 What shall we do with the 'writing' sailor?: Style-shifting and individual language use in a French navigation journal from the eighteenth century

#### Abstract

This chapter focuses on seventeenth- and eighteenth-century documents that were written during or after transoceanic voyages by French navigators. The main function of these logbooks and navigation journals as an information and documentation tool means that this genre leads the writer to follow linguistic, structural and content-related rules and routines (cf. Linzmeier 2022a). Nevertheless, the documents also include passages marked by stylistic variation in which the writers seem to flexibly change their linguistic repertoire depending on varying communication goals. Based on an exemplary qualitative analysis of a navigation journal from 1716 and its pragmatic functions, this study contributes to the current emphasis in historical sociolinguistics on structural and stylistic hybridity and flexibility, which can be observed in different writing contexts.

## 1 Introduction

This chapter deals with French logbooks and navigation journals from the seventeenth and eighteenth centuries. They were written during or immediately after transatlantic voyages by professional navigators, and this chapter examines to what degree the functional-pragmatic literacy linked to these writers allows for intra-writer and stylistic variation and flexibility within the same document. At first glance, the writers strictly obey linguistic, structural and content-related rules that the 'logbook' and 'navigation journal' text genre requires in order to convey nautical information. A qualitative analysis, however, reveals that style-shifting is a major component of this type of writing culture. This means that within the same document the writers switch back and forth between different discourse-traditional structures that can function, within passages, as nautical instructions and reports, letters of complaint or recommendation, commercial registers or even personal diaries (cf. Berthiaume 1990: 31–33). This chapter will subject these documents, which have not yet been extensively considered in linguistic research, to a text-linguistic analysis by looking specifically at the underlying functions of the individual text passages of a navigation journal from the early eighteenth century. This exemplary study is therefore in line with one of the focus areas of historical sociolinguistics, that is, the 'investigation of hitherto un- or under-explored data' (Säily et al. 2017: 2) – which not only comprise documents from private settings (such as private letters, diaries), but also official texts used in professional settings that allow for 'analyzing social and governmental schemata, furnishing details of the linguistic behavior of upper and upper-middle social strata' (Esteban-Segura 2012: 146).

The chapter begins with a brief introduction to research on French seafaring and its writing culture in the seventeenth and eighteenth centuries (Section 2). Section 3 is dedicated to the hypothesis that navigation documents can be seen as hybrid documents by nature. After presenting the theoretical framework (Section 3.1), this section presents an overview of these documents' historical background and their multi-functionality (Section 3.2). The empirical part consists of a description of the corpus, which is considered here as part of a larger project1 (Section 4), alongside an exemplary analysis of a *journal de navigation* from the beginning of the eighteenth century (Section 5). A brief conclusion summarizes the findings on style-shifting and hybridity in navigational documents (Section 6).2


2 Seafaring and the writing culture of navigational experts in the seventeenth and eighteenth centuries

The seventeenth and eighteenth centuries show a centralized expansion of the French navy and the professionalization of the training of navigators, provided by state institutions. Two *Ordonnances*, one from 1681 for the merchant marine and one from 1689 for the navy, were supposed to systematically reform the naval system and define the rights and duties of the crew in detail (cf. Lefrançois 2007: 0; Byington 2011: 14, 19f.; Vergé-Franceschi 1992). Amid an increase in transoceanic voyages, the authorities showed an emerging interest in science and technology and an obsessive desire with written, systematic documentation of new findings, knowledge and sea (trade)-related activities (cf. Soll 2008: 367f.). Maritime knowledge and terminology were collected in the form of reference books, manuals and dictionaries (cf. Ridel 2015). In this context, public schools, Jesuit-run secondary schools (*collèges*) and schools for the *Gardes de la Marine* were established to improve the training of navigators (cf. Russo 1964: 422).

As part of the *Ordonnances*, logbook keeping became an obligatory task on overseas travels, and the documents had to be handed over to the authorities afterwards. Manuals and guidelines were developed to lead the authors through the writing and documenting process (cf. Schotte 2013b: 99). This, however, demanded the promotion of maritime experts' reading and writing skills by establishing pedagogical measures, so that they became at least situationally routinized writers. Schotte (2013b: 99) points out that '[b]y the 1680s, the administration of Louis XIV began deploying varied pedagogical strategies to disseminate acceptable standards, commissioning textbooks, establishing schools, and instituting examinations'. In the *écoles de navigation*, textbooks helped to improve reading and writing skills, earlier logbook entries had to be reproduced, exams were established, while corrections and feedback were given by instructors (cf. Schotte 2013b: 99,

on findings from historical studies (esp. the works by Schotte). This article schematically summarizes the main points of this preliminary work in Sections 2 and 3.

106; Fauque 2000: 383). I therefore treat the seventeenth and eighteenthcenturies' maritime experts as part of schooled and situationally routinized groups of writers whose writing activity is primarily (but not exclusively) characterized by functional-pragmatic writing.

## 3 Logbooks and navigation journals: Hybridity and discourse-traditional variation by nature?

## *3.1 Theoretical framework*

This chapter examines the logbooks and navigation journals from a discourse-traditional and a text-linguistic point of view, which are closely interrelated. I follow Wilhelm's (2001: 470) well-established understanding of *discourse traditions* as 'habitualisierte, einem stetigen Wandel unterworfene Regelkomplexe' [habitualized rule complexes subject to constant change].3 Without detailing the research on this conceptual discussion here, in summary, every form of utterance (written and oral) is shaped by traditional and conventionalized patterns, the so-called *discourse traditions*. 4 Looking specifically at written material, we can make a further division of discourse traditions into non-literary texts (e.g. legal texts, instructions for use) and text genres that are literary and artistic in nature (e.g. poetry, prose) (cf. Schöntag forthcoming: sec. 3). Massicot (2021: 67), referring to Koch (1988a: 341f.), emphasizes, 'dass solche Traditionen als immer wiederkehrende historisch bedingte sprachliche Handlungsmodelle und Muster fungieren, an deren Regeln sich die Schreiber beim Verfassen von Texten halten (sollen)' [that such traditions function as recurrent historically conditioned linguistic models of action and patterns the rules of which writers (should) adhere to when composing texts].

4 Cf., for example, Aschenbrenner (2003: 1).

<sup>3</sup> All translations are my own unless otherwise stated.

From a text-linguistic point of view, these text genres are among other factors determined by certain underlying functions. Following Brinker (2010: 88, 93), texts may have several functions, but one is dominant and thus decisive for the categorization of the texts. Brinker (2010: 98–112), who only refers to non-literary texts, distinguishes the following five functions: *information* (e.g. reports, expert opinions), *appeal* (e.g. instructions, laws), *(self-)obligation* (e.g. contracts, guarantees), *contact* (e.g. congratulatory letters) and *declaration* (e.g. powers of attorney). These text functions are correlated with specific salient linguistic structures (e.g. conventionalized patterns or strategies, formulas, etc.) (cf. Brinker 2010: 91–93).5

## *3.2 Historical framing: The collective term 'logbook' as used for a complex reality and functionality*

In historical sociolinguistics it has already been observed – for example, for semi-private and semi-public contexts – that certain texts may consist of building blocks determined by different discourse traditions. One document may consist of parts belonging to different text genres and therefore may reveal a shift between different text functions. It has been shown that private and intimate thoughts can intertwine with work-related ones, for example, when diaries of simple merchants switch to components belonging to accounts and sales books (cf. Martineau 2013: 136). Diplomatic letters may reveal discursive hybrids composed of narrative, reporting, descriptive and argumentative passages (cf. Okulska 2006: 61).

In the seventeenth century, the pages of typical logbooks followed a table-like format in which several kinds of nautical and climatic information were noted (e.g. name of officer in charge, time, wind, route, sea state, latitude, longitude, etc.). These items of information are often in note form, number-heavy and elliptic. Additionally, there can also be individual observations in more or less complete sentences in a special commentary

5 Brinker (2010: 98–112) gives examples of explicit structures and formulas used to express the individual functions.

section or on individual pages. The contents of many logbooks were later – still on board or after returning to the port – summarized and enriched with further details in a second booklet: These documents summarized the day's events and observations at the end of the day or even later, and they have to be seen as clean and revised versions of the original logbooks (cf. Schotte 2013a: 287, 311f., fn. 28). Nevertheless, in the seventeenth century, the mainly elliptic logbooks and the revised journals as well were often both denominated *journal de bord* or *journal de navigation.* The term was also used to refer to maritime documents that should rather be called reports, declarations or the like. Often, logbooks were inserted into letters to the authorities when they were copied or followed completely different structures, which makes it even more difficult to compare them (cf. Berthiaume 1990: 32, 64; Linzmeier 2022a: 260f.).6 In the following, the reasons leading to discourse traditional hybridity and style shifting will be outlined:

*Job-related reasons for hybridity*: The main use of these documents is to function as reports or instructions addressed to other maritime experts or the authorities (e.g. the documentation of strategic, nautical and geographical knowledge to improve future journeys, and as a training tool for younger sailors), but they may also have been used as a source of evidence and proof, as justification tools, instruments of complaint or recommendation or as commercial registers (cf. Berthiaume 1990: 24–33, 79–88; Linzmeier 2022a: 262f.).

*Pragmatic reasons*: Writers did not necessarily have several booklets available and therefore used single pages or passages for different purposes. In addition, the documents can be seen as a deliberate attempt to condense and summarize the findings of the voyages – which lasted for months – for the readership (the navigation experts, as well as future navigators). Moreover, only a few journals (in revised versions) moved beyond professional circles. Mostly, the documents circulated only within the nautical network, which was familiar with the hybridity of the documents.

*Individual and personal reasons*: The textual structure also varies strongly depending on the specific situations experienced on board (e.g. did

<sup>6</sup> Berthiaume (1990: 32) and Sankey (2010) in particular highlight the hybridity and complexity of these documents.

many things worth mentioning happen and what kind are they: seafaringrelated? crew-related etc.?) and the subjective evaluation of these experiences by the navigator in charge (what do they consider worth reporting or leaving out?) (cf. Schotte 2013b: 111). Even though the logbook design had been increasingly codified and standardized by the authorities and experts in the form of regulations, writing guides and manuals (cf. Schotte 2013b: 112), giving the journal a *personal touch* – for example, by switching to a more narrative or individual style and integrating anecdotal or poetic passages (cf. Sankey 2010: 406f.) – did not seem to contradict the requirement of the authorities, which indeed was to deliver reliable information that was detailed and truthful and that served administrative claims (cf. Berthiaume 1990: 37; Schotte 2013b: 104, 115).

The preceding points show that hybridity and multi-layeredness are already characteristics of the seafarer's profession and personality: On the one hand, each trip is characterized by individual incidents and peculiarities. On the other hand, it depends on the navigator which observations he judges worth mentioning and which purpose he pursues with his text. This multilayeredness is directly mirrored in the documents' content which shows that the possibility for intra-writer variation and style hybridity is already inherent in the communication form itself and in its multi-functionality (cf. also Sankey 2010: 406–13).

## 4 A case study: Corpus compilation and the research question

Methodologically, my wider research project is based on a rich corpus of French *journaux de bord* and *journaux de navigation* from the seventeenth to the eighteenth century, especially from 1660–1760 (see Footnote 1). The documents are taken from the microfilmed *Fonds de la Marine* of the National Archives in Paris (navy and *Compagnie des Indes*)7 and the

<sup>7</sup> Series 4JJ, which includes 477 documents, MAR/4JJ/1–431, *Journaux de bord.*

*Archives départementales de Loire-Atlantique* (merchant marine).8 In this article, I will limit myself to an exemplary qualitative analysis of the *Journal du vaisseau le François, commandant de Voutron, de La Rochelle au grand banc de Bonnaventure, 1716* (Paris, MAR 4JJ/11/6).

Gédéon Nicolas de Voutron (1670–1733) was a French naval officer who, in the late seventeenth and early eighteenth centuries, undertook many significant long-distance voyages, including seven crossings to Canada (cf. Litalien 2010: 9). He came from a Huguenot merchant family in the La Rochelle area that succeeded in joining the petty nobility. After the Edict of Nantes, he converted to Catholicism in order to retain ownership of the family's property in France as well as his titles of nobility (cf. Laux 2010a: 27–37).9

The document can be classified as a journal, which in all likelihood was written after the voyage summarizing the main events and sent to the council: this becomes evident since Voutron emphasizes at the beginning that he is not going to describe his observations of the first month in detail, since *il ne Se paʃʃa rien dextraordinaire* [nothing special happened] (fol. 1r) (cf. also Laux 2010b: 134). The document describes the main points of the journey of the ship *Le François* from La Rochelle (France) to Bonnaventure (Québec, Canada), the stay there and the return to France (cf. also Berthiaume 1990: 37). The journal also serves the ship's commander in particular to highlight difficulties of the voyage and complaints to the council. The document is bound in a hard cover and consists of thirty-five pages (eighteen folios) containing a chronological report of four individual stages of the journey:


In analysing a sample of passages from Voutron's journal, I will focus on the questions of how style-shifting becomes evident within one document, and which underlying functions can be identified.11

## 5 Analysis

#### *5.1 Entries from 23 July to 31 August*

The document begins with the statement that the ship left La Rochelle on 23 July and that the writer (first person) fell ill with a fever for a month. He concludes that because of his illness he will not send the council a detailed journal covering all this time, as nothing exciting happened and the trip went surprisingly well.

In the section following he adds that they arrived on Monday, 24 August, and he provides the coordinates (longitude, latitude, etc.). From this point onwards, the writer switches to an abbreviated and elliptical bullet-point style for about two pages, providing information on wind strength, sea state and coordinates (and other additional information) of individual days (from 24 August to 31 August). The example shows the list-like character of the document section, which enables the writer to document certain observations and events without great effort. The external form and arrangement of the contents are purpose-bound (cf. Waldispühl

11 The transcription is as close to the original as possible, for example, punctuation, majuscule and minuscule use, word and morpheme boundaries, and accents are reproduced as in the manuscript. It should be noted, however, that the manuscript does not always allow for clearly distinguishing between *u* and *v* and the use of majuscule and minuscule. There are different *S*-graphs (*s, S, ʃ*), which are also not always clearly distinguishable, and stressed vowels marked with a dot rather than an accent, for example, *degrėz*.

2019: 203), that is, they serve the writer as well as the readership, that is the nautical professional circle, to quickly find essential nautical-related information.12 The reader is able to extract the information most relevant to him or link individual entries from the list according to his needs and purposes (cf. Koch 1988b: 33, 1990: 143). Since listing makes it possible not only to save but also to distribute information (with the possibility to rearrange or link individual contents), Koch (1990: 144) describes them as 'sehr praktische, fachlich orientierte, aber konzeptionell anspruchslose Sprachprodukte' [very practical, specialist-oriented, but conceptually undemanding language products]. Due to the simple structure of lists, the correct interpretation of the underlying pragmatic context plays an important role, showing that list-like documents depend on the common background knowledge of the readership, for example, the expert group (cf. Koch 1990: 141f.). See (1):

(1) du Mercredy 26 au Jeudy 27 vent de SSE a Ouest par le Nord gros vent groʃʃe mer. Serré Les huniers. bauCoup roulé et tangué : au SSO : 2 degrėz Sud 27 Lieux Latt obʃerveé 44 = 35. Long ariveé 332 = 7. var.on NO 15 vu pluʃ. rs navires; 40 braʃʃes. (fol. 1v) [from Wednesday 26 to Thursday 27 wind from SSE to West from North strong wind heavy sea. Furled the topsails. Travelled and pitched a lot: to the SSE : 2 degrees South 27 leagues Observed latitude 44 = 35. Longitude made 332 = 7. NW variation 15 sighted several ships; 40 fathoms.]

The verbalization strategies used include text-specifics, such as abbreviations, ellipses and formulaic structures, and routines that are typical of 'diaries undertaken for institutional purposes (such as business ledgers or

12 Waldispühl (2019: 203) emphasizes that the purpose of a list may, for example, be to locate and consult information, to catalogue items, to document events, to help as a memory aid, or to plan future events.

military records)' (Elspaß 2012: 158). In the present case, it must be emphasized that abbreviations and ellipses had been established by model books and instructions on how to write a logbook and thus can be considered as conventionalized structures (cf. Linzmeier 2022a: 282). The condensing style is even more evident in the use of infinite structures, such as the past participle (*Serré Les huniers. bauCoup roulé et tangué*), which places the focus on the most essential information, namely the (collectively) performed action, without having to express other grammatical categories (the subject etc.).13 This subsection clearly has an *informative function* and reporting character. The writer does not need to indicate this explicitly (e.g. through verbs such as 'I inform'), since contextual indicators, such as the attribution of the document to the discourse tradition of the navigation report, guide the addressees – who are familiar with these forms and structures – in recognizing the informational function. In Linzmeier (2022a: 285f.), I also argued that the lack of commas (e.g. *vent de SSE a Ouest par le Nord gros vent groʃʃe mer*) should not be taken as a lack of punctuation skills, but rather as a style of brevity accepted among navigational experts. It is likely that punctuation is not mandatory in these highly formulaic building blocks found in logbooks since the readership is familiar with these structures and with recognizing the intended semantic units. In addition, the list-like, that is, the visual, structure, helps to identify semantic units.

Because Voutron was ill, he does not provide more information about the month of August. Already here it becomes clear how strongly the structure of the document depends on the writer's personal condition: the writer limits himself to providing only the mandatory information, namely navigation-specific observations (esp. climate and coordinates), which will make the individual steps of the voyage transparent to the professional community that might need this information in the future.

13 For frequently occurring text-type specifics cf. Linzmeier (2022a: sec. 5.3): these include structuring patterns such as introductory formulas, entries according to day and time and the use of the page margin to store specific content in elliptical form. In addition, there are abbreviations and the use of majuscules in the (typo-) graphic area. Morphosyntactic specifics include, for example, the possible ellipsis of the subject pronoun and the heavy occurrence of participial constructions.

Figure 18.1. Journal du vaisseau le François, commandant de Voutron, de La Rochelle au grand banc de Bonnaventure, 1716 (MAR 4JJ/11/6, excerpt from folio 2r). Archives nationales (France), Service hydrographique de la Marine, Journaux de bord, sous-série Marine 4JJ.

*5.2 Septembre 1616*

To describe the details from the month of September, the writer – now well recovered – switches to a more narrative style with complete sentences delimited by punctuation and finite verbs. Figure 18.1 illustrates the transition from the list (30/31 August) to the narrative style (September). He now avoids ellipses and infinite structures, for example, *Lhoriʃon Estoit gras* [the horizon was overcast] (fol. 2r; see the last line in Figure 18.1) with a finite form of the verb *être*. In contrast to the descriptions from the month of August, the writer now enriches the narration of his action steps with further details. In these entries, which extend over about eight pages, a strong alternation at the level of functions begins.

On the one hand, these entries serve to describe impressions of the terrain. The text passage belongs to the discourse tradition of a report and has an informative function, as (2) shows:

(2) Bonnaventure neʃt quun rocher ʃans eau ou Il ne croist que duʃapinage. […] A La grand terre ou ʃe mettent Ceux qui arivent Les derniers on Est un peu mieux. Il ya quelches graves, de Leau Et dubois. (fol. 3r–3v). [Bonnaventure is just a rock without water where only fir trees grow […] In the *grand terre* where those who arrive last are staying, the situation is a bit better. There is some gravel, water and wood.]

The writer uses verbs in the indicative (*est*, *il y a*) and evaluates his observations (e.g. *on Est un peu mieux*). The informative function may therefore be accompanied by an 'evaluative' attitude (Brinker 2010: 99). At the same time, the writer gives advice for future trips (*appeal function*). (3) provides evidence of this:

(3) Il ne faut Jamais en montant ʃarester aLaCoste duʃud; Outre Les Courans, Les vents de NO y ʃont inutilles […] (fol. 4v) [You must never stop at the South Coast on your way up; besides the currents, the NW winds are useless there […]]

The appeal is made clear by the explicit impersonal structure *Il ne faut Jamais*. This is followed by naming the reasons designed to convince the reader of the correctness of Voutron's statement.

#### *5.3 Sejour de quebec*

From fol. 5v to 7v a detailed description of his time in Quebec follows. Since the ship does not move during this time, the purpose of this passage is not to describe navigation-specific steps, but rather, other actions taken during the stay – especially the unloading and loading of the ship.

#### *5.4 Retour en France*

Starting on fol. 7v, the writer titles the remaining pages *Retour En France* [Return to France]. The descriptions of the chronological, navigationspecific events of individual days begins on 21 October and lasts until 31 October (i.e. the arrival at *isle st. Paul*, fol. 9r). The passage serves to describe the wind and weather conditions and swell, and their aggravating effects on the voyage and the manoeuvre (*informative function*). It is composed largely of short, consecutive main clauses. The predominant tense is the *passé simple* (past definite), which describes completed actions. The writer describes the events as the collective actions of the crew, which is why he uses the first person plural (*nous*). See (4):

(4) Le 27. nous Continuasmes a EnfiLer lariuiere ʃans rien voir : (fol. 8v) [On the 27th we continued down the river without seeing anything:]

Immediately after the description of the navigational events of 31 October, the writer adds a synthesis of the accounts of the difficulties of the whole voyage (fol. 9r–9v). See (5):

(5) Je croy que Cest ici Le Lieux dexaminer les difficultéz deCette Nauiguation, Et les moyens quil y auroit a prendre pour La rendre moins dangereuʃe : Lon ne doit pas Se regler ʃur ʃe que Je viens de dėstailler : De toutes Les Campagnes qui Se ʃont faites, Il ni en a peut-etre pas une de Semblable dans une Saiʃon auʃi auanʃeé

Jay este Sept fois en Canada, Et quoy que ie men Sois bien tiré Joze aʃurer que le plus fauorable deʃes voyages ma donné plus deCheveux blancs que tous Ceux que Jay fait ailleurs.

Dans tous Les endroits ou Lon nauige ordinairemt on ne Sou ffre point Et Lon ne risque pas comme en Canada, Cest un tourment Continuel de corps Et desprit. Ji ay profité de Lavantage deConnoistre que Le plus habille ne doit pas conter Sur Sa Science : (fol. 9r–9v)

[I believe that this is the place to examine the difficulties of this navigation, and the steps that should be taken to make it less dangerous:

One should not settle for what I have just outlined: Of all the campaigns that have been made, there is perhaps none like it in such an advanced season

I have been to Canada seven times, and even though I did well, I dare say that the most favourable of these voyages has given me more white hairs than all those I have made elsewhere.

In all the places where one usually navigates, one does not suffer and does not risk as in Canada. It is a continual torment of body and mind.

I benefited from the advantage of knowing that the most skilful must not rely on his knowledge:]

The shift that is made here is clear: from a sober description of individual navigation steps, the writer moves on to discourse-traditional patterns of a *letter of complaint and recommendation* that is, explicit performative formulas and sentence patterns that perform the *appeal function* (cf. examples in Brinker 2010: 101–107). The phrase *Je croy* [I believe] introduces the passage. This can be considered one of the *verba dicendi*, and it makes it clear that from now on the writer will use the first person pronoun to reveal his opinion, while the preceding passages had served merely to report. The writer walks a tightrope between trying to live up to his role as a navigation expert and being taken seriously, justifying his actions or inactions while not losing face, and at the same time living up to the politeness norms of the time. Consequently, one of the main functions of this passage is to appeal: the writer wants to convince the professional community to take action to make travel to Canada safer. However, he does not begin by suggesting measures – these follow later – but first bolsters his subsequent remarks by emphasizing how arduous travel to Canada in particular is. Since he wants to be seen as a Canada-experienced navigation professional and to be recognized for it, he describes his experiences in the indicative, rendering them as non-doubtable facts (*on souffre, on risque, Cest un tourment*). At the same time, he softens the drastic quality of the complaint, in which, using the conditional, he emphasizes that measures 'should' be taken (*quil y auroit a prendre*).14 This becomes even

14 This shows that the informative function is not simply combined here with an evaluative attitude (cf. Brinker 2010: 99f.). The writer is clearly trying to influence the reader's opinion and behaviour, that is, reaction, which is why the appeal function is dominant in this text passage.

clearer when the writer emphasizes that, 'One should not settle for what I have just outlined', since every journey is somewhat different anyway. Thus, he weakens his direct appeal through these strategies – presumably to remain within the norms of politeness. Furthermore, the passage serves to justify non-successful steps in which the writer emphasizes the haphazard nature of a journey to Canada in general, in saying that 'the most skilful must not rely on his knowledge'. With the help of these structures the writer can emphasize his role as an expert (while avoiding self-praise – he speaks impersonally of 'the most skilful') and at the same time justifies non-successful steps to the council.

