Alexandra N. Lenz/ Mateusz Maselko (eds.)

# **VARIATIONist Linguistics meets CONTACT Linguistics**

# **Vienna University Press**

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Wiener Arbeiten zur Linguistik

Band 6

Herausgegeben von Alexandra N. Lenz, Melanie Malzahn, Eva-Maria Remberger und Nikolaus Ritt

Advisory Board: Peter Auer, Universität Freiburg, Deutschland Ina Bornkessel-Schlesewsky, Universität South-Australia,Australien Olga Fischer, Universität Amsterdam, Niederlande Junko Ito, UC Santa Cruz, USA Hans Kamp, Universität Stuttgart, Deutschland Johanna Laakso, Universität Wien, Österreich Michele Loporcaro, Universität Zürich, Schweiz Jim McCloskey, UC Santa Cruz, USA John Nerbonne, Universität Groningen, Niederlande Peter Trudgill, Universität Agder, Norwegen

Alexandra N. Lenz / Mateusz Maselko (eds.)

# VARIATIONist Linguistics meets CONTACT Linguistics

In cooperation with Manuela Lanwermeyer

With 101 figures

V&R unipress

Vienna University Press

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über https://dnb.de abrufbar.

#### **Veröffentlichungen der Vienna University Press erscheinen bei V&R unipress.**

Gedruckt mit freundlicher Unterstützung des Rektorats der Universität Wien, der Eidgenössischen Technischen Hochschule Zürich – Congressi Stefano Franscini – des österreichischen Wissenschaftsfonds (FWF) sowie der Österreichischen Akademie der Wissenschaften (ÖAW).

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Theaterstraße 13, D-37073 Göttingen Dieses Werk ist als Open-Access-Publikation im Sinne der Creative-Commons-Lizenz BY International 4.0 ("Namensnennung") unter dem DOI 10.14220/9783737011440 abzurufen. Um eine Kopie dieser Lizenz zu sehen, besuchen Sie https://creativecommons.org/licenses/by/4.0/. Jede Verwertung in anderen als den durch diese Lizenz zugelassenen Fällen bedarf der vorherigen schriftlichen Einwilligung des Verlages.

Umschlagabbildung: Beautiful row of vintage postal mailboxes in Fuerteventura (Spain), iStock

#### **Vandenhoeck & Ruprecht Verlage | www.vandenhoeck-ruprecht-verlage.com**

ISSN 2365-7731 ISBN 978-3-7370-1144-0

# Contents


# Historical Contact Scenarios


# When Variation(ist Linguistics) Meets Contact (Linguistics). To This Volume

# Introduction

This volume focuses on the inherently heterogeneous nature of language(s) as seen from the perspective of variationist linguistics and contact linguistics, which became established and internationally recognized sub-disciplines of (socio-) linguistics during the latter half of the 20th century. Over the last few years, each paradigm has considerably broadened the spectrum of the topics under investigation, however there has not yet been an extensive and satisfactory exchange between the two scientific fields.

An inter(sub)disciplinary meeting in the Italian-speaking Swiss Canton of Ticino, organized by the University of Geneva and Austrian Academy of Sciences, was expected to contribute to an amendment of these unfavorable circumstances by assembling young and experienced researchers from the areas of variationist and contact linguistics and by creating a platform for discursive exchange and future cooperation between these two linguistic fields.

This volume is one of the scientific impacts resulting from the homonymous International Conference "VARIATIONist Linguistics meets CONTACT Linguistics" which took place at the Congress Center Monte Verità in Ascona from May 20–23, 2018 (for more details, see the conference report by Pamela Goryczka and Virginie Gremaud in this volume). Selected (and peer-reviewed) papers published in the present book—for the most part—draw upon talks held during the conference.

# Areas of Research

Both variationist linguistics and contact linguistics are highly interdisciplinary research fields. They have been programmatically influenced by and have inspired other linguistic sub-disciplines (including: quantitative linguistics, corpus linguistics, historical linguistics, dialectology, sociolinguistics, ethnolinguistics or so-called cultural linguistics, linguistic typology, psycholinguistics, perceptual linguistics, gender linguistics, lexicography, pragmatics, contrastive linguistics) and neighboring disciplines (including: sociology, psychology, geography, history, anthropology, colonial studies, theology, literary studies, political and legal sciences). Although contact and variationist linguistics are closely related to other scientific areas, they clearly address their own different linguistic aspects and employ distinct methods.

Contact linguistics investigates the use of languages, which—like their speakers—are socially in contact with each other. This discipline is explored through different stages—varying according to phenomena, area, and period of the language contact process, including its inception, evolution, and results. At the same time, it considers the intralinguistic and extralinguistic parameters and control factors that determine the level of intensity of interaction and, therefore, of potential influence. Language contact occurs in a large number of phenomena, including: language convergence, borrowing, relexification, hybrid structures development, adoption, as well as degradation, and leveling of morphological and grammatical features, code-switching, stratal influence, and language shift. It can lead to the emergence of new varieties (such as pidgins, creoles, mixedlanguage varieties, interdialects—especially so-called language 'island' varieties' or linguistic 'enclaves', minority languages, ethnolects, guest workers' varieties, xenolects) or even fully-developed languages. Not least of all, a wide range of interests in language contact studies includes ('external') multilingualism—be that in the context of language border areas, non-monolingual communities, migration, or the workplace (cf. Bakker / Matras 2013; Darquennes / Salmons / Vandenbussche 2019; Hickey 2010; Hogan-Brun / O'Rourke 2019; Matras 2009; Myers-Scotton 2002; Plewnia / Riehl 2018; Riehl 2012; Thomason / Kaufman 1988). The main focus of contact linguistics is on exploring the interaction between different languages, which has caused this linguistics sub-discipline to be referred to as interlingual contact studies. In contrast, the mutual influence of various varieties representing the same language—so-called intralingual contact —is only marginally considered (cf. Kerswill 1994; Trudgill 1986; Wilson 2019).

Intralingual heterogeneity in its various forms is the main topic of variationist linguistics, which focuses on microstructural phenomena within the same language or within a variety (cf. Chambers 2003; Chambers / Schilling 2013; Felder 2016; Romaine 2000; Sinner 2014; Tagliamonte 2012). While research historically focused on the phonetic-phonological level, much attention is now being paid to higher linguistic levels, particularly to morphology and syntax, but also structures above sentence level, i. e. (the whole) text (see e. g., Adamzik 2016; Cornips / Corrigan 2005; Lenz / Patocka 2016; Schubert / Sanchez-Stockhammer 2016; as well as the thematic volume based on the previous "VARIATIONist linguistics meets …"-conference Adamzik / Maselko 2018).

Variationist linguistics focuses primarily on the research of coexisting variants of variables ('alternative linguistic characteristics')—be they phonological, grammatical, or lexical—within one single language. Aspects of typological or language contrastive nature that go above and beyond single languages are seldom taken into account. If we currently consider variationist linguistics as a single research field, it is the result of a merging of different special (sub-) disciplines. Of these, two should be mentioned in particular: first, dialectology with its traditional focus on local vernaculars; and second, sociolinguistics developed in the 1960s (cf. e. g., Labov 1966). The latter, after first focusing on language usage specific to different social classes, turned its interests to nonstandard varieties in general and then later to (spoken and also written) standard varieties as well. At present, variationist linguistics focuses on variation in all its dimensions. Variationist linguists take the whole variety spectrum into consideration, which ranges from small-scale (enclave) dialects to (regional) nonstandards and '(supra-)national' standard varieties (cf. e. g., Ammon et al. 2016; Dürscheid / Schneider 2019; Hickey 2012; Schmidt / Herrgen 2011). Instead of regarding the relation between sociodemographic parameters (such as age, gender, education, professional occupation, mobility, and political orientation), the so-called 'third wave sociolinguistics' emphasizes the linguistic creativity of speakers using the features of their repertoire as socio-symbolic entities according to their individual needs (cf. Bucholtz 2011; Coupland 2007; Eckert 2010; Eckert / Rickford 2001; Hernández-Campoy 2016; Tagliamonte 2016).

Besides the individual foci of variationist linguistics and contact linguistics, the two research paradigms show shared central characteristics (cf. Britain 2010; Britain 2018; Kerswill 2010; Léglise / Chamoreau 2013; Lenz 2016; Pfenninger et al. 2014; Ravindranath 2015; Schreier 2019): A common feature can be found in the emphasis on nonstandard language use. Both sub-disciplines focus on (vernacular) languages or varieties that are, in most cases, not codified and missing a homogenous and common orthographical writing system. This means that both variationist linguistics and contact linguistics are particularly, although not exclusively, interested in descriptive, usage-based approaches rather than in prescriptive, standards-based issues. A further commonality is the fact that both disciplines deal with co-existing linguistic alternatives, i. e., with variants that exist intralingually and interlingually. However, the boundaries between varieties and languages, and therefore between intralingual and interlingual variation, are fluid. While contact linguistics focuses on the relation between distinct languages, variationist linguistics illuminates the dynamics of varieties in contact with each other (cf. e. g., Auer et al. 2005). Despite these, and other parallels between the disciplines, we are convinced that there is a strong need for a transdisciplinary approach in future research.

#### Content

The present volume aims at giving an insight into the complex synergy between occurring linguistic contact constellation and variation in the parlance. A total of thirteen papers illustrate examples from diverse (linguistic) angles (e. g., phonetics and phonology, morphology, syntax, as well as perception, language policy, dialectology, neurolinguistics) discussing the following three topics: Firstly, what different forms of contact situation can (be) develop(ed); secondly, how coexistence of two or more languages (or their varieties) can favour the variation within them; and thirdly, how this interplay and mutual dependence can promote processes of linguistic change.

Taking into account the nature of the contact constellation, we can divide the articles into two major thematic areas. In order to create more clarity about the key topics of the published studies and their connections, they become subdivided according to the area and/or time criterion in the second step.

The first main part of the publication (I) deals with interlingual contact, i. e., the contact between different languages (which represent diverse language families). To begin, we focus on extraterritorial varieties of German and Norwegian and their contact languages, in particular (national/dominant) languages in the territories settled once by Europeans. Katharina Dück examines the language contact situation and the language attitudes of the Caucasian Germans living in the South Caucasus (and partly in Germany after remigration). The socalled Russian German also becomes an object of investigation in the contrastive study by Peter Rosenberg, who pursues the question of varying degrees of decomposition and simplification in terms of morphology. The language situation in Russian multilingual communities is compared to the situation in Brazil, which is again the focal point of interest in Mateusz Maselko's paper. Here, attention is focused on the grammaticalized verb give, which can be used in both the German 'island' dialect, Riograndese Hunsrik, and the contact language, Brazilian Portuguese, as an existential verb and a copula 'to become'. The last article in this section, written by Janne Bondi Johannessen, also deals with a variety spoken on the American continent, namely the Norwegian heritage language, and is dedicated to another syntactic category, the word order in subordinate clauses. The next section consists of two studies examining historical language contact constellations. Multilingual (writing) practices in the Ukrainian city Lemberg in the late 19th century become the subject of the article by Stefaniya Ptashnyk, in which language policy aspects gain importance. Jelena Zˇivojinovic´ takes interest in phonological and morphological issues of the language contact in Renaissance Ragusa, an aristocratic maritime republic centered on the city of Dubrovnik in Dalmatia, located today in southernmost Croatia.

The emphasis of the second main part (II) is on the intralingual contact, i. e. the contact between various dialects or varieties within the same language. The questions addressed in the initial section concern German (varieties) spoken in Austria. The paper by Johanna Fanta-Jende investigates the areal and social dimension of synchronic variation for a selected phonological variable, the Middle High German /ei/. Lars Bülowand Dominik Wallner discuss variation and change in structural aspects of the verb to be in Salzburg's dialects and examine its plural forms. Jutta Ransmayr's article concerns how teachers and students talk during German lessons and to what degree code-switching and code-mixing takes place in the Austrian classroom. A similar linguistic problem is pursued by Lisa Krammer's paper, which explores the attitudes of Viennese students towards the oral variational usage of the German language at universities in the context of the lecture. Perceptual aspects are also central for Wolfgang Koppensteiner and Agnes Kim, who pay close attention to layman attitudes towards language change and contact, as well as language ideologies, on German in Austria. The second part of the volume is supplemented by two studies scrutinizing German varieties spoken in Germany. Verena Sauer discusses the former inner-German border and examines dialectal dynamics in the border areas of South Bavaria and North Thuringia, as well as the impact of the extra-linguistic borders on the perception of inhabitants. The neurolinguistically and phonologically oriented article by Manuela Lanwermeyer explores the impact of dialect exposure on phoneme processing and provides evidence for cross-dialectal comprehension difficulties between speakers from two German dialects, the Bavarian-Alemannic transition zone, and Central Bavarian.

# Acknowledgments

The (in several respects) highly successful conference "VARIATIONist Linguistics meets CONTACT Linguistics" on the Monte Verità (Switzerland) gave us, the undermentioned Organizing Committee, an incentive for the current publication.

The conference was sponsored by:


The present volume is published with kind support of:

– Austrian Science Fund (FWF)

– Swiss Federal Institute of Technology in Zurich – Congressi Stefano Franscini


We sincerely thank all sponsors for their generous support.

We would also like to extend thanks to our extraordinary conference assistants, Pamela Goryczka (Vienna) and Virginie Gremaud (Geneva), as well as our special support and conference supervisor, Stefanie M. Moog.

Proposals for presentations (talks and posters) held, as well as the articles published in this thematic volume, were subjected to a double-blind peer-review process and evaluated by two specialists in the individual field. We would like to express our deep gratitude to the member of the Scientific Committee who provided valuable insight and expertise that significantly improved the conference and this publication:


Needless to say, our thanks are also extended to all authors of the published articles as well as all conference participants who enriched this scientific meeting with their multi-faceted, interesting and inspiring presentations, on the one hand, and contributed to productive discussions and thereby to the success of this volume, on the other hand.

Finally, we would like to show our gratitude to the editors, Advisory Board, and publisher of the publication series "Vienna Studies in Linguistics" (in German: "Wiener Arbeiten zur Linguistik") for the opportunity to present the highly topical results of important (international) studies on variationist and contact linguistics to the broader audience.

Alexandra N. Lenz & Mateusz Maselko Vienna (Austria) & Geneva (Switzerland), May 2020

# (Selected) Bibliography (on Variationist Linguistics and Contact Linguistics)


© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

#### Pamela Goryczka / Virginie Gremaud

# Report of the Conference "VARIATIONist Linguistics meets CONTACT Linguistics" (Ascona 2018)

The international conference (CSF Workshop) "VARIATIONist Linguistics meets CONTACT Linguistics" was held on May 20 to 23, 2018 at the Conference Center Monte Verità in Ascona (Switzerland). It was the second instalment of the conference series "VARIATIONist Linguistics meets …", whose first edition in the previous year had already proven to be an overwhelming success in bringing together researchers of two different sub-disciplines of linguistics, variationist linguistics and text linguistics. Adopting an interdisciplinary approach facilitated the discussion of overlapping research interests, as well as contrasting different methods employed by the sub-fields. Drawing on the challenging yet yielding experience, the Organizing Committee of this year's event, Alexandra N. Lenz (University of Vienna) and Mateusz Maselko (University of Geneva), decided to continue the exchange of ideas between researchers of diverse disciplines of linguistics, this time focusing primarily on the question of linguistic variation induced by language contact. In order to fully meet the demands of the heterogeneous and dynamic nature of language(s) and to offer a broad perspective on various different languages, the contributions were chosen in the run-up to the conference by a Scientific Committee, comprised of a numerous international experts. In general, the conference profited from its 27 participants who came from as many as ten countries (and three different continents).

On the day of arrival, a brief welcoming speech by the Organizing Committee was followed by a full-course dinner in the hotel's restaurant. Afterwards, an informal gathering took place in order to provide the basis for getting acquainted with fellow colleagues and their research interests. In the interactive introduction round, the participants were asked to engage in conversation with as many other researchers as possible, covering key questions such as "How many (and what) languages do you speak?" and "How many varieties of these languages do you speak?". At the end, in groups of two, the participants were asked to introduce their last respective conversation partner and offer a guess towards which pole of the two featured linguistic sub-disciplines their research interests were more inclined. Finally, every participant placed their name on a poster depicting the two linguistic poles, based on their own affiliation to one or the other field (or somewhere in between). This rather informal, albeit effective, approach offered a quick and interesting overview into the different research interests of all contributors before the official conference program had begun.

The conference itself kicked off with a joint introductory lecture by Alexandra N. Lenz and Mateusz Maselko. Initially, Alexandra N. Lenz outlined general observations regarding differences and potential common ground between the research interests, as well as methods of variationist linguistics and contact linguistics. To support this theoretical input, the focus of the talk then turned to the presentation of empirical data concerning lexical and syntactic phenomena of variation from the ongoing research project "Deutsch in Österreich. Variation – Kontakt – Perzeption" 'German in Austria. Variation – Contact – Perception' which tackles German and its varieties spoken in Austria. Mateusz Maselko completed the talk by focusing on the syntactic features of a specific German variety spoken in parts of Brazil, the so-called Riograndese Hunsrik, discussing its relevance in the context of the two featured linguistic approaches. Afterwards, the conference proceeded with talks by Dirk Frans Pijpops, who argued that language contact could cause lectal variation in regard to Belgian speakers using Netherlandic lexemes with the respective Netherlandic morphological variant and vice versa. Nour Efrat-Kowalsky, who presented an analysis of the assignment of gender in Dutch, then noted that variation is not solely a product of language contact. In the following combined discussion of the two talks, the contributors seized the opportunity to reinforce their hypotheses.

In the afternoon, a poster session gave ten researchers the opportunity to display and present their projects, which concerned a wide range of topics, such as more formal aspects of potential contact-induced change (Maia Duguine, Manuel Padilla-Moyano, Monja Burkhard), the societal impact on language loss or change (Zafeer Hussain Kiani, Veronika Milanova), observations regarding language attitudes in general (Lisa Krammer, Verena Sauer), neurolinguistic approaches as a way to explain language production and perception (Manuela Lanwermeyer), an insight into historical variationist linguistics (Fabian Fleißner) and variation between (standard) language and dialects (Pamela Goryczka).

Following the poster session, the next two presentations by Stefaniya Ptashnyk and Jelena Zˇivojinovic´ addressed contact-induced language change from a historical perspective. Stefaniya Ptashnyk outlined her analysis of 19th century nonfictional texts written in the linguistically highly diverse city Lviv and Jelena Zˇivojinovic´ presented the results of a synchronic study based on a literary corpus depicting the unequal use of language(s) used in Ragusan Renaissance. Subsequently, a discussion of the two presentations led to even more insightful observations on historical multilingualism and potential code-switching.

The day's closing plenary lecture was given by Péter Maitz, whose overview of the German-based creole language Unserdeutsch, a linguistic result of the German colonial period in the South Seas, showed that more in-depth analyses concerning contact-induced variation and language change is still required to make up for the lack of research interest to date.

The second day of the conference began with a plenary lecture by Janne Bondi Johannessen, who discussed methodological issues in obtaining data for the Nordic Syntax Database and also presented the main results of a study focusing on word order peculiarities in the Scandinavian heritage languages that have developed due to large immigration waves from Nordic countries to North America in 1825–1925.

Following the plenary lecture, the two presentations by Johanna Fanta-Jende and Lars Bülow both focused on the vertical and horizontal dimensions of variation in Austria. In her talk, Johanna Fanta-Jende presented results from a study on the variation of phonetic and phonological phenomena, resulting, as was shown from a variety of factors, most prominently from the intersituative language behaviour patterns of specific speakers with varying sociodemographic backgrounds. Lars Bülow presented empirical evidence for the intra- and interindividual variation of the plural forms of the German verb sein 'to be' in the Central/South Bavarian transition area highlighting the importance of contact variation not only as a result of areal-horizontal factors but also of social-vertical variables.

After a quick lunch break, the participants were invited to gather in small groups and seize the opportunity to do some sightseeing in the city centre of Ascona. With renewed vigour, the conference continued in the afternoon with talks by Jutta Ransmayr and Christian Efing.

Jutta Ransmayr outlined results of a research project conducted on code switching (in terms of standard language, colloquial and dialectal varieties) in Austrian schools, which could be used as a starting point for the development of adequate teaching materials and recommendations on how to address code switching in class. In his presentation, Christian Efing stressed the importance of language contact for the emergence of sociolects, such as the German Rotwelschdialects. In addition, differences between these sociolects and other ethnolects were discussed with regard to their pragmatic functions as well as the possibility to act as common denominators in creating community identities.

The second conference day ended with two presentations about a specific German variety spoken in Namibia, the so-called Namdeutsch. While Christian Zimmer presented results from a study concerning the perception of Namdeutsch regarding the (supposedly) more prestigious standard German, Birte Kellermeier-Rehbein offered new insights into lexical borrowing and morphosyntactic contact phenomena taking place between English, Afrikaans, Low German and Namdeutsch.

To conclude the busiest conference day, the conference dinner was held at the Restaurant Collinetta, situated on Lake Maggiore. The sophisticated atmosphere and impeccable service offered an exclusive dining experience that was enjoyed by all the participants.

The third and last day of the conference began with a plenary lecture by Peter Rosenberg who discussed all the main aspects of the conference—variation, contact and change—from the perspective of comparative language island research. In the next talk, Katharina Dück continued this line of research, presenting a study on the perception of Caucasian German both by Caucasus Germans living in the Caucasus region and those who have emigrated to Germany.

In the last presentation, by Justin Spencer Davidson, on a phonetic Spanish-Catalan contact phenomenon, it was shown that opting for a broader definition of contact-induced change could be useful to account for potential contactinduced variation.

The final discussion served as a creative think tank for unanswered questions, new ideas and proposals for future collaborations and meetings. In the accompanying awards ceremony the prizes for the best talk by a young researcher and the best poster presentation went to Johanna Fanta-Jende (University of Vienna) for her talk on "Social-vertical and areal-horizontal dimensions of 'inner multilingualism' in Austria" and to Fabian Fleißner (University of Vienna) for his poster entitled "Von der Rekonstruktionsphilologie zur historischen Variationslinguistik: Tempus- und Aspektwandel altgermanischer Dialekte". A Special Recognition Prize was awarded to Pamela Goryczka (University of Vienna) for her poster presenting her master's thesis with theworking title "Roboterbein oder dem Roboter sein Bein? – Adnominale Possessivität auf der Dialekt-Standard-Achse in ruralen Gebieten Österreichs".

From our perspective as participating student assistants, the meeting proved to be a wonderful opportunity to meet researchers from all over the world, learn about their projects and methodological approaches and realise that variationist linguistics and contact linguistics have a lot in common (one profiting from the other and vice versa). In the end, a really good conference turns out to be great when, after all the input you have received, you feel even more inspired to rush home and work on a (new) project. In that sense, we hope you had the pleasure to do so in the meantime. Rest assured that we are already looking forward to your contributions for the next instalment of "VARIATIONist Linguistics meets …". I Languages in Contact

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

Extraterritorial Varieties of German and Norwegian

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Katharina Dück

# Language Contact and Language Attitudes of Caucasian Germans in Today's Caucasus and Germany

Abstract: This article examines the language contact situation as well as the language attitudes of the Caucasian Germans, descendants of German-born inhabitants of the Russian Empire and the Soviet Union who emigrated in 1816/ 17 to areas of Transcaucasia. After deportations and migrations, the group of Caucasian Germans now consists of those who have since emigrated to Germany and those who still live in the South Caucasus. It's the first time that sociolinguistic methods have been used to record data from the generation who experienced living in the South Caucasus and in Germany as well as from two succeeding generations. Initial results will be presented below with a focus on the language contact constellations of German varieties aswell as on consequences of language contact and language repression, which both affect language attitudes.

Keywords: language contact, migration, variation, language attitudes, identity

Abstract: Im Zentrum der nachstehenden Betrachtungen stehen die Sprachkontaktsituation sowie die Spracheinstellungen der Kaukasiendeutschen, Nachfahren deutschstämmiger Einwohner des Russischen Reichs und der Sowjetunion, die 1816/17 in Gebiete Transkaukasiens ausgewandert sind. Nach Deportationen und Migrationen besteht die Gruppe der Kaukasiendeutschen heute aus denjenigen, die inzwischen nach Deutschland ausgewandert sind und denjenigen, die bis heute im Südkaukasus leben. Mithilfe soziolinguistischer Methoden wurden erstmalig Daten der im Südkaukasus und in Deutschland lebenden Erlebnisgeneration sowie zwei Nachkommengenerationen aufgenommen. Erste Ergebnisse werden nachstehend mit Schwerpunkt auf Sprachkontaktkonstellationen deutscher Varietäten sowie auf Folgen des Sprachkontakts und der Sprachrepression für die Spracheinstellungen vorgestellt.

Keywords: Sprachkontakt, Migration, Variation, Spracheinstellungen, Identität

#### 1 Introduction

For approximately 150 years (1817–1941), Swabians lived in the South Caucasus separated from the German-speaking world. Surrounded by other cultures and languages such as Georgian, Armenian, Azerbaijani, and Russian, they cultivated the culture and language they had brought with them from Germany. Some descendants of these Swabians live in Germany today, and others live in the former German settlements of Transcaucasia, where the current main languages of communication are Georgian, Azerbaijani or Armenian, depending on the local situation; additionally, Russian still plays an important role in everyday communication, especially among those who were born before the 1990s. However, both groups of descendants seem to have maintained a German memory culture and are constructing a transnational collective identity of the 'Caucasian Germans', which is linguistically linked to a variety of Swabian. This variety has been cultivated for over 200 years and shows an interesting language history with unique and astounding characteristics due to 150 years of persistently limited language contact with the outside world.

The following article provides an insight into current studies (e. g. Dück 2018) on the connection between the language and identity of Caucasian Germans living today and is part of the "German in the World" project at the Leibniz-Institute for the German Language in Mannheim (Germany). Its aim is to close the gaps in 'language island' and language contact research as well as in variational linguistics. In doing so, it seaks to determine the characteristic accumulations of the respective varieties caused by language contact, language skills, language attitudes, perceptions of the social environment, and the cultural situation of this minority group and put into context with the respective selfperceived and externally perceived identity. Of particular interest are the effects of the clash of the acquired 'conserved' Swabian variety with near Standard German and other languages as well as the effects on identity constructions. The following study presents initial results from this project with the help of newly collected language data from the speaker group, while focusing on the consequences of language contact for the Swabian variety and the language attitude of the informants.

#### 2 Historical Background and Analyzed Language Variety

The settlement areas of the Swabians in Russia and the former Soviet Union have been widely (and thoroughly) investigated (cf. e. g. Hoffmann 1905; Allmendinger 1989; Songhulaschwili 1997; Auch 2001; Haigis / Hummel 2002; Föll 2002; Springform 2004; Hertsch / Er 2017). Tsar Alexander I had continued the immigration policy of his grandmother Catherine I and again invited many foreigners to move to Russia. Due to his settlement policy, he promoted the settlement of the newly conquered southern Ukrainian territories. At the same time the settlement of Germans in Transcaucasia was enforced in order to protect the empire against the Turks in the south. In 1816 forty families followed Alexander I's invitation to the South Caucasus to flee economic hardship, political oppression and religious confrontations. In 1817, more than 1400 families followed (cf. Biedlingmeier 2005: 17)—mainly radical Pietists from Württemberg. They gathered in Ulm, crossed the Danube to the Black Sea, hibernated near Odessa and continued their journey in the countryside to the South Caucasus, where they founded the first German colony Marienfeld (today: Sartitschala) in early summer 1817, which is situated east of the capital of Georgia Tbilisi.

Figure 1: Emigration paths of Swabian radical pietists (figure based on <https://commons. wikimedia.org/wiki/File:Europe\_laea\_location\_map.svg>)

In the course of the following century more than twenty German settlements were established in Transcaucasia. This settlement development was interrupted during the Second World War, when the mother and daughter colonieswere dissolved in 1941 and their inhabitants were forcibly deported. Caucasian Germans who had entered into so-called intermarriages with Georgians, Armenians, Azerbaijanis, or other ethnic groupswere excluded from the deportations. In the deportation areas of Central Asia, the Caucasian Germans finally came into contact<sup>1</sup> with other so-

<sup>1</sup> Even before the deportation there were some contacts with Volga Germans, for example, when they came from the north to the Transcaucasian German villages in years of famine. Shortly before the deportations in August 1941, some Caucasian Germans also reported that Russians

called 'Russian Germans' 2 (cf. Berend / Riehl 2008: 22). Noawadays, Caucausian Germans are considered as Russian Germans.<sup>3</sup>

Thus, three different speaker groups of Caucasus Germans emerged, two of which spoke a Swabian variety. First, there are the Caucasus Swabians with the largest number of Caucasian Germans, who often settled in mixed Kazakh, Uzbek, Russian and German settlements after the deportations and migrated to Germany in the 1990s at the latest—partly to the villages and cities from which their ancestors had emigrated 200 years ago. In addition, there are those Caucasus Swabians who were exempt from forced deportations or were allowed to return to the Caucasian villages, which was only the case if they had a local partner.

Finally, the smallest faction of the Caucasian Germans are the descendants of the so-called 'Russian Germans', who had already settled in the oil metropolis of Baku as architects and engineers or in Tbilisi as merchants, pharmacists and doctors in the 18th and 19th centuries and mingled with the locals. Their descendants today speak a well-educated colloquial or near Standard German. This is probably due to the excellent, mostly academic education, to which the families attached great importance (cf. Note 5, on this subgroup of 'Russian-Germans').

The three groups also experience very different language contact constellations: The last described and smallest group has as L1 either a Standard German variety or Russian and as L2 either Russian oraStandard German variety, depending on what their parents learned first. Especially with regard to

brought Crimean Germans to the Caucasian villages to help with the harvest before everyone was finally deported to Central Asia at the end of the harvest.

<sup>2</sup> The term 'Russian Germans', which is commonly used today, is questionable. See especially Peterson / Weger 2017. More appropriate is the term 'Germans from X'. On the historical background of the Russian Germans see Wiens 1993; Eisfeld 1999; Eisfeld / Herdt 1996; Landmannschaft der Deutschen aus Russland e.V. 2006; Krieger 2013 and 2017.

<sup>3</sup> It is often overlooked that the group of so-called 'Russian Germans' is historically very heterogeneous, and must be divided into at least two groups from a linguistic perspective alone: On the one hand, there were those who had already immigrated to Russia since the middle of the 16th century for professional reasons and belonged primarily to higher classes of urban Germany, such as engineers, architects, doctors, officers and merchants. They mostly stayed temporarily in Russia (mostly living in metropolises such as Moscow, St. Petersburg, later also Odessa, Tbilisi and Baku), retained German citizenship, or often relinquished their Russian nationality and later abandoned their German language. On the other hand, there was a much larger group of people who followed the advertising and planned settlement policy of the Russian tsars between 1763 and 1824 and emigrated to Russia. This group had economic, social and religious motives, and mainly a rural, arable population, craftsmen and winegrowers (cf. Stumpp 1982: 6). They settled—permanently first since 1763—in newly conquered, mostly rural areas in southern Russia in mostly closed colonies at the Volga, then since 1780 around the Black Sea and on the Crimea and still later—since 1817—in the Caucasus region. Unfortunately, this cannot be discussed in detail here due to lack of space.

the oldest generation (I)<sup>4</sup> , the respective national language is added as L3. Because this group does not speak any Swabian variety, it is not considered hereinafter.

However, the groups of Caucasian Germans who have not left the Caucasian villages or were not allowed to return belong to the descendants of the Caucasus Swabians and thus speak a Swabian variety: In most cases the L1 is either a Swabian variety or the language of the non-German parent (Georgian or Azerbaijani). The L2 is then usually the language or variety of the non-L1 speaking parent. The L3 is usually Standard German for this group and the L4 is Russian. The L2 and L3 were learned in school. For most Caucasian Germans the L1 is without exception a Swabian variety, the L2 is the Standard German and the L3 is Russian, which generationsI and II usually learned in school. These observations coincide with Riehl's remarks on the language skills of the different generations of 'Russian-Germans' in Siberia (cf. Riehl 2017: 22–25). However, questions about the language use in the country of origin were added to the autors interviews.

In general, all three groups speak Russian very well. This is due to the language repression policy in the Soviet Union, which affected both the Caucasian and the Central Asian countries. Only after the collapse of the Soviet Union did the respective national languages such as Georgian, Azerbaijani and Armenian become more widely spoken, but they were still largely confined to the context of the family.<sup>5</sup> Until the 1990s, the first two generations spoke Russian in public. Finally, with the end of the policy of language repression, the respective family languages also entered the public sphere. However, the use of the respective variety does not only differ according to country and group affiliation, but also with regard to each generation.

#### 3 State of Research

There are already numerous studies on German language minorities in Central and Eastern Europe in general. Besides works about Germans in Ukraine (cf. Hvozdyak 2008, especially Trankarpatien-Ukraine cf. Melika 2002) and Romania (cf. Bottesch 2008; Scheuringer 2010, especially Banat Swabians cf. Scheuringer 2016), the following should be noted from Berend (1998; 2011), Berend / Jedig (1991), Berend / Riehl (2008), Blankenhorn (2003) and Rosenberg (1994), who are mainly dedicated to the Russian-German dialects in Russia respectively the

<sup>4</sup> For the determination of the generations see Figure 2 with the distribution of recordings (total).

<sup>5</sup> Had 'Russian Germans' not emigrated in the 1990s from Central Asia, the same might have been said for Kazakh and Uzbek, which would theoretically have affected the language of Generation II.

former Soviet Union. In addition, studies should be considered which deal with the language contact of Russian Germans in Germany, like Anstatt (2011), Meng (2001), Pfetsch (1999) and Rosenberg (2010).

Research on the Swabian variety of Transcaucasian settlements remains a desideratum to this day, although, in contrast to other 'Russian-German' varieties (cf., for example Berend 2011), Swabian hardly had contact with other ('Russian-German') varieties until the pre-war period of the 1930s due to the comparatively closed settlements of the Caucasus Swabians (Berend 2011: 103, 105). This data gap in language contact research as well as in variation linguistics —primarily the 'Russian-German' varieties and their description—needs to be closed. The author recorded voice data on audio and video of almost thirty Caucasian Germans of the experience generation and two generations of descendants: in Baku (Azerbaijan), in Tbilisi's districts Didube (formerly Alexandersdorf) and Tschugureti (formerly Neu-Tiflis), in Bolnisi (formerly: Katharinenfeld) for the first time, and in various German cities such as Landau, Neustadt an der Weinstraße, Offenburg and Schwaikheim (Rems-Mur-Kreis). The number of respondents is not representative in quantitative terms, but qualitative statements can be made with around 15 hours of footage.

#### 4 Theoretical Framework of Treated Topics and Methodology

19 questionnaire-based interviews with Caucasian Germans were collected in Southwest Germany, where the largest groups of Caucasian Germans live today. 7 interviews were collected in Georgia and 2 in Azerbaijan. The sociolinguistic interview-guide, which contains 40 questions, was designed to elicit quantitative and qualitative data. The interview-guide is inspired by Albert / Marx (2010) in consideration of Nortier (2008) and König (2014) and jointly discussed and developed within the framework of the aforementioned project 'German in the World' at the IDS. The surveys in the area of language attitude follow the methods of Gärtig et al. (cf. 2010) and Plewnia / Rothe (cf. 2012: 9–118). So, in order to make statements about the connection between language and identity construction as well as the effects of migration and repression on language acquisition and multilingualism, linguistic biographical data were incorporated and the test persons were asked about their language skills (in German, Swabian, Russian, Georgian, Armenian, Azerbaijani etc.), their respective language attitudes and the social, cultural and media situation.


Figure 2: Distribution of recordings (total)

The questionnaire-based interviews were conducted by the author herself from September to December 2017 in Azerbaijan, Georgia and Germany. These, as well as informal table discussions, were recorded over audio and in some cases also video recorder. Within the framework of this project the data are currently being transcribed and evaluated (following Schmidt / Schütte 2016). The approach to quantitative and qualitative data evaluation was primarily based on Berend (1998; 2003: 151–164), Meng (2001) and Anstatt (2011: 101–128). Afterwards the voice and video recordings will be processed technically and entered into the Archiv für Gesprochenes Deutsch ('Archive for Spoken German') of the Leibniz-Institute for the German Language.

# 5 Results

#### 5.1 Language Contact of Caucasian Germans in the South Caucasus

This chapter begins by describing the language contact constellations of the Caucasian Germans who were excluded from the forced deportations because they were, for example, married to a Georgian, Azerbaijani or Armenian person, and who still live in the German villages of the South Caucasus. Their language contact situation is described by the Caucasian German Balthasar Megrelishvili<sup>6</sup> below. His biography is quite exemplary. He was born in 1947 in Bolnisi (formerly Katharinenfeld, Georgia) and was raised with a Swabian variety (L2) by his German grandmother, with whom he spent most of his childhood. His grandfather was Georgian. His parents both spoke Georgian (L1) with him.

<sup>6</sup> The names of the informants were changed by the author.

[1] Balthasar Megrelishvili (BM), born 1947 in Bolnisi, talks about his language acquisition in the family and the first language contact with Russian (also present at the interview: the Caucasian German Gustav Armin (GA), born 1938 in Neustadt / Wstr.)—Interview Katharina Dück (KD) 2017.


7 /din jan/: incomprehensible and intelligible syllables that don't make sense in any of the three languages used by the informant.


Apart from the numerous hesitations and pauses, which the speaker explains with a lack of language practice, the simultaneous use of the Standard German and the Swabian variety is particularly striking. An example is the interchangeable use of /i/ and /ich/ for the first person. When the speaker remembers his earliest childhood memories with his grandmother, typical Swabian diphtongs such as /gwea/, /hau/ and /taun/ stand out. The latter two forms are overcorrections that other Caucasian Germans do not use. In addition to the Swabian variety as L2, his Swabian grandmother also taught him German songs and poems, such as Heinrich Heine's Loreley, which he recited freely—in what he called the "high language". It is uncertain whether the simultaneous use of the standard variety orignates from this, from German lessons at school, or from his media consumption—Megrelishvili stated that he still reads a lot in German on the Internet and also watches films and videos in German. He repeatedly and deliberately points out the difference between the Swabian "dialect" and the "high language" German.

He later refers to Georgian (L1) as his mother tongue, although here he still claims that he thought in German up to the first grade and, moreover, that a person is defined by the language he or she thinks in. On the other hand, he repeatedly describes himself as Georgian. In the family Georgian was always spoken except for the grandmother, who spoke German. In school, Russian became his L3, and accompanied him during his professional life as an aviator and flight instructor in the army, where Russian was also the main language of communication. Megrelishvili sums up the language contact situations in professional life as follows:

[2] Balthasar Megrelishvili (BM) reports on his language contact in professional life (also present at the interview Gustav Armin (GA))—Interview Katharina Dück (KD) 2017.


During this time, his Georgian also suffered from the repressive language policy in the Soviet Union. He reports that he usually spoke Russian with his Georgian colleagues and only used Georgian when they were alone in pairs. Only after he had retired and returned to his birthplace Bolnisi did Georgian become his main language of communication again.

Overall, it is difficult for the speaker to speak German, and he occasionally changes to Russian or Georgian during the interview:

[3] Balthasar Megrelishvili (BM) reports about his relatives in Germany language acquisition in the family and the first language contact with Russian (also present at the interview Gustav Armin (GA))—Interview Katharina Dück (KD) 2017.


8 /grusin/ Russian for 'Georgians'.


Like all other descendants of the Caucasus Swabians who are living in the Caucasus today, Megrelishvili exhibits frequent (functional and non-functional) code switching (cf. Lüdi 2004 and especially Riehl 2016: 25–27). Apart from transferences (cf. Gass 1996) and spontaneous borrowings (cf. Hoffer 1996 and especially Poplack 2004: 590–591), it is one of the group's most conspicuous variation phenomena, which can also contain insertions of individual elements in the form of discourse markers (cf. Blankenhorn 2003: 77) and modifiers such as /tak/ and /wot/ (cf. Blankenhorn 2003: 124) as well as changes of complete utterances or complex utterance units. The switch occurs mostly when informants talk about emotions such as disappointment, injury, war, captivity and escape traumas but also when recalling positive memories of earliest childhood. Members of this group switch even if they are unaware of the other person's ability to understand the other language.

For example, the interviewer was sometimes asked whether she understood the respective language after changing to Russian/Georgian. Despite all his efforts to speak German, Megrelishvili switched to Russian and Georgian without knowing that the interviewer speaks (and understands) Russian. As soon as he noticed this due to the author's use of the Russian consent particle /da/ (cf. Blankenhorn 2003: 110), he asked directly about the Russian knowledge, and used Russian during the remaining conversation more frequently for longer speeches, and above all for complex contents. Before that, he had tried to explain complex contexts in German and to use his passive German vocabulary. He emphasized several times that he lacks the practice in German. The phenomenon of the alternating use of elements of two or more languages—whether complete

<sup>9 /</sup>silno ljubimij ne budesch/: Russian for 'you won't get very popular' [like 'you can't force popularity / love'].

<sup>10 /</sup>da da/: Russian for ,yes, yes'.

<sup>11 /</sup>khartula dassesea/: Georgian for 'in Georgia it's not like that'.

<sup>12 /</sup>tzalitz tzaporeli weri knewi/: Georgian for 'you can't be nice on purpose' [like 'you're not necessarily considered nice'].

<sup>13 /</sup>tak/: Russian for 'well'.

<sup>14 /</sup>wot/: Russian für 'so'.

utterances or even inserts of individual lexemes—is well known to the informants:

[4] Balthasar Megrelishvili (BM) talks about his everyday experiences with code switching (also present at the interview Gustav Armin (GA))—Interview Katharina Dück (KD) 2017.


This example shows the code switching not only between languages (here Russian and German), but also between the Swabian variety (/daitsche/, /schir garts mir rausgflogerl/) and Standard German (/deutsch/, /nicht mehr sprech deutsch […] dann vergess ich/). The change from Swabian to Standard German is even more noticeable when the informant talks about domain-specific processes. When asked why he was not present at the German city festival the day before, he reports on the necessary grape harvest:

[5] Balthasar Megrelishvili (BM) reports from the vintage (also present at the interview Gustav Armin (GA))—Interview Katharina Dück (KD) 2017.

<sup>15</sup> As Note 19—here the discourse marker /wot/ has an interaction-strategic function as a prelude to the explanation of a communicative process, namely code switching itself.


Viticulture was a Swabian domain in the former German villages of Transcaucasia and not only in Bolnisi (formerly Katharinenfeld), where the informant was born and raised. In fact viticulture also existed in the South Caucasus before the German settlers, the Württemberg Pietists, came. They came from a winegrowing region, and significantly expanded and effectively improved it in the South Caucasus (again in 2001). The vocabulary of the wine-growing sector contains accordingly numerous Swabian lexemes, which are activated in the informant's speech about the grape harvest and initiate further Swabian variants such as /ropft/, /got alles zum grund/, /ghet/. At the same time, he still uses a speech relatively close to the standard German: /aber jetzt ist was eh: gewesen/, instead of the Swabian variants'isch' and 'gwea' , which he uses as well during the interview. It is unclear why the informant does not use the Swabian variety throughout the conversation. Perhaps it is because the interviewer does not belong to the network of "Caucasian Germans", and the informant tries to speak in the "Hochsprache" ('high-level language') out of courtesy. It is probable that the speaker moves within the range of his varieties and, depending on the interlocutor or domain, switches between his base variety and his standard variety (cf. Riehl 2006: 191).

#### 5.2 Language Contact of Caucasian Germans in Germany

Similar observations can also be made in interviews with Caucasian Germans in Germany. Likewise, the Standard German, in which the informant Georg Alles<sup>16</sup> tries to speak to the interviewer, obviously is not easy to follow:<sup>17</sup>

[6] The married couple Georg Alles (GA), born 1932 in Bolnisi, and Odette Alles (OA), born around 1938 in Bolnisi, report where they come from—Interview Katharina Dück (KD) 2017.


16 The names of the informants were changed by the author.

<sup>17</sup> About the context: The Alles couple were asked about their origins. The wife was previously told that I am looking for Caucasian Germans who still speak 'the old Swabian'.


The informant makes an effort to speak High German—especially the adjective /klein/ seems difficult to find; he first uses /klo/ for the Swabian variant 'kloi', but then corrects himself. Numerous hesitations, stuttering and many pauses occur, which prove that the informant struggles with finding the Standard German vocabulary. Although he knows how to use it, he cannot consistently maintain it (this concerns for example the lexem /isch/). Caucasian Germans use the language perceived as High German, usually referred to as "literary German" or "high language", in conversations with non-Caucasian Germans. It is a colloquial variety of German with Swabian colouring—mainly in its phonetic and some lexical Swabian peculiarities such as "schwätzen". As soon as Caucasian Germans are present, or when someone prompts him to use his variety, he changes to the Swabian variety and speaks much more fluently and freely. Compare the informant's wife in line 0031: /na schwätz schwäbisch wen du schoibest schwätze verzeele wilschst/. This variety is particularly evident when Caucasian Germans are among themselves (or have become accustomed to the interviewer):

[7] Ida Illig (II), born 1924 in Bolnisi, Alicia Vögele (AV), born 1926 in Bolnisi, and Ida Kromer (IK), born 1927 in Bolnisi, try to remember a song of praise on Katharinenfeld; about this and other former German colonies like Helenendorf, Annenfeld and Traubengrün (the Caucasian German Wilma Schülke (WS), born 1954 in Kazakhstan)—Interview Katharina Dück (KD) 2017.



The conversation shows an excerpt from a table talk of four Caucasus Swabians in a relaxed atmosphere in a private setting (the author was present but was hardly noticed over time). Being over 90 years old, three of the four female speakers belong to the generation of experience (Generation I). They were born in the 1920s in the former Swabian colony Katharinenfeld away from the Germanspeaking world, where they were neighbours over 70 years ago (as they are again today). They can still actively remember their childhood in Katharinenfeld.

Particularly impressive in this excerpt are the strongly overlapping, repeatedly confirming speeches—a sign that the participants in the discussion feel comfortable and uninhibited, talking to each other as they always do. Especially noticeable are the numerous lexeme variants of Swabian such as /isch/ for 'is' or 'I'; /oin/, /koin/ for'one' or'none'; /guat/ for'good' or /hochmiatig/ for'haughty'. There even are examples of words that are typical of the 'conserved' Swabian variety and are hardly ever used today, such as /ghet/ for 'had'; /gwea/ for 'been' and /ghept/ for'had'. The speakers only switch to Standard German to make sure that the interviewer still understands everything. Apart from that, the two and a half hours of conversation—with few exceptions such as a Russian job title for exterminators—show only a few obvious effects of language contact, which can probably be traced back to the early functional separation (diglossia) of the Swabian (L1) as low variety from Standard German as high variety (cf. Riehl 2014: 16).

The language contact situation of the Caucasian Germans in Germany (here only Generation I applies) differs substantially from that of those living in the South Caucasus today; they initially grew up alone with the 'conserved' variety of Swabian (L1) without a further language of a parent. As already mentioned, this generation learned Standard German, which they call "literary German" (L2), in the German village school in Katharinenfeld. They used this variation when talking to the teacher, reciting poems, or in singing lessons. According to their own statements, only the L1 was used in everyday life. After the deportations in adolescence, they used the L1 exclusively in their domestic environment and the L2 with other Russian Germans from the Volga or Black Sea regions, and they continue to do so today with other Germans.

This generation only encountered Russian (L3) as a foreign language at school. This changed in 1931 when the school language was changed to Russian after the summer holidays. From then on it was completely forbidden to use any language other than Russian in public spaces. At home, this generation continued to speak solely Swabian. They only came into contact with other languages, such as Georgian, when they had hired Georgian workers. Some report that they had played with the children of the cattle keepers or housekeepers and thus learned Georgian.<sup>18</sup> Of this closer domestic language contact, only a few words from the domestic and culinary fields have remained to this day: They use 'shish kebab' instead of 'spit roast', 'dolma' instead of 'cabbage rolls' or 'ajap-sandali' instead of 'vegetable stew'. As most people of this generation spent their entire professional life in the Kazakh, Kyrgyz or Uzbek deportation areas (until the 1950s it was forbidden to leave the assigned settlement), their Russian is at a good to very good language level: everyone can read, write and speak Russian. However, nobody of this generation (!) spoke Russian with me. Code-switching was limited to individual technical terms from former professional life when they reported about it in a free conversation.

The situation is different with Generation II of the Caucasus Swabians living in Germany, as with the spokeswoman WS, who is represented in the above with a short speech. She was born in 1954 in one of the mixed settlements in Kazakhstan and grew up there. Her language contact in German was primarily with Russian. The language repressions<sup>19</sup> in the Soviet Union also applied to Kazakh, so that there was little or no language contact here (cf. Berend / Riehl 2008: 23). On the other hand, there was variety contact, namely with other Russian-German varieties, so that this informant already shows a weakened Swabian variety both through the dominance of Russian in public life and the pressure of the other Russian-German variants. The fact that she still masters the 'conserved' Swabian variety quite well is probably due to the language education of her parents, who despite the ban on speaking a language other than Russian, insisted that the children speak Swabian as soon as they entered the house. Thus, she had the same L1 as the parents, but due to the stronger pressure of the language in her environment as well as other varieties, she developed a weakened Swabian which is characteristic for numerous speakers of this generation. At the same time, there are also some informants in this generation who, have been so strongly influenced by the Russian language contact and the related repressions that they name Russian as their mother tongue. The Caucasus Swabians, however, represent a

<sup>18</sup> The fact that Georgian or Armenian employees could be afforded was only possible at the beginning of the 20th century, when the Caucasus Swabians became wealthier with the bloom of the winegrowers' cooperatives.

<sup>19</sup> In 1929, in all parts of the Soviet Union the Russian language was established as the only language to be used (in public life) and from 1931 onwards it was prosecuted.

remarkably small proportion in contrast to other Russian-German speakers. In addition, the Caucasus Swabians, in general—and this is also remarkable—often answer the question about their mother tongue explicitly; their mother tongue, according to them, is Swabian and not German.

Moreover, the cohesion, culture and language of this group is particularly striking: while other 'Russian Germans'—like in the Volga area—mingled with each other, so that a process of koineization began quite early on (cf. Berend 1998: 10), the Caucasus Swabians sought future spouses almost exclusively among the Caucasus Swabians after the deportations. The internal social structure of the Caucasian Germans is characterised by a strikingly pronounced sense of togetherness (also across countries)—in contrast to other Russian Germans. After the partial rehabilitation of the 'Russian Germans' in the 1950s, the Caucasus Swabians had already settled in Kazakhstan and Central Asia in so-called 'Posjolki gorodskoga tipa' ("city-like settlements") and formed new varieties, and thus culture and identity communities. An example of these settlements is the Shelisinka area of Pavlodar territory in northern Kazakhstan, where many more settlements were situated (cf. Berend 2011:106). However, these settlements ceased to exist in the 1990s because most of the Caucasus Swabians emigrated to Germany.

Interestingly, a striking number of Caucasian Swabians have settled again in southern German regions—often not far from the emigration places of their ancestors. Finally, these settlement trends influence the development of the 'conserved' Swabian variety that should not be underestimated; it is still quite dynamic within the Caucasian Germans in Germany even among the third generation, of which some were born in the Federal Republic. Some of them also state "Swabian"—and not "German"—as their native language (although only two have been recorded in interviews so far). However, they are increasingly using the variety of Swabian which is already common in Germany. In communication with grandparents (Generation I), however, this generation uses exclusively the 'conserved' Swabian variety. None of the 3rd generation informants I spoke to speaks Russian, while their parents speak Russian fluently.

#### 5.3 Language Attitudes of Caucasian Germans

The cross-country identification with the Caucasian German group is reflected particularly strong in the language attitudes that prevail towards the German language: when asked about the German language in general or the Swabian variety in particular, all Caucasian German speakers exhibit strikingly positive reactions and attitudes, whether they live in Germany or in the Caucasus. Associations with the German language as well as the Swabian variety are expressed by terms such as "beauty", "love", and "home". All informants rate both languages as "good" or even "very good". The informants make the strongest distinction between the Swabian variety and the Standard German in the point of "speech melody". Both have a nice sound; but while Standard German sounds "beautiful", especially in poems and songs, Swabian is, in comparison, considered an "honest" and "direct" language.

The emotional aspect, often closely connected with the Swabian variety, cannot be emphasized strongly enough: Swabian is the "language of childhood", "family" and "friends", and is largely associated with the nostalgic feelings "familiarity", "protection", and "security"—especially by the Caucasian Germans in Germany. Emotional descriptions like these about Standard German do not occur. While the form of Standard German is mainly perceived favourably, the Swabian variety is primarily linked to its associative content.

Some informants are not sure whether it is possible to separate "Swabian" from "German". It is often mentioned by the Caucasian Germans in Georgia who show a strong contrast between the Swabian variety and Standard German in their speech that they "belong inseparably" together. Thisfinding is certainly due to the fact that the transitions between the Swabian and Standard German are not distinct (cf. Riehl 2006: 191), although the speakers are conscious of the distinction between the two varieties. Since, for this group of speakers, both varieties have the function as the "language of nearness" (Riehl 2006: 190).

#### 6 Conclusion and Research Desiderata

As has been shown, similar observations could be made in the interviews with the Caucasian Germans as Berend (1998; 2011), Blankenhorn (2003) and Riehl (2006; 2014; 2017). The most striking variation phenomenon in the language contact of Caucasian Germans is code switching, although it can be observed much more frequently in the language use of Caucasian Germans in the South Caucasus, and not specifically in the network. This stands in contrast to the Caucasus Germans in Germany, where code switching is network-specific and takes place less frequently. An essential factor for switching between language elements (nonfunctional code switching left aside) is the will to verbalize strong emotions. Another important factor is the lack of a counterpart in the situationally dominant language: either the term sought is more appropriate for what is meant in the other language or the informants cannot think of it and the other language is used to bridge a momentary lexical gap.

More ambiguous is the change between the 'conserved' Swabian variety and the Standard German. In general, the Caucasian Germans are aware of the difference ["dialect" is "dialect" and "Hochsprache" is "Hochsprache" ('high-level language' is 'high-level language')] and use the Swabian variety primarily in network-specific communication. Riehl (2006) has already described similar observations in her statements on Germans in Transcarpathia. Outside their network, the Caucasian Germans make efforts to use Standard German, even if it is sometimes difficult, but, nevertheless, their speech shows Swabian variants. Sometimes two different variants are used within one speech [/daitsch/ and /deutsch/; /i/, /isch/ and /ich/]. Further studies of the third generation of Caucasian Germans in Germany are required to investigate the 'conserved' Swabian variety that has been learned within the families and is currently in contact with the Swabian variety spoken in Württemberg today.

Apart from that, the Swabian variety plays an essential role both for communication and for the identity construction of "Caucasian Germans" across countries, as can be seen above all in the positive results of questions on language attitudes. At the same time there are those speakers who still speak a Swabian variety but are (or feel) isolated from other speakers due to the former language repression policy.

[8] Balthasar Megrelishvili (BM) tells of other inhabitants of Bolnisi with German ancestors (also present at the interview the Caucasian German Gustav Armin (GA)—Interview Katharina Dück (KD) 2017.


A question that remains open is the future (8: 0006) of the 'conserved' Swabian variety in particular, since a decline in the number of Caucasian Germans in the South Caucasus (8: 0017) is especially evident today. Since hardly anyone has handed over Swabianvariety to the next generation, the lanaguage in the Caucasus for these people is about to be lost.

#### Bibliography


Transkriptionssystem 2 (GAT2). <https://ids-pub.bsz-bw.de/frontdoor/index/docId/4 616> (July 05, 2018).

Songhulaschwili, Awtandil. 1997. Die Deutschen in Georgien. Tbilissi: Mezniereba.

Springform, Daphne. 2004. Deutsche in Georgien. Tbilisi: Goethe-Institut.


#### Appendix

Conventions of transcription according to FOLKER


Dr. Katharina Dück Leibniz-Institute for the German Language (IDS) Augustaanlage 32 68165 Mannheim Germany dueck@ids-mannheim.de http://www1.ids-mannheim.de/direktion/personal/dueck.html

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Peter Rosenberg

# The Impact of Variation, Contact, and Change on Case Morphology: What Can We Learn from Language Islands in the 'Flood'?

Abstract: Today, German language islands in Russia and Brazil are in the process of experiencing language shift. In the course of this shift, the varieties of these communities display certain varying degrees of decomposition and simplification in terms of morphology. Language decay within these communities is not just disorderly, not amorphous, but somehow structured. Regular and irregular morphology are developing differently: While reduction to a two-term (or even common case) system is the main characteristic of regular noun inflection, in personal pronouns two or three-term distinctions are maintained, mostly containing dative. This contribution focuses on comparative language island research, i. e. comparing the impact of linguistic variation and contact on language change: As the comparison of language use in German language islands in Russia and Brazil reveals, phenomena of rapid change occur which are accelerated by language obsolescence as a result of the dissolution of ethnic boundaries. These phenomena are interpreted as induced by the interaction of "internal" and "external" language change with external effects being the trigger of a basically internally structured process.

Keywords: language islands, language contact, language variation, language change, morphology

Abstract: Deutsche Sprachinseln sind heute nahezu überall auf der Welt dem Sprachwechsel zur Mehrheitssprache ausgesetzt. Ihre Sprachinselvarietäten zeigen in unterschiedlichem Ausmaß Erscheinungen des Morphologieabbaus und der Vereinfachung. Einige Abbauprozesse wie der Kasusabbau sind aber nicht amorph, sondern strukturiert und gerichtet. Die reguläre und die irreguläre Morphologie entwickeln sich unterschiedlich: Während die Reduktion zu einem Zwei-Kasus-System oder zum Einheitskasus ein Charakteristikum der Nominalmorphologie ist, erhält sich in der Kasusmorphologie der Personalpronomina ein Zwei- und Drei-Kasus-System, meist unter Bewahrung von Dativformen. Wie könnte diese unterschiedliche Entwicklung erklärt werden? Wo liegt ein Einfluss der Kontaktsprachen vor? Was kann als Konvergenz zur Kontaktsprache oder zu einer Kontaktvarietät gelten? Welche Rolle für die sprachliche Variation spielt die Struktur der Sprachinselvarietäten, welche das Distinktivitätsbewusstsein der Sprechergemeinschaft? Der Beitrag behandelt Ergebnisse aus vergleichenden Untersuchungen in den stark unterschiedlichen Sprachkontaktsettings in Russland und Brasilien und plädiert für eine vergleichende Sprachinselforschung und eine interdisziplinäre Perspektive unter Einbeziehung ethnologischer Ansätze.

Keywords: Sprachinseln, Sprachkontakt, Sprachvariation, Sprachwandel, Morphologie

### 1 Introduction

The first goal of this contribution is to present some findings of a research project exploring language change in German language islands in Brazil and Russia. The core idea of this project centers around the assumption that we can learn from language obsolescence (attrition, decay) as well as from language emergence (standardization, koineization, dialect merger) which has been the subject of linguistic research in the past. Therefore, the second goal is to discuss the question: What can we learn from language islands in obsolescence? The potential learning effect will affect linguistic as well as sociological and ethnological issues of variationist and contact linguistics.<sup>1</sup>

In the first part of this contribution, methodological aspects of comparative language island research (Chapter 2) are addressed. Chapter 3 portrays the results of linguistic change in regular and irregular case morphology of language islands in Russia and Brazil, which are drawn from a research project in Siberia and the Brazilian South. Possible impacts of the structure of the varieties, of the age of the speakers, and of language use in different settings are taken into account. Discussing the relation of contact, variation, and change, convergence as a result of language or of variety contact is analyzed. In Chapter 4, the boundary marking effect of language, i. e. the sociolinguistic and ethnological aspect of group formation, is assessed. The concluding Chapter 5 summarizes the results and advocates an interdisciplinary and comparative approach in language island research.

<sup>1</sup> The state of research on language islands, particularly in Russia and in Latin America, is given for the past in Russia in Rosenberg (2002b), Rosenberg (2005), Berend/Riehl (2008); and for Latin America in Rosenberg (2018), cf. for the Hunsrück variety in Brazil: Altenhofen (2016), Maselko (2013), and for Mennonite Low German in Latin America: Kaufmann (2011).

# 2 Methodological Aspects: Comparative Language Island Research

This chapter explores the methodology of the language island research in question, i. e. what is it about (subject), why (objectives and intentions), where (regions of investigation), who (sample), and how (elicitation and data processing).

The object of the investigation is to explore the development of regular and irregular morphology in German language islands in Russia and Brazil: its frequency, its distribution, and its functionality (depending on languages and language varieties in contact, sociolinguistic context). The intention is to describe the reduction or resistance of case morphology and to explain these phenomena in terms of convergence, language change and language shift.

The regions of the "language islands" 2 have been chosen with the aim of comparability: long existing and big German communities, historically under

<sup>2</sup> Language islands are—in the words of Wiesinger (cf. 1983: 901)—relatively small closed communities on a territory of a linguistically different community. Hutterer (cf. 1982: 178) calls language islands areas of settlement of a linguistic minority within a majority of a different language. Language island research is linked to Contact Linguistics and Variational Linguistics in many ways. A language island may be regarded as a special case of a linguistic minority. Of course, language islands present a vast variety of internally structured linguistic communities under extremely differing contact settings, and language island research is by all means a heterogeneous subject: Depending on the duration and separateness of settlement and the heterogeneity of linguistic varieties they display different levels of language variation systems: from local vernaculars to more or less mixed, leveled or merged dialects, koinés or regiolects up to urban vernaculars, sometimes with superregional usage. As common traits, most of them share a limited area, enclosing a linguistically different community linked by a dense communicative network which is to a certain extent more introverted than extroverted, and connected by attitudinal distinctiveness. Because of its delimitation they are mostly easily observable in time and space, explorable as a whole which makes them a fruitful object of research. The 'island' metaphor suggests a kind of isolation from the surrounding 'sea'. Of course, we cannot take this for granted without empirical analysis. In contrary, the ways and degrees of boundary marking, particularly its negotiation by linguistic means, are an object of investigation. Mattheier (1996) emphasizes, the most important sociolinguistic output of all intervening factors in language islands is an attitudinal structure of distinctiveness as the basis of non-assimilation. He discusses some important differentiations concerning the settings of language contact language islands are exposed to: Progressive (or expanding) language islands have been the prototype of the German colonization in Middle and South Eastern Europe. Regressive language islands have other characteristics than progressive ones, for instance the French-speaking community in Quebec in former times with its separatist attitudes related to the feeling to be the last bridgehead of the former French Belt. Introvertial language islands are self-contained groups, for instance religious language islands like the Mennonites in Russia, Paraguay (cf. Rosenberg 2018), Mexico and the United States (cf. Kaufmann 1997). These groups are most resistant to external influence, but differ considerably in the extent of separateness. On the other hand, mobility of the inhabitants and discontinuity of the settlement,

political pressure, but still rather stable, with Low and High German varieties<sup>3</sup> in contact: the "German National Rayon" in the Altai region in West Siberia (Russia), especially the village of Schumanowka, and the area around Pelotas, Rio Grande do Sul in South Brazil.

The communities mentioned above belong to the world's largest German language islands: Brazil has approximately one million German-speaking inhabitants; this is the second place among all German minorities in the world (the former Soviet Union ranked first with about two million). But they are—sociolinguistically spoken—language islands in the 'flood', i. e. they undergo an accelerated language shift which will be briefly summarized:

The language islands in Russia and Brazil have been chosen due to the occurrence of similar varieties and some parallels in history, yet stark contrasts regarding the structure of the contact language (cf. Chapter 4).

The methodology of the investigation implies the study of language usage in German language islands in different situations, i. e. in three settings of investigation: a set of 60 standardized sentences to be translated into the intended dialect; an interview narration; and a home conversation self-recorded by the speakers. This is done by three age groups of speakers (under 40 years old, 40 to 59 years old, over 60 years old) of two dialects in contact in each country: in Russia the East Low German Plautdietsch and a West Upper German variety called Catholic, in Brazil the East Low German Pomerano and the West Middle German variety Hunsrückisch.

The sample of the project on regular and irregular case morphology in German language islands in Russia and Brazil contains 125 speakers (about 60 in Russia, 60 in Brazil, each investigation with about 60 speakers in the 1990s and 60 in the 2000s).

The recordings were conducted in the village of Schumanowka, in the Altai region in West Siberia, Russia (a central village with speakers of Plautdietsch and the 'Catholic' variety), and in the colonies around Pelotas in Rio Grande do Sul (with speakers of Pomeranian and the Hunsrückisch variety).

With the exception of the youngest age group recorded in the 1990s, the age groups and the groups of dialect speakers are fairly balanced. The data contains a set of 60 sentences from the translation task (added by counting from 1 to 12 and from 10 two 100) of each speaker (with a duration of 8 to 50 minutes), an interview including a narration about customs and personal stories (taking from

ethnically mixed settlements or language islands dispersed among a different speaking population are factors which may induce assimilation (cf. Rosenberg 1994: 155).

<sup>3</sup> Two dialects in contact in each country: in Russia the East Low German Plautdietsch and a West Upper German variety called "Catholic", in Brazil the East Low German Pomerano and the West Middle German variety Hunsrückisch. Further information about the sample of the project will be given below.


TOTAL 6 25 27 23 23 21 125

30 minutes to 1 hours 30 minutes), and an everyday talk self-recorded by the speakers, without the presence of an explorer (with a length of time between 20 min. and 1 h 45 min.).<sup>4</sup>

Table 1: Sample of language island research in Brazil and Russia

The data processing can be visualized by the flowchart in Figure 1.

Figure 1: Flowchart of data processing

<sup>4</sup> The whole duration if the translation task recordings is 50 hours 33 minutes, of the interview recordings 99 hours 50 minutes, and of the self-recordings 29 hours 45 minutes. The translation task has been fully analyzed, from the interviews and the self-recordings a sequence of at least 200 potentially case-marked elements was chosen from the middle period of the talk. The self recordings were conducted systematically for the most part in the 2000s. Additionally, a real-time panel study has been arranged in the 2000s: tape-recordings of 16 speakers a second time after 15 years. Results of the real time study are not part of this contribution.

Our methodology had to be quite ambitious because, in 'drowning' language islands, it is difficult to control all intervening factors since all varieties are mixed varieties<sup>5</sup> , mostly undescribed, and speakers are insecure about the grammar of their varieties. Thus, it might be interesting to describe some of the methods to deal with these challenges.

The audios have been taken systematically in the three elicitation settings mentioned above.

For transcription and tagging annotation the EXMARaLDA program has been used.

The tagging system is an adapted version of the STTS system (cf. Schiller et al. 1999). It has been used the following way: On the left of an angle, the input is noted. On the right the output: word class, gender, case, number, person, weak or strong paradigm of inflection (and, in parentheses, some facultative features like reduction, declension paradigm shift, genitive replacement types etc.). Everything has a fixed digit position for computer aided analysis. All this is documented in a detailed manual.

The annotation program for tagging and the Coma program for sociodemographic and other metadata assignment were part of the computer-aided data processing. The annotationwas completed, in part, by hand and with the help of a drag-and-drop-panel (Annotation panel), which was written, in part, by an own software program (Auto Hotkey) for fully automatic tagging (with corrections afterwards). EXAKT is the program part for searching.<sup>6</sup>

<sup>5</sup> Mixed varieties are the result of dialect merger of German varieties as for the Russia case described by Viktor M. Schirmunski (1930) and Hugo H. Jedig (cf. Berend / Jedig 1991). Schirmunski (1930) called merger and levelling of Russian German varieties a "large-scale experiment on language history" and a "linguistic laboratory". Schirmunski discussed these phenomena under the notion of "primary" and "secondary" dialect features: Variety contact induced linguistic shift in the direction of High German standard features (or variety features which are closest to these). Even if in the Russian German colonies High German standard was rarely available as a criterion of dialect levelling, there has been some evidence in Ukrainian German varieties (of Hessian and Swabian origins). Andreas Dulson (cf. 1941: 93), however, added an important criterion to Schirmunski's assumption: Replacement of (primary) dialect features might be effective if compact groups of linguistic communities are associated with language usage. Otherwise mingling of dialect features instead of replacement occurs. Dialect merger and levelling have been a main subject of research in Russian German language islands since the combination of specific features might be the result of processes in the colonies rather than brought along from the regions of origin in the German homeland (even if such sets of features coincidentally resemble a particular dialect in the interior German speaking area).

<sup>6</sup> As an example the EXAKT search procedure for personal pronouns (PPER) with accusative input (A), which are realized as any demonstrative pronouns (PDS) takes the notation: PPER:.\.A.\*>PDS:.\*

# 3 Linguistic Aspects – Some Results

#### 3.1 Variation

#### 3.1.1 Systematic: Regular – Irregular Inflection

Regularity is understood to be a 'scalar' phenomenon, (more or less) in accordance with the (morphological) rules of a grammar, irregularity of nonaccordance (cf. Ramat 1985). A rule is "any statement expressing a linguistically significant generalization about the grammatical facts of a particular language" (Kiefer 2000: 297). Regularity concerns the input and the output of a rule: the input is considered regular if it can be seen as a (natural) class. The output is considered regular if it is predictable in form and compositional in content (cf. Kiefer 2000: 297). Related to language islands, this distinction is important because we could schematically relate to irregularity/regularity the following features and functional conditions (Table 2):


Table 2: Features and functional conditions of irregularity and regularity (Rosenberg submitted)

In the language islands, a general tendency of case reduction emerges. While genitive is almost entirely absent, dative is rarely realized, accusative is more frequent, and finally, only nominative is left over. Figure 2 displays the results of the translation task with dative input. The data is taken from 125 respondents, who (in the first elicitation setting) were required to translate 60 test sentences containing 270 potentially case marked elements. The morphological data base was categorized into 28,536 types. This enormous amount of nearly 30,000 types is due to the fact that nothing can be excluded because of the mixed character of the varieties, their undescribed status and the insecure knowledge of the speakers. That is why the tagging method was constructed in a probabilistic way ordered by well-defined rules (by verbal or prepositional agreement, then by semantics and last by other parallel structures in the utterance.) <sup>7</sup>

The abbreviations are D for dative realization, \_ for no case ending (for instance de), N for nominative, A for accusative, NA for a common form for nominative and accusative (die), DA for a common form for dative and accusative, + for an additional form which was not intended, 0 for missing realization of the element.

As can be seen, dative is rare in the output with less than one third.

Figure 2: Case marking (nouns, adjectives, possessive pronouns, demonstrative pronouns, determiners, indefinite pronouns, personal pronouns) on dative input (translation task into intended dialect, n = 6,218, in %)

<sup>7</sup> The tagging system was constructed as follows: In the translation task one input sentence was Vor ihm lag ein Hund ('A dog lay in front of him.'): In the output ihm ('him') was realized as eer which could be personal pronoun 3rd person singular feminine gender in dative case ('her') or personal pronoun 3rd person singular masculine gender in nominative case ('he'). The former is considered more likely than the latter. Therefore, the tag is: PPER:M.D.S.3>PPER:F.D. S.3∧PPER:M.N.S.3. On the right hand side of the angle > the conjunction symbol of a small roof (∧) is used to arrange the hierarchy of the output according to probability: In this case it is assumed that, in the first place, it should be a Low German expression of the personal pronoun dative feminine ihr ('her'), but it could also be the nominative masculine er ('he').

These results are for both regular and irregular morphology. As the figure shows, dative is not the most frequent choice on dative input. Unmarked, sometimes reduced forms prevail, and nominative or nominative-accusative forms are frequent. Additional forms include nouns which are doubled (for instance by personal or demonstrative pronouns). On the whole, a tendency of unmarking seems to occur. In regular inflection this is even more pronounced.

#### Regular morphology: noun inflection

Case distinction in regular morphology (noun inflection: nouns, determiners, adjectives, demonstrative, indefinite and possessive pronouns) is rarely marked. In adjectives, for instance, this is quite prominent (Figure 3 and 4)<sup>8</sup> .

Figure 3: Regular Morphology – adjective inflection: Case marking on dative input (translation task into intended dialect, n = 699, in %)

The output marks an oblique case if any case. Dative input or accusative input makes no difference in case marking: nominative (N) or common case (NA) are prevailing by far.

As concerns regular inflection so far, we observe two main tendencies which must be discussed:

<sup>8</sup> For more details concerning regular inflection cf. Figure 5, 17, 18, 20 (possessive pronouns); Figure 7 (nouns); Figure 21 (definite articles).