The underlying *appeal* is in the following pages reinforced by the criticism formulated of, for example, the available maps (6) and the training of the *pilotes* (7), again expressed by *verba dicendi* in the first person singular (*avancer, dire plus*) and the presentation of the facts in the indicative (*il n'y a aucune, il n'ya point de*):


After further critical remarks, the writer ends these long pages of complaint and recommendation with an apology for his expression of opinion or feeling (8):

(8) Jespere que Lon me Pardonnera ʃe petit Anthouʃiasme (que Loccaʃion fait naistre) […] (fol. 12r) [I hope that I will be forgiven this little enthusiasm (which the occasion gives rise to) […]]

At this point, the *contact function* clearly becomes visible since the writer addresses the readership (the council) directly using verbal structures such as *pardonner*.

From here on the description of the difficulties of the journey ends and the writer changes back to the elliptical logbook style, the naming of coordinates etc., with single-day specific descriptions as shown in (1) (*information*) (fol. 12r–14v). These entries end on 29 November with a final detailed and narrative account of the end of the journey (fol. 15r–16v).

The hybridity of this document becomes clear once again in the final part, which extends over two and a half pages (fol. 17r–18r). While in the preceding passages the reporting character clearly predominates, the conclusion serves to establish *contact* with the council and as an *appeal*. This becomes clear because the writer addresses the *Conseil* four times here. At this point, the document thus takes on the character of a letter to the authorities (9):

(9) Je ne dis rien auConʃeil ʃur mon ʃujet. Il ya aʃėz Lhontemps que Je Lentretiens de mes desmarches. Sil Les aprouue Cest La mon pris […]; Je nay pu faire mieux ni auec plus de dilligence. JeCroy mesme pouvoir auanʃer que lon ne vera point de Journal deCanada Si court que Celui cy : Je nay eu auCun Esgard a mon interest particulier qui mauroit pu porter ʃur deʃ pretexes aparent aproLonger mon voyage; il ne men auroit pas cousté dauantage et il men Seroit revenu plus dargent; Mais je nay pas Esté EsLeué dans ʃes principes : Je Leferéz toujours Connoistre quand Le Conʃeil voudra meContinuer La grace de Memployer estant prest daCheuer Le reste de ma vie auʃeruice Sans demander autre choʃe que de nestre pas Exclus comme ie Lay esté Jusques aCet heure Des honneurs a quoy bûte un gentilhomme et Sans lesquels apres un Certain temps il ne peut plus auoir de plaiʃir dans Le monde, ʃurtout quand il Se void commandé par une troupe dofficiers qui Certainement ne deuoient pas estre auanʃéz a Son prejudice : N. Voutron (fol. 17v–18r)

[I don't tell the council anything about my subject. I have been informing them long enough of my proceedings. If they approve them, that's my reward. […]; I could not have done it better or with more diligence. I think I can even say that there will be no Journal about Canada so short as this one: I had no intention of taking any particular interest which might have led me to pretexts to extend my journey; it would not have cost me more and I would have made more money; But I was not raised that way: I will always recognize it when the council wants to pursue the grace of employing me and being ready to complete the rest of my life in the service without asking for anything other than not to be excluded as I have been up to now from the honours to which a gentleman is entitled and without which after a certain time he can no longer have pleasure in the world, especially when he sees himself commanded by a troop of officers who certainly should not be promoted to his disadvantage: N. Voutron]

The balancing act between the necessary expression of respect and politeness towards the council and the simultaneous attempt to be recognized for his own efforts and to influence the council in making decisions becomes completely clear here. The section has a contact-intensifying function because the writer is concerned with the personal relationship with the council (cf. Brinker 2010: 110f.). This shows explicitly in verbal structures (*Je Leferéz toujours Connoistre*), but can also be deduced from the context as the writer constantly uses the first-person singular to remind the council of his accomplishments and his honourable attitude (*Mais je nay pas Esté EsLeué dans ʃes principes*). Although the relationship between the authorities and Voutron is asymmetrical, and he therefore needs to follow norms of politeness (e.g. the use of the conditional *Je Leferéz*, lexical choices: *meContinuer La grace*), the author is nevertheless aware that he is an important part of a network of professionals and is entitled to make demands. Therefore, the *obligation function*, which also comes into play here (as the writer offers to continue to commit himself to naval service: *estant prest daCheuer Le reste de ma vie auʃeruice*), is only an additional function, being clearly subordinated to the *contact* and the *appeal function*. The appeal is manifest in the writer's attempt to influence the council in its opinion and behaviour (cf. Brinker 2010: 101). This happens through explicit structures (*demander*) as well as through voicing criticism and distinguishing oneself from less capable employees. At this point, the journal ends abruptly with Voutron's signature.

The directness with which Voutron confronts the council is somewhat risky because of the hierarchical asymmetry, but it is not entirely surprising: this is a very hybrid text passage, which may be attributed simultaneously to the discourse traditions of *complaint*, *recommendation* and *petition*, and it can be read as a kind of letter to the authorities. Nevertheless, it does not contain any epistolary formulas typical of the time (e.g. a closing formula) because the document may also be viewed as a sort of military report. The 'tendency to condense letter endings' (Okulska 2006: 70) indicates once again that contact maintenance and appeal are not the only functions of this passage, but that the writer's intention is rather – as part of the network – to report to the navigational professionals on the journey, 'sharing knowledge' (Gotti 2006: 27), circulating insights and making recommendations for future undertakings.

## 6 Conclusion

On the basis of an exemplary qualitative analysis of a navigation journal, this chapter has focused on a very specific group of maritime writers and their writing culture, which has so far received little attention in linguistic research.

Logbooks and navigation journals can be classified as linguistically, textually and discourse-traditionally complex documents. Although the texts have a primarily informative function and therefore have to follow linguistic, structural and content-related rules to a certain degree, they are all different in form and vary depending on the individual underlying functions (e.g. information, appeal, etc.), pragmatic requirements (e.g. politeness) and personal reasons (such as personal interest and wellbeing). However, in the case discussed here, no main function can be identified. Even if an essential function of Voutron's journal is to inform the council through reporting, especially towards the end of the journal the text takes on the tone of a complaint, recommendation and petition. The positioning of these passages towards the end of the text and the multiple direct addressing of the council also show the importance that the writer attaches to these text passages and their functions. The fact that the journal was woven into a letter and had a direct addressee (the council) – instead of just an unknown later readership of nautical professionals – allowed for intra-writer variation, since it gave Voutron the chance to combine his obligation to document and transmit nautical details with other personal concerns (such as complaint, request and advice). Consequently, the document is puzzled together from subtexts of different discourse traditions and is therefore clearly characterized by polyfunctionality.

This exemplary study underlines the already accepted approach in historical sociolinguistics that a look at socio-pragmatic parameters (such as identities, social roles politeness, attitudes) and socio-historical factors (such as demographic and economic parameters, family, gender and cultural history) may enrich historical sociolinguistics research (cf. Nevalainen & Raumolin-Brunberg 2012: 27). This wider approach allows a focus on the impact of parameters such as the social and professional context, the network structure and the individual conditions, choices and attitudes of the writers and to describe specific (i.e. individual and group-related) writing contexts that may be mirrored in the form of structural and stylistic hybridity and flexibility.

## References

#### *Primary sources*

Archives Nationales (Paris), Service hydrographique de la Marine. Journaux de bord, sous-série Marine 4JJ: Journal du vaisseau le François, commandant de Voutron, de La Rochelle au grand banc de Bonnaventure, 1716 (MAR 4JJ/11/ 6; 18 folios).

#### *Secondary sources*


*Sprachgeschichte und Sprachkontakt – Münchner Beiträge zur Sprachwissenschaft*. Frankfurt: Lang. 245–303.


& Anita Auer (eds), *Exploring Future Paths for Historical Sociolinguistics*. Amsterdam: Benjamins. 1–19.


#### Oxel Uribe-Etxebarria

# 19 The linguistic choices of an early nineteenthcentury Basque writer

#### Abstract

This study analyses spelling choices in the work of Manuel Umerez (1757–1818), a Basque author of the early nineteenth century. The selected corpus is formed by two manuscripts by the same author, which makes the observation of variation across the lifespan of the author possible. For that purpose, I focus on three spelling variables and present evidence of the use of each variant in the two documents. The data are placed in a wider context within the Basque written tradition and the textual genre, that is, religious didactic literature. The results show how the frequency of the most innovative variants increased and their usage became more systematic before a standard Basque orthography was later widely used.

## 1 Introduction

The Catholic Church in the Basque Country used the local language as an instrument to spread Counter-Reformation ideas after the decisions of the Council of Trent (1545–63). As a result, an unprecedented Basquelanguage religious and devotional literature emerged first in the province of Labourd (on the French side of the Basque Country) during the seventeenth century (Zuloaga & Krajewska 2021). However, in the eighteenth century, the output of the Labourdin literature lessened and the southern or Spanish region of the Basque Country became the focal point of Basque literary production. As in the previous century, books printed in the Basque language were still mainly religious, while other literary, scientific or legal texts were written either in Latin or Romance languages.

The need for 'making the best use of bad data' (Labov 1994: 11) is therefore patent when analysing the history of the Basque language, partly due to the lack of a large historical corpus for the language, and also because the Basque written tradition is mainly composed of religious educational literature almost until the twentieth century (Lakarra 1997: 454f.). Consequently, the available written evidence is biased towards highly educated writers (i.e. clergymen) and formal registers.1 Furthermore, Basque writers, in general local clergymen, were bilingual in Basque and Spanish and literate in the dominant Spanish language within a diglossic background.

Those Basque authors who generally addressed local devotees, wrote either in their own local variety or used supra-local norms that gradually superseded the vernaculars and are known as the 'literary dialects' of Basque. Thus, the written tradition has not been continuous and uniform for all Basque-speaking regions. Although there had been prior proposals, a new unified standard Basque was not established until the late 1960s by *Euskaltzaindia* [The Academy of the Basque Language] (Hualde & Zuazo 2007). Previously, the spelling systems used by Basque writers were also based on the orthographic norms of the neighbouring Romance languages, namely French or Spanish. As a result, we find a high degree of inter- and intra-writer orthographic variation in Basque texts before the twentieth century.

The present study deals with intra-writer variation in the work of a Basque writer of the early nineteenth century. Manuel Umerez (1757–1818) was a clergyman who wrote two manuscript books concerned with religious and moral issues from 1805 onwards. The aim of this chapter is particularly to show spelling shifts in the two manuscripts and the changes made from one to the other to study orthographic variation across the lifespan of the author. I also investigate possible motivations of the spelling variation, such as the influence of prestigious patterns used by other educated authors (a 'change from above'), or the wish to approach the addressee using a closer language. For that purpose, first I present a short description of the selected corpus and what is known about the writer and focus

1 Fortunately, new documents of other genres and types, such as private letters, administrative and legal documents, continue to emerge, and recently, some studies approach these 'non-canonical' texts from the perspective of historical sociolinguistics (Krajewska & Zuloaga 2021; Padilla-Moyano 2015).

on the sociolinguistic contexts: I place Umerez's texts within the Basque written tradition of the time and consider the question of literacy, which determines who the audience of such writings could be.

## 2 Corpus

Manuel Umerez was a local priest born to a peasant family in Oñati, a town located in the province of Guipuscoa (Elortza 1996). After completing his studies he was a beneficed priest of the parish church of San Miguel in his hometown Oñati for some decades, where he occupied the position of parish secretary among others. He died there in 1818.

Umerez's manuscripts are two extensive books by this author, containing approximately 206,000 words. The first book was written in 1805 and the second somewhat later, although the exact date is still unknown. Scholars believe it was written later, because it seems a reformulation of the first manuscript, which includes many additions and corrections, for example, he uses a more purist vocabulary, replacing certain direct loans with equivalent native words (Altzibar 1992: 76).

Basque philologists have known about the existence of these documents and have recognized their linguistic importance for some time (Michelena 1958; Elortza 1977). They paid interest to the special dialectal variety that the author used, as I describe in Section 3.1, but the texts have not, as yet, been published or scholarly edited. Both manuscripts have been preserved together in a private household: Arrazola House, in the neighbourhood of Olabarrieta, Oñati (province of Guipuscoa). Recently the owner, who inherited the manuscripts, gave permission to the town council of Oñati to digitize and make them available on the internet. The Provincial Historical Archive of Guipuscoa was in charge of the digitization process and the facsimiles were published on the town council's website.2

2 The digital images of the two manuscripts are currently available via the following links: <https://www.xn--oati-gqa.eus/eu/udala/argitalpenak/osaba-instruccinuac\_1-alea.pdf>

In the two books that Umerez wrote, a priest gives religious advice and instructions to a young, newly married peasant. Thus, the content of both is similar, but the second is much more extensive. The books carry the following titles:


 [Instructions from an uncle to his married and peasant nephew for his and his family's good government, according to the holy law of God]4


In manuscript O, the addressee is the author's own nephew and in manuscript E, an assumed friend of the author, who is also depicted as a humble peasant. The total number of pages and words in the corpus is shown in Table 19.1.


Table 19.1. Total number of pages and words in each manuscript

and <https://www.xn--oati-gqa.eus/eu/udala/argitalpenak/errectore-instruccinuac2 alea.pdf> accessed 14 June 2022.

<sup>3</sup> I will use the labels O and E for the manuscripts as in Elortza (1977, 1996).

<sup>4</sup> All translations are my own unless otherwise noted.

## 3 Sociolinguistic context

#### *3.1 Basque writings of the late-eighteenth and early-nineteenth century*

The writings of Umerez should be viewed in a wider context. In the first half of the eighteenth century, a whole language codification project for Basque was undertaken by the Jesuit priest Manuel Larramendi (1690–1766) who published a work in defence of the Basque language in 1728, the first ever printed grammar of Basque in 1729 and finally his most important work, the trilingual Spanish-Basque-Latin dictionary, in 1745. The latter's date of publication is considered as the beginning of a new period in the history of written Basque, the so-called *First Modern Basque* (Lakarra 1997: 516f.; Urgell 2018).5

In the Southern Basque Country, the influence of Larramendi's pioneering work, indeed, caused a certain awareness among other clergymen in favour of using Basque for pastoral work. As a result, the number of publications (mainly Catholic literature) increased during the late eighteenth and early nineteenth century. The expulsion of the Jesuits from Spain (1767), the threat of the French Revolution (1789) and later the War of the Convention (1793–95) as well as the Spanish War of Independence (1808–14) also seem to have produced a reaction within the Basque Church. However, some of those religious writings remained unpublished due to the turbulence of the times and the prohibitions on publishing in Basque (Urgell 2018).

Larramendi did not attempt to homogenize dialectal variation by the development of a unified standard language, but rather he inspired the use of the different Basque dialects in writing. He shaped the 'literary Guipuscoan' dialect that he promoted for writing, which gradually became the most prestigious written variety and functioned as a supra-dialectal koiné for various purposes in the Southern Basque Country. In the neighbouring province, the 'literary Biscayan' variety was developed later, in the early nineteenth century by other writers. Although traditional language historiography is grounded mostly on printed materials, the unpublished

<sup>5</sup> It is also named Early Modern Basque in Ulibarri (2013).

work studied in the present chapter should also be connected to that literary tradition. The peculiarity (and so far, the main interest) of Umerez's work is that it is written in a previously unattested local variety, the one spoken in his home town of Oñati, an unusual practice for that time: although he was not the only clergyman of his town who wrote in Basque, the majority used the most prestigious Guipuscoan dialect, for example, the sermons written by Miguel Plaza (1771–1854) (Elortza 1996: 19–23; Pagola 1992: 605–794). Testimonies of that time also give evidence that the members of the community were used to hearing the preachers of the religious orders using the Guipuscoan dialect in the pulpit and even demanded its use (Zuazo 2019: 106f.).

In the second half of the nineteenth century the interest in regional languages and dialects grew and Louis-Lucien Bonaparte (1863) made the first classification of Basque dialects. He discerned eight major dialects each of them subsuming some more local varieties of Basque, which he first attested thanks to the translations that he entrusted to several collaborators all over the country. Bonaparte's map in Figure 19.1 shows the following dialects, from west to east: Biscayan (red), Guipuscoan (light blue), Northern Upper Navarrese (yellow), Southern Upper Navarrese (green), Labourdin (orange), Western Lower Navarrese (dark blue), Eastern Lower Navarrese (brown) and Souletin (grey).

#### *3.2 Literacy*

Illiteracy was still high in the Basque Country at the beginning of the nineteenth century (De Benito 1994). There was no compulsory elementary schooling and, in most cases, the only instruction children received were oral teachings on Christian doctrine, which was also the only content they received in Basque. The few men who were educated received their education in languages other than Basque: that is, Spanish grammar was studied in secondary school and Latin was necessary for higher education. It was common for a small portion of the country's elite to study abroad. The fact that some clergymen preferred to preach in Spanish rather than in their mother tongue (Madariaga 2014: 226–37), although

Figure 19.1. Map of Basque dialects, sub-dialects and varieties in the nineteenth century (Bonaparte 1863).

the majority of the population were monolingual Basque speakers, reflects a diglossic situation in the Guipuscoan society of the time.

Umerez's manuscripts include metalinguistic evidence and interesting information about child education in his town. Passages (1) and (2) point out that primary school education and learning to read was considered important for both boys and girls, but no further schooling was needed. The main objective was for them to be able to read the catechism and, as (3) shows, it must have been common to own a catechism and some other devotional books in Basque. The last passage (4) reveals the difficulties that Basque monolingual children had to face when taught in Spanish. It seems that the local boys' school had only Spanish speaking teachers. The type of education that Umerez defends is a practical and religious training that would give children a minimal preparation for their future life as farmers, herders and homemakers.


[from that moment on you will give each child their catechism booklet in Basque, and you will make them learn well all the doctrine […] You must also make them practice reading, so that it does not get rusty and forgotten, and they know how to read with sense some devout books in Basque]

(4) Umiac batez ere Muticuac icasi oi daube Erderaz Escolan beren Doctrina santuba; eta Euscaldunac dirian leguez, ez daube comprenditan ezercho; icasten dauben, pusquia daquie Loruen modura; eta onela guertaquetan da onerian Umiacaz, amabi edo amalau urte cumplitia, salvaitaco premina dan gara, Doctrinaric ez daquiela (ms. O, fol. 166r–v)

[Children, especially boys, learn the Holy Doctrine in Spanish at school; and since they are Basque speakers, they understand nothing at all; the little they learn they know it like parrots, and so it happens that these children turn twelve or fourteen without knowing enough doctrine]

## 4 Linguistic variables and results

In order to focus on intra-writer variation and the writer's stylistic choices, a selection of spelling variables are analysed in the corpus. Since they occur frequently in the texts, they are quantifiable and may be related to sociolinguistic factors. The possibilities of orthographic variation for historical sociolinguistic studies have been evidenced in previous research (Rutkowska & Rössler 2012) and have recently been applied to the observation of individual writers (intra-writer variation), such as in late Middle English (Hernández-Campoy & García-Vidal 2018).

Basque spelling, from the earliest texts, was based on French or Spanish orthographic norms and this caused various difficulties within both the Northern and Southern Basque Country. To cite an example, in the Southern Basque Country, the graphic distinction between fricative and affricate sibilants was spread only during the First Modern Basque period through the influence of Larramendi, so that many earlier texts lacked a specific spelling for affricates. As Spanish has a smaller inventory of sibilants (see the Basque sibilants in Section 4.2), several Basque writers used, for example, the grapheme <s> to represent both /s̺/ and /ts̺/. Variability is high, because even those writers who tried to differentiate the two sounds graphically, did not agree on the same solution. In the last decades, studies have shown spelling variation in the work of certain representative Basque authors of that period (see Altzibar 1992; Urgell 1987).

The current study considers three orthographic variables: (1) the use of <b> or <v> corresponding to the voiced bilabial stop /b/ (variable B); (2) the choice between <ce>, <ci> of the Romance spelling tradition vs the innovative variants <ze>, <zi> for the syllables /s̻e/, /s̻i/ (variable Z); and (3) the adoption of the new grapheme <s̃>6 for the representation of /ʃ/ (variable X).

#### *4.1 Variable B*

In the Basque consonantal inventory, as well as in modern Spanish, there is no voiced labiodental fricative /v/, but in Spanish orthography, two different graphemes (<b> and <v>) represent the voiced bilabial stop /b/. The Spanish standard maintains an etymological criterion for the distinction between those two spelling variants (rae, asale 2010: 92), so that

6 There was no consistent use before, as I explain in Section 4.2.2.


Table 19.2. Frequency of <b> in Umerez's manuscripts

in most words the spelling of the language they come from is kept: *acerbo* [acerbic] (from Latin *acerbus*) and *acervo* [heritage] (from Latin *acervus*) are homophones in Spanish. Although the Spanish Royal Academy (RAE) implemented the norm in the eighteenth century, confusion in its use still existed during the nineteenth century, when the orthographic reform was completed (Martínez 2010; rae, asale 2010: 32), regardless of the cases of anti-etymological <b> or <v> that are normative due to the tradition in the use of spelling, as *móvil* [mobile] (from Latin *mobilĭs*) or *abuelo* [grandfather] (from Medieval Latin *avolus*) (rae, asale 2010: 93). In Umerez's two texts the variant <b> is much more frequently used than <v> (see Table 19.2).

The usage is not conditioned by position (initial or interior of the word) since <b> is more frequent in all positions, but it is necessary to distinguish loan words from native words. The unmarked choice is <b>, which can appear in both borrowed and native words, while <v> is mostly used in loans that are spelled with a <v> in the donor language (e.g. *verde* [green] from Spanish *verde*]). The exceptions in the corpus are few in proportion, that is, few loans that in Spanish and/or Latin were spelt with a <v> appear written with a <b> in the corpus, as in the words *adoptibo* [adopted] (ms. E, fol. 147r), *baliente* [brave] (ms. O, fol. 177v), *benganza* [revenge] (ms. O, fol. 182v), *bigore* [vigour] (ms. O, fol. 129r), *fabore* [favour] (ms. O, fol. 87v): 59/2,540 tokens (2.3 %) for manuscript O and 26/3,902 tokens (0.7 %) for manuscript E.

Even more interesting is the case of the native words, which are mostly spelled with a <b>, but variation occurs in seven lemmas as Table 19.3 shows. Analysing those lemmas that show variation, the use of the variant <v> drops a little from the first to the second manuscript, but that is visible especially in the word *bitarte* [between, through]: the frequency of that word written with a <b> is 54 out of 125 (43.2 %) in manuscript O, while it is written systematically with a <b> in E: 318/319 (99.7 %). It is likely that the author based this on an erroneous etymology that related the Basque word *bide* [way] with Latin *via*, and *bitarte* [between, through] as derived from *bide*; although he might have rectified his error in manuscript E. In manuscript O there is free variation between *bitarte* and *vitarte*: the first instance of *bitarte* is on page 11. However, in the first part of the book (107 pages) the author writes *vitarte* 33 times (out of 41 = 80.5 %), while in the last 107 pages there is only one occurrence of *vitarte* (out of 23 = 4.3 %).

The only counterevidence of that tendency is the developments in the variants *bider/vider* [times], but the similarity of this word with Latin *vicis*, Spanish *vez* [time] could also explain the hesitation of the writer when spelling it.


Table 19.3. Native words that show <b>/<v> variation and number of tokens in each manuscript

The general trend suggests that the writer refined his orthographic criteria with regard to the variable B and used an increasingly uniform spelling. The variation was reduced and he kept the use of <v> almost exclusively as an etymological spelling, for which his knowledge of etymology also improved through time.

## *4.2 Spelling of the sibilants*

Basque has three voiceless sibilant phonemes, a lamino-alveolar /s̻/, an apico-alveolar /s̺/ and a prepalatal /ʃ/, which in modern orthography are respectively represented with the letters <z>, <s> and <x> (Hualde 2003: 16). Each fricative sibilant has an affricate counterpart: /ts̻/, /ts̺/ and /tʃ/. The spelling of sibilants has historically been one of the greatest challenges for Basque authors, when adapting the graphic system of the Romance languages to the phonemic system of Basque (Trask 1997: 77; Ulibarri 2013: 91).

In the present study I observe the representation of the lamino-alveolar /s̻/ and the prepalatal /ʃ/, which are the two fricatives that present a spelling variation in the corpus (see Table 19.4).


Table 19.4. Spelling of sibilants

#### 4.2.1 Variable Z

The lamino-alveolar fricative /s̻/ in the corpus, specifically the sequences <ze> and *<*zi> may appear spelled with a <c> as in the Latin tradition or with a <z>, matching the representation of the same sound in other positions.


Table 19.5. Frequencies for variable Z

Table 19.5 shows that <z> is much more frequent in manuscript E (97.8 %) than in O (55.2 %), but again the distinction must be made between loans and native words. We might assume that <c> is retained in borrowings and <z> appears more in native words, but there are exceptions of both in both manuscripts: that is, words of Latin origin written with a <z> and words of native origin spelled with a <c>. However, there is a strong difference between the two manuscripts: in manuscript O <z> is used only in 15 % of the loans, while in manuscript E the figure is 93 % (see Figure 19.2). Therefore, there is a definite preference for this spelling in the latter even in loanwords.