Figure 4: Regular Morphology – adjective inflection: Case marking on accusative input (translation task into intended dialect, n = 766, in %)

(1) The first one is the spread of den/-n as a default for all kinds of oblique case marking (analogous to accusative masculine) which is common and well attested also for neuter, where we should have -s in High German varieties or -t in Low German ones<sup>9</sup> :


<sup>9</sup> The following examples are sentences taken from the translation task of Low German speakers (No. 1: a speaker of Pomeranian, No. 2: a speaker of Plautdietsch, No. 3: a speaker of Pomeranian). Line (a) contains the output in the intended dialect, line (b) the Standard German input, line (c) an English interlinear translation, and line (d) the Standard English equivalent.

The use of accusative instead of dative, especially of -(e)n/den, seems to be widespread in German nonstandard varieties.<sup>10</sup> However, the development is highly speeded up, at least as described for Low German Plautdietsch.

However, 50 years ago, regarding the Plautdietsch variety of the Russian German Mennonites, Hugo Jedig, stated that dem is the most frequent oblique case marker:

"In general, the accusative (with masculine and neuter) is replaced by dative. […] [The accusative in adjectives; PR] is used by the older generation almost exclusively, by the younger generation, however, it is hardly ever found. […] In all other word classes (articles, pronouns, nouns) the accusative in the dialect is replaced by other cases, for the most part by dative, to some extent also by nominative" (Jedig 1966: 52, 71–72; translation: PR, cf. Table 3)<sup>11</sup>:


Table 3: Definite article in Low German Plautdietsch (cf. Jedig 1966: 52)

50 years later, the dative form de:m, әm, m is less frequent (with hardly 40 %) than no case marking [dә, ә] or a nominative-accusative common form [diː] or the accusative form [de:n, әn, n].

(2) If no oblique case marking is chosen, we find—as a second tendency further reduction to the common case marker -e: This can be recognized by the translation of the SG sentence Ich habe Löcher in meinen neuen Strümpfen ('I have got holes in my new stockings'). The figures below focus

<sup>10</sup> The accusative as the prevailing form of oblique case marking is widespread in the North of Germany (cf. Koß 1983; Panzer 1983). For the urban variety of Berlin, Lasch (cf. 1928: 267) reported on oblique case unmarking in NPs. Rosenberg (1986) detected a replacement of dative by accusative in NPs, frequently in masculine form -(e)n/den also for neuter. Personal pronouns frequently had dative. Schlobinski (1988: 224) concluded "dative with personal pronouns, accusative with noun inflection". Carol Pfaff (1994) detected a common object marking by den among Turkish L1 speakers of German in Berlin.

<sup>11</sup> "Der Akkusativ ist beim Maskulinum und Neutrum im allgemeinen vom Dativ verdrängt." Das Adjektiv im Akkusativ "wird fast ausschließlich von Vertretern der der älteren Generation gebraucht, bei Vertretern der jüngeren Generation dagegen findet sie sich fast gar nicht. […] Der Akkusativ ist in der Mundart in allen anderen Wortklassen (beim Artikel, Pronomen, Substantiv) von anderen Kasus verdrängt, zum größten Teil vom Dativ, zum Teil aber auch vom Nominativ." (Jedig 1966, 52, 71–72, cf. Berend / Jedig 1991).

on the different potentially case-marked elements in the translation task (possessive pronoun, adjective, noun), differentiated by varieties. As the figures show, a common structure occurs, however, with deviations according to the different varieties (which are subject of Chapter 3.1.2): The common case marker -e is prevailing in possessive pronouns (mein-e, Figure 5) and adjectives (neu-e, Figure 6); in nouns (Strümpf-e) it is frequent (while no ending is the most frequent: Strümpf-∅ / Strömp-∅, Figure 7).

Figure 5: Possessive pronoun inflection: Case marking on dative input (translation task into intended dialect, sentence: Ich habe Löcher in meinen neuen Strümpfen. 'I have got holes in my new stockings', n = 64, in %)

Figure 6: Adjective inflection: Case marking on dative input (translation task into intended dialect, sentence: Ich habe Löcher in meinen neuen Strümpfen. 'I've got holes in my new stokkings', n = 64, in %)

Figure 7: Noun inflection: Case marking on dative input (translation task into intended dialect, sentence: Ich habe Löcher in meinen neuen Strümpfen. 'I've got holes in my new stockings', n = 64, in %)

This tendency of simplification is common in the entire regular morphology. Interestingly, the direction of simplification corresponds exactly to the weak inflection paradigm in German, which is also used for constructions that would afford the strong inflection paradigm in Standard German (Table 4):


Table 4: Weak adjective inflection paradigm in German (cf. Helbig / Buscha 2001: 274)

#### Irregular morphology: personal pronoun inflection

The results yielded by irregular inflection (in terms of personal pronouns) are entirely different: While dative is only rarely realized in noun inflection it is very frequent in personal pronoun inflection (Figure 8).

Figure 8: Irregular Morphology – personal pronoun inflection: Case marking on dative input (translation task into intended dialect, n = 986, in %) D = dative realization, \_ = no case ending (for instance de), N = nominative, A = accusative, NA = common form for nominative and accusative (sie), DA = common form for dative and accusative (mi/di, euch), + = additional form, 0 = no realization.

A clear difference with approximately two thirds of the data realizing dative input as dative output is obvious when compared to the results of noun inflection mentioned above (with less than 30 %).

Figures 9 and 10 show case marking on dative input with singular personal pronouns (in interviews and narrations): Since the oblique case is not differentiated in 1st and 2nd person plural<sup>12</sup> and 3rd person plural is rare (see Figure 11) the focus is on singular inflection. As Figure 9 illustrates, dative output is prevailing by far, and almost obligatory in 3rd person singular (Figure 9).

Figure 9: 1st –3 rd person singular personal pronoun inflection: Case marking on dative input (interview and narration, n = 412, in %)

In 3rd person singular, dative output is even more frequent than demanded by input (Figure 10, arrow). Dative output on dative input in oblique case constructions is more frequent than accusative output on accusative input.<sup>13</sup> Where accusative is replaced by dative a two-term system prevails, where accusative is used a three-term distinction is maintained.

<sup>12</sup> us (Standard German uns, 'us'), juuch (Standard German euch, 'you') in Pomeranian (cf. Tressmann 2006: 508, 228), o˛ns / ons (Standard German uns, 'us'), jy:nt / junt (Standard German euch, 'you') in Plautdietsch (cf. Jedig 1966: 76; Siemens 2012: 151), ons (Standard German uns, 'us'), eich in Hunsrückisch and "Catholic".

<sup>13</sup> However, the sharp drop of the column of accusative output might be misleading (aswell as of nominative output): In the translation task accusative (or nominative) is demanded (for instance: sie 'she' 3 rd sing. fem. accusative, and it is realized by common nominative-accusative forms (NA, for instance: se, 'she' 3 rd sing. fem. nominative-accusative). A substantial drop goes into the direction of dative (D), reduction (\_), and replacement by Low German 1st and 2nd person in oblique case (DA: mi, di).

Figure 10: Personal pronouns of 3rd person singular (translation task into intended dialect: all cases (n = 1,762, in %)

Plural pronouns are treated different: 3 rd person plural in oblique case are frequently reduced and changed to singular or replaced by nouns (Figure 11).

Figure 11: Personal pronoun inflection (3rd person plural): Number marking on plural input (translation task into intended dialect, n = 527, in %)

However, personal pronouns are not entirely exempt from case reduction: Speakers often change word class in their output when replacing personal pronouns (ihm, ihr 'him, her') with demonstrative pronouns (dem, der 'this-DAT.M, this-DAT.F', Figure 12).

Figure 12: Realization of word class on dative input (translation task into intended dialect, n = 6,218, in %) N = Noun, PPER = personal pronoun, P\_S = substituting demonstrative or indefinite pronoun, ART = determiner, APPR\_ = preposition with determiner (contracted), ADJ = adjective, P\_AT = attributive demonstrative or indefinite pronoun, PRF = reflexive pronoun, +=additional form, CARD = numeral (STTS, adapted)

If changing the word class into a regularly inflected one, to some extent, speakers use the accusative (den, die 'this-AKK') or nominative, which they use when casemarking regular forms. This could be interpreted as a kind of regularization which might trigger a sequence of reduction, which comprises the stages regularization > simplification (+ defaultization) > morphological reduction (with subsequent substitution, e. g. by word order).<sup>14</sup>

<sup>14</sup> The sequence of reduction may take the succession as in the following example: Regularization: Personal pronouns (e. g. PDT ahm SG 'ihm', 'him') realized as demonstrative pronouns (e. g. PDT dem SG 'dem', the-dat.sg.m/n). Simplification: Demonstrative pronouns (e. g. PDT dem SG 'dem', the-dat.sg.m/n) realized as oblique case with accusative form (PDT den, SG 'den' the-acc.sg.m or PDT dat, SG 'das' the-acc.sg.n) or as common case with nominative form (PDT der, SG 'der', the-nom.sg.m or PDT dat, SG 'das' the-acc.sg.n). Defaultization: When realized as oblique case with accusative form even in neuter (e. g. PDT dat, SG 'das' the-acc.sg.n) often the accusative masculine is used (PDT den, 'den', theacc.sg.m). Morphological reduction: Loss of inflectional markers may take the form of PDT de (definite article without case marking) or PDTe (definite or indefinite article without case marking), i. e. without any inflection. Subsequent substitution (for instance, by word order): definiteness analogue to the Russian language by final position (cf. Rosenberg 2016a).

#### 3.1.2 Variety

Although descriptions of earlier stages of our language island varieties are rare, it is supposed that the dominant dialects of origin (West Middle German and West Upper German varieties) in general had more morphological distinctions than what we find today. Nowadays, we recognize a tendency of common case, of gender neutralization, plural rendered as singular, past tense realized as present tense, a widespread loss of any 3rd pl. pronoun inflection (cf. Rosenberg 2016a). These phenomena appear as a gradual breakdown of morphology in all communities investigated.

Of course, differences between the language island varieties are also important: To a certain extent, case marking is depending on the variety spoken (Figure 13).

Figure 13: Dative input: Case marking by varieties (translation task into intended dialect, n = 6,218, in %) HRX = West Middle German variety Hunsrückisch (Brazil), KAT = West Upper German variety called "Catholic" (Russia), POM = East Low German Pomerano (Brazil), PDT = East Low German Plautdietsch (Russia).

The Catholic community in Russia uses more dative forms than any other. While the weak inflection paradigm mentioned above (cf. Table 4) is common for the High German dialects, the Low German varieties have also their typical Zeroendings which can be recognized when accusative input is given (Figure 14).

Figure 14: Personal pronouns with accusative input realized as demonstrative or indefinite pronouns: Case marking by varieties (translation task into intended dialect, n = 332, in %)

#### 3.1.3 Age and Language Usage

The age factor is not significant in case marking of personal pronouns (Figure 15).

While age groups do not remarkably correlate with case distinction, criteria of language use, however, offer a more convincing correlation: the intergenerational usage of the language island variety spoken with grandparents correlates best with case marking on personal pronouns (proportionally depending on frequency of usage, Figure 16).<sup>15</sup>

<sup>15</sup> As a measurement of the language island variety use in a core domain we compared speakers who use the language island variety with their grandparents (which is the last 'stronghold' of this variety), and thosewho do not. The subsample comprisesthose of the 2000s' investigation (with information about language use with the grandparents): 70.6 % use the language island variety only talking to the grandparents, 18.1 % frequently, 4.5 % sometimes, 6.8 % not at all. Besides the lower dative scores, the data yield an (uncommon) replacement of dative by accusative in personal pronouns with those ones who do not or not very often use the variety with their grandparents. Those speakers who do not use the German variety with grandparents use accusative with double frequency compared to those who use the German variety with grandparents (22 % compared to 11 %).

Figure 15: Personal pronoun on dative input: Case marking by age groups (translation task into intended dialect, n = 396, in %)

Figure 16: Personal pronoun on dative input: Case marking by language usage (frequency of dialect use with grandparents, translation task into intended dialect, n = 398, in %)

#### 3.1.4 Formality (Situational Setting)

Finally, the situational setting of the investigation reveals some differences.

The results of interview style and narrations compared to the translation task yield some differences (in possessive pronouns) with a slightly higher tendency to nominative (with about 30 %) instead of dative or any other cases in the interview narrations which signifies an even more radical reduction of case marking (comparing Figure 17 and Figure 18)<sup>16</sup> .

Figure 17: Possessive pronoun: case marking on dative input – Translation task (n = 324, in %)

<sup>16</sup> Narrations tend to yield more informal speech than translation tasks. The expectation would be that the translation task should produce the maximal case distinction speakers are able to generate while narrations contain less case distinctions. This is partly due to the Standard German input in the translation task. Since we want to compare language use in different settings of elicitation a tertium comparationis is needed also in the narrations. As tertium comparationis we used again Standard German correspondents (like cases). Differences between translation task and narrations are not analyzed up to now.

Figure 18: Possessive pronoun: case marking on dative input – Interview and narration (n = 281, in %)

#### 3.2 Contact

One of the most disputed questions is: Is change caused by convergence (an interpretation which appears sometimes referring to the influence of Brazilian Portuguese or US-American English)?

At first, it should be asked: convergence with what? Are the phenomena observed results of convergence as a result of language contact to Russian and Brazilian Portuguese respectively—are they related to the variety structure or even of convergence with another language island variety?

#### 3.2.1 Convergence as a Result of Language Contact

The phenomena presented above emerge in all observed varieties, not only in those communities with intense contact to the majority language or to another German variety. Therefore, convergence is not very likely to serve as an explanation of this kind of change.

Case reduction occurs in communities with a morphologically 'rich' contact language (Russian), as well as with a 'poor' one (Brazilian Portuguese), in morphologically more 'conservative' varieties (Low German dialects), as well as in others (High German dialects).

The influence of the contact languages may, indeed, be referred to the following structural differences:

Russian hassix caseswhich must be case-markedwith little exception, in noun as well as in pronoun inflection, even in typical settings which trigger reduction processes: orality, colloquial and informal style, or allegro speech. Also, within personal pronouns, all cases are case-marked. Singular is more differentiated than plural, 3 rd singular masculine is most differentiated.

Brazilian Portuguese has case—as far as formally marked—only in personal pronouns: While in 1st and 2 nd person (singular and plural) merely an oblique case is marked, dative and accusative are differentiated in 3 rd person singular.

The concept of case is clearly established in both contact languages. Like other languages, singular (more than plural), 3 rd person (more than 1 st and 2 nd person), and accusative (more than dative) are morphologically most differentiated in terms of case- (and gender-) marking.

The comparison of the language islands in Russia and Brazil reveals some differences: Within the Russian German islands,speakers maintain dative even in case of the 'regularization' mentioned above (by change of word class from the irregular to the regular inflection paradigm (CAT and PDT both are Russian German varieties, Figure 19). This might be due to the influence of the morphologically more differentiated Russian language.

Figure 19: Irregular morphology replaced by regular morphology – Personal pronouns(on dative input) realized as demonstrative or indefinite pronouns: Case marking by varieties (translation task into intended dialect, n = 298, in %)

Interview narrations yield some more differences between the varieties—the Russian German language islands producing more datives than the Brazilian German ones (in possessive pronouns, Figure 20).

Figure 20: Regular morphology: Possessive pronoun case marking, corresponding to Standard German dative. (Interview and narrations, n = 281, in %)

#### 3.2.2 Variety Structure and Variety Contact

Of course, the grammatical structure of the varieties is of some significance. Furthermore, the linguistic communities differ in terms of boundary marking: The Plautdietsch speaking community of the Russian German Mennonites can be referred to as a real "group" (Brubaker 2002) in the sense of a dense communicative network with shared socio-cultural norms and a strong feeling of "belonging" (Pfaff-Czarnecka 2013). Their variety serves as a "boundary marker" (Barth 1969), distinguishing the community at socio-cultural, economic, and religious levels. Additionally, the speakers of Plautdietsch were the majority and the autochthonous inhabitants in the "central village" of Schumanowka composed in the 1970s out of several smaller villages speaking different dialects. This made them more resistant to external influence.<sup>17</sup> This is partly true also for the speakers of Pomeranian in the Pelotas region. However, in Brazil, Hunsrückisch can be regarded as a koiné among the German speaking population (cf. Koch 1971: 96; Altenhofen 1996).

Comparing the varieties reveals some differences concerning case reduction (Figure 21).

Dative output (D) in determiners is most frequent with the Catholic variety, and in general its more frequent with the High German varieties (Hunsrückisch and 'Catholic'), while reduction to de (\_) is slightly more frequent with the Low German varieties.

<sup>17</sup> At least, this was true at the time ofrecording. In the 2000s the Mennonite community largely emigrated to Germany (or other countries like Canada). But, again, the resettlers have built new communities in the Western parts of Germany (for instance, around Detmold).

Figure 21: Definite article on dative input: Case marking by varieties (n = 1,142, in %)

Regarding personal pronouns, dative output is more frequent in the 'Catholic' speaking community. Low German speakers, especially the Plautdietsch community, use the oblique form mi, di (DA), speakers of Hunsrückisch use also accusative (mich/dich, Figure 22).

Figure 22: Personal pronouns on dative input: Case marking by varieties (n = 986, in %)

Of course, there are some phenomena of convergence directed to the contact language or a contact variety within the language islands. However, the overall tendency is quite uniform. Hence, the case reduction presented above appears to be an accelerated—but ordinary—linguistic change all German varieties are subject to, not essentially a matter of contact-induced adoption or advergence.

#### 3.3 Explanations of Change

The observation of case reduction in noun inflection and resistance to change in personal pronoun inflection demands answers to two questions:

The first question to answer is: Why do these processes of an accelerated language change take place in the very moment of language shift?

The present situation of the German language islands characterized as in decline can be described as follows:

The language island varieties are (more or less) communities "in obsolescence" (Dorian 1989) which entails

	- a representation of grammatical relations less by morphology than by word order
	- form-function-patterns according to the 1:1-principle (cf. Andersen 1989: 386)
	- and a replacement of marked (case-)structures by unmarked (or less marked) structures (cf. Campbell / Muntzel 1989: 189; Jakobson 1936) or by 'natural' ones (cf. Dressler 2000; Mayerthaler et al. 1998: 167).

The second question to address is: Why are personal pronouns conservative in terms of case distinction? We can account for the following tentative answers:

– Personal pronouns are frequent.

– They have—as a core concept—animate reference.

<sup>18</sup> The term refers to the concept of norm opposed to system distinguished by Eugenio Coseriu (1974). Norm in this sense is a category of language use, the 'system of what is usual in a linguistic community' while the system is a category of the functional language, the 'system of what is possible in a language'.

	- cf. English him/her (Old English dative: him/hire)
	- cf. Danish ham/henne (Old Scandinavian dative honum/henni)
	- cf. French lui (Old French li, Vulgar Latin li, Latin illi)

The case semantics of dative as recipient case prevails while locative is less important (and comitative is insignificant because of few data).<sup>19</sup>

What is maintained in terms of case distinction seems to be in part a simplification and reduction of morphology to the point of a core concept of case that we could call a "resemantization" which in turn might be the backside of degrammaticalization (cf. Leiss 2004: 857).

<sup>19</sup> The figure contains the output results of the following test sentences: (Das ist ihre Tasche.) Gebt sie ihr zurück. '(That is her bag.) Give it back to her' (recipient), Der Direktor hat ihn ihm vorgestellt 'The director has introduced him to him' (recipient) Vor ihm lag ein Hund 'A dog lay in front of him' (locative), Die Kinder kletterten an ihr hoch 'The children climbed up on her' (locative), Mit ihm fahren wir nicht. 'We do not ride with him' (comitative). Only translations into 3rd person pronoun singular (with the gender requested) were counted (recipient: n = 66, locative: 72, comitative: n = 25). Beyond semantics a further important factor for case assignment could be seen in whether the case is governed verbal or prepositional: In our translation task, dative as the case for the recipient contains examples governed verbal, while the other dative functions (locative and comitative) are governed prepositional.

Figure 23: Personal pronouns with dative output (3rd person singular): Case semantics (translation task, n = 163, in %)

# 4 Sociolinguistic and Ethnological Aspects: Boundary Marking by Language

What is specific of these language islands in terms of sociolinguistics and ethnology?

As mentioned above, these communities belong to the biggest German language islands worldwide. But they are language islands in decline: Language shift in the Brazilian language islands is more intense, but the Russian language islands catch up quickly. The process of assimilation began earlier in the Brazilian German communities, but it is more rapid in the Russian German communities. This development will be traced in the following part:

Among German settlers in Brazil (since 1824), the historical conditions of colonization have had much in common with those of the Russian Germans (since 1763). Settlers were living separated in small isolated colonies just as in Russia. They used predominantly their dialect varieties for communication, and since these varieties were quite different, dialect convergence was an inherent trait of their development. Dialect varieties cover a wide range: from the prevailing (Rhine Franconian and Moselle Franconian) Hunsrück varieties over Swabian and Volga German to Low German Westphalian and Pomeranian varieties in Brazil (just to mention some of them), and from the widespread (Rhine Franconian and Hessian) Volga German varieties over Swabian and Bavarian to Low German Plautdietsch of the Mennonites in the former Soviet Union.

The High German standard language was told at school, but with the political restrictions since (at the latest) 1934 under Stalin in USSR and under the nationalist government of Getulio Vargas in Brazil (Estado Novo since 1937) all educational and political minority rights were suspended. Therefore, a redialectalization took place. In contrast to Russia, however, in Brazil a superregional (Hunsrück) variety emerged which was never the case in Russia. Obviously the numerical dominance of the Hunsrück speakers (about 50 p.c. of the first settlers) and the closer network of communication which was not as limited as in Russia have led to this important difference.

The Brazilian society is multiethnic and the German speaking minority is only one of a hundred ethnic communities. From the beginning, the German colonists were 'aliens' by mission: For about 100 years, they kept their distance to the surrounding population in terms of geography, language, culture, economy, religion and social structure. Since about 1940 'Brazilianization' has emerged, unifying the country (at least related to the 'white' Brazilians), in the last decades modernizing the society, and, hence, lowering the barriers of social contact. For a long time, the maintenance of the minority language and culture depended on the autonomous settlement. The German speaking settlements, however, became subsequently integrated into the society. Today, Brazilian Germans are primarily Brazilians, speaking Brazilian Portuguese, married with Brazilians of other ethnic descent, studying somewhere in the country. Among the younger ones, the German language has become a heritage language. The disintegration of the language islands began two generations ago. Nowadays, ethnic diversity is a familiar trait of all people but it is not a vital resource of social distinction.

The former USSR was and Russia is—by constitution as well as by societal awareness—based on ethnicity. Language served as a boundary marker since it represented a difference: the experience of social or cultural difference and of communicative belonging (as long as 'compact groups' were demarcated by language and code alternation structures were established). This was even true in the deportation camps and guarded villages.<sup>20</sup>

Although German settlement in the USSR has ever been discontinuous, the manifold migrations (voluntarily or not) of the Russian Germans did not affect the ethnically based belonging. However, since the 1970s some political liberations, the modernization of the country, and the construction of 'central villages' (with different German varieties spoken) enhanced the expansion of Russian among the younger generation.

<sup>20</sup> The history of the German language in Russia / USSR is described in more detail in Rosenberg (2002a; 2005), Berend / Riehl (2008), with respect to ethnic belonging in Rosenberg (2016b).

Since the breakdown of the Soviet Union, a disruptive language shift emerged among Russian Germans because of the mass emigration of resettlers ('Aussiedler') to Germany in the 1990s. Today, the majority language is dramatically expanding in public and private domains. Younger generations are quickly shifting to Russian, code-mixing is more frequent than code-switching<sup>21</sup> (as functionally alternating use of languages, cf. Auer 2009: 108), and the proportion of second-language learners outweighs native speakers. Intermarriage is steadily increasing, and getting vocational education outside of the village is common. Additionally, the ethnic composition of the villages is becoming more diffuse because of the replacement of the emigrated resettlers by non-Germans or nonlocals (immigrating Germans from the Central Asian republics of former USSR). Today, only about 10–15 % of the villagers are locally born and network clusters are dissolving. Being German is not a primary distinction anymore.

In a nutshell, language shift in the Brazilian language islands is more advanced, but the Russian language islands catch up quickly. The process of assimilation began earlier in the Brazilian German communities, but it is more rapid in the Russian German communities.

This is met by the results of our sociolinguistic survey in the colonies (with 85 items concerning sociodemographic data, social networks, language awareness, language acquisition, language competence and language use, attitudes, related to the varieties and languages in contact): Within the non-public language domains German is mainly used with the grandparents and parents, rarely with spouses and children. For speaking with the children the contact language is used mostly. The trendlines (for usage 'only') are almost inverted (Figures 24 and 25).<sup>22</sup>

As the item oldest child (replying) shows, there seems to be an intergenerational pattern with very high scores for German "never" and Contact language "only". If German is used, it is restricted to semi-public domains (with neighbors, friends, and colleagues at work).

Within the (semi) public language domains, German is hardly ever used (Figure 26: grey columns). In public the contact language is prevailing, the more outside the village and the more institutional the more Russian or Brazilian Portuguese is used in communication (Figure 27: red column). Again, the dis-

<sup>21</sup> Code-switching is here defined as functionally alternating use of languages or varieties in one single conversation while code-mixing is an alternating use of languages or varieties which is not perceived as functional (cf. Auer 2009: 108).

<sup>22</sup> Only 2000s' inquiry data in Russia and Brazil (n = 68, without 1 informant with "no response"). Oldest child is the first born child in the family: "oldest child" means mutually using the language with the other family members; "oldest child (replying)" means answering of the child only. In the Figures 24–29, only the informants with (grand)parents and children respectively were counted.

Figure 24: Sociolinguistics of the language islands: German language use in the family domain (n = 68, in %)

Figure 25: Sociolinguistics of the language islands: Contact language use in the family domain (n = 68, in %)

tribution of German language usage and the use of the contact language are almost inverted.

Comparing Russia and Brazil, some traits of an accelerated loss of the German language in Russia can be detected: If asked whether the speakers use the contact

Figure 26: Sociolinguistics of the language islands: German language use in (semi) public domains (n = 68, in %)

Figure 27: Sociolinguistics of the language islands: Contact language use in (semi) public domains (n = 68, in %)

language with their parents, more Russian Germans predominantly answer never (more than Brazilian Germans do; cf. Figure 28: grey columns).

But among Russian Germans the oldest children are talking to their parents in Russian 'only' or 'frequently' by a clear majority (Figure 29).

Figure 28: Sociolinguistics of the language islands: Contact language use in the family domain: Speaking with parents (n = 59, numbers)

Figure 29: Sociolinguistics of the language islands: Contact language use in the family domain: Oldest child speaking with parents (n = 54, numbers)

What is striking in our findings is that the acceleration in language change in the Russian-German language islands is not simply a consequence of language contact and imposition.Whatfind a loss of something different:the gradual loss of knowing and caring about what islinguistically 'ours' and whatis'theirs'. This is connected to a lack of intergenerational transmission of the German language, an increasing proportion of non-native speakers, and a common practice of code-mixing (frequently without any awareness of using elements of two languages). The determining factor which might have opened the gate for change in these communities could be calledaloss of normativity, i. e. of norm awareness and norm loyalty.

# 5 Conclusions and Further Considerations: What Can We Learn From Language Islands in the 'Flood'?

The results suggest some cautious conclusions and considerations:


Ethnic boundaries are not 'given' by the essence of origin, language or culture, but constructed by choice and evaluation of social relations.

This is the famous contribution made by Fredrik Barth (1969), the forefather of ethnological constructivism, who stated that the most important process is in ethnic boundary marking: "The critical focus of investigation from this point of view becomes the ethnic boundary that defines the group, not the cultural stuff that it encloses." (Barth 1969: 16)

Brubaker also warns against the tradition of 'groupism', which takes (ethnic) groups for 'things in the world' instead of appraising a shared and imagined 'groupness'. Groups are "what we want to explain, not what we want to explain things with" (Brubaker 2002: 165).

In postmodern terminology, Joanna Pfaff-Czarnecka (2011) suggests using the concept of 'belonging', not 'identity' with the aim of assessing the heterogeneous and multilingual group memberships and the degree of incorporation into different groups.<sup>23</sup>

The notion of construction has become widely accepted in social sciences and in interactional sociolinguistics. Furthermore, it is sometimes referred to in postcolonial studies, when construction appears as a deus ex machina, a kind of passe-partout explanation, which shifts the field of interpretation from interaction to mental states. However, since we must explain the emergence or obsolescence of boundaries (of social, ethnic or national groups), we still have

<sup>23</sup> Since belonging to a certain degree is an attitudinal matter the study of attitudes within the language island community is urgently needed. As Boas / Fingerhuth (2017) show, even speakers with positive attitudes switch to English. The diffusion of linguistic boundaries as well as complex feelings of belonging may play a significant role in these processes.

some questions remaining: What are the resources of this construction? What is the impact of language in boundary marking?

Boundaries may be differentiated according to three criteria: durability (stability), permeability (allowing crossing or not), and liminality (abrupt or smooth transition, cf. Schiffauer et al. 2018).

What causes the construction of a disruptive boundary orits state as a blurred one?

As Fredrik Barth (1994) conceded, 25 years after his programmatic proposition: Construction is not arbitrary, not just an imagination, but as a social organization of cultural difference, it consists in structuring experience along central cultural values.

Instead of disregarding the role of experience and social interaction, it is suggested that we take construction as a threefold process: A process including selection of experiential features (making them 'focused' in terms of Le Page / Tabouret-Keller 1985), their hierarchization (making them relevant) and—as far as attitudes are involved—their evaluation (making them highly valued)<sup>24</sup> .

Therefore, we must ask: under which circumstances does language serve as an ethnic boundary marker, under which does it lose this ability?

With the focus on ethnicity we could say: Language is an ethnic boundary marker if it serves a communicative need (in some language domains), if it displays a certain distinctiveness (Mattheier 1996), and if it is considered a legitimate distinction within an ethnic frame. An ethnic frame is chosen if social experience is reasonably focused as an ethnic structure (cf. Esser 1996a; 1996b) with prototypical actors of relevant 'ethnic' characteristics and if an ethnolinguistic vitality of the speech community is positively evaluated (cf. Giles / Bourhis / Taylor 1977). Then, ethnic (or multiethnic) varieties are maintained or even emerge.

Ethnic boundaries will vanish if the experience of difference is less prominent, if language loses the ability to focus this experience, and if differences no longer represent an ethnic "loading". Then, normativity decreases, and borders become diffuse—not when oppression is most severe, but when language and culture lose their discreteness. Then, language shift—and sometimes

<sup>24</sup> Barth himself, has admitted some oversimplifications when reviewing his approach of 1969 at a 25 years anniversary conference in 1993: The main objections have been the following: The claim, it wouldn't be the "cultural stuff"which defines the ethnic group, but the boundary was "overstated", and thus "people's choice of diacritica appeared arbitrary" (Barth 1994: 12). These diacritica of boundary marking are not constructed by a "mere act of imagining", but rather have "empirical properties" (Barth 1994: 13): they are based on "experience", and reflect "salient, major cultural discontinuities" (Barth 1994: 14). In this sense, "ethnicity is the social organization of culture difference" (Barth 1994: 13).

language decay—is the consequence. Losing boundaries is akin to losing norm awareness, norm institutions and norm loyalty, which opens the gate for obsolescence, assimilation, and change.

3. Keeping this in mind, the interdependence of boundary marking, language contact, variation and change requires an interdisciplinary methodological approach:

The description of morphological change may be the realm of structural linguistics. If we want to explain the impact of external factors, we would have to assess variety and language contact phenomena. Comparing different contact settings may provide evidence of cognitive and even typological processes of change. Since time and social forces of change are part of this explanation sociolinguistic and ethnological analyses are needed to assess the role of language in boundary marking and the signification of belonging by language use.

Figure 30: Interdisciplinary methodological approach in comparative language island research

A comparative language island research could be applied to contribute to the question of internal and / or external induced change. Research on language islands in obsolescence may serve just as well as the famous linguistic laboratory of language change Viktor Schirmunski (1930) addressed for emerging language island varieties.

This means to combine the study of internal (structural, typological and cognitive), external (variety and language contact), and attitudinal (sociolinguistic and ethnographical) features. If possible, a comparative investigation might be useful—of different models of contact, different types of speakers and different variational settings.

### Bibliography


Boas, Hans C. 2009a. The life and death of Texas German. Durham: Duke University Press.


(eds.): Dialektologie. Ein Handbuch zur deutschen und allgemeinen Dialektforschung. Vol. 2. Berlin / New York: de Gruyter, 178–189.


WDSA = Wolgadeutscher Sprachatlas, aufgrund der von Georg Dinges 1925–1929 gesammelten Materialien, bearbeitet und herausgegeben von Nina Berend. Unter Mitarbeit von Rudolf Post. Tübingen / Basel 1996: Francke.

Dr. Peter Rosenberg European University Viadrina Frankfurt (Oder) Faculty of Cultural and Social Sciences Große Scharrnstraße 59 15230 Frankfurt (Oder) Germany rosenberg@europa-uni.de

# Mateusz Maselko

# Grammaticalized GIVE as Existential and Copula Verb in Riograndese Hunsrik and Brazilian Portuguese: A Case of (Supportive) Language Contact?

Abstract: The paper contributes to the fields of syntactically oriented variationist linguistics, contact linguistics, as well as language typology and deals with an extraterritorial variety of German, Riograndenser Hunsrückisch 'Riograndese Hunsrik', and its primary contact language, Brazilian Portuguese. The article focuses on (uncommon worldwide) syntactic constructions occurring in the two mentioned languages in which the (grammaticalized) give-verb (in German: geben, in Portuguese: dar) can adopt the function of an existential verb and a copula with the semantics of 'to become'. Based on an analysis of pertinent entries in dictionaries and grammars as well as empirical (pilot) studies, we find that the verb shows a pronounced ingressive character in both languages but has a wider range of semantic possibilities in German. Furthermore, the article provides an excursus on the (potential) use of geben as a future tense auxiliary in Riograndese Hunsrik as well as a sociolinguistic-typological overview of this dialectal variety of German.

Keywords: language contact Riograndese Hunsrik German : Brazilian Portuguse, syntax, give-verb (geben, dar), existential verb, copula verb

Abstract: Der Beitrag ist im Bereich der syntaktisch ausgerichteten Variationslinguistik, der Kontaktlinguistik sowie der Sprachtypologie angesiedelt und setzt sich auseinander mit einer extraterritorialen Varietät des Deutschen, Riograndenser Hunsrückisch, und ihrer primären Kontaktsprache, der Nationalsprache brasilianisch Portugiesisch. Der Artikel fokussiert auf in den beiden genannten Sprach(varietät)en auftretende (weltweit äußerst unübliche) syntaktische Konstruktionen, in denen das (grammatikalisierte) give-Verb (dt. geben, por. dar) die Funktion eines Existenzverbs bzw. einer Kopula mit der Semantik 'to become (= werden)' übernehmen kann. Auf Basis von einer Analyse einschlägiger Einträge in den Lexika und Grammatiken sowie empirischen (Pilot-)Studien wird festgestellt, dass das Verb in den beiden Sprachen einen weit ausgeprägten ingressiven Charakter aufweist, im Deutschen jedoch über ein breiteres Spektrum an semantischen Möglichkeiten verfügt. Weiters bietet der Beitrag einen Exkurs zum (potentiellen) Gebrauch von geben als Futurauxiliar im Riograndenser Hunsrückisch sowie einen soziolinguistisch-sprachtypologischen Überblick über diese dialektale Varietät des Deutschen.

Keywords: Sprachkontakt Deutsch/Riograndenser Hunsrückisch : (brasilianisches) Portugiesisch, Syntax, give-Verb (geben, dar), Existenzverb, Kopulaverb

#### 1 Introduction

The paper<sup>1</sup> is interested in the syntactic-typological peculiarities of the West Central German dialects and Luxembourgish (as well as a few other languages) in which give<sup>2</sup> appears in its usual function as a trivalent full verb and also shows grammaticalized variants. The focus is on two roles, namely as an existential verb and as a copula with the meaning 'to become'. Remarkably—not to say astonishingly—we observe both of these particularities in the extraterritorial variety of German called Riograndenser Hunsrückisch 'Riograndese Hunsrik' and in its primary contact language Brazilian Portuguese, which has official language status and is the most widely used language in the affected area as well as in the rest of the country.

The questions addressed in this article concern various aspects of syntactically oriented variationist linguistics, contact linguistics, and language typology. Particular attention is paid to structure, semantics, (possible) situational context of use, and the areal distribution of the give-constructions mentioned above as well as their status in the languages concerned and their reference works such as dictionaries and grammars.

The article is structured as follows: In the first step, in Section 2, a brief sociolinguistic-typological introduction to the non-standard variety of German spoken in the Brazilian state of Rio Grande do Sul is given. Furthermore,we make the language contact situation occurring there a subject of discussion. Section 3 demonstrates the current state of research on Riograndese Hunsrik and the grammaticalized give-variants, especially in German and Brazilian Portuguese. Section 4 deals with the basic concepts of language contact research, such as borrowing and transference. In Section 5, we investigate the use of give as an existential verb and as a copula on the basis of the existing research literature and

<sup>1</sup> Many thanks to my friend Dave Britain (University of Bern, Switzerland) for his contribution to this article: helpful comments and linguistic correction. Ein herzliches Dankeschön!

<sup>2</sup> In line with Fillmore (1982), the semantically defined categories (concepts) are set in small caps (e. g., give), whereas italics are used to point out the verbs (forms) that refer to these action concepts (e. g., geben, dar).

empirical analyses. Moreover, the chapter reflects upon the German-specific function of give as a future tense auxiliary. The final Section 6 summarizes the findings and presents further potential research desiderata.

# 2 Riograndenser Hunsrückisch 'Riograndese Hunsrik': A Brief Overview

#### 2.1 Distribution, Origin, and Terminology

The paper focuses on Riograndenser Hunsrückisch 'Riograndese Hunsrik<sup>3</sup> ' (ISO 639–3 code: hrx), a (non-standard) extraterritorial variety of German. It is (actively) spoken outside of the contiguous European German-speaking area in Southern Brazil by at least 700,000 people (cf. Altenhofen 1996: 56).<sup>4</sup> Most of them live in the southernmost state of Brazil Rio Grande do Sul<sup>5</sup> . The dialect name became established in the research landscape at the end of the 1990s (see more 3) —mainly through the above-quoted publication of Altenhofen (1996). It refers, on the one hand, to the geographical origin of the German emigrants (Hunsrückisch ← Hunsrück is a mountain range in Rhineland-Palatinate, Germany) and, on the other hand, to their new 'home' in South America (Riograndenser ← Rio Grande do Sul). The label is, nevertheless, controversial: Firstly, because the immigrants came from various parts of the German-speaking area and not only the Hunsrück region; and secondly, because, besides Rio Grande do Sul, other regions of (Southern) Brazil such as Santa Catarina, Espírito Santo, and Paraná were also settled by German speakers. However, there is no doubt that the naming was determined by historical statistics (and facts): the largest (and very early) migration flow came from Southwestern Germany, especially the Hunsrück (and

<sup>3</sup> Hunsrik has been chosen by "Ethnologue"—an annual (English-language) reference publication which provides current statistics and other information on the living languages of the world—to designate the German variety in Southern Brazil (cf. "Hunsrik" 2019). Based on the term most commonly used in the German-speaking research landscape, I have decided to supplement the term Hunsrik with the local semantics of the adjective Riograndese (see my previous English-language publication on German spoken in Brazil: Maselko 2013b).

<sup>4</sup> "Ethnologue" even talks of 3,000,000 Hunsrik speakers (cf. "Hunsrik" 2019). However, this figure should be treated with caution. The—in all likelihood exaggerated—number seems to include passive speakers as well.

<sup>5</sup> Língua Hunsrik 'Hunsrik language' was declared an integral part of the historical and cultural heritage of Rio Grande do Sul by the state legislative assembly in 2012. It is also considered a co-official language in the municipality of Santa Maria do Herval (as well as in Antônio Carlos located in the state of Santa Catarina). The legal situation and measures taken to preserve, disseminate and standardize the minority language, to communicate the history and culture of the language community and to strengthen social identity led to the introduction of Hunsrik lessons in some schools in Rio Grande do Sul (cf. Maselko / Hamester Johann / Dewes 2014).

Palatinate), and settled mainly in Rio Grande do Sul. German immigration started in 1824 and lasted for more than a century (cf. Altenhofen 1996: 14; Engelmann 2004: 62).<sup>6</sup>

From the perspective of variationist linguistics, the (German) Hunsrück has been a highly intriguing area. Rather than a uniform dialect, a so-called West Central German dialect continuum between two varieties was discovered: Moselle Franconian (in German, Moselfränkisch) in the northwest and Rhine Franconian (Rheinfränkisch) in the southeast.<sup>7</sup> One of several distinguishing features separating these two dialects is the dat-das line, which is also referred to as the Hunsrück Barrier (in German, Hunsrück-Schranke). Dialects to the north of this isogloss have a consonant t in words dat 'that' and wat 'what' similar to what appears in English, while those to the south have an s in das and was as in Standard German.<sup>8</sup>

Both dialects, Moselle Franconian and Rhine Franconian, were spoken by the majority of the early settlers and form(ed) the basis of Hunsrik—the new variety that developed in Southern Brazil. Over time, it has become a supra-regional koine<sup>9</sup> and a kind of relatively uniform Dachsprache (usually translated as 'umbrella/roofing language') within the German-speaking community.<sup>10</sup> However, due to the social-geographical heterogeneity of the immigrants, several (relic) language markers of the many other original varieties can still be found in use—primarily in (family) sociolects and basilectal/micro-regional varieties. The research literature mentions the following dialects, which do not belong to the

<sup>6</sup> Following Koch (cf. 1974: 20) and Altenhofen (cf. 1996: 50, 75–76), we call the areas around São Leopoldo, Santa Cruz do Sul, and Agudo in Southeastern Rio Grande do Sul already settled at the beginning of the 19th century 'old colonies'. The 'new colonies' founded first in 1890 are located in the northwest of the state around the city of Santa Rosa near the border with Argentina. From a linguistic point of view, this distinction is essential insofar as some linguistic differences between the two areas can be observed.

<sup>7</sup> For the division of German dialects, see i.a. Wiesinger (1983).

<sup>8</sup> For this and other relevant dialectal isoglosses of the so-called Rhenish fan (in German, Rheinischer Fächer), see Beckers (1980: 468–469). Furthermore, the book edited by Russ (1990) can interest English-speaking readers as an introduction to German dialectology.

<sup>9</sup> According to Kerswill (2018: 519; italics in original), by the term koine, we understand an outcome of koineization: "a [language; MM] contact-induced process that leads to quite rapid, and occasionally dramatic, change. Through it, new varieties of a language are brought about as a result of contact between speakers of mutually intelligible varieties of that language. Koineization is a particular case of dialect contact (Trudgill 1986). Typically, it occurs in new settlements[,] to which people have migrated from different parts of a single language area", "in the context of increased interaction or integration among speakers" (Siegel 2001: 175). "By contrast with the original Greek regional koine, a new dialect in a new settlement is an immigrant koine, which, once established, becomes the vernacular of the new community." (Kerswill 2018: 520; italics in original)

<sup>10</sup> Auer (2005: 59) suggests the use of "the more neutral term S(outhern)B(razilian)G(erman) Koiné" (italics: MM) instead of the geographically motivated Hunsrik.

West Central German dialect group, including Moselle Franconian and Rhine Franconian: Pomeranian and Westphalian (Low German varieties spoken in the northern European German-speaking area), Alemannic, Bavarian, East Franconian, and Swabian (Upper German varieties spoken primarily in the South), as well as extraterritorial German varieties such as Volga German and Volhynia German (cf. Altenhofen 1996: 26; Fausel 1959: 7–9; Ziegler 1996: 45–46). Some descendants of Low German-speaking immigrants have continued to use their dialects of family origin all along (cf. Auer 2005: 59).

#### 2.2 Language (Contact) Constellation

In this section, we take a look at the language contact<sup>11</sup> constellation in Southern Brazil. In addition to the further dialects of German mentioned above, Hunsrik comes into contact with other minority languages such as Dutch, Guarani, Italian, Japanese, Kaingang, Polish, Spanish, and here, in particular, their nonstandard varieties.<sup>12</sup> Nonetheless, the major contact language is—of course— (Brazilian) Portuguese, which is the national and official language. Most phenomena that are an outcome of contact<sup>13</sup> between Portuguese and Hunsrik can be located in the lexis, as Riehl (cf. 2014: 96) suggested. According to her observations, the lexical level is followed by syntax, phonology/prosody, and morphology.<sup>14</sup> Although transfer phenomena occur in Hunsrik at all these levels,

<sup>11</sup> Following Riehl's (2004: 12; translation: MM) definition, I understand language contact as "the mutual influence of two or more languages[/varieties of language(s); MM]. […] There are two directions: on the one hand, the influence of the first language (in the sense of the first language learned) on the second language; and on the other hand, the influence of the second language on the first language. All types of language contact phenomena are manifested in individual utterances and the usage of a multilingual speech community." ["die gegenseitige Beeinflussung von zwei oder mehreren Sprachen. Dabei gibt es zwei Richtungen: einmal den Einfluss der Erstsprache (im Sinne der zuerst gelernten Sprache) auf die Zweitsprache und zum anderen den Einfluss der Zweitsprache auf die Erstsprache. […] Alle Arten von Sprachkontakterscheinungen machen sich sowohl in individuellen Sprachäußerungen als auch im Sprachgebrauch einer mehrsprachigen Sprachgemeinschaft bemerkbar."]

<sup>12</sup> "Ethnologue" provides a map showing the linguistic situation in Southern Brazil and reports some further languages spoken in Rio Grande do Sul (cf. "Southern Brazil").

<sup>13</sup> For the (fundamental) theoretical and terminological debate in the field of language contact research, see Section 4.

<sup>14</sup> It goes without saying that this 'order' is an attempt at generalization. The extent of language transfer depends on several factors. Not only the typology of partaking languages is relevant, but also a whole range of extralinguistic aspects. Significant can be, amongst others, degree of contact intensity, speakers' attitude towards languages/varieties concerned, age of the speakers, duration of the language contact situation to date, administrative-geographical conditions. Thus, the language contact and the interference may well diverge with respect to the same grammatical phenomenon depending on language, area, speech community and behave differently.

it is wrong—at least from the perspective of modern contact linguistics—to speak of this variety as a mixed language, as Bossmann (cf. 1953: 96–100) and Fausel (cf. 1959: 6–9) do, amongst others. In this context, it is worth noting that the mixed language debate is highly complex and there are currently several definitions (and questions) that are presented and discussed in a quite detailed and generally clear way; for example, by Matras (2000) and Matras / Backer (2003).

According to the research literature, "mixed languages do not fit within the genetic model and therefore cannot be classified genetically at all" (Thomason / Kaufman 1988: 3). These are varieties "that emerged in situations of community bilingualism, and whose structures show an etymological split that is not marginal, but dominant, so that it is difficult to define the variety's linguistic parentage as involving just one ancestor language" (Matras / Backer 2003: 1). Hunsrik does not meet these criteria: Firstly, it did not primarily emerge as a result of intensive contact between two languages. Secondly, the grammar base of the new variety is not mutually determined by both 'donor' languages. In conclusion, I suggest categorizing Riograndese Hunsrik as a supra-regional (dialect) koine that has a clearly recognizable West Central German base and is simultaneously exposed to the (significant) influence of the national language Brazilian Portuguese.

To give an understanding of the language situation in the German-speaking communities in Southern Brazil, we can use the model of Auer (cf. 2005: 59). It takes three (types of) varieties into account: Portuguese, Riograndese Hunsrik —called Southern Brazilian German Koiné, as mentioned above—and original German dialects like Westphalian and Pomeranian.

"The local repertoire has a two-dimensional structure, with Portuguese and German providing one dimension, and more and less prestigious forms of Portuguese and German respectively the other. Variation within present-day SBGK is only rarely due to the various traditional dialects the settlers brought along from Germany (horizontal variation). Rather, SBGK shows a strong acrolectal/basilectal variability, i. e. some forms are more prestigious than others. The latter happen to be closer to Standard German (a variety which is hardly present in the community today)." (Auer 2005: 59–60)

In a sense, this proposition validates the previously mentioned idea that Hunsrik cannot be seen as a mixed language. As demonstrated, Hunsrik has a clear German basis; and the impact of Portuguese appears in its more basilectal forms, principally. By comparison, the influence of German on Brazilian Portuguese is weaker, but still noticeable, primarily in everyday language.

Riograndese Hunsrik and Brazilian Portuguese do not 'compete' with regard to their functionality/'universality'(and prestige). In contrast, they coexist across various language domains. Nowadays, Hunsrik is spoken primarily in rural or small urban areas. Typical contexts are everyday conversations among family members, friends, neighbors, work colleagues, and members of local clubs/associations. Portuguese is preferred in (semi-)official situations that extend beyond the dialectal lexical inventory, as well as when in contact with foreigners, strangers, and Luso-Brazilians.

### 3 State of Research

A detailed and—at least at the time of publication—very up to date overview of the state of research on German (varieties) spoken in Latin America and in particular Brazil is given by Rosenberg (2018) in his article published in a handbook of German linguistic minorities overseas (Plewnia / Riehl 2018).<sup>15</sup> In the present work, the discussion is limited to the most relevant publications.

Despite a large number of speakers, there is very little awareness about the German ('language island') variety from Southern Brazil not only among 'linguistic laypeople', but also among linguists. A quite limited research literature contributes to this relative obscurity. After a few decades of scientific stagnation in this area<sup>16</sup>, the linguistic discussion on Riograndese Hunsrik experienced somewhat of a revival in the late 1990s as, within two years, four relatively extensive monographs, Altenhofen (1996), Ziegler (1996), Damke (1997), and Tornquist (1997), were published.<sup>17</sup> The mentioned studies are particularly concerned with the topic of language contact between the regional variety of German and 'surrounding' Brazilian Portuguese. In addition to language contact research, the focus is on various language system levels—especially lexicology, phonetics, and phonology—as well as on aspects of sociolinguistic.

The papers presented at the beginning of the 21st century, such as Rosenberg (2003; 2005) and Auer (2005) explore the goals of modern variationist linguistics and break with the traditional patterns of (historical) 'language island' research (Sprachinselforschung)—this has had a positive long-term effect on the Riograndese Hunsrik research. Although the number of publications still leaves much to be desired, (specific) studies from the morphological and syntactic areas are

<sup>15</sup> In addition, I refer readers to the chapter on the Hunsrik-speaking community in the significant work on the position of the German language in the world by Ammon (cf. 2015: 369– 380).

<sup>16</sup> After the publication of Koch's (1974) work, no essential papers can be named that explicitly address the linguistic aspects of Riograndese Hunsrik. The only publications from this period treat either historical questions of the German emigration to Brazil or the social situation of the descendants of German settlers. Some publications discuss incidental (socio)linguistic aspects such as the group- and identity-forming role of the dialect.

<sup>17</sup> There are also a few articles by Gärtner: e. g., Gärtner (1996; 1997), Born / Gärtner (1998), and Gärtner / Cunha (1998).

slowly gaining in importance, e. g., Maselko (2013a; 2015; 2017; 2018), Rosenberg (2016; in this volume). Beyond these concerns, the focus remains on (general) sociolinguistic and contact linguistic issues. In this area, the following publications are worth mentioning: Altenhofen (2016), Pupp Spinasse (2016), Kaufmann (2017). Recently, a study by Altenhofen / Morello (2018) has been published which I consider a milestone in research on the Hunsrik dialect. The study covers—to a varying extent—various (sub)disciplines: history, anthropology, sociology, sociolinguistics, phonology, and grammar.

The research literature on grammaticalized functions of the give-verb proves to be limited. In the empirical part of this paper, I will refer to the few existing individual studies as well as reference works on the investigated languages and their varieties (German and Portuguese) such as dictionaries and grammars. At this point, I shall mention three scientific works that form the theoretical basis of my research in this field: Newman (1996) examining several languages around the world, Lenz (2007) for German, and Martins Salomão (1990) for (Brazilian) Portuguese.

#### 4 Theoretical Principles on Borrowing

The coexistence of two (or more) languages or their varieties in one area may lead, on the one hand, to bi-/multilingualism of the local population and, on the other hand, to "a change in the structural inventory of at least one of the languages involved, and sometimes of both" (Matras 2009: 146). Bilinguals are not able to completely hide one of the spoken languages, which often results in "a kind of import of a structure or form from one language system into another. The process is best known as 'borrowing'" (Matras 2009: 146; italics: MM) or transference (cf. Clyne 1975:16), whereas "items affected by it are called 'borrowings', 'loans' or 'transfers', and the languages involved are frequently labelled, according to their roles, 'donor' and 'recipient'" (Matras 2009: 146; italics: MM).

In this paper, I refer to the terminology which is commonly used in the foreign/second language didactics or multilingualism research and initially goes back to Weinreich (1953).

According to him, adult speakers

"process their second language via their knowledge of their first language, at least initially. Their familiarity with the principles of complex expression of ideas in theirfirst language often serves as a foundation on which they can build when acquiring the L2; indeed some structures of their L1 may even be helpful in understanding corresponding structures of L2" (Matras 2009: 72).

Based on the degree of similarity of involved language systems, we can distinguish between positive and negative transfer/'borrowing'. Positive transfer occurs when certain (grammatical) components in the L1 and L2 are similar or the same, and thereby single phenomena may be transferred from the L2 into the L1 or vice versa (relatively easily). Negative transfer occurs when particular (grammatical) elements of associated languages are different, and phenomena from the L1 can be found in the L2, but seldom vice versa<sup>18</sup> (cf. Müller / Kupisch / Schmitz / Cantone 2011: 22).

While I do not reject the terminology presented, I consider it necessary to draw attention to the fact that, in specific (empirical) cases, it can cause difficulties and may not prove suitable. The reason for this is that the theory firstly requires clear assignments of contact phenomena to specified transfer types, and secondly, it leaves little room for 'interpretation' or alternative/extended defining. This is especially the case when a particular grammatical construction (potentially) exists in both language systems but to a different degree.

"In speech communities [like in the German-speaking area in Southern Brazil] speakers in which the native language [for our purposes better: L1] is always or nearly always spoken in a bilingual mode, since everyone is bilingual, the permanent license to integrate[—at first sight ']foreign['—]grammatical operators [from the L2] can lead to long-term integration of such operators into the recipient language." (Matras 2009: 151– 152; additions and adjustments: MM)

The integration of a specific construction is favored by contact with a language in which it is (more) frequently used. In other words, its use in the 'recipient' language becomes stimulated or activated through frequent use in the 'donor' language.<sup>19</sup> In such cases, talking about a 'real' transfer is at least controversial. For this reason, I suggest using an alternative term whose definition is more open and not as strict as transfer; in my opinion, supportive contact (phenomenon) is well suited for this purpose.

<sup>18</sup> As mentioned in 2.2, most phenomena that are an outcome of contact are lexical. But as Thomason / Kaufman (1988: 37) point out, "[i]f there is strong long-term cultural pressure from source-language speakers on the borrowing-language speaker group, then structural features may be borrowed as well—phonological, phonetic, and syntactic elements, and even (though more rarely) features of the inflectional morphology."

<sup>19</sup> "There is pressure on the bilingual to simplify the [constructions'; MM] selection procedure by reducing the degree of separation between the subsets of the repertoire, allowing the two 'languages' to converge." (Matras 2009: 151)

# 5 GIVE in German (Hunsrik Dialect) and Brazilian Portuguese

#### 5.1 Introductory Theoretical Discussion: Grammaticalized geben-Variants

In some (West Central) German varieties as well as in (Moselle Franconiandetermined) Luxembourgish, the function of geben 'to give' goes far beyond that of a trivalent full verb.<sup>20</sup> In the national language of Luxembourg, the give-verb has undergone the currently highest degree of grammaticalization. For this paper, the term "grammaticalization" is taken to mean the "transformation from a (more) lexical relatively autonomous sign to a grammatical, less autonomous sign" (Nübling 2006: 172; translation: MM)<sup>21</sup> and the "process of emergence and further development of grammatical morphemes up until their demise" (Szczepaniak 2011: 5; translation: MM)<sup>22</sup>, which is initiated by "speakers using lexemes that convey concrete contents for expressing grammatical, i. e., abstract contents" (Szczepaniak 2011: 5; translation: MM)<sup>23</sup> .

Table 1 shows a range of grammaticalized geben-variants. While in the standard language the verb geben has only three (German: [1]–[3]) or six functions (Luxembourgish: [1]–[5], [7]) respectively, at least two more ([4]–[5]) appear in some substandard regional dialects (cf. Lenz 2007: 53). The last function [6] may already be extinct or used only strongly archaic in (very few) German ('island') dialects. In this paper, we will turn our attention to [3] and [4] (as well as [6]—but only in the form of excursus in 5.1.2), namely geben as existential verb and copula (as well as future tense auxiliary).

<sup>20</sup> For the polyfunctionality of geben in the German standard language variety, see "Duden. Das große Wörterbuch der deutschen Sprache" (2012), and "Duden. Deutsches Universalwörterbuch" (2015).

<sup>21</sup> "Wandel von einem lexikalisch(er)en, relativ autonomen Zeichen zu einem grammatischen, weniger autonomen Zeichen".

<sup>22</sup> "Prozess der Entstehung und Weiterentwicklung grammatischer Morpheme bis hin zu ihrem Untergang".

<sup>23</sup> "Sprecher mit Hilfe von Lexemen, die konkrete Inhalte transportieren, grammatische, also abstrakte Inhalte zum Ausdruck bringen".


Table 1: Grammaticalized geben-variants in varieties of German and Luxembourgish

#### 5.2 GIVE-Existential Verb

#### 5.2.1 Types of Existential Clauses in German

As introduced in 5.1, one of the functions the grammaticalized verb geben has in German is as an existential verb. In the standard German language as well as in most diatopic varieties of German, existential clauses usually use the expletive pronoun es<sup>24</sup> and the verb geben in its finite form 3SING.<sup>25</sup> Due to the semantic-

<sup>24</sup> It "does not refer to any specific entity in the scene or event being described" (Newman 1996: 160).

<sup>25</sup> Especially the Alemannic (dialect-)speaking area in Southwestern Germany, Germanspeaking Switzerland and Western Austria constitute the exception to the use of existential clauses. While in other speech areas almost exclusively the construction es gibt is used, in the southwest of the German-speaking area, existential clauses are formed with the expletive pronoun es and the conjugated haben 'to have' 3SG: es hat (literally 'it has'). The—slightly older—distribution map by König / Elspaß / Möller (cf. 2015: 243) coincides for the most part with that of the "Atlas zur deutschen Altagssprache" (AdA) 'Atlas of German Everyday

functional motivation, we distinguish three types of the existential clause es gibt (literally 'it gives'). Aslisted in Table 2, the construction can be used to express [8] the occurrence of an event or the emergence of an entity, [9] generic existence or [10] presence/availability (cf. Lenz 2007: 56).


'In Australia there are kangaroos'

Table 2: Types of existential clauses in German (standard language)

The fact that all three types named occur in Riograndese Hunsrik (cf. Boll 2018: 53) is not self-evident because there are German dialects in which this is not the case. For example, in another West German-based ('language island') dialect, Pennsylvania German spoken in the United States and Canada, geben is used for ingressive statements, whereas sein 'to be' is used for generic.<sup>26</sup>

First of all, we shall take a look at the ingressive constructions. In a study on tense and mood in the German of Southern Brazil (Maselko 2013a: 116–117), the interviewed Hunsrik speakers (n=198) were asked for a preferred variant to announce tomorrow's rain, using a multiple-choice task ([11a–d]). However, even if the question was targeted at tense use in forward-looking utterances, we can make statements about the use of the es gibt-construction because it represented one of the possible answers, namely [11b].

Language' (cf. "geben/haben (Frage 4c)" 2011). An analogous grammatical construction including the have-verb we encounter in neighboring French (il y a), which may be an indication of language contact. Even the "Variantenwörterbuch des Deutschen" ('Variant Dictionary of German' which is a dictionary of German national and regional standard language varieties) leads under the lemma "haben" with its existential meaning, but only in the sense 'to be present/in place'. Its use is restricted to Western Austria, Switzerland, and Southwestern Germany, namely the Alemannic-speaking area (cf. Ammon / Bickel / Lenz 2016: 305). On the AdA-map, there is also isolated evidence for the third variant es ist/sind (literally 'it is/are'), first and foremost from Bavaria in Germany and Austria (cf. "geben/ haben (Frage 4c)" 2011).

<sup>26</sup> For the alternation between es gibt and es ist/sind from a morphosyntactic and semantic perspective, see Newman (1996: 161–164).

[11a] Es reent mooye.<sup>27</sup> it rains tomorrow [11b] Es kipt mooye reen. it gives tomorrow rain:ACC [11c] Es tuut mooye reene. it does tomorrow rain:INF [11d] Es weet mooye reene. it will:3SG tomorrow rain:INF



Table 3: Distribution of es gibt-existential clause in terms of a future occurrence

As can be seen in Table 3, more than one-third of the informants chose the answer variant with the existential clause. This quite high number would seem to suggest advanced grammaticalization of es gibt in Hunsrik—at least in the sense of '(future) occurrence'. The fact that the variant with existential clause was chosen most commonly is not unexpected, considering that the ingressive/inchoative meaning is a feature of the German future tense since it has emerged (cf. Bogner 2009: 101–104).

### 5.2.2 Excursus: geben-Future Tense in West Central German

As shown in 5.1, geben is (potentially) suitable as a future tense auxiliary verb. For this reason, the (syntactic-semantic) characteristic of the verb geben cannot be ignored when talking about its ingressive existential function.

The "Rheinisches Wörterbuch" (RhWB) 'Rhenish Dictionary' which covers, among other things, the area of (historical) origin of the German speakers in Southern Brazil<sup>28</sup>, indicates in its quite extensive article that in the southwest of

<sup>27</sup> When writing down the Riograndese Hunsrik's phrases, I follow the rules of Wiesemann (2008). Maselko (2013b) discusses these orthographic rules in detail and gives a contrastive overview of other writing systems regarding the German dialect of Southern Brazil.

<sup>28</sup> With its nine volumes, the RhWB is the most comprehensive dialect dictionary of West Central German. It lists (almost) all the words that were or are still in use in the dialect since the 19th century.

the Moselle Franconian and Rhine Franconian dialect continuum<sup>29</sup> , geben can replace werden 'to become' as an auxiliary verb for the future tense (cf. "Rheinisches Wörterbuch" 1931: 1075–1076). The example sentences presented in the RhWB are given as [12]–[14].

[12] Et get reənen. it gives rain:INF 'It will rain.' [13] E get sterwen. he gives die 'He will die.' [14] Es gebt gemacht gen. it gives made:PP give:INF 'It will be done.'

The publications cited above, Lenz (2009) and Newman (1996), can be considered as (synchronic oriented) reference works on the grammaticalized geben. However, they do not indicate its function as an auxiliary verb for the future tense. In all likelihood, the reason is that the future tense construction geben + INF no longer exists in the modern dialects spoken in Germany. As discussed, this reading seems to be confined to a tiny area of West Central German. Furthermore, the form was most likely accompanied by stigma at the time it was still actively used—also in the adjacent (dialect contact) areas. This is how I understand the final remark "not used to form the future tense" (translation: MM)<sup>30</sup> in the dictionary entry "gin" 'to give' of the "Luxemburger Wörterbuch" (LWB) 'Luxembourg Dictionary' (1955–1962), which may have a descriptive function primarily, but plays a prescriptive role as well.

Surprisingly, the (limited) use of the analytical geben-future can be demonstrated for Riograndese Hunsrik (cf. Maselko 2013a: 153–156). Although the number of collected data (two examples in [15]–[16]) may seem very low, the fact of being able to find some evidence at all should not be underestimated.

[15] Sii kep te plume waser kepe. she gives the:PL flowers:ACC water give:INF 'She will water the flowers.'

<sup>29</sup> Strictly speaking, the RhWB-article on "geben" names the following counties (in German, Landkreise): Ottweiler, Saarlouis, and Saarbrücken.

<sup>30</sup> "nicht zur Bildung des Futurums gebraucht".

[16] Ti Lucia kep tii plume nas mache. the:FEM Lucia:NOM gives the:PL flowers:ACC wet make:INF 'Lucia will water the flowers.'

This statement is particularly appropriate when the task type is taken into account. As an image description task was used, the respondents had no answer options to choose from. The informants should imagine the pictured activity of flower watering is going to happen tomorrow, and then describe the act with one sentence. For those two informants who expressed it by combing the auxiliary verb geben and an infinitive, this analytical construction seems to be the prototypical variant when referring to the future. Presumably (clearly) more informants would have chosen the geben-future tense if they only had to tick the potential expression variants and not to write them down themselves. However, it is out of the question that the futuristic or ingressive meaning is inherent in the verb geben.

#### 5.2.3 Ingressive dar-Existential Clause in Brazilian Portuguese

In addition to the German language, the impersonal give-constructions can only be found in two other languages, as Newman (cf. 1996: 160) states. While the existential meaning remains practically the same in all of them, the discrepancies appear at the formal level. Unlike in German, a subject phrase does not have an expletive it-pronoun in Brazilian Portuguese and Jakaltek<sup>31</sup>, but avoids one altogether.

Now we turn our attention to Brazilian Portuguese and look at two (Englishlanguage) grammars of Portuguese spoken in South America. Thomas (cf. 1974: 45) writes about the use of dar 'to give' in his publication focused on the spoken variety as follows: "This verb, in addition to its basic meaning, is used in a great number of idiomatic expressions, in many of which other verbs are used in English." Whitlam (cf. 2011: 175; italics: MM) makes a similar observation: "The verb dar is used impersonally with a number of different meanings, especially in the spoken language". <sup>32</sup> In contrast to Thomas (1974), Whitlam touches upon the

<sup>31</sup> Jakaltek, also known as Jakalteko or Popti', is a Mayan language of Guatemala spoken by 90,000 people especially in the county of Huehuetenango.

<sup>32</sup> The internationally recognized "Dicionário Aurélio da Língua Portuguesa" (2010) 'Aurélio Dictionary of the Portuguese Language' lists a total of 103 dar-readings. In addition, there are entries for 17 idiomatic expressions using the verb dar. Nevertheless, it is difficult to find a suitable entry for the existential construction in this quantity. Most likely, we can refer to the following readings: 'ir de encontro; bater' 'come upon; hit' (O navio deu no recife 'The ship hit the reef '), and 'bater, soar' 'to beat, sound'(E a noite ia se passando. Deram dez horas'And the night was going on. It was ten o'clock'). For some grammaticalized variants of dar, see, amongst others, Coelho / Paula Silva (2014), Martins Salomão (1990), and Souza (2016).

reading of dar we are focusing on.<sup>33</sup> However, he does not treat the phenomenon in the same way in both editions of his grammar and made a remarkable substantive amendment. While in the first edition the use of the dar-construction is presented as reduced to specific idiomatic expressions with a resultative meaning (Whitlam 2011)<sup>34</sup>, it is considered as a full equivalent to the English ingressive existential clause there is/are in the second edition (Whitlam 2017)<sup>35</sup> .

The ingressive semantics of impersonal dar in Brazilian Portuguese is an issue in Martins Salomão's study (1990: 67). According to her, "[e]xistentialconstructions with dar are based on the [c]ausation construction through [m]etaphor […] [']existing things are effects of their causes[']." (quotes added, italics: MM) Similar to Newman (cf. 1996: 160), she points out,

"that they are subject[l]ess [sic!] clauses: the structural slot saved for the [c]auseargument is left vacant and, differently from the full-fledged existential-causative patterns, cause is never named. Our [e]xpressability [p]rinciple would account for that via condition. The [e]xisting [o]bject, however, is pretty much similar to the [e]ffect of [c]ausation, and it comes to be treated as a [p]roperty acquired by the [a]ffected [p]arty." (Martins Salomão 1990: 67)

The use of an impersonal give-construction in the sense of describing the manifestation of a thing in Brazilian Portuguese is illustrated in [17]–[22].

[17] Se der algum problema, te aviso.

if gave:3SGSJVFUT some problem you:ACC announce:1SG

'If there is any problem, I will let you know.'

[18] Eu sabia que ela ia desistir, e não deu outra.

I knew:1SGPRET that she went:3SGPRET give up:INF and not gave:3SGPRET other:FEM

'I knew she would back out, and that is exactly what happened.'

[19] Deu zebra no jogo de ontem.

gave:3SGPRET zebra in:the:MASC game of yesterday

'There was an upset in yesterday's game.'

(Whitlam 2017: 225)

[20] Deu um barulho na televisão.

gave:3SGPRET a:MASC noice in:the:FEM television

'There wasastrange noise in the television.'

<sup>33</sup> "The Routledge Portuguese Bilingual Dictionary" which covers the Brazilian variety of Portuguese, remarkably does not list the existential reading in its extensive entry on the verb dar (cf. Allen 2014: 97–98). It is worth noting that the dictionary has the same publisher as the grammar by Whitlam (2011; 2017) cited here. It also suggests that the ingressive dar-existential clause is mainly a phenomenon of spoken Brazilian Portuguese.

<sup>34</sup> "'to result in', used in idiomatic expressions" (Whitlam 2011: 175).

<sup>35</sup> "'there is/are' in the sense of 'there occur(s)'" (Whitlam 2017: 225).

[21] Deu chuva o fim de semana inteiro. gave:3SGPRET rain the:MASC end of week entire 'It rained throughout the weekend.' (Martins Salomão 1990: 24) [22] Deu praga na goiabeira. gave:3SGPRET disease in:the:FEM guava-tree 'There is a disease in the guava-tree.' (Martins Salomão 1990: 68)

Sentence [21] reminds us of the Hunsrik task Example [11b] Es kipt mooye reen discussed in 5.2.1. Both are weather expressions describing the 'existence' of meteorological conditions. This seems to be a common context to use an impersonal give-construction. According to Newman (cf. 1996: 165–166), it applies at least to one more language, namely, Jakaltek. The so-called 'weather constructions' have a special status in the study on dar in Brazilian Portuguese presented by Martins Salomão (cf. 1990: 75) who regards these as a subtype of existential constructions with a modified meaning. Their semantics can be illustrated by the following metaphors: "locations are affected by the transference of existing beings into their domains" (Martins Salomão 1990: 73), and "the weather is an effect brought into existence" (Martins Salomão 1990: 73).

Even though the current state of research on existential constructions does not allow us to speak of an apparent contact-induced interference or borrowing from German (Hunsrik dialect) into (Brazilian) Portuguese, we can at least assume an existing language contact constellation which supports the occurrence of (ingressive) dar-existential clauses in Portuguese of Brazil and contributes to their relatively stable use—in particular—in the spoken variety. We are dealing with a kind of linguistic convergence situation and supportive contact phenomenon, as terminologically suggested in 4.

#### 5.3 GIVE-Copula

#### 5.3.1 geben-Copula in West Central German

As introduced in 5.1, another function the verb geben can have in the nonstandard varieties of (West Central) German<sup>36</sup> is that of the copula meaning 'to become'. Drenda (cf. 2019: 131–132) shows specifically for European Hunsrik (Hunsrücker Platt) that the geben-variant is quite commonly used there. How-

<sup>36</sup> The references proving the presence of the analyzed reading are mainly (historical) dialect dictionaries such as Schön (cf. 1922: 57) for the dialect of the Saarbrücken region, the LWB, and the RhWB.

ever, it occurs aside from werden, which is also anchored in the standard German language.

Regrettably, due to gaps in the research literature and not entirely-satisfactory results of historical studies, the origin of the construction is more or less hypothetical. The grammaticalization of the ditransitive main verb geben to inchoative copula presumably began in the 15th century (cf. Nübling 2006: 193).<sup>37</sup> In any case, this is likely to be a (relatively) recent language development and thus a much younger copula variant than werden (cf. Bellmann 1998: 247). However, as observed by Nübling (cf. 2006: 193), the agentive geben with its evident 'action perspective', does not represent the ideal 'candidate' for a copula, which clearly tends towards the 'process/occurrence perspective'. The verb geben "had initially ridden of two of its three actants (intransitivization) and at the same time had changed or neutralized the complex, causative forwarding movement, in order to reverse the polarity of directionality." (Nübling 2006: 194; translation: MM)<sup>38</sup> This seems to be the only possible route from German ich gebe dir etwas 'I give you something' to ich gebe krank 'I am getting sick', literally 'I give (= become) sick'. 39 In this context, it is relevant for further analysis to introduce a thesis by Nübling (cf. 2006: 195–197)<sup>40</sup>. According to Nübling (2006),aseries of diachronic and synchronous aspects of geben-use in West Central German, as well as in the vernacular and standard German language, testify that it did not immediately appear with a predicative adjective but had first come together with a predicative nominal.<sup>41</sup>

<sup>37</sup> For grammaticalization of geben, see also Gaeta (2005), Girnth (2000: 137–145), and Lenz (2018: 203–208).