It seems that the writer tended to generalize the use of <z>, on the one hand, to all positions in the word, avoiding an irregularity that exists in the Romance orthographic tradition (<c> before *e* and *i* vs <z> before *a*, *o*, *u*, any consonant and at the end of a word) and, on the other hand, to the whole Basque lexicon (both native and loan words).

However, the generalization to the loans is not complete, as 6.3 % (178/ 2,814) of them are still spelled with <c>, while only 0.1 % (8/5,580) of the native words are. No pattern is seen in the loan words that keep the spelling

Figure 19.2. Comparison of the percentages of use of the variants <ce>, <ci> and <ze>, <zi> in native words (*left*) and loanwords (*right*).

<c> in manuscript E and it does not seem that the writer had a preference in specific words. Several words show <c>/<z> variation throughout the manuscript, as, for example, *cancinuan* (fol. 1v) / *canzino* (fol. 73v) [song], *ceremonia* (fol. 125v) / *zeremonia* (fol. 20v) [ceremony], *confirmacino* (fol. 17v) / *confirmazinoco* (fol. 123v) [confirmation], *discipulu* (fol. 10r) / *diszipulu* (fol. 23r) [disciple], *errecivitaco* (fol. 126v) / *errezividu* (fol. 122v) [receive], *Franciaco* (fol. 48v) / *Franziaco* (fol. 41r) [of France], *juicio* (fol. 7v) / *juizio* (fol. 13r) [judgment], *principe* (fol. 42v) / *prinzipe* (fol. 21v) [prince] etc., and even the same word can contain both spellings, as *concienziagatic* (fol. 41r) [conscience] or *circunstanziac* (fol. 163v) [circumstances].

#### 4.2.2 Variable X

The inter-writer variation is higher for the voiceless prepalatal fricative /ʃ/ than for the other sibilants. This phoneme does not exist in modern Spanish and the grapheme <x>, which in Medieval Spanish represented the sound /ʃ/, was relegated almost exclusively to the pronunciation /ks/ after several orthographic reforms in the eighteenth and early nineteenth centuries (García 2011; rae, asale 2010: 32). As a consequence, the rep resentation of /ʃ/ caused more variation among authors in the south of the country (northern authors had the French spelling <ch>) and there was no unanimity. There were various proposals, such as <x>, <sh>, <ix>, <sy>, <(i)s>, (i)ss, <ʃ>, <ʃs>, though none of them succeeded definitively. Several authors used letters with diacritics <ś>, <ŝ> or <s̃> in their handwritten texts, which sometimes were later replaced probably due to a printer's decision (Urgell 2018: 608f.).

Furthermore, the use of the sound /ʃ/ in Basque is limited: in some cases, it is the result of a palatalized sibilant, when preceded by the vowel *i* or the glide *j*, and in other cases it is used with expressive or affective value (Trask 1997: 146). Because palatalization is considered secondary, it is not always consistently indicated in old texts (Michelena 1961: 184). In this case, the same grapheme could represent the prepalatal sibilant and its alveolar equivalent.

Umerez uses the special character <s̃>, a common <s> with a tilde on top (see Figure 19.3). In manuscript O it is only used in two words, *gais̃o* [ill] (pronounced [gajʃo] or [gaʃo]) and *pisũ* [weight] (pronounced [piʃu]),

Figure 19.3. Example of the grapheme <s̃> in the word *gais̃o* (manuscript O, fol. 178r).


Table 19.6. Frequencies for variable X

and occurs in twenty-five tokens (it is still not systematically used in all occurrences of these two words; see Table 19.6). In manuscript E it is used in eleven words, but at the same time, many of those words also appear in the text written with a common <s>: *ais̃uari*/*aus̃uari*/*as̃uari* [groceries], *gaisõ* [ill]*, guizaisõ* [poor], *isill ̃* [quiet], *laister ̃* /*laster ̃* [soon], *maisũ* [teacher], *marmaseta ̃* [murmur], *pisũ* [weight], *pusica ̃* [bladder], *pusquiñ ̃* [bit], usatu ̃ [to shoo] (11 words, 171 tokens).

Again, if we compare the percentage of use of <s̃>, it increases from 23.8 % in manuscript O to 68.7 % in manuscript E.

Some of those eleven words listed above are not used in manuscript O, but *isil*, *la(i)ster*, *maisu*, *pusica* and *pusquin* that do occur in O do not present variants with <s̃>. The comparison concerning variable X shows that the writer became more consistent in representing the distinctive sound and we must assume that he made a conscious attempt to spell more accurately, although that practice was not yet completely systematic.

## 5 Concluding remarks

In a discussion of the sociolinguistic situation surrounding Manuel Umerez, we identified that people of the province of Guipuscoa in that period were mostly illiterate and monolingual Basque speakers, a reality that affected the character and purpose of Basque texts written at that time and the potential readership of those works: the lack of readers with sufficient competence and who were used to reading in Basque, together with other factors, meant that only texts of religious instruction were produced, such as small catechisms translated for basic education or ascetic works, devotional books and sermons addressed to educated clerics who would be in charge of transmitting the teachings orally to the people.

The microscopic perspective used in the study of three orthographic variables may contribute to other macroscopic studies that compare spellings used by different authors throughout history and to a global understanding of how modern Basque spelling developed. Moreover, manuscripts and unpublished books offer extra information that printed books lack, because we eliminate the possible filter of printers on the spellings used, and we observe the orthographic shifts and innovation processes of the writer by comparing, for example, two manuscripts of the same author as has been carried out here.

The language of the texts studied in the present chapter is usually considered close to the vernacular variety of the author, but in the same way as in the philological tradition, knowing the *usus scribendi* of the author is an indispensable tool for the editor of a text. A sociohistorical approach, therefore, has been of benefit in this instance in the evaluation of the internal or external factors that determined the writer's different variants. The results showed that the selected variables are conditioned by the author's deliberate choices. The fact that the writer used a very dialectal language, in several phonological, morphological and lexical forms, which were only used in his local speech (Elortza 1977; Uribe-Etxebarria 2020), does not mean that his writings are less carefully constructed or closer to oral communication. Instead, his spelling changes involve an intention to achieve a more effective orthographic system. We observe how Umerez was adopting the innovative variants <b>, <z> and <s̃> and their usage increased and became more systematic from the first to the second manuscript. These innovations either avoid unnecessary irregularities or provide a more precise mode of representing a sound.

As regards <b> vs <v> for the voiced bilabial plosive /b/ (variable B), the general tendency among the best-known authors of the First Modern Basque period was to favour a phonetic spelling by a less frequent use of the grapheme <v>. Some prescriptivist works also recommended permanently removing the letter <v> from Basque orthography, but in general, most authors, similarly to Umerez, respected the etymological spelling and used <v> with loanwords (Urgell 2018: 608).

In the use of the spelling <ze> and <zi>, the orthographic shift became apparent when comparing the two manuscripts of the corpus. Although the variants <ce> and <ci> were not completely replaced, the change in the spelling of borrowed words occurred in a clear direction, meaning that Umerez gradually abandoned the etymological spelling in favour of a more regular and uniform spelling. This choice was also extended and finally <z> became the standard spelling for the /s̻/ sound in modern Basque orthography (Trask 1997: 75f.).

Not all Basque authors found an adaptable and common spelling for the prepalatal sibilant /ʃ/ but gave differing solutions. Umerez used the grapheme <s̃>, but contrary to the other variables, in this case, the variation occurs between the use and non-use of the special diacritic mark to mark that phonetic feature. Therefore, as the frequency of <s̃> notably increases in the second manuscript, we can understand that the writer was perfecting his spelling to adjust it to the phonemic system of Basque.

Lastly, noteworthy is a metalinguistic commentary that can be read in manuscript E, where the author wrote the following Latin text (the Angelus) using a Basque-like spelling: *Angelus Domini nunziabit Marie ed conzepit de Espiritu Santo* (fol. 69r) [The Angel of the Lord declared unto Mary, and she conceived of the Holy Spirit]. After transcribing the verse, he explained: 'I know, my friend, that I have misspelled the Latin words before the three Hail Marys; but I have done so, knowing that you will get it right this way and read them as if you knew Latin'. Interestingly, he adapted the Latin words using the letters <b> and <z> in *nunziabit* (for *nuntiavit*) and in *conzepit* (for *concepit*)*.* Despite being a relatively modern spelling at that time, the author associated it with the Basque language, considering that a Basque monolingual speaker would read it more easily in that manner. I have focused on these two spelling variants in the text, though other words also reflect the Hispanic pronunciation of Ecclesiastical Latin, such as *Marie* (for *Mariae*) or *Espiritu Santo* (for *Spiritu Sancto*).

The relationship of the writer and the addressee is close in Umerez's work: he addressed his nephew directly in the first manuscript and a close friend in the second. As the recipients of the two manuscripts seem to be different, we may think that the variation could be motivated by the different relationship between the writer and the addressees. However, the addressee of manuscript E is unknown: while in the text he is depicted as a 'friend' of the author, there is a possibility that this was simply a literary convention. Besides, if manuscript E has a more 'Basque-like' spelling and represents more often an expressive pronunciation with the prepalatal consonant, one could assume a closer relationship. There is no evidence in the text to support such a theory; indeed, the author refers to himself as 'priest' in the manuscript E and as 'uncle' in O. Finally, due to the diglossic situation described in Section 3.2 and the underdeveloped written tradition of Basque, it would be difficult to imagine a spelling accommodation of these characteristics.

These types of religious texts written in Basque might be aimed at readers of low status and little education, rather than the more educated people who were able to read an extensive variety of books in Latin or Romance languages. In these two books, moreover, the recipient of the instructions is clearly defined as a peasant already in the title. Because of the dialectal forms used throughout the text, apart from the orthography, we may believe that the language is relatively close to speech and vernacular style. However, the study on graphic variants showed that prestigious patterns used by other educated authors influenced the writer's spelling choices as a 'change from above'.

It will be interesting to consider this conclusion when evaluating other linguistic data in the corpus. The author may have used the local variety deliberately, because he was addressing a person from the same village and of a low social rank, but it seems that he was very careful in his choices, tried to make systematic use of the variants and gradually reduce variation. I focused on spelling in the present study, but phonological, grammatical and lexical variation is also substantial in the texts, thus, this approach would likewise be useful for examining the other linguistic features in the future.

## References


#### Christine Wallis

# 20 Linguistic repertoires and intra-writer variation in Old English: Hemming of Worcester

#### Abstract

This chapter explores intra-writer variation in the works of Hemming of Worcester, an eleventh-century monk whose hand has been identified in seven surviving manuscripts. A corpus compares selections from Hemming's written output alongside parallel selections from other textual witnesses. The resulting scribal profile builds on data in Wallis (2013a) to analyse his variation in four features which deviate from 'standard' or 'focused' Late-West-Saxon: <wæ> spellings, o+nasal, retraction of *-ward*, and unstable <h> (hdeletion and h-insertion). Hemming's variation is shown to be due to two main factors, *constrained selection* (influenced by exemplar forms) in 'local' texts, and his own preferred usage, based on ongoing sound changes in late-Old English.

## 1 Introduction

This chapter focuses on variation in the writings of Hemming of Worcester (*fl. c.*1095), whose hand has been identified in a number of late-eleventh-century manuscripts, writing both Latin and Old English (OE) (Ker 1985 [1948]; Tinti 2002).1 An experienced Anglo-Saxon scribe, Hemming is unusual in that his name, place of writing and certain

<sup>1</sup> That the same hand is responsible for the writings discussed in this chapter is not in dispute, though the identification of this hand with Hemming of Worcester is less clear; Ker prevaricates over whether Hemming is to be identified as his hand 1 or hand 2 (the hand under consideration here), before coming down in favour of hand 2: '[t]he arrangement of the five sections of Tib. II can only be due to Hemming if he is identical with the scribe of ff.119–25 […] That this scribe is Hemming seems to me probable.' (1985 [1948]: 56).

other biographical details are known to us. He was a monk at Worcester under Bishop Wulfstan II (1062–95), and is described by William of Malmesbury as a sub-prior, while his name appears alongside those of other Worcester monks in the Durham *Liber Vitæ*, in an entry dating to the time of Wulfstan II's successor, Bishop Samson (1096–1112). In the Worcester cartulary to which he gives his name, London, British Library MS Cotton Tiberius A.xiii, Hemming names himself as the monk and priest ('monachus et sacerdos') who compiled the collection of charters (f. 131v). Tinti reads the cartulary as a reaction against the social and political instability that followed the Norman Conquest, with a particular focus on the recuperation of lands that had been lost to the monks, and on the new Norman bishop Samson, whose behaviour is negatively contrasted with that of his two predecessors, described as the monastery's 'admirable benefactors' (2002: 60). Thus, as a rare instance of a known, named Anglo-Saxon scribe, and one responsible for the (partial) copying of several manuscripts, Hemming makes an ideal subject for a study of intra-writer variation in a period whose surviving textual evidence presents a number of practical and theoretical challenges for historical sociolinguistic analysis.

Despite our comparatively detailed picture of Hemming's later life however, we know little of his origins or of his scribal training. Ker describes Hemming's hand as 'old fashioned' (1985 [1948]: 41) when compared with the other hands contributing to the cartulary, indicating that he was perhaps older than his colleagues at the time of that manuscript's production in the 1090s. On the basis of his performance as the scribe of Cambridge, University Library Kk.3.18 (a copy of the Old English translation of Bede's *Historia ecclesiastica*; 'the OE *Bede*'), Wallis (2013a) identified Hemming as a translator scribe, albeit one who retained a number of relict forms in his copy.2 This chapter provides a more detailed assessment of Hemming's

2 Benskin and Laing (1981) identify three main types of scribal behaviour: *literatim* scribes copy their exemplars faithfully, retaining the spelling of the original; *translator* scribes replace exemplar forms with functional equivalents from their own repertoires, although occasional *relict* features may be transmitted from the exemplar and appear as 'show-throughs' in the new copy; a *mixer* scribe adopts both strategies, sometimes transmitting the exemplar faithfully and sometimes translating its OE scribal practice by assessing his variation across a number of his copied texts, namely selections from the OE *Bede*, two prayers, four charter bounds and a curse,3 totalling around 22,000 graphic units. A selection of features are examined which show variation in Hemming's writing: (a) spellings showing <wæ> rather than <we>; (b) o+nasal spellings; (c) retraction in words ending in *-ward*; (d) unstable <h>.

## 2 Theory and method

One of the challenges of using OE texts for historical sociolinguistic research lies in their status as copied texts. Unlike data from later periods, the majority of extant writing from the Anglo-Saxon period represents copies of pre-existing work, of which the author's original text rarely survives. In cases like this it is less easy to detect a writer using variation as a way of appealing to their audience, or as an act of self-fashioning in the way that, for example, Hernández-Campoy and García-Vidal (2018) demonstrate in late-Middle English letters by male members of the Paston family. Nevertheless, Hemming *does* show variation in his written output, in terms of dialect (Late-West-Saxon (LWS) vs Mercian), and in his selection of various features on a scale that could be labelled conservative-innovative (archaic vs modern, or exemplar forms vs trained

features, though the degree to which each strategy is pursued may change over the course of their writing. A *Mischsprache* scribe is a mixer who maintains their mixing behaviour consistently throughout the text.

<sup>3</sup> 'In medieval books, most notably in monastic libraries, an anathema was a curse or imprecation written into the book, usually by a librarian, calling down sorrows upon the head of anyone who stole it […] Some scribes also included anathemata in their colophons against unauthorized copyists into whose hands the manuscript might fall' (Beal 2008, s.v. *anathema*). Hemming's warning follows a Latin text outlining Bishop Wulfstan's motivations in having the cartulary compiled, and threatening excommunication on anyone interfering with Worcester's lands. The curse, then, could be read as referring both to the physical lands, and to the cartulary which was meant to record and protect Worcester's claim to them.

preferences) (Wallis 2018: 82). The question posed by this chapter, then, is: how are we to understand Hemming's variation? Is it an expression of regional or ethnic identity? Worcester was one of only two dioceses to keep its Anglo-Saxon bishop after the Norman Conquest, and is distinguished by its long-standing tradition of post-Conquest production and consumption of texts in English (Treharne 2007). Or, should we rather view Hemming's variation as having its roots in the mechanics of textual copying? Historical sociolinguistic analysis of OE texts is challenging, not least because of our lack of social information about the writers involved, but also because of the kinds of textual production involved in the majority of our surviving documentation of the language; as most of the texts are copies, we are able to see only how a writer reacts to an exemplar in front of them, meaning that we rarely get a glimpse of how that writer might have written a text that they were able to compose spontaneously or autonomously for themselves.

Hemming is the copier (rather than the author) of the texts in his hand, and so theoretical frameworks based on concepts such as audience design (Bell 1984) are less relevant to the study of this material. Therefore the corpus is instead investigated using a framework developed to interrogate Middle English scribally copied manuscripts for dialect evidence, using a scribal profile approach (McIntosh 1974; Benskin & Laing 1981). In this approach, a number of linguistic features are selected, and all tokens of each feature are collected from the sample texts. In this way, a scribal profile is created, with each feature adding further detail to the picture of the scribe's writing habits. In this study Hemming's variation is investigated in two ways: in the first, the text is used as a control; a number of the texts Hemming copied survive in multiple witnesses (e.g. the OE *Bede*, one of the prayers, and one of the charters), meaning that we can compare Hemming's own scribal choices with those of other copyists. This enables us to gauge the kinds of features that might lie behind Hemming's exemplars, and to detect relict forms (i.e. forms that have 'bled through' from the exemplar). The second method is to use the scribe as a control; comparing Hemming's output across a number of texts with different exemplars enables us to judge which features are representative of Hemming's training, and which may be attributed to his *passive repertoire* (Benskin & Laing 1981: 58).

Table 20.1 lists Hemming's surviving scribal output and shows the text selection used in this study. Manuscripts which provide alternative witnesses to texts copied by Hemming are numbered 1a, 2a, etc.

As part of his work at Worcester, Hemming copied both Latin and OE manuscripts (see Table 20.1), and much of his output is related to episcopal concerns; Hatton 114 (no. 5 in Table 20.1) is a collection of homilies probably written for and used by Bishop Wulfstan, while CCCC 391 (no. 2) was formerly known as *Wulfstan's Portiforium* and CCCC 146 (no. 7) as *Samson's Pontifical* (Da Rold et al. 2010). It has already been noted that Hemming's Cartulary seems to be part of an ongoing project by the cathedral's monks to restore lost or alienated property. It is also clear from Hemming's surviving manuscripts that he sometimes worked as part of a larger team of scribes: part 2 of Hemming's Cartulary (no. 3) was written by three main hands, CCCC 391 (no. 2) contains four hands writing OE, while at least six hands were responsible for the homilies in Hatton 114 (no. 5) and its sister volume (Hatton 113). Eleven scribes worked on the homilies in Junius 121 (no. 6). Hemming's copy of the OE *Bede* (no. 1), by contrast, was predominantly a solo project, with only a few chapter headings provided by the monk Coleman, and running heads supplied by a further hand (Ker 1957: 37; Da Rold et al. 2010).

The textual selection for this study aims to provide enough data to compare (a) Hemming's scribal output in a number of texts with different underlying exemplars (intra-writer variation), and (b) Hemming's scribal practice with that of other scribes copying the same text (inter-writer variation). To this end, the selections from Book 3 of the OE *Bede* (*OEB3*) used in Wallis (2013a) have been supplemented with further material in the form of samples from the text at the beginning of that manuscript (*OEB Intro*) and Book 1 (*OEB1*). The aim of including this additional material was to see whether Hemming has a 'writing in' period, where he adjusts his own copying to the language of his exemplar (Benskin & Laing 1981: 66).4 Only two of the four extant *Bede* manuscripts are complete at the beginning of the text, and Hemming's copy is one of these, meaning that we can see

4 See also Chapter 21 by Iyeiri in this volume, who utilizes Benskin and Laing's framework to explore progressive translation in Middle English texts.


Table 20.1. Manuscripts containing Hemming's hand

what Hemming's copying behaviour was like as he became accustomed to the language of his exemplar. His manuscript begins with Bede's *Preface*, followed by a West-Saxon regnal list and a list of chapter headings, before Books 1–5 of *Bede*. The transmission of the OE *Bede* is complex, however, and Whitelock suspected that the creator of the chapter headings was 'not likely to have been the translator of the work' (1974: 275). Furthermore, Miller (1890: lvi) doubted the presence of the regnal list in the original translation, while Waite (2015: 31f.), argues that 'the Preface is the product of a writer working some time after the *OE Bede* was completed, possibly in West Saxon circles'. This has implications for the language of the underlying exemplar; if the preface, regnal list and chapter headings (*OEB Intro*) were not part of the original translation, then their dialect(s) may have been different, and it will be more difficult to gauge Hemming's own contribution to the text at this point. Therefore, an additional sample was taken from the beginning of Book 1 (*OEB1*), where the underlying dialect is more likely to be consistent with that in the rest of the OE *Bede*.

For some of the texts in Table 20.1, further witnesses are available. The OE prayer in CCCC 391 (no. 2) appears in two mid-eleventhcentury copies (no. 2a), while charter S786 survives as a late-tenth or earlyeleventh-century single-sheet (no. 3a), in addition to the later cartulary copy (no. 3). Although short, these additional witnesses allow us to compare Hemming's copying with that of other scribes, in a variety of text types, dates and dialects.

Finally, items 5, 6 and 7 (highlighted in grey in Table 20.1) do not form part of this study. CCCC 146 contains only Latin texts in Hemming's hand, while facsimile copies of Hatton 114 and Junius 121 were not available due to Covid pandemic restrictions.5

<sup>5</sup> A further manuscript, the rather damaged British Library, MS Cotton Otho C.i.2, is listed by Da Rold et al. (2010) as possibly containing sections in Hemming's hand (ff. 149r–155v), however this assessment is not shared by Ker (1957). Therefore the manuscript has not been included in the present study.

## 3 Data

This section presents and discusses the data for each of the four features under investigation. It begins with <wæ> spellings, before considering o+ nasal, *-ward*/-*weard*, and unstable <h>. None of these features are part of what might be described as 'standard' Late-West-Saxon (LWS), the 'focused' variety of OE used in the late-tenth and eleventh centuries (Smith 1996: 66), and this section investigates what it means when Hemming uses such non-LWS features in his writings.

#### *3.1 <wæ> spellings*

Hemming is one of a number of OE *Bede* scribes to transmit variant Mercian spellings with <wæ> for <we>. Many such spellings derive from the i-umlaut of /a/ (e.g. <wærma>, <wærgan> for <werma>, <wergan>; Campbell 1959: §193a), while a further group of words exhibits a change from Mercian <e> to <æ> following /w/ (e.g. <wæg> for <weg>; Campbell 1959: §328). <Wæ> spellings in the first group remained only in the West Midlands, later becoming ME /wa/; however, the 'precise significance and cause [of group two <wæ> spellings] remain uncertain' (Hogg 1992: §5.179). Examples in *OEB3* include <wærgra> [accursed], <onwæg> [away] (T) and <wærigan> [accurse] (O), in positions where other scribes write <we>. To add to these instances, Hemming has <wærminge> [warming] (noun), <onwæg>, and this distribution, together with its absence from *OEB Intro* suggests that it was a feature of the OE *Bede* archetype.6

In addition to *OEB3*, <wæ> spellings can also be found in some of Hemming's charter bounds, as shown in Figure 20.1.

6 The spellings <godwæbbe> [fine woven material] and <frætwædnysse> [ornament] also appear in *OEB3*. These spellings do not appear in *DOEC*, however they each appear once in the attested spellings listed by *DOE*, with Hemming as their only source.

Figure 20.1. *wællan* in the Charters (total frequencies).

Alongside the LWS form <wyllan>, Hemming frequently writes <wællan> [well, spring]. This form is also found once in a second textual witness of charter S786, in British Library, MS Cotton Augustus ii.6; see (1):

(1) of sondburnan on sceadwællan of sceadwellan […] in clægwyllan of clægwyllan in æðelstanes graf [from sandburn into the shady well, from the shady well […] to the clay well, from the clay well to Æthelstan's grove]7

The scribe of this parallel version includes a variety of spellings, including LWS <wyllan>, alongside Mercian <wellan> and <wællan>. <Wællan> is an Anglian – and specifically a West Mercian – spelling: '<æ> consistent or sporadic is spread right across the Hwiccean region including E. Wark, […] N. Glouc. […] and all over Worc.' (Kitson 1990: 209, fn. 41). It appears, then, that in <wællan> we are dealing with a local spelling, reflecting a local document detailing the relevant land boundary. The fact that both Hemming and the Cotton Augustus scribe transmit Mercian forms indicates that the underlying exemplar possessed Mercian dialect features, including spellings like <wellan> and <wællan>.

None of Hemming's other texts contain <wæ> spellings, and the reason for this may well be that both the *Bede* and the charters are in some way 'local' texts; the *Bede* because its textual history places it firmly

<sup>7</sup> All translations are my own unless otherwise noted.

in a dialect area (Mercian) to which Worcester also belonged, and the charters because these were texts written at the behest of (and possibly by) Worcester's monks. It therefore seems likely that <wæ> spellings are a feature which is part of Hemming's *passive repertoire*: 'those forms which are not part of the active repertoire, but which are nevertheless familiar in everyday usage as the forms of other writers, and which the scribe does not balk at reproducing' (Benskin & Laing 1981: 58).

This would indicate that Hemming was familiar with Mercian dialectal features and so content to incorporate them into his own copy. Judging by his scribal output for Worcester, he was an experienced scribe, and one who was entrusted on at least one occasion with the task of copying a substantial text (the OE *Bede*) alone. His familiarity with Mercian written forms is therefore unsurprising.

#### *3.2 o+nasal*

In LWS words with <a> followed by a nasal, non-West-Saxon spellings often favour o+N, as in *<*hond> [hand]*, <*monn> [man]*, <*ond> [and], and this feature is found particularly in Mercian texts such as the OE *Bede* (Hogg 1992: §5.5). In *OEB3* the occurrence of o+N varies according to the scribe, as demonstrated by a selection of common examples in Figure 20.2. T's scribe retains the greatest number of o+N spellings across a variety of words (394/405 of the total words in Figure 20.2). In contrast, the later manuscripts (including Ca, written by Hemming), are much more variable and typically have far lower frequencies of o+N (O: 101/ 272; Ca: 60/270; B: 24/304).8

The fact that T has high overall counts for o+N is not surprising because it retains many other Mercian features from the OE *Bede's* archetype. Interestingly, Hemming has high use in some words but not others; across all

<sup>8</sup> The total number of words with potential o+N varies according to manuscript, accounted for by lexical variation, and a divergent translation of chapters 16–20 in O and Ca. Furthermore, as *and/ond* is frequently abbreviated in O, Ca and B, we are unable to determine their scribes' preferred spellings for this item.

Figure 20.2. o+N in *OEB3*.

*Bede* scribes, <monig> [many] is most likely to be transmitted with an <o> spelling, even in B (18/50), which habitually transmits LWS equivalents of Mercian archetype features. <Ealond> [island] is also usually spelled with <o> by Hemming (14/17), and <monn> [man] is the preferred spelling in nearly half (35/75) of all instances. While some of Hemming's frequencies are close to those of O and might be explained by that manuscript's closeness to Hemming's (e.g. <monn>, <noma> [name], *<*from>), the two scribes clearly do not pattern alike in their preferences for which words should have o+N. The O scribe's intermittent and erratic use of o+N should not surprise us as he is a *Mischsprache* scribe (Wallis 2013b).9

It should also be noted that the raw frequencies for each word vary widely; frequent words in Ca are <monig> (41), <monn> (75), <from> (69) with total o+N spellings at 100 %, 49 % and 6 % respectively. Among the less frequent words, many have only a+N spellings in Hemming's copy: <hand> [hand] (18), <angel> [Angle-] (17), <and> [and] (10), <land> [land] (13), while only one of the ten instances of <noma> has o+ N. Word frequency does not, therefore, seem to be the underlying factor in the difference.

9 That is, he is a mixer scribe, and this strategy is maintained throughout the text.

Given the rather intriguing results for o+N from *OEB3* (the middle of the text), an examination of Hemming's copying from the beginning of the manuscript might indicate whether he has a 'writing-in period', whereby when faced with an unfamiliar text 'the medieval scribe […] begins by copying closely, even *literatim*, until he reads his exemplar fluently and at a glance' (Benskin & Laing 1981: 66).

The advantage of Hemming's manuscript is that his copy of the *Bede* is complete. However, as noted above, the text at the beginning of the manuscript is probably not part of the original translation. This matters because we cannot be sure what the dialect of the exemplar was; in other words, we don't know what kind of language Hemming was responding to as he was getting used to his copying task. Therefore, this part of the study uses a sample of c. 2,500 graphic units from the beginning of the manuscript, incorporating the entire *Preface* and regnal list, and the first part of the chapter headings. A second sample of c. 2,700 graphic units was taken from the beginning of *OEB1*. As we are dealing with low frequencies the data is tabulated in Table 20.2.

The evidence from the beginning of the *Bede* suggests that o+N may indeed not have been a feature of the earliest parts of Hemming's exemplar (which would be consistent with Miller's (1890), Whitelock's (1974) and Waite's (2015) observations). Although Hemming has opportunity to use o+N spellings, he does not do so, and the absence of the feature in his


Table 20.2. Frequency of o+N in samples from the OE *Bede* (total frequency (o+N) + (a+N) in brackets)

exemplar may well be the reason for this. The *Preface* and regnal list have no instances of o+N. In the chapter headings *monig* (which had 100 % o+ N spellings in *OEB1* and *OEB3*) only appears with <a>, and the only <o> spellings occur in <monn> (2/9) and <ealond> (4/7). The three instances of <ealand> occur right at the beginning of the chapter headings, and this *might* indicate a writing-in period, before Hemming reverts to a preferred <ealond>, however the evidence for <monn> does not pattern so neatly, as the two occurrences appear in the middle of the selection. Compared with the evidence from *OEB3*, where <o> forms contribute a far higher proportion of spellings (<monn> 37/75, <ealond> 14/17, <monig> 41/ 41), it is evident that the language of the underlying exemplar is indeed different in the earliest section, and that is what accounts for the differing frequencies.

It is clear, then, that Hemming's use of o+N is to an extent dependent on his exemplar: the more frequently o+N appeared there, the more likely Hemming was to write it in his own copy. What we cannot tell from the OE *Bede* alone, however, is whether o+N was Hemming's preferred spelling. To answer this question the evidence of the charters and prayers is required.

From the three sets of charter bounds in Hemming's hand in Cotton Tiberius A.xiii, there are three relevant lexical items, <land> (x3)*, <*sand> (x5) and <andlang> (x3). On every occasion Hemming spells the relevant word with <a>. In the Harley charter he writes <land> (x3), <hand> and <and> (1 each), again with <a> (see Figure 20.3).

Figure 20.3. Comparison of o+N in Hemming's charters and in Cott. Aug. ii.6.

In the charters, Hemming's use of o+N is zero. The fact that o+N appears intermittently in the Cotton Augustus version of S786 (2/4 instances) suggests however that somewhere in that text's transmission there *was* o+ N, though it is not possible to tell whether Hemming copied his charter directly from the Cotton Augustus version or another. Nevertheless, we are left with the fact that although Hemming *could* have written <o> in words like <sand> and <land>, he did not. It is also notable that o+N never occurs in Hemming's copies of the prayers in words like <fram> or <nama>, and occurs only once in the short admonition on excommunication, in <noma>:

(2) ic bidde

⁊ eac on godes noman halsige ælc mann hine sylfne georne wið þisne curs warnige. ⁊ þissere stowe hold sy. ⁊ getreowe. ⁊ se þe elles do. hæbbe him wið gode gemæne. ⁊ swa swa þes curs swutelað. [I beg and also pray in God's name that each man should eagerly take heed of this malediction and be faithful and honest in this place. And he who does otherwise, let him account to God. And so this curse declares.]

Therefore the evidence points to o+N being an example of *constrained selection* on Hemming's part. Laing (2004: 63) describes constrained selection as 'when a scribe suppresses some of his own habitual spellings in favour of the (functionally equivalent) others that he finds in front of him'. These spellings must be part of a scribe's active or passive repertoire, and will skew the relative frequency of functionally equivalent forms. O+N occurs frequently in *OEB1* and *OEB3* and is, as we have seen, very likely to be a feature of the original translation, as it is transmitted to a greater or lesser degree in other OE *Bede* manuscripts. The fact that it rarely occurs elsewhere suggests that it is indeed only transmitted by Hemming when it occurs in his exemplar, and given the charter evidence, probably not always then.

#### *3.3 -ward spellings*

A further idiosyncrasy of Hemming's is the retracted spelling <ward>, alongside the more usual spelling <weard>, which shows breaking of the vowel before r+C (Campbell 1959: §144). Figure 20.4 shows a com parison of each *Bede* scribe's usage in *OEB3*: Hemming is the only scribe to use retracted spellings for words ending in *-ward*. Although breaking diphthongs appear in common adjectives like <toweard> [future], <wiðerweard> [contrary, adverse] and <ondweard> [present], each of these words also occurs with retraction. Less common words can occur with both spellings (e.g. <yrfeweard>/ <yrfeward> [heir], <inneweard>/ <inneward> [inward, internal]), only with retracted spellings (e.g. <æfward> [absent]), or only with breaking (e.g. <upweard> [turned or moving upwards]).

Retracted spellings are unlikely to have been part of Hemming's exemplar. They do not appear anywhere in the other manuscripts, including T, which is by far the oldest and most conservative of the *Bede* copies. In addition, retraction before r+C is a feature of Northumbrian; as the *Bede* is an originally Mercian text we would not expect to see this feature in its earliest witnesses.

Figure 20.4. Selected <ward> spellings in *OEB3*.

Another explanation for the presence of retraction lies in the unstressed nature of the syllable. Campbell (1959: §338 and fn. 1) states that breaking may fail in unstressed syllables, and Hogg (1992: §6.7) adds that 'the second element of obscured compounds' is a usual environment for the phenomenon.10 Notably, <ward> is also the preferred spelling of the Worcester scribe who copied London, British Library, MS Cotton Nero E. i, part 2. He consistently updates <weardes> in his exemplar (charter S1280) to <wardes>, a move which Wiles (2013: 232) interprets as reflecting a 'changing phonological situation' in the late eleventh century.

Retracted spellings appear to be part of Hemming's own preferred style; in addition to *OEB3*, <ward> occurs in the chapter headings, *OEB1*, the OE prayer and the Harley charter (3–5):


However, <ward> is rarely the only spelling of the element, and as in *OEB3*, broken forms occur alongside retracted ones (e.g. <toward>, <toweard> (OE prayer); <toweard> (Bilingual Prayer)).

#### *3.4 Unstable <h>*

In this section, two separate but related phenomena are examined: hdeletion and h-insertion, either in prevocalic environments in stressed syllables, or as part of the initial consonant clusters <hl hn hr hw>. As Lass and Laing (2010: 346) note, '[t]hese two phenomena are two aspects of the same process: loss of initial [h] and consequent "hypercorrect"

10 As support for this explanation, Hemming has only breaking spellings in stressed syllables in a similar labial environment, for example, <wearp>, <wearð>.

employment of the now "non-referential" *littera* "h" in positions where it is not expected to be associated with a *potestas*'.11

Hemming's writings show both h-deletion and h-insertion. Examples of the former from *OEB3* are <reas> [fell], <gerinan>12 [to touch] and <nescan> [soft] for <hreas>, <gehrinan>, and <hnescan>. In two cases, the omission is noted and corrected:

(6) ic sceal hraðe deað under\h/nigan [I will soon submit to death] (*OEB3*, f. 37r) (7) medmycelne dæl \h/lafes [a small piece of bread] (*OEB3*, f. 43v)

In each case, <h> is added in superscript. It is difficult, in such a brief intervention, to determine whether these corrections are in Hemming's hand or not; Da Rold et al. (2010) suggest that at least three correcting hands contribute to the OE *Bede* text, in addition to Hemming himself, and while the correction in (6) certainly looks like Hemming's hand, that in (7) is less certain. Clearly, either Hemming or another reader noticed the h-deletion and restored the words to their etymological spellings. Whether this was by comparison with the exemplar, or a spontaneous correction is difficult to tell. Nevertheless, it is striking that not every instance of Hemming's h-deletion was noticed and corrected.

In addition to h-deletion, Hemming is responsible for a number of instances of h-insertion. One such example is in the OE prayer (8), where Hemming's copy reads <hlæne> [lean, thin] instead of <læne> [temporary, transitory, frail]:

(8) Ac loc hwænne min tima beo. ⁊ þin willa sy. ic þis hlæne lif forlætan scyle. [And whenever my time is, and it is your will, that I should leave this thin (*hlæne*)/ transitory (*læne*) life] (OE prayer)

Hemming's text has a misreading here; the penitent is clearly thinking about the time when they will leave this *transitory* life. *Læne* is also used


to describe something that is on loan for a temporary time, and 'generally used as an epithet of things of this world when they are contrasted with those of the next' (*BT*, s.v. *læne*). As confirmation, the other two textual witnesses to the *OE Prayer* read <læne> at this point.

The *OE Prayer* also contains unetymological <h> in two readings of <onhliht> for <onliht> [to illumine, make shine] (a spelling which also occurs in *OEB3*), along with <gehrece> [rule, government], <hleoma> [limbs] and <hlihte> [alighted]. It is clear that both h-deletion and hinsertion are traits of Hemming's own writing, as they occur in more than one text, and are only occasionally corrected. Hemming is not alone in his treatment of <h>, however; the other two textual witnesses to the *OE Prayer* contain the spellings <genehxa> for Hemming's etymological spelling <gehnexa> (<gehnescian>, [to soften]), while the Cotton Tiberius A.iii scribe is a frequent h-dropper, whose spellings include <bereowsian> and <dægwamlice> for Hemming's etymological <behreowsian> and <dæghwæmlice>. For Hemming, unstable <h> appears in the consonant clusters <hl hn hr>; it is not a feature of <hw>, and it does not appear in prevocalic position.13

Hemming's uncertainty about the status of some words with etymological [h] reflects wider changes in English during the late-OE period. As Scragg (2012: 213) notes, there is plentiful evidence for h-loss in early Middle English:

The written history of English suggests that the loss of the sound had occurred by the early Middle English period, although the possibility that it was lost in late Old English and survived in eleventh-century writings largely because of the success of the spread of a standard written form of the vernacular in eleventh-century England has rarely been considered.

Scragg suggests that the widespread use of LWS (what he calls the 'standard written form') masks the beginnings of h-loss, and the evidence of Hemming's writings indeed indicates that these changes were already

13 Other scribes of this period, however, *do* show h-loss in these positions (e.g. the Cotton Tiberius A.iii scribe responsible for the OE prayer and *Monasteriales Indicia*).

under way in the late-eleventh century. The combination of h-deletion and h-insertion raises some tantalizing questions: do Hemming's (over) corrections indicate an awareness of h-loss, or variable pronunciation? And, further, do they indicate an awareness of a 'correct' or desired spelling, in contrast with his own speech?

## 4 Conclusion

Hemming's variation reveals a number of interesting patterns. Firstly, it seems that constrained selection is responsible for features such as <wæ> spellings and o+N. O+nasal occurs in the *Bede* but not in the charters or prayers; even where an alternative witness of one charter preserves o+ nasal, Hemming routinely writes Late-West-Saxon forms with a+N, and it seems that he only writes o+N where it appears in his exemplar, not spontaneously. On the other hand, <wæ> spellings occur in both the *Bede* and in the charters. This suggests that Mercian (and sometimes specifically West Midland) dialectal spellings are part of Hemming's passive repertoire, as the charters and the *Bede* (an originally Mercian text) can be classified as dialectally 'local' productions.

Secondly, -*ward* retraction and unstable <h> seem, from their distribution, to be representative of Hemming's own usage, as they are found across the corpus. These two features have their roots in ongoing sound changes in late OE, monophthongization in unstressed syllables (*-ward*) and the simplification of consonant clusters with initial <h>. It is possible that Hemming's use of <h> in unetymological positions and his self-correction of h-deletion represent an awareness on his part of a preferred or desirable spelling which is at odds with his own pronunciation of the affected words.

Hemming's variation, then, indicates a toleration of locally current written forms stemming from (historical) Mercian orthographical traditions found in the exemplars he copied from. At the same time he introduces spellings which are at odds with 'standard' or 'focused' Late-West-Saxon; these, however should be seen as the reflections of late OE developments in pronunciation, rather than as reflections of a local, specifically 'Mercian' or 'West Midland' orthography.

## References


Annina Seiler & Nicole Studer (eds), *Meaning in Words, Texts and Discourse in the History of English*. Bern: Benjamins. 19–38.


#### Yoko Iyeiri

21 Intra-text variation as a case of intra-writer variation: Middle English scribal behaviours, with a focus on the spelling variation of woman in MS Pepys 21251

#### Abstract

This study deals with intra-text variation as a special case of intra-writer variation, discussing different patterns in the shift of language within a single text. Intra-text variation tends to be observed in medieval texts, which are usually based on exemplars. Scribes may faithfully copy the language of the exemplar at the beginning but shift to their own language within the same text. This phenomenon, which is known as 'progressive translation', is usually gradual, whereas this study, mainly concerned with the spelling of woman in MS Pepys 2125, demonstrates that abrupt shifting is also possible, especially when the exemplar forms are already too archaic for the scribe. This study also suggests that different patterns of intra-text variation are linked to different stages of language change.

## 1 Introduction

It has been widely acknowledged that variation is an essential element in sociolinguistics, as it often reveals some social conditionings related to linguistic activities. It is also a major concept in historical linguistics, as it

1 This study was in part supported by JSPS *Kakenhi* (Grant Number 18K00645). I would also like to acknowledge with gratitude the permission given by Peter Lang and Keisuisha for the reuse of Figures 21.1 and 21.2 in my earlier publications, and by the Pepys Library, Magdalene College, Cambridge for the use of manuscript images (Figures 21.4–21.7). Moreover, I would like to express my thanks to the editors of this volume for a number of comments, which have helped me improve this chapter to a significant extent.

often demonstrates the process of language change. In this study, I discuss how analyses of variation, particularly variation within a single text, can contribute to the understanding of the historical change of language. I am especially concerned with the relationship between intra-text variation and more general intra-writer variation. While the former is a special case of the latter, there may be some additional specificities to address. The principal part of this study focuses on the spelling variation of woman in a Middle English manuscript.

This chapter is structured as follows: Section 2 describes the data and methodology of this study; Sections 3 and 4 provide a brief survey of relevant previous studies on Middle English texts in general and some cases of intra-text variation in them; Section 5 is the most extensive and forms the central discussion on the orthography of woman in the data, together with some discussion on the use of strokes for abbreviation in the manuscript; and Section 6 concludes the study with some discussion on the relationship between intra-text variation and intra-writer variation.

## 2 Data and methodology

The main text explored in this study is *The Chastising of God's Children* (hereafter simply *Chastising*), the first item in MS Pepys 2125, Cambridge.2 It goes back to the early fifteenth century and was copied by a single scribe probably in the Gloucestershire area.3 Since this particular version of *Chastising* has not been edited to date, I have transcribed the text directly from the manuscript and created a corpus of this text, which consists of approximately 36,400 words. I use it as the main corpus in this study, though I also return to its manuscript version when necessary.

<sup>2</sup> *Chastising* in this study refers only to the text in MS Pepys 2125, while *The Chastising of God's Children* is extant in a number of versions. For details, see Bazire and Colledge (1957: 1–37).

<sup>3</sup> For details of this manuscript, see Taguchi and Iyeiri (2019: xlvi–lvii).

Towards the end of this study, I also explore *Pepysian Meditations on the Passion of Christ* (hereafter *PMPC*) as a supplementary corpus. This is the second item in MS Pepys 2125 and was copied by the same scribe. The analysis of this text reveals that intra-text variation as found in *Chastising* is not necessarily observable even though the same scribe is involved. I propose that the different linguistic trends in the two texts may be related to whether the scribe was aware of his choice from among the existing variants. In the research into *PMPC*, I resort to its manuscript instead of its digital corpus, since the target item occurs only seven times in it.

The methodology in this study is essentially quantitative, in that the discussion is mainly concerned with the distribution of different spelling forms of woman in the two texts, though some qualitative inferences are induced when the (im)probability of otiose strokes in examples like *wo~men* is examined. Otiose strokes are those not intended as an abbreviation marker (explained in Section 5.3).

## 3 Intra-text variation in Middle English manuscripts

As an illustrative case of intra-text variation, the present study mainly investigates *Chastising* in MS Pepys 2125, an early fifteenth-century manuscript text. The advantage of exploring medieval manuscripts like this (and perhaps incunabula) is that their linguistic features are often, though not always, variable within the text. This linguistic inconsistency may not necessarily be ideal for medieval dialectology, but provides an excellent opportunity to examine how various possible linguistic options were available to, and handled by, a single language user. This is due to the fact that a large number of medieval texts are copies from exemplars. They illustrate how scribes (or compositors) handled the language of the exemplars and how they balanced the language of the exemplar and their own. See Wallis (Chapter 20 in this volume), who also examines the relationship between the language of the exemplar and that of the scribe in medieval manuscripts.

As for scribes in general, previous studies in medieval dialectology maintain that there are largely three types of practices in copying exemplars. Having the philosophy of *A Linguistic Atlas of Late Mediaeval English* (= *LALME*) in mind, Laing (1988: 85) states:

A scribe may (a) copy the spellings of his exemplar *literatim*, producing an exact transcription; (b) transform the language of his exemplar into his own kind of language, producing a complete translation; (c) produce a mixture of the spellings of his exemplar and his own spellings, creating a so-called *Mischsprache*, a form of language not consistent with any one regional variety.

She then continues that texts of types (a) and (b) provide useful dialectal information, whereas '[t]exts of type (c) are not usable as primary source material for dialect maps' (p. 85). Indeed, one does not like to draw inferences in dialectology on the basis of texts where different dialectal forms are concocted. In my view, however, there are often probable patterns in the mixture of language in type (c), the analysis of which will provide useful information for historical linguists. It is fairer in this context to mention that Laing (1988) also stresses the possibility of extracting dis tinct dialectal forms from texts of type (c), which she considers essential in exploring the early Middle English period, where extant texts are much more restricted than in later periods. I would go a step further and argue that texts of type (c) can in fact be more useful in historical sociolinguistic analyses, as they often provide *meaningful* intra-text variation.

So-called 'progressive translation', which has been noted in previous studies, is one such tendency. The phenomenon is described in *LALME* (I: 15):

A copyist whose habit is to translate text into his own dialect takes time to get used to the language of his exemplar. The phenomenon of 'working-in' when reading unfamiliar hands is probably well-known to any scholar who has transcribed text from old manuscripts; […] For the first few folios or so, he produces a text of which the language is not his own, but that of his exemplar. As he gets used to his copy-text, so he converts with increasing fluency the language of the subsequent text into his own. It may well be that in many such cases what happens is that the scribe moves from copying in a purely visual way to copying via 'the mind's ear'.

From the perspective of historical linguistics, such phenomenon can provide the opportunity to observe the process of language change, as it often involves older and newer linguistic variants in different quantities.

## 4 Further details of language shifting within a text: Two case studies in the previous literature

As argued in Section 3, progressive translation is certainly a pattern of language shift within a single text, which can possibly reveal the ongoing process of language change. It can, in fact, even alter the nature of the text from type (a) (the language of the exemplar) to type (c) (*Mischsprache*) or from type (c) (*Mischsprache*) to type (b) (the language of the scribe) (see above). In other words, texts of type (c) are not always consistently of type (c) throughout, but may involve features of types (a) and (b) when viewed from a dynamic perspective.4

As shifts from (a) to (c) and from (c) to (b) occur commonly in intratext variation and hence are foundational to any discussion of other cases, I would like to demonstrate them by referring to two of my previous studies, one for the shift from type (a) to type (c) and the other for the shift from type (c) to type (b). They both deal with the spelling variants of the pronoun it in Middle English. The variants explored are forms retaining initial *h* (e.g. *hit*, *hyt*) and those without (e.g. *it*, *yt*). It is known that the Old English form *hit* (including its spelling variants) shifted to *it* (including its spelling variants) in the course of the Middle English period, first in

4 The existence of intra-text variation of this kind itself has been noted in various previous studies. Discussing the shift from type (c) to type (b), for example, Laing (1988: 88) remarks: '[…] over the first few folios he may well create a (c) type or mixed text, reproducing in his copy some forms which do not belong to his own dialect but rather to the dialect of his exemplar(s). Indeed such relict forms may be very common, and even dominant, in the early part of a text, before being displaced absolutely by what is evidently the scribe's own usage. The (c) type text gradually evolves into a (b) type'.

unstressed positions and later more generally, though the older form with *h* is occasionally attested in the Modern period, particularly in dialectal varieties (see Brook 1958: 126f.; *OED*, s.v. *it*).

The shift from (a) to (c) is illustrated, in fact, by *Chastising* in MS Pepys 2125, the very text explored in this study. Iyeiri (2013) investigates the shift from *h*-forms to *h*-less forms of it within this text and finds a gradual intra-text shift as shown in Figure 21.1.

Figure 21.1. Raw frequencies of *h*-forms and *h*-less forms of it in *Chastising* (based on Iyeiri 2003: 3465 ).