<sup>38</sup> "zunächst zweier seiner drei Aktanten entledigt (Intransitivitierung) und gleichzeitig die komplexe, kausative Beförderungsbewegung entdirektionalisiert bzw. neutralisiert, um anschließend direktional umgepolt werden zu können".

Gaeta (cf. 2005: 196) speaks of an inevitable loss of semantic specificity, especially agentivity. 39 The transfer from a lexical, decidedly directional verb to a grammatical marker with an abstract, general movement meaning is also evident in the structural-syntactic level in the formation of the present perfect tense. While the main verb uses haben 'to have' as auxiliary (e. g. ich habe dir ein Buch gegeben 'I gave you a book', literally 'I have given you a book'), the copula forms the present perfect tense with sein 'to be'—and thus in the same way as the werden-copula in Standard German. Gaeta (cf. 2005: 196) and Nübling (cf. 2006: 194) consider this an indicator of the high degree of grammaticalization of the construction.

<sup>40</sup> The thesis is followed up, amongst others, by Lenz (cf. 2007: 68–69; 2018: 204) in her papers on the use of grammaticalized geben in (non-standard) German.

<sup>41</sup> It is worth to take a closer look at the circumstances in West Central German. Above all, the frequency and spatial aspects can help us to understand the phenomenon. According to the RhWB (cf. 1931: 1075), geben is used as a copula in Rhine Franconian, Moselle Franconian, and Ripuarian. Bellmann (cf. 1998: 247) provides a distribution map of geben as a copula with the adjective complement drawing upon data from the "Mittelrheinischer Sprachatlas" (MRhSA) 'Middle Rhine Language Atlas'. As is demonstrated there, the construction is used in the southwestern area of Moselle Franconian. The remaining areas as well as Rhine Franconian prefer werden similar to the standard language. Bellmann (cf. 1998: 259–262)

In Riograndese Hunsrik<sup>42</sup> , geben can be used in conjunction with both a noun and an adjective which distinguishes this dialect from, for example, another West Central German-based variety spoken in America, Pennsylvania German. Semantically, the verb shows—once again—its (obviously inherent) ingressive character. But unlike in Rhine Franconian, where there is a semantic-pragmatic geben/werden-opposition,<sup>43</sup> the prognostic sense is not a mandatory feature of the geben-copula in Hunsrik and constitutes only one of the semantic variants. This corresponds to the (southwest) Moselle Franconian grammatical system, in which geben encounters no variation and appears to be more established and grammaticalized as a copula.

Examples [23]–[30] presented below, which I collected in Alto Feliz ('old colonies' in Rio Grande do Sul) and Bom Princípio Baixo ('new colonies'), might clarify the correlation between the copula geben and its ingressive nature. As can be seen from most of the examples below, "we are dealing with a change in the state of a person which is not determined organically, but rather by other factors. In other words, 'becoming' may be construed as a type of emergence of a new form of a thing, but not restricted to organic growth." (Newman 1996: 169)

[23] Te kep xuul profesor. the:MASC gives:COP school teacher:PREDN 'He becomes a schoolteacher.'

shows that, in contrast to the copula with an adjective, no definite boundary can be drawn between the variants geben and werden with regard to the copula with a noun. In this case, we have to deal with continuity so that in northeastern Moselle Franconian and Rhine Franconian (at least on the left bank of the Rhine river), particular use of geben—especially in the fixed phrases—can be established, even if werden is dominant. Unfortunately, there is no map for the copula geben accompanied by a predicative nominal. Drenda (cf. 2019: 132) provides a general map of the geben/werden-copula distribution in the European Hunsrück area. He ignores the difference between a predicative adjective and a predicative nominal.

<sup>42</sup> The content of the two (popular science) Hunsrik dictionaries "Dicionário Português – Renano Hunsrik" 'Dictionary Portuguese – Rhenish Hunsrik' and "Dicionário Hunsriqueano Brasileiro – Portugueˆs" 'Dictionary Brazilian Hunsrik – Portuguese' diverge concerning copula 'to become'. Kuster-Cid / Kuster Cid (cf. 2018: 140) state both geben and werden as equivalent Hunsrik variants for tornar-se, ficar, virar (in Portuguese), whereas Boll (cf. 2018: 53, 120) mentions only the werden-copula.

<sup>43</sup> "The opposition is concerned with the possibility of realizing the dichotomy between realis and potentialis. […] [In Rhine Franconian; MM], in contrast to werden, this verb as a copula does not mean 'become per se' but 'maybe become' with more or less limited degree of probability." ["[E]s geht bei der Opposition um eine Möglichkeit der Realisierung des Gegensatzes zwischen Realis und Potentialis. […] [Im Rheinfränkischen; MM] bedeutet dieses Verb – im Gegensatz zu werden – als Kopula nicht 'werden schlechthin', sondern 'vielleicht werden' mit mehr oder weniger eingeschränktem Wahrscheinlichkeitsgrad."] (Bellmann 1998: 262; translation: MM)

[24] Te is pfarer geb. the:MASC is:COPAUXPRF priest:PREDN given:COPPP 'He became a priest.' [25] Tii kewe rayche loyt. the:PL give:COP rich people:PREDN 'They become rich people.' [26] Tan kepe ich huntert yoore. then give:COP I hundred years:PREDN 'Then I will be one hundred years old.' [27] Morkens kept ma frii vach. morning-SUF gives:COP you early awake:PREDADJ 'In the morning you get up early.' [28] Tu kipxt xwach un tan kipxt krank. you give:COP weak:PREDADJ and then give:COP sick:PREDADJ 'You become weak and then you get sick.' [29] Tas kep peeser. it gives:COP better:PREDADJ 'It gets better'. [30] Tii kinter sin katolix kep. the:PL kids are:COPAUXPRF Catholic:PREDADJ given:COPPP 'The kids became Catholic.'

The definite determination of copula's ingressive meaning 'realis' vs. 'potentialis' is mostly only feasible if the situational context is known. For semantic-formal reasons, however, the prognostic-speculative semantics is definitely to be excluded in the case of Examples [24] and [30], which are formulated in the present perfect tense.<sup>44</sup> In all likelihood, this can also be assumed from two further examples, namely [27] (an action taking place in the near future) and [23] (the consequence of a logical action/process).

While the copula geben appears in the first four phrases [23]–[26] with a predicative nominal, the other sentences [27]–[30] provide examples of the use with a predicative adjective. In the written questionnaire I did not foresee a separate task for the copula. Nevertheless, an image description task on progressive constructions<sup>45</sup> has provided quite impressive and qualitatively significant data, which I shall now discuss.

<sup>44</sup> As mentioned above, the use of the auxiliary verb sein indicates a high degree of grammaticalization of the geben-copula in Riograndese Hunsrik.

<sup>45</sup> The picture showed a woman in bed just waking up and yawning.

Out of 163 informants<sup>46</sup> who framed an answer at all, almost exactly half of them produced a copula construction. Considering the fact that the respondents had been given a free task without pre-formulated answer options, it is a considerable number. The answers vary slightly in wording or choice of tense, but all include the expression 'to become awake' (see Examples [31]–[32]).

[31a] Tii frau is im pet un kep wach. the:FEM woman is in:the bed and becomes:COP awake:PREDADJ 'The woman is in bed and wakes up.' [31b] Tii is wach am kepe. the:FEM is awake:PREDADJ on:the give:COPINF weak:PREDADJ 'She is waking up.' [31c] Tii is wach kep mit sayn despertador. the:FEM is:COPAUXPRF awake:PREDADJ given:COPPP with his alarm-clock 'She was woken by her alarm clock.' [32a] Tii frau wet wach. the:FEM woman becomes:COP awake:PREDADJ 'The woman wakes up.' [32b] Tii frau is wach woa. the:FEM woman is:COPAUXPRF awake:PREDADJ become:COPPP 'The woman woke up.'

From the figures presented in Table 4, it is apparent that geben represents the prototypical copula predicate variant, and there is minimal variation in Riograndese Hunsrik. In all age groups, the answers with the auxiliary geben clearly dominate. We observe a higher use in the elderly. It is also necessary to mention that evidence for the geben-copula were collected in all 16 exploration sites (see [31a–c]). In contrast, inhabitants of only three places produced the sentences with werden (Igrejinha: six proofs, Ivoti: one proof, and Arroio de Meio: one proof; all of them located in the 'old colonies'; example phrases, see [32a–b])<sup>47</sup> .

<sup>46</sup> The total number of participants in the syntax survey is 192, including 43 (22.4 %) aged 20–35, 78 (40.6 %) aged 36–59, and 71 (37.0 %) aged 60+. The task discussed here was not completed by 29 persons—most definitely because of the lacking or missing writing competence (in German).

<sup>47</sup> The first two places have a high percentage of the population that is Protestant. Maselko (2015) shows, step by step, that the denomination correlates with the choice of the passive auxiliary in Riograndese Hunsrik. Very clearly, the Catholics tend to use geben, and the Protestants werden. As far as the werden-copula construction is concerned, we are conceivably dealing with an exaggeration of properties from another morphosyntactic category, the videlicet passive. Apart from that, it is not out of the question that this is a hypercorrection effect, which leads some individuals to replace the universal variant geben with one that is closer to their (religious) social environment as well as part of the standard language variety.


Table 4: Distribution of geben-copula by age group

Unfortunately, I do not have statistical data on the use of the copula verb with a noun. The distribution scale of the geben-copula with an adjective on both the social-vertical and areal-horizontal levels, however, suggests that we would be dealing with similar values. It can be argued that the construction is fully grammaticalized and commonly used in the Southern Brazilian dialect of German. These circumstances undoubtedly make a potential interlingual transfer possible.

#### 5.3.2 dar-Copula in Brazilian Portuguese

Interestingly, and at the same time somewhat unexpectedly, give has a similar reading in Brazilian Portuguese, as demonstrated in 5.3.1 for the Hunsrik dialect. The examples illustrating its use in Portuguese of South America are given in [33]–[35] below.

[33] João deu em jogador. João gave:3SGPRET in player 'João became a player.' (Martins Salomão 1990: 91) [34] Esse namoro ainda vai dar em casamento. this courtship still goes give:INF in marriage 'This courtship will lead to marriage.' (Martins Salomão 1990: 92) [35a] Ele deu em linguista. he gave:3SGPRET in linguist 'He became a linguist, after all.' [35b] Ele dá prá linguista. he gives for linguist 'He may become a linguist.' [35c] Ele deu prá linguista. he gave:3SGPRET for linguist 'He became a linguist (as I had already anticipated).' (Martins Salomão 1990: 270–271)

Despite the evidence for its use, the construction does not find consideration in the descriptive grammars of modern Brazilian Portuguese. Neither the academic works written in Portuguese, such as Azeredo (2008), Bechara (2009), Cunha / Cintra (2017), Kato / Nascimento / Castilho (2009), Perini (2005)<sup>48</sup>, nor the English or German publications, e. g., Tyson-Ward (2010) and Whitlam (2011; 2017) mention the use of dar in a copula-like function. However, the reference to this particular reading can be observed in a few—but recognized—dictionaries of (Brazilian) Portuguese. The "Dicionário Aurélio da Língua Portuguesa" (2010) 'Aurélio Dictionary of the Portuguese Language'shows two readings in which dar assumes a copula function with the sense 'to become'. Example [36] below illustrates the first of them paraphrased by authors using reflexive 'fazer-se, transformer-se' which we may translate as 'to establish oneself (literally: to make oneself), to transform oneself '. While giving an understanding of the next darreading 'ter determinado resultado' which means 'to have a certain result/result in' (see Examples [37]–[38]), the authors point out its syntactic role to link the subject of a clause to a subject complement being called a predicative. Another reference work, the "Dicionário Priberam da Língua Portuguesa" 'Priberam Dictionary of the Portuguese Language' <sup>49</sup>, proceeds similarly and goes even further by classifying the verb dar precisely as a copula with the sense 'to become'—in Portuguese 'tornar-se' (see Example [39]). These semantic specifications correlate with those from the English–Portuguese dictionary Allen (2014).<sup>50</sup> Her speech example is given below as [40].

[36] Nesse mesmo dia encontrou Abreu que […] dera em jogador […].

on:this:MASC same day met:3SGPRET Abreu who […] give:3SGPLU in gambler […]

'That same day he/she met Abreu who […] became a gambler […].'

[37] Seu palpite deu certo.

his/her suggestion gave:3SGPRET correct

'His/her suggestion worked.'

<sup>48</sup> Likewise, the very extensive grammar of (European) Portuguese by Paiva Raposo / Bacelar do Nascimento / Coelho da Mota / Segura / Mendes (2013) does not make the particular function of dar as 'become' a subject of discussion. However, any mention there would be a surprise given that—according to the existing research literature and surveys—it is Brazil-specific. Castilho (cf. 2014: 411) mentions merely the occurrence of dar as 'helping' verb in the (fixed) phrase dar certo (see more below).

<sup>49</sup> It is an online universal dictionary that contains both Portuguese and Brazilian variants.

<sup>50</sup> A search for the reading 'to become' in the decidedly more extensive dictionary "Michaelis. Moderno Dicionário Inglês–Português, Português–Inglês" (cf. 2000: 1015) 'Michaelis. Modern Dictionary English–Portuguese, Portuguese–English' proves to be futile.

[38] O negócio deu errado. the:MASC deal gave:3SGPRET wrong 'The deal went wrong.' ("Dicionário Aurélio da Língua Portuguesa" 2010) [39] Ela deu uma boa profissional. she gave:3SGPRET a:FEM good professional 'She became a good professional.' ("Dicionário Priberam da Língua Portuguesa" 2008–2013) [40] Ele agora deu em estróina. he now gave:3SGPRET in spendthrift 'He became now a big spender.' (Allen 2014: 97)

Although the research literature treats the topic differently and to varying degrees, it should now be clear that the use of dar as a copula with the meaning 'to become' is quite possible in Brazilian Portuguese. In fact, we should not be surprised by the lack of mentions in reference works considering that we are dealing not only with descriptive but also—or mainly—prescriptive works that generally marginalize the local, non-standard, or less common forms in favor of supra-regional, standard-language, and general common variants.

Finally, we shall turn our attention toasub-aspect, which might have already been noticed by a closer look at the language examples above. Apart from three sentences presented earlier (see Examples [37]–[39]), the copula verb is followed by a preposition, in other words: the predicative expression is a prepositional phrase consisting of a preposition and a (predicate) noun<sup>51</sup> .

<sup>51</sup> The examples of dar-use cited here from the reference works as well as a random pilot study based on the online results would seem to suggest that the occurrence of the copula verb with (a preposition and) an adjective is uncommon and certainly not as common as it is with a noun. I believe that this fact is linked to the grammaticalization paths shown exemplarily for the German language in 6.3.1, according to which the give-copula must first be related to a predicative nominal before doing so to a predicative adjective. Only two examples given, [37] and [38], have a predicative adjective, namely, certo 'correct' and errado 'wrong'. If we take a closer look at the dictionary entries on the verb dar, we find out that in all likelihood, we are dealing with fixed phrases. Several reference works (cf. "Cambridge Dictionary"; "Dicionário Aurélio da Língua Portuguesa" 2010; "Dicionário Priberam da Língua Portuguesa" 2008– 2013; "Michaelis" 2000: 1015; "PONS" 2001–2019 as well as "Collins Portuguese–English Dictionary" 2007–2019 under the lemma "certo") determine dar certo (see Example [37]) as a set phrase/phraseme and do not list any other copula-adjective-examples of this kind and meaning. As a fixed expression (using a grammaticalized 'helping' verb), dar certo is also defined by Castilho (cf. 2014: 411) in his "Nova Gramática do Português Brasileiro" 'New Grammar of Brazilian Portuguese'. Interestingly, dar certo has its own entry in the "Dicionário Português – Renano Hunsrik" aswell; Kuster-Cid/Kuster Cid (cf. 2018: 52) translate the phraseme as ríchtich këpe (literally 'to give right'). Based on this research, it seems to me that we should speak here of a specific lexicalized (idiomatic) phrase rather than a universal

Interestingly, we observe variation in the choice of preposition, which, as Martins Salomão (cf. 1990: 98, 270) and Newman (cf. 1996: 170–171) state, is semantically conditioned. Examples [35a] and [35b] which differ in the choice of preposition shall serve to illustrate. While the second sentence with prá, a Brazilian everyday language variant of para 'for, to, towards', is a kind of judgment and represents a potential and perspective change of state, the first sentence with em 'in, into' expresses an actual (and completed) change of state that has happened without being (necessarily) planned.

This semantic difference can also be read out of Examples [35a] and [35c], which even have the same tense form of dar, the preterite. Both relate to a completed process and an already changed state but have a slightly differentiated meaning. In [35c], the preposition prá causes the emphasis of prediction, whereas in [35a], em indicates a change of state which is unexpected/fortuitous or less perspective. The first sentence may be used, for instance, in relation to a purposeful and goal-driven person whose motivation, belief, and competencies to become a linguist are beyond doubt from the beginning. By contrast, the second sentence can refer to an undecided person who has never actually shown an interest in linguistics, did not aspire to this profession, or has not been particularly successful in his education.

In the final stage of this section, the results of a corpus linguistic pilot study are presented. Firstly, the research shall provide 'real' examples demonstrating the verb dar in a copula-like function. Secondly, it aims to show some (semantic and structural) (frequency) tendencies regarding the construction's use and the (possible) choice of preposition. Thirdly, the study is intended to confirm or refute the thesis that the dar-copula is a phenomenon specific to the Brazilian variety of Portuguese.

To answer these questions, I firstly created a catalog of twenty popular/dream professions<sup>52</sup> since occupations represent a (proto)typical usage context for

syntactic pattern, as is the case of copula and noun. The renunciation of the preposition may also indicate this fact. The expression dar errado ([39])—mentioned only in a few dictionaries (cf. under the lemma "dar" in "Dicionário Aurélio da Língua Portuguesa"; "Dicionário Priberam da Língua Portuguesa" 2008–2013, as well as under the lemma "errado" in "Collins Portuguese–English Dictionary" 2007–2019; "PONS" 2001–2019)—cannot contradict this thesis. The use of certo's antonym errado, as well as the very closely related semantics of both phrases are indications of an analogy formation. The "Dicionário Priberam da Língua Portuguesa" (cf. 2008–2013) justifies the explicit mention of dar errado in its entry with the same reasoning.

<sup>52</sup> The selection is based on various polls conducted in Brazil and consists of the following professions (in the Portuguese language): advogado/-a 'lawyer', arquiteto/-a 'architect', ator/ atriz 'actor', cantor(a) 'singer', dentista 'dentist', deputado/-a 'deputy', diretor(a) 'director', jogador(a) '(soccer) player', jornalista 'journalist', médico/-a 'doctor', militar 'soldier', motorista 'driver', polícia 'policial', político/-a 'politician', presidente 'president', professor(a)

copula constructions with the meaning 'to become'. Afterwards, I used the "Google" search engine<sup>53</sup> to determine the occurrence of collocations and their frequency.<sup>54</sup> As previously shown, using examples from the research literature, the possible collocation in a give-copula construction looks like this: COP € PREP + N<sup>55</sup>. In order not to restrict the results to a single verb form, in my investigation I took into account four different finite forms of dar distinguished by two grammatical persons (1SG and 3SG) and two verb tenses (present tense and preterite). The (re)search has been narrowed to the websites with the country code top-level domains \*.br (Brazil) and \*.pt (Portugal) to enable areal-specific statements. This results in the following search formula I used to determine the collocations and to consider the mentioned factors: ′′dou/dá/dei/deu em/prá/ para/ø PROFESSION′′ site:br/pt.<sup>56</sup>

The method accompanies some problems that we should be aware of. First of all, the significant difference in the total number of Brazilian and Portuguese websites<sup>57</sup> does not allow undisputed contrastive analyses. The next point that should be considered is the fact that not every hit shown is a copula construction.<sup>58</sup> When looking at the numbers shown in Table 5 below, these facts must be taken into account in any event. Nevertheless, the results allow us to recognize individual relationships and tendencies.

<sup>&#</sup>x27;teacher', secretário/-a 'secretary', técnico/-a 'technician', vendedor(a) 'salesperson', vereador(a) 'councilor'.

<sup>53 &</sup>lt;https://www.google.com>.

<sup>54</sup> The reference day is October 19, 2019.

<sup>55</sup> In detail it has the following structure: COP (= finite dar) € PREP (= em, prá/para or ø)+N (= predicative nominal). The theoretical possibility of connecting the copula with a predicative adjective is omitted here for the reasons mentioned above.

<sup>56</sup> I.e. ′′dá prá jogador site:br′′. If the feminine form differs from the masculine, a search with a female profession designation was carried out additionally (i. e. "dá prá jogadora site:br").

<sup>57</sup> As of October 19, 2019, "Google" showed 3,510,000,000 pages for site:br and 371,000,000 for site: site:pt. That corresponds to the ratio 1:9.46. For the year 2015, the portal "domains.info" registered approx. 3.7 M domains with the country code \*.br and 0.7 M with \*.pt. At this time, there were at least five times as many websites registered in Brazil as in Portugal.

<sup>58</sup> For dá + N (expression without preposition), I do not specify the values in the table because —despite the correct entry into the search engine (viz. with acute accent diacritic and in quotation marks)—there were many results that did not match (i. e., with da which is a synaeresis form of the preposition de and the feminine definite article a). The figures on nonprepositional constructions are generally to be treated with caution due to the presumably large number of hits with dar as the main verb.


(Continued)


Table 5:"Google"results for collocationdar€PREP+ NPROFESSIONon the Brazilian and Portuguesewebsites

From the figures, it is apparent that there are more results with dar in the preterite (see Examples [41]–[43]) than in the present tense (see Examples [44]– [46]). However, when we try to match proportionally the number of the results which occur in the 3SG preterite and do not have a preposition with them realized in the 3SG present tense, then the difference would not be noticeable, and there would be even more evidence of the present tense.

The figures and the established linguistic evidence reveal that dar in its copula-like function can also occur without a preposition. Sentences [42]–[43] below follow the pattern known already from example [39]. Statistically speaking, we can even surmise that this is the most commonly deployed variant of the darcopula construction in Brazilian Portuguese.

If we exclusively take a look at the results containing a preposition, it is evident that "Google" delivers almost three times more hits with the preposition prá/para than with em (484 vs. 174). However, the discrepancy in the number of results does not prevent us from finding examples illustrating exactly the semantic conditions of both prepositions presented above. Examples [44]–[46] are—in my opinion—well suited to differentiate the meaning or—put another way—the usage motivation of prá/para and em in combination with dar-copula. In [44] and [46], which both show prá we deal with clear statements concerning simultaneously the reference person's predisposition to a specific occupation as well as (the realization potential of) their intentions/plans. While these are foreseeable changes of state, Example [45] using em explicitly points out the unpredictability of the change and its 'surprising' character.

[41] Ele estava junto com o amigo Ed Valentin, que também deu pra cantor. he was together with the:MASC friend Ed Valentin, who also gave:3SGPRET for singer

'He was with his friend Ed Valentin who became a singer too.' <http://magnopapagaio.blogspot.com.br/2011\_02\_01\_archive.html> [42] Não deu jogador de futebol e nem quer estudar. not gave:3SGPRET player of soccer and nor wants study 'He did not become a soccer player, nor did he want to study.' <(https://alfenashoje.com.br/noticia.asp?id\_noticia=3902> [43] Valente Lageano é grande e também deu jornalista. Valente Lageano is grand and also gave:3SGPRET journalist 'Valente Lageano is grand and became a journalist too.' <http://www.deolhonacapital.com.br/2009/02/01/intervalo-para-o-"meme"/> [44] La minha linda, sabe que você dá pra arquiteta. her my:FEM beautiful, know that you gives for architect 'My dear, you know that you may become architect.' <https://www.larissalieders.com.br/2010/05/28/tchara/> [45] Ou um indivíduo sem competência de qualquer espécie dá em político. or an:MASC individual without competence of any type gives in politician 'Or an individual without any competence becomes a politician.' <http://dragoscopio.blogspot.com.br/2007\_11\_01\_archive.html> [46] Eu não, não dou pra político. I not, not give:1SGPRES for politician 'I am not, not (!) going to become a politician.' <https://farofafa.com.br/2016/06/08/mateus-aleluia-voz-dos-recuados/>

Professions that usually attract more public attention or arouse the widespread interest of several social groups, such as jogador(a) '(soccer) player', político/-a 'politician', and professor(a)'teacher', show the most numerous "Google" results. However, on account of their 'polarizing' power in society and (online) media, I consider this finding to be not unexpected. This is supported by the fact that the isolated evidence for the use of copula dar collected from Portuguese websites is primarily associated with similar occupational groups, which can be seen from Examples [47]–[49].

[47] […] Mourinho comeu futebol desde que nasceu. Não deu para jogador, felizmente.

[…] Mourinho ate soccer since that born. Not gave:3SGPRET for player, fortunately

'[…] Mourinho has eaten soccer since he was born. He did not become a soccer player, fortunately.'

<https://bomsensoamiguinhos.blogs.sapo.pt/233911.html>

[48] Este homem dá para professor.

this:MASC man gives for teacher

'This man is going to become a teacher.'

<https://expresso.pt/politica/2017-09-18-Marcelo-voou-sobre-mar-picado-paraacompanhar-militares-portugueses-e-voltou-de-coracao-cheio>

[49] Joe Hart não dá para ator.

Joe Hart not gives for actor

'Joe Hart is not going to become an actor.'

<https://www.record.pt/jogo-da-vida/detalhe/joe-hart-nao-da-para-ator-7828 88>

The difference in the count of "Google" hits ending with \*.br and \*.pt is significant. Even considering the fact that there are more websites registered in Brazil than in Portugal, we can assume that the dar-copula is a construction typical of Brazilian Portuguese. "Google" delivers nearly fifteen times more hits with a Brazilian country code than with a Portuguese one (10,970 vs. 733), which does not quantitatively correlate with the number of existing websites (approx. five times more Brazilian web pages as shown in Footnote 57).

On the other hand, the isolated findings from European Portuguese do not reinforce my assertion that the dar-copula is a language contact phenomenon whose occurrence is supported by the significance of geben in the German dialect, Riograndese Hunsrik. However, it cannot be ruled out since among these few hundred hits for dar € PREP + N on \*.pt-websites, only in a minority of cases do we come across the grammaticalized copula construction. Furthermore, there can be no guarantee that texts providing evidence for this collocation were written solely by Portuguese people. In the era of globalization as well as the intense migration from South America to Europe, it is conceivable that some findings are attributable to speakers of Brazilian Portuguese.

#### 6 Conclusion and Research Desiderata

In Riograndese Hunsrik, the give-verb has undeniably reached one of the highest degrees of grammaticalization among varieties of the German language, and perhaps worldwide too. The German dialect of Southern Brazil uses geben in at least five of the seven cases mentioned in the paper: as light verb, perfective prefixed verb, existential verb, copula verb, and passive auxiliary verb; in the latter two, however, it encounters the competing werden-variants. At first glance, only Luxembourgish seems to have a higher grammaticalization 'level'—at any rate among the Germanic languages—since geben takes up the position of subjunctive auxiliary and has become invariable as a result of the elimination of werden-variants. Thus, even if we determined a more extensive use of geben as a future tense of Hunsrik, the national language of Luxembourg would still remain the most highly grammaticalized.

Although individual grammars draw attention to the semantic polyvalence of dar in Brazilian Portuguese, the current state of research on the equivalent phenomena in this Latin American variety of Portuguese seems even more limited than that for German. Nonetheless, the Brazilian research landscape provides stimulating and exciting clues to the use of darin an existential verb and copula-like function, which in any case should be pursued further. The extent of the empirical tests conducted in this article, and thus the conclusions drawn from them, may be—of course—unsatisfactory in a wider context. However, this article seeks to uncover the possibilities of variationist and contact linguistic research in this field and to encourage further empirical research—whether it be with the help of sociolinguistic methods such as interviewing, speech recording, language production tests or corpus linguistic ones, for which the comprehensive online "Corpus do Português" 'Corpus of Portuguese' <sup>59</sup> would be the logical and recommended access point for this. Needless to say, the demand for subsequent research in the verbal-syntactic field of grammaticalized variants of give applies to the German language too, where there is a lack both of basic studies and of reliable data in terms of the formal-semantic framework and variability, especially for individual (micro-areal) varieties. It would certainly have a positive effect on large-scale and more theoretically oriented research, which has so far convincingly addressed the relevant language-historical, linguistic-geographical, system-linguistic and semantic questions.

In the final paragraphs we should again focus on the results presented in this article and summarize them as briefly as possible: The verb give shows in several places its pronounced (probably inherent) ingressive character both in German and Portuguese. Especially in Brazilian Portuguese, the grammatical dar-variants are firmly tied to ingressive or inchoative semantics. In contrast, in German, the verb forming existential clauses may also have a generic or context-related reading. Interestingly, in both languages, weather descriptions and forecasts prove to be typical contexts for using existential constructions with give.

It is astonishing, given its agentive background, that give can play the role of a copula verb. The verb once again reveals the high degree of grammaticalization it has achieved in Riograndese Hunsrik. Firstly, unlike in other varieties of German, it can not only connect with a predicative nominal, but also with a predicative adjective. Secondly, it is invariable almost in the entire area in which it occurs. Thirdly, it does not adhere to the meaning of a 'pure prognosis' or a 'badly realizable speculation' and thus covers several semantic domains.

This goes hand in hand with only a small part of the Moselle Franconian dialect area from the European original 'homeland'. The copula-like dar-construction attracts little attention in Portuguese linguistics. As shown in the empirical study, almost exclusively in the Brazilian variety of Portuguese grammatical constructions consisting of dar and a predicative noun are possible. It is not unusual for a semantically determined preposition to appear between the two components. The preposition prá/para, which stands for predictability, is statistically more common than em used in connection with unpredictability and randomness. In only a few cases collocations with an adjective can be found; the rarity and extremely limited range of adjective candidates cause us to surmise that this solely happens in set phrases.

<sup>59 &</sup>lt;https://www.corpusdoportugues.org>. The corpus contains about one billionwords of data.

Unfortunately, it remains unclear why we observe a formal-functional and cognitive-semantic extension of the give-verb's features in South Brazilian German and Brazilian Portuguese (and otherwise only a very marginal number of languages in the world). This relationship must not be ignored in subsequent studies. At the present time, however, we can identify good prerequisites for the fact that German spoken by a large community in (Southern) Brazil supports—if not causes—the use of dar for existential and copula constructions in Brazilian Portuguese. Speaking of supportive contact phenomena in these cases may not be unfounded.

#### Bibliography


California at Berkeley. <https://escholarship.org/uc/item/0xs293hs> (September 26, 2019).


Michaelis. Moderno Dicionário Inglês–Português, Português–Inglês. 2000. São Paulo: Melhoramentos.


Tyson-Ward, Sue. 2010. Brazilian Portuguese. London: Hodder Education.

Weinreich, Uriel. 1953. Languages in contact. The Hague: Monton.


#### Index of Glossing Abbreviations


Mateusz Maselko, BA MA MA Université de Genève Faculté des Lettres Département de langue et de littérature allemandes Uni Bastions Rue De-Candolle 5 1211 Genève 4 Switzerland Mateusz.Maselko@unige.ch http://www.unige.ch/lettres/alman/de/enseignants/linguistique/maselko/

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Janne Bondi Johannessen (†)

# Dialect Syntax in Norway and America: Research Infrastructure and Research Results

Abstract: The paper shows how research infrastructure—two corpora and a set of recordings with transcriptions—can enhance research and be used to answer research questions. Taking Norwegian heritage language in America asastarting point, the questions asked are: Is the word order in Norwegian heritage language the same as in European Norwegian? If not, how can this be explained? The specific word order in subordinate clauses will be used as a case in point. Since a heritage language is typically a spoken language variety, any comparison would need to be undertaken with an equivalent language variety in the homeland. In order for this kind of research to be feasible, a research infrastructure containing relevant language data is helpful. I will show how the research questions can be answered using two corpora in addition to old recordings.

Keywords: research infrastructure, speech corpora, heritage Norwegian, European Norwegian, target and non-target word order

Abstract: Denne artikkelen viser hvordan forskningsinfrastruktur kan brukes til å besvare forskningsspørsmål. Med norsk nedarvingsspråk i Amerika som utgangspunkt stilles disse spørsmålene: Er ordstillingen i norsk nedarvingsspråk den samme som i europanorsk? Hvis ikke, hvordan kan det forklares? For å belyse spørsmålene vil ordstillingen i underordnede setninger bli undersøkt. Siden nedarvingsspråk typisk er talespråk, er det viktig for sammenligningens skyld at en tilsvarende språkvariant er undersøkt i språkets opprinnelsesland. I vårt tilfelle kommer forfedrene til nedarvingstalerne fra dalstrøkene i Øst-Norge, og dermed følger det at nettopp disse områdene må utgjøre standarden for sammenligning. For at en slik sammenligning skal være mulig, er det en stor fordel at det finnes relevant forskningsinfrastruktur. Ved å bruke to korpus i tillegg til gamle opptak vil forskningsspørsmålene bli besvart.

Keywords: forskningsinfrastruktur, talespråkskorpus, nedarvingsnorsk, europanorsk, standard- og ikke-standard-ordstilling

#### 1 Introduction

This paper<sup>1</sup> attempts to demonstrate how effective language sources, especially electronic, searchable corpora linked to audio, see Sections 5.2 and 5.3, can be used to find out what constructions are used and how common they are in a certain language variety. For example, if there are two corpora containing similar metadata, and there is sufficient detail of the metadata, researchers can narrow down their searches to make valid comparisons between two—or more—language varieties. Additional data, such as recordings not linked to transcriptions in a searchable corpus system, can be helpful to document a particular use.

In the paper, I demonstrate how non-target word order in subordinate clauses in heritage Norwegian—i. e., verb–adverb rather than adverb–verb, see Section 4—are not just arbitrary mistakes by certain individuals, but is systematic.<sup>2</sup> The question of why this order has come about to be part of the heritage Norwegian variety is answered by exploring four hypotheses. The corpora also play a role in testing these hypotheses. The remainder of this paper is structured as follows: Section 2 presents two language varieties; European Norwegian and heritage Norwegian. Section 3 provides a brief overview of previous research on the word order of subordinate clauses and word order generally in heritage Norwegian. In Section 4 some central ideas as well as specific hypotheses are presented. Section 5 explains why corpora are key in the current research and presents the old recordings by Einar Haugen, the Corpus of American Nordic Speech (CANS) and the Nordic Dialect Corpus (NDC). In Section 6, two types of results are provided. First, the results from the data sources—recordings and, especially, searchable corpora—showing that the non-targetword order has been part of the heritage language over 80 years, and that it is indeed part of the present-day heritage language as well, in contrast to the homeland variety in Norway. Second, the hypotheses are tested one by one, indicating that what is the cause of this word order is incomplete acquisition. The paper is concluded in Section 7. The research of Sections 4–6 is based on Larsson / Johannessen (2015a; 2015b).

<sup>1</sup> I would like to thank two very thorough and constructive anonymous reviewers for their very helpful comments, which have clearly improved the paper. I would also like to thank Mateusz Maselko and Alexandra N. Lenz for organizing a very interesting conference.

<sup>2</sup> Since Norwegians immigrated from many different parts of Norway, and settled in different parts of North America and Canada, it may seem wrong to regard them as one, single variety. However, I choose to do that in this paper, because the phenomenon I look at does occur in many different speakers with a variety of backgrounds. Also, there are reasons to regard heritage Norwegian as a common variety, as argued, on the basis of the lexicon, in Johannessen / Laake (2012; 2017) and Hjelde / Johannessen (2017).

#### 2 Norwegian: A Brief Overview

Norwegian (ISO 639–3 nor) is a Scandinavian language, in the North Germanic branch of the Germanic languages. It is spoken in Norway, a geographically vast country with a total area of 385,252 square kilometres, and with latitudes from 57° to 81° N. The country has a rugged coastline of long fiords, thousands of islands, and large mountainous areas throughout most of the country. It is perhaps for this reason that, its five million people speak a variety of dialects. Variation has also been caused by the fact that Denmark ruled the country for four hundred years, leading to Danish becoming the written language and also to aspects of Danish influencing the Norwegian language in the capital and later the rest of the country—for more, see Nordbø (2016) and Røyneland (2009).

Norwegian is also spoken in North America. 850,000 people emigrated to the USA and Canada between 1825 and 1930, and Norwegian was a thriving language for much of this time. Most of the immigrants came from rural areas in Norway and wanted to farm land, so they settled together in areas where they could build farms and cultivate the land. Here they founded Norwegian-language schools and churches and built what became tight-knit communities where their language would thrive (cf. Haugen 1953; Johannessen / Salmons 2015b: 10). This continued for a long time, until around the second world war, when English took over as the language of the institutions of education and worship, and Norwegian started losing its role in the communities. Today, this heritage variety is moribund, as the majority of the speakers are very old. However, the remaining speakers speak a language that is well worth studying<sup>3</sup> . They learnt it from their family, often as theirfirst language, and many did not learn English until they went to school. For all of them, English took over as their main language, and they are generally illiterate in Norwegian. The Norwegian heritage language in America has been spoken for several generations as a minority language in small communities next to a majority language. It is to be expected that it differs from European Norwegian along many dimensions, from the lexicon via phonology and morphology to syntax, see for example the overview article in Johannessen (2018) and individual papers in anthologies such as Bousquette / Brown (2018a; 2018b), Heegård Petersen / Kühl (2018), Johannessen / Salmons (2015a) and Page / Putnam (2015).

<sup>3</sup> The present author has met around two hundred speakers spread across small communities in the American Midwest and Canada. Altogether, a rough estimate may be that there are a couple of thousand speakers left.

#### 3 State of Research

This paper is based on research in Larsson / Johannessen (2015a; 2015b), which deals with word order in subordinate clauses in both Swedish and Norwegian, but the present paper concentrates only on Norwegian.

Word order in heritage Norwegian has also been studied previously: for subordinate clauses, see also Taranrød (2011) and Section 5, and for word order in main clauses, see Eide / Hjelde (2015) and Johannessen (2015b), who both observe that the general placement of the finite verb in the second position, found in European Norwegian is not as regular in heritage Norwegian. Westergaard / Anderssen (2015) found that the two available orders of possessive pronouns, with respect to the noun, are not distributed in the same way in heritage and European Norwegian. Other Scandinavian heritage languages have also been studied with a focus on word order: Arnbjörnsdóttir et al. (to appear) study main clause word order in North American Icelandic, as do Hedblom (1978) for the Hanebo dialect of Swedish, and Kühl / Heegård Petersen (2016, to appear) for main clause word order in North American Danish. Though many researchers examined the word order of various constructions in heritage Norwegian, variation in subordinate clauses has only been studied by Larsson / Johannessen (2015a; 2015b) and by Taranrød (2011) in her MA thesis.

# 4 Subordinate Clauses in American Heritage and European Norwegian

Subordinate clause word order in Heritage Norwegian differs from the word order in European Norwegian, in that there seems to be free variation between the order in [1a] and [1b] in the former, whereas the latter, in the relevant area of South Norway, only accepts adverb-verb order, i. e. [1a]. The two orders are illustrated in [2a, b], with the categories complementiser, subject, adverb and verb. (See Section 5 for information on how this data have been found.)

```
[1a] CMPL–SUBJECT–ADVERB–VERB (Heritage Norwegian, European Norwegian)
```
[1b] CMPL–SUBJECT–VERB–ADVERB (Heritage Norwegian)

[2a] det er mange som ikke har slutta og

it is many who not have stopped too 'There are many that have not stopped, too.' (Heritage Norwegian, coon\_valley\_WI\_02gm)

[2b] om du finner ikke ut så if you find not out then 'If you don't find out, then…' (Heritage Norwegian, sunburg\_MN\_04gk) It should be mentioned that there is one complementizer that allows for either word order even in European Norwegian, viz. at 'that'. If the embedded clause is asserted, it can have the word order in [1b], see Julien (2008; 2015). This complementizer is one of three that head the clauses studied in Section 6. If anything, its presence would make the similarity between heritage and European Norwegian stronger, but we shall see that this is not the case.

The research on heritage languages reveals that explanations for change can be found across many dimensions: language-external factors such as social structure, language attitudes, cognitive factors—including incomplete acquisition and attrition, transfer and convergence, processing, memory, complexity and overgeneralisation (related to frequency of occurrence of particular forms or constructions), language-internal, structural factors, typically related to syntax, phonology and morphology, and factors directly related to language contact, such as transfer—for overviews, see Benmamoun et al. (2013a; 2013b), Johannessen (2018), and Montrul (2015).

The situation of heritage Norwegian, in the communities where it is still spoken, is one of language contact where the speakers are bilingual, and their heritage language, which is typically their L1, has developed and is no longer their dominant language. The communities are undergoing language shift, as the heritage language has not been passed on to the next generation.

Larsson / Johannessen (2015a; 2015b) have several hypotheses for the linguistic changes (on how they found the changes, please see Section 5). Could the changes be due to the contact situation with English, i. e., influence of English, or even more specifically, transfer from an English construction to Norwegian? Could there be a verb second (V2) pattern generalised from main clause word order? Could there be attrition of heritage language? Could there be incomplete acquisition of the heritage language? The results will be presented in Section 6.

#### 5 Methodology: Corpora and Recordings

The study of heritage languages gives us, as linguists, a golden opportunity to study variation, stability and change, since we have a language variety that has been developing in a different geographical place (America) from the ancestral language from which it originated (Norway). We thus have two varieties, and can use the ancestral language (the European variety in our case) as a baseline for comparison. One might argue that it is wrong to compare a language spoken by a (for instance) fifth generation heritage language speaker in America against a baseline of a present-day European language. This of course depends on the goal of the study and the assumptions that are made. If it is important to study the heritage language of a living person as a direct consequence of input from the language of origin (in our case Norway), it is a study that cannot be done if the speaker is a fifth generation immigrant in America. Therefore, the displacement of generations must be taken into account. Regarding Norwegian, there is actually data on earlier generations of heritage speakers, and even of emigrant speakers, from recordings done by Professor Einar Haugen in the 1940s (see also Haugen 1953). For more information on the research material, see Section 5.1.

Taranrød (2011) investigated the word order in subordinate clauses in heritage Norwegian and European Norwegian. The material at the time was rather small, but she still found striking differences between the two. Out of twelve relative clauses in heritage Norwegian containing an adverb, six. i. e. half, had verb– adverb order. She also examined her findings in relation to the overall size of the text material. The relative clauses were found in a material of 60 000 words. In an early version of the Nordic Dialect Corpus she found only three subordinate clauses with verb–adverb order in a material of a total of 1.5 million words (cf. Taranrød 2011: 64).

Larsson / Johannessen (2015a; 2015b) considered it necessary to first determine whether there have actually been changes in the modern heritage language compared with the baseline language—the modern European Norwegian language—and then also study the language of past heritage speakers and immigrants from several generations ago. Only if a linguistic change can be established is there a point in looking for reasons for a particular change. For this they needed data from early immigrants and heritage speakers. Such data could be found in Einar Haugen's recordings and work. They also needed updated data from today's heritage speakers, which they found in the CANS corpus. Finally, they needed data from the particular geographical areas of Norway from which the ancestors of the present-day heritage speakers emigrated. They found this in the NDC corpus.

The existing infrastructure they could use thus consisted of two speech corpora—the CANS corpus and the NDC corpus—and one set of recordings– Haugen's–some of which were transcribed. They are available at the same website —at the Text Laboratory at the University of Oslo. In addition to simple recordings and word lists, this website also offers a variety of searchable corpora for written and spoken language, where the greater part are accessible by the same user interface. Both speech corpora have some important features in common: they contain conversations about everyday topics rather than formal interviews or metalinguistic discussions, and they in most cases contain two speakers from the same place, avoiding influence from other speech varieties. The recordings are transcribed and searchable, see Sections 5.2 and 5.3. Many of the recordings in the two corpora contain more than one type of transcription, i. e. both an orthographic and a phonetic one, making more specified search possible. The two corpora are grammatically tagged with part of speech tags and morhosyntactic features, such as plural and tense and are therefore searchable in many ways. Searches can be filtered by a long array of metadata, such as gender, age, place and year of recording. The corpora are built on the same corpus search system platform, Glossa (cf. Kosek et al. 2015; Nøklestad et al. 2017). The searches can be done in the same way, see Sections 5.2 and 5.3, and the results are also presented in the same way, where the user can choose between a concordance view, a frequency list view or a map view. However, the two corpora have rather different thematic contents. The CANS corpus contains only American Norwegian speakers, which means that many of the conversations touch upon topics related to the national identity of the speakers, their knowledge and use of the Norwegian language and their heritage more generally. The Norwegian part of the NDC contains only European Norwegians, and they talk about a much wider variety of topics, given that they, of course, can talk about anything in their own language. Though the speakers may sometimes reminisce about the past, the topics also concern holiday habits, the political or financial status of their local village, or cultural events. Given the different settings of the speakers and their history, the metadata are also very different for the two corpora. The CANS corpus gives the additional opportunity for the users to filter their searches through categories such as home place of the ancestors in Norway and number of generations in America, whether the speakers have been to Norway, what language of instruction was used at school and confirmation, and whether they can read and write in Norwegian. Haugen's recordings and the two corpora will be described below in Section 5.1–5.3. They are all available at theweb site of the Text Laboratory, University of Oslo, and their URLs are provided on the final page of this paper.

#### 5.1 Haugen's Recordings

Einar Haugen (1906–1994), Professor at University of Wisconsin–Madison and Harvard University was a pioneering sociolinguist whose work on the Norwegian language in the American Midwest in the 1930s–1940s has had lasting value. During his fieldwork he made numerous recordings with immigrant speakers and heritage speakers of Norwegian, some of which were also transcribed.

There are recordings of more than 220 speakers, and more than 90 transcriptions, in which a phonetic-like standard has been used. The recordings consist of conversations between Haugen or his assistant and one or more speakers. They deal, in part, with the sociolinguistic background of the speakers, such as questions on family language, and partly with descriptions of how they celebrate weddings or of dramatic episodes in their lives.

Although the material is not directly searchable, transcriptions and recordings are available on the web, which is helpful when the main goal is to find examples of particular constructions. For the present research it was easy to read through the transcriptions looking for relevant subordinate clauses. Listening through the recordings was obviously more time consuming, but still possible. [3] is an example of several Haugen recordings.

[3] Då me ha no blitt jipte så kjem hornaran when we have now become married then come horners. def 'When we now had become married, the horn players came.' (Winfield Krostu, Waupacs Co., born 1884 in Wisconsin, recorded in 1942.)

Such examples from the old recordings from 1930s–40s show that the non-European word order existed in heritage Norwegian 80 years ago and is not a new phenomenon.

5.2 The Corpus of American Nordic Speech (CANS)

The CANS corpus (Johannessen 2015) in 2018 consists of 250 000 words uttered in conversations by 69 speakers from two countries (Canada and the USA), covering 28 different locations, and two languages: heritage Norwegian and heritage Swedish. The recordings are from 2010 till 2018, but the corpus will continue to grow as more recordings and transcriptions are being made. The recordings partly consist of interviews between a Norwegian university professor —or other staff—and a heritage speaker, and partly between two heritage speakers. The interview is designed to ensure that all the speakers answer questions about their Norwegian language history and language use. The main goal of the conversation between speakers is to record the heritage Norwegian of the speakers, with as little influence as possible from the visiting Norwegians. They can speak about whatever they want, but since their vocabulary is generally limited, they often find it difficult to talk about modern topics, and are happy to talk about the olden days, which often means life as a child or agricultural methods.

The corpus system is very well suited to do searches for linguistic phenomena in a linguistic aswell as an extra-linguistic context. On the left side of the interface are metadata categories through which the search can be filtered. This includes general categories such as gender, age, place and area, but also special heritage categories such as number of visits to Scandinavia, school language, literacy in the Scandinavian language, place of origin in Scandinavia and number of generations in America. In the middle of the page is a window for simple searches —basically single words—in the transcribed text, but it is possible to switch to an


#### Figure 1: The user interface of the CANS corpus

extended interface where search fields can be added for more words, where features can be selected, such as grammatical categories and features, and where it is possible to switch from orthographic to phonetic search, and where it can be indicated whether the search is for parts of words (first or last), or whether the search string should be sentence-initial or sentence-final. The searches are conducted using buttons or menus, in addition to the fields where the writing of words or parts of words are written; the interface has been developed to maximally user-friendly.

A suitable search (using the extended version) for the present problem would be to find a complementiser (subjunction) followed by up to three words and then a suitable adverb such as ikke 'not', see Figure 2.

This search would give a number of relevant and irrelevant hits, from which only relevant can be picked out and studied. The results are presented in Section 6. The corpus also gives opportunities for visualizing the results in a map, which for some purposes is very useful, see Figure 3. The results can also be shown as a


Figure 2: A broad search for a complementiser followed by the adverb ikke at most three words to its right

frequency list, and the researcher can furthermore, for each occurrence, test the validity of the data by clicking on a button and get a video clip from that particular part of the recording.

Figure 3: A map of the hits from the search in Figure 2

The map shows that the results from our search for a complementizer followed by the adverb ikke 'not' within three words to the right of the complementizer is spread across three states: North Dakota, Minnesota and Wisconsin.

#### 5.3 The Nordic Dialect Corpus (NDC)

The 2015 version of the NDC corpus (Johannessen et al. 2014) contains 3.1 million words uttered by 873 speakers from five Nordic countries covering 238 locations and five languages—Danish, Faroese, Icelandic, Norwegian and Swedish.

The majority of the Norwegian recordings in the corpus were completed between 2007 and 2013, during the "Norwegian Dialect Syntax" project, which was fully funded by the Research Council of Norway. There is a vast dialectal variation in Norway, which is why we chose to do recordings in as many as 165 locations and with 575 speakers. For most locations there are two speakers of each gender and two age groups—young, under the age of 30, and old, over 50. Each speaker participated in an interview, as with the recordings in the CANS corpus, and in a conversation with another dialect speaker. This way we obtained different speech situations, where the interviewwas semi-formal, while the conversation was free.In order to ensure that the corpus could be used freely forresearchwhen finished, the participants were asked to not talk about things that would breach person protection rights. Instead, they were given a list of neutral topics to talk about, such as film, literature and theatre, holidays, sports and travels, in addition to more local topics, such as agriculture and traditional baking.

The search options in the user interface are very similar to those described for CANS, but there are fewer metadata categories and they contain other subcategories. For example, there is no category here for school language, generation or visits to other places, but there are five countries to choose between, not just two as in CANS, and there are 237 places, not just 28 as in CANS. Figure 4 shows the user interface.


# 6 Results

#### 6.1 Corpus Results

The CANS corpus only contained 50,000 words at the time of the research for Larsson / Johannessen (2015a; 2015b). In order to make the two speaker groups from the corpora comparable, a subselection of the Nordic Dialect Corpus was made, in which only old speakers were chosen, and only from some places in the Oppland county: Brandbu, Brekkom, Gausdal, Jevnaker, Nordre Land and Skreia —areas where most of the ancestors of the CANS speakers were from amounting to 49,000 words. This had the additional advantage that the two sources of data were directly comparable both with respect to size and linguistic background. The search results from the CANS corpus and the Nordic Dialect Corpus are given in Figure 5 and Figure 6, respectively.


Figure 5: Results from a search in the Corpus of American Norwegian Speech (from Larsson / Johannessen 2015a: 249)


Figure 6: Results fromasearch in the Nordic Dialect Corpus (from Larsson / Johannessen 2015a: 250)

The results show a clear difference between heritage Norwegian (the CANS corpus, Figure 5) and European Norwegian (The NDC, Figure 6). The former has 22 occurrences of verb–adverb order as against 10 with adverb–verb, i. e. there are twice as many of the unexpected word order as of the expected one. The latter contains 14 with verb–adverb order and 40 with adverb–verb. This number is striking since the data includes sentences with the complementizer that 'at', which allows verb–adverb order in European Norwegian, see Section 4. In other words, in European Norwegian, around a quarter have verb–adverb order. The difference between the two varieties is thus colossal. There is nothing in the European Norwegian data that suggests that the verb–adverb order is acceptable in subordinate clauses following the complementizers om 'whether' and som 'which/who/that', which have 0 hits out of 5 occurrences and 1 out of 17, respectively. This is in sharp contrast to the heritage variety, which has 1 of 3, and 6 of 13, respectively. Although the results are striking, the overall numbers are small, and the investigation could be repeated with more data, but the results are striking. A qualitative impression based on the old data in Haugen's material supports the finding: Given that this material is not easily searchable, examining the recordings manually would be a too arduous task, but sentences such as [3]), from Haugen, show that the change towards what we can now see as a facultative word order in subordinate clauses started at least 80 years ago.

Some information on the heritage speakers in the corpus with respect to the results is in order. The corpus at the time of the investigation continued ca. 20 speakers, with a total of 50,000 words, as mentioned at the beginning of this section. They did not produce the same number of words, and it was only the most talkative amongst them that also produced subordinate clauses containing an adverbial. This could indicate that the unexpected word order was something idiosyncratic for a few of the speakers, or that there was a systematic variation pattern based on sociolinguistic background. From the data it is clear that there is no systematic inter-speaker variation. One of the speakers, westby\_WI\_03gk, a woman from Westby, Wisconsin, uses both patterns.<sup>4</sup> Most of the speakers in the corpus, as mentioned above, had an ancestral background from the same area of Norway, and also lived in the core areas of the Midwest. The speaker westby\_WI\_03gk is a typical speaker. She was born in 1922 and was 88 years old at the time of the recording. She was third generation Norwegian—both her parents had been born in the USA, she had never been to Norway and had no relatives in Norway that she knew of. She was confirmed in English and had to speak English at school. This is also when she learnt English, at the age of six.

#### 6.2 Analysis Based on Research Questions

How can we explain that such a change has taken place in the heritage variety? I will go through the hypotheses posed in Section 4 one by one, in accordance with the arguments in Larsson / Johannessen (2015a; 2015b), though with examples found specifically for the present paper.

<sup>4</sup> It could be mentioned that a brand new version of the CANS corpus (too new to be included in the present paper) contains 185 speakers and 700,000 words, and that a search similar to the one in Figure 2 gives many hits. A quick glance at the results shows that intra-speaker variation can be found with others, too, e. g. a man from Decorah, Iowa, decorah\_IA\_01gm, has three relevant subordinate clauses (excluding at 'that'), of which two are with the unexpected word order.

Transfer. To test this hypothesis, we are looking for some English constructions that would be a pattern for the new word order in subordinate clauses, i. e., verb– adverb. The most frequent adverb is ikke 'not', and we will start with this.

English negative subordinate clauses have do-support, [4], unlike heritage Norwegian ones, which have nothing like that, [7].

[4] a man who didn't work up there

The heritage variety has no do-support or similar construction, so there is no pattern that has been transferred from English here.

Other adverbs in English are actually in a preverbal position, which makes English much more similar to European Norwegian than heritage Norwegian.

```
[5a] a man who actually/often/rarely worked up there ADV-V (her. and Eur.Norw.)
[5b] * a man who worked actually/often/rarely up there V-ADV (her. Norw.)
```
English distinguishes between auxiliary verbs and main verbs. Auxiliary verbs do not need do-support, while main verbs do, see [6a–c].

[6a] a man who couldn't work up there

[6b] \* a man who worked not up there

[6c] a man who didn't work up there

In heritage Norwegian there is no do-support, no fixed word order adverb–verb, and no difference between auxiliary and main verbs, as exemplified in [7].

[7] Det var mange ord der som vi bruker ikke it was many words there which we use not 'There were many words there which we do not use.' (westby\_WI\_03gk)

There is therefore nothing in the new word order that could be seen as transfer from English.

A generalized V2 pattern. The new subordinate clause verb–adverb order resembles the main clause word order in heritage and European Norwegian, as seen in [8].

[8] det går ikke it works not 'It doesn't work.' (aal\_01\_um)

Could the new subordinate word order simply be a generalisation from the main clauses? If they were, we would also expect subordinate questions to have generalised the V2 pattern. Consider the main clause questions and subordinate questions of European and heritage Norwegian in [9]:


The sentence in [9a] shows that subordinate questions have the canonical subordinate order with the non-V2 order, Wh-subject-verb, while [9b] shows that the main clause questions have the canonical V2 order, Wh-verb-subject. When subordinate questions therefore show no development towards main clause order, it is unlikely that the word order of other subordinate clauses should have been generalized from main clauses.

Attrition. Could the reason for the verb–adverb order be that the speakers are attrited? According to Montrul (cf. 2008: 261), loss or attrition of a first language-L1-leads to lexical retrieval delays, but that core syntax remains intact. Håkansson (1995) studied young expatriates who had not spoken Swedish since childhood and should be categorized as attrited. Though their morphology was affected, their syntax remained unchanged. When Larsson / Johannessen (cf. 2015a: 254–257) found attrited speakers in their recordings, the speakers had a simplified language, with no embedded sentences with adverbs, and had very few sentences with the target V2word order in main clauses. Those that actually had a complex syntactic structure with embedded clauses containing adverbs, had target L2 in main clauses and were generally fluent. They conclude that attrition is not a factor that can explain the non-target subordinate clause word order.

Incomplete acquisition. Finally, it is possible that the subordinate word order is difficult to learn, and therefore is not properly acquired by the heritage language speakers. It has been shown that in European Norwegian, L2 learners struggle with this word order, L1 speakers acquire it very late. Two Scandinavian investigations show that at age 4 in one study and age 5:9:18 in another, both the nontarget verb–adverb order aswell as the target adverb–verb. Consider example [10] from North Norwegian.

[10] huske du koffer han Karsten var ikke i barnehagen? remember you why he Karsten was not in nursery.def 'Do you remember why Karsten wasn't in nursery?' (Iver 5;9.18) (Westergaard / Bentzen 2007: 285)

Larsson / Johannessen (2015a; 2015b) explain the late acquisition of this construction on the basis of steps of acquisition with respect to verb movement: Children first learn that the verb moves to the IP projection (due to all ordinary subject-initial main clauses), and at a later stage onto the CP projection (due to the existence of object-initial main clauses). However, subordinate clauses represent a further step, in which the child has to learn that there is no movement in embedded clauses, thus allowing the verb to stay behind the adverb. In heritage language environments, it will be even more difficult to learn this construction, since subordinate clauses are less frequent than in the homeland, given a general attrition of many speakers. Furthermore, and most importantly, the heritage children start school at the same time as they were going to learn the final step of the subordinate clause word order. The consequence of this is that they will hear much less than before of their heritage language, and that learning a new language (the majority language English) will interfere with the finer details of their first language. They therefore conclude that incomplete acquisition is the reason for the non-target subordinate clause word order. Note that the target order also exists, so both orders are learned, but the speakers have not yet gained the nonlearning (i. e., the non-movement) represented by the last step.

#### 7 Conclusion and Research Desiderata

The goal of this paper was to show how good sources of language data, especially speech corpora, can be used to find out what constructions are used and how common they are in a certain language variety. As a case in point, I chose research based on Larsson / Johannessen (2015a; 2015b), on the non-target word order in subordinate clauses in heritage Norwegian, which is not just arbitrary mistakes by certain individuals, but is systematic in this variety. The research question of why this order has come about to be part of the heritage Norwegian variety, was answered going through four hypotheses, where searchable corpora and other data (old recordings by Einar Haugen, the Corpus of American Nordic Speech, CANS, and the Nordic Dialect Corpus, NDC) were used to answer some of them.

A research topic that naturally presents itself is the word order in expatriate Norwegian children. There are many Norwegian families outside of Norway, in Europe as well as in the USA, and the language of children in these families would be extremely interesting in the light of the findings discussed here. If Larsson / Johannessen (2015a; 2015b) are right, children with a comparable sociolinguistic biography (having learnt Norwegian as an L1 and then at some point close to school age learning and using a different L2 language, which subsequently becomes their dominant language) should also end up with a syntactic pattern in their Norwegian similar to that described here for American heritage Norwegian.

# Bibliography


Janne Bondi Johannessen, Cand.philol. and Dr.philos. University of Oslo Center for Multilingualism in Society across the Lifespan & the Text Laboratory Department of Linguistics and Scandinavian Studies P.O.Box 1102 Blindern 0317 Oslo Norway jannebj@iln.uio.no https://www.hf.uio.no/multiling/english/people/core-group/jannebj/index.html

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

Historical Contact Scenarios

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Stefaniya Ptashnyk

# Language Variation in Multilingual Historical Settings: Multilingual Practices at the University of Lemberg in the Late 19th Century

Abstract: This article explores language variation in multilingual written texts. The multilingual practices in historical texts from the domain of the university administration are presented, with lecture catalogues from the University of Lemberg published between 1848 and 1918 being used as the corpus for analysis. The dynamics during the investigation period are also presented. One can observe that language shift and language variation are connected to contemporary language policies. In addition, the functions of variation are analyzed, and the historical material shows that language choice is dependent on the language laws and decrees, and also reflects the societal prestige of the respective languages. Moreover, it can provide evidence for some uncertainties in language use, especially during the periods of changes and transitions.

Keywords: language politics, multilingual practices, language of instruction, lecture catalogues, biscriptality

Abstract: Дана стаття присвячена явищам варіантності у багатомовних письмових текстах. На базі корпусу каталогів лекцій Львівського університету, котрі були опубліковані між 1848 і 1918 роками, шляхом мовного аналізу показано палітру багатомовних практик в історичниx текстах із галузі адміністрації вищої освіти та їх динаміка протягом досліджуваного періоду. На базі проаналізованого матеріалу можна констатувати, що перехід від однієї мови до іншої та мовна варіація до певної міри пов'язаніз тогочасною мовною політикою. Окрім того, у статті проаналізовані функції мовної варіації, і тут історичний матеріал показує, що вибір мови залежить не тільки від приписів мовного законодавства, але й відображає суспільний престиж дaної мови, а також може свідчити про невпевненість, відсутність рутини у використанні певних мов, зокрема в періоди змін і переходів.

Keywords: мовна політика, багатомовні практики, мова викладання, програма університетських лекцій, диграфія

# 1 Preliminary Remarks

'VARIATIONist Linguistics meets CONTACT Linguistics': This slogan was the starting point of the successful linguistic meeting in Ascona in May 2018. At the same time, this slogan can be interpreted as evidence for a long linguistic research tradition: Whereas the term 'variation' was coined in the context of sociolinguistics, variation linguistics and dialectology, it describes phenomena within one language (cf. Christen 2010: 145–160; Glück 2005: 746–747; Lenz 2010: 295; Schmidt / Herrgen 2011: 14), the term 'language contact' instead refers to linguistic constellations involving two or more languages (cf. Riehl 2004). As a consequence, for an extensive period of time variation phenomena and language contact phenomena have belonged to different subfields of linguistics, or it was at least suggested as such.

Nevertheless, in recent research, the concept of variation has also been used within contact linguistics and multilingualism research with regard to multilingual communities which involve two or more languages present in linguistic interaction. In multilingual contexts, variation includes both—different varieties of a language as well as two or more typologically different languages (cf. Franceschini 1998: 12; Veith 2002: 135). This is because 'multilingual variation' can show many parallels with variation within a historical language.

"Whoever speaks selects; the result of this process is variation. Variation is the visible consequence of the fact that several regular options exist in a speaker's space of linguistic possibilities. Each linguistic action is preceded by numerous decisions and selection, conscious and unconscious (and semi-conscious). This concerns different levels, from phonology to system grammar; on these different levels selection takes place, which—inter alia—also serve as a means of social positioning of the speakers." (Lenz / Plewnia 2018: 1; translation: SP)<sup>1</sup>

This statement is primarily related to language variation within a language but is also valid for multilingual contexts: In a multilingual, as well as a monolingual situation, speakers are confronted with the need to adapt their speech to the extra-linguistic (social and situational) factors. Furthermore, they are required to choose the appropriate way of expression, using the appropriate variety or language. In such instances, monolingual speakers can choose between different varieties of the standard–nonstandard continuum (dialects, styles, or registers), while multilingual speakers can additionally use the available (contact) lan-

<sup>1</sup> "Wer spricht, wählt aus; das Ergebnis ist Variation. Sie ist die sichtbare Folge der Existenz von regulären Optionen im sprachlichen Möglichkeitsraum eines Sprechers. Einer jeden sprachlichen Handlung gehen zahlreiche Auswahlentscheidungen, bewusste und unbewusste (und halbbewusste), voran. Das betrifft die verschiedensten Ebenen, von der Phonologie bis zur Systemgrammatik; auf diesen verschiedenen Ebenen finden dann Auswahlentscheidungen statt, die u. a. auch der sozialen Positionierung der Sprecher dient."

guages which belong to their individual linguistic repertoire. Multilingual speakers have competences in two or more languages as well as in their subsystems, and all these languages and varieties may interact—as a result of language contact—with each other at all levels and may exert influence on each of them. The specific nature of the investigation of variation in multilingual contexts results from the fact that in multilingual communication every single language as well as its varieties (standard variety, dialects, local colloquial languages, etc.) must be considered.

Variation in multilingual settings has been intensively investigated in urban language research, especially in the research of communication in multilingual cities (see e. g. Kolde 1981; Löffler 2001; Schwitalla 2010). These investigations have shown that language choice and the switch from one language to another are related to different extra-linguistic factors resulting from individual preferences, competences of the speakers and the addressees, the speakers' attitude towards their own language(s) etc., as well as the circumstances of the communicative situation, for instance the language repertoire of the speakers' community, language loyalty, the prestige of the languages in the society, etc. (cf. Löffler 2005a: 159; Pütz 2004: 227). Barbour / Stevenson emphasize that language choice can also be influenced by political factors, and in many communities the use of certain languages in certain domains is decided by legislation (Barbour / Stevenson 1998: 249). Language planning, language laws as well as social ideologies are factors that determine and influence variation in a multilingual society (Barbour / Stevenson 1998: 17 f.)

Whereas in the last few decades the change from one language to another in spoken discourse has been very much in the centre of linguistic research, there are few studies on variation in written texts. However, from the early 1990s onwards, a number of scholars began to consider historical mixed texts as an interesting field of research. In this respect, the research in the English language is more advanced than German studies (Schendl / Wright 2011: 17).

In the following text I will use the term 'variation' as a broad definition and examine the phenomena of variation in a multilingual context from a historical perspective. My research focuses on the question of which forms of multilingual variation or 'multilingual practices' (for this term see Pahta et al. 2018) can be observed in historical texts and which functions they have during a given period.

A great variety of research material for such a question is provided by historical multilingual cities; one of them being Lemberg (today: Lviv / Львів) which belonged to the Austrian Empire from 1772 and 1918 as the capital of the Crownland of Galicia and Lodomeria. In the past, multilingual cities were places of multilingual communicative practices; numerous examples for this can be found in historical documents which were produced in different domains of social life. In particular, newspapers from the 19th century can be mentioned as a representative example for such phenomena (cf. Ptashnyk in press). Whereas there is a sufficient number of publications about the historical press, there is little research on the area of higher education. For this analysis, I will use written historical material from the field of education, namely the lecture catalogues from the University of Lemberg published between 1848 and 1918.

### 2 Structure of the Contribution

In the following article I will first provide a brief overview of the linguistic, ethnic and religious situation in Lviv in the 19th century. I will look at the most commonly used languages in the city and the linguistic situation at the University of Lemberg and its dynamics in connection with the language policy of the Habsburg state. Next, I will present my corpus and methodology of my work. In the following steps, I will then demonstrate the multilingual practices in historical documents between 1848 and 1918 and analyze their dynamics. To conclude, I will attempt to explain the functions of multilingual variation in the investigated material during the period of study.

# 3 Lemberg and its University between 1848 and 1918: Sociolinguistic Situation and its Dynamics

Lemberg in the 19th century contained several nationalities who lived in close proximity to each other. Different languages were used within this multilingual urban society. First of all Polish, Ruthenian/Ukrainian<sup>2</sup> and German. Besides these, Latin and Old Church Slavonic, Hebrew and Yiddish were used for written communication or in private situations (cf. Isajevycˇ et al. 2007; Ptashnyk in press). At that time, there was intensive contact between the respected languages. Some of the languages were represented in the domain of higher education, as it will be shown later. Since the focus of my investigation lies on printed documents regarding the University, its historic and linguistic situation must be mentioned.

Regarding many of the historiographical works on the University of Lemberg, the following phases are highlighted for the period up to the First World War (cf. Dybiec 2011; Finkel / Starzyn´ski 1894):

<sup>2</sup> Galician Ukrainian or Ruthenian was elaborated on the basis of the Galician vernacular, but, until the end of the 19th century, the standardization of Ukrainian was not completed (cf. Fellerer 2003: 109).


This period is marked by a very complex linguistic situation which was strongly influenced by the Viennese and by the local language policies. From the 1820s onwards, the Viennese government pursued the 'Germanisation' of the University of Lemberg. This process was interrupted due to the so-called 'spring of nations' 1848–1849. In 1848, national aspirations of the peoples arose in Galicia and in the whole monarchy. Postulates of equal rights for languages and nationalities ('Volksstämme') were formulated, although those were not legally guaranteed at that time. This process of language 'liberalisation' was stopped again, because in the following years, absolutism was restored. From 1852 to 1859 the Austrian emperor resolved the problem of different ethnic groups by restoring the pre-revolutionary hegemony of the German speaking elites. Between 1850 and the late 1860s, we can witness the advancing Germanisation of Galician secondary and higher education as well as of other areas of public life. Subjects at the University of Lemberg were mostly taught in German.

At the beginning of the 1860s, Austria returned to its constitutional principles regarding the national and language rights. Following the Austro-Hungarian Compromise in 1867, a rapid 'Polonization'started in Galicia due to the so called 'Galician autonomy': Even if the Article from the Constitution of the December 19, 1867 guaranteed equal rights to all nationalities and their languages (so called 'Landesübliche Sprachen', literally: 'languages usually used in a particular regions'), the new Galician language policieswere in favour of the Polish nationality and its linguistic rights. In 1869, the status of the Galicia's universal language Polish was provided, while German was obligatory only for official correspondence with the imperial administration in Vienna (Supreme Court, ministries, etc.).

Regarding the history of the university, the beginning of the 1870s is considered to be the phase of the language shift from German to Polish. According to the Decree of the Minister of Culture from July 11, 1871 (Z. 523/Präs.) to the Governor of Galicia and to the Academic Senate of the University of Lviv, the restrictions, which had until now prevented lectures in Polish and Ruthenian at the Faculty of Law, Political Science and Philosophy at the University of Lviv, had to be completely abolished (Beck von Mannagetta / Kelle 1906: 17). In 1871 Polish was introduced into the administration of the University of Lemberg. We can observe the increasing dominance of Polish as the language of instruction at the University of Lemberg after 1874. At the beginning of the 20th century, most of the university's lectures were held in Polish.

#### 4 State of Research, Corpus and Methodology

While the multilingualism of the Habsburg Monarchy has often been the focus of historical and linguistic research, it is almost surprising that there is very little research on the sociolinguistic situation at the University of Lemberg. Most historians argue that the shift from Latin to German as the language of instruction occurred soon after 1824 (cf. Roskau-Rydel 1999: 47). In the late 19th century, the university was completely polonized. Recent research shows instead, that the teaching process at the university was multilingual during the whole Austrian period (cf. Ptashnyk in press).

Even fewer empirical studies are available concerning the administrative language use in higher education in Lemberg. In my research I attempt to close this 'gap'. As already mentioned above, my analysis is based on institutional documents, namely on lecture catalogues (in Latin called catalogi lectionum or catalogi praelectionum), a text genre that has so far been insufficiently investigated (Rasche 2009: 44).

The historical lecture catalogues are considered to be interesting material from a linguistic point of view for several reasons: Firstly, they document the contemporary language use of the university authorities and teachers. Furthermore, they enable the discovery of the processes and factors of language shift in institutions such as universities.

Regarding the drafting of the lecture catalogues it is important to mention —according to Clark (2006)—that the production took place under the supervision of the rector or vice-rector. The composing itself would usually be undertaken by the professor with the most competent knowledge of Latin, usually the professor of rhetoric or Latin philology, as was the case in Berlin or Bonn. "This professor typically had to pen or at least edit documents published in the name of the university, for example, encomia on the sovereign's birthday and, later, the lecture catalogue." (Clark 2006: 33) By the decree of May 14, 1810 concerning the publishing of the lecture catalogues, the Court of Audit Committee (Studienhofkommission) ordered that every catalogue had to contain the following information: "what, when, for how long, by whom and following what kind of reading books will be taught at the university throughout the academic year". (Sr. k. k. Majestät Franz des Ersten politische Gesetze und Verordnungen […] 1911: 98; translation: SP)<sup>3</sup>

In 1850/51, the question of the lecture catalogues at the Imperial-Royal Universities was regulated by two further legal acts—by the Decree of the Ministry of Education of June 30, 1850, Z. 5432/178 and by the Decree of the Ministry of Education of February 3, 1851, Z. 1147/125. In these documents the procedure for the compilation of catalogues was described in detail. Thus, the government aimed for a further uniformity of the lecture catalogues for all Austrian universities.

Exploring the language variation in a multilingual context, I will use the lecture catalogues from the University of Lemberg, which were regularly published, starting with the academic year of 1808 (cf. Finkel / Starzyn´ski 1894: 217). The available documents, which originate from the 1820s, are entitled "Ordnung der öffentlichen Vorlesungen". After the Polonization of the public domains in Lemberg as well as of the University, the lecture catalogues are entitled in Polish "Program wykładów". The size of the lecture timetables constantly increased: in 1848, it had only 20 pages. In 1860, it already contained 32 pages, and in 1910, the Polish version contained 52 pages.

My analysis is based on catalogues published between 1848 und 1910. The surveys were carried out in 5-year increments as well as in selected additional years in which special changes were recorded. However, due to a lack of space, only selected examples will be shown. The lecture catalogues usually have one main language ('frame language'), but overall, they are multilingual. My research focuses on text passages in which several languages are used side by side. In order to analyze the texts,I will investigate the dynamics in frequency and in patterns of language variation: Does the language choice for a given text fragment follow a certain scheme or principle? Besides this, I will attempt to analyze the functions of these multilingual practices aswell as the function of every occurring language.

<sup>3</sup> "was, wann, durch wie lange Zeit, von wem und nach was für Vorlesebüchern an der hohen Schule im Verlaufe des ganzen Schuljahres gelehrt werde."

#### 5 Analysis and Results

#### 5.1 Lecture Catalogues around 1848–1850

The lecture catalogue from the academic year 1848 is entitled

"Ordnung der öffentlichen Vorlesungen, welche an der Weiland Seiner kais. königl. Majestät Franz I. allerhöchsten Namen führenden Universität zu Lemberg im Schuljahre 1848 gehalten werden. Lemberg: aus der k.k. galizischen Aerarial-Druckerei." 'Organization of public lectures given at the Weiland of His Imperial-Royal Majesty Francis I, the Highest Name leading University of Lemberg, in the school year 1848. Lemberg: from the Imperial-Royal Galician Aerarial Printing Company.'

The main language (frame language) is German. In several text passages, however, Latin and Polish are chosen to impart certain information.

The first part (Personalstand, 'staff ') is written in German, but one encounters a variation phenomenon, which can be described as biscriptality. The surnames of the teachers, Polish as well as Ukrainian and German. (e. g. Lewicki, Grassl, Mauss, Z˙ erdzin´ski), are all written in Antiqua. For the first names, as well as other details, Fracture is used (cf. Figure 1):

Figure 1: Lecture Catalogue of the University of Lemberg for the academic year 1848, page 4

The actual list of lessons begins on page 9. In addition to the German frame language, Polish and Latin are used, albeit for a specific function: the latter two languages appear exclusively in the titles of textbooks and in the names of their authors.

As we can read on page 9, "Klein's: Historia Ecclesiae Christianae (Tom. II)" the subject "Church History" was taught in Latin. On page 10 it is stated that "Moral Theology" was taught in Latin, "Ambrosii Stapf: Theologia moralis". At that time, Latin textbooks were not only used for theology lessons, which still mainly took place in Latin. Also for lectures with German as the means of instruction, for example at the Faculty of Philosophy, Latin textbooks were still often used in 1848, and their titles are indicated in the present document in Latin; cf.: "Reine Elementar-Mathematik" was taught in Latin, according to "Appeltauer's: Elementa matheseos purae (Viennae, II. Tom)" (page 17).

Comparable co-occurence of two languages is further proven with the combination German and Polish; for example:

"Das ehemalige polnische Recht, nach Ostrowski: Prawo cywilne narodu polskiego, in lateinischer Sprache" 'The former Polish law, after Ostrowski's Civil law of the Polish people, in Latin language' (page 12; translation: SP)

Another example on page 15:

"Theoretische Geburtshilfe in polnischer Sprache, nach eigenem Werke: Zasady sztuki połoz˙niczej, dla niewiast tejz˙e sztuce sie˛ oddaja˛cych (Lemberg 1837) … lehrt Professor Felix Pfau" 'Theory of Obstetric, in Polish language, after the own work: The principles of the art of midwifery, for women who devote themselves to this art (Lemberg 1837) … teaches Professor Felix Pfau' (page 15; translation: SP)

In the analyzed catalogue from 1848, the lecture titles are always given in German. But this fact does not necessarily mean that the lectures are held in German: around 1848, lessons are still taught in Latin to a large extent. There are two 'techniques' for providing information on the language of instruction. In many cases, the language of instruction is explicitly given as in the next passages:

"Die Dogmatik … lehrt in lateinischer Sprache Herr Professor Sebastian Michael Tuczyn´ski" 'Dogmatics … is taught in Latin language by Professor Sebastian Michael Tuczyn´ski' (page 10; translation: SP)

"Exegetische Vorlesungen, in lateinischer Sprache, … hält Herr Professor Jacob Gierowski" 'Exegetic lectures, in Latin language … are held by Professor Jacob Gierowski' (page 10; translation: SP)

"Das ehemalige polnische Recht, nach Ostrowski: Prawo cywilne narodu polskiego …in lateinischer Sprache" 'The Former Polish Law, after Ostrowski's: Prawo cywilne narodu polskiego … in Latin language' (page 12; translation: SP)

These examples are clear evidence of multilingualism in the university system and demonstrate the discrepancy in language use for administrative communication and teaching. Whereas the language of administration is predominantly German, the lectures are often held in Latin and in very few cases in Polish. In the last example, even trilingualism can be observed: German as the language of administration is the descriptive language of the document (its frame language), Latin is mentioned as the language of instruction, and Polish appears as the textbook language.

The second technique for providing information on the language of instruction can be demonstrated by the following example: On page 16, the question of the language of instruction is explicitly commented on in the footnote:

"Zur Lehrsprache in den gesammten Gegenständen der philosophischen Studien ist durch den §. 31. des mit allerhöchsten Entschließung vom 28. September 1824 genehmigten philosophischen Lehrplans die deutsche Sprache vorgeschrieben, mit Ausnahme der lateinischen Philologie, über welche der Vortrag in lateinischer Sprache gehalten wird. (§ 8. des Studien-Plans)" 'German is prescribed as the language of instruction in all subjects of the Faculty of Philosophy by §. 31 of the philosophical curriculum approved by the supreme resolution of September 28th, 1824, with the exception of Latin philology, where the lectures should be held in the Latin language. (§ 8 of the Curriculum)' (page 16; translation: SP)

This note refers to the above-mentioned regulation of the language policy from 1824. Accordingly, one can assume that German served as the language of instruction in all subjects at the Faculty of Philosophy, with the exception of Latin philology. However, one encounters another exception: on page 19 we find the information that the subject "Polish language and literature" is taught "in Polish language" in adherence to the textbook "Gramatyka je˛zyka polskiego" 'Grammar of Polish language'.

The lecture catalogues for both terms of the academic year 1850 show similarities with the catalogue of 1848, as well as some differences. The aspects which they share include the phenomenon of biscriptality: the same as in 1848, also in the catalogues for 1850 all teacher's surnames are written in Antiqua. For other information the Fracture font has been used consequently. All lecture titles and subject descriptions are still given in German. In contrast to the edition of 1848, the references to the textbooks have completely disappeared. Accordingly, there was no longer 'space' for Latin or Polish in this function. German is the only description language of those documents. Therefore, in 1850, we deal with the very rare example of a monolingual catalogue at the University of Lemberg, whereas at the same time the teaching continued to be multilingual.

Five years later, the phenomenon of biscriptality appeared again, but in a new form: Fracture was no longer used, but owing to the use of Ukrainian, the Cyrillic font appeared. The Cyrillic alphabet was generally used for the announcement of Ruthenian lessons; in very few cases we also come across Antiqua in transliteration of Ukrainian indications.

In 1855, the catalogues become multilingual again: Besides German, Ukrainian, Polish and Latin are used as description languages. However, remarkably, for some lecture titles a translation into German is given, cf. page 12 on Figure 2: The subject "Dogmatics" was taught in Ruthenian; the announcement of this lesson starts in Ruthenian, but at the end of the catalogue entry a German translation with all information is given. Similar phenomena can be observed not only for Ruthenian, but also for Polish lessons.


Figure 2: Lecture Catalogue of the University of Lemberg for the academic year 1855/56, page 12

#### 5.2 Lecture Catalogues in the 1860s

Next, we look at the catalogue for the academic year 1860/61. The title, which is in German, is

"Ordnung der öffentlichen Vorlesungen an der k.k. Universität zu Lemberg im Winter-Semester des Studien-Jahres 1860/61"

The frame language is German. Furthermore, Ukrainian, Polish and Latin are used as description languages.

Latin as the language of announcement appears to be the rule at the Faculty of Theology: all information concerning the lectures (title, name, date, time, room, etc.) is provided in Latin. In some rare cases in 1860, the theological lectures were taught in German—this is the case of "Erziehungskunde" 'educational science'. All information related to this subject is provided in German as well. Furthermore, Polish also appears as the language of description, for example for the subject "Teologia pasterska" 'pastoral theology' which was taught in Polish. In contrast to the catalogue of 1855, it is announced in Polish only, without any translation. Therefore, we can assume that the author of the catalogue seems to assume that its readers can understand Polish.

The case of the Ukrainian announcement is different, even though it concerns the same subject. On page 13 it states that "Богословіє пастђрскоє" is taught in Ruthenian; at the same time all the information is also announced in German (cf. page 13 on Figure 3).

Figure 3: Lecture Catalogue of the University of Lemberg for the academic year 1860/61, page 13

In contrast to the Faculty of Theology, the lectures at the Faculty of Law and Political Science are announced (and held) exclusively in German. Furthermore, for lectures at the Faculty of Philosophy, the absolute dominance of German can be noted. A special feature is the announcement of the lecture on "Ruthenian Language" and "Ruthenian Literature" on page 20 which is bilingual (Ukrainian with German translation). The announcement of the lecture on the "History of Polish Literature and Stylistic Exercises" is also written in Polish with a German translation. It contrasts with the case of "Teologia pasterska" as mentioned above. There seem to be uncertainties in the use of Polish as language of description.

To sum up for the academic year of 1860, we generally observe considerably less language variation within the individual entries in the course catalogue than in the previous evaluated editions. The entries show more regularity and more certainty regarding the choice of the description language in the course catalogue. Nonetheless, there is some indecision as to whether or not a translation into German is required.

In contrast to the earlier documents, the overview in table form ("Tabellarische Übersicht") in the course catalogue of 1860/61 is designed differently: it is multilingual, containing Latin (especially the announcements in theology), German (mainly at the Faculties of Law and Philosophy), Polish (for individual subjects such as "Teologia pasterska" 'pastoral theology', and finally Ukrainian (e. g. the lecture in "Богословіє пастирскоє" (Bohoslowje pastyrskoje, i. e. 'pastoral theology', see page 22), the latter however in Latin transcription.

The use of Antiqua for Ukrainian announcements offers various interpretations: It may have been pure convenience for the typesetting on the one hand; on the other hand it could also be interpreted—especially in the context of the so called 'alphabet war' <sup>4</sup>—as a deliberate rejection of the Cyrillic alphabet. The latter thesis can only be maintained to a limited extent, since the Cyrillic alphabet was used in several passages in the list of lectures on the previous pages. In any case, there is a lot uncertainty to observe in the choice of the font.

The analysis of the occurrence of the description languages shows strong differences regarding their frequency, which can be linked to the differences in their presence as languages of instruction. There is also a striking unsteadiness in Ukrainian and Polish announcements: Sometimes they are translated into German, sometimes not; there is no clear pattern and no clear strategy. In my opinion, this is because the two languages were relatively new in the sphere of academic communication at that time and they were used infrequently for teaching purposes. The choice of Ukrainian is interconnected with most uncertainties and it manifests itself in biscriptality—the Ukrainian remarks are partly written in Antiqua and partly in Cyrillic.

At the same time, in comparison with the catalogue of 1850, some peculiarities of the catalogue of 1860 indicate increasing unification: In general, in 1860 there is no longer explicit information regarding the language of instruction. There appears to be further clarity regarding this question. Language of instruction is either regulated by law for the all faculties, or it can be derived from the language of the announced lecture title.

#### 5.3 Lecture Catalogues after 1867: German and Polish Parallels in 1870s

As mentioned above, the beginning of the 1870s is considered to be a period in which the language shifts from German to Polish as the official language in city administration, as well as in higher education in Lemberg. During this phase, the lecture catalogues were published as parallel editions in both Polish and German.

<sup>4</sup> In 1859, the administration of the Galician Governor Gołuchowski proposed that the Latin alphabet, instead of the traditional Cyrillic one, had to be introduced for all Ukrainian publications. This provoked the so-called Second Alphabet War in 1859–1861. Finally, these efforts were aborted following concerted protest by Ukrainian intellectuals and clergy that blocked the implementation of any alphabet reform (cf. Magocsi 2002: 19).

The practice of parallel announcement of university lectures in two different languages has a long tradition. The coexistence of Latin and German lesson catalogues in European Universities in the 18th century was already mentioned previously. In this respect, for Lemberg this was not a 'revolutionary' practice, only a kind of 'translation' into the contemporary Galician realities.

Now, I would like to go into detail about two editions that appeared in the mid-1870s and were even printed in the same printing house:

The Polish edition of 1876:

"Wykaz Wykładów odbywac´ sie˛ maja˛cych w c.k. Uniwersytecie imienia Cesarza Franciszka we Lwowie w letnim pólroczu 1876. Lwów. Z I. Zwia˛zkowej drukarni, Hotel Z˙ orz˙a, 1876."

The German edition of 1876/77:

"Ordnung der öffentlichen Vorlesungen an der k.k. Universität zu Lemberg im Sommer-Semester des Studien-Jahres 1876/77. Lemberg. Aus der Ersten Vereins-Buchdruckerei im Hotel George, 1877."


Figure 4: Polish edition of the lecture catalogue of 1876 (Wykaz Wykładow … w letnim pólroczu 1876), page 4

The main language of the first document is Polish, and a maximum amount of information in Polish is provided here—the title of the lecture, the names and titles of the teachers, the time and place of the lecture. The language of instruction is explicitly indicated using the Polish description language. Therefore, regarding the administrative language use, in the Polish lecture catalogue for the academic year 1876, one can observe the tendency towards monolingualism. At the same time, the lectures were taught in four languages: For example, all titles of lectures at the Faculty of Theology are given in Polish, but the language of instruction was mainly Latin (for example, canon law), sometimes also German (e. g. for lectures in pedagogy), rarely Ukrainian or Polish (e. g. "Katechetyka i Metodyka") which is explicitly mentioned in the lecture catalogue; cf. Figure 4.

On page 4, note 2, it is explicitly mentioned that lectures at the Faculty of Theology are held in Latin, with the exception of pastoral theology, catechetic and methodology, which are taught in Polish and Ruthenian; pedagogy is held in German as we could observe already in Figure 4:

"Wykłady teologiczne odbywaja˛ sie˛ w je˛zyku łacin´skim, z wyja˛tkiem teologii pastoralnej, katechetyki i metodyki, przy których uz˙ywa sie˛ je˛zyk polski i ruski, przy pedagogii zas´ je˛zyk niemiecki." 'Theological lectures are held in Latin, with the exception of pastoral theology, catechetics and methodology, where Polish and Russian are used, while German is used for pedagogy.' (page 4; translation: SP)

Also lectures at the Faculty of Law are provided with explicit information on the language of instruction, cf. page 5:

"Historya prawa i pan´stwa niemieckiego … w je˛zyku niemieckim" 'Law History and State History … in German Language' (page 5; translation: SP)

"Prawo kanoniczne … w je˛zyku polskim" 'Canon Law … in Polish Language' (page 5; translation: SP)

In addition, it is explicitly noted on page 7 that lectures at the Faculty of Law are held in Polish with a few certain exceptions:

"Uwaga. Wszystkie wykłady w fakultecie prawniczym odbywaja˛ sie˛ w je˛zyku polskim, z wyja˛tkiem tylko wykładów P. profesora Dr. Buhla, który wykłada po niemiecku, i PP. Docentów Dra Dobrzan´skiego i Dra Ogonowskiego, którzy wykładaja˛ po rusku." 'Please note: All lectures in the Faculty of Law are held in Polish, with the exception of the lectures of Professor P. Dr. Buhl, who teaches in German, and PP. Docents Dr. Dobrzan´ski and Dr. Ogonowski, who teach in Ruthenian' (page 7; translation: SP)

This document presents the double 'explicitness' by providing information on the language of instruction. This is unsurprising, because many subjects during this period underwent a major shift from German to Polish as the language of instruction, and this process of language shift was accompanied by many difficulties and uncertainties. A few years later, most of the lectures were held in Polish. The multiple—explicit and implicit—indication of the language of instruction in lecture catalogues demonstrates the effort that this transition was associated with.

The German version of the lecture catalogue of 1876/77 titled as

"Akademische Behörden an der k. k. Universität zu Lemberg, sammt der Ordnung der Vorlesungen an derselben"

follows the established tradition of multilingual practices which we know from the lecture catalogues of the previous decades. Therefore, lectures in the field of theology are mostly announced in Latin: "Dogmatica generalis, quotidie ab hora 10–11 mane, Dominus Professor publicus ordinarius Dr. Sylvester Sembratowicz. (in Audit. IX.)" (page 3)

In contrast to previously analysed German-language editions from the past, all administrative information for the lectures, including location (room number), is given in Latin. There is no language variation within the individual entries, a stronger stringency is achieved.

The lectures were held mostly in Latin with very few exceptions, as we can read on page 3. The following lecture was held in German:

"Erziehungs-Wissenschaft, Dienstag von 10–12 Uhr Vormittags, k.k. ord. öffent. Uni. Prof. Herr Dr. Franz Kostek. (Sall Nr. XI.)" 'Education Sciences, every Tuesday 10–12 am, Imperial-Royal public Professor Ordinarius Dr. Franz Kostek. (Romm Nr. XI).'

As this example shows, all information on the German-language lectures, including the room number, is given in German. In contrast, the announcements of the Ruthenian and Polish lectures are usually multilingual; see the following text passages:

"Богословіе пастырское, 9 годинъ тыжднево въ понедђлокъ, середу, пятокъ, и соботу отъ 9–10 зъ рана, въ вторникъ отъ 9–10 зъ рана, Благород. Господинъ Проф. публ и порядочныІй Др. Фр. Кoстекъ (въ сали І.) (Pastoraltheologie, in ruthenischer Sprache, 9 mal wöchentlich, am Montag, Mittwoch, Freitag und Samstag von 9–11 Uhr, am Dienstag von 9–10 Uhr Morgens, k. k. öffentl. Univ. Prof. Herr Dr. Franz Kostek, (Saal I.)" 'Pastoral theology, in Ruthenian Language, 9 times a week, on Monday, Wednesday, Friday and Saturday from 9–11, on Tuesday from 9–10 a.m., Imperial and Royal Public University Professor Dr. Franz Kostek' (page 4)

First, the title, the extent and the days of lectures as well as the personal information and the venue (room 1) are given in Ukrainian, followed by its translation into German; the same pattern can be found in the announcement of the Polish-speaking "Teologia pasterska", "Katechetyka i metodyka" (cf. also page 4).

Explicit information on the language of instruction at the Faculty of Theology can be found in note 2 on page 4. In this way, there are multiple assurances given that the language of instruction has been communicated to the students with maximum clarity:

"Die Vorträge über die theologischen Lehrfächer werden in lateinischer Sprache abgehalten, mit Ausnahme jener über Pastoraltheologie, Katechetik und Methodik,welche in zwei Landessprachen (der polnischen und ruthenischen) dann der Erziehungskunde, welche in deutscher Sprache statt findet." 'The lectures on theological subjects are held in Latin, with exception of those on pastoral theology, catechetics and methodology, which are held in two national languages (Polish and Ruthenian) and educational science, which is held in German' (page 4; translation: SP)

At the Faculty of Law, the German-language lecture titles are announced in German, while the Polish and Ukrainian-language lecture titles are written in Polish and Ukrainian, respectively, with a German translation of each entry. Latin is completely missing in this list of lectures, because at that time, it was no longer used as a language of instruction at the Faculty of Law. Cf. the following examples

"Ueber deutsche Reichs- und Rechtsgeschichte, k .k. ordentl. öffentl. Universtäts-Professor Dr. Buhl, täglich von 10 bis 11 Uhr. (Im Saale Nr. VII)" 'About the German Imperial and Legal History, k.k. public full University Professor Dr. Buhl, daily from 10 to 11 o'clock. (In room no. VII)' (page 5; translation: SP)

"Prawo cywilne powszechne austryackie (Oesterreichisches allgemeines Privatrecht), k. k. ord. öff. Uni. Prof. Dr. Fangor, täglich von 8–9 Uhr. (Im Saale Nr. IV)" 'Austrian general private law, k.k. public full University Professor Dr. Fangor, daily from 8 to 10 o'clock. (In room no. IV)' (page 5; transation SP)

"Австрійскоє поступованье карноє (Oesterreichischer Strafprozeß), Privatdocent Dr. Dobrzanski, täglich mit Ausnahme der Donnerstage von 7–8 Uhr Früh. (Im Saale Nr. V.)" 'Austrian Criminal Trial, Privatdocent Dr. Dobrzanski, daily with the exception of Thursdays from 7 a.m. to 8 a.m. (In room no. V.)' (page 6; translation: SP)

Other information, i. e. the details of the lecturer, the room and time, are all written in German. In that academic year, the majority of teaching in law took place in Polish, but there were still a considerable number of German-language lectures.

In the table overview only Latin (for information about the lecturers) and German (for information about the lectures themselves) are present as descriptive languages. If the teaching had to take place in Ukrainian or Polish, this was explicitly noted, e. g. "Theologia pastoralis, lingua polona feria II. Audit VIII" (cf. Figure 5).

In general, during the late 1870s we can still observe the clear dominance of Latin for theological lectures as well as for their announcements, and the dominance of German for other faculties. The German version of lecture catalogues of the University of Lemberg was published until the academic years 1878/1879 after which their publication was discontinued. From then on, Polish was the only frame language of the Lviv lecture catalogues.


Figure 5: Lecture catalogue of 1876/77, the German edition (Ordnung der Vorlesungen … Sommer-Semester 1876/77), page 15

#### 5.4 Lecture Catalogues at the End of the 19th and at the Beginning of the 20th Century

About 15 years later, the University of Lemberg printed its lecture catalogues only in Polish. They are entitled as

"Uniwersytet imiena Cesarza Franicszka I. we Lwowie. Skład Uniwersytetu i Program Wykładów w [YEAR]. Lwów. Z I. Zwia˛zkowej drukarni we Lwowie [YEAR]".

Analysing the lecture catalogue of 1894/5, we note that this document presents a different picture as in the 1870s or in the 1880s. On page 2 there is a remark on the language of instruction, which is defined as faculty-specific: Latin predominates in theology; Polish dominates in the other two faculties (medicine is not yet part of the university):

"Wykłady na wydziale teologicznym odbywaja˛ sie˛ w je˛zyku łacin´skim, z wyja˛tkiem teologii pastoralnej, katechetyki i metodyki, tudziez˙ pedagogii, w których sie˛ uz˙ywa je˛zyka polskiego i ruskiego. Na wydziale zas´ prawnym i filozoficznym wykłady odbywaja˛ sie˛ w je˛zyku polskim, z wyja˛tkiem wykładów prawa karnego prof. dr. Stebelskiego i dr. Dobrzan´skiego, historyi powszechnej prof. dr. Gruszewskiego, które odbywaja˛ sie˛ po rusku, tudziez˙ niemieckiej filologii i literatury prof dr. Wernera, ktore sie˛ odbywaja˛ po niemiecku. Prócz tego istnieje takz˙e jedna katedra prawa cywilnego, tudziez˙ katedra je˛zyka i literatury ruskiej z wykładem ruskim, ale te obecnie wakuja˛" 'Lectures at the Faculty of Theology are held in Latin, with the exception of pastoral theology, catechetics and methodology, as well as pedagogy, in which Polish and Ruthenian are used. At the Faculty of Law and Philosophy, lectures are held in Polish, with the exception of lectures on criminal law by Prof. Dr. Stebelski and Dr. Dobrzan´ski, on common history by Prof. Dr. Gruszewski, which are held in Ruthenian, as well as on German philology and literature, which are held in German by Prof. Dr. Werner. In addition, there is also one chair of civil law, as well as a chair of Ruthenian language and literature with Ruthenian as the language of instruction, but these are currently vacant' (page 2; translation: SP)



Figure 6: Lecture Catalogue of the University of Lemberg for the academic year 1909/10, page 4

Regarding the Faculty of Theology, the lecture announcements follow the same pattern that was already common in the past: The language of instruction is the language of the whole announcement, e. g. Latin for "Dogmatica generalis" (page 26), Polish for "Pedagogia" (page 27) or Ukrainian for "Педагогія" (page 27). Consequently, the name of the lecturer, the time and the location of the lecture are given in the same language as the title.

Since Latin, Polish and Ukrainian are the official languages of the lectures at the Faculty of Theology, all three languages can also be found in the course catalogue. Latin is dominant in this case.

The lecture announcement at the Faculty of Law and Philosophy follows the same principle: the language of instruction corresponds to the language of announcement. See page 28: "Austriackie poste˛powanie karne" in Polish, "Австрійске поступованье карне" in Ukrainian, etc.


Figure 7: Lecture Catalogue of the University of Lemberg for the academic year 1909/10, page 26

In the late 19th century, German can only be found at the Faculty of Philosophy (page 33 of the lecture catalogue of 1894/95), it was in use for lectures on "History of German Literature in the 18th Century", on "The so-called younger Romanticism" and "Seminar for German Philology". In the analysed catalogue, their titles as well as the information about the lecturer (all three are read by Prof. Dr. Werner), the time and place are all given in German.


Figure 8: Lecture catalogue of the University of Lemberg for the academic year 1909/10, page 43

Furthermore, in the "Table Overview" of Lectures, Polish is used as the framework language and the lecture titles are provided in the respective language of instruction. The explicit details of the lecture language are now omitted.

The lecture catalogues for the academic years 1906/07 and 1909/10 are both titled and written in Polish. With regard to their structure and multilingualism, the lecture catalogues from the beginning of the 20th century hardly differ from the catalogue of 1894/94. The announcement of the lectures is made in Latin, Polish and Ukrainian, as well as German and follows the same principle: the announcement language corresponds to the language of instruction (cf. Figure 6). There are no further explicit remarks about the language of the lecture.

Finally, we look at the table overview in this catalogue on page 24: The frame language is Polish; the entries are made in the respective languages of instruction. As we can see in the Figure 7, the document varies between Latin, Polish and Ukrainian.

German language is finally found on page 43; the department of Philology offered lessons on "Deutsche Metrik" 'German Metric' and on "Schiller's Leben und Werk" 'Schiller's life and work' which were taught by Professor Schatz.

#### 6 Conclusion and Research Desiderata

By analysing the lecture catalogues of the University of Lemberg I tried to show how the societal multilingualism of the urban society manifested itself in documents of institutional communication. Between 1848 and 1914 the shift of the language of instruction at the University of Lemberg can be observed which reflects the dynamics of the sociolinguistic situation in the city. However, not only that, but much more, the administration of the University experienced specific dynamics in language use. During the whole period that was examined one can observe certain changing language dominance at the multilingual institution.

The societal multilingualism has left traces in such 'administrative texts' as lecture catalogues, in the form of multilingual variation which specific features and frequency varied during that time. Around 1850, at the beginning of the investigation period, we were dealing with a very wide variety. The language choice did not necessarily follow a uniform and clear scheme. Into the 1870s, there were still many uncertainties, frequent variations, and switches from one language to another and vice versa within the individual announcements.

Great uncertainties can be observed in the use of Polish and—even for a longer time—in the use of Ukrainian which appeared in lecture catalogues after 1850. In the earlier documents, there was no uniform handling for entries regarding Ukrainian and Polish lessons; sometimes German or Latin translations were added to Ukrainian and Polish announcements, sometimes not.

In the early years of investigation one often finds explicit information about the language of instruction, sometimes even twice—both individually for the respective lecture and then again as an overall rule for the respective faculty (e. g. in the form of references). This is also an evidence that there was a lot of uncertainty for external readers.

During a short 'interim period' in the 1870s in which the shift from the German to the Polish language as the official language took place there wasatendency towards monolingualism in the Polish editions. After this 'transition period' the multilingual scheme finally prevailed: a uniform frame language (it was consequently Polish after 1875) with multilingual lecture announcements which implicitly informed the reader in which language the lessons took place.

Around 1890, explicit information regarding the language of instruction disappeared. On the one hand, the choice of the language of instruction has now been secured by legal regulations. On the other hand, the experience with multilingual teaching during those years led to an established scheme of lingual handling of the announcements: The language of the title announcement revealed in which language the lesson would be held. The explicit language specifications were no longer necessary. The last lecture catalogues that were examined contained hardly any fluctuations: Only one language for every entry was used. In total, for the whole lecture catalogue, four languages were used, however, therewas no variation within a single entry; but—if necessary—the switches took place between the entries.

As already mentioned, all four languages—Polish, Latin, Ukrainian and German—are represented in the catalogues during the whole period of investigation, although in different quantities. Besides Latin and German, Polish and also Ruthenian achieved a certain 'normality' in their presence in the last decades of the monarchy. The lecture announcements in these latter two languages were no longer perceived as 'exceptional' or 'extraordinary', therefore the translations into Latin or German became redundant.

Because it is a specific feature regarding variation in Lemberg's lecture catalogues, the phenomena of biscriptality must be mentioned. Throughout the entire period of investigation, the analyzed documents contained at least two script systems. In the beginning they were Fracture and Antiqua, later Antiqua and Cyrillic. Regarding Lemberg, the Hebrew script must also be mentioned as well, although it was not used at the University until the end of the Austrian Empire.

At the beginning of the investigation period, Fracture and Antiqua coexisted in written documents; this fact can be attributed to the co-existence of German, Latin and Polish in the Galician multilingual society. The variation between the Fracture typeface and Antiqua disappeared very early from the lecture catalogues; already in 1855 there were no more passages in Fracture font. The Ukrainian language, which appeared in the catalogues after 1850, introduced the Cyrillic alphabet. During the 1860s and 1870s there were instances of Latin transliteration for Ukrainian text passages; therefore the biscriptality can be observed in a new shape. The later documents reveal no more transliterations. This can be seen as a result of the processes of language standardization, however at the same time, metalinguistic discussions also had a considerable influence on the more consequent use of the Cyrillic alphabet.

Reflecting on the presence and the predominance of certain languages in the lecture catalogues, we can infer that the linguistic situation at the University of Lemberg mirrored the contemporary language policies as well as the (unequal) prestige of the contact languages. While the languages with a higher social prestige were used more often, the others were less represented, and they produced more uncertainties which led to more variation in writing.

The analyzed material has already clearly demonstrated the trends in the development of multilingual practices in the educational administration in Lemberg, and how the use of contact languages has changed over time. Nevertheless, only a fraction of the communicative reality within an urban communicative system has been captured with the investigated material. Further research into other social domains is needed to complete the picture.

# Bibliography

#### Secondary Literature


Handbook of Linguistic Variation. Vol. 1: Theories and Methods. Berlin / New York: de Gruyter Mouton, 295–315.


Riehl, Claudia Maria. 2004. Sprachkontaktforschung. Eine Einführung. Tübingen: Narr. Romaine, Suzanne. 2000. Language in Society. An Introduction to Sociolinguistics. New

York: Oxford University Press.


#### Primary Literature


Dr. Stefaniya Ptashnyk Heidelberger Akademie der Wissenschaften Deutsches Rechtswörterbuch Karlstraße 4 69117 Heidelberg Germany stefaniya.ptashnyk@adw.uni-heidelberg.de http://www.deutsches-rechtswoerterbuch.de/ueber.htm

# Jelena Zˇivojinovic´

# Language Contact in Renaissance Ragusa

Abstract: The aim of this article is to provide the reader with a general overview of language change and results of language contact in Renaissance Ragusa. I will present the results of a synchronic study, based on the analysis of qualitative data pulled from a literary corpus reflecting the linguistic society of Ragusan Renaissance. Furthermore, the articles present a comparison with a smaller corpus of diachronic data, analyzed by Dragica Malic´. Calques, phonological adaptations, and new combinatorial morphological creations containing a Romance root and a Slavic derivational morpheme, such as: It. spacciarsi > spacˇati se or It. accomodarsi > akomodavati se, will be explored through additional examples in order to highlight the underlying contact and change strategies.

Keywords: contact, Renaissance Ragusa, Dubrovnik, morphological change, transmorphemization

Abstract: Lo scopo di questo articolo è quello di presentare una panoramica generale sul mutamento linguistico eirisultati del contatto linguistico nella Ragusa rinascimentale (oggi Dubrovnik). Presenterò i risultati di uno studio sincronico basato sull'analisi qualitativa dei dati tratti da un corpus letterario che rispecchia la società linguistica del Rinascimento ragusano, oltre ad un corpus di dimensioni più ridotte dei dati diacronici, analizzati da Dragica Malic´. Calchi, adattamenti fonologici ed innovazioni morfologiche contenenti una radice romanza e un morfema derivazionale slavo, quali it. spacciarsi > spacˇati se oppure accomodarsi>akomodavati se, verranno approfonditi attraverso ulteriori esempi in modo da mettere in luce le strategie di mutamento e il contatto sottostante.

Keywords: contatto, Ragusa rinascimentale, Dubrovnik, mutamento morfologico, transmorfemizzazione

#### 1 Introduction

Renaissance Dubrovnik was a city of great importance, particularly due to its strategic position. It was a crossroad for merchants and fishermen, but also scholars from the entire Mediterranean. It is a characteristic example of how multiple languages coexisted within a relatively small area. During an initial stage (approx. 15th century) we identified three main components, two of which are Romance languages and one is Slavic. Namely, Ragusan, Venetian and a Slavic vernacular<sup>1</sup> held very specific sociolinguistic positions. Unfortunately, there are no remaining written records of Ragusan<sup>2</sup> ,aDalmatian-Romance language, whose genealogical classification is still debatable<sup>3</sup> . We do, however, have a wide literature from the Renaissance period in Ragusa, which clearly illustrates the sociolinguistic stratification of this time.

My focus in this research will be to provide an overview of some aspects of language contact in Dubrovnik. Key aspects of this topic will be described, starting from the causes of contact, both social and linguistic predictors—that should result in a context of multilingualism. My goal is to conclude this insight in the Ragusan language contact by presenting the analysis of a literary corpus of data provided by Marin Drzˇic´, along with the the analysis of the Vatican Croatian Prayer Book and the Ragusan Academic Prayer Book, by Dragica Malic´.

I will start by displaying some major considerations in a socio-historical context, by describing legends and stories about the birth of Ragusa in Section 2. Section 3 will provide a brief explanation on the linguistic stratification and an overview of the spoken languages in Renaissance Dubrovnik. Section 4 will illustrate the state of research, while Section 5 presents some contact parameters and the filter used for the analysis of data. Namely, the phenomena of transphonemization and transmorphemization, developed by Rudolf Filipovic´ will be introduced. Finally, Section 6 will summarize some results of both synchronic and diachronic corpuses in the perspective of contact-induced change.

<sup>1</sup> There is not a specific label for this Slavic Shtokavian variety spoken in Dubrovnik. Certain documents, however, mention the term lingua serviana, such as Diversa de Foris 26, fol. 20' from 1613, stating: "Hic infra erunt registratae infrascriptae duae litterae ex lingua Serviana et charactere, recognitae prout ad mobilia extra 1613 folio 35 interpretatae per ser Paschalem de Primis cancellarium linguae Slavae". It is often represented as Serbian, Croatian or even Serbo-Croatian. Nevertheless, such terminology is anachronistic.

<sup>2</sup> There is, however, a poor, yet significant witness of an illustrious personage, Philippus de Diversis in the manuscript named "Descriptio Ragusina edita ab eximio magro Philippo de Diversis Quartigianis Luccensis, A.D. MCCCCXL ab eo confecta" that is located in the Franciscan monastery in Ragusa.

<sup>3</sup> For further information on genealogical classification of Ragusan check Gordon (2005), Muljacˇic´ (1962; 1999; 2001a), Voegelin (1976) and Zamboni (1976) among others, who offer different views and perspectives.

#### 2 Socio-historical Facts of Ragusa: A Brief Overview

Starting at the very beginning, the birth of Ragusa is considered emblematic regarding the issue of contact. As a matter of fact, a legend named Pavlimir, written by Junije Palmotic´, states that the protagonist of the story was Prince Pavlimir, whose ancestors were forced to flee to Rome due to local intrigues. The young prince was later called by local authorities to return to the motherland, defeat the opponents and restore the ancestral regnum Sclavorum situated in the Western Balkans. His effort paid off and Pavlimir was able to build a new town named Dubrovnik, by creating a symbiosis of Romance and Slavic cultures, which are characteristic of the Dalmatian coast.

Besides Palmotic´' drama, there is also a well-known historical version. Namely, Dubrovnik (as we know it today), which is a union of two different towns. In one instance, a town named Ragusium was built in the 7th century by a Latin population fleeing from a close town Epidaurus due to continuous attacks and thefts. They found shelter on a local island and in a short time they created ex novo a new homeland of great potential. In another instance, Slavic populations settled on the hills on the mainland, naming their town Dubrovnik. Recent archaeological excavations in the area of Pustijerna confirmed this version of a Latin and a Slavic component, where the remains of an early Christian church were found. In fact, the first attestations of Ragusa chronologically overlap the end of Epidaurus and the rise of a new maritime puissance. However, it should be emphasized that my topic of study is not the ethnicity of the population, but rather the linguistic component. Nonetheless, apart from the presence of numerous foreigners<sup>4</sup> , natives used to define themselves as Ragusans<sup>5</sup> , despite the continuous change of power at the top<sup>6</sup> . There was no patriotic attachment to the language<sup>7</sup> and this attitude made the difference concerning the linguistic situation and use of language.

Ragusa (Croat. Dubrovnik) was the perfect candidate for the development of language contact. Owning such a strategical geographical position, in a period when the Mediterranean was the center of European trade, Ragusa was a crossroad of merchants, craft workers, notaries and scholars. It wasapoint

<sup>4</sup> One of the most known communities was the one from Prato, well described in Bettarini (2007; 2012).

<sup>5</sup> Marino Darsa or Marin Drzˇic´, who will be introduced in Section 4, in fact, used to sign his works as "Marino Darsa raguseo" or "M.D.".

<sup>6</sup> Since its birth, Ragusa was governed, at times dominated and at times protected by several different powers that followed one another for centuries. At first Byzantium (ca. 7th century-1204) and sequentially, Venice (1204–1358), Hungary (1358–1526), the Ottoman Empire (1526–1684), and finally, Austria (1684–1806).

<sup>7</sup> Kuncˇevic´ (2012), Resˇetar (1933c).

bridging east and west, the hinterland of the Balkans and the Sea. Although there were numerous attempts of submitting the town to major powers, among others Venice and the Ottoman Empire, in order to eliminate its rise and limitate its financial power, a wise way of ruling and stipulating alliances with neighbouring authorities was the key for survival. Furthermore, Dubrovnik was able to obtain a wide range of autonomy (although it was officially ruled by a foreign authority, first Venetian, then Hungarian, later Ottoman and finally, Austrian), which lasted for centuries, until the capture by Napoleon in 1808<sup>8</sup> . Such conditions, in addition to the Ragusan skilled diplomatic abilities in maintaining the most politically neutral position possible created the perfect environment for the development of a prosperous and flourishing culture.

#### 3 A Multilinguistic Territory

Renaissance Dubrovnik is a characteristic example of how multiple languages can coexist along with a strong patriotic feeling. In this micro linguistic area a situation of multilingualism persisted for centuries. Even though this will not be the core and main topic of this article, the understanding of the diastratic dimention is key for the interpretation of language contact and change in this area. It is possible to detect several components (languages) whose use changed drastically over time by shifting from a trilingual society (made of Ragusan, Venetian and a local Slavic vernacular) to a multilingual one (made of Italian, Venetian, the Slavic vernacular and partially Ragusan). This shift can be confirmed by empirical evidence. I will now provide a brief insight into each of the mentioned languages.

Ragusan, sometimes considered as a dialect of Dalmatian and sometimes as one of the Dalmatian languages, represents the oldest variety of this very same language. Unfortunately, no written records of this language have been preserved. We are relatively certain about the fact that it was extinguished roughly around the end of 15th century, due to the predominance of both Venetian and the Slavic vernacular. It is important to mention that Ragusan had been the official language of Dubrovnik, even if for a very short period, in an attempt by the local authorities to revive the language. At first, we could consider it to be the only sociolinguistic substrate in Dubrovnik, whereas in the early Renaissance it backs up the Slavic vernacular as substrate.

A slightly different treatment was reserved to the Slavic vernacular, Italian and Venetian, which outlived the disappearance of the local autochthonous language

<sup>8</sup> Further information on the historical background of Dubrovnik can be found in Carter (1971), Krekic´ (1961; 1972; 1980; 1997) and more.

and coexisted in an uneven symbiosis. The Slavic vernacular, initially adstrate language along with Venetian, slowly took over as the new substrate, replacing Ragusan. The reasons for this change are still unknown. We know that Dubrovnik was the destination for numerous migratory communities, both Slavic and Romance, but there does not seem to be a linguistically-driven motivation.

Venetian was the lingua franca as language of trade. The local population did not have a high proficiency of the language. Prior to diastratic change, it was an adstrate language, along with the Slavic vernacular—later a superstrate. It is reasonable to assume that, due to tipological relation, Venetian could have replaced Ragusan (as it did with Dalmatian in the remaining part of Dalmatia). However, as mentioned above, Venetian was restricted to a commercial use in the context of trade and common inhabitants did not have a high proficiency, which can also be confirmed by empirical evidence.

Italian held the prestige of a literary language, used for the sake of literature, especially in prose works, whereas Slavic, on the other hand, appeared in poems<sup>9</sup> . Further information on diastratic change can be found in Zˇivojinovic´ (in preparation).

#### 4 State of Research

"We speak of language contact when two or more languages (or varieties of languages) interact with one another." This is how Campbell (2013: 298) started his chapter on language contact in a handbook, which is still considered as the reference point for historical linguistics studies. It is important to include in this definition because, as Thomason (cf. 2001) points out, the fact that two or more languages are used in the same place and at the same time. Considering both definitions, which are very general, I will attempt to describe certain points of Slavic-Romance contact in a literary context by presenting a synchronic analysis of comedies written by Marino Darsa (1508–1567), a Ragusan playwright and prose writer. However, Jakobson (1987: 16) notes that "any evolution possesses a systemic character and that any system is dynamic in nature". Therefore, the set of both diachronic and synchronic considerations is needed to provide a clearer picture of language contact. In order to strengthen my point, I will add a smaller diachronic component by incorporating the analysis of certain manuscripts, in particular the Vatican Croatian Prayer Book<sup>10</sup>, presented in Malic´ (2006, 2015).

<sup>9</sup> However, we do not lack poems written by Ragusan authors in Italian, such as Girolamo Papal, born in Split in 1460, Nicola Naljeskovic´ (ca. 1510–1587) and Nikola Gucˇetic´ (1549– 1610).

<sup>10</sup> Città di Vaticano, Biblioteca Apostolica, Barb.Lat.370.

On one hand, the texts I have examined are illustrative of the linguistic situation in Ragusa in the Renaissance period. These written records retrace the oral production and examplify diastratic and diaphasic variation of the questioned area. Nonetheless, Darsa resorted to linguistic stereotyping, the poetic function in this linguistic heterogeneity of the text serves to foreground the humor related to language interference and code-switching. However, certain works, such as Dundo Maroje<sup>11</sup> , reveal a realistic urban sociolinguistic situation. Therefore, it is reasonable to assume that this work faithfully represents the language spoken in Ragusa in this period. His knowledge of the Slavic vernacular, Italian and Venetian<sup>12</sup> was unquestionable, due to his multiple and extended stays, first in Siena and later in Venice, where he died. His works bear witness of extensive use of code switching, illustrating a high multilinguistic proficiency of his characters. For example, Italian propositions are used as larger units to replace the Slavic vernacular ones, but, in the same way, we also find commutations on a smaller scale. However, both single adjuncts and larger propositions are used only as unbound forms. In fact, infra-sentential code-switching (also called code-mixing<sup>13</sup>) is tendentially very rare and it can easily be argued that it might potentially be considered an example of Darsa's literary virtuosity or, more likely, a reflection of a multilingual society where code-switching is used to connote semantically and/or syntactically different concepts. In fact, we find the same issue, but with a slightly different outcome, in the following examples:


<sup>11</sup> First performed in 1551, the year of its composition remains unknown.

<sup>12</sup> Besides the Slavic vernacular and Italian, he also knew other languages: Latin, German and Turkish, which is evident from the use of certain specific elements in his works.

<sup>13</sup> The terminology of code-switching and code-mixing, as used here, belongs to Muysken (2000: 250–278), who assumes that code-mixing is a distinctive phenomenon occuring in "all cases when lexical items and grammatical features from languages appear in one sentence."

PIJERO: […] Stiamo a veder kao c´e ova rabota proc´. Let's to see how will this job go (pass by) 'Let's see how this deal goes'

Such examples allow us to understand the type of switching and mixing that this bilingual, or even trilingual community had chosen. The level of the linguistic competence is very proficient, though not a native level. This leads us towards drawing a possible hypothesis on the type of multilingualism characterized by asymmetry due the use of a specific type of commutation, which only involves larger chunks and avoides code-mixing at the morphological level.

On the other hand, the manuscript analyzed by Malic´ is the oldest Croatian vernacular prayer book and the finest example of the early shtokavian vernacular literary idiom. Therefore, such data present a precious set, helping us to explore diachrony. In particular, Dragica Malic´ (2006; 2015) provides a decomposition of lexicon of the two oldest Ragusan Latin<sup>14</sup> prayer books so far. Her work relies on many articles written by Resˇetar<sup>15</sup> who conducted an investigation on the Ragusan monuments. She focused on loanwords and calques, also called tudjinke<sup>16</sup> in Resˇetar (1952), and provided a brief insight in a basic structure and etymology of theword. What will result from a comparison of the two types of data will allow us to draw a conclusion with respect to the interaction between variation and contact in this specific area.

#### 4 Methodology and Parameters of Contact

The inspection of both the synchronic and diachronic corpuses has been conducted by applying the theories of transphonemization and transmorphemization on the level of morphological and lexical analysis.

#### 4.1 Transphonemization

First introduced in Filipovic´ (1978), the term of transphonemization has often been used by Yugoslav, or better, Serbian and Croatian scholars to designate the function of substitution on the phonological level. In particular, it has been used by Rudolf Filipovic´ in his studies on anglicisms in Serbo-Croatian. The concept

<sup>14</sup> By Latin here I mean the Latin (or Roman) alphabet, as opposed to the manuscripts written in Cyrillic.

<sup>15</sup> She specifically mentions Resˇetar (1907; 1933a; 1933b; 1933c; 1936; 1938; 1938b; 1952).

<sup>16</sup> In S ˇulek (1990), tudjinke simply stand for Fremdwort. There is no mention of any specific analysis in terms of types of borrowings.

refers to the process of adaptation of loanwords, or better of their switch into a replica. It splits into three distinguished types: 1) complete transphonemization, 2) partial or compromise transphonemization and 3) free transphonemization.

Firstly, in complete transphonemization, the description corresponds to phonemes in the borrowing language. For example, Engl. /dʒ/ vs. Croatian grapheme dzˇ, as in jeep /dʒi:p/ vs. dzˇip /dʒip/.

Secondly, partial or compromise transphonemization retraces the original phonemes that differ in part from the ones in the borrowing language. For example, Eng. /æ/ vs. Croatian e, as in jam /dʒæm/ vs. dzˇem /dʒɛm/. Filipovic´ claims that several consonants illustrate partial transphonemization very well, that is, the English phoneme transphonemizes by allowing a free change of place of articulation, but by maintaining the same manner.

Lastly, free transphonemization occurs very frequently and depends on the similarities and differences on the phonological systems of the borrowing language and the lending language. Namely, it occurs when the phonological form of the replica is formed according to the orthographic model of the donor language and not its pronunciation. For example, Engl. /ǝ/ allows several outcomes in Croatian. We find Engl. Yorkshire /'jɔ:kʃǝ/ vs. Croat. Jorkˇsir /'jorkʃir/, or Engl. pyjamas /pǝ'dʒa:mǝz/ vs. Croat. pidzˇama /pi'dʒama/.

#### 4.2 Transmorphemization

The process of transmorphemization occurs in contexts of morphological substitution. Filipovic´ (1979; 1980), who invented this term, uses it

"to cover changes occurring when a morpheme of the donor language, according to the basic principle of morphological adaptation begins with the formation of the citation form of the loan, and goes on in the creation of inflected forms, whatever the part of speech may be."

Similar to the above-mentioned process of transphonemization, this one as well splits into three different types: 1) zero transphonemization, 2) compromise transmorphemization and 3) complete transmorphemization. All three types are explained below, along with the original tables of examples from Filipovic´(1980).

Firstly, zero transmorphemization occurs when the borrowing language takes a free morpheme with a zero bound morpheme.


Table 1: Zero transmorphemization

Secondly, in a compromise transmorphemization a loan keeps a final bound morpheme of the donor language. This bound morpheme does not conform to the borrowing morphological system and therefore, maintains a compromise replica.


Table 2: Compromise transmorphemization

And lastly – as a complete transmorphemization, a bound morpheme is replaced by a borrowing language bound morpheme.


Table 3: Complete transmorphemization

# 5 Results of Contact

#### 5.1 Borrowings and Their Phonological Adaptation

Investigations on language contact usually prefer the lexical sphere, which is the most sensitive area regarding the influence of foreign elements. One of the two main outcomes of contact-induced change is borrowing, which is usually associated with situations of language maintenance and is defined as "the incorporation of foreign features into a group's native language by speakers of that language" (Thomason / Kaufman 1988: 37). There is a continuum in borrowing, from words that remain relatively foreign and unassimilated in pronunciation and spelling, through those that become more or less acclimatized to forms that have been assimilated so fully that their exotic origin is entirely obscured. There are numerous Italian lexemes in Darsa's comedies, but the largest part of this group underwent a long process of phonetic/phonological adaptation. In fact, it is very rare to find foreign, non-adapted words, because almost all the foreign lexemes are loanwords, integrated words, whose orthography was adapted if compared to the receiving language form. Tagliavini (1942: 379–381) synthetized the Slavic-Italian contact in the following way:

"Dovunque esistono contatti tra due popoli e due lingue si determinano influssi reciproci; nel caso dell'italiano e del croato<sup>17</sup> e dei loro rispettivi dialetti, gli influssi linguistici e le penetrazioni lessicali sono avvenute però quasi unicamente in un solo senso, e cioè dall'italiano sul croato, mentre i dialetti italiani sono rimasti immuni da influssi slavi […]. La differenza quantitativa e qualitativa tra i due flussi di scambi lessicali si deve al diverso prestigio delle due lingue e al fatto che mentre, sul litorale e nelle isole di Dalmazia, gli Slavi hanno generalmente conosciuto l'italiano, gli Italiani solo molto raramente hanno conosciuto e parlato il croato."

Such a synthesis partially confirms my previous hypothesis on the distinction of substrate/ superstrate that attributes a major percentage of distribution to the local Slavic vernacular (which is assumed to be the substrate language) and a minor one to Italian (superstrate). Moving back to loanwords and their orthographic adaptation, Zˇupanovic´ (2008) noticed that there are (more or less) ten different innovations that need to be spotted. They are: transphonemization zero, degemination, the change /o/ > /u/, /e/ > /i/, /i/ > /e/, /ie/ > /i/, the insertion of the grapheme j inside /ia/ and /io/, an exchange between fricatives and affricates, an exchange of palatals and an exchange of sibilants. I will now present a few examples of phonological adaptation from either Venetian and/or Italian into Slavic, some of which have been detected by Zˇupanovic´ (2008). At first, transphonemization occurs in type zero, free morpheme + zero bound morpheme, as in the following examples:


<sup>17</sup> Many scholars have addressed the Slavic component in Ragusa as to Croatian, to reconnect to today's political-linguistic distinction.

Secondly, degemination is a very common feature that we often find in the passage of lexemes from Italian to Neo-Shtokavian, which is also typical of Venetian.


Probably the most common vowel change is /o/ > /u/ that we find in numerous examples in, as Zˇupanovic´ (2008) specifies, both stressed and unstressed syllables. Again, this feature was very common in Venetian as well.


Italian diphthongs are subjected to a change that splits into two different directions. On one hand, they are simplified into one single vowel, so that we obtain /ie/ > /i/ as in the following examples:


The diphthong /ua/ instead, undergoes the transformation into /va/, as in:


On the other hand, certain /ia/ and /io/ diphthongs are separated by the intervocalic element /j/.


Several consonantal changes are not to be forgotten. Mostly fricatives and affricates have undergone this change.


Dealing with consonants, we also find the velar sibilant /s/ transforming into a post-velar sibilant /ʃ/.

<sup>18</sup> An example of complete transmorphemization.


Some other transformations are worth being mentioned, even though the following examples do not occur as frequently as the previous ones:


#### 5.2 Morphological Change in Synchronic Data

All the three processes of transmorphemization can be detected in the morphological change that occurred as a consequence to the Slavic-Italian contact. It mostly involved the categories of nouns, verbs and adjectives. Starting from zero transmorphemization, we assist in a process of morphological adaptation through the loss of the final bound morpheme. Such characteristics are common in lexemes containing the suffixes -o and -e, usually marking masculine singular lexemes, as in the following examples, all taken from Dundo Maroje, involving the categories of nous and adjectives:


The process of compromise transmorphemization is the one that probably occurred more often in the Italian-Ragusan Slavic vernacular change. As it is possible to notice in the following examples, Italian bound morpheme is maintained in the final Slavic outcome. Due to the discrepancy between the two languages, only degemination materialized, as in the following nouns:

vedoela /vedo'ela/ veduvela /vedu'vɛla/ vedovella /vedo'vɛlla/ osto /'ɔtse/ oˇste /'ɔʃte/ oste /'ɔste/ alegrezza /ale'gressa/ alegreca /ale'gretsa/ allegrezza /al'legretstsa/

mascarada /maska'rada/ maˇskerata /maʃke'rata/ mascherata /maske'rata/

During morphological adaptation of both nominal and adjectival categories, it is relatively common to find a graphemic insertion -a-, a feature standing between a phonological and a morphological use, applied in order to break longer chains of consonants that usually do not occur in the standard Shtokavian form. For example:


Concerning the adaptation of Italian suffixes, we have to mention certain common nouns, usually ending with -tur and deriving from the Italian -tore or -ijer, coming from -iere. Here are listed some examples found in Dundo Maroje:


Along with compromise transmorphemization, we also find a conspicuous amount of complete transmorphemization in verbal borrowings, where the standard Italian suffix is substituted with its Slavic equivalents -iti and -ati<sup>19</sup> . What is fundamental to mention is that in this category we find integrated loanwords, that resulted from the application of the morpheme marking the infinitive. Also important to mention is that we can notice a phonetic and graphic adaptation of such loanwords, whose meaning does not change, except for fewer exceptions that will be seen below:


<sup>19</sup> We find a very similar pattern in Zˇupanovic´ (2008). However, she spotted a conversion into verbal suffixes -at and -it, whereas in Drzˇic´, we find the integral version that is also used nowadays.

<sup>20</sup> The additional morpheme -va- is also called imperfective morpheme, whose function is only aspectual and serves to derive imperfective verbs from the perfective ones.

#### 5.2 Evidence from the Diachronic Corpus

Dragica Malic´ has conducted an extensive research on a conspicuous number of manuscripts in the domain of Slavic lexicology and lexicography. Of great importance are her above-mentioned articles on Ragusan loanwords that resulted out of her detailed examination of the Croatian Vatican Prayer Book<sup>21</sup> (V<sup>1</sup> from this point further) and Ragusan Academic Prayer Book<sup>22</sup> (A from this point further). These two manuscripts represent the oldest written Ragusan prayer books and here I only consider the non-Slavic corpus analyzed by Malic´. Despite lexical adaptations, it is possible to detect all three types of transmorphemization in her examples.


Such examples fit perfectly within the framework developed by Rudolf Filipovic´. Furthermore, there is an evident similarity between data pulled out from both the synchronic and diachronic corpuses, which seem to be using the very same pattern of loanword adaptation. Because of such congruency, it is likely that synchronic data find confirmation in diachrony, and therefore, the hypothesis of contacted-induced change is confirmed.

# 6 Conclusion and Research Desiderata

Ragusa is undoubtedly a cultural centre with a very interesting historical and linguistic background, characterized by multidimensional complexity. Its sociocultural status in the period of Renaissance reflected the sociolinguistic situation, which is well illustrated in the Darsa's work. The combination of data pulled out from some of the most important literary works in the Slavic Adriatic area and some of the oldest written records in Ragusa provides an intriguing

<sup>21</sup> Dating back to year 1400 ca.

<sup>22</sup> Mid-15th century manuscript.

insight into Renaissance language use. The work by Rudolf Filipovic´ acts as a filter to the analysis of adaptation of loanwords and as a unifier of the two corpuses. The result of this binomial is a confirmation of language change due to contact. However, in the perspective of a future work, it would be necessary to broaden the corpus of diachronic data, by analyzing Cyrillic manuscripts as well.

#### Bibliography


Jakobson, Roman. 1987. Language in literature. Cambridge / Massachusetts: Harvard University Press.

Krekic´, Barisˇa. 1961. Dubrovnik (Raguse) et le Levant au moyen âge. The Hague: Mouton.


Resˇetar, Milan. 1933c. Jezik Marina Drzˇic´a. In: Rad JAZU 248, 99–240.


Jelena Zˇivojinovi c´, BA MA University of Verona Department of Foreign Languages and Literatures Lungadige Porta Vittoria 41 37129 Verona Italy jelena.zivojinovic@univr.it

UiT The Arctic University of Norway Department of Language and Culture (ISK) / CASTL Teorifagbygget 4.367, UiT Campus Tromsø 9019 Tromsø Norway jelena.zivojinovic@uit.no

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# II Varieties in Contact

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

Varieties of German in Austria

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Johanna Fanta-Jende

# Varieties in Contact. Horizontal and Vertical Dimensions of Phonological Variation in Austria

Abstract: The main objective of this paper is to investigate the 'horizontal' (geographic) as well as the 'vertical' (social) dimension of synchronic variation for a selected phonological variable (Middle High German /ei/). The empirical input consists of spoken language data from 13 villages representing Austria's major dialect regions (i. e. Central Bavarian, South Bavarian, South-Central Bavarian and Alemannic). The data point to different processes of language change currently going on in Austria that affect the entire dialect–standard language spectrum. Regarding the dialectal parts of the spectrum, regional dialect variants are shown to be encroaching on the older base dialectal variants in the eastern parts of the country. Moreover, also with respect to the standard-near pole of the axis, a process of 'Viennese Monophthongisation' is observable that favours urban variants from Austria's capital Vienna over standard German variants.

Keywords: German in Austria, dialect–standard axis, language change, speech repertoires, horizontal and vertical dimension, phonological variation

Abstract: Ziel dieses Beitrags ist die Untersuchung der 'horizontalen' (arealen) sowie 'vertikalen' (sozialen) synchronen Variationsdimension eines ausgewählten phonologischen Merkmals (mittelhochdeutsch /ei/). Die empirische Grundlage liefern multivariat angelegte Datenerhebungen an 13 ländlichen Ortspunkten Österreichs, die vier oberdeutsche Sprachräume repräsentieren (mittelbairisch, südbairisch, südmittelbairisch und alemannisch). Die Daten deuten auf divergierende Sprachwandelprozesse hin, welche die gesamte Dialekt-Standard-Achse im gegenwärtigen Österreich betreffen. Hinsichtlich des dialektalen Bereichs des Spektrums kann der Rückgang von ursprünglich weitverbreiteten basisdialektalen Varianten zugunsten neuerer Formen konstatiert werden. Darüber hinaus sind – von Österreichs Hauptstadt ausgehend – auch Wandelprozesse in Hinblick auf den standardnäheren Bereich des Spektrums feststellbar, die sich als sukzessive Ausbreitung der "Wiener Monophthongierung" anstelle der standardsprachlichen Variante manifestieren.

Keywords: Deutsch in Österreich, Dialekt-Standard-Achse, Sprachwandel, Varietätenrepertoires, horizontale und vertikale Dimension, phonologische Variation

#### 1 Introduction

The present paper is set in the current tradition of empirical sociolinguistic studies that strive to capture, describe and explain synchronic variation in German language use by plotting it in a two-dimensional model that features an 'areal-horizontal', regional-dialectological axis and a 'social-vertical' dialect– standard axis (see Auer 2005; Lenz 2010); the latter dealing particularly with the individual linguistic repertoires of speakers. In recent years, there have been several thorough empirical studies focusing on the 'vertical architecture' of variation in specific regions in Germany (cf. Kehrein 2012; Lanwer 2015; Lenz 2003). However, apart from few empirical investigations on specific locations (cf. Scheuringer 1990; Scheutz 1985; Unger 2014), similar endeavours still represent a major research desideratum in the Austrian context. Furthermore, most previous research in Austria has focused primarily on only one 'pole' of the dialect– standard axis at a time;<sup>1</sup> and, thus far, little attention has been paid to speakers' overall language spectra and speech repertoires.

The present paper addresses this apparent gap, setting out to shed light on the complex dynamics of the current situation regarding the use of the German language in Austria. In this context, what is commonly referred to as the phonological variable of Middle High German /ei/<sup>2</sup> seems to be arguably a worthwhile object of study for the purposes of the present undertaking, since it exhibits variation on both the vertical as well as the horizontal dimensions. Thus, in addition to a regional patterning of variants, it is presumed to currently undergo a language change as a consequence of varieties in contact. In fact, this does not

<sup>1</sup> See especially Moosmüller (1991) and Moosmüller et al. (2015), who describe Standard Austrian German, and the GOBA-database (Georeferenzierte Online-Bibliographie Areallinguistik) for a selection of local dialect monographies <https://regionalsprache.de/GOBA/Catalo gue.aspx> (June 3, 2020).

<sup>2</sup> It has become customary in German dialectology to analyse phonological matters relating them to an (idealised) historic sound framework, i. e. in the case of vowels usually the Middle High German vowel system. By doing so, variants of different current dialects can be compared based on their (assumingly) historic common root. The spoken standard language would not be suitable as a frame of reference, since it constitutes the spoken variety of a semi-artificial written language, which developed comparably late (between the 15th and the 18th century) in the German speaking countries as some sort of koiné on the basis of different regional varieties (see Löffler 2003: 65–68 and Wiesinger 1983: 813 for the discussion of the reference system).

only apply to the 'lower' (rather dialectal) parts of the dialect–standard axis, but also to the 'upper' (standard-near) parts of the spectrum.<sup>3</sup>

On the 'lower' end of the axis, for example, where in Bavarian dialects originally mostly /oa/ diphthongs were attested in usage, these days, the spread of /aː/ monophthongs can be observed, e. g. [haːs] instead of [hɔɐ̯s] heiß 'hot' (see Section 3.1). Regarding the standard pole, such change involves notably the socalled 'Viennese Monophthongisation', which describes a shift from standard /aɛ̯/ via intermediate /æe̯/ to monophthongised /æː/, indicating for the previous example heiß 'hot' the monophthongised variant [hæːs] instead of [haɛ̯s] (see Section 3.2).<sup>4</sup> Both change processes pattern variants in a regionally differentiated manner, rendering an analysis across the various Austrian dialect regions particularly fruitful (see Section 5). Furthermore, in Scheutz's (1985) study of the variational spectrum in the Central Bavarian town of Ulrichsberg, situational context as well as degree of formality seemed to play a major role in the production of different variants of MHG /ei/ in individual speakers' utterances, suggesting a social-vertical analysis to be potentially highly informative.

Against this background, by using a multimethodological approach combining more controlled as well as 'free' conversational settings, the following central research questions can be derived:


<sup>3</sup> Note that my usage here and in the remainder of the paper follows common convention in the literature, which metaphorically places dialect at the 'lower' pole and standard at the 'upper' pole of the social-vertical axis of variation. The underlying principle is that native speakers of German in Austria usually make use of a whole range of linguistic repertoires (see also Section 2.1). The terms'lower' and 'upper' parts of the dialect–standard axis are not to be mistaken for categories of societal stratification or as an evaluation of prestige.

<sup>4</sup> Note that, contrary to the commonly used rendition of the diphthong in Standard German (e. g. /aɪ̯/, /ai̯/ or /ae̯/), Moosmüller et al. (cf. 2015: 344) propose that /aɛ̯/ is best used to describe the sound for the Austrian Standard Language. Following their notation convention, in the present paper /aɛ̯/ is used for the standard form, /æe̯/ is used for the semi-monophthongised form (as in their example of Seide 'silk', see Moosmüller 2015: 344) and /æː/ represents the fully-monophthongised variant (same as /ɛː/ in other studies, e. g. Scheutz 1985).

The outline of the paper is as follows: Section 2 provides a concise overview of the sociolinguistic situation in Austria, as well as the general, basic structure of its dialectal variation. Section 3 then zooms in on the particular phonological characteristics, historical particularities and hitherto assumed geographical distribution of the variable selected for analysis here, MHG /ei/. In this section, also previous relevant studies will be reviewed. The empirical study is introduced in Section 4, where the methodological approach to data selection and processing are presented. Section 5 then reports the results of two rounds of data analysis, with 5.1. presenting a study that focuses on the horizontal dimension, 5.2 offering a study focusing on the vertical dimension of variation in the use of MHG /ei/ and 5.3 discussing invariant lexical particularities. The paper concludes with Section 6, in which findings are summarised and discussed and an outlook is provided regarding potential future developments and change in the use of MHG /ei/ in Austria.

# 2 A Brief Overview of the Sociolinguistic Situation Regarding German in Austria

#### 2.1 German in Austria on a Macro-sociolinguistic Level

From a macro-sociolinguistic perspective, the Austrian language situation regarding the use of German can be characterised as one of so-called 'inner multilingualism' ('innere Mehrsprachigkeit'; see Wandruszka 1979), allowing speakers to switch and shift effortlessly up and down a 'vertical' dialect–standard axis (cf. Ammon et al. 2016: XLVII; Auer 2005; Lenz 2019). Accordingly, a quite complex constellation of varieties in contact has to be assumed, in which any non-disjunctively conceptualised varietal speech layer ('Sprechlagen', see Lenz 2010; Schmidt / Herrgen 2011) hypothetically interacts with at least one other variety of the spectrum. This includes interplay on the vertical level between a dialect and the higher parts of the spectrum, but also on the horizontal dimension in two ways: either between one standard variety and a (competing) other standard variety (e. g. Standard German German (SGG) and Standard Austrian German (SAG), see Moosmüller et al. 2015) or between dialects originating in different regions.<sup>5</sup>

Due to these highly dynamic contact constellations, different forms of language evolution are observable in the overall system, where strikingly stable

<sup>5</sup> Of course, in a discussion in greater depth of such theoretical assumptions, actual Kookkurrenzregeln ('rules of co-occurrence'; translation: JFJ) within varieties also have to be considered (cf. Scheutz 1999).

variants of base dialects exist alongside forms highly reactive towards influence of the standard, as well as new "dialectal phenomena […] diverging from both the standard and the dialect" (Auer / Schmidt 2010: 207). One possible result of dialects in contact is 'dialect mixing' or "the partial merging of lexicons and grammars of different but related dialects" (Hinskens et al. 2005: 9)—which applies, of course, also to phonological phenomena. According to Trudgill (1986) frequent dialect mixing typically leads to either

"(a) quantitative variation between the 'old' and the 'new' variants; or (b) the occurrence of the 'new' variant, in some words, but not in others, […]; or (c) intermediate, phonetically approximate, forms, the resulting variety being a fudged dialect" (Trudgill 1986: 62–78, as cited in Hinskens et al. 2005: 45–46; emphasis in original: JFJ).

Trudgill developed these considerations in the context of discussing the language situation of new colonies, where contact between settlers of different origins and diverse linguistic backgrounds lead to the formation of a new dialect. Yet, arguably, his theoretical assumptions also apply in the complex Austrian scenario of language contact. Indeed, how this dynamic plays out in the case of one particular phonological variable, MHG /ei/, is the subject of the present paper.

Sociolinguistic research routinely points to the influence of the situational context for an individual's selection of a variety or set of features (beginning with Labov's work in the 1960s—see e. g. Labov 2006); and, naturally, this also applies in the Austrian situation (see e. g. Soukup 2009). In other words,

"which of the various variants within the spectrum gets to be chosen by the language member depends on pragmatic factors, e. g. on the communicative partner, the setting, the topic, the intended effect on the addressee as well as which variants are available to the speaker" (König 2011: 135; translation: JFJ).<sup>6</sup>

Therefore, in any analysis of the system of variation in a certain sociolinguistic context, not only the quantities of variants produced are of significance, but also the question of when they are used, with whom and to what degree. Certain situational settings may yield different forms or differing amounts of comparable existing forms. For dialect–standard spectra in particular, it is to be assumed that "[w]ith an increasing public orientation and formality of the situation, [also] differences in the respective language varieties become apparent" (Mattheier 1980: 96)—implying, typically, a shift towards the standard pole.

As an additional factor dynamically affecting the Austrian sociolinguistic context, trends of regional diffusion of features also have been identified, in

<sup>6</sup> "Welche der vielen Varianten innerhalb des Spektrums von den Sprachteilhabern jeweils ausgewählt wird, hängt von pragmatischen Faktoren ab, z. B. von den Kommunikationspartnern, von der Sprechsituation, vom Thema, von der beabsichtigten Wirkung auf die Ansprechpartner, aber auch von der Tatsache, welche Varianten dem Sprecher überhaupt zur Verfügung stehen."

which Austria's capital Vienna seems to constitute a key centre of influence on, and of the origination and spread of, language change processes (cf. e. g. Auer 2005: 20–21; Moosmüller 1991). As Wiesinger puts it,

"Much of the force of innovation in Central Bavarian has since the latter part of the twelfth century come from Vienna […]. Right up to the present, Vienna has remained the 'heart of Austrian dialects'(Kranzmayer) by being the linguistic model which shapes and influences the east and south of the country with Lower and Upper Austria, Burgenland, Styria and Carinthia" (Wiesinger 1990: 464–465, 456).

Yet, the particular trajectories of diffusion of individual forms also seem to reflect the kinds of societal norms associated with certain linguistic innovations. What Hinskens et al. (2005: 8–9) refer to as the 'urban hierarchy' describes innovations "jump[ing] from large, influential cities to smaller, less influential ones" (in Austria, for example from Vienna to provincial capitals and so on), a trajectory representing the application of 'norms' and forces external to the destination area. By contrast, a contra-hierarchical diffusion trajectory of a feature is also possible, rather indicating a "revitalisation of traditional norms" indigenous to an area (Hinskens et al. 2005: 8–9). Also, covert conceptions regarding what constitutes prestigious or stigmatized forms and varieties play a significant role in linguistic developments, determining if one particular variant will stay constant and the same, experience change or 'fudging', rise to a new function, or will be entirely replaced by another (cf. e. g. Koppensteiner / Lenz 2017; Lenz 2010).

At least all these listed dynamics and factors are at play in the Austrian sociolinguistic context in general, and in the phonological system of German in Austria, which is the focus of this paper, in particular, creating a scenario of high intricacy and complexity that is difficult to disentangle. Indeed, in part due to this complexity, the standard pole of the vertical axis especially is probably the lesser described and theorized one for Austrian German language use, while the dialect pole has been subject to a rather long-standing tradition of dialectological inquiry. The next section (2.2) thus recaps specifically whatwe know about dialectal variation in German in Austria.

#### 2.2 German in Austria: An Overview of the Main Regional Dialects

Contrary to the notoriously complex and diverse social-vertical language usage situation which is only beginning to be elucidated in detail, Austria's base dialects can be fairly straightforwardly described on the regional-horizontal dimension (cf. Lenz 2019; Wiesinger 1983 and 1990). Thus, the dialects spoken in Austria can be assigned either to the Bavarian ('Bairisch') or to the Alemannic ('Alemannisch') branch of the Upper German ('Oberdeutsch') dialect family.<sup>7</sup> While Bavarian covers the vast majority of the country, the Alemannic dialects (with respective subdivisions on a small scale) are restricted to the western parts, mainly to the political state of Vorarlberg (see Figure 1). 'Traditional' dialectology has primarily focused on the diverging phonological systems between Alemannic and Bavarian, only recent studies focus particularly on the (morpho-) syntactic variation between the Bavarian and Alemannic parts of Austria (see Breuer/ Wittibschlager 2020; Fingerhuth / Lenz in press; Korecky-Kröll accepted; Lenz 2019: 331–338; Lenz et al. 2019). In fact, this divergence often leads to difficulties of mutual comprehensibility among Austrian speakers from the east and west.<sup>8</sup> In particular, the respective realisations of vowels developed differently out of Middle High German ('Mittelhochdeutsch')—as exemplified in Section 3 for MHG /ei/—set Alemannic and Bavarian apart (cf. Wiesinger 1990: 456). Yet, also an Alemannic–Bavarian transition area exists, where features from both dialect regions occur.