This is close to a model case of progressive translation and shows the shift from type (a) to type (c). The text begins with the older *h-*forms, which were probably inherited from the exemplar, moving gradually to the mixture of *h*-forms and *h*-less forms, and finally shifting to the state where *h*-forms are increasingly marginalized, though still existent. I note in this study that the marginalized *h*-forms on later folios were probably selected very carefully for the purpose of emphasis, as they tend to be

5 Iyeiri (2013) had missed a few relevant examples of it in calculation, and this has been rectified in Iyeiri (2018). Hence, the graph has been slightly modified according to the revised statistics, though the shape of the graph looks largely the same. Since there are as many as 360 examples of it in *Chastising*, the argument stays the same before and after the revision of statistics. The statistics are based on the counting of the subjective and objective cases of it only, since the rise of the genitive form *its* in the history of English encompasses a separate issue.

attested sentence-initially and in some cases with *H* enlarged (see Iyeiri 2013: 346). It is probable that *h*-forms were rather archaic for the scribe but that they were still acceptable to him to a reasonable degree, though often reserved for special purposes.6 As mentioned in Section 2, the text probably goes back to the early fifteenth century.

Towards the end of the same century, by contrast, we find texts with a different pattern of variation between *h*-forms and *h*-less forms of it, which in my view illustrates a different stage of language change. Iyeiri (2018) demonstrates that *hit* disappears in the middle of the printed text of Nicholas Love's *Speculum Vite Cristi* (1494), altering the types from (c) to (b). Figure 21.2 depicts this intra-text variation. Unfortunately, this is a case of an incunabulum, whose compositors are difficult to identify – in the case of a manuscript it is easier to tell whether a text was copied by a single scribe or not.7 Still, this graph shows a notable shift between quire h and quire i, switching the text from type (c) to type (b), though the older forms with *h* pop up later in quire o, which may simply be accidental. It is quite possible that *hit* was already marginal and therefore perhaps too archaic for someone involved in and responsible for the text, for example, compositor, who followed the exemplar at the beginning of the text but shunned the older form after a while. Furthermore, it does not seem to be reserved for special purposes, unlike the case of *hit* in *Chastising* discussed above.

Detailed analyses of language shift within a text like the above can reveal useful information about the ongoing language change, and this is


Figure 21.2. Raw frequencies of *hit* and *it* in different quires of *Speculum Vite Cristi* (based on Iyeiri 2018: 100).

perhaps a fruitful area for further exploration in historical sociolinguistics. Medieval manuscripts (and perhaps incunabula) are particularly useful for this purpose, since they are relatively free from post-editorial activities and therefore reflect the true shift of language within a text. The following sections present an additional and yet different case by investigating the orthography of woman in *Chastising* in MS Pepys 2125.

## 5 The orthography of woman in *Chastising*

#### *5.1 Some preliminaries*

The orthography of woman in the history of English is complex. As it is essentially a compound of wife and man, it goes back to forms with *f* and *m*, such as *wifmann*, *wyfman* and *wifmon*. In the development of this compound to the present-day form *woman*, two major changes have taken place: (1) the replacement of *i*, *y* by *o*; and (2) the shift from the consonant cluster *fm* to *m*, which took place through the intermediate stage of *mm*. The same applies to the plural form *women*, except that its pronunciation still reflects earlier *i*, *y* (cf. *OED*, s.v. *woman*, *n*; Campbell 1959: §484).8

The alteration of the vowel seems to have taken place earlier in western dialects, as Upward and Davidson (2011: 186f.) state:

The anomalous *o* in ModE *woman*, *women* (ME *wimman*, *wimmen* < OE *wifman* 'wife-man', *wifmen*) did not arise until the 13th century, first of all in texts in western ME dialects in which the original -i- had become a rounded vowel. There have been a variety of spellings and pronunciations in the past, including, for example, a 16thand 17th-century pronunciation of *woman* which allowed a play on words between *wo*- and *woe*.

*Chastising*, localized in the Gloucestershire area in the Southwest and going back to the early fifteenth century, already shows a consistent use of *o* instead of *i*, *y* both in the singular and plural forms of woman. It is, therefore, not necessary to deal with the vowel any further in the following discussion, however it may have been pronounced.

By contrast, the shift from *fm* through *mm* to *m* is relevant. Of the three variants, the combination *fm*, which is the oldest, is no longer evidenced in *Chastising*. 9 Hence, the discussion hereafter concentrates on the variation between *mm* and *m*. The singular and plural forms of woman are treated together, since the issue of *mm* and *m* is relevant to both. The question is how *mm*- and *m*-forms are distributed in the corpus under analysis and, if there is a shift within the text, whether the shift has any historical implications.

<sup>8</sup> Campbell (1959: §484) lists a number of cases where assimilation of consonants took place, including the shift from *fm* to *mm*.

<sup>9</sup> The *fm* spellings are attested in Middle English, but in a very restricted manner and mainly in early Middle English. See *Middle English Dictionary* (s.v. *womman*).

## *5.2 The distribution of the variant forms of* woman *in* Chastising *in MS Pepys 2125*

As mentioned in the previous section, woman appears both with *mm* and *m* in *Chastising*. Table 21.1 gives the full inventory of the orthographic variants of woman including its plural forms, together with their raw frequencies:


Table 21.1. The forms of woman in *Chastising*

On the whole, the spelling of woman in *Chastising* has reached a consolidating stage. Apart from the *a* and *e* variation in the second syllable, which is related to the singular and plural distinction as in present-day English, variation is restricted to the issue of *mm* and *m* only. The genitive plural form *womens* occurs only once:

(1) othir mennys preyers and *womens* beth nedful: (15v) [other men's and women's prayers are necessary] (my translation)

As Table 21.1 demonstrates, double *m* forms are already rare in *Chastising*, indicating their marginal status in the language of the text. Since they are the older forms, if a shift is present in the text, a plausible shift is from *mm* to *m*, representing a case of progressive translation. In other words, *mm* is likely to be the variant inherited from the exemplar, whereas simple *m* belongs to the scribe's language.

The distribution of the forms in *Chastising* indeed makes this probable. As Figure 21.3 shows, forms with *mm* are attested only at the beginning of the text, more specifically only on the first two folios. Supposing that the *mm* forms have been inherited from the exemplar, the shift in this text is from type (a) (the language of the exemplar) to type (b) (the language of the scribe), although the entire text, when viewed as a whole, is of the (c) type (*Mischsprache*). It is interesting that the shift takes place without the intermediate stage where both *mm* and *m* are attested in a mixed manner. This abrupt shift hints at the possibility that *mm* forms were already too

Figure 21.3. Raw frequencies of *mm* and *m* forms of woman in *Chastising* (folios 1–28).

old-fashioned for the scribe. Perhaps he copied them from the exemplar at the beginning of his work but suddenly decided to abandon them, as it was nonsensical for him to keep doing so.

To generalize from this case, abrupt shifting is also a probable pattern of intra-text variation. This appears to be observed when the language of the exemplar is too archaic for the scribe, though other motivations may also be possible. A plausible pattern is that the older forms are attested only at the beginning of the text, but this does not continue for a long time. The older forms are abandoned quite abruptly, perhaps by the decision of the language user, i.e. the scribe. Hence, the shift in this case is 'from above' in sociolinguistic terms.10 From historical linguistic perspectives, this pattern can possibly be viewed as a sign of the marginalization of the older variants. In other words, this is a pattern that is likely to be observed at the final stage of the *S*-curve (cf. Swann et al. 2004: 268f.), where the older variant is very close to extinction but still remains as a residual form.

10 I use the term 'from above' in the Labovian sense, that is, above the level of conscious awareness, though 'from above' and 'from below' are more frequently used in the phrases 'change from above' and 'change from below'. These are concepts widely utilized in sociolinguistics, both historical and contemporary. Cf. Trudgill (2003: 20–21) and Swann et al. (2004: 35–36).

## *5.3 Otiose strokes in the spelling variants of woman?*

The above section has shown the distribution of the six examples of *mm* forms of woman as against the *m* variants, which are the newer and dominant forms in *Chastising*. Since three of the six examples of *mm* are found abbreviated with a stroke, which I have expanded in the above analysis, some additional comments on this point are in order. More specifically, the six examples of *mm* forms are: *wo~man* (1 example), *wommen* (2 examples), *wo~men* (1 example), *womme~* (1 example) and *wo~me~* (1 example). Here and throughout this chapter, *o~* represents *o* with a stroke on it and *e~* represents *e* with a stroke on it. See Figures 21.4 and 21.5, which have been transcribed as *wo~man* and *wo~me~* respectively in this study:11

Figure 21.4. *wo~man*, MS Pepys 2125, 1r.

Figure 21.5. *wo~me~*, MS Pepys 2125, 1v.

For reference's sake, see Figure 21.6, which illustrates an unequivocal case of *mm*.

Figure 21.6. *wommen*, MS Pepys 2125, 1r.

11 Figures 21.4–21.7 in this study have been reproduced by permission of the Pepys Library, Magdalene College, Cambridge.

Considering the fact that the strokes of this kind are usually expandable as *m* or *n*, 12 I have done so in the previous section: *wo~man* has been interpreted as *womman*; and *wo~men*, *womme~*, and *wo~me~* as *wommen*, all counted as *mm* variants in the statistics.

Middle English palaeography, however, remarks that strokes of this kind may or may not be expandable, suggesting that one has to take a cautious stance on this issue. Commenting on so-called otiose strokes, Parkes (1979: xxix) argues: '[…] the fifteenth century scribes used otiose strokes as a feature of calligraphic decoration, and some of these strokes can look very much like marks of abbreviation […]'.13 Despite this possibility, however, I would take the view that the strokes found with woman in *Chastising* are functional and therefore need to be expanded for the following reasons. First of all, woman with strokes is attested only on the same folios where expanded *mm* forms are attested, namely folios 1 and 2, and disappears thereafter. Table 21.2 exhibits the distribution of the forms with clear *mm* and those with strokes in the text:

Table 21.2. Distribution of the *mm* forms and those with strokes in *Chastising*


As shown in this table, there are only three examples with *o~*, all of which are witnessed in the environment where *mm* is attested. Secondly, the two examples with a stroke on *e* on folio 1v, which obviously needs to be expanded, suggest that strokes in general were most probably functional


for the scribe. The existence of *wo~me~* (1v) is particularly supportive for the functionality of the stroke, since it will be only natural for both strokes to be functional instead of the one on *e* only. Furthermore, the overall functionality of strokes in this manuscript has been confirmed by additional research. See the extensive discussion by Taguchi and Iyeiri (2019: lxxvii– lxxx).14 Overall, it is safe to regard the above forms with strokes as examples of *mm* forms, though I would also admit that the occurrence of forms with strokes such as *wo~man* in a way represents an intermediate stage. It is certainly possible that they worked as a mediator to the following stage where double *m* experienced a complete deletion.

#### *5.4 Additional comments on the spelling of woman in MS Pepys 2125*

Before moving on to the conclusion, it is perhaps fairer to touch on the state of affairs in *PMPC*, though it neither supports nor weakens the hitherto argument. It is the second item in the same manuscript, that is, MS Pepys 2125 (28v–38v), and known to have been copied by the same scribe as *Chastising*.

Although it is reasonable to assume that the same kind of language shift takes place in *PMPC* as well, this is not necessarily the case, which may partly be due to the unfortunate fact – unfortunate for the purpose of this study – that there are only sparse attestations of the lexical item in question. *PMPC* gives only seven examples of woman in total: *wymen~* (31r), *wy~men* (32v), *woman* (34r), *woman* (36r), *wymen~* (37r), *wy~men* (37v), *wy~men* (38v). The retention of *y* is an archaic feature of the orthography of woman and attested only in the plural in *PMPC*. Of the seven examples, only two are in the singular and both appear in the form *woman*. The rest are in the plural and with *y* either with or without a stroke on it.

14 Additional relevant cases in the same manuscript such as *da~pnable* for *dampnable* (8v) and *torme~tith* for *tormentith* (24v), where the expansion of the stroke is necessary, are discussed in Taguchi and Iyeiri (2019). It is also mentioned that the strokes in *Chastising*, especially in middle position, are very sparingly attested and, when they occur, tend to be written with a careful touch. Hence, the strokes in this text were perhaps written when needed and with conscious attention by the scribe.

As shown in Figure 21.7, the stroke is written with a fairly careful touch and in the arch form in two of the examples of *wy~men*, 15 and therefore I would expect it to be intended as a marker of abbreviation, which yields *mm* in expansion:

Figure 21.7. *wy~men*, MS Pepys 2125, 32v.

As for the stroke in final position, that is, *wymen~* (31r) and *wymen~* (37r), it is difficult to judge its expandability. As it is quite lax, it may simply be otiose. All examples of *wymen* are with a stroke either in middle or end position.

All in all, the abrupt shunning of *mm* as observed in *Chastising* is not encountered in *PMPC*, though the scribe involved is the same in both texts. In *PMPC*, relevant examples are very restricted in frequency, and they consistently yield *o* and *m* in the singular. In the plural, archaic *y* is retained followed by either *mm* or *m*, supposing that the stroke on *y* is expandable. One probable explanation is that, as I have argued elsewhere, the scribe seems to have followed the exemplar more strictly in copying *PMPC* than *Chastising* (Iyeiri 2013: 345–47). First of all, it is likely that *PMPC*, whose content is more directly linked to the life of Christ, was more highly valued by the scribe than *Chastising*, even though *Chastising* is also a religious text. A second possibility, which is in my view more relevant to historical sociolinguistics, is that there needs to be a reasonable frequency for a particular lexical item to show an abrupt shift within the text. Or, to incite the scribe's awareness, the same lexical item may at least need to appear concentrated in a cluster, if not so frequent in the entirety of the text. While *Chastising* yields as many as 78 examples of woman, *PMPC* provides only seven. Furthermore, the examples are very sparsely distributed within the text. Under such circumstances, the scribe involved may not have been given a chance to consider the modernization of the

15 The remaining example gives a rather lax horizontal stroke.

archaism in the exemplar. The retention of *y* in *PMPC* suggests that the language of the exemplar of this text was probably more archaic than that of *Chastising*, to begin with. Even so, the scribe probably kept copying the archaic forms quite mechanically. The same perhaps applies to *mm* and *m*.

## 6 Concluding remarks with some discussion

As thus far examined, the scribal behaviour towards the exemplar in Middle English can be various, depending upon, for example, the lexical item, whether it appears repeatedly in the text, how archaic it was for the scribe, and how the exemplar text was valued by him or more generally. Concerning the variant spellings of woman in *Chastising*, it is perhaps safe to state that the forms with double *m* were already marginal in the early fifteenth century. Hence, the scribe abandoned them quite abruptly after copying some from the exemplar. He was likely to be conscious of his linguistic behaviour. In other words, this is a case of shifting 'from above', which can be contrasted with the gradual shifting of language within manuscripts, which may be a shift 'from below'.

I consider that the identification of various patterns of linguistic shifts within a single text like this provides useful information as to the stage and process of the ongoing language change. Most typically, shifts within a text are from older forms to newer ones,16 as often discussed under the term 'progressive translation' in medieval dialectology. The present study has highlighted a case of the shift from (a) to (b), which is quite abrupt. This is an additional pattern to the often-discussed shifts from (a) to (c) and from (c) to (b). In theory, there remains the possibility for the language to shift from type (a) through type (c) to type (b) within a single text, which I leave for future studies.

16 While the present study has focused on the patterns under the category of 'progressive translation', there are occasional cases where language comes to be progressively archaized. See Laing (1992: 576), where some cases of the latter kind are mentioned.

Before closing this study, I would like to return to the relationship between intra-writer variation and intra-text variation. Intra-text variation as examined in the above discussion is obviously a special case of intra-writer variation, since the same language user shows a shift of linguistic features in his or her linguistic activities. One has to be aware, however, that style shift as often observed in intra-writer variation is usually a response to external factors, especially the intended audience, whereas intra-text variation as found in medieval texts is likely to be a response to the exemplar, or more specifically, to the dialectal differences or ongoing language change in the linguistic environment of the scribe.

## References


#### Phil Beier, Gohar Schnelle & Silke Unverzagt

# 22 Intra-writer variation in Old High German and Old Swedish: The impact of social role relationship on constructing instructions

#### Abstract

Do social differences affect the choice of morphosyntactic means to express instructions? This corpus-based explorative study on intra-writer variation in an Old High German [= OHG] text by Notker of St Gall and Old Swedish [=OS] texts by Birgitta of Sweden aims to investigate the relevance of social role relationship [=SRR] for the determination of historical register. The corpora were enriched with information on addressor, addressee and their SRR. Three morphosyntactic features functioning as instructions in both OHG and OS were examined for a systematic correlation with SRR. Imperatives were found to be independent of SRR and serve as a ubiquitous means to express instructions. In contrast, subjunctives and modal verbs are dependent on upward and downward SRR, respectively.

## 1 Introduction

Registers in a speech community provide a window on the structure of society and form a predictor for linguistic variation. Thus, the study of linguistic variation is of particular interest when contexts of language use change, as was the case when vernaculars started to be used in written contexts.

In this chapter, we present the results of a corpus-based inquiry on register-dependent variation in Old High German [=OHG] and Old Swedish [=OS] texts by two individuals, Notker of St Gall (c. 970–1022) and Birgitta of Sweden (1303–73), respectively. Notker's Psalter and Birgitta's revelations were chosen as landmarks in the emergence of vernacular writing traditions in their respective languages.

We focus our study on the linguistic variants within instructive contexts, a relevant register component in our resources.1 Instructions are functional instantiations of communicative contexts with the communicative goal of influencing the future behaviour of the addressed person (Neumann 2014: 59). We define *instruction* from the text typology perspective (Werlich 1976: 121ff.) as an underlying contextual function influencing language use, in this case the use of forms serving this instructive function, which can then be realized by different linguistic means (= variants), knowing that the term *directive* (Searle 1979: 13f.) looks at this linguistic phenomenon from the perspective of Speech Act Theory, while the term *request* refers to a specific case of directives based on a narrower definition of the resulting social act (Márquez Reiter 2000: 13).

Communication in instructive contexts has been considered in earlier sociolinguistic studies, especially focusing on education and language acquisition (Bernstein 1990; Cloran 2005; Painter 2005; Chiang et al. 2021) and systemic functional semantic modelling (Cloran 2016). Within instructive contexts different linguistic means are used to realize this communicative metafunction. In both OS and OHG, the imperative, subjunctive finites and modal verbs serve this task. Philological studies of the individual historical languages have primarily looked at their formal linguistic properties (Schrodt 1983; Donhauser 1986; Petrova 2008). In this study, we want to explore how this linguistic variable is linked to the situational-functional variable *social role relationship* [=SRR].

The chapter is structured as follows: In Section 2, we explain what we mean by *register*, *social role relationship* and *intra-writer variation*. We also introduce the *Field-Tenor-Mode model* and our intention to focus on the intertwining of two components of this model (*instruction* and *social role relationship*). Section 3 describes our resource BiNoKo [Birgitta-Notker-Korpus] and the operationalization of the research question. The quantitative findings are discussed in Section 4. Section 5 presents the summary and an outlook for future research strands.

1 This study is part of the Sonderforschungsbereich 1412: Register: Language-Users' Knowledge of Situational-Functional Variation, project B04: Register Emergence and Register Change in Germanic.

2 Register, Social Role Relationship (SRR) and the expression of instruction

Register refers to the systematic co-occurrence of certain quantitative linguistic feature constellations with recurring situations of use (Halliday 1978: 31–35; Neumann 2014: 16; Neumann & Evert 2021: 144; Seoane & Biber 2021: 1). Registers are promising explanatory variables for the investigation of historical variation (Biber 2012). Thus, registers feature prominently in variationist research on both contemporary languages (Szmrecsanyi 2019; Tagliamonte 2006) and historical languages (Kytö 2019). However, the analysis of register variation in the earliest attested stages of a language like OHG and OS, with limited sources and different writing cultures, poses a methodological challenge (Schnelle 2018).

A problem in historical sociolinguistics is keeping extralinguistic parameters constant. Particularly, we must be confident that variant distributions are not influenced by diatopic or diachronic factors. To address this problem, we start with the exploration of intra-writer variation in individual texts. Systematic variation can then be expected to be dependent on situational dimensions.

The present study is embedded in our project's general research question on how written registers emerge at a time when the written vernacular was just beginning to establish itself for the purpose of distant communication. For this period in the histories of both OHG and OS, two aspects must be kept in mind. Firstly, Latin (as the predominant variety used for written communication) is expected to have a large influence on the choice of a particular variant in either OHG or OS. Secondly, the first written texts were almost exclusively produced in religious contexts. The surviving Christian texts serving as linguistic and textual role models for text production go back at least several centuries to multilingual literary traditions (Ganslmayer 2018). Religious genres were omnipresent in medieval education, theological discourse and religious practice and must have been associated with certain varieties of language use. Both parameters will be kept in mind during the discussion of the results. For historical languages, a set of relevant parameters must still be established. Since these languages were spoken in cultures alien to us today, this is a difficult task.

To model the recurrent situations of use, we use Halliday and Matthiessen's (2014) systemic functional Field-Tenor-Mode model [= FTM-model]. The FTM-model was designed within the functionally oriented Systemic-Functional-Linguistic framework [=SFL] with the purpose of providing a universal descriptive model for the characterization of communicative situations (Neumann 2014: 15; Hasan 2014). Based on the assumption that registers are constitutive of language use independently of historical and/or cultural backgrounds (Halliday 1978: 110f.), the FTMmodel has the potential to be applied to different languages and cultures, as well as to be flexibly extended with language-specific parameters (Biber 1994; Wegener 2011).

The FTM-model predicts that every instance of language use can be situationally and functionally described on three interdependent dimensions: (1) the functional *Field*-dimension describes the content and effect of the communication, (2) the social *Tenor*-dimension describes the participants of communication, and (3) the physical *Mode*-dimension provides a framework for describing the concrete and abstract material used for communication.

One of the most common situational parameters applied in register studies is different *Tenor*-dimensions, since social differences are always reflected in language use (Bell 1984). The subdimension SRR is a universally proven factor in language variation (see, e.g. Simon 2017), being further subdivided into the parameters of *level of authority*, referring to social hierarchy, and *level of expertise*, referring to educational/professional hierarchy (Neumann 2014: 136). These subdimensions also play a role in our text material: both Notker's Psalter and Birgitta's revelations consist mostly of dialogues. In most cases, the protagonists are identifiable on the basis of information given directly or indirectly in the text. The SRR between them varies and can be determined by levels of authority and/ or expertise. The hierarchical gap between addressor and addressee (with dialogues, e.g. between Jesus and a Prophet or Jesus and Birgitta) is not only pervasive but, as a factor in formality, also very likely an important parameter of language use.

The *Field*-dimension is further divided into several subdimensions, amongst others *goal orientation* [=GO], describing on an abstract level the change in the world intended by the addressor (Neumann 2014: 54), like instruction, argumentation, narration and evaluation. The GO we are examining is instruction, which is defined as communication that aims to change the addressee's behaviour. From our close reading experience, it emerges that instructions play an important role in both corpora, since inferiors are instructed by hierarchically superior characters to behave or act in a certain way (e.g. Jesus Christ tells Birgitta what she should do). On the other hand, superiors are instructed by inferiors, too (e.g. a Prophet complaining about his misfortune and asking God to behave more kindly towards him). As an important but also delicate social practice, instructions are expected to be influenced by social parameters.2 For this reason, we triangulate the Tenor variable SRR with the Field variable GO and investigate how the formal variants functioning as verbal expressions of an instruction are distributed regarding SRR.

We focus on instructions uttered directly to the addressee. Instructions in both OHG and OS are expressed either by the *imperative* (cf. 1), the present *subjunctive* in the second person (cf. 2) or the present indicative of *modal verbs* in the second person (cf. 3).



We start with the hypothesis that different instantiations of SRR somehow affect the linguistic realization of instruction. We intend to explore whether the SRR may influence language use, focusing on the particular example of the choice of morphosyntactic means to express instructions.

## 3 Data and methods

For our research on intra-writer and register-dependent variation in OHG and OS, we created the Birgitta-Notker-Korpus v 1.0 (BiNoKo; Beier et al. 2023a), consisting of Notker's Psalter and Birgitta's revelations. Both parts of the corpus were designed and built according to the same principles.

birgitternnorska. <https://project2.sol.lu.se/fornsvenska/> accessed 10 February 2022 (Delsing 2002).

#### *3.1 Historical sources and state of digitization*

The OHG part of BiNoKo, Notker's Psalter, is taken from the Old German Reference Corpus (ReA = Referenzkorpus Altdeutsch; Donhauser 2015), which is a deeply annotated multilayer corpus of all OHG and Old Saxon texts (excluding glosses).5 ReA is accessible via the search and visualization platform ANNIS (Krause & Zeldes 2016; Krause 2019) and is stored in the LAUDATIO repository6 (Guescini & Odebrecht 2020). The psalter consists of 150 psalms with a total of 145,737 text tokens and is a didactic textbook in which the Old Testament text is translated and commented on for teaching purposes (Glaser 2016; Glauch 2013).

The OS part consists of two manuscripts of Birgitta's revelations: A 65 and E 8902.7 A 65 contains three revelations while E 8902 comprises 34 revelations.8 The revelations consist of divine wills, calls to repentance for secular and religious dignitaries, and invocations for the pope's return to Rome (Nyberg 2008: 1610). Both manuscripts represent an early stage of editing: A 65 is a first-hand account by Birgitta written in Rome in the 1360s (Högman 2009: 70). For E 8902, however, it remains unknown who produced it and where. It has been suggested that the revelations in E 8902 were copied from a Swedish source by Norwegian scribes either in Sweden (most probably Vadstena) or Norway (Oslo, Hovedøya or Munkeliv), possibly sometime between 1380 and 1420 (Adams 2016: 74f.). The manuscripts differ from each other both in content and linguistic characteristics. Both manuscripts are accessible through 'Fornsvenska textbanken' (Delsing 2002), which is a database of machine-readable editions of OS texts. Taken together, A 65 and E 8902 contain 37 revelations, with a total length of

<sup>5</sup> A description of the corpus is to be found at <https://www.deutschdiachrondigi tal.de/> accessed 10 February 2022.

<sup>6</sup> <https://doi.org/10.34644/laudatio-dev-WiWkDnMB7CArCQ9CyBEw>.

<sup>7</sup> A 65 is located at the National Library of Sweden (Stockholm) and E 8902 at the National Archives of Sweden (Stockholm).

<sup>8</sup> Except for four revelations in E 8902, all revelations in E 8902 and A 65 are found in the Latin, Swedish or in both traditions (Adams 2016: 77–79). The four exceptions and the portion of the revelation in I:22 embedded in I:25 have been examined in context with their surrounding texts.

approximately 32,000 tokens. In contrast to the OHG part of BiNoKo, there was no linguistically annotated database available, so the texts were annotated before conversion to ANNIS.

#### *3.2 Annotation*

The corpus architecture of BiNoKo was designed to model SRR. We added up to five annotation layers: addressor, addressee, srr, section and litForm. These layers are organized in spans, which include text token. Labels for addressor and addressee identify the protagonists giving or receiving instructions, like Jesus, God or sinner. The social role relationship is determined according to level of expertise and level of authority as up(wards), down(wards) or equal. In addition, we included span annotations for Notker's threefold organization of the text in Vulgate text, translation and commentary (section: origo, trans and comm). Thus, we will be able to test if the degree of textual autonomy is relevant for the distribution of the formal means to realize instructions. Finally, a theological classification of psalms was annotated in the litForm layer, addressing the possibility that, for example, laments or praises behave differently linguistically. For the Birgittine texts, the reference to a Latin source (section) and a classification of revelations are not available in the corpus. All annotation principles, annotation practices, case-by-case decisions and technical details are laid out in the comprehensive annotation guidelines (Beier et al. 2023b), which will be accessible online and include multiple examples.

For the grammatical features, the OHG part was able to rely on the annotations of the ReA. However, the Birgitta part had to be annotated from the ground up based on text files. Tokenization, lemmatization and relevant part-of-speech annotations were modelled according to the corpus architecture of the ReA. The OHG data were obtained by search queries via ANNIS, while the OS data were collected manually. The differences in text sizes will be considered when interpreting the empirical findings.

## 4 Data analysis

The distribution of the variants imperative, subjunctive and modals will be presented and discussed separately for OHG and OS in the following sections and then compared in the summary.

## *4.1 Old High German*

For a first quantitative approach, let us consider the overall number of annotations assigned for SRR, not only restricted to instances of instruction. In Notker's Psalter, there are 4,843 instances of a protagonist addressing another individual. As shown in Table 22.1, the constellation of unequal relations occurs far more frequently than equal relations. Thus, it appears that the situational parameter SRR is of analytical value in the given text. In unequal relations, communications directed towards a social inferior (=down) appear almost three times as often compared to upward communications, when counted based on spans annotated for SRR (70.7 %). Apparently, communications directed downwards are shorter than their upward counterparts. Therefore, the 70.7 % of SRR spans for downward relations represent only about 60 % of the tokens on the annotation layer text of the Psalter.


Table 22.1. Proportion of the SRR-categories in spans and in tokens9

#### 9 Numbers are given as 'relative (absolute)' counts.


Table 22.2. Distribution of instruction variants according to SRR in OHG (absolute numbers in brackets)

Table 22.2 shows the distribution of the three variants according to SRR. Most instances occur in an up-SRR (63.7 %). The imperative with a total of 1,172 occurrences is by far the most prominent variant.

As can be seen in Figure 22.1, the smaller absolute counts of modals and subjunctives come with a complementary distribution in the SRRcategories: whereas thirty-six of forty-two subjunctives are used within instructional communication towards a social superior, this distribution is mirrored in the distribution of modal verbs, which occur mostly within instructions directed towards a social inferior.10

<sup>10</sup> The Cramer's V measure for the effect of SRR on the occurrence of all three variants is 0.1053, indicating a weak effect (*p*-value is < 0.00001, significance level at *p* < 0.05). The Cramer's V measure of the effect of SRR on the occurrence of subjunctive or modal verb is 0.6623 (strong effect).

Figure 22.1. Distribution of the variants imperative, modal verbs and subjunctives in social role relationship categories. Scaling is different for each variant visualization.

The subjunctive is preferred in SRR-up-contexts11 like (4) (here: David to God).

(4) *Dîn ána-siûne ne-chêres dû fóne dînemo chínde.* your face NEG.avert.SUBJ.SG.2 you from your child [You shall not avert your face from your child.] (Ps 68, 238, 19)

The exceptions to this pattern are ambiguous. The subjunctive of these occurrences could also be analysed as a marker of subordination.

With a total of fourteen occurrences, the absolute count of modal verbs is even smaller.12 Compared to the subjunctive instructions, the modal verbs seem to form a complementary pattern: almost all modal verbs are

<sup>11</sup> This finding corresponds with Concu (forthcoming).

<sup>12</sup> Twelve instances contain *sculan*, while two contain *durfan*. Since there are so few occurrences, we cannot discern any relevance based on the choice of the modal verb.

used when talking to an inferior (down). Nevertheless, there is a very interesting exception to this pattern, which shows in a quite straightforward way that social macro-categorizations (as reflected by our SRR annotation) are contradicted by individual social acts (in this case the linguistic behaviour). (5) is uttered by an addressor group introduced as *heretici* (categorized and annotated as *heretic*) to the prophet David, who as a prophet stands above all other people.


From an absolute macro-perspective based on medieval social hierarchy structures, the SRR for this constellation was therefore categorized as up. The fact that this exception is found in the utterance of a person who disregards social structure might be reflected by unconventional language behaviour on a micro-level – by using the modal verb to realize the instruction function, he may want to show that he thinks of himself as superior to David. In fact, only two examples of modal verbs in instructive contexts directed towards a superior were found.13

As said before, this result must be discussed in the context of Latin dependency. Table 22.3 shows that the original Vulgate text units (origo), translated text units (trans) and the commenting text units (comm) are included in equal parts. This reflects Notker's consistent recurrent text preparation for didactic use. However, token counts within these text units are quite different: 51 % of all text tokens are comments, while, as expected, the Latin original and translation include approximately the same number of tokens.

13 The other case besides (5) could also be read as a future expression.


Table 22.3. Proportion of section-annotation in spans and tokens

Table 22.4. Distribution of sections trans and comm according to variant of instruction


Without performing a deeper analysis of the Latin text at this point, we evaluate the occurrence of the variants within one of the OHG section-categories trans and comm as an indicator of the degree of Latin dependency.

Table 22.4 shows that the distribution of subjunctives and modal verbs in trans and comm15 is different too: 83.3 % of the subjunctives are included in translation parts, meaning there is a Latin original they could have been influenced by, whereas most modal verbs are included in commentaries, where no Latin original exists.

As (6) shows, the subjunctives within trans are adoptions of Latin subjunctives.16

14 inscr (headings) are not considered in the analysis.


(6) origo *Non me derelinquas usquequaque*. NEG me forsake.SUBJ.SG.2 ever trans *Niêner ferlâzest du mih.* never forsake.SUBJ.SG.2 you me [You shall never forsake me.] (Ps 118, 440, 19–20)

(7) illustrates one of the seven instances included in comm:


As indicated by the initial words *daz chît* 'this means', this example is a paraphrase added to the translation part. Such additional paraphrases are analysed as commentaries (Beier et al. 2023b), although it may be debated whether the Latin form could still be influencing the variant choice within the paraphrase.17 Additionally, the use of modal verbs differs from that of

17 Of the seven occurrences in commentaries, two are paraphrases, and the remaining five are addressed to God: God is addressed directly again by the subjunctive.

the subjunctive in that they mostly occur in comm-sections and thus lack a Latin prototype.18

To sum up, we see that the imperative is the variant used in the overwhelming number of cases. It can be used in all SRR-relations. Subjunctive examples occur far more rarely and are preferred when the SRR is up, while the most infrequent variant is the modal verb, preferably used in SRR-down-situations.

#### *4.2 Old Swedish*

For OS, the analysis was carried out in two stages. Firstly, the total number of the instructions in A 65 and E 8902 were categorized according to the SRR-categories up, down, equal and na [= not available] (see Table 22.5). Secondly, we counted the distribution of each variant across SRR-categories (see Table 22.6). In both cases, the instructions in A 65 and E 8902 were taken together since the instructions in A 65 are too few in number.

Table 22.5 illustrates the total number of instructions and their absolute and relative frequencies in each SRR-category. Of the 140 instructions, only one (0.7 %) is used in communications between equals, and 13.6 % are directed towards a superior, while the majority (81.4 % of the instructions) are directed towards an inferior individual. As far as the addressor-addressee-relationship is concerned, most cases in the category up are found in the communication between Fiend and Christ.19 For the category down, it is the communication between Christ and Birgitta which is most common, and for the category equal, the communication is between workers. The category na, where the SRR cannot be identified, is

<sup>18</sup> The one occurrence included within a trans-span was analysed as trans following the annotation guidelines (Beier et al. 2023b), without having a Latin model.

<sup>19</sup> In all the examples examined, Fiend appears to Christ only as a suppliant before the court, therefore the SRR was categorized as up. In a larger body of text, the Fiend may assume more diverse roles.



Table 22.6. Distribution of instruction types according to SRR-categories


Figure 22.2. Distribution of the variants imperative, modal verb *skula* and subjunctives in social role relationship categories. Scaling is different for each variant visualization.

also worth noting: 4.3 % of the examples belong to this category. In most cases, these concern dialogues between voices from hell and a soul.

Table 22.6 and Figure 22.2 present the distribution of the 140 instruc tions across their variants and SRR. As we can see, the imperative is the first choice for forming instructions. It clearly emerges that the examples with subjunctives and modal verbs as well as the greater part of the examples with an imperative are used with the SRR-category down. The SRR-categories up and equal only occur with imperatives.20

Our survey has shown that the imperative is the primary choice for instructions, regarding its quantity and due to its occurrence in all three SRR-categories. It is also worth noting that the SRR-down context seems to influence the use of subjunctives and modal verbs. Both occur exclusively within this particular category. Moreover, the SRR-down context prevails in all instruction variants. This is probably what one would expect in revelatory literature. Birgitta's revelations have an educational function, illustrated for instance by divine wills and calls to repentance for secular and religious dignitaries, which imply a superior instructor and an inferior audience.

In addition to the direct instructions discussed above, we also looked at indirect instructions (i.e. instructions that do not refer to a second person but to a third person who is not involved in the communicative situation). They are formed with the subjunctive or a modal verb. 129 indirect instructions occur in the examined data – 33 with a subjunctive and 96 with a modal verb. (8) contains an indirect instruction from A 65.

(8) *byri fyrst pafin mz sik sialfum* begin.PRES.SUBJ.SG.3 first the pope with himself [First the pope shall begin with himself ] (Bir. IV:49)21

21 The OS text is based on the text from 'Fornsvenska textbanken': Äldre Fornsvenska (1225–1374): Birgittaautograferna (Delsing 2002); Cf. <https://project2.sol.lu.se/ fornsvenska/> accessed 10 February 2022.

<sup>20</sup> The Old Swedish figures are too low to compute the effect of SRR on type of instruction. We leave this for future research.

In this example, we witness a communicative situation in which the addressor is telling the addressee what the pope should do. Such instructions, expressed by a present subjunctive, are a common feature in medieval Swedish law texts (Pettersson 2005: 17f.) – a text type which Birgitta might have been familiar with, as she was both daughter and wife of a lawyer (Pernler 1993).

## 5 Summary and outlook

In our analysis we have shown that SRR indeed has an impact on the choice of a variant for giving instructions. For both Notker's Psalter and Birgitta's revelations in A 65 and E 8902, the imperative is the most frequent variant for instructions and can thus be seen as the default for both languages, while subjunctives and modal verbs seem to be dependent on the social role relationship. The factors that trigger the choice between the default and the other variants need to be determined in future research.

In the case of OHG, the two less frequently used instruction variants of subjunctive and modal verb show a complementary distribution. While the subjunctive is especially used when an up-instruction has been translated from a Latin original, modal verbs, though rare, are used for down-instructions given to an inferior and within a commentary section. In the OS material, however, the use of subjunctives and modal verbs differs from the one in OHG. Both the subjunctive and modal verbs appear exclusively in down-instructions. However, there is a distinction in the preference between these two. We see that a modal verb is preferred to the subjunctive and this preference applies not only to direct but also to indirect instructions.

Since we are looking at data from a period of media adaption in the vernacular, the distributional patterns of the variants may be considered a reflex of the emergence of written registers. Latin as the dominant written language is expected to have had an impact on this process. In this context, the subjunctive seems to be a variant closely related to the Latin original and associated with a particular teaching context. Both the orientation towards Latin as the established written variety and the use in instructional communication are indicative of a possible connotation carried by the subjunctive in communication contexts in which formal language use is expected. This consideration must be extended to other Latin-dependent linguistic features. This can be achieved by a quantitative comparison with the Latin Vulgate original as well as with other Latin texts from Notker's contemporary, contrasting the findings with distributions of autonomous features. Finally, a comparison with Middle High German texts would be fruitful.

For Birgitta, it could be of interest to first compare the results with a larger body of texts. In addition, a comparison of the instruction variants in A 65 and E 8902 with the Latin translation and the later Swedish redactions of these revelations would be useful in order to see whether the instruction type changes over time. Furthermore, looking at the use of instructions by specific individuals, such as Birgitta, Christ and Mary, may reveal a preference for a specific variant in a specific situational context. And finally, a comparison with other revelatory literature like Mechthild's revelations both in OS and Latin would add to our understanding of how instructions are expressed in Swedish revelatory literature.

In general, we have been able to prove that the Tenor-dimension is a relevant factor in the linguistic realization of instructions. Hence, an investigation of the effect of other linguistic means might be interesting to add to the picture of the linguistic profile used within this situational instance.

## References


#### Oliver Currie

23 On the indexical meaning of literary style shifting: The case of word order variation in the sixteenth-century Welsh Bible translations1

#### Abstract

The sixteenth-century Welsh Bible translations show striking intra-writer variation between the poetic books of the Old Testament, which have frequent verb-initial order in positive declarative main clauses, and the prose books of the Old and New Testament, where this construction is rare. The Welsh Bible translators may have exploited existing linguistic variation – such verb-initial order was rare in prose but common in poetry – in a novel way, using a construction associated with poetry to give a poetic quality to prose translations of Biblical Hebrew poetry. While the meaning indexed by this style shifting appears to be primarily literary, the process of the style shifting is essentially sociolinguistic and the chapter argues for a rapprochement of sociolinguistic and literary concepts of style.

## 1 Introduction: Literary style and sociolinguistic style

Given that the key sociolinguistic term *style* was adopted from rhetoric and literary studies (Hernández-Campoy 2016: 3–31) and that literary texts also form a significant part of historical (socio)linguistic corpora, it is somewhat of a paradox that there has been s0 little cross-disciplinary discussion of the literary and sociolinguistic concepts of style.2 The


'Third Wave' sociolinguistic conception of style, which posits the individual as the locus of stylistic variation and in turn emphasizes the creativity of individual language users as agents (Coupland 2007: 84; Eckert 2000: 44f., 2008: 456f.), is inherently compatible with the literary con cept of style, which also perceives style as a characteristic feature of individual writers and an expression of their literary and creative identity. Nevertheless, when as historical sociolinguists we study stylistic variation and style shifting in written texts, including literary texts, the question still arises: to what extent do we need to and how do we take into account literary factors in our analysis?

One aspect of this question is philological in nature and relates to the interpretation of the textual evidence: observed linguistic variation in texts which may appear to be *prima facie* sociolinguistic may also have been influenced by other factors which are at least in part literary, such as a given writer's use of different sources or literary or discourse traditions. Another aspect of this question touches upon a more fundamental theoretical issue. If we conceive of sociolinguistic stylistic variation and style shifting as 'the individual's internalization of broader *social* distributions of variation' (Rickford & Eckert 2001: 1; my emphasis) and the individual's exploitation of such social distributions of variation in their use of language to index existing *social* meanings or even to create new meanings, what do we understand by the term 'social', both as the source for individual linguistic variation and its result in the indexed social meaning? Do we understand 'social' in a narrower sense as encompassing primarily the prototypical 'social' parameters analysed in sociolinguistic field studies such as class, status, region, ethnicity, gender, community, group identity, etc. and any values which may be associated with them, or do we understand 'social' in a broader sense as denoting the whole environment in which the individual lives, including all its cultural and literary dimensions?

In some cases, for example, the phenomenon of *literary dialect*, where non-standard speech is represented using dialect (or approximations of it) in English literature, typically in dialogue (Mair 1992; Beal 2006), the linguistic variation in literary texts seems to parallel (without necessarily replicating) style shifting observed in sociolinguistic field studies, in that it exploits 'social distributions of variation' to index the meanings, values and perceptions associated with such variation as well as for literary effect (Hakala 2017). However, style shifting in written texts may also draw on 'distributions of variation' which are not obviously or exclusively 'social' in the usual sense. In the sixteenth-century Welsh Bible translations, for example, which are the subject of this chapter, we find striking word order variation between, on the one hand, the books of the Old Testament translated (into Welsh prose) wholly or mostly from Biblical Hebrew poetry and, on the other hand, the books of the Old Testament translated (into Welsh prose) wholly or mostly from Biblical Hebrew prose as well as the books of the New Testament translated from Greek prose. In the poetic books of the Bible (in particular the Psalms), we find a much more frequent use of a verb-initial construction – *absolute-initial verb order*, where a finite verb comes in absolute-initial position in a positive declarative main clause – compared to the prose books. This frequent use of *absolute-initial verb order* in prose texts was an innovative feature of the sixteenth-century Bible translations (Currie 2013, 2016). In other sixteenth-century prose as well as in earlier (Middle Welsh) prose texts the construction was rare, though it was frequent in sixteenth-century and earlier Welsh poetry. It is possible – and such a hypothesis is advanced in this chapter – that the sixteenth-century Welsh Bible translators – first William Salesbury in his 1567 New Testament and Psalms then William Morgan in the 1588 complete Bible translation – noted this existing variation between poetry and prose and exploited it for stylistic effect, potentially to give a poetic quality to their prose translations of Biblical Hebrew poetry.

The remainder of this chapter analyses and discusses the significance of this word order variation, exploring in particular the question whether it represents sociolinguistic style shifting and what significance such 'literary' style shifting has for the concept of sociolinguistic style. First, Section 2 provides an overview of the construction *absolute-initial verb order* and Section 3 presents the corpus and methodology. Section 4 then analyses the word order variation in the corpus, while Section 5 discusses the stylistic dimension of the variation. Finally, in Section 6, the conclusion discusses the significance of the stylistic variation in terms of broader theoretical issues, in particular whether we can combine the concepts of literary and sociolinguistic style.

## 2 Absolute-initial verb (AIV) order: Definition and historical overview

The word order variation centres on a specific verb-initial construction, which I term *absolute-initial verb order* (AIV order), where a finite verb comes in absolute-initial position in a positive declarative main clause, illustrated in (1a). The categorization 'absolute-initial' is necessary to distinguish AIV order from other verb-initial constructions where a preverbal particle precedes the verb in absolute clause-initial position, such as the dummy subject (*ef*, *fe*, *e* – derived from the 3SG masculine personal pronoun *ef* [he]) in (1b).


The restriction of AIV order to positive declarative main clauses (PDMCs) reflects a major historical word order split between positive declarative main clauses, on the one hand, and negative main clauses and subordinate clauses, on the other. In Middle Welsh prose (c. 1100–c. 1500), the word order of positive declarative main clauses was predominantly verb-medial with at least one constituent (subject, object, verbal noun object, adverbial phrase) fronted before the verb (Poppe 2000). In subordinate and negative main clauses, however, the word order was predominantly verb-initial (conjunction/negative particle + VSO). The word order of subordinate and negative main clauses has also remained stable throughout the history of Welsh, while that of positive declarative main clauses has undergone a major shift from verb-medial in Middle Welsh to verb-initial in Modern Welsh (c. 1800 to the present day). The development of verb-initial patterning in Modern Welsh involved the emergence and/or increase in frequency of several different verb-initial constructions and this chapter focuses on just one such construction – AIV order in Early Modern Welsh (c. 1500–c. 1800).

In Middle Welsh, AIV order is rare in prose – with no examples in some texts and often only one or two where it is attested – but common in poetry (Currie 2013: 48–51, 2016: 157f.).3 AIV order becomes more frequent in prose texts from the second half of the sixteenth century and William Salesbury's 1567 translation of the Psalms is the first continuous prose text in Welsh with relatively frequent AIV order (20 % of PDMCs in the corpus sample), though AIV order is significantly less frequent in his 1567 New Testament. William Morgan's complete 1588 Bible translation amplifies and extends William Salesbury's pattern of intra-writer variation, with more frequent AIV order than Salesbury in the Psalms as well as in other poetic books of the Old Testament not translated by Salesbury, but less frequent AIV order in the prose parts of the Old Testament and (prose) New Testament. While there is a more general increase in the frequency of AIV order in prose in the seventeenth and eighteenth centuries (Currie 2013; Willis 1998), there is also marked variation between individual writers throughout the early modern period. For example, at one extreme, Charles Edwards and James Owen (writing in the last third of the seventeenth century) use AIV order as the predominant order (at times in over 50 % of PDMCs), while, at the other extreme, contemporary writers such as Rondl Davies and William Jones avoid the construction altogether, and others such as Thomas Williams show intermediary patterns of usage of AIV order (approximately 20 % of PDMCs) (Currie

3 AIV order is more frequent in PDMCs in Old Welsh prose (c. 800–c. 1100), though the textual evidence is only fragmentary. The Middle Welsh verb-medial (or in some analyses verb-second) patterning in PDMCs is likely to have developed diachronically from a cleft structure, which may have arisen at a time when verbinitial order was predominant in PDMCs and which then came to be generalized (Willis 1998: 100; Currie 2013: 52). The more frequent use of AIV order in Middle Welsh poetry may reflect linguistic conservatism in the Welsh poetic tradition as well as the fact the availability of AIV order as a variant word order was metrically useful for Welsh poets, as it gave them more flexibility in syllable count and assonance patterns.

2013: 60–69). The intra-writer variation in the frequency of AIV order which we see in William Salesbury's and William Morgan's Bible translations is therefore part of a wider pattern of inter-writer variation as well as of diachronic change.

## 3 The corpus

The corpus of sixteenth-century Bible translations analysed in this study comprises extracts from the 1567 and 1588 Bible translations:


The extracts included in the corpus are shown in Table 23.1, which also gives the number of tokens of PDMCs and the percentage frequency of AIV order in each corpus text. The analysis focuses on intra-writer variation in the two main translators, William Salesbury and William Morgan. Accordingly, the corpus was designed to include (where possible) texts by both writers which were translated from Greek and Hebrew as well as from both poetry and prose, and also texts of different genres, such as Old Testament songs or lyric poetry (Psalms and Song of Songs), narrative prose (Gospel of Mark, Acts of the Apostles) and expository prose (Paul's Epistle to the Galatians). However, since William Salesbury only translated one Old Testament book, the Psalms, we cannot compare in his case translations from Hebrew poetry with those from Hebrew prose, as we can with William Morgan. Since Morgan also translated the whole Bible, we also have texts by him in a wider variety of genres which are also included in the corpus: Isaiah (prophecy), Ecclesiastes (sermon, gnomic), Revelation (vision).