Bavarian in Austria consists of two subcategories, namely Central Bavarian ('Mittelbairisch'; primarily in the states of Upper Austria, Lower Austria and Vienna) and South Bavarian ('Südbairisch'; mainly in the states of Tyrol, Carinthia and parts of Styria). There is also a substantial region of gradual changeover, the South-Central Bavarian transition area ('Süd-Mittelbairisches Übergangsgebiet'; particularly in most parts of Salzburg, Styria and Burgenland, and partially in Tyrol) (see Figure 1). One difference between South Bavarian and Central Bavarian lies for instance in the local upkeep of 'centring diphthongization' ('fallende Diphthongierung') of MHG /ê/, /œ/ and /ô/ in South Bavarian, as in [ʃnɛɐ̯], the equivalent to New High German (NHG, 'Neuhochdeutsch') Schnee 'Snow', [b ˚ ɛɐ̯s] for NHG böse 'bad' and [ʁɔɐt] for NHG rot 'red' (cf. Bülow et al. 2019; Lenz 2019: 238; Wiesinger 1990: 457; Zehetner 1985: 62). In general, South Bavarian—predominantly in the mountainous south and west of the country—is frequently regarded to be linguistically more conservative than Central Bavarian and "thus a relic area of linguistic features which formerly occurred in all Bavarian dialects" (Wiesinger 1990: 456). Central Bavarian on the other hand, has the status of being the most open and permeable towards innovations of all Bavarian dialects (cf. Wiesinger 2003: 208), rather readily adopting 'newer' processes, e. g. the vocalization of /l/, e. g. in [fyː] NHG viel 'a lot', or [sɔe̯d ˚ ͡s] NHG Salz 'salt' (cf. Kranzmayer, 1956: 29; Lenz 2019: 330; Voll-

<sup>7</sup> As the name suggests, Bavarian is not only used in Austria, but also in the southeast of Germany (predominantly in Bavaria). Alemannic, in turn, is also spoken in the neighbouring countries of Liechtenstein, Switzerland and the southwest of Germany.

<sup>8</sup> Regarding comprehensibility, note that difficulties often seem to be unidirectional, as Bavarian varieties are more present in media and TV, and thus typically more familiar to Alemannic speakers than vice–versa.

mann et al. 2017). These are, of course, quite general statements about Austria's dialect regions, neglecting the abundance of a wide range of differing features and current general processes of dialect transformation and language change.

Figure 1: Alemannic and Bavarian dialect regions in Austria (CC-BY-SA Melanie Seltmann dioe.at | regionalsprache.de)

On the background of this general overview of dialect areas in Austria, it is now opportune to discuss the status and development of the variable to be analysed here, MHG /ei/, in greater detail, which Section 3 now sets out to do. Within this Section, 3.1 and 3.2 recap the available information on MHG /ei/ in the Bavarianspeaking areas, while Section 3.3 separately deals with the situation in the Alemannic area.

# 3 MHG /ei/ in Austria: The Current State of Research

As a background to the subsequent empirical analysis of MHG /ei/, the following sections now detail the phonological characteristics, historical developments and known geographical distributions of variant realizations of the variable first for the Bavarian and then for the Alemannic spectrum of Austrian language use.

#### 3.1 MHG /ei/ in the Bavarian Dialects: The 'Lower' Part of the Dialect– Standard Axis

In the case of the phonological variable MHG /ei/, the widespread dialectal variant /oa/ is found in almost the entire area of Bavarian in Austria, constituting some sort of 'central feature' for the Bavarian context ("bairisches Leitmerkmal", Scheutz 1999: 118; see also Lenz 2019: 328–329). Its written form has been attested in documentary evidence since 1220, as an alteration from /ai/ via /oi/ to /oa/ (cf. Kranzmayer 1956: 59). Hence, in the Bavarian area, for example NHG heim 'home', weiß 'know' (first person singular) and klein 'small' are expected to be produced as [hɔɐ̯m], [vɔɐ̯s] and [ɡ˚lɔɐ̯] in their respective 'default realisations' ("Normalrealisierungen", Scheuringer 1990: 235).<sup>9</sup> However, since 1350, a small number of words seem to have started following the standard pronunciation of the diphthong (cf. Kranzmayer 1956: 63), marking explicit lexical exceptions to the rule. Thus, Wiesinger (1990) reports that the "theological terms Fleisch, heilig, Geist, geistlich, rein 'flesh', 'holy', 'spirit', 'spiritual', 'pure' and legal words such as Eid, Meineid, beleidigen, eigen, Kaiser 'oath', 'perjury', 'insult', 'own', 'emperor', […] have the diphthong [aɪ]" (1990: 450–451; emphasis in original: JFJ). Similar developments apply also to the word Mai '(the month of) May' (yet as a compound as in Maibaum a diphthong pronunciation might still occur; see Patocka 2010), to technical words, such as Heizung 'heating'(although when used as a verb, as [hɔɐ̯tsn̩ ] heizen 'to heat', an /oa/ diphthong may indeed be used; see Zehetner 1985: 81), or words of local administration as soon as they are used in a (semi-)official function, e. g. Gemeinde (instead of [ɡ˚mɔɐ̯] 'municipality') or [bʏɐ̯ɡ˚ɐmaɛ̯sd ˚ ɐ] 'mayor' (as opposed to dialectal [bʊɐ̯ɡ˚ɐmɔɐ̯sd ˚ ɐ]; see Zehetner 1985: 82–83).<sup>10</sup>

Outside of such influences of the standard on the pronunciation of MHG /ei/ in certain lexical items, phonological and sociolinguistic reasons have, especially in Vienna and the southern parts of Carinthia, led to the use of an /aː/ monophthong instead of the dialectal /oa/ diphthong, resulting in forms such as [haːm], [ɡ˚laːn] and [vaːs] instead of the above-mentioned 'defaults' of [hɔɐ̯m], [vɔɐ̯s] and [ɡ˚lɔɐ̯n] (cf. Wiesinger 1990: 465–477).

<sup>9</sup> Note an exception to this default that has to do with the above-mentioned process of /l/-vocalization, so that, in stressed position before /l/, there is likely to be a shift towards rounded pronunciation, e. g. in NHG Teil being realized as [d œ˜ ɛ̯ː].

<sup>˚</sup> 10 As Bülow et al. (2019) argue variationist linguistics has often neglected to pay attention to lexical particularities and the general influence of lexical diffusion on language change. The present paper's emphasis is on the description of a rather macro-structural model of the vertical and horizontal dimensions in Austria and its methodological implications; consequently, an in-depth analysis of lexical diffusion is outside of my scope. I will provide, though, a list and discussion of lexical particularities assumedly resistant to variation in Section 5.3.

Concerning Vienna, several hypotheses have been proposed for explanation. One simplified theory purports that a similar but more closed /æː/ monophthong could have been 'imported' by Swiss nobility who relocated to Vienna from 1272 onwards with Rudolf I and his son Albrecht I of the house of Habsburg (cf. Kranzmayer 1956: 60). This is, at least, also the time when the first written documents of the monophthong appeared in Vienna (see Kranzmayer 1956: 60). The sound was then supposedly modified by the Viennese aristocracy to /aː/ and passed on as a prestigious innovation to what Wiesinger calls 'Herrensprache', "the upper-society language of the nobility and higher bourgeoisie" (Wiesinger 1990: 465). Regarding Carinthia, one explanation may lie in the fact that, being part of the South Bavarian dialect region, a phonological conflict between /oa/ for MHG /ei/ and the previously mentioned centring diphthongization of MHG /ô/ to [ɔɐ̯] would have arisen (see also Section 2.2), potentially resulting in homonyms like [ɾɔɐ̯sn̩ ] reisen 'to travel' and [ɾɔɐ̯sn̩ ] Rose 'rose'. <sup>11</sup> Therefore, already around 1300, an adoption and 'Bavarianization' ("Verbaierung") of presumably Upper Franconian /a/ and /ä/ monophthongs may have occurred (cf. Kranzmayer 1956: 60–61). As Kranzmayer (cf. 1956: 61) argues, this may have been caused and influenced by the historic landownership of the Prince-Bishopric of Bamberg—a region of consistent /a/ monophthongs—whose higher officials had their permanent residency in Carinthia (especially in the towns of Villach and Feldkirchen). Wiesinger rather supports the idea of an indirect change due to a different kind of language contact, namely between German and Slovene: "As early as the Middle Ages, bilingualism in this region led to the formation of a colloquial language based on the articulatory habits of the Herrensprache" (Wiesinger 1990: 477; emphasis in original: JFJ).<sup>12</sup> He does not clarify how exactly this process took place but gives some information about the regional diffusion of the phenomenon: "This [= the /aː/ monophthong; JFJ] has reached as far forward as the Drau, Lesach and lower Möll valleys of upper Carinthia" (Wiesinger 1990: 477).

Resolving such origin issues is beyond the scope of the present paper. But at any rate, on the background of the accumulated information, it is possible to map the distribution of the traditional dialectal variants of MHG /ei/, including the historical areas of the /aː/ monophthong, as shown in Figure 2. This map is based on the Wiesinger Ergänzungskarte (WEK, Wiesinger 1962–1969), depicting all

<sup>11</sup> No information is provided, though, how this 'conflict' has been tackled in the other parts of the South Bavarian dialect area (e. g. Tyrol), given that it appeared to be a conflict at all.

<sup>12</sup> What Wiesinger refers to here is bilingualism and concomitant comprehensibility problems involving German and Slovene. The Carinthian Slovenes are an accepted autochthonous minority in Carinthia. Their ancestry goes back to the 7th and 8th century, eventually leading to centuries of bilingualism and language contact between Slovene and German in the area (cf. Pohl 2005).

variants of the lexeme heim 'home' as they occur in the dataset collected in Austria in the late 1920s.<sup>13</sup>

Figure 2: Historical (1920s) distribution of variants of MHG /ei/ in Austria (without the state of Vorarlberg), based on the WEK for the lexeme heim 'home' and also showing Austria's major dialect regions (in colour shading, see also Fig. 1)

Yet, as already intimated, on a more detailed level, local studies have led to new assumptions about an ongoing language change concerning the areal-horizontal as well as the social-vertical dimension of the Austrian dialect–standard spectrum.

For example, in 1910, Bíró, describing the "heanzische Mundart" in the village of Neckenmarkt in South-Central Bavarian Burgenland, only mentions the use of the overall Bavarian diphthong /oa/ as a realization of MHG /ei/. There are still exceptions, which, however, are restricted to the kinds of lexemes already mentioned above (e. g. Geist, Kaiser, Fleisch, Mai), plus a few minor supplements: (1) beleidigen 'to insult' and leisten 'to achieve' both follow the articulation of NHG /aɛ̯/, (the former in contrast to its root word leid, typically realized as [lɔɐ̯t]); and (2) heilig 'holy', pronounced [haɫi] 'with an /a/ monophthong. The latter, however, may not constitute an expected adoption of the Viennese form, but rather a phonological adjustment before velarized /ɫ/, which, as Bíró claims, 'doesn't support a preceding i' ("das ɫ duldet bekanntermaßen kein i vor sich", Bíró 1910: 69; translation: JFJ). However, in the same year, Pfalz (1910) identifies

<sup>13</sup> As Figure 2 and the "SAO" (= Sprachatlas von Oberösterreich, see SAO Volume 2, map 65 and 66) exemplify, certainly, a small number of additional variants for MHG /ei/ can be identified in Austria (e. g. /ua/ or /oi/ diphthongs). Since they are usually restricted to a very small spatial scale, which does not coincide with locations of interest selected for the present empirical investigation, they cannot be considered in this paper.

the beginning of a transformation process in the small town of Deutsch-Wagram, located at the north-western city borders of Vienna. Here, not only 'intentionally used minor variants' ("absichtlich gebrauchte nebenformen [sic]") already feature the /aː/ monophthong, but also the entire social group of local 'school youth' ("Schuljugend") has started adopting this new, prestigious variant from Vienna (cf. Pfalz 1910: IX; translation: JFJ). Unger (2014) shows this process to have been completed when re-investigating the dialect of Deutsch-Wagram again some 100 years later. Thus, today, the use of /aː/ has extended to all linguistic constellations and sociolinguistic groups, resulting in the full abandonment of the diphthong /oa/ except for specific agricultural terms (cf. Unger 2014: 53–54).

Scheutz's (cf. 1985: 242–243) findings from the 1980s add insights in vivo about the dynamics of this change from /oa/ to /a:/ by analysing conversational data produced under different conditions of formality, from the Central Bavarian town of Ulrichsberg in Upper Austria, around 190 km (beeline) west of Vienna. His results demonstrate that speakers do not necessarily just switch between dialectal /oa/ and standard /aɛ̯/ in different situations but may also access the 'Viennese variant' /a:/. Since the /a:/ variants are preferably used in formal settings, this leads to the assumption that this monophthong is perceived as less dialectal, eventually leading to a tripartite paradigm with the monophthong as the intermediate vernacular form between base dialectal /oa/ and standard /aɛ̯/ (cf. Scheutz 1985: 243). Additionally, the results indicate that the preference for /aː/ realisation are connected to certain specific lexemes, e. g. numerals such as ein(s), zwei 'one, two', or the highly frequent conversational discourse marker (ich) mein(e) '(I) mean' (cf. Scheutz 1999: 118).

Scheutz (cf. 1999: 117) concludes that even though the /a:/ monophthong is clearly associated with Vienna and nowadays frequently stigmatised, the expansion of this variant across the whole of Austria and especially to the cities is in full swing. Also, based on the studies reviewed here, we can assume that irrespective of the exact details of its origin, the use of the /aː/ monophthong has been gradually expanding across the Bavarian part of Austria for about a century now, at the same time slowly replacing the formerly widespread /oa/ diphthong (cf. Kranzmayer 1956; Lenz 2019: 329). The empirical studies reported in the present paper (see Section 5) are set to shed further light onto areal developments such as the one just described.

But not only the 'lower' parts of the dialect–standard axis are affected by current processes of language change in realizations of MHG /ei/; also the 'upper sectors of the spectrum seem to be undergoing a phonological transformation, albeit a more gradual and probably less salient one, which is described in the next section.

#### 3.2 MHG /ei/ in the Bavarian Dialects: The 'Upper' Part of the Dialect– Standard Axis

As mentioned earlier, the Bavarian dialectal realisation of MHG /ei/ is either the former diphthong /oa/ or the newly spreading /aː/ monophthong. The corresponding standard variant is NHG /aɛ̯/. However, /aɛ̯/ occurs in a much wider range of lexemes than /oa/ and /aː/. This is because the dialect preserves the option of separating sounds deriving from MHG /ei/ from sounds deriving from MHG /î/, allowing speakers to differentiate between [hɔɐ̯m]/[haːm] NHG heim 'home' for the variable of MHG /ei/ on the one hand, and [aɛ̯s] Eis 'ice'; [ʃvaɛ̯n] Schwein 'swine, pig' or [laɛ̯çd ˚ ] leicht 'light, easy' for realizations of MHG /î/ on the other hand. However, both MHG sound classes are covered by New High German /aɛ̯/, rendering the underlying MHG reference sounds opaque. Note, then, that for the purposes of the present paper and its scope, the analysis is limited only to forms traceable to MHG /ei/ to investigate the entire dialect– standard axis. Yet, of course, a comprehensive analysis of variation in standard usage in particular would also investigate sounds deriving from MHG /î/ as equally interesting objects of study (cf. Moosmüller / Vollmann 2001).

Standard /aɛ̯/ is also presumed to undergo a change, which was first described by Gartner (1900) for the beginning of the 20th century. Without actually referring to the historical sound system, Gartner (1900) characterises the Viennese dialect as featuring the /aː/ monophthong in the cases of "tswa zwei, has heiß" ('two' and 'hot', derived from MHG /ei/) and an "ä i " diphthong in the cases of "trä<sup>i</sup> drei […], klä<sup>i</sup> gleich, sä<sup>i</sup> tn Seite" ('three', 'soon, alike' and 'side', relating back to MHG /î/; Gartner 1900: 142–143; emphasis in original: JFJ). For the latter group, he explains that the falling diphthong is realised 'with an a that approaches [open] e˛ [and is therefore transcribed with ä; translation: JFJ]' ("ä <sup>i</sup> mit einem a, das sich dem e˛ nähert", Gartner 1900: 143; emphasis in original: JFJ). This peculiarity of pronunciation is what would later in time become known as 'Viennese Monophthongization', relating not only to cases of MHG /î/ but also instances of MHG /ei/.<sup>14</sup> It describes an assimilation of the first and second diphthong parts in the diphthongs /aɛ̯/ and /au̯/ (see Moosmüller / Scheutz 2013: 83), resulting, for example, in [b ˚ ʁæːt] NHG breit 'broad' or [hɒːs] NHG Haus 'house'. As Luick (1996: 39; translation: JFJ) states in 1904, with respect to the "most articulated Viennese no diphthongs can be found anymore but instead only very open e and

<sup>14</sup> Note that the previously described process transforming /oa/ to /aː/ could also be labelled 'Viennese Monophthongization', for its supposed origin and centre of spread. However, so as not to cause confusion, I will only speak of 'Viennese Monophthongization' when referring to the 'upper' parts of the dialect–standard axis.

[in the case of /ɔɛ̯/] o sounds" <sup>15</sup>. In addition, Kranzmayer (1953) describes the process of Viennese Monophthongisation as a phonological shift from the already rather opened diphthong "ä ˙ e˛" towards the pure monophthong "ä ˙ ", attested especially in younger Viennese speakers (cf. Kranzmayer 1953: 209–210).

Gartner (1900) also notes that such a 'simplification' ("Vereinfachung") of the diphthong usually occurs among the younger generation, especially within lower social classes (cf. Gartner 1900: 143). Bíró (1910: 22; translation: JFJ) on the other hand, describes the "æi" to be a feature among the 'Austrian educated class' ("gebildeter Kreis") opposing the local pronunciation in the "heanzische" dialect in Burgenland, where the "diphthong ai is equally pronounced as ai, ei in theatre German". 16

Having evolved at the beginning of the 20th century, after an era of slightly open vowels prevalent in Vienna, the shift towards Viennese Monophthongisation is said to have been completed around 1940, leading to the general adoption of the new vowels to all structural positions (while they formerly occurred just in unstressed positions without compensatory lengthening, see Moosmüller 2002: 100), as well as to all social classes (cf. Moosmüller / Scheutz 2013: 83). Especially for young Viennese speakers of the standard variety,<sup>17</sup> a total takeover of Viennese monophthongised forms can be assumed (cf. Moosmüller / Vollmann 2001: 46), yet again with a trend towards intense spreading to other parts of Austria, particularly the Central Bavarian region and to a lesser degree to the South Bavarian area (cf. Moosmüller 1998: 10).

In the course of this development, the socio-normative status of the 'Viennese form' remains still quite indistinct, being used by speakers in their everyday life but being still considered controversial in professional contexts. Looking at the usage of diphthongs during news casts, Ulbrich (2003) found twice as many monophthongised forms in the pronunciation of five Austrian news broadcasters (31 %) in comparison to three analysed German announcers; however, the analysed sample was rather small, and no insights were shared on the regional backgrounds of the selected speakers and the respective perceptual reception. Krech et al. (2009) identify clear diphthongs in all regards in the usage of 'higher' standard pronunciation by professional speakers', with slightly monophthongised forms occurring only in their 'lower' standard pronunciation ("gemäßigte Standardaussprache"). As for non-professionals, the respective regional

<sup>15</sup> "Im ausgesprochendsten Wienerischen werden überhaupt keine Diphthonge mehr, sondern bloß sehr offene e und [im Falle von /au̯/] o dafür gesprochen".

<sup>16</sup> "Der diphthong ai wird ebenso gesprochen, wie bühnendeutsches ai, ei".

<sup>17</sup> Due to general standardisation processes, the local intended standard is often the primary variety of these young urban Viennese speakers, with dialectal and vernacular forms only holding ironic, expressive or imitational functions (cf. Soukup 2009: 39; Glauninger 2012: 113; Fanta 2017).

dialect influences their standard pronunciation, often leading to total monophthongisation among speakers from Vienna, Lower Austria and Burgenland (cf. Krech et al. 2009: 238). Moosmüller et al. (2015) suggest that monophthongisation just occurs in the "Viennese variant of SAG" (Standard Austrian German); however, in general Standard Austrian German, it is "restricted to unstressed positions". Since the goal of the present paper is to shed light on the entire dialect–standard axis and all 'vertical' variants of MHG /ei/, both positions are of interest. This also includes lexemes such as the above-mentioned heiß, beleidigen or Gemeinde, where the /ei/ is stressed. These particular lexemes, as already discussed, supposedly showa standard-pronunciation preference even in dialectal context, as a function of their semantic field (see Section 3.1). I further pursue this point in Section 5.3 on 'Lexical particularities'.

#### 3.3 MHG /ei/ in the Alemannic Dialects

So far, only the Bavarian phonological characteristics for MHG /ei/ have been described in the present paper, while the Alemannic traits have not been dealt with yet. One reason for treating them in a single section, now, is that there are still many research lacunae concerning the language situation in the Alemannic western part of Austria. In addition, regarding MHG /ei/, a broad range of variants have been attested in Alemannic Vorarlberg (cf. Gabriel 1994 ff.: 306–368; Lenz 2019: 329). Hence, for reasons of scope and availability of data, only a (regionally) limited overview can be given here. Klausmann (cf. 2012: 56) illustrates the distribution of variants for MHG /ei/ at the border of Austria, Switzerland and Germany, based on the findings of the linguistic atlas of Vorarlberg "VALTS" (= Vorarlberger Sprachatlas mit Einschluss des Fürstentums Liechtenstein, Westtirols und des Allgäus; 1994 ff.). Exemplified by the lexeme breit 'broad', we find evidence for differences in the quality of diphthongs and monophthongs on a very small scale, yielding for instance "braat" and "broat" next to open "bròòt" and an open /ɛ/ sound as in "bräät" (see Figure 3).

The present study draws on the data collected in the village of Raggal located in the Great Walser Valley ('Großen Walsertal', see centre of Figure 3) being part of the Walser dialects<sup>18</sup> in the Highest Alemannic dialect area. There, mainly "/äi/" diphthongs are expected in realizations of MHG /ei/, e. g. in zwei 'two' and Seife 'soap' (cf. Lenz 2019: 329), however, this /äi/ bears no likely connection to

<sup>18</sup> The Walser were an Alemannic people from the Swiss region of Wallis in the uppermost Rhône valley in present day Switzerland. Already around 1300, they started populating different parts of Vorarlberg and Liechtenstein, bringing their own dialect, whose relicts can still be traced until today (cf. Gabriel 1987: 18–19; Schallert 2013).

Figure 3: Variation of MHG /ei/ in the Alemannic state of Vorarlberg, bordering Switzerland and Liechtenstein (Klausmann 2012: 56)

the Viennese predecessor/intermediate variant cumulating in Viennese Monophthongisation (see Section 3.2). The VALTS similarly lists "e˛i" <sup>19</sup> and additionally "ae̯" for the village of Raggal in its map II 84, based on analysis of the words breit 'broad', Eiße 'furuncle', heiser 'husky', Laib 'loaf '. Especially the standardconform /aɛ̯/ pronunciation (or /ae̯/ and /ai̯/ respectively) as in "[tai̯l]" Teil 'part, piece', "[sai̯l]" Seil 'rope' and "[lai̯tərə]" Leiter 'ladder' is said to be a distinctive feature in dialectal speech for the Walser people (cf. Gabriel 1987: 26; Schallert

<sup>19</sup> The VALTS makes use of the Teuthonista transcription system, in which "˛" represents open vowels, hence "e˛i" could be translated to [ɛi] according to the International Phonetic Alphabet (IPA 2010).

2013: 90–91). Additional variants can be found with regard to specific phonetic contexts, such as closed "e ˙ ¯" monophthongs before nasals (e. g. in [heːm] heim 'home', see also VALTS II/88 and VALTS II/90a), or an /i/ monophthong solely in the case of the lexeme [kliːn] klein 'small, tiny' (VALTS II/97), which might have developed diachronically in a parallel process along with MHG klein (cf. Gabriel 1994 ff.: 353–357).

From this it appears that at least for the village of Raggal three phonological variants, namely /ai, ae/ (henceforth as /aɛ̯/), the /äi/ diphthong (henceforth as /ɛe̯/ to differentiate from Viennese /æe̯/) and the /eː/ monophthong (before nasals) can be assumed as realizations of MHG /ei/ for all parts of the vertical dialect–standard spectrum (though with unknown shares), even though diverging transcribing conventions in the literature additionally impede an exact classification of existing data. Outside of such limited investigations as we possess on regional variation in MHG /ei/ in Vorarlberg, there is virtually no information about language shifting in relation to different situational settings or social class. Of course, one important aspect to note here is a general, ongoing discussion about the very nature and 'architecture' of the dialect–standard axis in Vorarlberg. While the linguistic situation is often assumed to be diglossic similarly to the situation in Switzerland, newer studies promote the concept of a dialect–standard continuum comparable to the Austro-Bavarian model (cf. Kaiser / Ender 2015; Schönherr 2016).

Section 5 further below shows how this plays out in the data analysed in the present paper. Before that, however, Section 4 sets up the empirical part by introducing the study methodology.

#### 4 Methodology and Corpus

In this section, I present the data analysed to gain insight into the horizontal and vertical patterning of variation in the selected variable of MHG /ei/. As already discussed, sociolinguistic variation in the Austrian dialect–standard spectrum is marked by considerable complexity. To gain insights while paying tribute to this complexity, the data used in the present study were collected under a multimethod approach that applies a range of speech elicitation settings in all research locales, featuring both more controlled as well as'free' conversational contexts of production. This set-up facilitates the identification of particular, situationally influenced speech patterns, allowing for conclusions to be drawn regarding the overall, general dynamics on the social-vertical axis.

These data are taken from a large speech corpus created under the Special Research Programme (SFB) "German in Austria: Variation – Contact – Perception" (in German "Deutsch in Österreich. Variation – Kontakt – Perzeption"), a multi-year project on language use in Austria financed by the Austrian Science Fund (FWF F60-G23).<sup>20</sup> In corpus compilation, a variety of speech elicitation methods were used, so that 'standardised' as well as'free' conversational data are included (for more detailed information see Lenz 2018). The conversational data consisted of a formal interview conducted by a non-local academic and an informal conversation among friends without the researcher being present. Furthermore, in the course of the inquiry, the informants were asked to perform various tasks, such as to translate the so-called Wenker Sentences<sup>21</sup> into their own local dialect or into their standard variety. For this task, each sentence was presented orally, in a recording, in one variety, and then had to be translated into the respective other variety, i. e., in a first step from Standard Austrian German (spoken by an Austrian TV-newscaster) into the local dialect, and then in a second step from the local dialect (spoken by a young local) into the standard variety. During two read-aloud tasks, the informants were asked to read isolated words (in randomised order), as well as a short text, entitled "Nordwind und Sonne" ('The North Wind and the Sun' <sup>22</sup>), with the aim of capturing their most standard-near pronunciation. The methodological toolkit used here is similar to ones used in other variationist linguistic projects (e. g. the German REDE project)<sup>23</sup>. Conveniently, the selected design allows for naturalness on the one hand, but also for standardisation and comparability within and across studies on the other (for a methodological discussion see Schmidt / Herrgen 2011).

For each informant, about four hours of speech material were audio-recorded. For the purposes of the present paper, 25 female speakers were selected for analysis. The places of origin of these 25 informants are rural villages with a population of 500 to 2,000, evenly spread across Austria (from west to east: Raggal, Tarrenz, Tux, Weißbriach, Oberwölz, Hüttschlag, Passail, Neckenmarkt,

<sup>20</sup> The present data and results were developed in the framework of project part 03: "Between dialects and standard varieties: Speech repertoires and varietal spectra." (FWF F06003, principal investigator: Alexandra N. Lenz). See the project homepage: <https://dioe.at/en/> (June 3, 2020). Further information is provided by Lenz (2018) and Budin et al. 2019.

<sup>21</sup> 'Wenker Sentences' are a tool for the elicitation of forms of spoken language that were introduced as a data collection method for dialect surveying by Georg Wenker in the 19th century. Wenker sent "questionnaires to schools where he had them filled out by local informants. Their task […] was to translate a series of Standard German sentences into the local dialect by means of the "allgemein gebräuchliche Alphabet" ['common use alphabet']. The sentences were construed in such a way as to elicit particular phonological and select grammatical features of the dialects in the translation" (<https://regionalsprache.de/en/en/ contents-wenker-questionnaires.aspx> (June 3, 2020).

<sup>22</sup> 'The North Wind and the Sun' is a text routinely used in linguistics for the illustration and elicitation of a range of sound realizations. Based on an ancient fable, it exists in translations for many languages andwas originally employed for illustration of the International Phonetic Alphabet by the International Phonetic Association (1949).

<sup>23 &</sup>lt;https://www.regionalsprache.de/en/> (June 3, 2020).

Taufkirchen a.d. Pram, Steyrling, Neumarkt a.d. Ybbs and Allentsteig and Gaweinstal; see Figure 4). First, to investigate the horizontal dimension of MHG /ei/, the translations into dialect of all 25 speakers were analysed, 13 of them being grouped together as 'young' (age between 18 and 35 years) with high formal education (high school 'Matura' graduate), and 12 being characterised as classical 'NORFs' (non-mobile, old, rural, female; aged above 60; see Chambers / Trudgill 1993: 29).<sup>24</sup> Second, in a follow-up analysis, the vertical dimension was taken into account by focusing on all six situational settings for one young and one old informant from only four of the selected locations, each functioning as a representative of one of Austria's major dialect regions: Raggal in Vorarlberg representing the (Highest) Alemannic dialect area, Weißbriach in Carinthia for South Bavarian, Neckenmarkt in Burgenland as a sample from the South-Central Bavarian transition area, and Neumarkt/Ybbs in Lower Austria representing Central Bavarian (see highlighted locations in Figure 4).

Figure 4: Sample locations for this study, evenly distributed across Austria. First analysis investigates all locations, second analysis focuses only on selected villages (highlighted in red)

During data processing, the 'natural' conversational data were orthographically transcribed with the transcription software EXMARaLDA Partitur Editor<sup>25</sup>, and the 'standardised' elicitation data were prepared for further analysis by reorganising the required items and linking them to the respective audio samples. In a second step, auditory phonetic transcriptions (according to the IPA) and annotations regarding the selected phenomena were performed in both datasets,

<sup>24</sup> For the village of Taufkirchen, no older female informant could be recruited so far; this results in the present asymmetrical distribution between 13 younger and 12 older speakers in my sample.

<sup>25</sup> See Schmidt / Wörner 2014 and the website: <https://exmaralda.org/en/> (Junee 3, 2020).


yielding a total number of approx. 2,657 tokens for the two analyses (see Table 1, for further information on the settings see Lenz 2018: 272–273).<sup>26</sup>

Table 1: Methodological matrix for the two rounds of analysis of realizations of MHG /ei/

As Table 1 shows, the first round of analysis comprises around 287 tokens of MHG /ei/ from only one elicitation setting (the translation into dialect), while the second round features about 2,370 tokens, collected in six different elicitation settings.

# 5 Results

#### 5.1 Study 1: The Horizontal Dimension of MHG /ei/

By comparing the dialect translations based on the Wenker Sentences, the frequencies of variants deriving from MHG /ei/ become observable for all 13 Austrian locations. The Wenker Sentences (WS) provide nine potential places of occurrence for realisation of MHG /ei/, all of them in stressed positions. Three additional sentences were added in the framework of the SFB "German in Austria" to incorporate specific (morpho-syntactic) Austro-Bavarian and -Alemannic features, which could not have been captured by the original Wenker Sentences. Out of these added sentences, the case of Wahrheit 'truth' is the only

<sup>26</sup> At this point, I would like to express my heartfelt thanks and appreciation to the student research assistants Barbara Binder and Florian Tavernier for their help in preparing, transcribing and annotating the data, and their valuable input in any of our technical discussions.


one with /ei/ in unstressed position. Table 2 gives an overview of all selected sentences for the translation tasks.<sup>27</sup>

Table 2: Potential places of occurrence for realisations of MHG /ei/ in the Wenker Sentences and three additional sentences

Since the translation stimuli were only presented orally, informants often undertook certain modifications in their translations. The lexeme Kleider 'clothes, dresses', for example, was more often than not changed to Gewand,amore frequent variant in the south of the German speaking countries. Sometimes, informants also unconsciously picked additional lexemes in other Wenker Sentences, expanding the list of potential places of occurrences, e. g. exchanging zu Hause by daheim 'at home' (WS\_15). Note, of course, that this leads to differing token numbers (n) among the selected informants in the present analysis. Furthermore, some words were consistently realized in the exact same way, precluding most kinds of variation. Thus, for example, Fleisch 'meat' and meisten, 'most', as well as the suffixes {-ei} (i. e. {-eien} in plural) and {-heit} as in Kindereien 'childish pranks' and Wahrheit 'truth' were consistently realised with a standard-near variant. In the case of Eier 'eggs', speakers used primarily the standard variant or the dialectal diphthong /oa/, monophthongised [ɑːɾ] was only pronounced once by a young informant from Tarrenz. As such lexical particularities may also be a part of the described language processes, they were not

<sup>27</sup> The Wenker Sentences here follow the orthography of the original phrases (see https:// regionalsprache.de/en/contents-wenker-questionnaires.aspx (June 3, 2020)). Note that orthographic <ei> does not necessarily have to be MHG /ei/, lexemes such as deiner 'your', meiner/-m 'my' or treiben 'to wreak sth.' relate back to MHG /î/ (see Section 3.2).

excluded from the analysis; instead, they are discussed separately at the end in Section 5.3.

The map provided in Figure 5 depicts the results for the translations into dialect of all 25 informants, across all sample regions (n of tokens = 287). In line with the existing literature (see Section 3), seven different phonological categories could be identified: /oa/ as the 'original' Bavarian variant (blue), /aː/ as the 'newer' form originating in Vienna and Carinthia (red), the Alemannic (Raggal) /eː/ monophthong (yellow) and /ɛe̯/ diphthong (orange), as well as the standarddiphthong /aɛ̯/ (including also other standard notation conventions, e. g. [aɪ̯]; white) with its'partway Viennese monophthongised' variant /æe̯/ (including also [æɛ̯]; medium grey) and the fully Viennese monophthongised /æː/ (including also [ɛː]; dark grey).

Figure 5: Realizations of MHG /ei/: Results of the areal-horizontal analysis of the translations into dialect by all 25 informants across Austria, regarding the entire dialect–standard axis

The map in Figure 5 illustrates which variants define the interindividual repertoires for each location. Focusing on the west, only in Raggal, the monophthong /eː/ (yellow) and the /ɛe̯/ diphthong (orange) are in common use, representing more than half of all variants in this location (56,5 %). These do not constitute attested forms in the Bavarian areas, which confirms a systematic phonological difference between the Alemannic and Bavarian parts of Austria. Yet, within Bavarian, differences between west and east also lead to diverging language behavioural patterns. On the 'lower' part of the dialect–standard axis, we find a clear preference for the /oa/ diphthong (blue) in Austria's western states Tyrol and Salzburg (which is even an exclusive preference in Tux and Hüttschlag). The only exception in the southwest is Weißbriach, where no single use of the /oa/ diphthong was attested; this, however, follows exactly Wiesinger's findings for Carinthia. In addition to Weißbriach, the informants from Gaweinstal also seem to produce the /aː/ monophthong (red) invariably as their only dialectal variant. Proximity to Vienna may be a major influencing factor in this regard.

Concerning the 'higher' parts of the dialect–standard axis (shades of white/ grey), a similar east-west-opposition can be found: While the standard diphthong (white) is predominant in the west, Viennese monophthongised forms (dark and light grey) constitute the main variants in the east. Here, geographic location is not the only factor that seems to play a role; the underlying dialect region also appears of importance, as the highest percentage of Viennese /æː/ monophthongs (dark grey) is attested in the Central Bavarian dialect area.

Figure 5 shows only the regional distribution of the selected variants. Thus, it does not allow for conclusions about individual language repertoires or patterns by sociolinguistic parameters, such as age. Figure 6 captures this aspect in detail.

Figure 6: Realizations of MHG /ei/: Results of the areal-horizontal analysis of the translations into dialect by all 25 informants across Austria, regarding the entire dialect–standard axis; comparison by informants' age group ('old': 60+; 'yng': 18–35); analysed dialect areas: AL = Alemannic, AL-B = Alemannic–Bavarian, SB = South Bavarian, S-CB = South-Central Bavarian, CB = Central Bavarian

In Figure 6, for each location, the results for the older speakers are shown on the left side and those for the younger speakers on the right side of the x-axis; and the respective percentage of the different variants used are depicted on the y-axis. Again, the gradual increase of both current monophthongised forms (/a:/ and especially Viennese /æ:/) is evident in the trajectory from left (west) to right (east). In many villages, the individual translations of the Wenker Sentences lead to quite similar results among the older and the younger generation, as in the cases of Tux (TUX), Allentsteig (ALLE) and Weißbriach (WEISS). In other cases, patterns of minor difference emerge that suggest an intergenerational (apparenttime) decline of specific forms and/or the establishment of 'new' variants (e. g. in Raggal (RAGG), Tarrenz (TARR) or with regard to the 'higher' parts e. g. Steyrling (STEY)). This applies not only to different compositions of the 'higher'(shades of white/grey) and 'lower' (other shades) parts of the individual spectra, but also to the general scale between both ('higher' vs. 'lower') components. Indeed, some villages demonstrate considerable intergenerational differences, thus contributing viable evidence for the described processes. While the /oa/ diphthong is the only or main variant for the older informants in Oberwölz (OBER), Neumarkt/Ybbs (NMYB) and Neckenmarkt (NECK), a strong shift towards the use of the /aː/ monophthong is attested among the younger speakers in these locations. Focusing on the relation of /oa/ and /a/, only in Passail (PASS), a contrary process seems to be taking place, with the young female showing higher numbers of the /oa/ diphthong in comparison to her older counterpart. Otherwise, general differences between the older and younger informants do not seem to be strong. Note that, being part of the South-Central Bavarian language area, Passail is located exactly between the two centres of expansion of the /aː/ monophthong, Vienna and Carinthia. Thus, interfering concepts about what constitutes the 'old' and 'new' variant may be responsible for the young person's production. Certainly, further sociolinguistic factors may also play an important role, such as individual mobility, concepts of identity and/or an individual's linguistic orientation towards urban or rural areas. These require further investigation, which is, however, beyond my present scope.

#### 5.2 Study 2: The Vertical Dimension of MHG /ei/

Connected to the above-mentioned implications of situation as an important sociolinguistic parameter the question arises of how speakers may vary their language production across different situational settings. Hence, a second round of analysis now zooms in on the vertical dimension of variation in realizations of MHG /ei/, in the four villages of Raggal, Weißbriach, Neckenmarkt and Neumarkt/Ybbs, each selected as a representative of one of Austria's major dialect regions. Figure 7 reproduces the results from the translations into dialect already described earlier, now limited to the 8 respective speakers (see top row of bar graphs in Figure 7). To this, the results from the free conversational tasks, i. e. Interview and Conversation among friends, are added (see bar graphs in the centre and at the bottom of Figure 7). The average number of tokens for these two situational settings is n = 128 and n = 126, which should yield results that are even more robust than those from the translations into dialect (n of tokens = 11–13).<sup>28</sup> For these settings, the category 'other' was added to the variants, consisting of very few additional forms (n = max. 10 per person and setting) that eluded clear categorisation (e. g. no full vowel as in [vɐs] for NHG (ich) weiß 'I know')

Figure 7: Realizations of MHG /ei/: Results of the vertical-social analysis for 8 informants in four locations across Austria; comparison across the speech situations of translations into dialect, Interview and Conversation among friends; values of n are per person

Examining all three situational settings in general, the Alemannic–Bavarian contrast again becomes apparent. Apart from very few instances of /aː/ monophthongs in the Raggaler Conversation among friends (n = 5 in total); occurring

<sup>28</sup> The particle nein 'no' and the indefinite article ein 'a/an' were left out of the analysis due to their excessive usage in spoken German and frequent occurrence in unstressed positions of the sentence, which could have led to imprecise transcription and annotation. Note also that some variants (e. g. /i/ monophthongs) occur so rarely that their frequency is not displayed in the graphic.

mostly in highly-frequent conversational particles such as [vaʃ] weißt (du) 'you know'), the intra-individual spectra in these two language areas seem to consist of entirely different components regarding the 'lower' part of the dialect–standard axis. Moreover, the general frequency of rather 'dialectal' forms differs considerably between these two groups, indicating that generally less dialectal variants of MHG /ei/ are used or the dialectal variants rather coincide with the standard forms in the Alemannic area of Austria (see also the results regarding 'Lexical particularities' presented in Section 5.3). Thus, we observe a gradual reduction of /oa/ diphthongs and /aː/ monophthongs across all settings in the Bavarian villages, with a surprisingly high number of remaining 'dialectal' variants in the supposedly quite formal setting of the Interview (approx. 40–60 %, see also Fig. 7). By contrast, in Alemannic Raggal, no single use of an /eː/ (or /iː/ 29) was found in this setting. Arguably, such avoidance of certain Alemannic traits may be attributed to some form of accommodation that is due to the Interview being perceived as a rather formal situation, which elicits standard-near speech. Furthermore, considerations of mutual intelligibility may play a role, especially with an interviewer who is a non-local from outside of the Alemannic state of Vorarlberg, as was the case here (the interviewer being from Vienna).<sup>30</sup> These findings support the idea of a diglossic language situation in Vorarlberg corresponding to the Swiss linguistic situation, where code switching rather than gradual code mixing occurs (see Section 3.3). At the same time, when looking at the Alemannic diphthong /ɛe̯/ (orange), only the young person from Raggal performs any inter-situational modifications at all, while the old person continuously uses a similar percentage of this variant in all three settings. Consequently, complexities arise for classifying the Raggaler /ɛe̯/ sound on the vertical spectrum. I will return to this issue later in my discussion of the remaining situational settings.

As for the Bavarian speakers, there are patterns of differences in variant use that clearly set the villages of Neumarkt/Ybbs and Neckenmarkt apart from Weißbriach. Thus, while both speakers of Weißbriach stick with an exclusive use of the /aː/ monophthong in all three settings, there is a gradual reduction of the /oa/ diphthong in Neumarkt/Ybbs and Neckenmarkt from the translations into dialect to the Conversation among friends up to its very limited use in the Interview, where /aː/ also becomes the dominant dialectal variant. Regarding the

<sup>29</sup> As mentioned in Section 3.3, in addition to the previously discussed (Highest) Alemannic /eː/ monophthong, four instances of an /iː/ monophthong (dark yellow in Figure 7) are attested in the data, being restricted to specific lexemes, i. e. klein/e 'small' (2x) and (ent)scheiden/ (aus)scheiden 'to decide/ to quit' (2x). These might constitute a diachronically incorrect link back to MHG /î/ or derive from two forms of simultaneous historic development.

<sup>30</sup> For a broader discussion on the impact of the interviewer's nationality on German in Austria see Stiglbauer / Koppensteiner / Lenz (in preparation).

development of the variants over apparent-time, there is a strong tendency of decline for the diphthong among the young generation, indicating a potential process of change. One possible interpretation would be that the young informants still use the diphthong in their 'intended dialect' during a rather controlled dialect translation task, according to their concept of what the 'most original local dialect' sounds like. In their 'unguided free conversation', though, the variant is diminished to 3–16 % in the Conversation among friends and has basically vanished in the Interview. A related assumption would be that the perceived degree of situational formality also plays a role, promoting the decline, if speakers perceive the monophthong as less dialectal.

Let us now turn to the 'higher' parts of the dialect–standard axis (shades of white/grey). Here, the conversational data confirm the previously described divergence in variant use across the country, such that standard /aɛ̯/ is used more in the west/south, and /æe̯/ and /æː/ in the east. The other situational settings provide further results in this regard, which are visualised in Figure 8, featuring the translations into standard (as a counterpart to the prior translations into dialect), as well as the two Read-aloud tasks, one using the text called Nordwind und Sonne and the other using the isolated words from the Word list.

The underlying intention of including these settings in the study design was to elicit standard-near pronunciation. The fact that the speakers analysed do not use a single variant of the previously discussed dialectal /oa/ diphthongs or /aː/ and /eː/ monophthongs in these settings suggests that the method was successful in this regard. Arguably, it can furthermore be concluded that the forms we find are presumably evaluated as standard-near by the informants themselves and correspond to their individual normative concept of their own 'intended standard language'.

Quite surprisingly, actually, the young woman from Neumarkt/Ybbs uses Viennese monophthongisation consistently and to a very high degree during the translations into standard and reading tasks, and a lot more than her local counterpart from the older generation or the informants from Neckenmarkt. Arguably, the Viennese influence thus becomes apparent for this Central Bavarian village. In Vorarlberg, we observe the Alemannic /ɛe̯/ diphthong in the utterances of the older informant from Raggal during the translations into standard. However, there are no occurrences at all in the reading tasks. It seems that the variant /ɛe̯/ may be used consistently (though not exclusively) in all oral settings, but is clearly disfavoured in reading contexts. Hence, it might be a marked dialectal form regarding the 'highest' parts of the dialect–standard axis, while its social standing and saliency remain fairly unclear for the rest of the axis.

In general, there is an overall trend of an increasing frequency of standard /aɛ̯/ for almost all informants from the Standard translations via Nordwind und Sonne to the Word list. This again suggests that the methodological set-up with its

Figure 8: Realizations of MHG /ei/: Results of the vertical-social analysis for 8 informants in four locations across Austria. Comparison across the speech situations of translations into standard, the Nordwind und Sonne read-aloud task and the Word list read-aloud task

range of settings has the potential to successfully access and elicit a wide range of the inter- and intraindividual language spectra.

#### 5.3 Lexical Particularities

The final analysis section presented in this paper now tackles the issue of lexical particularities. As described in the literature review, specific words do not seem to be translatable into the local dialects due to their historical association with specific domains of use (e. g. a clerical or juridical context). To empirically compile a comprehensive list of such words, all lexemes produced in a standardnear way in our informants' Interviews and Conversations among friends were analysed, yielding the list of 'mostly invariable' words presented in Table 3.<sup>31</sup>

<sup>31</sup> Translations to English: left column: 'actually'; 'club/society'; suffix -heit('childhood, safety');


Table 3: List of 'lexical particularities' in the Interview and Conversation among Friends sorted by frequency, i. e. lexemes most consistently and invariably produced in a standardnear way regarding MHG /ei/ (exceptions are marked with \*)

Of course, presumably, with a bigger corpus, the list shown in Table 3 would need to be extended. Furthermore, it is likely that, with the inclusion of more speakers, the chances of encountering exceptions to the rule would increase. Yet, it is

<sup>&#</sup>x27;part, partially, advantage, divided, to participate'); 'most'; 'own/distinct, features, peculiar, solitary person, sovereignty/autonomy'; 'clean, to clean'; suffix -keit ('activity/occupation, difficulty'); 'unfortunately passionate, misery, to suffer'; 'nursing home direction, 'local/ autochthonous, incredibly/eerily'; 'travel'; 'joint, together, nasty/mean, generally'; coll. 'awesome'; 'administration, to manage, to accompany, postcode'; suffix -ei ('bakery, chancery, party, library'); 'to name/label sth., to draw, marked'. Right column: 'to suffice, to reach/ achieve, accessible/attainable'; 'circle, roundabout, clique/circle of friends, circuit'; 'municipality'; 'opinion'; 'several'; 'to decide, to distinguish, to quit'; 'to master, master, championship'; 'keen/enthusiastic, spirit/ghost, spiritual/clerical'; 'prepared'; 'to grab/to grasp, comprehensive/overall'; 'cure/healing, holy/sacred'; 'May'; 'to acquire sth., event'; 'to refuse'; 'ribbon/loop, polisher/grinder'.

striking that in our dataset specific lexemes occur (almost) invariably throughout all analysed conversations with only few deviations (marked with an asterisk in Table 3): Even though more than 50 tokens of words ending on {-heit} were attested as produced in a standard-near way, there was still one case of [kʁɔŋkad ˚ n̩ ] Krankheiten 'illnesses' with an /a:/ monophthong (produced by an older speaker from Neumarkt an der Ybbs). As for the 'sacral word' heilig 'sacred', one out of four instances is produced with an /aː/ monophthong, and thus realised as [haɭɪçn̩ ]—a finding that corresponds with Bíró's descriptions from 1910. Morphology constitutes an additional factor in the case of Heim 'home', where compound words (Pflegeheim 'nursing home', Heimleitung 'nursing home direction') seem to impede the use of /aː/ or /oa/. Also, additional phonological processes must be taken into account, as for instances of Teil, teilweise etc. 'part, partially', which constitute potential places of occurrence forl-vocalisation with a rounding of the preceding vowels in a great part of the Bavarian area; these cases were consequently not included in the analysis. Generally, conclusions and generalizations concerning lexical particularities must be developed very carefully.

In fact, taking these particularities out of the dataset, the amount of general 'dialect' use increases considerably for most analysed speakers. The bar graphs featured in Figure 9 show the differences between the distribution of variants on the 'higher' (light grey) and the 'lower' part (dark grey) of the individual spectra before and afterfiltering out the listed lexical particularities. Since the translation sentences consist only of a small total number of lexemes and do not target lexical variation in their methodological conception, Figure 9 just focuses on the settings of Interview and Conversation among friends, where a high token count of realizations of MHG /ei/ occurs.

For most speakers, we see a difference in their ratio of dialectal vs. standard variant use between the Conversation among friends and the Interview in both analyses—filtered as well as unfiltered (on average, for the unfiltered analysis, a difference of around 11 %, and for the filtered version around 8 %). This seems to constitute evidence of the fact that the informants draw a contrast between the two settings and the associated formality, insofar as it can be assumed that they express this in their overall 'dialectality' rates. The picture is more complex for the informants from Weißbriach, who also show patterns of increased 'dialectal' usage in the Interview than in the Conversation among friends. In addition, for the Bavarian informants, it becomes apparent that the listed particularities represent the majority of standard-near variants in their speech production. This seems to indicate that the speakers do not necessarily vary so much on the entire dialect–standard axis, but rather in their lexical choices for different situational settings. In other words, they might produce the invariant lexemes to a slightly greater extent during the Interview, compared to the Conversation among

Figure 9: Realizations of MHG /ei/: Results for the settings of Interview (INT) and Conversation among friends (CaF), once without filtering of lexical particularities (above), and once with filtered graphs (below)

friends. As seen with the word Kreisverkehr 'roundabout' as another example for a compound (Kreis 'circle' + Verkehr 'traffic'), the phonological choices might be connected to the general preference for certain morphological and syntactical structures and their frequencies in specific situational settings.

In general, then, we find evidence for a very high usage of rather dialectal forms for all analysed Bavarian speakers from rural regions of Austria. When those words that are invariably pronounced in a standard-near manner are taken out of the data, the overall proportion of realizations of MHG /ei/ on the 'lower' part of the standard axis reaches 88.1 % for the Conversation among friends and 80.3 % for the Interview.

Figure 9 also shows that the filtering out of the lexical particularities does not affect the dialect rate of the Alemannic speakers to a great extent. Arguably, this has to do with the limited absolute amount of /eː/ and /iː/ monophthongs (24 tokens for the older and 30 for the younger informant). As it turns out, also among those remaining dialectal records, approx. 74 % can be traced back to the two lexemes heim and kein, reoccurring multiple times as [heːm] or [keːn] for both speakers (older: n of tokens = 17; younger: n of tokens = 15). As described by Gabriel (cf. 1998: 353–357), the use of these variants is motivated by the phonetic environment of MHG /ei/ being followed by a nasal, hence, arguably, the lexical particularities are of less importance in comparison.

Additional analysis on the lexical and phonetic characteristics of the presented data, as well as multidimensional statistical analysis capturing the impact of region, age, gender, the situational settings (and other parameters) would facilitate describing the complex dynamics occurring in the German language in Austria, however, they are beyond the scope of the present paper.

#### 6 Conclusion and Research Desiderata

The aim of this article has been to describe the dynamics and patternings of variants of the phonological variable of MHG /ei/ in Austria, in their full sociolinguistic complexity, by considering both the areal-horizontal as well as the social-vertical language dimension. As the results show, the 25 informants from different rural regions across Austria whose language variation was analysed by means of auditory phonetics have recourse to diverse individual speech repertoires and variation spectra, allowing them to code-switch and -shift flexibly on the dialect–standard axis. Areal distribution and situatedness in a specific dialect region appeared as an important influencing factor, yielding great differences between the Alemannic and the Bavarian parts of Austria but also between different areas of the Bavarian dialect family.

Furthermore, current processes of language change arguably induced by continuous contact between the prevalent varieties have been shown to lead to innovative language behavioural patterns among the analysed speakers. This has become apparent not only in departures from former observations on regional variation (e. g. in dialectological maps of regional distribution and descriptions of local base dialects), but also in the results from an apparent-time comparison of two different age groups in each selected location. While some villages remain mostly homogeneous in terms of their language behaviour patterns, others demonstrate clear intergenerational differences between the respective older (60+) and younger (18–35) speakers. This applies not only to the 'lower', dialectal parts of the dialect–standard axis, but also to the 'upper', standard-near parts of the spectrum.

As a detailed follow-up-analysis has shown, especially in the Central and South-Central Bavarian villages of Neumarkt/Ybbs and Neckenmarkt two processes of monophthongisation appear to have continuously spread out from Vienna, pushing formerly used variants slowly aside. By comparing data elicited in different situational settings involving 'natural' conversational data, rather controlled language translation tasks as well as reading tasks, also the socialvertical dimension of these changes could be captured. While the informants still produce the /oa/ diphthong seemingly according to their imaginary concept of their respective villages' base dialect in dialect translation tasks, a radical decline of this form is taking place in the context of 'natural' conversations, where the /aː/ monophthong is on the rise.

Attributing this phenomenon to dialect convergence or divergence is actually quite difficult, since it is not quite clear if this originally 'Viennese' /aː/ variant can be regarded as an external influence when it already represents a stable part of the speakers' individual repertoires. The main question is rather now whether the original form is successively being replaced by the supposedly more prestigious, monophthongised variant, or whether it is undergoing a form of reallocation, eventually covering different socio-communicative functions (cf. Britain / Trudgill 2005: 205), i. e. in form of an idealised 'dialect reminiscence' serving an expressive, ironic or imitational purpose.

On the 'upper' part of the dialect–standard axis, Viennese monophthongisation of standard /aɛ̯/ seems to be spreading, already constituting a legitimate form even during the reading tasks for the young speaker of Neumarkt/ Ybbs. Here, the situation is somewhat the reverse of that involving /aː/, since the 'new' /æː/ variant has no clearly assigned standard status but is used nonetheless in clearly standard-related contexts. This development is quite surprising, considering the fact that most base dialects (especially in the west) hold /aɛ̯/ as their 'default form' and the medial and economic influence from Germany has strongly intensified in the last twenty years, both sources of linguistic opposition to the promotion of Viennese /æː/. Again, the allocated function and the associated beliefs will play an important role for the future development of this feature, as it may either become increasingly stigmatised as a symptom of 'Viennese provincialism' (in contrast to an international self-presentation), or as an intentional, affirming expression of Austrian identity.

Future research will have to focus particularly on these perceptual aspects of both ongoing processes involving realizations of MHG /ei/. Additionally, in terms of the 'higher' parts of the axis, acoustic measurements might be advantageous, in order to corroborate the results presented here.

In summary, and returning to Trudgill's proposed tripartite set of options in the outcome of dialect mixing (see Section 2.1), we can now note, based on the gathered evidence, that actually all three seem to apply in the Austrian context of dialect–standard contact in the case of MHG /ei/. Thus, we find quantitative differences in the patterning of older and newer variants, as well as newer realizations in certain lexemes only. Finally, we could even argue that 'dialect fudging' (the generation of phonetically intermediate variants) occurs, in the way that transitional forms like [æɛ̯] represent a somewhat mixed/fudged version mediating between standard [aɛ̯] and Viennese monophthongized [æː] and projecting the onset of Viennese monophthongization.

In any case, the dialect–standard spectrum of German language use in the Austrian context has once more been found to be highly complex and intricate, and to provide a highly dynamic laboratory for sociolinguistic investigation.

#### Bibliography


VALTS = Vorarlberger Sprachatlas mit Einschluss des Fürstentums Liechtenstein, Westtirols und des Allgäus. 1985 ff., bearbeitet von Eugen Gabriel. Bregenz: Vorarlberger Landesbibliothek.

Vollmann, Ralf / Seifter, Thorsten / Hobel, Bettina / Pokorny, Florian. 2017. l-vocalization in Styria. In: Sylvia Moosmüller / Manfred Sellner / Carolin Schmid (eds.): Phonetik in und über Österreich. Wien: Österreichische Akademie der Wissenschaften, 123–135.

Wandruszka, Mario. 1979. Die Mehrsprachigkeit des Menschen. München: Piper.


Zehetner, Ludwig. 1985. Das bairische Dialektbuch. München: Beck.

Mag. Johanna Fanta-Jende University of Vienna Department of German Studies Porzellangasse 4, Stiege 1, 3. Stock, Raum 367 1090 Wien Austria johanna.fanta@univie.ac.at

# Dialect Contact in Salzburg. The Case of sein ('to be')

Abstract: In Salzburg's base dialects the verb sein ('to be') shows a wide range of variation in its plural forms. In addition to the alternation of the h/s-anlaut and the a/ai-variation of the stem vowel in the 2PL, this chapter also deals with variation in the suffixes in the 1/3PL. In order to describe and explain dialect contact, we conducted real- and apparent-time studies on different datasets. For the real-time study, we compared data collected in a dialectological survey in the 1970/80s with data from a recent 2016/17 survey that contributed to a linguistic dialect atlas of Salzburg. We contrasted the real-time evidence with two apparenttime studies: The first is based on the dialect atlas data, while the data for the second study originates from a third corpus ("Deutsch in Österreich" 'German in Austria' 2016/17) which was also gathered in 2016/17. The analysis of these datasets shows why change, previously considered as merely horizontal contact between dialects, cannot be explained without vertical convergence towards the standard variety.

Keywords: language contact, dialect contact, convergence, dialect-to-standard advergence, language variation and change

Abstract: Die Pluralformen des Verbs sein zeigen in den Dialekten Salzburgs ein breites Variationsspektrum. Neben der h/s-Anlautalternation sowie der a/ai-Stammvokalvariation in der 2.Ps.Pl. steht die Suffigierung der 1./3.Ps.Pl. im Fokus dieses Beitrags. Dabei werden Daten aus verschiedenen Erhebungszeitpunkten und -kontexten miteinander verglichen, um Wandelprozesse zu identifizieren. Für eine real-time Studie werden Daten zweier dialektologischer Erhebungen aus den 1970/80er Jahren und von 2016/17 miteinander verglichen. Der real-time-Evidenz stellen wir zwei apparent-time-Untersuchungen gegenüber, in denen wir jüngere und ältere Sprecher kontrastieren. Die erste apparent-time-Studie basiert ebenfalls auf den Daten von 2016/17, die für den "Sprachatlas Salzburg" erhoben wurden. Um die inter- und intra-individuelle Variation an bestimmten Ortspunkten zu verdeutlichen, wird zudem eine detaillierte apparent-time Analyse anhand eines dritten Korpus ("Deutsch in Österreich" 'German in Austria' 2016/17) von 2016/17 vorgestellt. Auf Basis des Materials wird argumentiert, dass Wandelprozesse, die bisher als horizontal-diatopische Konvergenz betrachtet wurden, nicht ohne vertikal-diastratische Advergenz gedacht werden können.

Keywords: Sprachkontakt, Dialektkontakt, Konvergenz, Advergenz, Sprachvariation und -wandel

#### 1 Introduction

Language contact has been seen in the last decades as an essential prerequisite in triggering linguistic innovation, i. e. language variation (short-term dynamics) and contact-induced change (long-term dynamics). Contact provides the context for variation and change, in making features of one variety accessible to speakers of another. This contribution is concerned with the subject of dialect contact, i. e. horizontal and vertical contact between varieties in one language.<sup>1</sup> Therefore, bidialectal speakers with access to the features of at least two varieties serve as a link between the two systems, providing a conduit of innovation from one variety to another.

The effects of dialect contact can be described in terms of convergence, divergence, advergence, and reallocation (e. g. interdialect formation) between the varieties (cf. e. g. Auer 2018; Britain 2010; Dahl 2009; Trudgill 1986). In this contribution, we want to illustrate and discuss the case of co-emergence of some of these developing patterns in a complex dialect contact situation, taking horizontal (convergence between dialects, reallocation) and vertical contact (dialectto-standard advergence) into account, as "[u]sually, these two developments go hand in hand, leading to leveling" (Auer 2018: 159). This chapter concerns, however, only a very limited set of variables: the verb sein ('to be') and its features, in particular its plural forms (present, indicative). In its function as an auxiliary, the verb sein is insofar special as it has the second highest token frequency of all verbs in German (cf. Nübling 2000: 12–14; Philipp / Weider 2002: 13). It is followed by sein in its function as copula. Even if a high token frequency can lead to relatively stable forms in the paradigm, there has always been variation and change in the paradigm of sein (cf. e. g. Koch 2004; Nübling 2000).

The main goal of this contribution is to describe and discuss variation and change in three of sein's structural aspects in Salzburg's base dialects, which can

<sup>1</sup> Note that we here—positioning ourselves in the Anglo-American tradition—equate the term 'dialect' with the term 'variety'.

be assigned to three broader Bavarian dialect regions: Central Bavarian, South Bavarian and South-Central Bavarian, a larger transition zone which is surrounded by the first two regions (see Figure 1).

Figure 1: Bavarian dialect regions in Austria according to Wiesinger (1983)<sup>2</sup>

The first feature we want to examine is the anlaut: Bavarian plural forms of sein in the area under investigation can have a h- or a s-anlaut (e. g. for the 1PL: mia han(d) vs mia san(d) 'we are'). The second feature is the stem vowel in the 2PL: Dialect speakers in Salzburg use either the monophthong /a/ or the diphthong /ai/ (e. g. es h/sadds vs es h/saidds 'you are'). The third structural aspect is morphological and concerns the relation of inflectional forms in the plural paradigm (present, indicative) ofsein. Plural verb paradigms in the Bavarian base dialects of Salzburg can be formed either as three-form plurals (1PL vs 2PL vs 3PL) or as twoform plurals (e. g. 1/3PL vs 2 PL). However, the actual picture is more complicated (see Section 2.2). As Bülow et al. (2019) and Mauser (2007) illustrate in their recent papers, speakers in this area vary between two different variants of threeform plurals and two different variants of two-form plurals (see Table 1; Section 2.2 for more details).

<sup>2</sup> The Lungau (a region of Salzburg) is highlighted because it is reported to be another transition zone which comprises the highest degree of South Bavarian dialect features in the South-Central Bavarian transition zone (cf. also Mauser 1998; Wiesinger 1983).


Table 1: Plural paradigm variants in Salzburg's base dialects according to Bülow et al. (2019)

In this chapter we examine and discuss how plural forms of sein developed in Salzburg's dialect regions over the past hundred years. Therefore, in Section 2, we report the current state of research and how it had been in the 1920/30s, as illustrated in Wiesinger (1989). We then, in Section 3, outline our methodological approach. In Section 4, we analyse the data from the 1970/80s (4.1) and compare them with the 2016/17 Salzburg dialect atlas dataset (real-time trend survey and apparent-time survey) (4.2). Furthermore, in Section 4.3, we illustrate with an indepth analysis the high amount of inter- and intra-individual variation found in the DiÖ dataset ("Deutsch in Österreich" 'German in Austria') also recorded in 2016/17.<sup>3</sup> Based on the results of these analyses, we will, in Section 5, answer our research questions and discuss the role of different forms of dialect contact regarding both inter- and intra-individual variation. We conclude the chapter with a summary (Section 6).

#### 2 Theoretical Aspects

The verb sein ('to be') is a so-called Wurzelverb and belongs to this class of special verbs (cf. Paul <sup>25</sup>2007: 279 f.). All forms of sein were and are frequently in use. Therefore, its (strong) suppletive forms in the paradigm are not surprising (cf. Nübling 2000). Perhaps because of that reason, there has always been a constant reorganization of the dialectal paradigms of sein (cf. Koch 2004; Nübling 2000; Rabanus 2008; Scheuringer 1990).

For the Bavarian dialects found in Austria, forms of sein were gathered and documented in several dialect atlas projects (e. g. Sprachatlas von Oberösterreich (SAO) 'Linguistic Atlas of Upper Austria' and Sprachatlas Salzburg 'Linguistic

<sup>3</sup> The authors gratefully acknowledge support by the Austrian Science Fund (FWF): Project SFB F60.

Atlas of Salzburg' (cf. Scheutz 2017)), historical descriptions of (archaic) base dialects (so-called Ortsgrammatiken 'local dialect grammars', e. g. Schatz 1897; Lessiak 1903), and dialect geographical overviews documenting historical dialects in the first and second decades of the 20th century (e. g. Mauser 1998; Reiffenstein 1955; Weitzenböck 1942; Wiesinger 1989). These projects, however, only covered selected areas of Austria. They were primarily descriptive, and most of them presupposed the existence of 'homogeneous' base dialects. To our knowledge, larger areal analyses for the Bavarian dialects in Austria were only carried out for a few forms of sein (cf. Mauser 2007; Wiesinger 1989 and 2004). Furthermore, there has been no comprehensive investigation of sein forms, neither in a real- nor in an apparent-time study. It is remarkable that the situation for Bavarian dialects outside Austria is not much better. Koch (2004) focusses on sein (and haben 'to have') in the Bavarian dialects of Lower Bavaria, but this is a rare example.

Paradigms of sein show phonetic and morphological variation and change on various levels. We will illustrate this by using the paradigms reported in Koch (2004: 128) and Mauser (2007: 70) who analysed the situation for Lower Bavaria (L-BAV) (Germany) and the Austrian-German (Salzburg-Bavarian) border region to the north of the city of Salzburg (SBG).


Table 2: Paradigms of sein-plurals in Lower Bavaria (L-BAV) according to Koch (2004: 128) and in the Salzburg-Bavarian border region (SBG) according to Mauser (2007: 70)

The differences between the MHG and NHG paradigms show diachronic change, whereas the differences between the NHG and the Bavarian (BAV) paradigms indicate variation on the dialect-standard continuum. The differences which are reported for the BAV dialects of Lower Bavaria and the Salzburg-Bavarian border region indicate diatopic variation (see Section 2.1).<sup>4</sup>

For the present chapter we focus on the alternation of the h/s-anlaut, the a/aivariation of the stem vowel in the 2PL, and the suffixes in the 1/3PL. As indicated in Table 2, the suffix of the 2PL seems to be stable over the dialectal paradigms. In what follows, we first concentrate on sound change and then describe the state of research on plural verb paradigms in Salzburg's base dialects.

<sup>4</sup> It partly also reflects diastratic variation (cf. Section 2.1 and 2.2).

#### 2.1 Anlaut and Stem Vowel

According to the Viennese dialectal school ('Wiener dialektologische Schule') and Wiesinger (1983), Bavarian dialects in Austria can be subdivided into three larger dialect regions (see Figure 1): Central Bavarian, South-Central Bavarian, and South Bavarian. Furthermore, some recent investigations indicate that the Central Bavarian dialect region should be further subdivided into a West-Central and an East-Central Bavarian area. Wiesinger (2004: 24) lists the anlaut variation in sein-plurals as one of the distinctive features: East-Central and South-Central Bavarian dialects have s-anlaut, whereas West-Central Bavarian dialects show h-anlaut. Scheuringer (1990: 322) mentions that the h-anlaut area forms a triangle between Munich, Nuremberg (both Germany), and Linz (Austria). This is roughly in line with Wiesinger's (2004: 22) map, which he had drawn according to map 108–111 of the DSA (Deutscher Sprachatlas) (see Figure 2).

Figure 2: Distribution of h- and s-anlaut in sein-plurals in the Bavarian dialect regions (Wiesinger 2004: 22)

Koch (cf. 2007: 392), who analysed the dialect atlas of Lower Bavaria (SNiB = Sprachatlas von Niederbayern 'Linguistic Atlas of Lower Bavaria'), shows that the h-anlaut prevails in the West-Central Bavarian dialects of Lower Bavaria (Germany).

The h-anlaut is, however, solely a base dialectal feature, whereas the s-anlaut is also used for intermediate speech-levels in the dialect-standard continuum (see Figure 3). Scheuringer (cf. 1990: 289–327; 1993: 76–77) points not only to the areal distribution of the anlaut variation but also to its (vertical) sociolinguistic dimension. He found for the city of Braunau,arather small town<sup>5</sup> at the Austrian-German border region in Upper Austria, that the h-anlaut was only used by informants from the 'Grundschicht' (mostly self-employed craftsmen) in 43 % of cases, whereas informants from the 'Mittelschicht' (civil servants, service occupations, commercial employees) and the 'höhere Schicht'(civil servants with high school graduation, academics) almost exclusively favoured the s-anlaut (cf. Scheuringer 1990: 289–327; 1993: 76–77).

Furthermore, age seems to be a relevant factor. Scheuringer's oldest informants (> 65 years) still showed h-anlaut in 35 % of the cases, the middle-aged informants (35–65 years) had 24 % h-anlaut and the youngest group (< 35 years) nearly exclusively used the s-anlaut (in 98 % of cases).

Scheuringer's (1990; 1993) and Koch's (cf. 2007: 393) surveys indicate the prevalence of s-anlaut in the urban surroundings of the West-Central Bavarian dialect region of Upper Austria and Lower Bavaria. There are, however, no recent data which show that the s-anlaut also spreads through Salzburg's rural West-Central Bavarian base dialects. Mauser (cf. 2007: 61, 70) recorded mostly h-anlaut in the northern parts of Salzburg but also refers to the vertical pressure on the hanlaut in rural areas. For him, dialect-to-standard advergence seems to be the most likely scenario regarding anlaut change. As already mentioned, the s-anlaut is used in intermediate speech levels and Standard Austrian varieties (see Figure 3). A possible development towards the s-anlaut is furthermore supported by horizontal dialect contact as the s-anlaut is prevalent in the East-Central aswell as in the South- and South-Central Bavarian dialects. Due to this vertical and horizontal pressure, we assume an on-going change from h- to s-anlaut in the West-Central Bavarian dialects of Salzburg.<sup>6</sup>

The manifestation of the stem vowel in the 2PL ofsein (present, indicative) is not a distinctive feature to separate West-Central and East-Central Bavarian dialects but it does distinguish Central from South (Central) Bavarian dialects. The SAOdata (Sprachatlas von Oberösterreich 'Linguistic Atlas of Upper Austria': Vol. II, Map 36) show exclusively a-monophthong forms (e.g. hadds / sadds) over the

<sup>5</sup> Braunau's population fluctuated between 16,000 and 17,000 inhabitants between the 1980s and 2018.

<sup>6</sup> A similar change is attested for the North Bavarian former h-anlaut areas in Bavaria (cf. Koch 2007: 392).

entirety of Upper Austria. Only in the south-west and south-east of the city of Salzburg are some forms with the /ai/-diphthong (e.g. haidds / saidds) listed. The distribution for the South Bavarian part of Salzburg (Lungau) also seems to be fairly clear. Mauser (cf. 1998: 318) only recorded /ai/-forms for the Lungau in his study. Therefore, the variation of the stem-vowel instead seems to separate the Central Bavarian dialect region (monophthong /a/) from the South-Central Bavarian transition zone and the South Bavarian dialects (diphthong /ai/).

To understand the development of the stem vowel we again have to take into account vertical dialect contact. Here, the situation is much more complex compared to the anlaut. Base dialectal /a/-forms correspond with forms that are used in Central Bavarian and Viennese vernaculars (e.g. sadds 'you are') but not with the Standard Austrian form seid ('you are') which has a diphthong. The opposite can also be said: Base dialectal /ai/-forms correspond in this particular feature with the Austrian Standard form but not with the vernacular (see Figure 3).


Figure 3: Variation in the dialect-standard continuum

Because the regiolectal form sadds ('you are') seems to prevail in all the bigger cities of eastern Austria (e. g. Vienna, Linz, St. Pölten, Salzburg), we assume that the form sadds ('you are') is more prestigious and influential in Austrian base dialects than the Austrian Standard form seids ('you are') (cf. Moosmüller 1991; Vergeiner 2019). Therefore, we predict for the 2PL of sein that the monophthong /a/ spreads into the South (Central) Bavarian dialects.

Regarding both the anlaut and stem vowel distribution, our goal is to answer the following research questions:

RQ 1: Is the s-anlaut going to replace the h-anlaut in the rural West-Central Bavarian dialects of Salzburg?

RQ 2: Is the /a/-monophthong or the /ai/-diphthong spreading in Salzburg's base dialects?

RQ 3: Are there any correlations between the developments in the anlaut and the stem vowel?

The third variable that we investigate in this chapter is morphological and concerns the plural form of sein.<sup>7</sup>

# 2.2 Plural Verbs in German Varieties

In general, three types of plural verb paradigms can be distinguished for German varieties (cf. Bülow et al. 2019; Rabanus 2008 and 2005; Schirmunski 1962 [2010]: 522): the one-form plural, the two-form plural, and the three-form plural (see Table 3).


Table 3: Types of plural paradigms in different varieties (cf. Bülow et al. 2019)<sup>8</sup>

<sup>7</sup> The suffix -TS for the 2PL has always remained clearly distinguishable from the suffixes of the 1/3PL. The morpheme {-(e)ts} (-TS) for the 2PL emerged due to an enclitic process of the personal pronoun eß (MHG ëz) with the MHG morpheme {-(e)t}. This process took place almost regularly in most Bavarian dialects. MHG {-(e)t} (-ET) is only preserved in some South Bavarian dialects (cf. Scheutz 2016: 84–85).

<sup>8</sup> As in Rabanus (2004; 2008), the capital letters -EN, -TS, and -ENT symbolize suffixes. These letters indicate sound classes which correspond to the morphemes {-en}, {-ts}, and {-end}. Depending on the stem final sound and the dialect region, their allomorphic realization can contain very different sounds.

In Standard German, for instance, we find plural paradigms with two distinct suffixes – one for the 1/3PL (-EN) and one for the 2PL (-T) ('two-form plural').<sup>9</sup> Plural verb paradigms with only a single suffix for all forms ('one-form plural'), such as in Standard (British or American) English, are characteristic of, for example, some eastern High Alemannic dialects around Zurich (cf. Rabanus 2004: 342). Plural paradigms with three different suffixes—one for each personal form—('three-form plural') are about to disappear in recent German varieties. They were representative, for example, in MHG for strong and weak verbs (1PL vs 2PL vs 3PL—EN vs -ET vs -NT) and are reported, however, to be still present in some Alemannic and Bavarian base dialects. Lessiak (cf. 1903: 203), for example, shows how three-form plurals in South Bavarian dialects are used. Wiesinger (1989) and Mauser (2007) point to three-form plurals in Salzburg's base dialects. Based on real- and apparent-time studies, Bülow et al. (2019) also found recent evidence for the use of three-form plurals in Salzburg. Their results, however, indicate that the three-form plural has lost ground over the past hundred years, whereas the use of two-form plurals has largely increased.

Following Rabanus (2008), we identify certain types of suffixes with capital letters: Suffixes with nasal and without dental of the type -EN get the letter A, suffixes for the 2PL get a B, and suffixes with nasal and dental of the type -ENTor the type -NT get a C. Thus, the typical Standard German plural verb paradigm corresponds to the notation ABA. Note, however, that the plural forms of sein in Standard German do not correspond to ABA but CBC. The suffix -NTof the form sind (1/3PL 'we/they are') is historically equivalent to MHG suffix -ENT.

As mentioned above, Bülow et al. (2019) and Mauser (2007) point out that we have to distinguish different variants of three- and two-form plurals for sein in Salzburg's base dialects (see Table 1): Variant 1 (ABC = -EN, -TS, -NT) is a threeform plural and corresponds with the MHG three-form plural (cf. Paul <sup>25</sup>2007: 240–242; Table 3). Variant 2 (CBA = -NT, -TS, -EN) is also a three-form plural, but it is in the historically reversed order compared to the MHG system. Variant 3—a two-form plural (ABA = -EN, -TS, -EN)—is structurally similar to Standard German and the regional Bavarian vernaculars (see Table 3). Variant 4 (CBC = -NT, -TS, -NT) is also a two-form plural, but it has a suffix with dental in the 1/ 3PL.

The suffix -ma for the 1PL like in mia ha-ma or sa-ma ('we are') (see Table 2) shall not be considered as these forms are more or less restricted to the Central Bavarian dialects of Bavaria (Germany) and the South Bavarian dialects of Carinthia (Austria) (cf. Rabanus 2008; Wiesinger 1989). The suffix -ma is only occasionally expected in Salzburg's border regions (cf. Scheuringer 1990: 265).

<sup>9</sup> Rabanus (cf. 2004: 345) reports on dialects in Bavarian Swabia that have the combination 1PL vs 2PL = 3PL.

In summary, there has so far been little research into the different plural paradigms of sein in Salzburg's base dialects. Therefore, we aim to answer the following research questions:

RQ 4: What types of plural paradigms are used for sein by the informants, and how do they vary?

RQ 5: Are there correlations between the use of particular plural paradigms and socio-demographic factors like age and location?

#### 3 Methods

Most dialectological fieldwork relies on apparent-time studies, implying that synchronic comparison of different age cohorts reflects the actual diachronic change (cf. Cukor-Avila / Bailey 2013). It is "the use of the present to explain the past" (Labov 1975: 825). This assumption is at best questionable, as even external linguistic factors such as age and social class interact with each other over time.

Furthermore, even with social grouping it remains unclear whether the language use of different generations reflects the same speech level within the dialect-standard continuum in a diaglossic situation. Thus, according to Labov (1966: 200) "the ideal method for the study of change is diachronic: the description of a series of cross sections in real time". Although real-time studies are more difficult to implement, its benefits are clearly visible. Real-time studies "can provide crucial data for studies of innovation, diffusion, social transmission, mechanisms of change, and many other fundamental concerns" (Chambers / Trudgill 2009: 149). On reflection, it is our considered opinion that a combined real- and apparent-time approach provides best insights into the mechanisms of language variation and change.

The following analysis is based on three different sets of data (see Figure 4). Our oldest set contains transcripts from a dialect survey from the 1970/80s that will be compared in a real-time trend study with data that were gathered in 2016/ 17 for a dialect atlas of Salzburg (cf. Scheutz 2017). With the latter dataset we also conducted an apparent-time study, as older and younger informants were recorded. Our third dataset consists of recordings whichwere carried out within the SFB-project "German in Austria" in 2016/17 (cf. Budin et al. 2019).<sup>10</sup> In the following, we use the German abbreviation DiÖ (Deutsch in Österreich) to refer to this project. These data allow an in-depth analysis of five additional locations.

<sup>10</sup> For more information, see PP02 "Variation and Change of DialectVarieties in Austria (in Real and Apparent Time (F 6002-G23)".

Figure 4: Research design and datasets

#### 3.1 Survey Procedures, Material and Stimuli

All analysed data were gathered directly by trained fieldworkers using a dialect questionnaire (Dialektfragebuch). Although those questionnaires show broad similarities, the interviews were conducted under varying circumstances: The 1970/80s survey was carried out by Herbert Tatzreiter, Werner Bauer, Franz Patocka and Hermann Scheuringer. The dialect questionnaire was completed by several informants at each location, and all answers were immediately transcribed in Theuthonista during the interview. In addition, every transcript includes comments on the informants as well as general remarks on their specific dialect, but unfortunately, audio recordings were only occasionally made. The other two surveys were conducted by two trained fieldworkers (Hannes Scheutz, Dominik Wallner), and all of these interviews were tape-recorded. Here, every informant had to answer all the given questions. This procedure lasted between two to three hours.

To provide congruent material for a better comparison, the dialect questionnaires that have been used, closely follow those of larger dialect atlas projects (cf. Sprachatlas von Niederbayern (SNiB) 'linguistic atlas of Lower Bavaria', Sprachatlas von Oberösterreich (SAO) 'linguistic atlas of Upper Austria', Deutsch in Österreich (DiÖ) 'German in Austria'). The majority of the items consists of translation tasks and cloze tests. The verb sein was always prompted within its complete paradigm but for the DiÖ-data (2016/17) we also included complete sentences and phrases to ensure a larger quantity of items but also to embed the target items in a broader syntactical context (see Table 4). With these contexts, we prove for the influence of sentence-phonetic restrictions. Furthermore, the entrenchment of this high-frequency verb into syntactical structures supports the avoidance of learning effects and distraction caused by questioning verb paradigms.


Table 4: Items and target sentences in the DiÖ-questionnaire (2016/17)<sup>11</sup>

#### 3.2 Informants and Locations

All of the informants were chosen by specific socio-demographic parameters. Though the 1970/80s survey only gathered typical NORMs / NORFs (> 65 years), the latter two studies also inquired data of a younger generation. To ensure a solid base-dialectal quality, these younger informants were characterized as young professionals (mostly of an artisanal background) under 35 years old, strongly connected to their localities, and without higher school education.

For the 1970/80s survey, interviews were conducted in 57 locations throughout Salzburg (9 Central Bavarian, 9 South Bavarian, 39 South-Central Bavarian transition zone). On average, six individuals were polled at each location to complete the questionnaire.<sup>12</sup> For the Salzburg dialect atlas project (2016/17), one

<sup>11</sup> As Bavarian dialects originally do not use simple past forms, the items No. 43, 87 and 135 are normally expressed in the past perfect tense, with an auxiliary form of to be and a past participle. Nevertheless, informants may use simple past forms in these items—in that case, they were not included in our analysis.

<sup>12</sup> Due to this method, only one person per location produced the sein-paradigm.

informant of each generation was recorded at 32 locations (9 West-Central Bavarian, 5 South Bavarian, 18 South-Central Bavarian transition zone). Accidentally, one older informant (Fuschl\_alt) did not complete the questionnaire, so we could only include 63 interviews in this dataset. The DiÖ-recordings (2016/17) encompass 20 informants. Considering gender-induced effects, we gathered data from each male and female informants of both age cohorts at five locations. Table 5 gives an overview of the number of informants and locations included in each dataset.


Table 5: Number of locations and informants in each dataset

### 4 Results

As the present contribution aims to investigate variation and change of seinplurals in Salzburg's base dialects, we will, in the following section, first analyse the data from the 1970/80s (see Section 4.1). We then compare these data in a realtime trend study with the NORMs/NORFs which were recorded for the Salzburg dialect atlas 2016/17. The dialect atlas dataset is also suitable to conduct an apparent-time study (see Section 4.2). Lastly, we present an in-depth analysis of data from 20 informants that were gathered at five locations in 2016/17. These DiÖ-data also allow us to compare two generations (see Section 4.3).

#### 4.1 Forms of sein in the 1970/80s Dataset

The 1970/80s dataset showsaclear areal distribution of the investigated parameters of sein: In the Central Bavarian dialect region, we only find the h-anlaut, while the s-anlaut was exclusively used in the South Bavarian region (see Table 6). A chi-square test reveals that the use of h- or s-anlaut significantly depends on where the informants are located within the three dialect regions of Salzburg (χ² = 22.326, n = 57, df = 2, p < .001, Cramer's V = .626).


Table 6: Distribution of h- vs s-anlaut over the dialect regions (1970/80s)

Informants from the South-Central Bavarian transition zone prefer the s-anlaut (72 %), but a clear north-south division of the transition zone should be recognized (see Figure 5). The further north the informants are located, the more likely it is that they use the h-anlaut. Note, however, that the informants from the city of Salzburg already used the s-anlaut.

Figure 5: Distribution of h- vs s-anlaut over Salzburg (1970/80s)

Regarding the stem vowel of sein (/a/ vs /ai/), we can also find a spatial distribution in the 1970/80s dataset. In the Central Bavarian area, the use of the amonophthong dominates (89 %), whereas the ai-diphthong prevails in the South-Central Bavarian transition zone (72 %) and the South Bavarian dialect region (78 %) (see Table 7). A chi-square test also demonstrates that the dialect regions



Table 7: Distribution of a- vs ai-stem vowel over the dialect regions (1970/80s)

Therefore, the use of the stem vowel strongly corresponds with the anlaut. Informants who prefer h-anlaut also prefer a-monophthong and informants who use s-anlaut seem to prefer ai-diphthong (compare Figure 5 and 6).

Figure 6: Distribution of a- vs ai-stem vowel over Salzburg (1970/80s)

Also, the distribution of the plural paradigm variants corresponds to Salzburg's dialect regions (χ² = 17.413, n = 57, df = 6, p = .008, Cramer's V = .39). Note, however, that this correlation is weaker than the correlation between the dialect regions and the two phonetic variables. Overall, the variant 4 (CBC) two-form plural dominates in Salzburg (cf. Bülow et al. 2019). It was applied in 77 % of cases and was, for example, exclusively used in the Central Bavarian dialect region, and 82 % of cases in the South-Central Bavarian transition zone. Less clear is the dissemination over the South Bavarian dialect region, where variant 1 (ABC) and variant 4 (CBC) were equally used (see Table 8).


Table 8: Distribution of the plural paradigm variants according to dialect region (1970/80s)

In sum, the 1970/80s dataset shows relatively clear spatial patterns for the distribution of the different sein plural forms. In the Central Bavarian area, we mostly find h-anlaut, a-monophthong in the second person plural and the plural paradigm variant 4 (CBC). In contrast, informants from the South Bavarian area preferred the s-anlaut and the ai-diphthong. Here the three-form plural in the form of variant1(ABC) is still present in 50 % of cases. The South-Central Bavarian transition zone shows a north-south division for the analysed features of sein. The typical patterns of Central Bavarian dialects prevail in the northern parts, whereas the South Bavarian features are present in the southern parts of the transition zone.

Regarding the anlaut distribution the results are in line with Wiesinger's (2004: 22) observations which are based on the map 108–111 of the "Deutscher Sprachatlas" (cf. Section 2.1, Figure 2).

The data, however, challenge Wiesinger's (1989) comments (based on Wenker's questionnaires and about 50 local monographies) on the distribution of the plural verb morphology. According to Wiesinger (cf. 1989: 36–50), the three-form plural in the form of variant 1 (ABC) should be prevalent in Salzburg's base dialects. This is definitely not the case for the 1970/80s dataset.<sup>13</sup> These data are more consistent with Scheuringer's (1993: 78) observation that the two-form plural in the form of variant 4 (CBC) prevails in both the West-Central Bavarian dialect region and the South-Central Bavarian transition zone.<sup>14</sup>

<sup>13</sup> For an in-depth analysis of 22 plural verbs, the reader is referred to Bülow et al. (2019).

<sup>14</sup> "Während die mittelbairischen Verkehrssprachen heute überwiegend unter standardsprachlichem Vorbild zu Gleichklang zwischen 1. und 3. Person mit Ausgleich nach dem {en}- Morphem der 1. Person tendieren, tritt in einem großen Gebiet im Westen des Mittelbairischen und Südmittelbairischen, das sich in etwa mit dem salzburgischen Einflußbereich in

Even if the plural variants of sein seem to be well distributed over the dialect regions in the 1970/80s, recent data, indicate that contact-induced change took place. Vertical language contact has already been insinuated by Scheuringer (cf. 1990: 322; 1993: 78), who states that only the most conservative dialects preserve variant 1 (ABC), thus, a three-form plural. Scheuringer (cf. 1990: 322) also applies this observation to the anlaut distribution. He assumes that the hanlaut is only used by very conservative speakers of the base dialect.<sup>15</sup>

In what follows, we compare the 1970/80s data in a real-time trend survey with recent dialect atlas data from 2016/17 to trace ongoing change. With the help of the 2016/17 dataset, we also conduct an apparent-time survey in which we compare NORMs and NORFs with younger informants.

#### 4.2 Forms of sein in the 2016/17 Salzburg Dialect Atlas Dataset

As all informants in the 1970/80s dataset were NORMs/NORFs, we can compare them in a real-time trend survey with the NORMs/NORFs of the 2016/17 Salzburg dialect atlas survey. These two groups seem to behave similarly regarding the anlaut distribution and the use of the stem vowel (see Figure 7) (anlaut: χ² = .859, n = 88, df = 1, p = .354, Cramer's V = .099; stem vowel: χ² = .185, n = 88, df = 1, p = .667, Cramer's V = .046).

Also, the spatial distribution of the anlaut and the stem vowel is similar between the two groups (compare Figure 5 and 8). What counts for the 1970/80s data is still valid for the NORMs/NORFs of the 2016/17 dataset (see Table 9): Whereas the h-anlaut dominates in the West-Central Bavarian dialect region (100 %), the s-anlaut prevails in the South Bavarian dialect region (100 %) and in the South-Central Bavarian transition zone (67 %) (χ² = 14.849, n = 31, df = 2, p = .001, Cramer's V = .692).

diesen Räumen deckt, in den beiden Personen Ausgleich zugunsten der 3. Person und ihrem {ent}-Morphem ein." 'While Central Bavarian vernaculars in general—due to standard advergence—established unsion in the flection of 1st and 3rd person plural, following the {en} morpheme of the 1st person, a large area in the western part of Central Bavarian as well as South-Central Bavarian, which roughly coincides with the Salzburg area, shows leveling within the inflection in favor of the 3rd person plural and its {ent}-morpheme.' (Scheuringer 1993: 78)

<sup>15</sup> "Wie schon erwähnt, ist die h-Lautung in ihrem ganzen Hauptverbreitungsgebiet im Dreieck München–Linz–Nürnberg als rein basisdialektal einzustufen (Freudenberg 1974, S. 752 'fernab der Verkehrssprache')." 'As mentioned before, the h-sound has its main distribution area in the triangle between Munich–Linz–Nuremberg and can be classified as a purely basedialectal feature. (Freudenberg 1974, p. 752 'Far from the vernacular').' (Scheuringer 1990: 322)

Figure 7: Anlaut and stem vowel variation in the 1970/80s and the 2016/17 dialect atlas dataset

Figure 8: Spatial distribution of the h/s-anlaut NORM/Fs 2016/17 Salzburg dialect atlas dataset

The a-monophthong still dominates in the West-Central Bavarian dialect region (62.5 %), whereas the ai-diphthong prevails in the South Bavarian dialect region (100 %) and the South-Central Bavarian transition zone (72 %) (χ² = 5.896, n = 31, df = 2, p = .052, Cramer's V = .436).


Table 9: Anlaut and stem vowel variation according to dialect region in the 2016/17 Salzburg dialect atlas dataset

However, we found a rather strong correlation between the use of plural paradigms and the two group of NORM/NORFs (χ² = 16.847, n = 85, df = 3, p = .001, Cramer's V= .45). As Table 10 illustrates, the informants from the 2016/17 dataset used variant 3 (ABA) to a much higher degree, i. e. the two-form plural that has -EN suffix for the 1/3PL. The younger informants and the NORMs/NORFs from the 2016/17 Salzburg dialect atlas dataset less often applied variant 4 (CBC) and variant1(ABC). Within the NORMs/NORFs of the 2016/17 data the distribution of the plural paradigm variants over the dialect regions is not significant (χ² = 9.245, n = 28, df = 6, p = .16, Cramer's V = .406).


Table 10: Variants of sein plural paradigms according to the 1970/80s and the 2016/17 Salzburg dialect atlas dataset

An apparent-time study reveals striking differences for all variables between the older (NORMs/NORFs) and the younger informants of the 2016/17 Salzburg dialect atlas dataset. For the anlaut, a chi-square test demonstrates a weak correlation between the age groups and the anlaut variants (χ² = 3.946, n = 63, df = 1, p = .047, Cramer's V = .25). The younger informants use more s-anlaut than the older informants (compare Figures 8 and 9). This means vice versa that the older informants use the h-anlaut more often (see Table 11). Note, however, that 8 of the 63 informants (5 old and 3 young) show intra-individual variation in using the anlaut.<sup>16</sup> These informants use both the h- and the s-anlaut in the same paradigm (e. g. san–sadds–han 'we/you/they are').

Figure 9: Spatial distribution of the h/s-anlaut younger informants 2016/17 Salzburg dialect atlas dataset


Table 11: Use of anlaut according to age-group in the 2016/17 Salzburg dialect atlas dataset

<sup>16</sup> A chi-square test reveals no significant difference between the two age groups (χ² = .648, n = 63, df = 1, p = .421, Cramer's V = .10).

Furthermore, an interesting change begins to emerge regarding the use of the stem vowel for the 2PL. We found a significant correlation between the stem vowel variants and the two age groups (χ² = 9.908, n = 63, df = 1, p = .002, Cramer's V = .397). The younger informants use the a-monophthong more frequently than the older informants. The older informants prefer the ai-diphthong (see Table 12).


Table 12: Use of the stem vowel according to age-group in the 2016/17 Salzburg dialect atlas dataset

The areal distribution of a-monophthong or ai-diphthong over the dialect regions is still significant within the younger group (χ² = 9.354, n = 32, df = 2, p = .009, Cramer's V = .541): Whereas the a-monophthong dominates in the West-Central Bavarian dialect region (67 %) and the South-Central Bavarian transition zone (89 %), the ai-diphthong still prevails in the South Bavarian dialect region (83 %) (see Table 9).

Regarding the use of the plural paradigm variants, variant 3 (ABA) is frequently used in the 2016/17 dataset. Whereas the older informants stick much more to variant 4 (CBC, 42.9 %) which has -NT suffix for the 1/3PL, the younger informants clearly prefer to use variant 3 (ABA, 71 %) which shows -EN suffix for the 1/3PL. Variant 1 (ABC) is only used by the older informants in 14.3 % of cases (see Table 10). A chi-square test demonstrates a significant correlation between the two age cohorts and the use of plural paradigm variants (χ² = 8.336, n = 59, df = 3, p = .040, Cramer's V = .38).

As mentioned above, the use of anlaut variants significantly correlates with the three dialect regions for the NORMs/NORFs of the 1970/80s and the 2016/17 dialect atlas data. Interestingly, this no longer counts for the younger informants (χ² = 5.744, n = 32, df = 2, p = .057, Cramer's V = .47). Here, the s-anlaut clearly prevails in all of Salzburg's dialect regions. For the younger group, plural paradigm variant 3 (ABA) also dominates in all of these dialect regions (χ² = 5.920, n = 31, df = 4, p = .201, Cramer's V = .31), while variant 4 (CBC) remains strong in the western parts of the South-Central Bavarian transition zone. It is, however, striking that the younger informants from the West-Central Bavarian region only use variant 3 (ABA) (see Figure 10).

Figure 10: Distribution of sein plural paradigms (younger informants 2016/17)

#### 4.3 In-depth Analysis of sein Plurals in the DiÖ-dataset

Our third dataset provides recordings from five locations: Berndorf (BD), Rußbach (RU), Hüttschlag (HÜ), Maria Alm (MA) and Lessach (LE). These locations represent the rural areas of Salzburg's five districts (Flachgau, Tennengau, Pongau, Pinzgau, and Lungau). Four informants (2 NORM/NORFs and 2 younger speakers) were recorded at each location. The verb sein was tested in two contexts. Firstly, it was surveyed in its isolated (plural) paradigm and secondly, its forms were embedded in a broader syntactic context (see Table 12). In total, 255 tokens were analysed for this dataset.<sup>17</sup>

Regarding the anlaut distribution for the NORM/NORFs the results show a clear distinction between the Central Bavarian region and the other two dialect

<sup>17</sup> Unfortunately, 25 out of 280 occurrences did not qualify for evaluation due to incorrect answers or bad audio quality.

areas. The NORM/NORFs in Berndorf (Central Bavarian) exclusively used h-anlaut in both contexts (paradigm and syntactic context) whereas all other NORM/NORFs in Rußbach, Hüttschlag, Maria Alm (all South-Central Bavarian) and Lessach (South Bavarian) used s-anlaut in both contexts. The apparent-time effects towards the use of the s-anlaut are also confirmed in this dataset. Only one younger informant from Berndorf used the h-anlaut in three cases (see Table 11).<sup>18</sup> All other informants invariably realized s-anlaut in both contexts, giving evidence that the tendency towards s-anlaut is strongly prevalent in this area.

Figure 11: Use of anlaut according to location and age

Furthermore, the stem vowel of the 2PL seems to be strikingly stable. Informants from Berndorf and Maria Alm only use the monophthong /a/. In Lessach, Hüttschlag and Rußbach, the diphthong /ai/ prevails.

The DiÖ-data also reveal variation for the realization of the stem vowel of the 1/3PL. Informants used either an a-monophthong (e. g. mi(a) san), an emonophthong (e. g. mi(a) sen) or a diphthong (e. g. mi(a) sain).<sup>19</sup> This variation is restricted to the South Bavarian informants from Lessach and the syntactic context (see Table 13).

<sup>18</sup> BD\_male\_young uses the h-Anlaut three times for 3PL when embedded in syntactic structures (Item No. 82, 135, 533).

<sup>19</sup> Further evidence for the use of the e-monophthong and the ai-diphthong in sein plural forms can be found in Mauser (1998).


Table 13: Variation of sein plural forms in Lessach in the 2016/17 DiÖ-dataset (sentence context)

The youngerinformants always applied e-monophthong forthe 1PL, whereasthe NORM/NORFs varied between all three forms. Both older informants showed intra-individual variation: LE\_fem\_old (NORF) varied between e- and a-monophthong, LE\_male\_old (NORM) used both ai-diphthong and e-monophthong. The same three variants are also used in both age groups when forming the 3PL, showing a remarkable amount of intra-individual variation (see Table 14).


Table 14: Use of the stem vowel for 3PL in Lessach in the 2016/17 DiÖ-dataset

The use of the stem vowel is not related to certain items or syntactical contexts; in other words, there are no coincidences. What becomes clear is that the amonophthong, which is dominant in all other dialect regions, plays a minor role in the South Bavarian dialects (only five realizations).

Regarding the realization of the plural paradigms, we also found a considerable amount of intra-individual variation, particularly for the syntactic context. Here, every informant showed intra-individual variation using either -EN or -NT suffix to form the 1/3PL (see Table 15). We found less intra-individual variation in Berndorf and Lessach insofar that six informants (the two NORM/ NORFs from Berndorf and all four informants from Lessach) showed consistency in using only one suffix for the 1PL. As displayed in Table 15, for example, the informants LE\_female\_old and LE\_male\_old consistently used -EN suffix to form the 1PL. This result indicates for the conservative South Bavarian dialect area a higher degree of intra-individual stability. Informants from Lessach only varied suffixes for the 3PL, although -EN forms (female: 6, male: 8 times) are preferred to -NT (female: 2, male: 1 times). The same accounts for the West-Central Bavarian dialects in Salzburg (e. g. BD\_female\_old), whereas the South-Central Bavarian dialects exhibit the highest amount of intra-individual variation: For example, informant HÜ\_female\_old shows a clear tendency towards -NT suffix for the 1PL and 3PL but also used -EN suffix for both personal forms. Thus, informant HÜ\_female\_old seems to vary between variant 1 (ABC), the three-form plural, and the two-form plural variants 3 (ABA) and 4 (CBC). These dynamics might be considered as further indication of ongoing language change in this region.


Table 15: Individual realizations of -EN and -NT suffix forms for the 1/3PL of sein

To sum up, the data show that the s-anlaut is spreading into the West-Central Bavarian dialects in Austria while the a-stem vowel (2PL) is spreading into the South-Central Bavarian dialects. Furthermore, the two-form plural variant 3 (ABA) is spreading. The high amount of intra-individual variation in the data, however, indicates that the change is still ongoing.

# 5 Discussion

Section 4 has provided variation and clear trends in the data. To answer our first research question, the real-time trend survey (Section 4.1 and 4.2) indicates that the s-anlaut will replace the h-anlaut in the rural West-Central Bavarian dialects of Salzburg. Whereas we found a clear pattern of regional distribution in the 1970/80s data (see Figure 5), the younger informants recorded in 2016/17 (dialect atlas data and DiÖ-data) clearly preferred the s-anlaut across Salzburg. The hanlaut is only used in a few cases by older informants (see Section 4.2 and 4.3). This means that the s-anlaut spreads into the West-Central Bavarian dialects. Therefore, this feature seems to lose its status to clearly separate West- from East-Central Bavarian dialects in Austria. It would be interesting to see whether or not this development stops at the Austrian-German (Salzburg-Bavarian) or Salzburg-Upper Austrian border. It is well known that political borders influence the dialect landscape in many European regions (cf. Auer et al. 2015; Auer 2018: 163– 164 and 2004).

To answer research question two, the a-monophthong seems to replace the aidiphthong in South-Central and South Bavarian dialects. Even in the South Bavarian Lungau, the younger informants prefer /a/—over /ai/-forms.

As a consequence, we find correlations between the developments in the anlaut and the stem vowel for the 2PL. To answer research question three, in the 1970/80s dataset h-anlaut appears very frequently with a-monophthong, whereas s-anlaut is more likely to appear with ai-diphthong. This seems to change over the years. For the younger generation, s-anlaut is now very likely followed by amonophthong. Thus, the most common form for the 2PL in the 2016/17 Salzburg dialect atlas dataset is sadds ('you are').

In simple terms, both developments indicate that (East-)Central Bavarian dialect features are spreading into Salzburg's base dialects.<sup>20</sup> These dialect features very often correspond with features that prevail in the regional vernaculars. That means that we have to consider any convergence between both the different dialect areas (horizontal dialect contact) and the dialect-to-standard advergence (vertical dialect contact). In particular, the spread of sadds forms (2PL 'you are') illustrates the importance of the Viennese-influenced regional vernaculars (cf. Moosmüller 1991: 50 f.; Vergeiner 2019: 156).<sup>21</sup>

Regarding research questions four and five, a very similar development holds for the plural paradigm. We found a clear tendency towards a variant that corresponds structurally (ABA two-form plural = -EN, -TS, -EN) to a variant that is well established in the regional Bavarian vernaculars (see Table 3). Whereas the two-form plural variant4(CBC) dominates in the 1970/80s dataset (except in the South Bavarian region, see Section 4.1), the two-form plural variant 3 (ABA) prevails in the 2016/17 datasets, particularly among the younger informants (see Section 4.2 and 4.3). The in-depth analysis (see Section 4.3), however, reveals that this change is far from complete. In addition to the trend towards the two-form

<sup>20</sup> Note, however, that the s-anlaut is also dominant in the South and South-Central Bavarian dialect regions, whereas the /a/-monophthong is a general Central Bavarian dialect feature.

<sup>21</sup> This example underlines the diaglossic situation in the Bavarian parts of Austria which is characterised by a dialect-standard continuum with intermediate forms (cf. Auer 2005: 22; Scheutz 1999: 108; Wiesinger 2014: 76–84).

plural variant 3 (ABA) we found a considerable amount of inter- and intraindividual variation. The three-form plural variant 1 (ABC), for example, is not only used for sein-plurals all over Salzburg's base dialects but also other plural verbs (cf. Bülow et al. 2019). It is, however, most commonly used in the South Bavarian dialects. Furthermore, Bülow et al. (2019) point out that, particularly in the older generation, all forms were used variably, and it is frequencies that change over time.

#### 6 Conclusion

The studies of sein-plurals in real- and apparent-time clearly show that the sanlaut, the a-monophthong (2PL) and the two-form plural variant 3 (ABA) are becoming more prevalent among younger speakers and are replacing the hanlaut, the ai-diphthong (2PL), the three-form plural variant 1 (ABC) and the two-form plural variant 4 (CBC). Those forms are associated with older speakers or, in the cases of the ai-stem vowel and the three-form plural (ABC), with speakers from the South Bavarian dialect region. The conclusion to be drawn from this is, that the base dialectal features of sein become more similar with both regional Bavarian vernaculars (dialect-to-standard advergence), which are strongly influenced by Viennese colloquial language and with East-Central Bavarian base dialects (convergence), which themselves are linguistically influenced by Vienna. Even if the data show clear trends (Sections 4.1 and 4.2), we found a large degree of inter- and intra-individual variation to indicate ongoing change (Section 4.3).

In sum, we attest for base dialectal sein-plurals a development towards regiolectal forms. We explain this process as leveling through the loss of base dialectal features. "Traditional dialect features are being replaced by regiolectal ones, and these in turn by near-standard ones." (Auer 2018: 169) This movement involves a continuous process of horizontal and vertical dialect contact-induced change. Therefore, dialect leveling in Austria almost seems to be in line with a general trend that is apparent in many European dialect-standard constellations (cf. Auer 2018). For Austria (except Vorarlberg), we can observe an ongoing emergence of more regional dialects.

#### Bibliography

Auer, Peter. 2005. Europe's Sociolinguistic Unity, or: A Typology of European Dialect / Standard Constellations. In: Nicole Delbecque / Johan van der Auwera / Dirk Geeraerts (eds.): Perspectives on Variation. Berlin: de Gruyter, 7–42.


zur deutschen und allgemeinen Dialektforschung. Vol. 2. Berlin / New York: de Gruyter, 807–900.


Dr. Lars Bülow University of Vienna Germanistik Universitätsring 1 1010 Vienna Austria lars.buelow@univie.ac.at

Mag. Dominik Wallner University of Salzburg SFB 'German in Austria' Erzabt-Klotz-Straße 1 5020 Salzburg Austria dominik.wallner@sbg.ac.at

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Jutta Ransmayr

# Classroom Talk in Austria. Standard and Non-standard Varieties and Language Preferences in German Lessons

Abstract: Until recently, hardly any data regarding which varieties—from Standard German to dialect—Austrian teachers and students use in classroom talk during German lessons were available. For the first time, a large-scale study investigated how teachers and students talk during German lessons and to what degree code-switching and code-mixing take place in the Austrian classroom. Concerning the standard variety, Austrian Standard German has been reported to be strongly influenced by German Standard German, especially among the younger generation. If and how generational language preferences are present in Austria among students and teachers was another issue that was investigated for the first time. The respective results will also be outlined in this contribution.

Keywords: Austrian German, classroom talk, dialect-standard continuum, codeswitching and code-mixing, generation-specific language preference.

Abstract: Bisher gab es keine fundierten Daten dazu, welche Varietäten – vom Standard bis zum Dialekt – österreichische DeutschlehrerInnen und SchülerInnen in der Unterrichtskommunikation verwenden. In einer breit angelegten empirischen Studie wurde erstmals untersucht, wie diese während der Deutschstunde sprechen und in welchem Ausmaß es dabei zu Code-Switching- und Code-Mixing-Phänomenen kommt. Was zudem die Ebene der Standardsprache betrifft, so unterliegt österreichisches Standarddeutsch Beobachtungen zufolge einem starken Einfluss des bundesdeutschen Deutsch, besonders unter der jüngeren Generation. Ob und inwiefern es dabei zu generationsspezifischen Sprachpräferenzen in Österreich bei SchülerInnen und LehrerInnen kommt, wurde in der angesprochenen Studie ebenfalls erstmals untersucht. Die entsprechenden Ergebnisse werden in diesem Beitrag thematisiert.

Keywords: österreichisches Deutsch, Unterrichtssprache, Dialekt-Standard-Kontinuum, Code-Switching und Code-Mixing, generationsspezifische Sprachpräferenzen

#### 1 Introduction

The first aim of this article is to provide an insight into patterns of language use during German lessons at Austrian schools. Until recently, not much had been known about which varieties—from Standard German to dialect—Austrian teachers and students use in classroom talk during German lessons. In the study presented in this article, data on language use and language attitudes in Austrian schools were gathered and analysed for the first time on a large-scale empirical basis.

The second aim of this article is to present initial results on generational language preferences in Austria according to teachers' and students' self-reported language use. Austrian Standard German has been reported to be strongly influenced by German Standard German in recent years. Any indications of language change taking place in Austria were investigated for the first time in the study presented in this article.

The role of the German language in German lessons in Austrian schools is a complex matter and there are many reasons for this. For one, the German language functions above all as the medium of instruction and education in Austrian schools. At the same time, the German language in German lessons is the object of study itself, the school subject taught and studied (Steinig / Huneke 2015). The fact that German consists of numerous varieties adds another layer to the picture: German is presumed to be one of the richest languages regarding variation among European languages (Barbour / Stevenson 1998). Variation comprises standard varieties as well as non-standard varieties. The resulting in inner or language-internal multilingualism<sup>1</sup> , encompassing the whole spectrum from standard at the one end to dialect at the other, is characteristic of Austria (Ammon 2005; Löffler 2005; Muhr 2013; Steinegger 1998; Wiesinger 1985 and 2010). Therefore, one can presume that inner multilingualism plays a role in class just as it does outside of school. Moreover, there is not just one standard variety in German. In the majority of relevant linguistic publications (Ammon 1995 and 2005; Ammon et al. 2004; Ammon et al. 2016; Clyne 1992, 1995 and 2005; Kellermeier-Rehbein 2014; Schmidlin 2011 to mention a few), German is conceptualised as a pluricentric language—just like many other languages, such as English—when describing the standard language, consisting of three main standard varieties: German Standard German (GSG), Austrian Standard German (ASG) and Swiss Standard German (SSG). Recently, GSG has been noted to gain influence on ASG and this consequently plays a growing role in the realm of language awareness in the German classroom (de Cillia / Ransmayr 2019). To add

<sup>1</sup> On inner multilingualism, see: Busch 2013; Wandruszka 1979. Some publications also use the term 'language-internal multilingualism', f.ex. Reich / Krumm 2011.

yet another factor to the rather complex teaching and learning situation in German classrooms in Austria, despite being the first language (L1) for the majority of pupils, German is L2 or L3 for a significant number of pupils: According to school statistics in 2013<sup>2</sup> , between 14 and 15 % of students at upper secondary level spoke languages other than German as their first language; in primary schools the number was above 50 % for Vienna and around 26 % for all of Austria (cf. de Cillia 2010: 10).

This article explores German language use in German lessons from various perspectives: To begin, the degree of multilingualism present in the Austrian classroom will be outlined. This will lead us to the dialect-standard continuum and its presence in daily classroom communication. We will present examples of how Austrian speakers in Austrian schools, students and teachers alike, switch light-footedly between dialectal utterances or colloquial Austrian German and ASG. In addition, we will address the question of whether speakers are aware of their code-switching and code-mixing.

Then, this article will focus on language change concerning ASG: Results on preferential, self-reported use of ASG or GSG among Austrian students and teachers will be presented, and the influence of the factors 'age' and 'television consumption' on 'language loyalty' towards ASG will be discussed.

#### 2 Austrian German: A Brief Overview

Austria's official language is German, as stated in article 8, paragraph 1 of Austria's Federal Constitutional Law. According to the last census (carried out in 2001), 95.5 % of Austrian citizens stated that their first language was German.

In Austria, both the standard language and dialects are and can be used in many domains, and there are numerous 'in-between' forms between these two, which are usually referred to as colloquial German (Umgangssprache) (Ammon 2003: 166–168). The model most commonly used to describe linguistic German variation in Austria is that of the so-called 'dialect-standard continuum'. This model differentiates between Austrian Standard German, transregional colloquial language (Umgangssprache) and dialect. Within this model, a three- or four-level polyglossia encompassing dialect (Basisdialekt, Verkehrsdialekt), colloquial language and standard language is assumed (Steinegger 1998; Wiesinger 2006 / 2014). Within this continuum, smooth shifting and switching between these levels is possible and quite common. Although a precise description of the colloquial language (Umgangssprache) is problematic, Ender/ Kaiser (2009) have shown that Umgangssprache as a variety between dialect and standard is regarded

<sup>2</sup> Statistik Austria: migration & integration. zahlen.daten.indikatoren 2013.

as important by Austrian speakers, thus legitimising the three-level modelling including Umgangssprache.

As stated in the introduction, one of the topics to be covered in this article, apart from switching phenomena within the dialect-standard continuum in the realm of Austrian classroom communication, will be language contact between Austria and Germany and their respective standard varieties (ASG and GSG), as observed among school teachers and school students. In the literature there are references made to ongoing language change in Austria that indicate ASG is strongly influenced by GSG. It is often assumed that this could be due to the influence of the German media, predominantly German TV-channels, on Austrians' linguistic behaviour and their 'language loyalty' – or more precisely their loyalty towards the Austrian standard variety (Ebner 2008; Muhr 2003; Pohl 2011). Another reason might be that Germans have been the biggest group of immigrants to Austria since 2009<sup>3</sup> , showing a notable presence in Austria, especially in the big cities, with the effect that GSG can be frequently heard not only on TV or on YouTube, but also in everyday life.

#### 3 State of Research

In the article at hand, both the dialectal as well as the standard level of German in Austrian schools will be taken into account. Concerning the standard language, German has been conceptualised as a 'pluricentric language' for nearly three decades.<sup>4</sup> When it comes to teaching German as a foreign language (GFL), the pluricentric concept has been included in the development of teaching material and in GFL-teacher-training for many years and can be seen as widely established in the field. However, not much has been known about the approach taken in Austrian schools in the school subject German (as a first language). A study by Davies (2017) looked into the question of whether teachers in North Rhine-Westphalia in Germany were familiar with the pluricentric concept and came to the conclusion that they had hardly ever heard of it and found it irrelevant for their teaching practise – a finding which confirms the asymmetry between Dnations<sup>5</sup> and O-nations<sup>6</sup> (cf. Clyne 2005: 297): Speakers of aD-nation have trouble understanding the concept of pluricentricity. A study by Pfrehm (2007; 2011), in which he investigated the differences in attitudes among Austrians and Germans

<sup>3 &</sup>lt;https://www.statistik.at/web\_de/statistiken/menschen\_und\_gesellschaft/bevoelkerung/ bevoelkerungsstruktur/bevoelkerung\_nach\_staatsangehoerigkeit\_geburtsland/index.html> (August 19, 2018).

<sup>4</sup> For the term 'pluricentric language' see Chapter 4.

<sup>5</sup> D-nation: linguistically dominant nation (Germany).

<sup>6</sup> O-nation: linguistically non-dominant nation (Austria, Switzerland).

towards (standard) expressions which are known and used on both sides of the Austrian–German border, also confirmed this asymmetry: Pfrehm was able to show that the Austrian respondents ranked these (standard) expressions as being standard language (e. g. heuer 'this year'), while the German respondents classified them as substandard.

As for attitudes towards the standard varieties of German, and especially towards ASG in non-German-speaking countries, studies have shown that ASG is often regarded as dialectal and non-standard (Markhardt 2005; Ransmayr 2006).

Apart from references being made towards ambivalent attitudes and insecurity among Austrians concerning their linguistic self-confidence (Clyne 1995; Muhr 2005), as well as assumed exonorm-orientation among German teachers in Austria (Ammon 1995; Heinrich 2010; Legenstein 2008), no largescale studies exploring what is actually happening in the area of Austrian education (teacher training, curricula, text books, language attitudes, the role of dialectal and standard varieties in school, varieties used by German teachers and students, language change etc.) have been carried out. All this occurs, despite the fact that in German didactics the topic of language variation is—among many other things—regarded as an important one by educationalists, such as Neuland / Peschel, who stress that both the topic of language norms, which are fundamental for German teaching, as well as the topic of language variation offer "a linking point not only to the learning field 'language reflection', but for all fundamental learning fields of language teaching." 7 (Neuland / Peschel 2013: 211; translation: JR). Steinig / Huneke also emphasise the importance of language awareness and knowledge about varieties for German teachers because teachers are above all linguistic norm authorities (Sprachnormautoritäten): "[…] a German teacher has to be able to provide comprehensible reasons for the assessment of oral or written utterances of his students." 8 (Steinig / Huneke 2015: 18; translation: JR).

A study on language use at schools in Bavaria, North Rhine-Westphalia and Mecklenburg-Vorpommern was carried out by Hochholzer (2004), which investigated dialect use in schools in Germany. A more recent study in Switzerland (Steiner 2008) provided insight into dialect and Standard German use during class in Swiss schools. For Austria, apart from a small field study (Rastner 1997), no comparable data had been available, which was one of the reasons for the development of a research project on the questions mentioned in this chapter. The research project "Austrian German as a language of instruction and edu-

<sup>7</sup> "Anknüpfungspunkte nicht nur für den Lernbereich 'Reflexion über Sprache', sondern für alle grundlegenden Lernbereiche des Sprachunterrichts".

<sup>8</sup> "[…] ein Deutschlehrer muss fähig sein, nachvollziehbare Begründungen für die Bewertung mündlicher und schriftlicher Äußerungen seiner Schüler geben zu könne".

cation" (de Cillia / Ransmayr 2019) was the first project to fill this research gap with a large-scale empirical study.

#### 4 Theoretical Principles and Research Questions

As for the underlying variational linguistic theoretical approach, the pluricentric concept formed the foundation of the research project "Austrian German as a language of instruction and education" 9 , from which the results presented in this article are derived. The pluricentric concept can be used to describe a standard language which has official status in more than one national centres (Germany, Austria and Switzerland being 'full centres' of German), following Ammon's terminology (Ammon 1995). According to the concept of pluricentric languages, on the level of the standard language German consists of three equivalent, statebound varieties: Austrian Standard German, German Standard German and Swiss Standard German (Ammon 1995 and 2005; Ammon et al. 2004; Ammon et al. 2016; Clyne 1995 and 2005; Kellermeier-Rehbein 2014). Ever since the publication of the Variantenwörterbuch (Ammon et al. 2004; Ammon et al. 2016) the assumption of three national standard varieties has its set place in the variational linguistic research discourse.

This model of description has proved useful for describing the level of the standard language, especially when it comes to normatively oriented contexts such as administration, school and education, where the underlying circumstances are defined by the state (e. g. curricula for schools and universities, teaching material). Another variational linguistic approach, with regard to regional or 'areal' variation on the basis of dialect regions as opposed to looking only at formal standard language, would be to describe German as being 'pluriareal' <sup>10</sup>. The theory of pluricentricity and the model of pluriareality<sup>11</sup> are different approaches to describe German, but do not contradict each other per se and can be seen as compatible with each other, as stated by Ammon already back in 1998: A "characterisation of the German language from different angles" <sup>12</sup> is possible (Ammon 1998: 320; translation: JR).

Some research questions for findings presented in this paper were based on the assumption of German being a pluricentric language on the standard level. These questions were:

<sup>9 &</sup>lt;https://oesterreichisches-deutsch.bildungssprache.univie.ac.at/home/>.

<sup>10</sup> For a critical discussion of the plurareal concept see Dollinger (in press [2019]).

<sup>11</sup> For recent publications on the pluriareal concept see f.ex. Dürscheid / Elspaß / Ziegler (2015); Niehaus (2017).

<sup>12</sup> "nach Blickwinkeln differenzierte Charakterisierung des Deutschen".


This article will also present results concerning the use of dialect and standard language and the degree of code-switching and code-mixing found in the German classroom. As with Auer / Eastman (2010), Gumperz (1982) and Heller (1988), among others, 'code-switching' in this article is understood to be a phenomenon of language contact, in which more than one language or variety, which can also apply to a dialect, are used by a speaker in a conversation. Auer / Eastman claim that the often-found additional terminological distinction between code-switching and code-mixing can be misleading (Auer / Eastman 2010: 86): Instead of the conventional distinction between code-mixing being used to refer to sentence-internal switching and code-switching being "reserved for cases of language alternation within the independent syntactic units", Auer / Eastman propagate a distinction between the two terms with a focus on the meaningfulness of the codes to the participants. Kaiser (cf. 2006: 278) points out that for the context of a dialect-standard continuum the additional term 'code-shifting' seems appropriate, as it describes the gradual transition between varieties, as opposed to code-switching, which describes a sudden change of code. In this article, code-mixing is used for intrasentential changes of code, whereas codeswitching is used referring to intersentential switches.

Based on the assumption that both teachers and students bring their inner multilingualism with them to the classroom and switch back and forth between dialect and standard and may even mix the two, we tried to find out the following:


#### 5 Methodology and Corpus

The FWF-funded research project "Austrian German as a language of instruction and education" <sup>13</sup> was carried out between September 2012 and April 2015 at the University of Vienna. Using methodological triangulation, this project investigated the role of Austrian German and other varieties of German in the context of German school teaching in Austria by creating different corpora and consequently varying research angles. In one project module, we analysed documents relevant for school teaching: All school curricula for German, from primary to upper secondary level,were analysed, as well as teacher training curricula and the most commonly used German school books. In addition, in the second project module we employed quantitative and qualitative means of research: An Austrian-wide survey looked into the language attitudes and conceptualisations of German among German teachers (of all school types, n = 165) and upper secondary school students (n = 1253). Furthermore, we carried out 21 interviews with German teachers, two group discussions with teachers and students for communicative validation of the survey results and seven participatory classroom observations. We analysed the survey data using statistical means (SPSS<sup>14</sup>), tests employed were the chi<sup>2</sup> -test, non parametric tests (Mann-Whitney-U-test, Kruskal-Wallis-test), comparison of means (T-test, analysis of variance). The interviews and group discussions were recorded, transliterated (following HIAT transcription guidelines<sup>15</sup>) and analysed using discourse analysis. During participatory classroom observation, the observer took notes following a structured analysation grid—filming and / or recording the lessons were unfortunately not permitted by school authorities.

# 6 Results on Linguistic Variety and Language Preference in the German Classroom in Austria

Before investigating attitudes and conceptualisations regarding varieties of German among teachers and students, we tried to get a picture of the linguistic variety present in everyday classroom life. In our survey, we asked German teachers of all school types about the languages and varieties they deal with in their classrooms. Answers to choose from were "dialects", "colloquial language"

<sup>13 &</sup>lt;https://oesterreichisches-deutsch.bildungssprache.univie.ac.at/home/>; de Cillia / Ransmayr (in press [2019]).

<sup>14</sup> Statistical Package for the Social Sciences.

<sup>15 &</sup>lt;https://exmaralda.org/wp-content/uploads/2015/12/General\_Transcription\_conventions\_ HIAT\_Handout\_DE.pdf>.

(Umgangssprache), "Austrian Standard German" (österreichisch geprägtes Standarddeutsch), "German Standard German" (bundesdeutsch geprägtes Standarddeutsch), "Swiss Standard German" (schweizerisch geprägtes Standarddeutsch), other languages of origin than German (andere Herkunftssprachen als Deutsch).<sup>16</sup> Figure 1 shows the teachers' estimations: The variety which was mentioned by 100 % of our probands was colloquial language, indicating that all German teachers in our survey noticed some degree of colloquial language use by students in classroom situations. The overwhelming majority of teachers also chose to pick the answersAustrian Standard German (96.2 %) and dialects (93.8 %). While a significantly smaller percentage of teachers claimed to notice some degree of German Standard German use among their pupils (44 % in total, of which 40.8 % stated "a little" and 3.2 % "frequently"), hardly anyone had observed Swiss Standard German in the classroom (3.8 %). A total of 87.9 % of teachers stated that their students brought other languages than German to the classroom (as shown in Figure 1). The picture we get, in terms of variety within German and its representation in a German classroom in Austria, clearly points to the prevalence of the dialect-standard continuum in the form of a nearly equal omnipresence of colloquial German, dialects and Austrian Standard German, the latter being slightly challenged by German Standard German.

Figure 1: Students' languages and varieties according to teachers' estimations; multiple answers permitted

<sup>16</sup> Multiple answers to this question were permitted. In the survey, teachers were given the option of differentiating their answers concerning each variety between "frequently", "a little" and "not at all". As this article aims to introduce a general picture, both positive answers "frequently" and "a little" were added up to form one group.

#### 6.1 The Dialect-standard Continuum in German Lessons among Austrian Students and Teachers

In substantial parts of the interviews conducted with teachers and of the group discussions with teachers and students, as well as in the survey, we collected data on the questions of which variety / varieties are primarily used by teachers while teaching, if they switch between varieties during classroom interaction, and if they do, whether their students are aware of their teachers' switching back and forth and the contexts in which this might occur.

#### 6.1.1 Students' Dialect Use during German Lessons and Teachers' Reactions

In the survey, we asked teachers about their estimations concerning the extent of students' dialect use in the following contexts: (1) subject-related classroom communication between teacher and students, (2) group work and (3) giving a presentation. The given options for answers were "nearly all of them / many" (use dialect), "a few / nobody" (use/s dialect). Figure 2 shows that the number of students who use dialect for giving a presentation is rather small, at 11 %, while nearly all or many students stick to Standard German when giving a presentation (88 %) according to the teachers' estimations. When it comes to doing group work, a clear majority of teachers (74 %) state that nearly all / many students use dialect in this work setting. As for subject-related classroom communication, numbers of dialect users and non-dialect users are equal.

Figure 2: Amount of students' dialect use during German lessons (teachers' estimations)

Bearing in mind that it is laid down in the Austrian school curriculum for German that students are to be guided towards a competent use of the standard language, we were consequently interested in the teachers' reactions to their students' choice of variety. In the survey, we asked teachers what they usually did when dialect use occured in the three situations mentioned above. Figure 3 illustrates that teachers stated to be the strictest with their students when it came to giving presentations: Nearly 90 % said that they interfered with student dialect use by correcting the student and / or pointing out that dialect use was not appropriate during presentations. In the group discussion, teachers brought up the fact that for many students it can be quite challenging to give a talk in Standard German. This teacher states:<sup>17</sup>

"When it comes to giving a presentation, […] they giggle a lot and feel ridiculous or, uhm, weird, or alienated, but the second time it already works much better. It's as if a switchwas switched on and they try, uhm, really to speak in some formal, uhm, standard language, uhm, well, or in a formal colloquial language. I have realized that it simply takes courage for the students. It also has something to do with habits. As soon as it comes to chatting or telling stories, they fall back into dialect, the language they feel at home in." (translation: JR)<sup>18</sup>

When studentswork in groups and use dialect to communicate with each other in order to fulfill a given task, teachers tend not to interfere by interrupting the students: Only 15 % of teachers stated they would correct students or draw their attention to the 'appropriate' use of language, while the vast majority of teachers said they would not step in (85 %). Matters are slightly different when teachers and students communicate in open classroom interaction. The majority of teachers clearly favour standard use, and the percentage of teachers who would not let dialect use pass uncommented in open classroom communication is clearly higher than in the group work setting: 61 % of teachers said they insisted on the use of the standard language and would correct students' dialectal utterances, 39 % did not mind their students using dialect in this setting (see Figure 3).

<sup>17</sup> HIAT guidelines were used for transliterating the group discussions and the interviews.

<sup>18</sup> "Wenns heißt, ja, es ist ein Referat zu halten […] dann kicherns herum und fühlen sich lächerlich ode:r, äh, seltsam, ode:r befremdlich, beim zweiten Mal gehts dann schon. Da wird der Schalter umgelegt und dann wird schon versucht, ä::hm::, wirkli in:, in aner gehobenen, ah, Standardsprache, ah/ eben, also, oder gehobenen Umgangssprache zu: sprechen, u::nd, ah, da hab i festgestellt, dass es für die Schüler anfoch a Überwindung/ es ist a Gewohnheitseffekt auch, a::hm:, wobei, wenns zum Plaudern kommt oder wenns ins Erzählen kommt, sie sofort dann wieder in den Dialekt hineinfallen. In die Sprache einfach, in der sie sich zuhause fühlen."

Figure 3: Teachers' reactions to students' dialect use

#### 6.1.2 Teachers' Use of Varieties during Class

In the survey, we asked teachers for a rough estimation of their choice of variety (Standard German<sup>19</sup>, colloquial German, dialect, other answer)<sup>20</sup> when teaching and communicating with students in German class. Teachers stated that they hardly used any dialect when speaking with and to students: 41.5 % said they used predominantly Standard German, 20.7 % stated they spoke colloquial German with students, and only 1.2 % said they used dialect. When asked to reflect on their choice of variety in different teaching situations (explaining and presenting new content; givingatask; disciplinary matters; organisational matters), a more differentiated picture evolved. None of the teachers stated that they used dialect for explaining or presenting new content—85 % said they strictly spoke in Standard German. The survey yielded similar results for the language used when explaining to the students what to do in a specific task: Nearly two thirds (65.6 %) claimed they did this in Standard German, 23.1 % said they used colloquial language. However, when it came to disciplinary matters, the picture changed and Standard German was no longer the predominantly used teacher-variety: 44.7 % of teachers said they used colloquial German in

<sup>19</sup> In order to use terms for linguistic varieties which both linguistic experts (German teachers) and linguistic lay persons (school students) would be familiar with, we offered the terms "Standard German" or "Standard German / High German" in all parts of the teacher and student questionnaires when referring to the standard language.

<sup>20</sup> Multiple answers were permitted.

such occasions, 6.9 % said they used dialect, and merely 33.3 % claimed they spoke Standard German when there were disciplinary issues. As for organisational matters, both Standard German (44.1 %) and colloquial German (36.6 %) were mentioned by teachers, and only a marginal group of teachers (3.7 %) said they used dialect. What becomes clear when considering these results is that German teachers seem to regard themselves as mostly Standard-German-oriented, partly colloquial-language-speaking teachers, who hardly use dialect during German lessons. In the interviews, matters concerning dialect were not as clear-cut: Numerous teachers referred to dialect as the appropriate means of addressing misbehaving students, and made a clear distinction between more formal, neutral subject-related conversations and emotional moments when they switched to a dialectal variety. The following passage illustrated this: "Yes, well, when in class I use formal colloquial language. And only when it gets emotional, when I have to scold someone <laughs>, then uhm it is very clear that it is in dialect." (translation: JR)<sup>21</sup>

Teachers also point out the function of the standard language as a language or variety of distance, and the 'closer', less distanced dialectal varieties and their effectiveness in certain classroom situations: "Well, actually I speak, I speak standard language with the children, but not when a student is throwing a pencil through the classroom for the tenth time. Then I come into his language, which he is used to." (translation: JR)<sup>22</sup>

What also emerged during the interviews, however, is that some teachers seem to have difficulty putting a term on the language or variety they use. This could be either because the process of consciously reflecting their use of varieties made them aware of their continuous switching between standard, colloquial language and dialect, or because they were not familiar (enough) with the appropriate terms. This example of a teacher from Salzburg<sup>23</sup> thinking out loud about his classroom language shows how 'multi-coloured' the picture often is:

"In my case it is a mix of a trace of Upper Austrian dialect, and of course as a German teacher you have this specific role and one has of course also this connection to dialect, one is somehow right in between. Yes? – - Yes, so now/ no Lungauer dialect. As you can

<sup>21</sup> "Ja. Also im Unterricht verwende ich schon gehobene Umgangssprache. Und nur, wenns emotional wird, also wenn ich schimpf <zum Beispiel (lachend)>, dann ah is des/is des gonz kloar, doss i des/des im Dialekt."

<sup>22</sup> "Na ja, ich spreche, ich sprech mit den Kindern immer in der Standardsprache, a::ber ich breche sie dann, wenn ein Schüler zum zehnten Mol mir, wos was i, den Bleistift durch die Gegend schleudert. Dann komm ich in seine Sprache, die er gewohnt ist."

<sup>23</sup> What makes this teacher's use of language particularly interesting is the fact that he/she teaches in the federal state of Salzburg in the Lungau district, but is originally from Upper Austria. Upper Austrian dialect and Lungauer dialect would be notably different from each other.

hear… but in normal conversations of course with dialectal applications." (translation: JR)<sup>24</sup>

A teacher from Upper Austria defined her way of speaking in class as a mix: "Uhm colloquial language. Partly with elements of dialect, but more colloquial language." (translation: JR)<sup>25</sup>

That fact that the dialect-standard continuum exists not only in everydayconversations outside school but also in the classroom became clear during our participatory classroom observations. The observations confirmed the teachers' self-estimations to some extent, in that they do use standard language mostly for explaining new contents and subject-related communication, and dialect is mostly used for less formal purposes. But what also became clear, was that teachers performed a small, but noticeable and recurring degree of codeswitching and code-mixing in all communication situations, not only in less formal contexts, but also during formal Standard-oriented communication – this happens probably to a much larger extent than teachers are aware of.

Examples for the insertion of dialectal<sup>26</sup> chunks into a stream of standard language when giving explanations on a topic during subject-related classroom communication would be "Remember this please!" (translation: JR)<sup>27</sup> right in the middle of going on explaining something in standard language. Another teacher, just after having begun the lesson and introduced the lesson's topic using standard language, switches to dialect for one sentence, being highly irritated when a student comes in late for class: "This pisses me off, and quite a bit!" (translation: JR)<sup>28</sup>

Code-switching and code-mixing were also frequently noticeable during the formal interviews: Apart from switching between standard language and dialect by inserting chunks of sentences in dialect into a standard phrase ("So during class I do use formal colloquial language. And only when it gets emotional, so when I scold someone <for example (laughing)>, than uhm it is/ it is totally clear

<sup>24</sup> "In meinem Fall is des a Mischung aus eben schon leicht verschwundenem oberösterreichischen Dialekt, und natürlich als Deutschlehrer is man do irgendwie a (in) einer gewissen Rolle (drinnen) natürlich irgendwie auch im Dialektalen irgendwie verhaftet, irgendwie so mittendrin. Ja? – - Jo, also jetzt/ kein Lungauer Dialekt. Wie man hört… aber in normalen Gesprächen natürlich mit dialektalen Verwendungen."

<sup>25</sup> "Ahm Umgangssprache. Mit teilweise vielleicht Dialektelementen, aber eher Umgangssprache."

<sup>26</sup> Dialectal insertions/dialectal pronunciation of single words in all the following German quotes and in their English translations are marked in italics.

<sup>27</sup> "Meakts eich des bitte!"

<sup>28</sup> "Des föht mi o, oba net wenig!"

that I […] this/ this in dialect.") (translation: JR)<sup>29</sup>, teachers also often pronounced individual words within a sentence in a dialectal manner, thus mixing codes:

"[…] The switch is flicked and then they try to speak, uhm, really in a formal, uhm, standard language, uh/ actually, so, or formal colloquial language, and, uh, I noticed that it is simply a challenging task for the students/ it is also a besetting effect […]." (translation: JR)<sup>30</sup>

In the group discussion with the student group, the students commented on their teachers' linguistic behaviour and confirmed our observations from participatory classroom observations as regards code-switching and code-mixing carried out by teachers. The students proved to be highly accurate observers and were able to offer plenty of examples of situations in which teachers would switch from standard to dialect, thus portraying a function-oriented pattern for the switching. The following excerpt from the group discussion illustrates this:

"The [subject]-teacher switches [between varieties] quite a lot. Well, it always depends on the topic. He explains something in High German to make sure we all understand, and then he makes this joke, which only half of the class understand because it is SO in dialect." (translation: JR)<sup>31</sup>

The students also commented in a slightly critical way on the standard competence of some of their teachers: "Teacher X speaks High German, teacher Y exclusively in dialect. Some cannot really speak High German." <sup>32</sup> This individually varying competence in Standard German or the lack thereof was something students had also observed and they were easily able to provide examples. What students noticed in particular was the conscious effort some teachers seemed to be making to meet linguistic expectations for German lessons, therefore attempting to speak Standard German but not being able avoid dialectal words every now and then—typical examples of (unintended) codemixing. The following quote illustrates this:

<sup>29</sup> "Also im Unterricht verwende ich schon gehobene Umgangssprache. Und nur, wenns emotional wird, also wenn ich schimpf <zum Beispiel (lachend)>, dann ah is des/is des gonz kloar, doss i des/des im Dialekt […]."

<sup>30</sup> "[…] Da wird der Schalter umgelegt und dann wird schon versucht, ä::hm::, wirkli in:, in aner gehobenen, ah, Standardsprache, ah/ eben, also, oder gehobenen Umgangssprache zu: sprechen, u::nd, ah, da hab i festgestellt, dass es für die Schüler anfoch a Überwindung/ es ist a – Gewohnheitseffekt auch […]."

<sup>31</sup> "Der [Fach]-Lehrer wechselt ganz oft ab. Also es kommt immer ganz aufs Thema drauf an. Er erklärt schon hochdeutsch, damits ja alle verstehen, und dann schiebt er einen Schmäh rein, den aber nur die halbade Klasse versteht, weil der einfach SO im Dialekt ist."

<sup>32</sup> "LehrerIn X redet Hochdeutsch, LehrerIn Y nur im Dialekt. Manche können auch nicht Hochdeutsch so richtig."

"Yes, they make an effort. Yes, I noticed this, especially with our [subject]-teacher, that as she is from the country, really often she has to concentrate and from time to time something slips in, but she really tries." (translation: JR)<sup>33</sup>

To sum up, the degree of code-switching and code-mixing among students and especially among teachers—consciously or unconsciously—is surprising, considering that teachers are regarded as norm authorities and (linguistic) role models by didactics and by society.

#### 6.2 Language Change due to Language Contact? ASG and GSG among Students and German Teachers

In one part of the survey, we gave students and teachers sentences containing 26 pairs<sup>34</sup> of ASG and GSG variants<sup>35</sup>). The probands were asked to underline the variant (ASG or GSG) they would rather use in a written text. Variants included were not only lexical: Grammatical variation was also targeted, such as morphological standard variation<sup>36</sup>, prepositional variation<sup>37</sup>, variation in formation of the perfect tense<sup>38</sup>, the use of the definite article before first names and relational titles<sup>39</sup>, variation in word order in subordinate clauses with three-part verbs<sup>40</sup> or gender variation<sup>41</sup>. For some pairs of variants, only one of the variants was distinctly marked as GSG by dictionaries<sup>42</sup>, while the opposing part was not distinctly ASG but unmarked German<sup>43</sup>, or vice versa<sup>44</sup>. These pairs of variants were nevertheless included in the selection, as favouring or disfavouring of one of the two variants also points in a certain direction of language preference.

<sup>33</sup> "Ja, sie bemühen sich. Ja, das ist mir schon aufgefallen, besonders bei unserer [Unterrichtsfach]-Lehrerin, dass sie/ da sie auch vom Land kommt, wirklich oft, dass sie sich konzentrieren muss und ab und zu rutscht ihr was rein, aber sie bemüht sich sehr."

<sup>34</sup> Whenever a pair of variants is mentioned in this chapter, the first term is always the ASG variant and the second term is always the GSG variant, f.ex. ein Cola (ASG) / eine Cola (GSG).

<sup>35</sup> E. g. Nach dem Aufstehen trinke ich gerne ein Cola / eine Cola, um wach zu werden. Als du mich angerufen hast, bin ich / habe ich gerade bei der Bushaltestelle gestanden. ('After getting up I like drinking cola to wake up. When you rang me, I was standing at the bus stop.')

<sup>36</sup> E. g. linking elements / 'Fugenmorpheme' as in Schweinsbraten / Schweinebraten ('pork roast').

<sup>37</sup> E. g. auf dem Laufenden / am Laufenden ('updated').

<sup>38</sup> E. g. bin / habe gestanden ('was standing').

<sup>39</sup> E. g. der Andreas / Andreas, die Oma / Oma.

<sup>40</sup> E. g. sagen hätte wollen / hätte sagen wollen ('would have liked to say').

<sup>41</sup> E. g. der Einser / die Eins ('school grade A'); das / die Cola ('cola'), das / die SMS ('short text message').

<sup>42</sup> E. g. lecker ('delicious').

<sup>43</sup> E. g. schmeckt gut ('tastes good').

<sup>44</sup> E. g. Wimmerl / Pickel ('pimple'), Bub / Junge ('boy').

The choice of variants in our study was based on the principles that the variants (both ASG and GSG) should be known to the probands, they should be taken from the probands' everyday lives, and that the sample should contain variants which the team of investigators had observed to be prone to language change.

The aim of this survey question was to identify tendencies concerning differing language preferences among students and teachers and see whether GSG expressions were preferred among the younger generation instead of ASG expressions. The corresponding research questions regarding possible language change using these sample sentences were if age-related tendencies pointing to language change could be found, and whether such age-related language change tendencies, if found, were correlated to TV-watching habits.

The analysis of our results showed unexpectedly strong and statistically significant differences among teachers and students concerning their preferred variants (p<0.000): Not only was the average number of preferred GSG variants higher among students than among teachers, but students turned out to prefer more GSG variants in total (54 %) than ASG variants (46 %) compared to teachers, of whom a clear majority (61.2 %) preferred ASG variants to their GSG counterparts (38.8 %). This supports the hypothesis of stronger'language loyalty' or rather 'loyalty towards the ASG variety' among language experts (teachers) and the older generation (as shown in Figure 4).

Figure 4: Teachers' and students' preference of Austrian Standard German (ASG) or German Standard German (GSG) expression

In order to be able to look more closely at generation-specific differences, in a next step all probands, students and teachers, were split into three age cohorts (14–21-year-olds, 22–40-year-olds, 41–65-year-olds), so that two 'generations' for the purpose of a relational generation concept could be constructed and analysed. In the next step, two of these generations, the youngest age cohort (14– 21-year-olds) and the oldest age cohort (41–65-year-olds) were compared: A statistical analysis was performed again with the aim of spotting the actual difference between the younger and the older generation in our sample as regards ASG or GSG preference in the 26 variant pairs. Again, we found significant differences. Figure 5 shows the variant pairs where 'young' and 'old' differed most significantly (p<0.01) concerning their preference of GSG terms.

Figure 5: Generational differences in preferences of some GSG expressions

Figure 6: Students' preferences of ASG or GSG expressions in correlation with TV-watching habits

One of the research questions aimed to find out if therewas a correlation between a tendency towards GSG variants among younger people and their media consumption. Therefore, one survey question was concerned with the TV channels the probands most frequently consumed. In a next step, we checked for a statistical correlation between language preferences and media consumption.<sup>45</sup> As it turned out, we found a significant correlation (p<0.000): Those students (14–21 year-olds) who had said to watch mostly Austrian TV-channels showed a slight preference for ASG variants (51 % ASG variants preference, 49 % GSG preference), whereas students who said they watched mostly German TV-channels tended to clearly prefer GSG variants (60 % GSG variants, 40 % ASG variants), as shown in Figure 6. To conclude, results cautiously indicate that TV-consumption may influence language use significantly and can therefore be considered a vital factor in language change.

#### 6 Conclusion and Research Desiderata

The research presented in this article was the first of its kind to fill gaps in the field of classroom talk in Austrian schools concerning the dialect-standard continuum, as well as the generation-specific language change which is supposedly occurring in Austria.

In this article, it was first of all shown that the results of participatory classroom observation point towards a considerable extent of both code-switching and code-mixing between standard and dialect taking place during German lessons in Austrian schools. Teachers frequently stated in the interviews that they consciously chose a variety (dialect, colloquial language or Standard German) depending on the classroom situation. According to their self-reported behaviour, teachers tend to switch between Standard German and less formal varieties when they move from a formal teaching situation (explaining new content or having teacher–student class discussions) to emotional classroom situations, especially when disciplinary action is necessary. The student group discussion showed that students seemed to be accurate observers as to when and how their teachers switch between varieties. During participatory classroom observation, it also became clear that a great deal of intra-sentential code-mixing was occurring. The latter results indicate that teachers switch and mix more than they state to do.

However, a certain degree of vagueness in the results remains regarding how the probands actually comprehended the terms 'dialect', 'colloquial language', 'standard language', 'Austrian Standard German' or'German Standard German', since it is known that the linguistic concepts of lay persons and experts can differ substantially and that particularly linguistic lay persons (but also experts) can experience difficulty pinpointing and delineating these terms.

<sup>45</sup> Of course, these results need to be interpreted with caution, since the information given on TV consumption as well as the information on the preferred ASG or GSG variants are selfreported data.

In addition, as stated in the introduction, some of the data represent selfreported rather than actual language use and the lessons observed in participatory classroom observation were not recorded or filmed due to restrictions by school authorities. Therefore, in order to acquire actual language use data rather than self-reported estimations and perceptions, additional research needs to be conducted in a follow-on study, preferably with a design similar to Steiner's (2008), who examined switching between Swiss Standard German and dialects in Swiss schools and recorded as well as filmed school lessons.

Secondly, language preferences concerning ASG and GSG were scrutinised. When putting a finger on the commonly assumed language change taking place due to the influence of GSG in Austria, itwas found that in a small given sample of variants there were significant differences regarding language preferences in ASG and GSG among teachers and students. Upon closer examination and confirmed by statistical tests again, it transpired that there were indications of generationspecific language preference, pointing to some GSG variants replacing ASG variants among the younger generation. Although this had been frequently suspected in recent years by experts and lay people, this study produced first empirical data to support it. Particularly the statistically significant correlation between preference for GSG variants and media consumption proved to be a highly interesting result demanding further investigation.

Even though the results were able to support the assumption that these generational language preferences are present in Austria, of course a study with a sample of variants this small has mainly got exploratory character. For a truly representative study, a much bigger sample is needed, preferentially supported by a corpus-linguistic study<sup>46</sup> using a large linguistic corpus, such as the Austrian Media Corpus (AMC)<sup>47</sup> .