The focus of this study is to understand the variation in frequency of AIV order evident in both writers between the prose and poetic books, specifically to what extent it can be explained in terms of sociolinguistic style shifting. Salesbury uses AIV order in over 20 % of PDMCs in his translation of the Psalms, but in only 2.3 %–6.6 % of PDMCs in the three prose New Testament books included in the corpus (Mark, Acts of the Apostles and Galatians). Morgan not only uses AIV order twice as frequently as Salesbury in his revised translation of the Psalms (41 % of PMDCs), but also uses the construction frequently in other poetic books of the Hebrew Old Testament (47 % of PDMCs in the Song of Songs and 25 % in Isaiah). On the other hand, Morgan uses AIV order much less frequently in the Old Testament books translated from Hebrew prose – 4 % in Ecclesiastes and 9 % in Esther – as well as in the four prose New Testament works translated from Greek (ranging from 2 % in Revelation to 7 % in Acts).

To investigate whether syntactic factors may have constrained the variation, each PDMC in the corpus was also tagged to include the following information:



Table 23.1. Composition of the corpus

\* In 1567 the Book of Revelation was translated by Thomas Huet.

† OT = Old Testament, NT = New Testament.

No comparative analysis has been performed of the word order in Welsh translations and the original Hebrew or Greek; this will need to be the subject of a separate study. Since verb-initial order was frequent in Biblical Hebrew, there is a possibility that verb-initial order in the original Hebrew may have also influenced the use of AIV order in the Welsh translations. Thomas (1988: 115) indeed notes by way of example a series of correspondences between AIV order in the Welsh translation of Psalm 105: 20–32 and verb-initial order in the Hebrew original. It should be stressed, though, that there is marked variation in the frequency of AIV order between the Welsh translations of different Hebrew books of the Old Testament in the corpus, ranging from 3.9 % in the mostly prose Ecclesiastes to 46.8 % in the poetic Song of Songs. Moreover, Song of Songs immediately follows Ecclesiastes in the Bible, so that this marked variation in word order is actually juxtaposed. It would seem, then, that if there is linguistic interference from Biblical Hebrew, it is not a straightforward case of the Welsh translators slavishly following Biblical Hebrew word order, but rather their being sensitive to it and taking it into account when translating, possibly along with other linguistic, stylistic and theological factors.

## 4 Possible syntactic constraints on the word order variation

Differences in genre and text type between the books of the Bible – in addition to the overarching prose-poetry distinction – also entail syntactic differences which can in turn impinge on word order variation. One of the key syntactic differences associated with text-type is the relative frequency of the different persons of the verb, as shown in Table 23.2 for the 1567 and in Table 23.3 for the 1588 Bible translations respectively. For reasons of space, impersonal subjectless constructions (0 %–6 % of PDMCs) have not been included in the tables, so the percentages do not add up to 100 %.

Three of the corpus texts – Mark, Acts and Esther – are narrative texts written primarily in the third person. In five other texts – in particular Revelation, Ecclesiastes and Song of Songs, to a lesser extent Galatians and the Psalms – first person verbs are especially frequent. Isaiah is more mixed: as a prophetic text with the voice of the prophet expressed in the first person, it has relatively frequent first person verbs (9 % of PDMCs), but still more frequent third person verbs, as the subject matter of the prophecy is narrated in the third person. The Psalms and Galatians are the only texts with a significant number of second person verbs: second


Table 23.2. Relative frequency (%) of different persons of the verb in Salesbury 1567\*

\*Abbreviations: *Imp/pass* = impersonal constructions with a logical personal pronoun subject and passive constructions with personal pronoun subject. *NomS* = verbal construction with nominal subject.


Table 23.3. Relative frequency (%) of different persons of the verb in Morgan 1588\*

\*Abbreviations: *Imp/pass* = impersonal constructions with a logical personal pronoun subject and passive constructions with personal pronoun subject. *NomS* = verbal construction with nominal subject.

person singular in the Psalms (praise and supplication addressed to God) and second person plural in Galatians (the addressees of Paul's epistle). If we then examine the breakdown of the occurrences of AIV order by the person of the verb, certain distinct syntactic patterns emerge.

Tables 23.4 and 23.5 show the relative frequency of AIV order compared to Personal Pronoun Subject + Verb (PSV) order (the single most frequent competing construction) when the subject of the verb is a personal pronoun (either expressed or as a null subject). Where there are fewer than ten examples for a given person of the verb in a text, no percentage frequency is shown, and for this reason 1PL and 2PL columns are also not included in the tables. (For reasons of space, only the percentage frequencies of AIV and PSV orders are given in these tables, so the percentages do not add up to 100). Similarly, Tables 23.6 and 23.7 show the frequency of AIV order compared to the main competing constructions in clauses with nominal subjects by text, first for Salesbury then for Morgan.

We can observe the following patterns in the distribution of AIV order based on the data in Tables 23.4–23.7:

i. Irrespective of text type, AIV order is much more frequent in both Salesbury and Morgan when there is a personal pronoun as opposed to a nominal subject; subject verb order is predominant in all texts when there is a nominal subject.


Table 23.4. Percentage frequency (%) of AIV vs PSV orders in Salesbury 1567


Table 23.5. Percentage frequency (%) of AIV vs PSV orders in Morgan 1588

Table 23.6. Percentage frequency (%) of AIV order in PDMCs with nominal subjects in Salesbury 1567


\*Abbreviation: *Adv+y+V =* Adverb + particle *y(r)* + Verb.

ii. In Salesbury, there is further significant variation in the frequency of AIV order between different persons and numbers of the verb. Salesbury uses AIV order relatively infrequently (< 5 %) with 3SG verbs with a personal pronoun subject in all texts, however he uses AIV order relatively frequently in the Psalms with 1SG, 2SG and 3PL verbs (ranging from 31 % to 53 %).


Table 23.7. Percentage frequency (%) of AIV order in PDMCs with nominal subjects in Morgan 1588

\*Abbreviation: *Adv+ y+ V* = Adverb + particle *y(r)* + Verb.


New Testament prose text Revelation, although 1SG verbs are common in both these texts.

These observations suggest that the more frequent use of AIV order in the poetic books was in part influenced by syntactic factors associated with text and discourse type, since verbs with personal pronoun subjects (in particular 1SG, 2SG, 3PL forms), with which AIV order seems in general to have occurred more frequently, happened to be more common in the poetic books. Yet at the same time, the fact that we find significant variation in the frequency of AIV order for the same types of verbal constructions (1SG, 3PL) between different texts by the same writers suggests that writers had a significant degree of flexibility in being able to choose between AIV order, on the one hand, and Personal Pronoun Subject + Verb order, on the other. There is evidence for the interchangeability of AIV order and Personal Pronoun Subject + Verb order not only in the intra- and inter-speaker variation in the frequency of the constructions, but also in examples of the substitution of AIV order for Personal Pronoun Subject + Verb order in different versions of the same text, such as Morgan's revision of Salesbury's translation of Psalms 5:6 in (2a) and (2b).

(2a) PSV order:

*Ti ddestrywy y rei y ddywedant gelwydd* You-2SG destroy-2SG.PRES the ones who say-3PL.PRES lie [You destroy those who tell lies] (Salesbury 1567; Psalm 5:6)

(2b) AIV order: *Difethi y rhai a ddywedant gelwydd* Destroy-2SG.PRES the ones who say-3PL.PRES lie [You destroy those who tell lies] (Morgan 1588; Psalm 5:6)

Such variation between AIV order and Pronoun Subject + Verb order seems to have been enabled by the apparent functional equivalence of the two constructions. In contrast, there seems to have been a significant functional difference between subject verb order and verb subject order for nominal subjects. The fronting of nominal subjects was used, for example, to express topic shift as well as focus, so AIV order was not interchangeable with Nominal Subject + Verb order. AIV order does, however, seem to have been perceived as interchangeable with Dummy subject + Verb + Nominal Subject constructions – as in (1) where Morgan has substituted AIV order for the dummy subject in his revision of William Salesbury's translation of Mark 1: 40 – as well as with dummy subject impersonal constructions.

## 5 Style shifting in the use of verb-initial order in the Bible translations

The more frequent use of AIV order in the poetic books also appears to be stylistically salient. Its occurrences are often concentrated: we find sequences of the construction in consecutive clauses and verses. In such cases, as illustrated in example (3) from Psalm 18: 36–38, the repeated use of the AIV order, juxtaposed with sequences of other word order patterns, helps render the characteristic parallelism of Biblical Hebrew poetry. In such parallelism, we typically find parallelism of content – the repetition of ideas often with verbal echoes – matched by parallelism of form, frequently involving the repetition and contrast of different word order patterns (Berlin 2008: 3–17).

(3a) AIV order:

*Ehengaist vy-cerddiat y danaf, ac ny lithrawdd* Broaden-2SG.PAST my path under-1SG and NEG slip-3SG.PAST *vy sodlau* my ankles AIV order: *Erlidiais vy-gelynion ac eu daliais* pursue-1SG.PAST my enemies and them catch-1SG.PAST *ac nyd ymchwelais tragefyn nes ym ei diva* and NEG return-1SG.PAST back until to-me their destroy AIV order: *Archollais hwy val na allent sefyll*: injure-1SG.PAST them so that NEG can-3PL.IMPERF stand

```
AIV order:
       cwympesont y dan vy-traet
       Fall-3PL.PAST under my feet (Salesbury 1567; Psalm 18: 36–38)
(3b) AIV order:
       Ehengaist fyng-herddediad tanaf: fel na
       Broaden-2SG.PAST my path under-1SG so that NEG
       lithrodd fy sodlau.
       slip-3SG.PAST my ankles
       AIV order:
       Erlidiais fyng-elynnion, ac ai goddiweddais:
       pursue-1SG.PAST my enemies and them catch-1SG.PAST
       ac ni ddychwelais nes i mi eu difa hwynt.
       and NEG return-1SG.PAST until to me their destroy them
       AIV order:
       Archollais hwy fel na allent sefyll:
       injure-1SG.PAST them so that NEG can-3PL.IMPERF stand
       AIV order:
       syrthiasant dann fy nhraed.
       Fall-3PL.PAST under my feet (Morgan 1568; Psalm 18: 36–38)
       [You gave a wide place for my steps under me, and my feet did not slip. 
       I pursued my enemies and overtook them, and did not turn back till they 
       were consumed. I thrust them through, so that they were not able to rise; 
       they fell under my feet.] (English Standard Version/ESV 2007; Psalm 
       18: 36–38)
```
The salience of frequent AIV order in the poetic books would seem to suggest that it is a distinctive stylistic feature of the text, both from the point of view of the translators, who chose to use the construction at a time when it was still infrequently used by contemporary prose writers, and for those who read the Psalms or listened to them being read. The function of the frequent, salient use of AIV order in the Welsh translation of Biblical poetic texts may have been twofold. First, the availability of AIV order as a productive syntactic construction facilitates the stylistic use of word order variation and therefore the rendering of parallelism. Second, the fact that AIV order was rare in contemporary prose (and probably also in contemporary spoken discourse), but frequent in contemporary and earlier

poetry, meant that AIV order could be perceived as a poetic feature. AIV order was a salient feature of Middle and sixteenth-century Welsh poetry not only because it occurred noticeably more frequently than in prose, but also because it was often incorporated into key formal features of poetry such as metre (especially syllable count) and assonance patterns.

A particularly interesting and pertinent example of the variation in the frequency of AIV order between poetry and prose is the fourteenthcentury Welsh text *Gwasanaeth Meir* (Roberts 1961), a translation of the Latin liturgical text *Officium Parvum Beatae Mariae Virginis* (known in English as 'The Little Office of Our Lady'). *Gwasanaeth Meir* is a mixed prose and verse text, with extracts of the gospels in prose and hymns and psalms in poetry. The poetic sections of *Gwasanaeth Meir* including the psalms have frequent AIV order, but there are no examples of AIV order in the prose sections, even where there is verb-initial order in the Latin original (Currie 2016: 157–59).

William Salesbury's innovative use of AIV order in his translation of the Psalms in contrast to the prose books of the New Testament, where the construction is relatively infrequent, may therefore have been influenced by poetic style. By using the construction in his translation of the Psalms, he could give his prose translation of the Biblical Hebrew poetry a poetic quality, creating a form of poeticized prose. In the case of historical texts, it can be more difficult to identify the source as well as to interpret the indexical meaning of such style shifting. Salesbury did not leave any commentary on his translation strategy or stylistic choices. We do know, however, that Salesbury had extensive access to Welsh poetic models: not only was he in close contact with contemporary poets such as Gwilym Hiraethog but he also had a strong antiquarian interest in earlier Welsh manuscripts, possessing his own collection. Salesbury also knew the text *Gwasanaeth Meir*, and as Mathias (1970) and Thomas (1976: 172–74, 1988: 101f.) have shown, he seems to have consulted *Gwasanaeth Meir* when translating the Psalms, as his translation has some verbal echoes of *Gwasanaeth Meir*.

We do, nevertheless, have evidence of the perception of the potential indexical meaning of Salesbury's use of AIV order in the Psalms in the form of William Morgan's response to it in his 1588 Bible. While on the whole Morgan significantly modified Salesbury's language in his revised translation of the New Testament and Psalms, eliminating in particular archaic, idiosyncratic and inconsistent features in Salesbury's orthography, morphology and vocabulary, he chose to retain Salesbury's frequent use of AIV order in the Psalms, while maintaining its relatively infrequent use in the prose New Testament (Thomas 1976: 348–55; Currie 2022). In fact, Morgan amplifies and extends this pattern of stylistic variation, since he uses AIV order twice as frequently as Salesbury in the Psalms and also uses it frequently in other poetic books of the Old Testament (Songs of Songs and Isaiah), but not the prose books of the Old Testament (Esther and Ecclesiastes). Morgan's use of AIV order in other poetic books, however, poses a different interpretative challenge.

The concept of Biblical Hebrew poetry – particularly the distinction between Biblical Hebrew prose and poetry – is a controversial one and has changed considerably over time both in Jewish and Christian traditions. As Kugel (1981) and Berlin (2008) have shown, there is not a categorical distinction between Biblical Hebrew poetry and prose, but rather a stylistic continuum between the two. Parallelism, for instance, is found in both prose and poetry, but is more systematic in poetry. The Psalms and Song of Songs could be perceived as unambiguously poetic from the perspective of either the source or target culture as they are songs, and songs are characteristically metrical. However, the prophetic books (e.g. Isaiah), which are mostly poetic in form in the Hebrew original, may not necessarily be perceived by target cultures as intrinsically poetic.

If the hypothesis that Salesbury's and Morgan's frequent use of AIV order correlates with poetic style is correct, then Morgan's frequent use of the construction in Isaiah and its avoidance in Ecclesiastes may reflect either a contemporary scholarly perception of the distinction between Biblical Hebrew poetry and prose or a target-language influenced perception of different biblical genres. In other words, from a Welsh perspective Morgan may have perceived prophecy (like Isaiah) as more inherently poetic, motivated perhaps by the existence of Welsh literary tradition of prophetic poetry, and sermons (like Ecclesiastes) as inherently prose. Although it is common practice in modern Bible translations and editions to distinguish typographically between prose and poetry, this was not generally the case in Renaissance and Reformation Bible translations (Norton 1993: 162–76), so the linguistic distinction between Biblical prose and poetry maintained in the sixteenth-century Welsh Bible translations by using a poeticized prose in the poetic books may also be innovative in the wider history of Bible translation.

## 6 Conclusion: Indexical meaning and sociolinguistic significance of style shifting in the Welsh Bible translations

While the more frequent use of AIV order in the poetic books of the sixteenth-century Welsh Bible translations would appear to be a clear instance of style shifting, the question remains how comparable it is, because of its predominantly literary nature, to cases of style shifting in contemporary sociolinguistic field studies. If we interpret the meaning indexed by the more frequent use of AIV order in the poetic books as evoking a poetic style, it may seem less typically social and therefore less sociolinguistic. In essence, however, the stylistic variation in the sixteenthcentury Bible translations seems to reflect precisely the phenomenon of individual agency and creative use of linguistic resources to form new meanings and identities which is encapsulated in 'Third Wave' sociolinguistic conceptions of style. For example, in Eckert (2000: 214) 'style is a process of bricolage – an appropriation of local and extra-local resources in the production of not just a pre-existing persona but of new twists on an old persona', and in Coupland's (2007: 84) conception of *styling*, speakers 'can frame the linguistic resources available to them in creative ways, making new meanings from old meanings'. The Welsh Bible translators, William Salesbury and William Morgan, seem to have exploited as their resource existing linguistic variation between prose and poetry and in particular the association of a verb-initial construction, AIV order, with poetry in a novel way, using AIV order more frequently in their translations of the poetic books of the Bible to give a poetic quality to their prose translations of Biblical Hebrew poetry.

The Welsh poetic tradition which Salesbury and Morgan seem to have drawn on was also a prestigious one: it enjoyed greater continuity than the prose tradition, was practised by a professional class of poets under patronage from Welsh nobility and, unlike prose, was codified in bardic grammars. Social and cultural prestige may thus have been an additional motivating factor underlying Salesbury's and Morgan's style shifting, but prestige does not in itself correlate with or on its own explain the variation in the use of AIV order. To try to understand the motivations for and indexical meaning of such stylistic variation, which appears to be both literary and sociolinguistic in nature, we need to consider the full range of potential factors which impinge on the variation – social, literary and cultural – and also have a conception of style and its (social) meaning broad enough to encompass both sociolinguistic and literary perspectives.

The broader conception of style advocated here is, I argue, compatible with a 'Third Wave' approach to sociolinguistic variation as, for example, outlined by Eckert (2008), which takes (social) meaning as the point of departure and has a broad conception of the meaning of a sociolinguistic variable as 'an *indexical field*, or constellation of ideologically related meanings, any one of which can be activated in the situated use of the variable' (Eckert 2008: 454; original emphasis). Unlike 'First Wave' approaches, which have typically sought to establish correlations between variation in the use of linguistic forms and pre-determined macrosocial categories (Eckert 2008: 454f.), the concept of the indexical field in 'Third Wave' approaches is broad enough to encompass all meanings associated with a variable, literary included. A dilemma arises over the meaning of the 'social' nature of stylistic variation in historical sociolinguistic studies such as this one essentially because of the different type of data – written, literary texts – which is the subject of analysis compared to sociolinguistic field studies of contemporary societies, where in practice, whether in 'First, Second or Third Wave' approaches, the parameters of variation investigated have tended to be more stereotypically social. However, we also see patterns of inter- and intra-speaker stylistic variation in historical written texts comparable to those observed in sociolinguistic field studies of contemporary societies, underpinned by the same processes of individual agency and creativity, which exploit 'broader social distributions of variation' (Rickford & Eckert 2001: 1) and which in turn potentially index a broader range of meanings – literary as well as more narrowly social. It seems logical, then, to understand 'social' in a more explicitly holistic sense, as encompassing the whole environment in which an individual lives including its cultural and social dimensions – in Eckert's words 'the social is eminently about the content of people's lives' (Eckert 2008: 456) – and thus to combine sociolinguistic and literary concepts of style.

## References


# Notes on contributors

lucia assenzi is a Postdoctoral Researcher at the Department of German Studies of the University of Innsbruck. She obtained a joint PhD degree in German Philology at the Universities of Padua and Heidelberg with a dissertation on the theory and practice of translation in the *Fruchtbringende Gesellschaft* [Fruit-bearing society]. Now, her main research interests are German historical (socio)linguistics and the history of German metalinguistic reflection. She specializes in language variation and change in the early modern period. She is particularly interested in the interconnections between societal changes, tendencies of metalinguistic reflection and language change at the syntactic-pragmatic interface.

anita auer is Full Professor of English Linguistics at the University of Lausanne, Switzerland. She is a historical sociolinguist and corpus linguist with a keen interest in interdisciplinary research, particularly the correlation between language variation and change, dialect and language contact scenarios, and socio-economic history. Her current research focuses on alternative histories of the English language, that is, the role of historical urban vernaculars in standardization processes; the language of the labouring poor in Late Modern England, as well as language maintenance and shift amongst Swiss heritage speakers past and present in North America. She has published widely on these subjects.

phil beier is a Research Assistant in CRC 1412 'Register: Language-Users' Knowledge of Situational-Functional Variation' at Humboldt-Universität zu Berlin. His research interest is in historical linguistics, especially Nordic philology. His focus lies on language and literature in Medieval Sweden. However, in general his interest covers the Nordic countries' language, literature, and history from sixth century CE to sixteenth century CE. Furthermore, he is interested in Yiddish literature production in Sweden.

joshua brown is a Senior Lecturer in Italian Studies at The University of Western Australia. After his PhD, he completed a postdoctoral fellowship at Stockholm University before returning to Australia. His main areas of interest are historical sociolinguistics, philology, language contact and digital humanities. He also has interests in Italian as a second language, language policies, and the teaching of Italian. Recent publications include a co-edited volume on *Languages and Cross-Cultural Exchanges in Renaissance Italy* (Brepols), 'Women's literacy in a late medieval religious community: organization and memorialization at Santa Marta in Milan, 1405–1454' (*Journal of Religious History*, 2021), 'Towards the elaboration of a diastratic model in historical analyses of koineization' (*Sociolinguistic Studies*, 2020) and 'Language history from below. Standardization and Koineization in Renaissance Italy' (*Journal of Historical Sociolinguistics*, 2020).

oliver currie is an Assistant Professor in the Department of English and American Studies at the University of Ljubljana, Slovenia. He has research interests in historical linguistics, historical sociolinguistics, translation studies and folklore, with a focus on the Celtic languages, English and French. His publications include articles and book chapters on word order variation and change in Early Modern Welsh, the sixteenth-century Welsh Bible translations, the standardization of Welsh, nationalism and language conflict in early modern Europe as well as nineteenth-century folklore collection and translation in the United Kingdom and France.

nora dörnbrack is a Doctoral Research Fellow at the University of Oslo, Norway. Her main research interests are historical sociolinguistics, intra-writer variation in private ego-documents from the eighteenth century and migrants' language use with a particular focus on Late Modern English and women migrants. Her current research is focused on the private writings of the Scottish emigrant to America, Mary Ann Wodrow Archbald (1762–1841), which form the starting point for an investigation into longitudinal intra-writer variation within the framework of a language history 'from below'.

christine elsweiler is a Lecturer at Ludwig-Maximilians-Universität München. In 2019 she completed her habilitation project entitled 'From Shared Meaning to Divergent Pragmatics: A Comparative Study of the Modal Auxiliaries *May*, *Can*, *Shall* and *Will* in Scottish and English Letters (1500–1700)'. Her research covers language variation and change in different historical periods of English, from early Middle English up to Late Modern English. She specializes in Scots and its historical stages, also from a comparative perspective with other varieties of English, focusing in particular on the intersection between Historical Pragmatics and Historical Sociolinguistics.

tamara garcía-vidal is a Lecturer at the University of Murcia, where she teaches English language, historical linguistics and sociolinguistics. She received a BA and a PhD from the University of Murcia. Her main research interests are in the history of the English language, historical sociolinguistics and corpus linguistics. She has published articles in leading journals such as *Studia Anglica Posnaniensia*, *Folia Linguistica Historica* and *Journal of Historical Sociolinguistics*.

anne-christine gardner is a Senior Researcher at the University of Lausanne in the SNSF-funded research project 'The Language of the Labouring Poor in Late Modern England'. Her research interests include qualitative and quantitative approaches to (historical) sociolinguistics, dialectology and word-formation, with a focus on the uncovering of unheard voices, speaker choice and identity construction in the past. Her publications include a monograph on *Derivation in Middle English. Regional and Text Type Variation* (2014), articles on lexical change in Old and Middle English, as well as articles on self-corrections and identity construction in Late Modern English handwritten ego-documents.

veronika girininkaitė works in the Manuscript Division of the University Library of Vilnius University (Lithuania). She studied at Vilnius University and defended her thesis in 2019 in the field of Historical Sociolinguistics on the basis of qualitative research of an idiolect, *Kalbų ir kultūrų sąveika keliakalbio idiolekte. Atvejo analizė: Vytauto*  *Civinskio Dienoraštis (1904–1910)* [Interaction of languages and cultures in the idiolect of a multilingual: A case study of 'The Diary' (1904–1910) by Vytautas Civinskis (Witold Cywiński)], supervised by Ala Lichaciova. She works on various aspects of manuscripts in Polish, Russian and other languages created between the eighteenth and the twentieth century.

katharina gunkler-frank is a Research Assistant at Friedrich-Alexander-Universität Erlangen-Nürnberg. Her main research interests are Historical Sociolinguistics, syntax and text linguistics. She is part of the junior research group 'Flexible writers in language history' that compiles a corpus of historical patient texts from the nineteenth and early twentieth century. She is currently working on the linguistic flexibility of these patients in letters to different addressees regarding the syntactic level and conjunctions, but she has also investigated other aspects of these texts, for example, the orthographic level.