Following thisfirst systematic study on (mostly) reported language behaviour presented here, a large-scale third-party-funded project investigating actual language behaviour in German and other lessons in Austrian schools should be headed for. This could well be an addition to the SFB-project "Deutsch in Österreich" 48 .

Why is this kind of research important at all for Austria and Austrians, apart from variational linguistic interests? For one, studies have shown that a positive attitude toward one's own language / variety as well as identification with one's own language / variety is important for language maintenance: Language loyalty

<sup>46</sup> A corpus-linguistic study (Konecky 2017) investigated the actual use of the variants of our survey sample sentences with the AMC (Austrian Media Corpus) and came to the conclusion that Austrian newspapers were altogether rather 'loyal' towards ASG variants, GSG variants were only used by very small percentages.

<sup>47</sup> AMC = Austrian Media Corpus. For information on the AMC see Ransmayr 2018.

<sup>48 &</sup>lt;https://dioe.at/>.

is regarded as an essential socio-psychological factor for the survival of languages or varieties when 'threatened' (Brenzinger 1997; Dressler / de Cillia 2006; Fishman 1964; Löffler 2016). Furthermore, teachers and students will benefit from the research gathered in this project and in future follow-on projects: Results can be taken directly into teacher training at universities and integrated into measures for professionalisation of aspiring and in-service German teachers.

#### Bibliography


und Gesellschaft. 2nd completely revised and extended edn. Vol. 1. Berlin / New York: de Gruyter, 296–300.


Jutta Ransmayr, Ass.-Prof. Dr. Mag. University of Vienna German Department and Department of Teacher Training Universitätsring 1 1010 Vienna Austria jutta.ransmayr@univie.ac.at

# Lisa Krammer

# Frau Professor, wo is'n die Anwesenheitslist'n?<sup>1</sup> Attitudes Towards the Variational Usage of German Language in the Context of University Lecture. An Empirical Study at Viennese Universities

Abstract: This paper explores the attitudes of students, from different fields of study at three Viennese universities, towards the oral, variational usage of German at universities in the context of lecture. Methodologically, the study employs an online-questionnaire. While the results indicate that inter-student communication mostly features non-standard tendencies, conversations between students and lecturers follow a different pattern. The self-reported use of standard German decreases in every examined field of study in the following order: conversations with lecturers during teaching sessions, conversations with lecturers after teaching sessions and conversations with lecturers during office hours. The use of non-standard German is associated with the factors intimacy, spontaneity and non-public contexts.

Keywords: Language attitudes, variationist linguistics, German language, Viennese universities, context of lecture

Abstract: Den Untersuchungsgegenstand bilden Einstellungen der Studierenden unterschiedlicher Studienrichtungen an drei verschiedenen Universitäten in Wien zum mündlichen, variationsspezifischen Gebrauch der deutschen Sprache im Kontext der universitären Lehre. Methodologisch stützt sich diese Untersuchung auf eine Online-Fragebogenerhebung. Während die Ergebnisse in Bezug auf die Verständigung unter Studierenden eine standardferne Tendenz aufweisen, lässt sich bei den Gesprächen mit Lehrenden ein deutliches Muster erkennen. Der vermeintliche Gebrauch von Standarddeutsch nimmt in allen untersuchten Studienrichtungen gemäß folgender Reihung ab: Gespräch mit Lehrenden während der Lehrveranstaltung, Gespräch mit Lehrenden nach der Lehrveranstaltung und Gespräch mit Lehrenden in der Sprechstunde. Die Verwendung von Nonstandard-Deutsch ist mit Faktoren wie Vertrautheit, Spontaneität und Ausschluss der Öffentlichkeit verknüpft.

<sup>1</sup> deu 'Professor, where is the attendance list?'

Keywords: Einstellungsforschung, Variationslinguistik, deutsche Sprache, Wiener Universitäten, universitäre Lehre

#### 1 Introduction

Selected universities of Vienna—educational and research institutions themselves—have been subjected to an empirical analysis, in order to close a gap in the field of linguistic research. Personal observations of the usage of many varieties of the German language in German studies-teaching sessions at the University of Vienna provided this research project with the first impetus. Universities are places of mobility, encompassing both international orientation and regional ties. The diverse geographic and linguistic range of students is characteristic of Viennese Universities (cf. STATcube 2018).

The objective of this paper is to explore the attitudes and perceptions of students from different fields of study at three Viennese universities towards oral interactions in an academic setting, specifically in the context of lecture. This empirical study is based on students'self-assessment and aims to ascertain which 'linguistic forms of appearance' (style/register/variety) they use at university. Which variety of German is used in which specific situation (When?), under which conditions (How?) and in interaction with which communication partners (With whom?) (cf. Fishman 1965)? This general question is interconnected with another: Which variety of German fulfils which functions and is associated with which competences and interests?

[2] deu Sprache ist für mich etwas sehr Persönliches. Diese muss/ soll nicht um jeden Preis versteckt werden. Man darf schon auch hören, woher wer kommt. (Krammer 2016: 3) 'Language is very personal for me but should not be concealed. One should be able to hear where somebody originally comes from.'

[3] deu Ich halte den universitären Rahmen für äußerst angemessen, um eine 'schöne Sprache' zu pflegen und deutlich zu kommunizieren. Außerdem verliert eine intellektuelle Unterhaltung meiner Meinung nach stark an Glanz, wenn sie durch Dialekt und Umgangssprache verunreinigt wird. Privat is ma recht wurscht. Ich will nur nicht, dass ich immer schlampert daherred. (Krammer 2016: 3) 'It is appropriate to use a 'beautiful' form of language and to communicate very precisely at university. In my opinion an intellectual conversation can be tarnished by dialect and colloquial language use. At home I do not care. I just do not want to speak sloppily all the time.'

These two statements from respondents of the conducted survey highlight many essential aspects of the research project. Prestige/stigma of varieties, appropriateness, comprehensibility and the difference between public and private contexts are recurrent themes in the respondents' answers. The switch from standard to non-standard German in the second statement [3] additionally intensifies the meaning of the statement. A significant part of the analysis is dedicated to identifying and systemising such supposed switches between standard and nonstandard German.

Linguistic expressions, like those quoted ones above, and statements about language use must be considered separately (cf. Neuland 1993: 723). The latter do not necessarily reflect real usage patterns of language. However, to properly explore these linguistic concepts and variational patterns of the German language, "it is necessary to return […] the speaking and experiencing individuum back into linguistics" (Busch 2017: 22; translation: LK)<sup>2</sup> .

#### 2 Theoretical Principles

Sociolinguistic research exploring language attitudes<sup>3</sup> has been established as an approach in its own right, next to objective (i. e., for example, recorded) linguistic data (cf. Lenz 2003: 265).

From a social-psychological point of view, attitudes can be described as "lasting positive or negative evaluation[s] of an attitudinal object" (Hartung 2010: 61; translation: LK)<sup>4</sup> . The herein examined attitudinal object is the language use at Viennese universities in the context of lecture. Attitudes are not directly observable; they exist neither in a "social vacuum" nor are they "stable and immutable frames of reference" (Giles / Watson 2013: 18). The same attitude can have different functions not only for different people but also for the same person at different times (cf. Vogel / Wänke 2016: 15–16). It is difficult to draw a line between the multidimensional and dynamic concepts of attitudes and perceptions; they are interconnected and interdependent (cf. Anders 2010: 73; Lenz 2003: 267).

This study is based on two models: Firstly, the context-sensitive theory from Tophinke / Ziegler (2006) in which attitudes are embedded in different contexts: the macro context is provided through participation in society (standards, values of a culture), the meso context through specific situations (roles, role patterns, proximity-distance-relation) and attitudes are also integrated in single interactions (intention, self-concept, expectations) (micro context) (cf. Tophinke /

<sup>2</sup> Um laienlinguistische Konzeptualisierungsstrategien und Perzeptionsmuster der deutschen Sprache zu erforschen, muss "das sprechende und erlebende Subjekt in die Sprachwissenschaft" zurückgeholt werden.

<sup>3</sup> An overview of the state of research: Albarracín et al. (2014). Regarding to the history of language attitudes: Casper (2002: 15–23).

<sup>4</sup> Sozialpsychologisch stellen Einstellungen "relativ überdauernde, positive oder negative Bewertungen gegenüber einem Einstellungsobjekt" dar.

Ziegler 2006: 5–8). Secondly, the pragmatic-constructivist theory of attitudes —the REACT-model from Purschke (2015: 49; emphasis in original: LK)—which states that: "Attitudes are relevance-driven targeting and evaluation routines on a high level of activation that sediment in an individual's stock of knowledge and are situationally (re)constructed in interaction". Language attitudes are therefore not stable mental dispositions, but rather dynamic and dependent on the current context and specific situation.

#### 3 State of Research

In 1975, Schlieben-Lange (cf. 1975: 107) was astonished that there were no studies on communicative behaviour at universities. Even at this point in time, hardly any linguistic studies of the institution university in Austria have been published. Weber (cf. 1980: 171–172) analysed the language usage of studentsin interviews. Apart from Ransmayr's thesis (2005), which tackles the role of Austrian (standard) German at foreign universities, and the research project "VAMUS" 5 (Dannerer / Mauser 2018), no other projects about language attitudes in the context of lecture in Austria exist.

The "VAMUS" study focuses on language use, practice and attitudes of students, lecturers and administrative staff. In addition, aspects of language policy concerning 'internal' and 'external' multilingualism at the university of Salzburg are being investigated. Their data consists of questionnaires, semi-structured interviews, recordings and written university documents (cf. Dannerer / Mauser 2019: 389). Results from the questionnaire indicate that 93 % of students consider standard German as the appropriate form of language during lectures. In contrast, the use of dialect is perceived as more adequate in interactions with peers (cf. Dannerer / Mauser 2019: 393).

In contrast to "VAMUS", this study is based in Austria's capital Vienna and focuses on students from three different fields of study at three Viennese universities. In Germany,Gogolin et al. (2017) examined ('external') multilingualism at the university of Hamburg. Additionally, attitudinal data about the language use at universities in Switzerland has been reported (Lüdi / Werlen 2005). In the German speaking part of Switzerland, due to diglossia, it is common that lecturers have a professional conversation amongst each other in dialect but switch to standard German when their next seminar begins—patterns of communication that would supposedly, according to Schmidlin (cf. 2011: 102), not be conceivable in Germany or in Austria. It is clear that the linguistic situation of

<sup>5</sup> An acronym for "combined analysis of multilingualism at the university of Salzburg" —"Verknüpfte Analyse von Mehrsprachigkeiten am Beispiel der Universität Salzburg".

Switzerland and Austria cannot be compared unreservedly. However, the results detailed herein suggest that similar conversational constellations and usage of varieties also occur at Viennese universities.

Studies<sup>6</sup> about language attitudes in Austrian, German and Swiss schools are more numerous but are of limited relevance to the conducted study. Soukup / Moosmüller (2011: 44) expounded that at "university level, no explicit official mention of teaching language could be found". Therefore, the subject matter constitutes a veritable research gap in linguistics—specially concerning Viennese universities—and it is of utmost importance to find out how students think about the current linguistic situation at university.

#### 4 Research Questions

In the scope of this survey, students are required to think about and reflect their language use in the context of a lecture and to ask themselves: How do I speak at university?—Specifically, in different settings like lectures, seminars, laboratory tutorials, field trips and situational constellations such as group work, presentation<sup>7</sup> , discussion, oral examinations<sup>8</sup> , conversations before / during / after teaching sessions with lecturers and during office hours<sup>9</sup> . The study focuses on an analysis of oral interactions but there are also academic analyses of written communication at universities.<sup>10</sup>

Many people perceive universities to be public areas, where people are interacting homogenously in standard German (cf. Löffler 2003: 5). Does this envisioned monolingual conception of universities correspond to the individual perceptions of students at universities? According to the attitudes of the respondents, is standard German<sup>11</sup>the communicative standard at Viennese universities? Or do their attitudes reflect that other varieties of the linguistic repertoire are, in addition, spoken in different situations and teaching sessions? To settle these questions there are two aspects worth considering: First, the sociosymbolic meaning of the institution university for the questioned students must

<sup>6</sup> A selection of relevant sources: Davies/ Langer (2014); Hochholzer (2004); Knöbl (2012); Lenz (2014); Steinegger (1998).

<sup>7</sup> Further details about presentations in the context of lecture: Guckelsberger (2005).

<sup>8</sup> More information regarding an analysis of oral final exams at university: Meer (2011).

<sup>9</sup> A selection of investigations of office hour conversations at university: Boettcher / Meer (2000); Kiesendahl (2011); Limberg (2010); Meer (2011); Rudolph (1994: 199–230); Zegers (2004).

<sup>10</sup> A selection of written interaction (via email) in the context of university: Biesenbach-Lucas (2007); Hiller (2014: 233–258); Kiesendahl (2011).

<sup>11</sup> Differences between national varieties of standard German, especially between Austrian and German standard German, are playing a particular role at Viennese universities, but could not be considered due to the scope of this study.

be considered. Second, it is essential to focus on the socialisation and linguistic biography of the respondents. Assuming a monolingual polyglossia within the German language, coupled with "internal multilingualism" (Wandruszka 1979) and socio-demographic heterogeneity of the students, the focus is on students' attitudes regarding language use in situational contexts with social and biographic features. Although it is not the main goal of the research project, it is nevertheless necessary to keep the counterpart to "internal multilingualism" the concept of "external multilingualism" (Wandruszka 1979) and the role and status of English as lingua franca—in mind.

In accordance with current literature and individual observations, the following research questions were derived:


#### 5 Methodology

Austria's federal capital, Vienna, is home to nine public universities and six public universities of applied sciences and arts (cf. BMWFW 2014). The sample consists of 579 students in total, selected from different fields of study at three universities: German studies students from the University of Vienna, students (all fields of study) from the Technical University of Vienna and human medicine

<sup>12</sup> Are they feeling more uncertain or is rather the usage of standard German as an academic language challenging? In the investigation of Moosmüller (cf. 1991: 163) the respondents often emphasized the laissez-faire component of the dialect.

students from the Medical University of Vienna (see Table 1). The diversity of the fields of study (humanities, medicine, technology) and German as the main language of their courses (lectures with English as the main language are only optional) are of vital importance for the selection process. German studies students are chosen specifically because of their interesting in-between status—they are neither linguistic lay people nor linguistic experts. Courses in linguistics and linguistic history are obligatory in their studies. To yield relatively homogenous data, it is paramount for the different types of courses<sup>13</sup> and the general sociodemographic conditions of the respondents (being born in Austria and having German as mother tongue) to be comparable. The survey has, altogether, reached mainly students from the following federal states in descending order: Vienna (175 respondents), Lower Austria (153), Upper Austria (116) and Burgenland (55).


Table 1: Composition of the sample.

The operationalisation of the research questions was conducted using an onlinequestionnaire, a quantitative method of elicitation, and data analysis was performed afterwards using the spreadsheet program Microsoft Excel® 365 (Microsoft Corporation, Redmond, WA, USA). This quantitative approach was chosen due to the varied possibilities of the design of the questionnaire and the resource-saving practicability that allows researchers to reach large groups of people. Following a quantitative paradigm is a more confirmatory than an exploratory approach. The researcher is, as much as possible, uninvolved in data collection, increasing the objectivity of the study. Nonetheless, passiveness is also a disadvantage because the researcher is unable to directly observe and inquire. Additionally, it is impossible to make further adjustments while the survey is being conducted.

To guarantee the comparability of the attitudinal objects, definitions of standard and non-standard German are offered in a way that is understandable for lay-people at the beginning of the questionnaire. Within the majority of the

<sup>13</sup> Definitions regarding to curricula and more detailed information from the students' union.

Austrian linguistic scene, excluding the Alemannic west, the existence of a dialect-standard-continuum has been established (cf. Ammon 2003; Moosmüller 1991), while the possible segmentation of the continuum is still heavily debated (cf. Barbour / Stevenson 1998: 149; Bellmann 1983: 123; Danesˇ 2005: 47; Durrell 1998: 25). Can the transitions between varieties within this continuum be classified as being abrupt or rather blending into one another? This study ascribes to new insights regarding fluid varieties within the continuum (cf. Berruto 2004: 190). The terminology and definitions of dialect,standard German and colloquial language by Schmidt/Herrgen (cf. 2011: 59, 62, 66) form the basis of this research project. Hence, dialect is defined in the online-questionnaire as: "A way of speaking that is typical for a place or region and has many linguistic peculiarities. People who are not living in this place or region cannot easily understand this way of speaking." Colloquial language: "A way of speaking between dialect and standard German. It is more extensively spoken (and understood) than dialect." Standard German: "A way of speaking that is understood in all German-speaking areas and has no regional characteristics. Newscasters in TV and radio are speaking standard German." In addition, there is also the option of choosing: accommodation<sup>14</sup> from dialect to colloquial language, accommodation from colloquial language to standard German and accommodation from dialect to standard German. It is important to reflect and take into consideration that the terminology implies a concept of language as clear, distinct units. This is a limiting factor that doesn't correspond the idea of a dialect-standard-continuum nor with the personal comprehension of language and varieties in the respondents' everyday life. A possible solution to this methodical challenge will be discussed in the chapter Conclusion and Research Desiderata. Another challenge is that respondents often answer in a way that is socially desired (cf. Garrett 2010: 44). This study employed methodical countermeasures to reduce these disruptive factors, such as neutral wording, an adequate order of the questions and appropriate transitions from one question to another.

The online-questionnaire consists of five main topics, which are subdivided into further questions. The first topic is centred around which varieties of the German language students use, for which reason, in different didactic and situational constellations at university. Perceptions of the change of the students' personal language usage at the university during their studies and the aspect of appropriateness regarding the usage of non-standard German at university constitute the second and third set of questions. The next thematic block of questions covers the students' perception of oral language use from lecturers in different teaching sessions and situational constellations at university. The last

<sup>14</sup> For a more detailed description of the communication accommodation theory: Bell (2009: 265–275); Giles (2009: 276–286).

type of question is formatted as a scale, divided into six intervals. The respondents should declare their agreement ("I agree totally") and disagreement ("I do not agree at all") towards nine general statements about language use at university. At the end of the questionnaire, sociodemographic data is collected. Pretests with students from the three different fields of study are conducted. The duration of the online-questionnaire is approximately ten minutes.

# 6 Results

The results<sup>15</sup> will be discussed in the same order as the research questions (cf. Research Questions).

1) The following results demonstrate a perceived difference in the usage of standard and non-standard German between formal / informal, official / unofficial, public / private and prepared / spontaneous contexts.

Selected situational constellations: presentation, discussion and oral exam The perceived usage of standard German in different contexts decreases in every examined field of study in the following order: presentation, oral exam and discussion. Because of spontaneity, involvement of emotions and an emphasis on personal experiences, "standard German" seems to be less frequently used during discussions. The results indicate that German studies students (68 % and 57 %) and medical students (70 % and 55 %) use "standard German" with similar frequency during presentations and oral exams. According to the students' selfassessment, "Standard German" is spoken by 50 % of technical students during presentations and 38 % during oral exams. Furthermore, the number of students, who claim to use "colloquial language" during presentations, was the highest with technical students (18 %), compared to medical (10 %) and German studies students (8 %).

#### Conversations with lecturers

Non-standard usage during teaching sessions correlates with the students' field of study and, additionally, with the degree of closeness (cf. Koch / Oesterreicher 1985)<sup>16</sup> between students and lecturers. The self-reported usage of "standard

<sup>15</sup> The conducted data and results have, in a similar form, been presented in my diploma thesis (cf. Krammer 2016).

<sup>16</sup> The model of Koch/Oesterreicher (cf. 1985: 19–43) embraces a multidimensional proximitydistance-continuum: The dichotomous parameter like dialogue / monologue, closeness / foreignness of the communication partners, spontaneity / reflection can be analysed and due to the frequency assigned as category language of proximity or language of distance.

German" in conversations with lecturers during teaching sessions decreases in the following order: German studies students (54 %), followed by medical students (43 %) and technical students (37 %). Conversations with lecturers after teaching sessions and conversations with lecturers during office hours are more frequently conducted in "colloquial language" by technical students (39 % and 32 %), when compared to the two other groups. Also, compared to the other groups, more German studies students picked the options "accommodation from colloquial language to standard German" (37 %) and "standard German" (32 %). The results indicate that German studies students seem to be more likely to use "standard German" in interactions with lecturers, both during and after courses.

Figure 1: Differences in language use in different contexts, when conversing with lecturers. (D = dialect; C = colloquial language; S = standard German; D & C = accommodation from dialect to colloquial language; C & S=accommodation from colloquial language to standard German; D & S = accommodation from dialect to standard German)

Applying Koch / Oesterreicher's model (cf. 1985: 23), the usage of non-standard German in conversations with lecturers after teaching sessions and conversations with lecturers during office hoursseems to be related to or caused by the factors of closeness, dialogue and elimination of the public. Spontaneity, a free flow of conversational topics and less planned and elaborated speech patterns are characteristic for private conversations.

2) Attitudes of students regarding 'internal' multilingualism at university comprise various institutional, social (role patterns, idea of a norm, expectations, conventions) and individual (authenticity, professionalism) facets. The optional free-text fields in the questionnaire offer respondents the opportunity to explain their reasons for using "standard German", "colloquial language" and "dialect" at university in more detail. The various answers have been arranged into overarching thematic categories. Tables 2–4 summarise the most frequent reasons and offer exemplary quotes from the respondents.


Table 2: Comparison of reasons for using "standard German" at university.

Students of German studies mention an increased comprehensibility and a high level of appropriateness as distinct reasons for using "standard German" at university. Results are similar for technical and medical students, with the latter also associating increased prestige with the usage of "standard German". In contrast, technical students emphasise the importance of using "standard German" in written communication.


<sup>17</sup> 'best possible comprehensibility' / 'effort to act formally correct' / 'to appear more professional and it is a must in written communication'.

<sup>18</sup> 'to avoid misunderstandings' / 'expression of respect and good manners' / 'when I try to appear more serious and official'.

<sup>19</sup> 'standard German is the only acceptable choice' / 'that everybody can properly understand me' / 'it is more polite towards lecturers and more formal'.

<sup>20</sup> 'because I either know the lecturer very well or the situation and the content of the conversation don't require that [standard German]' / 'that is the most natural way of speaking in my opinion' / 'the most comfortable way of conversing'.

<sup>21</sup> 'because one has to think less about what to say next' / 'normal language use' / 'it sometimes appears to be more authentic and likeable'.


#### (Continued)

Table 3: Comparison of reasons for using "colloquial language" at university.

German studies and technical students most often associate the usage of "colloquial language" with a more "private or relaxed context". Students from all three fields of study mention ease and habit as important factors of "colloquial language" use at university.


<sup>22</sup> 'because the atmosphere is more casual, relaxed and the conversation will not be assessed' / 'to communicate more effectively than in dialect' / 'one doesn't have to exert oneself (Yes, as a person from Styria it is strenuous to speak in standard German all the time)'.

<sup>23</sup> 'only with people from my hometown' / 'I do not use dialect at university – because one runs the risk of being pigeonholed' / 'because my opinion is that dialects need to be cultivated'.

<sup>24</sup> 'because I am talking to somebody with a similar dialect' / 'because it is my mother tongue and I prefer to speak that way' [written in dialect] / 'because it feels best to me and I do not have to think a lot about what I am saying'.

<sup>25</sup> 'I only use dialect in interaction with people who also live in the countryside' / 'more security, more self-confidence, […] I do not have to put on an act' / 'out of spite for the people from Germany' / 'because it is fun'.


#### (Continued)

Table 4: Comparison of reasons for using "dialect" at university.

Students from all investigated fields of study agree that the usage of "dialect" is most appropriate when conversing with colleagues (and in general: people) who also speak the same dialect. Medical and technical students also cite habit as a driving force for the usage of "dialect" at university. Interestingly, only students of German studies are of the opinion that the usage of a "dialect" should be avoided at university.

What do you think if students do not speak standard German during teaching sessions at university? 53 % of German studies students, 36 % of medical and 35 % of technical students answered that it "depends on the situation". The most frequent answer from medical students is that they "do not mind" (43 %), contrary to the attitudes of German studies students: Only 24 % "do not mind" other students using non-standard German. In a free-text field, medical and technical students have only classified non-standard German as inappropriate if it is not comprehensible to others. German studies students find non-standard German use during presentations and discussions inappropriate. Subsequently, the aspect of comprehensibility is the second most frequent category mentioned.

Figure 2: Attitudes about other students' language use at university.

3) An analysis of the free-text field "reasons for using 'standard German' at university" revealed that perceived politeness/respect and clear/precise modes of expression are associated with standard German and cause switches from nonstandard to standard German. Respondents who perceive non-standard German to be their "mother tongue" are an exception in the sense that non-standard German seems to be recognised as the most authentic and expressive form of speaking (cf. Löffler 2016: 142). Over 60 % of the questioned students from every examined field of study agree with the statement that "scientific content can be explained more appropriately and clearly in standard German".

4) Fundamental aspects of prestige/stigma of standard and non-standard German, attributed by students, correlate with the duration of their studies, their sociodemographic background and field of study.

#### Duration of their studies (newer/advanced students)

The majority of students from the examined fields (44 %–51 %) do not think that their language use has changed during their studies. While the attitudes of medical and technical students concerning potential changes in language use seem to be quite similar, those from students of German studies differ prominently: they supposedly have a stronger tendency towards "standard German" (38 %) and a reduced development towards "colloquial language" (11 %) during their studies. The assumption that newer students, in order to fit in, tend to speak standard German more often at university than advanced students could not be proven.

Figure 3: Students' experienced change of their personal language use during the course of their studies.

#### Sociodemographic background (state, dialect loyalty)

In every examined field, students from Upper Austria show higher dialect loyalty when interacting with their peers during lectures (in conversations and group work) than students from Lower Austria and Vienna.

#### Field of study (humanities, medicine, technology)

In general, students of German studies seem to have a more distinctive awareness and a more pronounced understanding of 'internal' multilingualism of the German language in the context of a lecture. They are more likely to use standard German and question the use of non-standard German at university more often than medical and technical students. Further investigations are necessary to examine if these attitudes are related to the specific linguistic focus in their studies or the fact that many students from the sample analysed herein want to become teachers. In this survey the sample of German studies students consists of 99 BA- and MA-students and twice as many teaching students (233). In addition, teaching degrees do not exist in the other fields of study examined, therefore complicating the comparability of the sample.

5) According to the results of the online-questionnaire, lecturers from Austria are perceived to be using the spectrum of German varieties more extensively than lecturers not originating from Austria. German studies students do not think that there is a link between the age and the language usage of lecturers. In contrast, medical and technical students seem to perceive lecturers under 40 years to more often use "colloquial language" and "standard German" less often than 40 +-year-old lecturers.

# 7 Conclusion and Research Desiderata

This empirical investigation explored the heterogeneity, dynamic nature and the 'internal' multilingualism of the German language in Austria from the perspective of students' self-assessment in the, so far, hardly explored Austrian tertiary education sector. For the first time, students' attitudes about their language use at Viennese universities were analysed, with new insights paving the way for further linguistic, as well as interdisciplinary, research projects. According to the results, a hypothesised mono-variational, standard German manifestation of language use (cf. Gogolin 2008) does not correspond to the subjective, communicative world of the questioned students. One-dimensional linguistic research approaches seem to be inadequate when accounting for language-based social functions and pragmatic requirements at university. The analysis of data highlights that the questioned students from all fields of study are more likely to use non-standard German in more personal contexts at university. In contrast, standard German is associated with presentations and conversations with lecturers during teaching sessions, i. e. more public contexts.

In a future questionnaire-based study, a scale with only two extreme poles, labelled dialect and standard German, might be better suited to integrate the theoretical background of the dialect-standard-continuum and to avoid segmenting the continuum any more than necessary. Methodologically, the research could be enriched by an additional qualitative survey. Semi-structured interviews with students and lecturers from the universities could be conducted. Audio recordings of different course types and during office hours would also be an interesting way to approach the research question.

Do explicit/implicit language-policy-strategies in the context of university exist (cf. Flach 2016; Maier 2016)? Is an orientation towards a linguistic norm required in the university statutes?<sup>26</sup> In addition to "language practices" and "language beliefs or ideology", aspects of "language intervention, planning or management" (Spolsky 2004) could be investigated via the study of relevant written sources.<sup>27</sup> The university as an area of research itself might best be investigated by employing the triangulation of language attitudes, language policy (and ideology) and language practices research.

The results of this study might, in the future, influence university teaching and teaching methodology. Students and lecturers might be more aware of the varieties within the German language and encouraged to foster or develop a reflective perception and handling of language. Further investigations should be conducted at universities of applied sciences and arts and at teacher training colleges.

#### Bibliography


<sup>26</sup> The workshop education policy and language policy at schools and universities by Busch (cf. 2003: 161) revealed that Austrian universities do not seem to have any explicit guidelines concerning language policy.

<sup>27</sup> A selection of possible documents: laws regulating university studies, development plans of the universities, curricula.


Mag. Lisa Krammer, BA University of Vienna Department of German Studies Universitätsring 1 1010 Vienna Austria lisa.krammer@univie.ac.at

# Wolfgang Koppensteiner / Agnes Kim

# Perspectives on Change: Language (Varieties) Contact and Language Ideologies on German in Austria

Abstract: This paper investigates layman attitudes towards language change and language contact of German in Austria. The data stems from 34 semi-structured interviews that were conducted in small rural locations throughout Austria. In a three-step methodological approach, the paper aims at identifying inter-individual attitudinal patterns, which may hint at language ideologies. First, the informants' conceptualizations and evaluations regarding aspects of language change and contact are categorized intra-individually. In the second step, an explorative cluster analysis based on step one identifies three clusters: the 'optimists', 'standard pessimists' and 'dialect pessimists'. In the last step, the informants in the single clusters provide an argument for different factors of language change, which we interpret as diverging discourse archives.

Keywords: language change, language contact, German in Austria, language attitudes, language ideologies

Abstract: Der vorliegende Artikel untersucht Spracheinstellungen von Laien zu Sprachwandel und -kontakt des Deutschen in Österreich anhand von Daten aus 34 halbstrukturierten Interviews, die in kleinen ländlichen Orten aus ganz Österreich durchgeführt wurden. Ziel ist die Identifikation von interindividuellen Einstellungsmustern, die wir als Sprachideologien interpretieren. Zunächst werden die von den InformantInnen geäußerten Konzeptualisierungen und Evaluationen von Sprachwandel und -kontakt intraindividuell kategorisiert. Aufbauend darauf können in einer explorativen Clusteranalyse drei Cluster identifiziert werden: Die 'OptimistInnen', 'StandardpessimistInnen' und 'DialektpessimistInnen'. Abschließend zeigt sich, dass die InformantInnen der einzelnen Cluster unterschiedliche Faktoren für Sprachwandel nennen, was wir im Sinne divergierender Diskursarchive interpretieren.

Keywords: Sprachwandel, Sprachkontakt, Deutsch in Österreich, Spracheinstellungen, Sprachideologien

# 1 Introduction

This paper<sup>1</sup> investigates the role of language contact as a factor for language change in laymen's language ideologies. Specifically, it analyses semi-structured interviews among speakers of German in Austria from small rural communities in order to trace their attitudes towards language change. It describes factors that these informants consider influential for language change and determines the importance of multilingualism, language and/or variety contact. The paper discusses the following research questions:


Methodologically, this paper combines qualitative and quantitative approaches. It explores the possibilities to utilize explorative statistical analyses for clustering informants according to their attitudes towards certain topics as expressed in interviews. First, the independent variables for this statistical procedure are generated by intraindividual content analyses of the interviews. Based on the clusters, interindividual patterns are then described with content and discourse analytical methods on trans-textual basis (cf. Spitzmüller / Warnke 2011).

When focusing on layman attitudes towards language change and language contact regarding German in Austria, this paper highlights an aspect that has not yet been investigated. In the analysis and interpretation, it therefore relates to concepts and findings from various linguistic subdisciplines. In this context, the paper is structured as follows: Section 2 provides a brief overview on German and other languages in Austria, focusing particularly on attitudinal and ideological aspects. Consequently, Section 3 illustrates the state of research regarding language attitudes and ideologies towards language change and language contact. Relevant theoretical and methodological issues for this paper are presented in

<sup>1</sup> This paper and the underlying researchwere supported by the Austrian Science Fund (FWF). It presents research results of the project parts "PP06: German and the Slavic languages in Austria: Aspects of language contact" (FWF F06006, principal investigator: Stefan Michael Newerkla, cf. Kim / Prochazka 2019; Kim 2020) and "PP08: Standard varieties from the perspective of perceptual variationist linguistics" (FWF F06008, principal investigator: Alexandra N. Lenz, cf. Koppensteiner / Lenz 2017) of the Special research programme (SFB) F60-G23 "German in Austria (DiÖ): Variation – Contact – Perception" (cf. Budin et al. 2018). The authors thank Stefan Michael Newerkla and two anonymous reviewers for their comments.

Section 4 and 5 respectively. Section 6 each describes the steps in the analysis. Section 7 concludes the results and identifies issues for future research.

#### 2 German and Other Languages in Austria: A Brief Overview

The term 'German in Austria'refers to several varieties, registers and styles of the German language (or rather: generally subsumed under the glottonym 'German') used in Austria. Therefore, a multidimensional description must consider at least (a) the standard register of German in Austria, (b) the non-standard registers, and in this context also (c) the relation of German to other languages in Austria. In the largest parts of Austria, Central or South Bavarian dialects are spoken, whereas in Vorarlberg and the westernmost parts of Tyrol Alemannic dialects prevail (cf. Lenz 2019). Excluding Vienna, local dialects and other non-standard registers have a considerable importance in everyday communication. Particularly for Bavarian parts of Austria, the vertical varietal spectrum is expected to be structured diaglossically, i. e. as a continuum with several 'concentration zones' (Verdichtungsbereiche; cf. Lenz 2010) in-between the local dialects and the standard. The Austrian standard variety of German is distinct from the German or Swiss standard variety in several features on all linguistic levels (cf. e. g. Lenz 2020 for phonology, Dürscheid / Elspaß / Ziegler 2018 for morphology and syntax, Ammon / Bickel / Lenz 2016 for the lexicon). Reliable extrapolations on the use of various registers of German and contemporary multilingualism in Austria are not available.

From an attitudinal point of view, we encounter reasonably non-transparent conceptualizations of 'standard' (as opposed to 'dialect') 'in the minds' of Austrian speakers and listeners (cf. Soukup 2009: 42). Typically, 'standard' is associated with (rather) public / formal domains (cf. Soukup / Moosmüller 2011); however, there are numerous aspects left to clarify. These include questions on types of prestige assigned to 'standard', its (possible) delimitation from other forms of German 'standard' registers (especially against the backdrop of Germany) aswell as whether there is more than one 'standard' in Austria (cf. Herrgen 2015; Koppensteiner / Lenz in print).

Attitudinal insights into registers of 'dialect' also offer quite heterogeneous results. In rural areas, layman conceptualizations highlight either (rather) diatopic aspects, whereas in larger cities and especially Vienna (rather) diastratic aspects prevail (cf. Breuer in preparation; Koppensteiner / Breuer in print; Steinegger 1998). Generally, 'dialectal' registers are allocated to (rather) privateinformal domains (cf. Soukup / Moosmüller 2011).

Regarding the construction of a specific Austrian (linguistic) identity, there are other languages besides German that also play a decisive role. The Second Republic of Austria frequently refers to and utilizes its history as part of the multilingual Habsburg monarchy in order to construct a linguistic identity that differentiates itself from Germany (cf. De Cillia / Wodak 2006; Kim 2020): On the one hand, the Austrian constitution declares German as the state's language (cf. BGBl. Nr. 1/1920, Art. 8). On the other, it recognizes its "linguistic and cultural multiplicity having grown" and thus essentially its historical multilingualism, while explicitly excluding multilingualism that resulted from migration movements throughout the late 20th and early 21st century. The historical multilingualism represents the remains of societal and individual multilingualism from the Habsburg monarchy. This applies to the autochthonous minorities and their languages, i. e. Slovenian, (Burgenland-)Croatian, Hungarian, Romani, Czech and Slovak. However, even considering these, Fischer / Doleschal (2011: 69) observe a development from being languages of autochthonous minorities to becoming neighboring languages and thus foreign languages. Austrian language policy generally does not support equal bilingualism, especially not regarding the many other languages spoken in Austria not officially recognized (cf. e. g. the data collected by Brizic´ / Hufnagl 2011 in primary schools in Vienna).

# 3 State of the Art: Language Change and Language Contact

Linguistics generally distinguishes language internal factors for language change from language external, i. e. sociolinguistic factors. However, these factors mutually support and influence each other: The internal factors determine the linguistic course and therefore the product of change, whereas the external factors play a decisive role in the actuation and propagation of change (cf. Hickey 2012: 403). Within multilingual groups and/or societies—including situations, in which a great majority of this group's or society's members learns one or more prestigious foreign languages—multilingualism and language contact play a crucial role in the course of language change (cf. community-external change in Hickey 2012).

Various sociolinguistic disciplines agree that speakers are the actual agents of linguistic change and that change begins with variation in language usage by these speakers. Variation is then explained by optimization strategies that speakers undertake when communicating with others, whose linguistic competencies differ (cf. e. g. Schmidt / Herrgen 2011 for theory from variationist linguistics and Muysken 2013 and Onysko 2019 for sociolinguistically or cognitively oriented contact linguistic accounts). Socio-pragmatic conditions influence the choice of languages or registers from the (multilingual) linguistic repertories of speakers and thus eventually trigger variation and change.

From a layman's perspective, language change is often evaluated as (gradually) negative. Regarding youth language, it is perceived as a decay of language (cf. Neuland 1993: 726) and in cases of (increasingly used) Anglicisms in German as a threat to the integrity of a language (cf. Spitzmüller 2005). The latter study discloses the pronounced use of metaphors within layman metacommunication on language change. These include: (a) implementing language as an organism, which is (life-)threatened by excrescences of language change, (b) language as a substance, which possibly might be dissolved by language change, (c) language as a container, becoming aware of impacts from outside, (d) language as an artefact, understood as created in conflict with possibly corruptive effects of language change (cf. Spitzmüller 2005: 210). As Spitzmüller points out, such layman conceptualizations are ultimately based on individuals' imaginations and perspectives. They assist in the orientation and fundamental categorization of the/ their world (cf. Spitzmüller 2005: 364–365).

#### 4 Theoretical Framework: Language Attitudes and Ideologies

Before turning to notions of language ideologies and language myths, this section briefly elaborates on language attitudes. Empirically "accessing" language attitudes is still an unresolved issue (cf. Garrett 2010: 29; Soukup 2019). This is grounded in complex parameters such as intra- and interindividual variation, the vast dynamics and forms of social interaction, the important role of (individual) knowledge of and on language, the importance of framing language attitudes within/of certain contexts, (interindividual) difficulties accruing from attempts of verbalizing attitudes at all, as well as interindividual processual-cognitive handling due to varying degrees of consciousness (cf. Lenz 2003: 265–267; Koppensteiner/ Lenz 2017; Preston 2010; Purschke 2014, 2015; Tophinke / Ziegler 2006). These issues can only be tackled by applying methodological triangulation that generates multivariate data (cf. Koppensteiner / Lenz 2017; Koppensteiner / Lenz in print). Referring to Purschke's (2015: 49) definition of attitudes as "relevance-driven targeting and evaluation routines on a high level of activation that sediment in an individual's stock of knowledge and are situationally (re)constructed in interaction", both the aspect of 'sedimentation' and its interactionally determined approach applied in this paper's methodological focus. In analyzing interview data, the informants' attitudinal utterances are interpreted as hinting at attitudinal fragments. Ultimately, aggregating such fragments highlights (inter-individually comparable) attitudinal patterns (e. g. shared sets of metacommunications).

Attitudes (including beliefs) of individuals do not, however, emerge from nowhere. They are rather embedded into and constructed by discursive repetition of these attitudes in social contexts (cf. Liebscher / Dailey-O'Cain 2009; Tophinke / Ziegler 2006). Through repetition, a discourse instantiates language ideologies in the neutral sense of the term, i. e. "shared bodies of commonsense notions about the nature of language in the world" (Rumsey 1990: 346; for the distinction of neutral and critical notions of 'linguistic ideology' cf. Woolard / Schieffelin 1994: 57). Moreover, hegemonic discourses and their underlying ideologies turn into "the law of what can be said about language" and are thus transformed into discourse archives in the sense of Foucault (cf. Watts 2012: 601). In this context, inter-individually recurring expressions of attitudes can be conceived of as items in a certain discourse archive. Therefore, attitudinal patterns analyze and describe language ideologies. This paper suggests and evaluates an explorative methodology to identify ideologies through the characterization of attitudinal patterns. In this approach, the analyzed interviews are understood to be both snippets and reflections of a general discourse on language (change and contact).

#### 5 Methodology and Corpus

The data sample<sup>2</sup> analyzed in this pilot study includes interviews of 34 informants from eight locations in total, representing all major dialect regions of Austria: Central Bavarian (Allentsteig [AL], Neumarkt/Ybbs [NY], Taufkirchen an der Pram [TK]), South-Central Bavarian transition area (Neckenmarkt [NM]), South Bavarian (Tux [TU], Weissbriach [WB]), Alemanic-Bavarian transition area (Tarrenz [TA]) as well as High Alemanic (Raggal [RA]).

All informants in the sample are 'autochthonous', defined as having been raised in and lived in the location for more than 50 % of one's life with at least one parent coming from the location as well. The informants were further divided according to sociolinguistic categories age, gender as well as the level of formal education. Regarding the parameter age, two groups where defined: adults between 18 to 35 ('Y'[oung]) and elderly persons aged over 60 years ('O'[ld]). Concerning the level of formal education, the informants were divided into those with a general qualification for university entrance ('H'[igh]) and those without ('L'[ow]). Informants of age group 'O' generally do not have a high school degree. Additionally, the informants were rated with regard to their overall regional

<sup>2</sup> The data was collected by the project parts 3 and 8 of the SFB "German in Austria". For the methodological details see Budin at al. (2018), Koppensteiner / Lenz (2017), Lenz (2018) and Lenz et al. (2019).

mobility and their degree of (professional) communication<sup>3</sup> (the higher the number from 1–4 the more mobile / communicative the informant). Table 4 in Appendix A gives an overview of all informants in our sample.

The semi-structured interviews were conducted in formal settings. Formality was established by certain parameters: The explorer spoke 'intended Standard' (i. e. the individual standard-next register, cf. Lenz 2003), and addressed the informant, whom he / she had not met before, formally. The interview guideline oriented towards gaining a wide range of attitudinal utterances on various aspects of German in Austria. It is comprised of various types of questions (both open and closed ones and supported using scales<sup>4</sup> ). Furthermore, one of the main aims of the interview was to ensure to elicit the informant's individual 'vertical spectrum' in a way that was as unbiased as possible, which also concerns (the use of) terms for registers and varieties. Therefore, any terminological influence of the explorer on the informant was avoided. Only (lay linguistic) terms proposed by the informant himself / herself were referred and tied up to in the course of the interview.

This paper analyzes questions dealing with 'dialect', 'standard' and '(historical) language contact', eliciting the informants' attitudes towards language change and language contact (cf. Appendix B). Most of the questions primarily targeted rather general aspects of language change. Regarding 'dialect' the informants' assessment of their local dialect's future was discussed (Q32) and evaluated whether they believed, that these (specified) developments were restricted to their hometown exclusively (Q32A). Concerning the 'standard' register, the interview assessed whether this register is subject to influences and language change, if (any) concrete changes are explicitly verbalized (both Q54) and how language change regarding this register is evaluated (Q113A). Similar to the 'dialect' section, another question targeted the informants' attitudes towards the future of the 'standard' register (Q54B).

Regarding language contact from a contemporary perspective, the informants were asked whether language contact triggers language change (Q103). Other questions targeted at influential languages (Q103A) and registers of German that are prone to change due to contact (Q103B). Regarding beliefs about historical language contact, the informants were confronted with the fact that German had

<sup>3</sup> 'Communication degree' defines as the informant's regular language use in professional contexts with various groups of people (apart from persons from the informant's own location with e. g. customers, colleagues and sales persons from other regions etc.).

<sup>4</sup> Scales are supposed to support the quantification of primarily qualitative data. In the interview, an informant is not just asked on his competence and usage of a certain variety, their answers will also be rated on scales. Scales support the informants in their individual determination of answers, e. g. if the informant is not sure or if problems occur with regard to verbalizing metalinguistic thoughts.

consistently been in intense contact with other languages throughout the history of Austria. They were asked to name contact languages (Q95) and to give their opinion, whether these 'left their traces' in German in Austria (Q95A).

Methodologically, three steps were taken to approach the data and identify patterns in the informants' attitudes towards language change and language contact:


# 6 Analyses

#### 6.1 Intraindividual Content Analyses

To structure the heterogeneous amounts of data (due to primarily open questions in the interviews) from a content analytical perspective, Mayring's (2015) methodological approach towards summarizing and bundling content was used. In this process, data was categorized inductively relying directly on the interview content. As demonstrated in Section 5, the interview guideline was designed to capture several aspects of attitudes towards certain objects of research interest. Nevertheless, we avoided deductive formation of (prefabricated) categories based on (to a certain extent expected) content / answers of the corresponding questions in order to deal with the data in a way that was as unbiased as possible. Technically, this step was supported by the CAQDAS (Computer-Aided Qualitative Data Analysis Software) ATLAS.ti (version 8; cf. Friese / Soratto / Pires 2018). In the following section, we will highlight selected recurring findings within our data on exemplary basis.<sup>5</sup>

The assessment of the answers regarding standard register's dynamism (cf. Q54, Q113A) reveals that the informants either perceive the standard register as (a) dynamic, i. e. in change, or (b) static. Informants, who belong to group (a) hold various factors co-responsible for the development of the standard. These include community-internal factors such as inter-generational differences between speakers and—more prominently—community-external factors, e. g. regional differences important for variation (cf. [1a]) and immigrational as well as the influence of English (cf. [2a]). These conceptualizations are accompanied by evaluations, that add up to the individual attitudinal picture of each informant. The influence of English is quite often assessed as (rather) negatively (cf. [2b]). Levelling processes as described in [1a] are, on the other hand, assessed as inevitable (cf. [1b]).

[1]


[2]


<sup>5</sup> For pragmatic reasons, we cannot discuss all major strands of conceptualization and evaluation found in the data but focused on selected examples as representative as possible for the data pool.

<sup>6</sup> Disclaimer: The given examples are meant to mirror the original quotes in colloquial (rather formal) German as authentically as possible. They thus reproduce German constructions on purpose and might violate English syntax and idiomatics.

However, the standard register does not necessarily bear dynamic potential. The reasons for such assumptions oscillate around the conceptualization of 'standard as invariant register' with various argumentative backgrounds (cf. [3], [4]). These static assessments are often expressed as positivistic facts. Therefore, explicit evaluations are substituted by more neutrally, indifferent answers as [4] illustrates.


Another set of questions deal with the future of standard [cf. Q54B] and dialect [cf. Q32]. In contrast to the first set of questions, these did not trigger clearly separate answers regarding conceptualization and evaluation. The answers, however, generally, include both components. With regard to standard, the indicated stasis on dynamic processes is transferred onto the future as well. In such cases, secure supra-regional comprehensibility is highlighted and implicitly evaluated as positive:

[5] I think this will be as it is forever. yes. I think that this is really necessary that really everybody can understand it. […] for sure the the writing, if you read it. you also read it in High German. NM0028\_YM

However, the future of standard is conceptualized as dynamic as well. Again, several factors are considered responsible for change. In example [6], language use of younger people is related to an opaquely conceptualized 'loss of linguistic quality'. However, processes of change are not necessarily evaluated as negative. In example [7], English is conceptualized as something productive.

[6] I think that this/. yes, I think that this is getting worse, if you then hear the young people talking, ahm, that is then really a completely extreme language. ((laughing)) but yes, I don't think this is good really, yes. RA0505\_YM\_H

[7] there is fairly a large amount of words that that simply from English, if now some/ there are English words that don't yet have a German equivalent. and basically the English words are "germanized". that way roughly I (could) describe it. […] I think this is basically good, but I think that is simply a matter of time. that is just that's just how it is. I don't think this is bad. AL0051\_YM\_H

(Meta-)communications that conceptualize the future of dialect [cf. Q32] dynamically highlight the influence of other registers of German or foreign languages as well as urbanization. Negative evaluations of these developments are frequent (cf. [8]). Other informants conceptualize the future of their local dialect as static and in all cases evaluate this as positive. They name certain parameters that remain constant, e. g. aspects of identity and tradition (intertwined with the "localness" of the variety) or the necessity to hand over dialect to future generations (cf. [9]).


Finally, the set of questions explicitly dealing with language contact as a factor for language change reveals the following insights: The vast majority of the informants (meta-)communicate linguistic influence of other languages, especially of English (cf. [10]). This is argued by referring to globalized markets, media usage and (travel-induced) mobility. Contact with other foreign languages, e. g. Turkish, Croatian and other Slavic languages was indicated as well (the latter group of languages especially from a historical perspective). Some comments also deal with different varieties of the German language. Other informants perceive influence of other languages. However, they either cannot point towards certain languages'responsible' for such contact (cf. [11]) or refer to more tangible (linguistic) categories (such as the written language cf. [12]). This indicates that both language contact and language change are not trivial concepts to informants and reveal individual hurdles to (meta-)communicatively grasp them: Contact is predominantly evaluated neutrally to negatively [cf. 10]:


Another group of informants does not indicate language contact induced language change. Here, views dominate that see no changing potential through other languages, that language does not change in general or that only the usage of certain varieties changes (cf. [13]).

[13] no, I don't think so. I think, because of the fact that in recent times a relatively high number of foreigners came to Austria. ahm the the people speak with the foreigners simply more High German. or just more elevated dialect. but otherwise I don't think so, no. TA0245\_YF\_L

These findings demonstrate that the informants' answers to the questions in focus can be categorized content-wise according to certain parameters for reasons of comparability. This includes the informants' conceptualization of the

<sup>7</sup> Here, the informant probably refers to the abolishment of the so-called German scripts (Fraktur and Kurrent, the "old script") and its replacement with Latin scripts (the "new script") in 1941.


aspects of language change and language contact tackled in the interviews and on the other, their evaluation of each of the aspects. (cf. Table 1).

Table 1: Aspects, questions and categories

The following chart (Figure 1) plots the conceptualization of each of the aspects against its evaluation. The comparison of the aspects allows for some general observations:

First, the majority of the informants conceptualizes language as dynamic. With regard to all three aspects (A1, A2, A3) that target the dynamism of either the standard register or the local dialect, 21 to 23 informants (61.76 % to 67.65 %) express that opinion.

Second, even a greater percentage (76.47 %), i. e. 26 informants, is convinced that other languages influence at least one register of German in Austria. However, not a single informant assesses that fact positively.

Figure 1: Correlation of conceptualization and evaluation for the four aspects

Third, the informants express clear evaluations more frequently if asked more detailed questions about the future of a certain register (A2, A3) than if asked, whether the standard register is subject to change at all (A1). In the first cases, three informants (at most) expressed neither a classifiable conceptualization nor evaluation of the respective aspect. In the latter, however, 12 informants (35.29 %) did not evaluate their conceptualization of the general dynamics of the standard register. This pattern can be traced back to the abstract concept of language change elicited by question Q54 and evaluated in question Q113A in comparison to the quite concrete expectations evoked by the questions regarding the future of certain registers. An alternative interpretation suggests that throughout the sample, informants generally do not evaluate the dynamic or static character of the standard register independent from how they conceptualize it. As soon as not specifically asked for possible future developments, they rather accept the standard register as it is. The second interpretation is supported by the fact, that also the positive, neutral and negative assessments of both a dynamic and a static conceptualization of the standard register are distributed evenly.

Fourth, and most important for the paper's content: Out of 34 informants, only a single one assesses the future of his/her local dialect neutrally, whereas 10 are neutral with regard to the future of standard register. This supports the idea of the dialect being the language of proximity, to which the speakers are emotionally attached (cf. Section 2). Eventually, the conceptualization of a register as static does not co-occur with a negative evaluation of its future. This indicates that language change is generally—like language contact—assessed rather ambivalently.

The question arises, whether the attitudes towards these aspects of language change and contact are distributed evenly throughout our sample or whether they correlate with the sociodemographic aspects captured in the survey (cf. Section 5). The results of the statistical test for correlations (Fisher's exact test) can be found in Appendix C. Here, only the most important results are highlighted.

Most importantly, the discrepancy described above regarding the future of the standard and dialect registers is reflected in the results: A2, the conceptualization and evaluation of the future of the standard register, does not correlate with any sociodemographic variable. A3, its counterpart regarding the local dialect, however, strongly correlates with the location and the communication degree, aswell as with the regional mobility of the informants. Informants with a higher communication degree or regional mobility seem to imagine the local dialect's future more positively, whereas such with a lower communication degree or regional mobility assessit negatively (and dynamically).Regarding the location, single places seem to be prone to certain attitudes towards the dialect's future. For instance, all informants from Weißbriach evaluate their dialect's future negatively, since it is considered as developing, whereas the majority of the informants from Neckenmarkt conceptualize it as dynamic, too, without lackingapositive attitude towards its future. Informants from Raggal, on the other hand, consider their dialect to be static and evaluate its future positively. Regarding A4, language contact and language influence, we cannot clearly distinguish between an age effect and an effect of education degree due to the survey design. However, younger informants tend to be more neutral with regard to influences from other languages and so do informants with a general qualification for university entrance.

These results demonstrate that certain attitudes and beliefs about language change and language contact overlap with areal, generational, educational and professional factors in our sample. However, the analysis has so far only captured single aspects of language change and language contact, even though these aspects probably interact. Attitudinal patterns should therefore be identified and described based on these interactions. At this stage, an exclusively qualitative study would proceed by grouping informants based on the above-described categorization, relying on reasonable and well-founded yet individual decisions of the scholar on which aspects are most important. This pilot study, however, explores an alternative approach using statistical analyses to form meaningful (yet as coherent as possible) groups of informants, who share certain attitudinal patterns and ideologically-loaded expressions. This can facilitate a more objective<sup>8</sup> , i. e. inter-individually reproduceable approach towards subsequent analyses.

#### 6.2 Explorative Statistical Analyses

To group our informants according to their expressed attitudes on several aspects of language change and contact, we conducted a TwoStep cluster analysis using SPSS 23 (for details cf. Appendix C). This procedure yields a three-cluster solution (cf. Table 2). Even though the silhouette coefficient (s = 0.2) suggests little structure within the data set, a qualitative interpretation reveals the hermeneutic value of the proposed clusters.


<sup>8</sup> That 'objectivity' has, of course, limitations on its own. The scholar's position can (probably) never be eliminated completely in such analyses. Here, this is the case in formatting and structuring content according to certain categories as well as in aggregating and setting variables as basis of statistical analyses (cf. Section 6.2).


(Continued)

Table 2: Cluster size, centers and importance of variables

Besides the number of cases (i. e. informants) within each cluster, Table 2 lists the variables from the most to the least important. V2 (future of the standard register) proves to be the most important determinant in the clustering process, followed by V4 (language contact and language influence). To characterize the clusters specifically, we take a closer look at the exact distribution of variables per cluster. With respect to V2 (i. e. the conceptualization of the standard register's future), the clusters deviate primarily with regard to the informant's evaluation of the underlying aspect A2 (cf. Figure 2): Most informants (11, i. e. 73.33 %) in Cluster 1 express a positive opinion, whereas Cluster 2 mainly comprises informants, who believe that the standard register will develop negatively (6, i. e. 60 %). The informants in Cluster 3 are rather neutral in this respect (7, i. e. 77.77 %).

Figure 2: Distribution of V2 (future of the standard register) specifications across the clusters

Apart from V2, the other variables extracted (and their specific combinations of both conceptualization and evaluation) play a decisive role in differentiating clusters as well. As to V4 (language contact and language influence, cf. Figure 3), Cluster 1 is characterized by a high degree of neutral acknowledgement of language contact as a driving factor in the change of German in Austria (12, i. e. 60 %). The informants in Cluster 2 predominantly acknowledge contact as well (9, i. e. 90 %). However, in contrast to Cluster 1, the majority assesses such influence on German in Austria negatively (5 out of 9, i. e. 55.55 %). This is also the case for Cluster 3: 55.55 % of the informants evaluates language influence as negative. The rest of the informants, however, does not acknowledge language contact at all.

Figure 3: Distribution of V4 (language contact and language influence)-specifications across the clusters

Regarding the third important variable in the cluster analysis, V1 (dynamics of the standard register), the clusters are characterized as follows (cf. Figure 4): 66.66 % of the informants in Cluster 1 conceptualize the standard register of German in Austria as dynamic and assess its dynamism either positively or neutrally. Similarly, the majority of the informants in Cluster 2 (8, i. e. 80 %) assesses that register as dynamic. However, half of them evaluate development of the standard negatively. Most informants in Cluster 3 (6, i. e. 66.66 %) conceptualize it as being static.

Finally, Figure 5 illustrates the composition of the clusters with regard to V3 (future of the local dialect). Cluster 1 is characterized by a positive evaluation of the local dialect's future (12, i. e. 80 %), whereas the informants in Cluster 3 mostly assess it negatively (6, 66.66 %). Cluster 2 remains rather undefined, even

Figure 4: Distribution of V1 (dynamics of the standard register)-specifications across the clusters

though a negative evaluation of a dynamic conceptualization forms the cluster's center. That indifference may be explained by the fact that V3 plays the least important role in the clustering process.

From these observations follows that the three resulting clusters do have hermeneutic value and help to aggregate insights into attitudinal patterns and language ideological positioning resources. According to the distributions of specifications, Cluster 1 can be strikingly conceived as the cluster of 'optimists', Cluster 2 as the one of 'standard register pessimists' and Cluster 3 of 'local dialect pessimists'.

Having seen that the attitudes of the informants overlap with their sociodemographic characteristics, the question arises whether the clusters reflect the sociodemographic setup of our sample, too. Therefore, we tested the clusters for correlations with the sociodemographic variables used. Out of these, only the informants' communication degree proved to significantly correlate with the clusters at the p < 0.05 level (χ <sup>2</sup> = 10.666, p = 0.048\* / V = 0.382, p = 0.125). An interpretation of Figure 6 reveals, that this correlation is likely due to the fact, that in Clusters 1 and 2 the majority of the informants work or worked (i. e. before being retired) in professions with higher (i. e. 3) communication degree (Cluster

Figure 5: Distribution of V3 (future of the local dialect)-specifications across the clusters

1: 10, i. e., 66.66 %; Cluster 2: 7, i. e. 0.7 %). That is true for only two informants in Cluster 3 (i. e. 22.22 %).

Generally, the clusters seem to be hardly predictable with the sociodemographic variables used. Thus, we conclude that the three-cluster solution actually does group the informants according to their attitudes towards language change and language contact.

Figure 6: Informants with professions of certain communication degrees in the three clusters

#### 6.3 Inter-individual Discourse Analyses

The described attitudinal patterns are grounded in a comparatively small set of conceptualizations and evaluations of aspects of language change and contact. If these truly hint at a shared ideology, one would expect certain themes and argumentation strands to recur in the informant's utterances of each cluster. As seen in Section 6.1, the informants most commonly explicate reasons or factors for language change in their answers to the analyzed questions. Therefore, this section more specifically questions whether similar factors are highlighted in the proposed clusters.

Methodologically, we observe how themes unfold within our conversational data by analyzing 'theme development cues' (Themaentwicklungshinweise, cf. Spitzmüller / Warnke 2011; Hausendorf / Kesselheim 2008). The methodological setting (i. e. semi-structured interview) already purports certain themes inherent to the posed questions, e. g. the future of the local dialect (cf. Section 5). These themes are then literally 'developed' by our informants—eventually by explicating reasons for language change. Following this approach, the informants' perspectives on the themes were extracted inductively and grouped systematically with regard to the given reasons/factors for language change, which are (both implicitly and explicitly) expressed. In total, eight decisive theme-strands could be extracted (cf. Table 3)<sup>9</sup> . However, some of them exclusively occur combined with others. These strands will be discussed in the following.

Strikingly, the most frequently mentioned themes consistently deal with language contact. Therefore, the informants conceptualize language contact as the major driving force for language change. This applies to the theme-strand labeled as'contact with prestigious languages', its sub-strands'media' and 'youth language' as well as both strands dealing with migration (i. e. 'contact caused by migration' along with sub-theme 'tourism' as well as 'variety contact caused by migration').

21 of 34 (61.76 %) informants mention 'contact with prestigious languages', the most prominent theme-strand found. Here, 'cultural' dynamics leading to language change were discussed and a strong connection of language (change) to a (vastly unspecified) concept of 'culture' was detectable. Especially English as foreign language is seen as a major influential factor in both private and professional domains. Correspondingly, English is assessed with regard to its status as'world language', i. e. lingua franca, also being of high importance for all kinds of (globalized) economic developmental processes. The informants do not conceptualize the contact with English as the contact of multilingual speakers or

<sup>9</sup> Side note: Each informant can be (and actually was) assigned to more than one theme strand, of course.

speakers of German and English. Instead of being understood asamedium for communication, English is rather described as a discreet organism. It seems to somehow slowly 'ingress' into 'our' culture, which typically is—for above reasons—hardly avoidable (cf. [14–16]).


Youth language and the media are frequently mentioned in the context of 'contact with prestigious languages'. Often, these two factors are connected contentiously. Approximately every fifth informant refers to youth language as a reason for language change. Arguments focus on intergenerational differences (cf. [17]) and are raised especially by older informants (older than 60 years), complaining that 'the young generation' does not care enough about 'proper' language use (with regard to 'standard' registers) or gradually loses competence in 'dialect' registers. Younger informants perceive their use of language as dynamic and influential yet affected by language use in the media (cf. [18]). 29.41 % of the informants indicate that language changes through media. The term often remains unspecified; informants hardly differentiate between 'classic' media like TVor radio and 'new' media like internet-based media channels. Many connect it to English, though (cf. [19]). Influences of other (standard) varieties, especially from Germany mediated through TV-stations discussed as well, though with a lower frequency.

[17] Well it will change as well, because because the youth again and again/ this just is cool and modern yes. One has to implement other terms/ with regard to the older people, I think, there will not change that much, but regarding the younger people, when they grew up once, when they are old, they will not speak like us. There will be a lot of changes NY0209\_OF\_L


The second theme strand involving language contact tackles 'language contact caused by migration' and was discussed by 47.06 % of the informants. However, there is a fundamental difference to 'contact with prestigious languages'. In these cases, informants trace influence back to speakers of other languages mainly referred to by the informants as'refugees' or (other)'immigrants'. Comments are sometimes masked in terms like "globalization" or sketched in allegedly global "movements" (cf. [20]) as a strategy to precisely denominate groups of speakers. Often, these statements are loaded metaphorically negatively; hence, with regard to language (change), 'immigration' is often perceived negatively (cf. [21]). This especially applies to languages of the Eastern European (Slavic languages and Hungarian) and Arabic areas bearing little prestige. Occasionally, this view is intensified by indicating that 'their own' language is maintained or passed on (in in-group-communications)—despite of divergent expectations to assimilate linguistically by the (clearly Austrian framed) surroundings (cf. [22]). In these cases, 'foreigners' or 'immigrants' are conceptualized as people exhibiting tendencies to resist to 'domestic' habits of language use or (unspecified normative) conventions of 'our' (linguistic) behavior.

[20] now because of the whole globalization lots of people are coming from there […] well, just because of the globalization, and and you can see it anyway, nowadays it takes you two hours from Africa to here; and that's and and and yes, and it intermixes WB0056\_OM\_L


Tourism is another cause for contact with speakers of other languages or varieties of German: However, only 8.82 % of the informants mentioned it as reason for language change. Linguistic accommodation for tourists is considered appropriate to a certain extent. In order to be a 'good host' and demonstrate a professional attitude, one must adapt to tourists linguistically. Thus, language contact with tourists promotes (near-)standard registers (cf. [23–25]). With regard to dialect registers, the situation appears rather vague, as both accommodation and its refusal are (meta-)communicated occasionally.


A third of the informants considers 'variety contact caused by migration' as a reason for language change (understood as contact between varieties of the same language). Due to mobility, relocating and/or commuting within Austria dialectal registers are "prone to" change in leveling processes. (cf. [26–28]).

[26] mobility will increase, just asI move away from [location] and thus carry my dialect to somewhere else, there will come other people to [location] from somewhere else and carry their dialect to us AL0031\_YM\_H


According to our informants, normative authorities also play a decisive role for language change, although to a lesser extent: Every fifth informant brought up this theme during the interview. However, the lines of argumentation on that topic are rather heterogeneous: Some informants express the position that 'our own' language (i. e. German) has to be 'preserved', which is something that the government and—put slightly unspecific—'everybody' (being member of an indistinct in-group) should take care of (cf. [29, 30]). Others emphasize the role schooling authorities play or should play (which according to the informants they do not satisfactorily, cf. [31]). That strand refers to the classic notion of standard registers being learned at schools properly (if not for the first time at all).


Six informants (17.65 %) 'naturalize' language change and therefore argue that language development is inevitable. These statements suggest that linguistic change is 'unstoppable', something that occurred, occurs in the present and will continue to occur in the future. Change in this context is not evaluated as being particularly threatening and is considered as'normal' and assessed as inherent to language (on constant basis, cf. [32–34]).


So far, this section has identified numerous themes, namely factors or reasons for language change. Table 3 illustrates the distribution of the factors for language change within the clusters: The numbers for each cluster (C1–C3) indicate how many of the informants provided that reason, e. g. 'contact with prestigious languages' was named by 10 informants in C1 (67 % of all informants in this cluster), 7 in C2 (70 % of all informants in this cluster) and 4 in C3 (44 % of all informants in this cluster).


Table 3: Theme strands correlated to the three clusters

The clusters behave quite differently, as the bars indicate. These represent the percentage of informants of each cluster arguing on a certain factor for language change. For example, they show that in C3 only 44 % of the informants mentioned contact with prestigeous languages, which is under the average of 62 % of all informants giving that reason for language change. The cells highlighted in grey mark the theme strands, which the informants of a certain cluster refer to more often than the total average (yellow bars). Cl ('the optimists', blue bars) over average naturalize language change as distinctive marker without major further conspicuities, being on average or quite near to average with regard to the other reasons. Cluster 2 ('the standard-pessimists', green bars) is over average on any topic dealing with language contact. In contrast to the other clusters, these informants hardly take something like 'naturalization' of language change into account—changes with regard to language predominantly seem be the result of contact situations of different kinds. The third cluster ('the dialect pessimists', red bars)—nomen est omen—over average consider variety contact caused by migration as a reason for language change (i. e. in C3 probably to a certain extent reduced to dialectal registers). Apart from that, C3 behaves quite on average, with the exceptions of cultural language contact, youth language and standardizing institutions, that score under total average.

# 7 Conclusion and Research Desiderata

Our findings in the initial CAQDAS-aided intra-individual content analysis indicate (Step 1, cf. Section 6.1) that the standard register is conceptualized heterogeneously as either dynamic or static. The grounding of such conceptualizations suggest that different focal points are crucial for such assessments. Dynamic processes of the standard register are associated with speakers of either German (community-internal factors for change, e. g. intergenerational differences) or other languages (community-external factors). However, static evaluations reify the standard register by associating it with normative, linguistic publications (e. g. dictionaries, grammar). Broadly speaking, these conceptualizations are also transferred to the assessments of the future development of the standard register.

In contrast to the dialect register, these future perspectives are not perceived as noticeably negative but as rather productive (e. g. through the integration of English words). Dynamic processes regarding the dialect register are quite often characterized as dangerous. This evaluation is based on conceptual connections of dialect register to abstract notions such as '(local) identity' and 'tradition'. In contrast, the informants stress (supra-regional) 'comprehensibility' as the dominant attribute of the standard register. The content analyses of Step 1 further revealed that language contact is conceived as a major factor in language change, which is never explicitly evaluated as positive but rather as negative or at best neutral. (Meta-)communication on language contact, including giving examples, however, posed a challenge to a number of informants. Although such answers remained on a rather implicit level, categorization processes in the course of the content analysis could uncover these attitudinal utterances—like those on the future of various registers of German in Austria—consistently comprise two aspects: conceptualization and evaluation.

This consistency throughout the data and several aspects of language change and language contact facilitated the implementation of a data-driven coding system by which each informant's attitudes towards these aspects were classified. Tests for correlations of these categories with sociodemographic information strikingly revealed that informants with a higher regional mobility and a higher communication degree evaluate the future of the (local) dialect more positively than less mobile informants in a non-communicative profession. Additionally, in certain places, particular evaluations of that aspect prevail. The other aspects, e. g. the future of the standard register, did not show comparable overlaps with the sociodemographic background of the according informants.

The coding system also provided ground for Step 2 (cf. Section 6.2), an explorative statistical analysis. In this step, the paper deviates from other qualitative analysis, putting the scholar in focus of intellectually grouping informants according to certain (individually ranked) parameters. In order to raise reproducibility, we clustered the data with a statistical model (TwoStep, via SPSS) to detect groups of informants with similar attitudinal patterns towards language change and language contact. The hermeneutical value of the three proposed clusters was then tested in thorough interpretations of the model results.

Surprisingly, the aspect 'future of the standard register' turned out to be the most relevant in the clustering, while the 'future of the dialect', even though emotionally discussed by some informants, turned out to be the least relevant. Cluster 1, the 'optimists', include informants that assess the future of the standard register rather positively, as they do with regard to standard dynamics. They perceive language contact and influence as a neutral yet decisive (linguistic) factor and evaluate the future of the dialect register as rather positive. Cluster 2 comprises of 'standard register pessimists', convinced that the standard register faces negative development. Language contact is evaluated similarly negatively. Regarding the dynamism of the standard register, both dynamic and static evaluations are balanced, as do evaluations on the future of the local dialect. The informants in Cluster 3 might be put as 'local dialect pessimists'. They are rather neutral with regard to the standard register, both against its alleged dynamic and future development. Language change and language influence does not concern these informants; most of them assess these parameters as rather static. It is the negative evaluation of the local dialect's future that informants of this cluster share.

Regarding the sociodemographic parameters, the informants are distributed across the clusters heterogeneously. Therefore, the clusters reflect attitudinal aspects rather than sheer sociodemographically-driven indications. However, the parameter 'communication degree' turned out to significantly correlate with the clusters. This finding might cautiously be interpreted as the higher the communication degree the more optimistic views on the variables tested.

In Step 3 (cf. Section 6.3), an inter-individual discourse analysis, the three clusters were interpreted as attitudinal patterns shaped by different language ideologies. In order to support this assumption, step 3 'dug deeper' into the attitudinal conceptualizations expressed by our informants and investigated, whether they correspond to the clusters defined in step 2. Corresponding to the paper's aim, it focused on the extraction of factors for language change. In our informants' opinion, contact proved the most decisive factor for language change. However, it requires further differentiation, namely into 'contact with prestigious languages' especially with English, 'language contact caused by migration' both involving bilingualism in German and another language, and 'variety contact caused by migration', i. e. contact of various varieties of German. The first two clearly stand out with regard to frequency. These factors differ from each with regard to essential conceptualizations:

The object of reference (i. e. the language itself vs. speakers of that language): Informants refer to the 'language' itself (e. g. English) and tend to conceptualize it as an organism, if the (individual) argumentation structure evolves around notions of prestige. Contrarily, they pertain to actual (yet quite indistinct) speakers of certain (foreign) languages if they discuss 'language contact caused by migration'. In these cases, they (indirectly) express (perceived) needs of delimitation against an (alleged and/or stereotypically constructed) out-group. Typically, this is the case if that group speaks a language with minor prestige: 'immigrants' are never conceptualized as speakers of German (e. g. from Germany) or English (e. g. from the United Kingdom) but as speakers of Eastern European (e. g. Slavic) or extra-European (e. g. Arabic) languages.

The evaluation of change: Prestigious languages such as English are conceptualized as'migrating' into German to a certain extent. More specifically: they leave their traces in German language use, e. g. through loanwords. However, that development does not generally 'devaluate' the informants' 'own' language. This is, on the other hand, the case for other, less prestigiously evaluated languages.

Besides these community-external factors for language change, we found numerous other aspects including, e. g. the role of standardizing institutions or a simple naturalization of language change. The three clusters found in step 2 differ from each other with regard to the factors for language change that the informants argue on: informants in Cluster 1 naturalize language change more often than informants from other clusters. Those in Cluster 2, however, highlight language contact both with prestigious languages and as a result of migration, whereas informants in Cluster 3 are concerned with variety contact. The heterogeneous and vague verbalization strategies on factors of language change could thus be mapped onto the clusters.