juan manuel hernández-campoy is Professor in Sociolinguistics at the University of Murcia, where he teaches English Sociolinguistics, Varieties of English, and the History of English as well as sociolinguistic research methods. Similarly, his research areas include sociolinguistics, dialectology, the history of English, and socio-stylistics, with special interest in language variation and change, where he has published books such as *Sociolinguistic Styles*, *Style-Shifting in Public* (with J. A. Cutillas-Espinosa), *The Handbook of Historical Sociolinguistics* (with J. C. Conde-Silvestre), *Diccionario de Sociolingüística* (with P. Trudgill), *Metodología de la Investigación Sociolingüística* (with M. Almeida), *Sociolinguistics and the History of English: Perspectives and Problems* (with J. C. Conde-Silvestre), and numerous articles in leading journals.

judith huber works in English Historical Linguistics at Ludwig-Maximilians-Universität München. Her research interest is variation and change in the history of English from a usage-based perspective, including syntax, lexicology, pragmatics and language contact. She has published a monograph on the expression of motion in the history of English (*Motion and the English Verb. A Diachronic Study*, Oxford University Press 2017) and various articles on issues in the history of English, such as Middle English BE/HAVE variation or the loss of the pronoun *thou*. She has recently edited a ZAA special issue on diachronic construction grammar (with Th. Herbst, 2022).

mark iten is a doctoral student in English Linguistics at the University of Lausanne in the SNSF-funded research project 'The Language of the Labouring Poor in Late Modern England'. He is interested in doing research in the fields of historical sociolinguistics and corpus linguistics. In the current research project, he is investigating pauper petition letters from a normative perspective, by comparing the lower-class language to the language of letters by the contemporary middling sorts and the elite, and by taking the socio-historical context into consideration.

yoko iyeiri is Professor of English Language at Kyoto University. Her research interests are in the areas of the history of English, historical sociolinguistics and corpus linguistics. She is the author of *Negative Constructions in Middle English* (Kyushu University Press, 2001) and *Verbs of Implicit Negation and their Complements in the History of English* (Benjamins, 2010), and co-editor with Mayumi Taguchi of *Pepysian Meditations on the Passion of Christ: Edited from Cambridge, Magdalene College, MS Pepys 2125* (Winter, 2019).

andreas krogull is a Postdoctoral Researcher at the Leiden University Centre for Linguistics, currently working on the NWO-funded research project 'Pardon my French? Dutch-French Language Contact in the Netherlands, 1500–1900'. His research interests include historical multilingualism and language contact, language standardization, and language policy and ideologies, particularly in the history of Dutch. He has published his research in journals like *Journal of Historical Sociolinguistics*, *Language Policy* and *Dutch Crossing*, as well as in several edited volumes.

laura linzmeier is a Research Associate at the Department of Romance Studies of the University of Regensburg. Her research focus is on minority and lesser used languages (esp. spoken in Sardinia) as well as contact- and (historical) sociolinguistics. She is a co-organizer of the CITAS-funded research network MS ISLA (Mediterranean Studies on Island Areas) at the University of Regensburg. Currently, she is interested in the effects of ports and ships on the processes of maritime communication and in the functional literacy linked to maritime communities in France during the seventeenth and eighteenth centuries.

theodore markopoulos is Assistant Professor of Linguistics at the University of Patras. He received his PhD on Historical Linguistics from the University of Cambridge (*The Future in Greek*, Oxford University Press, 2008), and has written extensively on a variety of issues related to the history of Greek, including the grammaticalization of modal and comparative constructions, as well as language contact in Late Medieval Greek. His main research interests are morphosyntactic change from a usage-based perspective, historical sociolinguistics (especially concerning variation and linguistic ideologies), language contact and the diachrony of Modern Greek dialects.

jill puttaert is a Postdoctoral Researcher at the Leiden University Centre for Linguistics. She is currently working on the NWO-funded project 'Pardon my French? Dutch-French Language Contact in the Netherlands, 1500–1900'. She has research interests in the history of Dutch, language history 'from below', language choice and language variation and change. Her doctoral thesis (funded by FWO at Vrije Universiteit Brussel) dealt with the written language use in letters by soldiers from the Napoleonic era and by paupers asking for financial support from their local authorities.

gijsbert rutten is Professor of Historical Sociolinguistics of Dutch at the Leiden University Centre for Linguistics. He directs the NWOfunded research project 'Pardon my French? Dutch-French language contact in the Netherlands, 1500–1900'. Recent publications include *Language Planning as Nation Building. Ideology, policy and implementation in the Netherlands, 1750–1850* (Benjamins, 2019), and *Revisiting*  *Haugen: Historical-sociolinguistic perspectives on standardization*, a special issue of *Language Policy* co-edited with Rik Vosters (2020).

markus schiegg works in German Linguistics at Friedrich-Alexander-Universität Erlangen-Nürnberg. He leads the junior research group 'Flexible writers in language history' that compiles a corpus of historical patient texts from the nineteenth and early twentieth century. His research focuses on historical sociolinguistics, language variation and change, and the history of German. He recently completed his habilitation project on intra-writer variation in historical patient letters, published as a monograph: *Flexible Schreiber in der Sprachgeschichte* (Winter, 2022). Another recent publication is a volume on *Intra-individual variation in language* (co-edited with Alexander Werth, Lars Bülow and Simone E. Pfenninger, de Gruyter 2021).

martina schmidl is a Postdoctoral Researcher at the University of Münster, Germany. She is currently working in the ERC-project 'Governance in Babylon: Negotiating the Rule of Three Empires' (grant agreement no. 101001619) under principal investigator Prof. Dr Kristin Kleber. She studied Assyriology and Religious Studies at the University of Vienna, Austria, and wrote her PhD thesis on Late Babylonian letters from two temple archives. Her research interests are the economic and social history of Babylonia in the first millennium BCE, Late Babylonian epistolography, historical sociolinguistics and the form and structure of literary texts as meaning-making devices.

gohar schnelle is a Research Assistant at the Department of History of the German Language at Humboldt-Universität zu Berlin. Her research focuses on historical sociolinguistics and historical corpus linguistics especially in Old High German and the Early New High German period. She is currently writing her PhD-thesis on registers in early Old High German. As a research supervisor, she is also in charge of two historical resources of German: the Old German Reference Corpus (OHG) and the Ridges-herbology-corpus (ENHG).

eleonora serra is a FWO Postdoctoral Fellow at Ghent University. Her research interests, which focus on the linguistic situation of sixteenth-century Italy, include the language of private letters, the relationship between prescription and usage, and the direction of change in social terms. She has recently begun a project on the use and social functions of epistolary formulae in Renaissance Tuscany. Her PhD, attained at the University of Cambridge, has explored dynamics of linguistic prestige in the context of sixteenth-century Italy, examining the reception of the codified variety across a range of social groups and contexts.

doris stolberg is a Senior Researcher at the Leibniz Institute for the German Language, where she heads a project on lexical-semantic change in extra-territorial varieties of German. Her research interests include language contact studies, historical sociolinguistics and (post)colonial linguistics. Among her main publications are: *Canadian heritage German across three generations: A diary-based study of language shift in action* (2019); *Changes between the lines. Diachronic contact phenomena in written Pennsylvania German* (2015); *A travel guides' view on a (post)colonial place: Tsingtau/Qingdao, China* (2021); and *The influence of German on the lexicon of Tok Pisin* (2017, co-authored with S. Engelberg).

silke unverzagt is a Research Assistant in project B04 'Register Emergence and Register Change in Germanic' of the CRC 1412 'Register: Language-User's Knowledge of Situational-Functional Variation' at Humboldt-Universität zu Berlin. She has a Master's Degree in Historical Linguistics from Humboldt-Universität zu Berlin and is currently working on Old High German texts. She wrote her master thesis on cliticization of the definite article and prepositions in Old High German.

oxel uribe-etxebarria is a doctoral student at the University of the Basque Country (UPV/EHU). He graduated in Basque Linguistics and Philology, and his doctoral research is a critical edition and a linguistic analysis of the work of an early nineteenth-century Basque author. His overall research interests include the history of Basque, language variation, historical dialectology and philology.

sarah van eyndhoven is a doctoral student at the University of Edinburgh. Her main research interests include eighteenth-century written Scots, historical correspondence, and the interactions between language, politics and identity among the ruling families of Scotland in the 1700s. Her previous publications include a socio-historical analysis of a Scots spelling change in the Helsinki Corpus of Older Scots, and a quantitative sociolinguistic comparison of Scots language choices between the literate Scottish public of the eighteenth century and a number of politically active individuals. Her main project focuses on building a corpus of early eighteenth-century Scottish letters.

christine wallis is Wissenschaftliche Mitarbeiterin at Friedrich-Alexander-Universität Erlangen-Nürnberg. Her PhD thesis (University of Sheffield) investigated scribal behaviour in shaping the text of the Old English translation of Bede's *Historia ecclesiastica*, and she has worked in digital humanities for the *Unlocking the Mary Hamilton Papers* and *Eighteenth Century English Phonology database* projects. Her current work focuses on language variation and change in historical English, examining how wider notions of linguistic norms or language standards impact on individual writers' training and output. Her research and teaching interests lie in the areas of historical sociolinguistics, philology, manuscript studies, textual editing and digital humanities.

## Index

accommodation 8, 13, 85–101, 107–29, 181–97, 249–67, 347, 385– 404, 448 address term 41, 46, 95, 184 addressee 39–53, 59–81, 85–101, 107–29, 152, 161, 171–72, 175–76, 181–97, 201–19, 229, 236–37, 240–43, 249–67, 271–91, 317–34, 425, 432, 434, 448, 491–509 adjective comparison 107–29 age 6, 13, 18, 59–81, 86, 95–96, 101, 187, 295, 301, 327, 339–56, 359–78 agency (of individual speakers) 9, 17, 290, 531–32 apparent time 18 appellation (*see* address term) attention to speech/attention to writing 8, 14, 99, 117, 188, 201– 19, 242, 265, 281, 349, 486 audience design theory 8, 12, 16, 96–101, 108–09, 188–94, 218, 223, 229, 236–37, 243, 250, 273, 280, 385– 404, 454 auditor 21, 100, 189, 194, 197, 393, 397– 98, 403 Babylonian 39–53 Basque 431–48 Berkshire 133–53 Bible translations 513–33 bilingualism 328, 344, 351–52, 355, 359– 78, 387, 402, 432 borrowing 112, 175, 320, 359–78, 440,

443, 447 *Briefsteller* (*see* manual) business writing 59–81, 99, 191–92, 407–26 chancery style/chancery documents 201–19, 259, 261, 266 clitics 223–43 code fluctuation 15 code-switching 13, 18, 241, 299, 323, 329– 30, 339–56, 359–78 codification 15, 64, 95, 134, 296, 413, 435, 532 Communication Accommodation Theory (*see* accommodation) community 1, 3, 8–10, 16–18, 78, 86–87, 91, 101, 108–09, 112, 118, 133–34, 152, 224, 312–13, 318, 321, 333, 358–78, 387, 402, 417, 421, 436, 491, 514 comparative adjectives (*see* adjective comparison) complexity 66, 80, 107–29, 167, 175, 202, 205–06, 209–19 concentration 4, 11, 14 constrained selection 451–70 contact 111, 249–378, 387, 402 convention 9–10, 47, 59–81, 181–97, 201–19, 242, 276–77, 312, 353, 407–26, 448 convergence 117, 188–89 copy/copying 11, 18, 165, 186, 189, 299, 301, 412, 451–70, 473–89, 497 correction (*see* self-correction) cyrillic (*see* script, cyrillic)

deferential devices 182, 184, 193, 195, 196

developmental variation (*see* variation, developmental) dialect 7, 89, 142, 205, 225–26, 232, 241, 252, 254, 259, 295–314, 342, 349, 353, 432–37, 446–48, 451–70, 476–78, 481, 514 dialectology 1, 17, 475–76, 488 diary 15, 18, 318–19, 321, 339–56, 359–78, 408, 411, 416 diglossia 387, 395, 403, 432, 437, 448 direct request strategies (*see* request) discourse-ending formulae (*see* formulae) discourse tradition 407–26, 514 divergence 188, 219, 281–82, 287–88 Dorset 133–53 draft 66, 157–77, 340 Dutch 299, 312, 317–34

education 16, 60, 62–64, 133–53, 162– 65, 235, 266, 295–96, 299, 304, 312, 314, 327, 362, 404, 431–48, 494, 507

emotion 42, 48–52, 65, 202, 235, 241–43, 342, 351–52


epistolary formulae (*see* formulae)

exemplar 18, 451–70, 473–89

expert writing 416, 422 external/social factors (*see* variation, functionalized)

Field-Tenor-Mode model 491–509 focused variety 458 formality (*see* variation, functionalized) formulae 14–15, 45, 59–81, 182, 185, 193– 94, 202–03, 206, 216, 218–19,

260, 287, 368, 385–404, 411, 416– 17, 421–22, 424 formulary (*see* manual) French 89, 111–12, 120, 203–04, 250, 252, 317–34, 340, 345–46, 348, 355, 374, 407–26, 432, 439, 444 Gallo-Italic 249–67 gender 6–7, 207, 227–28, 235, 242, 322, 324–26, 332, 334, 344 German 5, 12, 14, 15, 201–19, 223–43, 340, 345–47, 349, 351–53, 355, 359–78, 491–509 German cursive script (*see* script, German cursive) Greek 385–404, 515, 518–20 guide (*see* manual) Guipuscoan 431–48

*h*-deletion/dropping/insertion 134–35, 141–46, 151–53, 467–69 hierarchy 39–53, 162, 201–19, 327, 424, 494–95, 502 historical paradox 16 hypercorrection 100, 142–44, 152, 263– 64, 466

identity/identity construction 7–8, 10, 12–13, 42, 45, 53, 59, 80, 157–77, 181, 188–89, 202, 218–19, 224, 241–43, 258, 271–91, 333, 343, 344, 355, 374, 387, 396, 398, 426, 454, 514, 531 ideology 7, 9, 271–91, 532 idiolect 254, 295, 339–56, 371 imperative 46, 50, 184, 195, 491–509 indexical field 9, 532 indexical meaning/variants 10, 108, 513–33 indirect request strategies (*see* request)

inexperienced writers 14–15, 56–81, 133– 53, 205, 223–43, 353 informational maximalism 16 instruction 408, 410–12, 434, 446, 491–509 irregular preterites 305–07, 313 Italian 59–81, 203, 209, 249–67, 386–89, 394, 402–403

language change 2–3, 7, 10, 13, 15–18, 85–101, 107, 135, 204, 207, 225, 318–20, 333, 360, 458, 468–69, 474, 477, 479, 488–89, 518 from above 90, 432, 448, 483, 488 from below 483, 488 in the lifetime 13, 17, 19, 59–82, 85–101, 147, 151, 194–95, 197, 204, 207, 254, 295–314, 339–56, 359–78, 389, 431–48 language choice 317–34, 339–56, 359–78 language dominance 319, 361–62, 365, 369, 371, 373, 377–78, 432, 493, 508 language history 'from below' 12, 295 language mixing (*see Mischsprache*) Latin 89, 111–12, 185, 251–53, 257, 266, 340–41, 346, 352–53, 365, 431, 436, 440–43, 447–48, 493, 498, 502–05, 508–09, 529 lenition 257, 260–61 letter-writing guide (*see* manual) lifespan change (*see* language change, in the lifetime) list 258, 260, 265, 297, 369, 415–17, 419, 457, 462–63 literacy/illiteracy 65, 90, 100, 137, 140, 205, 296, 343, 363, 387, 407, 432– 33, 436–38, 446 literary style 60, 64, 257, 264, 432, 435– 36, 448, 493, 513–33 Lithuanian 339–56

loan (*see* borrowing) logbook 407–26 long -s 133–53 longitudinal variation (*see* language change in the lifetime) L2 learning 5, 256, 265, 352–54

mandatory forms 5, 225 manual 60, 63, 133, 165, 186–87, 196, 202, 204, 219, 298, 302, 304–05, 310, 313, 409, 413, 417 maritime writer 407–26 merchant 63, 74, 209, 249–67, 411 Mercian 451–70 migration 13, 15, 18, 295–314, 319, 355, 361–63 Milanese 249–67 *Mischsprache* 360, 453, 461, 476–77, 482 modal verb 193, 399–403, 492, 495, 499–508 montage 352 morphosyntactic variation 320, 343, 385– 404, 491–509 multilingualism 251, 317, 319, 321, 323, 339–56, 387, 403, 493

navigation journal 407–26 network 61, 65, 73, 76, 79–80, 93, 273, 289, 412, 424–26 non-standard 63, 141, 205, 233–35, 295– 314, 514 norm 8, 10, 14, 15, 89–90, 116, 134, 143, 151–53, 182, 187, 189, 197, 203, 205, 218, 281, 297, 342, 347, 351, 359, 421–22, 424, 432, 439–40 notary book 385–404 oath 42, 47–49

orthography 14–15, 17, 61, 69, 80, 85– 102, 135, 143, 147, 149, 151, 206, 209, 263, 266, 276–78, 297–98,

341–44, 347–49, 353–54, 374, 432, 431–48, 469–70, 529 other-language item (*see* language choice, code-switching) otiose stroke 475, 484–87 passive repertoire 454, 460, 464, 469 pauper petition 133–53 person reference 360–61, 364–65, 374–78 persona 10, 174, 196–97, 219, 273, 285, 290, 531 place names 262, 360, 364, 373–74, 376–77 poetic style 241, 356, 413, 513–33 Polish 339–56 politeness 41, 101, 188, 287, 421–22, 424– 26, 495 prescriptivism 134, 218, 297, 302, 304–05, 310–11, 313, 447 progressive translation 476–78, 482, 488 range of variation 4, 96, 117, 253 rank 47–48, 52, 59–81, 92, 94, 98–100, 107–29, 166, 181, 194, 196, 204, 208, 216, 219, 279, 285, 287, 296, 318, 320, 341, 399, 448 real time 13, 18 Referee Design 96–101, 189, 206, 218 region 1, 6–7, 90, 142, 146, 152–53, 205, 223–43, 252, 262, 276, 431–32, 436, 454, 459, 476, 514 register 4, 8, 65–66, 71–73, 76, 79–81, 96–101, 111, 124, 175, 206, 258, 281, 360–61, 363, 365, 370, 386, 394–95, 403, 432, 491–96, 508 relationship 41, 44–45, 48, 52–53, 60, 65, 66, 95, 101, 110, 112–13, 115, 117, 120, 124, 129, 160–61, 168–70, 176, 181, 208, 216–18, 229, 237, 240–41, 243, 254, 262, 272–73,

287, 301, 322–24, 327, 329–30, 332–34, 360, 375, 424, 448, 491–509 Renaissance 59–81, 249–67 request 14, 42, 46, 49, 52, 72, 152, 164, 168–69, 176, 181–97, 240, 287, 425, 492 rhetoric 42, 44, 48–49, 52, 111, 172, 176, 265, 289, 355, 513 rhotacism 259–60, 265 Roman script (*see* script, Roman) runic script (*see* script, runic) Russian 321, 339–56 Scots/Scottish English 181–97, 271– 91, 297–98 scribe 39–53, 398, 451–70, 473–89 script cyrillic 339–56 German cursive 359–78 Roman/Latin 359–78 runic 87–90 script choice 349–50, 364–65, 369– 70, 373–78 self-correction 14, 18, 145, 157–77, 366, 372, 375, 469 shift 13–14, 147, 310, 353, 362, 366, 375, 377–78, 388, 402, 432, 446– 47, 473–89 situational parameter 4–5, 8, 15, 95, 99, 116, 160, 164, 181, 187–88, 196, 343, 412, 492–94, 499, 509 social network (*see* network) social rank (*see* rank) social role, social role relationship 171, 182, 190, 196–97, 229, 426, 491–509 speaker accommodation (*see* accommodation) speaker design 9–10, 12, 42, 87, 160, 183, 188–90, 194–96, 218–19, 250

speech community (*see* community) spelling (*see also* orthography) 14–15, 69, 88–90, 145, 150, 152, 166–67, 170, 172, 206, 217, 231, 275–77, 283, 298, 312, 313, 348, 351, 353, 365–66, 368, 371–72, 431–48, 451–70, 473–89 stance 170, 272 standard 5, 14, 64, 69, 73, 95, 143–45, 205, 206, 225, 232, 235, 239, 250, 266, 272, 279, 281, 295–314, 348– 49, 353, 402, 432, 435, 439, 442, 447, 458, 468–69 standardization 15, 89, 135, 206, 271, 276, 297, 413 status (*see also* rank) 13, 48, 50, 52, 59–81, 85–101, 107–29, 152, 181, 187, 193, 197, 272, 279–80, 282, 287–88, 320, 353, 388, 448, 514 style axiom 99, 118 style shifting 3, 8, 13, 85, 87, 96, 98–99, 101, 108–10, 116–18, 124, 128, 189, 217–19, 407–26, 513–33 subjunctive 402, 491–509 Swedish 491–509 syncretism 296, 302–05, 313 syntactic constraints (on variation) 5, 110–11, 224, 238–39, 519, 521–27 Systemic-Functional-Linguistic framework 492, 494 text function 59, 65, 91, 191, 407–26, 492 text linguistics 407–26 text-structuring formulae (*see* formulae) third-wave of variation studies 8–9, 42,

108, 514, 531–32

```
translation 185, 203–04, 209, 218, 278–
        79, 351, 352, 372, 436, 446, 452, 
        457, 460, 462, 464, 497–98, 
        502–04, 508–09, 513–33
Tuscan 63, 76, 249–67
```

```
uniformitarian principle 15–16, 98, 
        124, 134
```
variation conditioned 4–5, 10, 14, 22, 88, 96, 108, 110, 112, 117–19, 238, 440 developmental (*see also* age) 364–66, 369, 378 distance/immediacy 12, 72, 202–06, 218, 223–24, 235–37, 240, 242– 43, 281, 283, 290, 343, 359, 377–78 functionalized 3, 5–7, 10, 14, 65, 118, 177, 240–43, 250, 320–21, 331, 345, 365, 368, 371–74, 378, 398, 495, 526, 528 free/non-conditioned 5–6, 10, 14–15, 153, 441 longitudinal (*see* language change, in the lifetime) social (*see* variation, functionalized) stylistic (*see* variation, functionalized) verbal -s 295–314 verb-initial 513–33

Welsh 513–33 West-Saxon 451–70 writing experience (*see also* inexperienced writers) 14–15, 223–43, 451–60 writing guide (*see* manual)

Yiddish 340, 346, 351, 355

# Historical Sociolinguistics

*Studies on Language and Society in the Past*

## series editors

*Nils Langer Stephan Elspaß Joseph Salmons Wim Vandenbussche*

The interdisciplinary field of Historical Sociolinguistics seeks to reveal the impact of language development on society and the role of individuals and society in the changing forms and usage of language. This book series is aimed at sociolinguists and social historians who are keen to publish studies on the social history of languages, the interaction of linguistic practices and society, and the sociological significance of linguistic variation with a historical dimension. The purpose of the series is to provide empirically supported studies that will challenge and advance current language historiographies, which often continue to present the history of particular languages as necessarily leading to the creation of a standard or prestige variety. Of particular interest are topics such as the following: language myths and language ideology, historical multilingualism and the formation of nationstates, the sociolinguistics of minority and regional languages, the rise of urban vernaculars, immigrants and their languages, the role of prescriptive grammarians, and the social history of pidgins and creoles.

Book proposals from historians and linguists working on any language in any period are welcome, in particular those that include a comparative dimension as well as those with a strong empirical foundation. The language of publication is primarily English, though other languages may be considered. The editors guarantee that all publications in this series have been subjected to external and anonymous peer review. The four series editors and twenty-six members of the advisory board are all members of the Historical Sociolinguistics Network (HiSoN).

## advisory board

Anita Auer (Lausanne), Wendy Ayres-Bennett (Cambridge), Andrea Cuomo (Ghent), Steffan Davies (Bristol), Ana Deumert (Cape Town), José del Valle (CUNY), Martin Durrell (Manchester), Jan Fellerer (Oxford), Elin Fredsted (Flensburg), Róisín Healy (Galway), Juan Hernandez-Campoy (Murcia), Kristine Horner (Sheffield), Ernst Håkon Jahr (Agder), Mark Lauersdorf (Kentucky), Anthony Lodge (St Andrews), Nicola McLelland (Nottingham), Miriam Meyerhoff (Oxford), Agnete Nesse (Bergen), Terttu Nevalainen (Helsinki), Taru Nordlund (Helsinki), Gijsbert Rutten (Leiden), Joachim Scharloth (Waseda-Tokio), Peter Trudgill (Agder), Marijke van der Wal (Leiden), Rik Vosters (Vrije Universiteit Brussels), Laura Wright (Cambridge)