This paper set out to close a research gap by providing recent insights into layman conceptualizations alongside corresponding evaluations on both language change and language contact of German in Austria. It suggested a datadriven methodological approach to the operationalization of qualitative data for quantitative analysis, the results of which then served as the basis for further indepth qualitative explorations. This approach does not 'quantify' qualitative data for the pure sake of quantification. Instead, this paper intertwines these two strands in three steps that meaningfully complement each other. Each step was inspired by a methodological approach that has proven their worth in numerous studies before. Their combination broadens the perspectives with regard to both the single steps and the overall analysis at the same time. Thus, it promotes processes of analysis and fosters further perspectives on the data (types) themselves.

The results of this study indicate that layman (meta-)communications on the given topics are quite consistent insofar as they comprise both aspects of conceptualization and evaluation: e. g., languages and their varieties are generally either conceptualized as dynamic or stable. Based on this consistency, qualitative data can be operationalized for further quantitative analyses. Cluster analyses can then be applied in order to detect attitudinal patterns across the informants and thus increase the reproducibility of the analyses in comparison to traditional content analytical approaches, in which the scholar must define such groupings. In the following interpretation and a subsequent discourse analyses, these clusters proved to have hermeneutical value. These discourse analyses have, however, only focused on one aspect, namely factors for language change. Many other aspects such as frequent narratives and topoi remain to be investigated. If also these can be mapped to the clusters, this would support the hypothesis that each cluster represents certain attitudinal patterns that are shaped by distinct concurrent language ideologies.

Independent from these first results, thorough deepening analyses must be conducted subsequently in the future. Our study included only a small set of variables that must be enlarged by both including further questions of the interviews and further types of data (e. g. 'conversations among friends', cf. Koppensteiner / Lenz 2017). Evidently, such data enrichment inevitably will boost analytical complexity, exceeding the pragmatic range of this smaller study with piloting character. The mixed qualitative-quantitative research design proposed in this context will undoubtedly support such a more complex and comprehensive coverage of the interview data. It is, however, necessary to point out, that the quantification of qualitative data always needs to be reflected and requires careful interpretation. Most importantly, we need to keep in mind that the statistical analyses suggested in this paper only describe the data sample under analysis and that it is not valid to infer on a larger population from their results. They are helpful in dealing with a complex and interacting data. Since this study could illustrate and discuss the benefits and limitations of the proposed approach, we are confident that this approach is worth to be taken on larger amounts of data as well.

#### Bibliography


Handbuch der Sprachvariation. Language and Space – German. An International Handbook of Linguistic Variation. Vol. 4. Berlin / Boston: de Gruyter, 318–363.


# Appendix A – Informants


#### (Continued)


Table 4: Overview on informants included; location abbreviations: Allentsteig = AL, Neckenmarkt = NM, Neumarkt/Ybbs = NY, Raggal = RA, Tarrenz = TA, Taufkirchen = TK, Tux = TU, Weißbriach = WB; missing values are coded with 0

# Appendix B – Interview Questions

The square brackets illustrate the methodological intention to influence the informant with regard to terminology used as little as possible. The content, i. e. term was filled in by the explorer corresponding with what was said by the informant in the course of the interview.

#### Q32

Wie sehen Sie die Zukunft von [Dialekt-Terminus] in [Erhebungsort]? How do you see the future of [dialect-term used by informant] in [hometown of informant]?

#### Q32A

Würden Sie sagen, dass das eine Entwicklung ist, die auf [Erhebungsort] beschränkt ist? Warum?

Would you say that this development is restricted to [hometown of informant]? Why?

#### Q54

Würden Sie sagen, dass [Standard-Terminus] ebenfalls Einflüssen unterliegt bzw. sich verändert? Wenn ja: Können Sie mir Beispiele nennen? Welche Veränderungen stellen Sie fest?

Would you say that [standard-term used by informant] is also subject to influences respectively undergoes change? If yes: Please give examples. Which changes do you notice?

#### Q113A

Auch [Standard-Terminus] unterliegt einem ständigen Wandel. Wie beurteilen Sie selbst diese Tatsache?

[standard-term used by informant] is also subject to constant change. How do you judge this fact?

### Q54B

Wie sehen Sie die Zukunft von [Standard-Terminus]? How do you see the future of [standard-term used by informant]?

#### Q103

Verändert sich Ihrer Meinung nach die deutsche Sprache in Österreich durch den Kontakt mit anderen Sprachen und den Einfluss von anderen Sprachen? In your opinion, does the German language in Austria change because of contact with other languages and due to the influence of other languages?

#### Q103A

Welche Sprachen sehen Sie hier als besonders einflussreich an? Bitte nennen Sie Beispiele.

Which languages do you consider as especially influential? Please give examples.

### Q103B

Welche der Sprachformen des Deutschen, über die wir gesprochen haben, betrifft das Ihrer Ansicht nach?

According to your opinion: Which of these types of German we talked about does this affect?

#### Q95

Deutsch war in der Geschichte Österreichs immer schon in intensivem Kontakt mit anderen Sprachen. Welche waren aus Ihrer Sicht entscheidend?

Throughout the history of Austria, the German language has always been intensely in contact with other languages. From your point of view, which were decisive?

#### Q95A

Hatten Ihrer Meinung nach auch diese Sprachen Einfluss auf die deutsche Sprache in Österreich? Können Sie Beispiele nennen?

Do you think these languages also had influence on the German language in Austria? Can you give examples?

# Appendix C – Details on Statistical Analyses

a) Correlations of the aspect-specific conceptualization–evaluation pairs with the sociolinguistic variables (cf. Section 6.1)

In Table 5, only those correlations with a p-value of Fisher's exact test below 0.1 are included. Such a value indicates, that the possibility of a correlation between the two tested variables is 90 %.


Table 5: Correlation of the aspects in focus with the sociodemographic variables; \* indicates a significant correlation at the p < 0.05 level, \*\* indicates a highly significant correlation at the p > 0.01 level, \*\*\* indicates a very highly significant correlation at the p < 0.001 level, [VS] indicates a very strong effect size

#### b) Clustering procedure (cf. Section 6.2)

The TwoStep<sup>10</sup> cluster analysis in SPSS can readily be run with categorical variables, as in our case, continuous and mixed variables as long as they do not correspond to each other strongly. The interpretation of Figure 1 (cf. Section 6.1) suggests a correlation of the conceptualization and the evaluation most of the aspects. In three out of four cases (A2, A3, A4), Fisher's exact test indicates highly significant correlations with high to very high symmetric effect sizes. This means that in these cases the two variables (i. e. conceptualization and evaluation) very likely predict each other (cf. Table 6).


Table 6: Correlations of the sub-variables for each aspect, \*\*\* indicates a highly significant correlation at the p < 0.001 level

Due to the close correlation of three of these four conceptualization-evaluation pairs, they are combined according to the four aspects. The explorative statistical clustering is thus conducted with four instead of eight variables (with nine possible specifications each, cf. Table 7).<sup>11</sup>

<sup>10</sup> In the TwoStep clustering, the AIC (Akaike information criterion) was selected to automatically determine the number of clusters. Itwas chosen over the BIC (Bayesian information criterion), since its target model "is one that is specific for the sample size at hand" (Burnham / Anderson 2004: 299).

<sup>11</sup> Logically, 12 specifications would be possible. Since there are no cases of 'no conceptualization' but still 'evaluation' in the data, we do not list these three potential cases (i. e. n.a.-pos, n.a.-neu, n.a.-neg) in Table 7.


Table 7: Variables used for the two-step cluster analysis

Wolfgang Koppensteiner University of Vienna Department of German Studies Universitätsring 1 1010 Wien Austria wolfgang.koppensteiner@univie.ac.at

Agnes Kim University of Vienna Department of Slavic Studies Spitalgasse 2, Hof 3 1090 Wien Austria agnes.kim@univie.ac.at

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

Varieties of German in Germany

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440

# Verena Sauer

# The Inner-German Border and its Effects on the Former Dialect Border Areas in Bavaria and Thuringia

Abstract: The German reunification was one of the most important events of the 1900s. East Germany (GDR) and West Germany (FRG) were reunited as the Federal Republic of Germany after 40 years of separation. The effects of political borders on linguistic borders is an issue which is frequently discussed in German dialectology. The goal of my work is to show firstly how dialectal dynamics in the former border areas in South Bavaria and North Thuringia took place between 1930 and 2014. Secondly, I want to analyze how the extra-linguistic borders affect the perception of the inhabitants. Therefore, I compiled data from real-time and apparent-time analyses.

Keywords: Itzgründisch, dialect border areas, inner-German border, dialect geography, perceptual dialectology

Abstract: Eines der bedeutendsten Ereignisse der 1990er Jahre war die deutsche Wiedervereinigung. Nach 40 Jahren der politischen Isolation voneinander trat die Deutsche Demokratische Republik (DDR) der Bundesrepublik Deutschland (BRD) im Jahr 1990 bei. In diesem Zusammenhang wurde innerhalb der Linguistik vor allem diskutiert, welche Auswirkungen die politische Grenze auf die dialektalen Grenzen hatte. Das Ziel der vorliegenden Untersuchung ist es, zum einen auf die dialektale Dynamik innerhalb der itzgründischen Sprachlandschaft in den Jahren von 1930–2014 (Real-Time-Vergleich) einzugehen. Zum anderen wird auch die Dialektwahrnehmung von Sprechern aus drei Altersgruppen aus dem ehemaligen Grenzgebiet untersucht (Apparent-Time-Vergleich).

Keywords: Itzgründisch, Grenzdialekte, innerdeutsche Grenze, Dialektgeographie, Wahrnehmungsdialektologie

#### 1 Introduction

This project<sup>1</sup> was conceived during my time working as a PhD student at the TU Dresden and the CAU Kiel. As a former inhabitant of the so-called 'Itzgründisch' dialect area, I learned to speak 'Itzgründisch' as my first variety. For my PhD project, I worked closely with native speakers from these border regions in Sonneberg (South Thuringia) and Coburg (North Bavaria).<sup>2</sup>

The Sonnebergers and the Coburgers<sup>3</sup> belong to the same cultural area (Franconians), the same economic area (toy industry) and denomination (protestants). Their common dialect 'Itzgründisch' is an East Franconian variety (cf. Steger 1968: 333–399). However, after about 800 years of shared history, the Sonnebergers and Coburgers were separated from each other for about 40 years (1949–1989) and were not allowed to communicate (cf. Schwämmlein 1999: 95). In the context of German reunification, the political border between the two areas no longer existed (cf. Schwämmlein 1999: 95). Thus, in the 1990s a joint project took place between scientist of the Thüringisches Wörterbuch and the Sprachatlas von Nordostbayern in the former border areas (cf. Fritz-Scheuplein 2001; Harnisch 2015; Krämer-Neubert 1995; Lösch 2000) and speech recordings were produced. Besides that, very few studies<sup>4</sup> investigated the impact of the perception of the speakers on their dialect and how this affected their way of speech.

The main issues addressed in this paper are: a) the former history and structure of the 'Itzgründisch' dialect area before the separation (until 1961), b) the present dialect-geographical structure of 'Itzgründisch', c) the present perceptual dialectological isoglosses and d) the synthesis of both structures.

The first question is: What effect did the inner-German border have? The goal of my work is to show how dialectal dynamics in the former border areas occurred between 1930 and 2014. Firstly, I give a brief overview of the history of the 'Itzgründisch' dialect area in Section 2 and secondly, I critically summarize the present state of affairs in the field of research. I then briefly discuss the role of real- and apparent-time comparison in linguistic research today and show some new approaches to combine both methods with each other. In Section 5 I will present the methodology and the corpus of the current study. Subsequently, I ask in Section 6: How do the extra-linguistic borders affect the perception of the inhabitants? Iinvestigate the perception of 'Itzgründisch' in people's minds using

<sup>1</sup> This paper is based on the larger study Sauer (2018) that is written in German and has been published already. From the point of view of a wider dissemination, this paper written in English reports the most important findings.

<sup>2</sup> For further information see Figure 1: The 'Itzgründisch' dialect area.

<sup>3</sup> The rural as well as the urban public are meant as 'the Sonnebergers and the Coburgers'.

<sup>4</sup> Harnisch investigated some perceptual dialectological aspects in his work (cf. Harnisch 2008: 93).

a questionnaire on the social data of my informants (Section 6.1) and speech samples to identify the perceived dialectal characteristics (Section 6.2).<sup>5</sup> This task is in apparent-time and covers three different age groups. Furthermore, I comparatively analyze the dialectal way of speech of the former East German and West German inhabitants in Section 6.3. This research is based on a variant analysis in real-time. Finally, the conclusion in Section 7 draws together the objective, dialect geographical space and the subjective, perceptual dialectological space. In this way I want to solve an important desideratum clarifying the role of the political border on the objective and subjective structure of 'Itzgründisch'.

#### 2 'Itzgründisch': A Brief Overview

The 'Itzgründisch' dialect area includes the former West German region Coburg (Bavaria) in the south and the former East German region Sonneberg (Thuringia) in the north. Adjacent regions are Hilburghausen (Thuringia) in the geographical west and north, whose inhabitants speak so-called 'Hennebergisch', which is another East Franconian dialect in Thuringia, and Kronach (Bavaria) in the geographical east, whose inhabitants speak an Upper East Franconian dialect, which is an East Franconian variety as well.<sup>6</sup> The research area 'Itzgründisch' belongs to the Upper German region (cf. Steger 1968: 333–399).

In the northeast, the 'Itzgründisch' dialect area abuts on the Middle German region. The Rennsteig, a mountain region in the Thuringian Forrest, as acted as a natural border in earlier times, separating the speakers. Thuringian varieties were in the north of the Rennsteig and in the south there are East Franconian dialects. In the south of the research area 'Itzgründisch', the so-called 'Coburger Schranke' is situated, which separates the Lower East Franconian dialect area (e. g. Coburg and Sonneberg) from the Upper East Franconian dialect regions (e. g. Kronach). Between the districts Hildburghausen and Sonneberg / Coburg so-called 'Südhennebergische Staffelung' runs along. The 'Hennebergisch' dialect area is on the western side (at the center Hildburghausen) and on the eastern side the 'Itzgründisch' dialect area (at the center Sonneberg and Coburg). Because of its special linguistic-geographical position and the different influences of the ad-

<sup>5</sup> Throughout this paper, the term 'perceived dialectal characteristics' will be used to refer to linguistic and non-linguistic characteristics that come in the minds of informants while hearing the speech samples.

<sup>6</sup> Please also see Figure 1: The 'Itzgründisch' dialect area.

Figure 1: The 'Itzgründisch' dialect area. In Figure 1 the different political areas (districts) are marked by different hatching lines

jacent dialect areas, it is called a 'Binneninterferenzraum' 7 (cf. Klepsch / Weinacht 2003: 2769).

Between 1949 and 1989/90 two German states (GDR & FRG) existed which were hermetically sealed by the inner-German border since 1961<sup>8</sup> . As a result, the dialect area 'Itzgründisch' was separated into two parts and the inhabitants of Sonneberg were forbidden to get in touch with their neighbors in Coburg for 40 years. The inhabitants of East Germany (GDR) in general were not allowed to be in contact with their former neighbors in West Germany (FRG) or other people from so-called 'capitalist foreign countries' (cf. Grimm 1983: 9–36).

Before the separation, the Sonnebergers and Coburgers shared a history of over 800 years, beginning in the eighth century. Both cities belonged to the diocese of Würzburg and were therefore influenced by the East Franconian culture. The toy industry was the most important economic sector in this area until the Second World War and can be traced back to influences of the city of

<sup>7</sup> 'Binneninterferenzraum' means a language area, whose structure is highly influenced by the neighboring language areas. 'Itzgründisch' or even the Franconian dialect area as 'Binneninterferenzräume' are not clearly defined by linguistic similarities but rather by their linguistic contrasts to other language areas.

<sup>8</sup> Since the 1950s the communication between the inhabitants of GDR and FRG was rarely possible, and then only under onerous conditions.

Nurnberg in the fourteenth and fifteenth century. From the eighteenth century on, Sonneberg was the center of the toy industry worldwide and new jobs (mainly in homework) were generated for both the Sonnebergers and Coburgers. In the course of the power take-over by the National Socialists, the war and subsequently the division of Germany, Sonneberg lost its leading role in the toy industry and became a 'grey city' (cf. Schwämmlein 1999: 95). At the same time, Coburg, as a member of the FRG, gained more economic power. Since the 1950s, as a result of so-called 'Wirtschaftswunder' and the establishment of large companies such as BROSE and HUK Coburg, Coburg has developed into an economically prospering location (cf. Habel 2009: 100–126; Schrinner 2012: 3).

Following the German reunification, Sonneberg and Coburg entered into a formal city partnership in 1990 and, in 2014, Sonneberg became the first non-Bavarian member of the Europäische Metropolregion Nürnberg ('European metropolitan region Nuremberg'). Today, many Sonnebergers commute to Coburg or other Bavarian cities for work, but only a few Coburgers go to Sonneberg or Thuringia. This results from the weaker economy of Thuringia in comparison to the Bavarian economy. Nevertheless, both areas belong together in terms of history, culture and economy (cf. Grimm 1983: 9–36; Habel 2009: 100–126; Schrinner 2012: 3). The aim of this study is to clarify whether or not the speakers of both areas use the same variety and how they perceive themselves and theirway of speech.

#### 3 State of Research

The first serious discussions and analyses of dialect areas on the former inner-German border emerged between 1991 and 1994, in the course of the joint DFGproject Erhebungen zur Dialektsituation im thüringisch-bayerischen Grenzgebiet ('Studies on the dialect situation in the Thuringian-Bavarian border region') between researchers of the Thüringisches Wörterbuch ('Thuringian Dictionary') at the University in Jena (Thuringia) and researchers of the Sprachatlas von Nordostbayern ('Linguistic Atlas of North-East Bavaria') at the University in Bayreuth. Their research area included 42 locations along the inner-German border, situated in Thuringia and Bavaria. The two research groups used different methods and pursued different research goals, which will be described in the following section.

The researchers from the University in Jena examined the use of the different varieties in the everyday lives of the inhabitants. Their goal was to analyze how non-linguistic factors affected the speaking pattern and language in those areas, which were divided by the former inner-German border (cf. Lösch 2000: 156). Therefore, they asked so-called 'Ortsexperten' 9 to classify how the inhabitants of their village speak. The Ortsexperten used population registers to write down the typical way of speech for every inhabitant. They could choose between three different varieties (Dialektsprecher 'dialect speakers', Sprecher der Mischsprache 'sub-standard speakers' or Hochdeutschsprecher 'standard speakers'). Lösch found that 49 percent of the Thuringians and only 39 percent of the Bavarians claimed to use the dialect as everyday language. Besides that, 43 percent of the Bavarians and 40 percent of the Thuringians claimed to use the default language as everyday language (cf. Lösch 2000: 160).<sup>10</sup> As a basic trend, he noted that the dialects are abandoned in favour of the default language (cf. Lösch 2000: 163). In addition, he identified that: "Durch die Grenzziehung ist die grundsätzliche sprachliche Einheit 'Thüringisch' wie auch im Itzgründischen praktisch kaum beeinträchtigt worden" (Lösch 2000:163; translation VS)<sup>11</sup>.Amajor criticism of Lösch's work is firstly that he used 'Ortsexperten' to collect his data and did not consult the speakers directly. Secondly, his data is only based on the views of the 'Ortsexperten', therefore it is not objective data (cf. Herrgen 2001: 223). Furthermore, nothing is said about the structure of the dialects – whether they separated as a result of the isolation of the speakers or not.

The main goal of the study at the University Bayreuth was to determine the dialect competence of the speakers in the border region. Therefore, the researchers formed four different age groups (born after 1920; born circa 1940; born circa 1955; born circa 1975) and interviewed speakers from the former West and East German border regions. Furthermore, the analysis included a translation task in which the informants had to translate 25 standard phrases into their dialect.<sup>12</sup> The project was continued in 2005 by Rüdiger Harnisch from the University in Passau. He and his team analyzed and evaluated the collected data from the 1990s. Overall, the researchers concluded that many of the speakers, in Thuringia as well as in Bavaria, used a semi-standard or a standard variety instead of the dialect. In addition, the dialectal forms of the Bavarian and the Thuringian

<sup>9</sup> Lösch, Head of the Research Group gave the following explanation of 'Ortsexperte': "[…] Ortskundige mit verläßlichen Kenntnissen über die Ortsverhältnisse und die gesamte Einwohnerschaft" (Lösch 2000: 158; translation: VS), "people who know the area very well and have reliable knowledge about the inhabitants".

<sup>10</sup> Lösch admitted: "Der hohe Prozentanteil der Hochdeutschsprecher in beiden Populationen ergibt sich daraus, daß in den Grundgesamtheiten auch die Vorschulkinder, die Schüler und die Jugendlichen enthalten sind. Sie wurden von den Ortsexperten meist pauschal als Hochdeutschsprecher eingeordnet" (Lösch 2000:160; translation: VS), "the high percentage of standard speakers in both populations is caused by the fact, that all school children were defined as standard speakers by the Ortsexperten". For additional information please see Sauer (2018).

<sup>11</sup> "The essential unity of the variety Thuringian aswell as'Itzgründisch' is hardly affected by the former inner-German border."

<sup>12</sup> Further information given in Harnisch (2015) etc.

speakers varied. For instance, the use of the front-rolled trill [r] was most common in Bavaria, but was increasing in Thuringia (cf. Harnisch 2008: 83–97). Harnisch states:

"Das Projekt konnte also zeigen, dass die innerdeutsche politische Grenze in doppelter Weise auch zur sprachlichen Grenzbildung geführt hatte: durch Schaffung neuer faktischer Dialektgrenzen, die kongruent mit der politischen wurden, und durch die Projektion alter mentaler Sprachlandkarten auf neue politische Grenzen zwischen hüben und drüben". (Harnisch 2015: 236)<sup>13</sup>

These results can be explained using the assumption that the speakers reoriented themselves towards their own political environment, e. g. the Coburgers focused more closely on their Bavarian neighbors and the Sonnebergers on their Thuringian neighbors (cf. Harnisch et al. 2008: 209). In the course of this, both increasingly used semi-standard varieties and aligned their dialectal forms.

In 2001 Fritz-Scheuplein published the research results of her dissertation project Geteilter Dialekt? Untersuchungen zur gegenwärtigen Dialektsituation im ehemaligen deutsch-deutschen Grenzgebiet ('Divided dialect? Studies on the current situation in the former German border regions'). Her research area included fourteen places in Thuringia and Bavaria next to the former inner-German border. Fritz-Scheuplein consulted two different age groups (average age: 72 and 32) as well and used a fully structured questionnaire for the group interviews (cf. Fritz-Scheuplein 2001: 54). Similar to the result of Harnisch, she pointed out that the dialects are abandoned in favour of the default language in the group of the younger informants. Nevertheless, there were no differences between the dialectal forms of the Thuringians and the Bavarians. In contrast to Harnisch, Fritz-Scheuplein argues that:

"Basisdialektal sind die alten wie die jungen Sprecher der beiden Grenzregionen nur schwer auseinanderzuhalten. Erst bei der Ersetzung des Basisdialekts durch eine höhersprachliche Varietät wird ein Ost-West-Unterschied deutlich, der wohl auf die unterschiedlichen Lebensumstände, verursacht durch die ehemalige innerdeutsche Grenze, zurückzuführen ist" (Fritz-Scheuplein 2001: 195)<sup>14</sup> .

Furthermore, it transpires that there are different scientific findings exploring whether the inner-German border did have an effect on the dialects of the inhabitants in the border regions or not. During current research, new methods will

<sup>13</sup> "The study has shown that the former inner-German border have led to the formation of a linguistic border in two ways: by creating new dialect borders, which are congruent with the political border and by creating mental borders." (translation: VS)

<sup>14</sup> "The dialectal way of speaking of the older and younger inhabitants of the border regions is hard to distinguish from each other. Only if they change into sub-standard, differences between the speaker from GDR and FRG can be detected caused by the former inner-German border." (translation: VS)

be used to answer this question and to explain which of the introduced hypotheses are confirmed.

Much of the current literature on dialectal border regions (cf. Palliwoda / Sauer / Sauermilch 2019; Palliwoda 2019; Sauermilch submitted) focuses particularly on the speakers' perception of dialects. This development started with the studies of Anders (2010) and Purschke (2011). Palliwoda (2019) focusses on the former inner-German border and the spatial localization and valuation of speech samples. She uses so-called 'priming method', which sets different primes while the informants listen to the speech sample. The goal of this study is to examine whether primes affect the localization and valuation of the informants or not.

Sauermilch (submitted) combines these two approaches by collecting objective and subjective data. Her study took place in the Low German language area and includes three different age groups (born between 1940 and 1945; born between 1970 and 1975; born between 1990 and 1995).

However, there have been no (completed) studies so far which compare differences in the objective space (in real-time) and the subjective space (in apparent-time). Therefore, the present paper represents a research desideratum.

#### 4 Real-time and Apparent-time Comparisons

This study is based on a real-time comparison in order to investigate the objective structure of the 'Itzgründisch' dialect area and an apparent-time comparison to evaluate the subjective structure of the speaker's perception. In the following section, I will provide a definition of both methods and identify critical issues. To conclude, I will describe the 'integrating method', which builds the basis of the current study and combines the real-time and the apparent-time method with each other.

Labov defines "observations in real time, that is, to observe a speech community at two discrete points in time" (Labov 1994: 45–46). The speech of different age groups at different times is the basis for a real-time comparison. The apparent-time analysis is different. Here, all the data is collected at a specific date from speakers of different age groups. This can be defined as a "distribution of linguistic variables across age levels" (Labov 1994: 45–46). To evaluate language change it is difficult to only use apparent-time data. Sometimes so-called 'agegrading', "a regular change of linguistic behavior with age that repeats in each generation" (Labov 1994: 46) can lead to misinterpretations. For example, speakers from the younger age group can display differences in their way of speech in comparison to speakers of the older group. This can be interpreted as a language change, although it is simply 'age-grading'. Of course, it is not always possible to conduct a real-time study, e. g. because of a lack of comparable data from different points in time.

Therefore, it can be helpful to use a combination of written and oral data or different types of text, like the 'panchronic approach' by Lenz (2012). Lenz understands language as a highly dynamic system and combines speech material from several centuries, different regions and written as well as oral sources. As a result, it is possible to show a geographical, a medial-communicative and a historical distribution of dialectal variants.<sup>15</sup>

In the current study I used an 'integrating approach' to combine objective realtime data with subjective apparent-time data. The real-time data includes speech samples from different points in time (1930s, 1960s, 1990s and 2014) and speakers of different age groups. This data is used to investigate the dialect-geographical structure and isoglosses of the 'Itzgründisch' dialect area. Furthermore, perceptual dialectological data was collected to evaluate the subjective structure of 'Itzgründisch', e. g. the dialect perception of the inhabitants. In a second step, both structures can be compared and evaluated to determine whether they contain similarities. If the objective structure of 'Itzgründisch' does not show any isoglosses between Sonneberg and Coburg, and the informants cannot perceive any difference between the speech samples from Sonneberg and Coburg, it can be assumed that there was no dialectal change that lead to a split of the 'Itzgründisch' dialect area. The 'integrating approach' is a helpful method in avoiding misinterpretations and understanding dialectal change better.<sup>16</sup>

#### 5 Methodology

In previous studies (cf. Fritz-Scheuplein 2001; Harnisch 2015; Krämer-Neubert 1995; Lösch 2000) the dialectal change at the political border was investigated mainly using dialect-geographical methods, such as questionnaires and translation tasks. Recently, subjective data and perceptual dialectological methods have been developed to explore the border dialects (Palliwoda 2014; Sauer 2017; Sauermilch 2016). Data from various points in time (so-called 'real-time comparison') or from different age groups at one point in time (so-called 'apparenttime comparison') were collected. The methodological design of the present study integrates subjective data from an apparent-time comparison and objective data from a real-time comparison. I used objective data for the dialect-geographical analysis in order to gain insights into the intra-linguistic structure of the 'Itzgründisch' dialect area. In addition, I collected subjective data which

<sup>15</sup> Further information given in Lenz (2013).

<sup>16</sup> Further information given in Sauer (2018).

includes an investigation of the extra-linguistic structure such as the auto stereotype or the origin of the speakers (former East or West German area).

#### 5.1 Apparent-time Analysis

Sixty-seven informants from different border areas in Coburg (Bavaria) and Sonneberg (Thuringia) were recruited for this study. Based on their age, they were divided into three groups: all informants of the older group were born between 1933 and 1949, thus their language socialization was finished before the total separation in 1961 and therefore they are called 'pre iso' <sup>17</sup>. The informants of the middle group, born between 1955 and 1971, were socialized during isolation and are called 'iso' <sup>18</sup>. The members of the last and youngest group were born between 1987 and 1996 and are called 'post iso' <sup>19</sup> (see Table 1).


Table 1: Overview informants by origin and sex (apparent-time analysis)

The data for the apparent-time analysis were based on a dialect-sociological questionnaire and non-linguists' assessments of four different speech samples of speakers from Coburg, Sonneberg and the neighboring Kronach, which does not belong to the 'Itzgründisch' dialect area. The speech samples were used to assess perceived dialect features. Therefore, I analyzed the salience of single features and feature clusters and investigated potential salient features. Besides these factors, the informants were asked to localize the sample.<sup>20</sup>

<sup>17</sup> 'Pre iso' stands for 'pre isolation' as the informants were socialised before ('pre') isolation ('iso').

<sup>18</sup> 'Iso' stands for 'isolation' as the informants were socialised during isolation ('iso').

<sup>19</sup> 'Post iso' stands for 'post isolation' as the informants were socialised after ('post') isolation ('iso').

<sup>20</sup> For additional information please see Sauer (2018).

# 5.2 Real-time Analysis

The real-time analysis is based on five different corpora covering a period from 1934 to 2014. The oldest corpus, the Niederlöhner corpus (1937), was collected by Niederlöhner in 1934/1935 in 178 villages in South Thuringia and North Bavaria (cf. Niederlöhner 1937: 184). All participants were school children in seventh or eighth grade. His questionnaire included 500 single lexemes focusing mainly on the vocalism. Unfortunately, no speech recordings were made, thus only the written data and dialect maps are available today. Besides that, recordings from the DR and the Zwirner corpora were included in the real-time analysis. The DR corpus was created by researchers of the GDR and the integrated recordings were produced in 1964 (cf. IDS 2018a). The recordings of the Zwirner Corpus are from 1957 and were made by researchers of the FRG (cf. IDS 2018b). A total of ten samples were taken from the DR- and Zwirner corpora. Recordings for this study were also examined from the SPRiG<sup>21</sup> corpus, which was collected in 1992 and 1994. A sample of 20 recordings was chosen for the present study. To cover the present as well, I collected 67 speech recordings in 2014 for the apparent-time analysis, as mentioned above.


Table 2: Overview informants by origin and sex (real-time analysis)

The real-time analysis consists of 97 recordings from four different corpora and written data from the Niederlöhner corpus (see Table 2). It covers a period of 80 years. The informants were asked to complete a task in which they translated standard words and phrases into their dialect. The comparison included about 130 single lexemes and involves 146 linguistic features, thereof 124 relating to vocalism and 22 variables relating to consonantism.<sup>23</sup>

<sup>21</sup> SPRiG stands for 'Untersuchungen zur Sprachsituation im thüringisch-bayerischen Grenzgebiet' (Linguistic studies on the language situation in Thuringian and Bavarian border regions).

<sup>22</sup> For the Niederlöhner corpus there is only written evidence but no speech recordings of the speakers.

<sup>23</sup> For additional information please see Sauer (2018).

# 6 Results

#### 6.1 Dialect-sociological Results

A questionnairewas used to assess the dialect-sociological structure of my corpus (Sauer corpus). The questions were divided into three categories: language biography (A), regional loyalty (B) and language knowledge (C). In this paper, I focus on category B and show some results of the represented regional auto stereotypes. In this context I ask the question:

#### How important is the designation of origin Franconian to you?

Answer options: very important / important / partially / unimportant / very unimportant /Iam not a Franconian (see Table 3).


Table 3: Designation of origin Franconian: Overview of answers given by the informants (apparent-time analysis)

Considering the political classification of the informants, the Sonnebergers belong to Thuringia and are therefore Thuringians. The Coburgers, however, belong to Bavaria and thus are Bavarians. In terms of a linguistic classification, both speaker groups belong to the East Franconian language area. Additionally, in terms of culture, both share Franconian traditions (e. g. Osterbrunnenfest 'Easter fountain celebration', Kirchweih 'fair' etc.) (cf. Sauer 2018: 64). Table 3 provides an overview of the results according to the regional auto stereotype Franconian. It is apparent that most of the Coburgers (28 of 33) consider being a Franconian important or very important. Only three participants believe that this designation of origin is partially important or unimportant. Those two who answered 'I am not a Franconian' were born in the area of Sonneberg but raised in Coburg and still live there. Therefore, they do not consider themselves Franconians. Table 3 shows that the Sonnebergers identify themselves as Franconians aswell, although politically they are Thuringians. Compared with the data from the Coburgers, it seems less important to the Sonnebergers to be Franconian. However, at least four participants consider it important or very important to be called Franconian and a further nine informants answer that it is partially important to them. In total, 17 out of 34 Sonnebergers identify themselves as Franconian. Mostly they belong to the middle or younger speaker group. Regarding the older speakers, it appears more important to be a Thuringian than a Franconian. This is perhaps because they equate their political identity with their cultural identity and do not distinguish between the two.

The results of this analysis show that the Coburgers think of themselves as Franconians and at least one third of the Sonnebergers also do. Data from this Table 3 can be compared with the data in Table 4 and Table 5 which show the importance of the designations of origin Bavarian and Thuringian.


Table 4: Designation of origin Bavarian: Overview of answers given by the informants (apparent-time analysis)


Table 5: Designation of origin Thuringian: Overview of answers given by the informants (apparent-time analysis)

Only a small number (8 of 33) of those interviewed Coburgers consider it very important or important to be referred to as a Bavarian. The most surprising aspect of the data is that nine of the Coburgers believed that they are not Bavarian. The reason for this opposition can be traced back to the historical inheritance dispute between Bavaria and Franconia 200 years ago. Franconian territories fell to Bavaria and thus the Franconians lost their territorial dominion. This might be the reason<sup>24</sup> why some of the Coburgers do not identify themselves as Bavarians.

However, the Sonnebergers have a significantly close relationship to Thuringia: 23 of 34 informants find it very important or important to be a Thuringian.<sup>25</sup> Therefore, it can be stated that most of the Sonnebergers identify as Thuringians, but some also see themselves as Franconians. Some of the informants from Coburg (7 of 33) and from Sonneberg (2 of 34) have a connection to the Thuringian or Bavarian identity, likely resulting from reasons related to family. For example, some Coburgers were born in Sonneberg and therefore identify themselves as Coburgers and Sonnebergers.

In summary, these results show that both Sonnebergers (Thuringia) and Coburgers (Bavaria) refer to themselves as Franconian. While the middle and younger Sonnebergers have a strong connection to Franconia, the older Sonnebergers also relate to Thuringia. Nonetheless, the Franconian identity connects both speaker groups with each other.<sup>26</sup>

#### 6.2 Perceptual-dialectological Results

The 67 informants of the study were asked to localize four different speech samples from Coburg, Sonneberg and Kronach. The recordings had a length of approximately thirty seconds. All speakers were inhabitants of the area, had been for several generations and had never left this region for more than two years. The first speech sample was recorded in Stockheim / Kronach<sup>27</sup>,aneighboring region to the 'Itzgründisch' dialect area. The speaker uses an Upper East Franconian variety. The three other samples are all Lower East Franconian dialect and recorded in Hassenberg / Coburg<sup>28</sup> (speech sample 2), Malmerz / Sonneberg<sup>29</sup> (speech sample 3) and in Neustadt / Coburg<sup>30</sup> (speech sample 4).

As Figure 3 shows, the speech sample 1 from Stockheim / Kronach has been correctly localized by 33 of 67 informants as a Franconian variety from Kronach. It is highly probable that they recognize salient dialect markers for the Upper East Franconian dialect area: The informants noted lexical markers such as Bub 'boy'

<sup>24</sup> Besides the assignment of Coburg to the former Duchy of Sachsen-Coburg-Gotha from 1826 to 1918 must be kept in mind.

<sup>25</sup> In this context it is interesting, that only the older Thuringian speaker group (6 out of 16 informants) call their dialect a Thuringian variety instead of a Franconian variety. The middle and younger Thuringian speaker said consistently, that their dialect is Franconian.

<sup>26</sup> For additional information please see Sauer (2018).

<sup>27</sup> See Figure 2.

<sup>28</sup> See Figure 4.

<sup>29</sup> See Figure 6.

<sup>30</sup> See Figure 8.

Figure 2: Localization of speech sample 1 (Stockheim / Kronach)

Figure 3: Speech sample 1 (Stockheim / Kronach): Upper East Franconian dialect

instead of Junge 'boy' or Kappe 'cap' instead of Mütze 'cap'. They state that the forms Bub and Kappe are often used in Kronach but are less common in Sonneberg and Coburg. Besides that, they also find phonetic markers such as the monophthongizations [ɑ] instead of standard [ae] in the form klein 'little'. In addition, they recognize the increase of the dialectal sound [i:] instead of standard [e:] in Schnee 'snow'. Nevertheless, another 14 informants believe they heard a Franconian variety from Coburg. Therefore, it can be noticed that there are many similarities between the Kronacher and the 'Itzgründisch' variety.

The data in Figure 5 shows that the localization of the speech sample from Hassenberg / Coburg was not as easy for the informants as the localization of speech sample 1. 22 of 67 Informants recognize a Thuringian variety from

Figure 4: Localization of speech sample 2 (Hassenberg / Coburg)

Figure 5: Speech sample 2 (Hassenberg / Coburg): Lower East Franconian dialect

Sonneberg, another 20 informants state correctly that they heard a Franconian variety from Coburg and 15 say that they noticed a Franconian variety from Sonneberg. Thus, two interesting aspects can be found: Firstly, the informants recognize that it is an 'Itzgründisch' variety from Sonneberg or Coburg but not from Kronach. Secondly, the informants are unsure about categorizing the Sonnebergers variety as Thuringian or Franconian. This is due to the fact that Sonneberg politically belongs to Thuringia, but dialectically belongs to the East Franconian language area. Therefore, it is difficult for the informants to define clearly.<sup>31</sup> In this context, the speech sample from Hassenberg / Coburg was lo-

<sup>31</sup> Further information given in Sauer (2018).

cated within the 'Itzgründisch' dialect area by most of the informants. But it is also apparent from Figure 5 that it is almost impossible for the informants to assign the sample to Sonneberg or Coburg.

Figure 6: Localization of speech sample 3 (Malmerz / Sonneberg)

Figure 7: Speech sample 3 (Malmerz / Sonneberg): Lower East Franconian dialect

Figure 7 shows that it is difficult for the informants to identify the speech sample from Malmerz / Sonneberg, too. Most of the informants (22 of 67) state that they hear a Franconian variety from Coburg and another 17 informants identify it as a Thuringian variety from Sonneberg. Only four informants localize the recording correctly as a Franconian variety from Sonneberg. Here, it is especially noticeable that six informants are incapable of doing the task at all. Some of them explain that the used variety was not a dialect but a regional language. Nonetheless, most informants are able to identify it as an 'Itzgründisch' speech sample and differentiated it from the Kronacher variety.

Figure 8: Localization of speech sample 4 (Neustadt bei Coburg)

Figure 9: Speech sample 4 (Neustadt / Coburg): Lower East Franconian dialect

Figure 9 illustrates that in speech sample 4 from Neustadt / Coburg the informants are unsure about assigning the variety to Coburg or Sonneberg. While 19 of 67 informants state correctly that they hear a Franconian dialect from Coburg, another 15 informants identify it as a Thuringian variety from Sonneberg. As previously shown for the samples from Hassenberg / Coburg and Malmerz / Sonneberg, they can categorize the speech sample as 'Itzgründisch', but they are not able to define whether the dialect heard is from Sonneberg or Coburg.

A comparison of the results of the localization of the speech samples reveals: Both Sonnebergers (TH) and Coburgers (BY) are not able to differentiate between the Coburgers and the Sonnebergers way of speech. The most interesting finding was that most of the informants think that the speakers are using 'their way of speech' when they try to differentiate the speech samples. With this in mind, the Coburgers believe that the speech samples from Coburg and Sonneberg are like their own way of speech—the same applies to the Sonnebergers.<sup>32</sup>

# 6.3 Dialect-geographical Results

In Sauer (2018) 146 linguistic features were examined in total. The study has shown that the dialectal variants of the Coburgers cannot be differentiated from those of the Sonnebergers. Apart from the dialectal unity between both areas, vertical as well as horizontal linguistic change could be identified:

1. Dialectal relict forms are gradually abandoned in favor of dialectal main forms. (Horizontal change)

For example, the variable mhg. ei (e. g. variant Seife 'soap') forms different dialectal variants within the 'Itzgründisch' area. In the 1930s and 1960s the speakers used the monophthongs [e:] and [ɛ:] as well as the diphthongs [e:ɐ] and [ɛɪ] to express mhg. ei. Since the 1990s mainly the monophthong [ɛ:] is pronounced by the speakers of all three speaker generations.<sup>33</sup>

2. Since the 1990s it begins to emerge that there is a vertical change in favour of socalled 'regiolects' (sub-standard varieties). (Vertical change)

The dialectal variants Hefe 'yeast', Kette 'chain' and Beet 'bed' (variable mhg. e) are only used by the older speaker group 'pre iso' and the middle group 'iso' but not by the younger speakers 'post iso'. The younger generation appears not to know the dialectal forms because these terms do not belong to their everyday life anymore. Therefore, they only use standard forms.<sup>34</sup>

In this paper the variant distribution of mhg. ë in the lexeme Wetter 'weather' will be used as an example to illustrate the differences between the Lower East

<sup>32</sup> For additional information please see Sauer (2018).

<sup>33</sup> For additional information please see Sauer (2018: 199–212).

<sup>34</sup> For additional information please see Sauer (2018: 244–260).

Franconian (Sonneberg and Coburg) and the Upper East Franconian dialect area (Kronach).

Figure 10: Dialectal realization of the lexeme Wetter 'weather' (Niederlöhner corpus, 1937)

Figure 10 presents the results obtained from the variant analysis of the Niederlöhner corpus. It is apparent that the speakers in Sonneberg and Coburg use the variant [va:də] with an open front unrounded [a:]<sup>35</sup> as the root vowel. The informants in Kronach use either the rising diphthongization [ɛɪ] <sup>36</sup> or the openmid front unrounded vowel [ɛ]. The differences between Lower East Franconian and Upper East Franconian dialect area are clearly identifiable in the corpus data from the 1930s.

All three study regions from the Zwirner- and the DR corpus show the 'Itzgründisch' variant [va:də] (see Figure 11). Thus, it can be stated for the 1960s that the research area 'Itzgründisch'was homogenous with regard to mhg. ë in the lexeme Wetter.

<sup>35</sup> Regarding the use of the long vowel [a:] in the Niederlöhner corpus and the short vowel [a] in the other four corpora: The present study is based on two sources: On the one hand on speech recordings from the 1960s, 1990s and 2014, which I transcribed by myself using the vowel [a] and on the other hand on written evidence for the 1930s, meant are transcripts by Niederlöhner who was using the vowel [a:].

<sup>36</sup> Niederlöhner used the Teuthonista transcription system. In the present study, his transcriptions were carefully translated into IPA.

Figure 11: Dialectal realization of the lexeme Wetter 'weather' (DR / ZW corpus, 1964/1957)

Figure 12: Dialectal realization of the lexeme Wetter 'weather' (SPRiG corpus, 1992)

A first language change can be noticed in the corpus data from the 1990s (see Figure 12). The dialectal form [va:də] in Sonneberg and Coburg and the form [vɛɪdə] in Kronach remain unchanged, but there is a 'new' variant [vɛdə] within the 'Itzgründisch' area. It appeared in the western part of Sonneberg and seemed to be more like a regional language. The root vowel [ɛ] in the regional variant [vɛdə] is the same as in the standard form [vɛtɐ]. For this reason, it seems obvious that a new regional language variant emerged in the early 1990s.

Figure 13: Dialectal realization of the lexeme Wetter 'weather' (Sauer corpus, 2014)

The vertical change, beginning in the 1990s, continues and expands in the corpus data from 2014 (see Figure 13). Besides the dialectal forms [va:də] in Sonneberg and Coburg and [vɛɪdə] or rather [vɛdə] in Kronach, the additional regional variant [vɛdə] can be noticed in some regions of Coburg. It is unlikely that this variant appears as a dialectal assimilation from a Coburger variant [va:də] to a Kronacher variant [vɛdə]. The analysis of other forms with mhg. ë shows that the speakers of the 'Itzgründisch' dialect area are using the root vowel [a] and especially the younger speakers sometimes switch between the dialectal form and the more standard related variant [ɛ].

To summarize: The dialectal form [va:də] of Wetter is used in Sonneberg and Coburg. Furthermore, it is different from the Kronacher variant [vɛɪdə]. Thus, there is a distinct isogloss between the 'Itzgründisch' and the Kronacher dialect area. A vertical linguistic change can be derived from the data of the more current studies (1990s / 2014).<sup>37</sup>

#### 7 Conclusion and Research Desiderata

The aim of this study was to assess what effect the inner-German border had on the dialect of the inhabitants. Therefore, firstly the subjective space 'Itzgründisch' was examined. I asked the question: Did the extra-linguistic borders affect the perception of the inhabitants—and if so, how? The results of the apparent-time comparison among three different age groups showed that both Sonnebergers and Coburgers call themselves Franconian. Besides that, they consider it (more or less) important to be a Thuringian or Bavarian, but they are interconnected as Franconians. Furthermore, the subjective space 'Itzgründisch' is homogenous in the perception of the Sonnebergers and the Coburgers. The informants were not able not differentiate between the speech samples from Sonneberg and Coburg. Though, they localized the speech sample from Kronach very well.

Secondly,I asked: Did dialectal dynamics in the former border areas take place between 1930 and 2014—and if so, how? The variant analysis showed that the objective space 'Itzgründisch' is homogenous as well and clearly separated from the Upper East Franconian dialect area. There are no significant phonetic differences between the way of speech of the Sonnebergers and the Coburgers. However, a beginning vertical linguistic change can be recognized in the data from the 1990s and 2014. In the young age group some informants use a regional language instead of the dialectal forms.

The following conclusion can be drawn from the present study: The political border did not consequentially create a linguistic border within the dialect area 'Itzgründisch'.

This is firstly because the investigation areas Sonneberg and Coburg have an 800-year history, which they share, and therefore the speakers have a shared culture as well. Therefore, the political border could not transform into an emotional border for the inhabitants and neither divides the cultural area 'Itzgründisch' into two separate spaces. Secondly, this study focused mainly on the subjective and objective structure of the dialect 'Itzgründisch' and the formation of new isoglosses between Sonneberg and Coburg after the political isolation. It will be necessary to evaluate the structure of the regional language and its use in this area. In addition, it is important to find out, how the dialectal structure in other areas near the border has developed. More research is needed

<sup>37</sup> For additional information please see Sauer (2018).

for a better understanding of the effects of the inner-German border on dialectal isoglosses in specific and the linguistic usage in general.

#### Bibliography


(eds.): The Dialect Laboratory. Dialects as a testing ground for theories of language change. Amsterdam et al.: Benjamins, 121–138.


Dr. Verena Sauer CAU Kiel German linguistics Leibnizstraße 8 24118 Kiel Germany sauer@germsem.uni-kiel.de verenasauer.org

# Manuela Lanwermeyer

# The Impact of Dialect Exposure on Phoneme Processing: Electrophysiological Evidence from German Dialects

Abstract: In linguistics, there is an ongoing debate about the factors that motivate sound change. One trigger currently discussed is the misunderstanding caused by different dialects in contact. Based on an EEG experiment, this paper provides evidence for cross-dialectal comprehension difficulties between speakers from two German dialects: the Bavarian-Alemannic transition zone and Central Bavarian. Because the dialects contrast phonemically, some lexemes have different meanings in both areas or are unknown in one of the two. This study aims to provide evidence of dialect-related comprehension difficulties of these lexemes on a neural level. The results demonstrate enhanced neural costs in Central Bavarian listeners while listening to Bavarian-Alemannic lexemes. Furthermore, the results support the idea that exposure to dialect variability generally influences speech perception.

Keywords: neurodialectology, phoneme change, electroencephalography, forward models, cross-dialectal comprehension

Abstract: In der Linguistik gibt es eine langanhaltende Debatte über die Auslöser von Lautwandel. Ein in diesem Zusammenhang wichtiger Faktor sind durch Dialektkontakt ausgelöste Missverständnisse. Basierend auf einem EEG-Experiments liefert der vorliegende Beitrag einen Nachweis für interdialektale Verstehensprobleme zwischen Sprechern verschiedener deutscher Dialekte: dem bairisch-alemannischen Übergangsgebiet und dem Mittelbairischen. Da die Dialekte phonologisch kontrastieren, haben einige Lexeme unterschiedliche Bedeutungen in beiden Regionen oder sind in einer der beiden unbekannt. Ziel der Studie ist es, für solche Lexeme neuronale dialektrelatierte Verstehensschwierigkeiten nachzuweisen. Die Ergebnisse zeigen, dass bairisch-alemannische Lexeme erhöhte neuronale Kosten bei mittelbairischen Hörern auslösen. Weiterhin legen die Ergebnisse nahe, dass der Kontakt zu dialektaler Variabilität die Perzeption per se beeinflusst.

Keywords: Neurodialektologie, Phonemwandel, Elektroenzephalographie, Vorwärtsmodelle, Interdialektales Verstehen

#### 1 Introduction

In linguistics, which factors influence phoneme change has been a long debated matter. One key factor is the contact between different varieties, e. g., different dialects. As dialects differ in structure, comprehension difficulties may occur during supra-regional communication, subsequently leading to phoneme change. This assumption is supported by Labov (cf. 2010: 35–36), who describes misunderstandings as contact phenomena which have the potential to trigger sound change. For example, he explains how the expansion of the low back merger in words such as cot and caught in Pennsylvania is caused by repeated misunderstandings between native speakers and immigrants who experienced difficulties in acquiring the phonological distinction between short and long /o/ due to their reduced phoneme systems. In addition, Schmidt / Herrgen (2011) describe comprehension difficulties resulting from dialect contact as an important trigger for sound change. In their linguistic dynamics approach they claim that in the speakers' desire to be understood, they actively and interactively synchronize their individual competencies. This synchronization involves acts of speech production and speech comprehension inasmuch as speakers try to match their partners' communicative expectations and abilities to understand, while listeners try to decode the input based on their competence. An important factor during synchronization is the feedback of the interlocutors. If speakers receive positive feedback, since their utterances fully match the listeners' communicative expectations, the conversation partners' competencies are stabilized. In contrast, if speakers receive negative feedback from interlocutors, this results in them modifying their language production strategy, affecting a restructuring of their individual competence. The negative feedback may appear in the form of a signal demonstrating that an interlocutor did not, or partially, comprehend an utterance. Thus, synchronization is defined as the calibration of competence differences during interaction resulting in either a stabilization or modification of the active and passive competencies involved (cf. Schmidt 2010: 212; Schmidt / Herrgen 2011: 25–30). Over time, such modifications manifest themselves and progressively spread over the entire speech community.

So far, it is open to question, whether such comprehension difficulties occur on a neural level during cross-dialectal comprehension. Answering this question adds to the superordinate issue of whether misunderstandings can be held responsible for triggering sound change. Therefore, this paper discusses supraregional comprehension in dialect contact settings in Germany. An ideal study region to investigate cross-dialectal misunderstandings is Central Bavarian including the Bavarian-Alemannic transition zone, as several structural (phonemical) differences occur between these dialects. This affects the Middle High German<sup>1</sup> (MHG) ô phoneme, leading to different form-meaning associations in the dialect regions. While in the Bavarian-Alemannic transition zone, MHG ô and MHG ei historically merged to /o͡a/ (e. g., /ro͡asn̩ / Rosen 'roses' and Reisen 'journeys'), Central Bavarian speakers distinguish /o͡u/ > MHG ô (e. g., /ro͡usn̩ / Rosen 'roses') and /o͡a/ > MHG ei (/ro͡asn̩ / Reisen 'journeys'). Due to this development, several lexemes have different meanings in the two dialects, which can potentially lead to cross-dialectal misunderstandings. In addition, diachronic analyses demonstrate an ongoing change of /o͡a/ to /o͡u/ or /oː/ in lexemes like Brot 'bread', groß 'big', hoch 'high' and roten 'red' in the Bavarian-Alemannic transition zone.

According to Schmidt / Herrgen (cf. 2011: 189–212), this phoneme change may be triggered by the aforementioned misunderstandings. As this assumption could not yet be supported by empirical data, this paper presents a neurolinguistic experiment examining cross-dialectal comprehension between these dialect regions on a neural and behavioral level. The primary research question is whether the theoretically postulated cross-dialectal comprehension difficulties can be proven by using the electroencephalography (EEG) method. The experiment is conducted in order to answer the following questions:


In addition, the experiment helps answer general questions about how nonnative dialect variants are processed and whether dialect contact influences neural speech processing.

The results will be interpreted neurolinguistically and more theoretically in the framework of the Pickering / Garrod (2013)'s integrated theory of language production and perception. Thus, this contribution discusses the influence of dialect contact on phoneme processing from a theoretical and empirical perspective.

<sup>1</sup> In order to investigate and compare the development of dialect phoneme systems, in dialectology typically historical (normalized or reconstructed) sound reference systems are used, namely Middle High German for vowels and West Germanic for consonants (see e. g., Schmidt / Herrgen 2011: 92).

#### 2 The Bavarian-Alemannic Transition Zone:ABrief Overview

In southern Germany, two major dialect regions can be separated: Alemannic and Bavarian. Due to a bundle of isoglosses representing phonological differences between the two regions, a Bavarian-Alemannic transition zone emerges, stretching east of the Lech. This zone is characterized by a gradual transition of Central Bavarian and Alemannic or Swabian dialect variants (cf. Kranzmayer 1927: 1). Despite the lack of a clear dialect border, three main isoglosses structure the continuum of the Bavarian-Alemannic transition zone. According to Schweizer (1956), these run from north to south along the Lech, the Ammersee and the Würmsee. The following phonological differences are separated by the isoglosses (western variants are mentioned first): The Lech isogloss distinguishes MHG ei (/ɔ͡i/, /a͡i/ vs. /ɔ͡a/, Seife 'soap'), MHG î (/iː/, /e͡i/ vs. /a͡i/, Eis 'ice') and MHG û (/uː/, /o͡u/ vs. /a͡u/, Haus 'house'). The isoglosses near the Ammersee differentiate MHG a (/a/ vs. /ɔ/, Flasche 'bottle'), MHG ä, äˆ (/ɛ/ vs. /a/, jäh 'sudden') and near the Würmsee MHG ô (/ɔ͡a/ vs. /ɔ͡u/,rot 'red') and MHG ê (/ɛ͡a/ vs. /ɛː/, Schnee 'snow') (cf. Nübling 1991: 281–282; Kranzmayer 1927). These phonological differences are also accompanied by morphological and lexical differences. For example, -en, which is pronounced /ə/ in Swabian, but /n/ in Central Bavarian. Examples of lexical differences for example are Heigl vs. Stier 'bull', Eber vs. Bär 'bear' and Aftermontag vs. Mörchta 'Tuesday'. As previously stated, there are further phonological differences in addition to these main isoglosses, which are discussed in detail by Schweizer and Kranzmayer (see Kranzmayer 1927; Kranzmayer 1927/28; Kranzmayer 1929/30; Schweizer 1928/29; Schweizer 1930/ 31). Furthermore, Bohnenberger (1928), Fried (1966) and Wiesinger (1983) present dialect differences between Swabian and Central Bavarian (for a precise overview of all dialect classifications of the Bavarian-Alemannic transition zone see Lanwermeyer 2019: 89–101). In total, the Bavarian-Alemannic transition zone can be characterized as a broad, repeatedly subdivided interference zone (cf. Wiesinger 1983: 839).

From a diachronic perspective, the Bavarian-Alemannic transition zone appears to be a region that is sensitive to language change. This was demonstrated by Lameli (2013), who compared data from the "Sprachatlas des Deutschen Reichs" (1880) with data from the "Bayerische Sprachatlanten" (1980). Lameli (2013) correlated the similarity of 371 variables from 70 Bavarian districts between the two periods. This correlation provides an indication of the diachronic stability or diachronic dynamics in the districts between 1880 and 1980. Overall, the results demonstrate a stable Bavarian and Swabian core zone on the one hand and an exclusive zone with comparatively less diachronic stability in between on the other. Interestingly, this zone of diachronic dynamics fits the BavarianAlemannic transition zone of Wiesinger (1983), but stretches even more to the east, as far as Munich.

One phenomenon which is not examined by Lameli (2013) but is subject to change in this region is the dialect pronunciation of MHG ô in lexemes like Brot 'bread', groß 'big', hoch 'high', roten 'red', tot 'dead', etc. As previously mentioned, MHG ô is pronounced /o͡a/ in the Bavarian-Alemannic transition zone, while the dialect variant is /o͡u/ or /oː/ in Central Bavarian. Diachronic analyses between the "Sprachatlas des Deutschen Reichs" (1880) and the "Bayerische Sprachatlanten" (1980) (Eichinger 2011; König 1998) demonstrate a reduction of the /o͡a/-diphthong in the Bavarian-Alemannic transition zone, which is either replaced by /o͡u/ or by /oː/ (cf. Lanwermeyer 2019: 109–119). This phonological change is lexically gradual, as different lexemes containing the MHG ô phoneme showa different stage of change. For example, in the 1980s, no single /o͡a/ could be found for the lexeme Brot 'bread', while even today adjectives like groß 'big' or hoch 'high' are often pronounced with /o͡a/. Furthermore, the areal distribution of /o͡a/ and /o͡u/ differs a great deal between the lexemes Rose 'rose' and Mutterschwein (MHG lôse) 'sow' (see Figure 1). While Mutterschwein 'sow' is pronounced like /lo͡as/ in the Bavarian-Alemannic transition zone and /lo͡us/ in Central Bavarian, many /oː/-monophthongs occur for the lexeme Rose 'rose' in the eastern Bavarian-Alemannic transition zone.

Figure 1: Dialect realizations of MHG ô in the lexemes Mutterschwein 'sow' (left) and Rose 'roses' (right) in 1980 ("Bairische Sprachatlanten")

The reason for this phonological change may be due to cross-dialectal misunderstandings between Bavarian-Alemannic speakers and Central Bavarian listeners. These occur due to structural reasons, since a historical merger of MHG ô and MHG ei occurred in the Bavarian-Alemannic transition zone, which is absent in Central Bavarian. Due to this merger, a different phoneme-to-lexeme assignment has developed in each of the dialect regions. In the Bavarian-Alemannic transition zone, minimal pairs differing only in the historical phonemes MHG ô and MHG ei have become homophones, e. g., /bro͡at/ (Brot 'bread' – breit 'broad'), /ro͡asn̩ / (Rosen 'roses' – Reisen 'journeys'). In Central Bavarian MHG ô and MHG ei did not merge but developed into /o͡u/ (< MHG ô) and /o͡a/ (< MHG ei) instead.

Regarding cross-dialectal communication, these structural differences lead to faulty decoding, since Central Bavarians try to map the Bavarian-Alemannic variants onto their own competence. Three different settings may occur:


According to Schmidt / Herrgen (2011)'s linguistic dynamics approach, misunderstandings and incomprehension lead to negative feedback during crossdialectal comprehension. In response to this negative feedback, and because humans want to be intelligible, speakers change their pronunciation word for word in order to "make individual words sound the same as when they are pronounced by speakers of the target variety" (Trudgill 1986: 58).

Thus, the phoneme change of /o͡a/ could be triggered by the previously mentioned misunderstandings leading to a change in the Bavarian-Alemannic speakers' competencies. In the following, this assumption of cross-dialectal misunderstandings is further examined by collecting neural data on dialect phoneme processing using the EEG method.

# 3 State of Research

So far, minimal research has been conducted on dialect phoneme processing. Existing studies, using the EEG method, predominantly examine the phoneme discrimination of dialect contrasts and largely make use of Mismatch Negativity (MMN).<sup>2</sup> The MMN is a frontocentral negative component, usually peaking at 150–250 ms from change onset. This component is elicited when infrequent deviations (deviant stimuli) are inserted in a series of frequently repeated sound patterns (standard stimuli)<sup>3</sup> in a passive oddball design. In such an oddball design, isolated phonemes, syllables or lexemes are presented to participants while they are watching a silent film, e. g., pa – pa – pa – ba – pa – pa (see Näätänen et al. 2007; Picton et al. 2000 for an overview). The MMN reflects a preattentive response to any change in auditory stimulation. The standard stimuli create a representation, namely a short-term memory trace in the auditory cortex, against which the deviant stimuli are compared. Thus, the MMN is the reflection of a discrimination process as the representation is violated by an infrequent deviant, indicating that the deviant is found to be incongruent with the memory representation of the preceding standard stimuli (cf. Näätänen et al. 2007).

A listener's ability to differentiate phonemes is largely dependent on their native phoneme inventories (see e. g., Buchwald et al. 1994 concerning the discrimination between /r/ and /l/ in Japanese). This is supported by several oddball studies, which prove language-specific memory traces for phonemes. The MMN component increases when the deviant is a phoneme of the subject's native language in contrast to non-native phoneme categories (cf. Näätänen et al. 1997; Winkler et al. 1999). This can also be applied to non-native regional phonemic contrasts within a language. For example, Brunellière et al. (2011) compare the regionally merged /e/–/ɛ/ contrast in word-final open syllables (e. g., /epe/ 'sword' vs. /epɛ/ 'thick') in merged and unmerged French speaker groups. The

<sup>2</sup> Components can be described by different factors: the polarity in µV leading to positive and negative effects, the latency indicating the time after starting measurement (typically the onset of the critical item), the amplitude displaying the strength of an effect and the topography which shows at which electrodes an effect is elicited.

<sup>3</sup> Note that in this contribution the term 'standard' is used in the explained sense—frequently repeated sound patterns in a passive oddball design—and not in terms of the standard language.

results demonstrate processing differences, indicating that in contrast to unmerged speakers, merged speakers associate both forms with only one semantic representation (homophones). This supports the assumption that access to lexical meaning in spoken word recognition depends heavily on the listeners' native regional accent. In another oddball study concerning the /e/–/ɛ/ contrast, Brunellière et al. (2009) find that regionally merged vowels are differentiated less successfully than stable phoneme contrasts. Interestingly, this also holds true for speakers who differentiate these phonemes themselves but are exposed to merged dialect variants in daily life. This result is further supported by Conrey et al. (2005) who focus on mechanisms of semantic integration and phonological decision processes concerning the regionally merged /ɪ/–/ɛ/ contrast in American English (pin-pen merger). Using a same-different task<sup>4</sup> , they show that merged and unmerged dialect groups process the stimuli differently at a conscious, decisional level. Remarkably, in accordance with Brunellière et al. (2009) the results indicate discrimination difficulties concerning merged variants in contrast to supra-regional stable contrasts even in unmerged listeners explainable by dialect contact with the regionally merged variants.

Using an oddball experiment with a same-different task, Dufour et al. (2013) examine how southern French speakers process the standard French vowels [o], [ɔ] and [u]. In contrast to [ɔ] and [u], [o] does not occur in the southern French phoneme inventory. The neural data show that southern French speakers can differentiate all the contrasts. By contrast, the behavioral data indicate difficulties in the discrimination of [o]–[u] and [o]–[ɔ]. Therefore, the native phoneme inventory affects vowel perception at a late decisional stage of processing.<sup>5</sup>

In sum, these studies demonstrate that the dialect phoneme inventory and exposure to regional variability influences phoneme processing. Furthermore, EEG studies using an oddball design seem to be quite promising into examining dialect phonemes, since phonological and lexical processing stages are highly influenced by the listeners' regional accent.

<sup>4</sup> In a typical experiment with same-different-task, which focuses on phoneme discrimination, participants are exposed to auditory or orthographical stimuli. After that, a critical item is presented auditorily and participants have to decide whether it differs from the previously stimuli or not.

<sup>5</sup> For further EEG studies examining dialects see Bühler et al. (2017a), Bühler et al. (2017b) Miglietta et al. (2013), for phonetic dialect variation and Fournier et al. (2010), Werth et al. (2018) for tone accents.

# 4 The Impact of Dialect Exposure on Phoneme Processing: a Theoretical Approach

In order to understand the effects of dialect contact on phoneme processing, a model which combines speech production and speech perception is required. Pickering / Garrod (2013) propose that producing and understanding are tightly interwoven and that this interweaving enables people to predict their own and others' actions. Therefore, "actors construct forward models of their actions before they execute those actions, and […] perceivers of others' actions construct forward models of others' actions that are based on their own potential actions" (Pickering / Garrod 2013: 332).

Initially, forward models were assumed in order to understand the problem of sensorimotor control. The underlying question is how motor commands interact with sensory signals to perform accurate movements, e. g., lift the arm. Therefore the motor system forms "a loop in which motor commands cause muscle contractions, with consequent sensory feedback, which in turn influences future motor commands" (Wolpert et al. 2003: 593). To perform accurate movements without delay, forward models in the central nervous system (CNS) simulate the causal relationship between actions and their consequences (cf. Wolpert / Ghahramani 2000: 1212). In this context, forward models "aim to mimic or represent the normal behaviour of the motor system in response to outgoing motor commands" (Miall / Wolpert 1996: 1265). Therefore, a motor command initiates two processes (see Figure 2): First, it acts on the actual sensorimotor system resulting in a movement followed by sensory feedback of the movement (reafference). Second, the current state of the motor system and a copy of the motor command (efference copy) pass into cascading forward models, which act as neural simulators of the musculoskeletal system and the environment, and estimate the new state and the sensory feedback (cf. Wolpert et al. 2003: 595). This prediction is then compared to the reafference of the movement. If the estimated sensory feedback does not fit the actual sensory feedback (prediction error), the forward model and / or the motor command can be adapted. Thus, "forward models are not fixed entities but must be learned and updated through experience" (Wolpert / Flanagan 2001: R729).

It is important to note that forward models can also estimate contexts. People are able to perform movements in many different settings. Therefore, it can be assumed that the CNS selects from multiple coexisting forward models. For example, when a person lifts an object with unknown dynamics, the context must be identified in order to choose the right motor command.

Pickering / Garrod (2013) apply forward models to speech production (see Figure 3). Again, two processes run parallel: on the one hand, the utterance is

Figure 2: The comparison of estimated sensory feedback by means of cascading forward models with actual sensory feedback (reafference)

generated based on a production command, which is fed into the production implementer. Speakers process an utterance themselves in the comprehension implementer and generate a percept of the produced utterance. On the other hand, the efference copy is fed into the cascading forward production model and forward perception model. The result is a predicted utterance percept, which then is compared to the actual utterance percept. When this comparison presents a discrepancy (prediction error), the forward model or the ensuing production command can be adapted. For instance, this happens when speakers misspeak. They recognize a mismatch between their intended utterance and their actual output (prediction error) and correct themselves. It is worth noting that speakers construct representations associated with different linguistic levels, such as semantic, syntactic and phonological representations. Thus, the predicted and actual utterance percepts are compared separately due to semantics, syntax and phonology (cf. Pickering / Garrod 2013: 338).

It is important to note that forward models can be assumed for perception as well as actions. The basic idea is that "others' actions are decoded by activating one's own action system at a sub-threshold level" (Wolpert et al. 2003: 593). Thus, during action perception perceivers draw on experiences they had with their own bodies in the past (simulation route). In simple terms, one can imagine the process as follows: In the case of an arm movement, perceivers recognize the start of the actors' movement and then determine how they would continue if it were their arm. They predict their partners' movement by using the mechanism they

Figure 3: Model of speech production, using a snapshot of speaking at time t (Pickering / Garrod 2013: 338)

would use to move their own hand (covert imitation) (cf. Pickering / Garrod 2013: 334–335).

During speech perception, thissimulation route is based on experiences of the comprehenders producing utterances themselves. Thus, by using forward models they determine what they would say under these circumstances. To do this precisely, they must consider the context, e. g. information about differences between the language systems of the speakers involved. In order to estimate an utterance in line with the simulation route, the comprehenders covertly imitate the speakers' utterances in order to derive the production command they would use if they were to produce the utterance themselves. Then they use the same forward modelling as if they were producing the utterance themselves in order to predict the speakers' utterances and utterance percepts (see Figure 3) (cf. Pickering / Garrod 2013: 342). For instance, in a situation where two children have received presents, namely the boy an airplane and the girl a kite, the girl utters I want to go out and fly the. In order to predict the utterance, the boy covertly imitates the girl's production I want to go out and fly the and derives the production command he would use to continue the utterance. He then derives the production command he would use in order to pronounce the word the girl would say (kite) and runs forward models to derive his predicted utterance percept. This means that the boy derives the production command of what he assumes the girl is likely to say (kite), rather than what he himself would be likely to say (airplane) (cf. Pickering / Garrod 2013: 341).

Apart from the simulation route, people can draw on their past perceptual experience with other people (association route). For example, perceivers can predict someone else's arm movement by assuming that it is compatible with their former experiences of perceiving other people's arm movements (cf. Pickering / Garrod 2013: 334). During speech perception, the association route is based on former experiences in comprehending others' utterances.

During speech comprehension, perceivers emphasize either the simulation route or the association route. When the interaction partners appear to be similar concerning cultural and dialect background, the simulation route is emphasized since in these cases simulation will tend to be accurate. In contrast, when the interaction partners seem to be less similar, e. g. different dialect speakers, the association route is emphasized (cf. Pickering / Garrod 2013: 346). Most likely, comprehenders can combine the association route and the simulation route in speech comprehension. Thus, it is possible to predict utterances by combining experiences of how similar people have spoken with experiences how one has spoken oneself under similar circumstances (cf. Pickering / Garrod 2013: 346).

It is worth noting that prediction errors on different linguistic levels have an online effect on speech perception, which can be measured with the EEG method. For instance, sentence-final lexemes elicit an N400, when they do not fit the context or are less predictable and thus violate the previous built up expectation (for further explanations of the N400 see Section 5.1).

#### 5 Methodology

#### 5.1 Experiment Design

In order to examine cross-dialectal comprehension between Bavarian-Alemannic speakers and Central Bavarian listeners, an oddball design containing full sentences and a semantic rating task was chosen. So far, few studies have adapted the oddball design by using full sentences to examine phonemic and semantic processes during sentence processing. For instance, Menning et al. (2005) examine semantic and syntactic errors by comparing the standard Die Frau düngt den Rasen im Mai 'The woman fertilizes the lawn in may' with the deviants Die Frau düngt den Riesen im Mai 'The woman fertilizes the giant in may' and Die Frau düngt den Rosen im Mai 'The woman fertilizes the roses in may'. The results showa sensitivity towards complex linguistic material insofar as the deviants are compared to the predicted continuation of the sentences built up by the standard. Furthermore, Boulenger et al. (2011) support that during natural speech processing, the brain extracts abstract regularities from the continuous signal and forms memory traces in the auditory cortex. Thus, it seems that memory traces can also develop for complex sentences, which include large-scale details about phonetic features of the speech signal.

The EEG design for this study follows Bendixen et al. (2014), who examine the standards Lachs 'salmon' and Latz 'bib' in comparison to omissions (La) during sentence comprehension. Listeners heard 30 different sentences which prime the meaning 'salmon', 'bib' or are neutral in this regard. During the experiment, every sentence was presented 6 times: 2/3 of the sentences ended with the expected lexeme, while in 1/3 of the cases the sentence-final lexeme was replaced by the omission La. Furthermore, participants were required to perform a semantic rating task. An important adjustment of this study in contrast to Bendixen et al. (2014) is the integration of semantic violations to examine higher levels of processing like semantic integration (misunderstanding). Typically, effects of semantic integration are indicated by the N400, a rather late negative component peaking at about 400 ms after stimulus onset. It is most pronounced over centroparietal sites and is, for example, elicited by semantically anomalous sentencefinal words (cf. Connolly / Phillips 1994; Kutas / Hillyard 1980; Kutas / Hillyard 1984; Kutas/ Federmeier 2011; Lau et al. 2008). The N400 displays the violation of predictions and expectations built up by the preceding sentence context, since predictable words are easier to access from memory and it requires more resources to process an implausible or infrequent continuation (cf. Lau et al. 2008).

#### 5.2 Conditions

#### Condition 1: Misunderstanding

In cross-dialectal comprehension, misunderstandings occur when a lexeme has different meanings in different dialect regions. In dialect contact between the Bavarian-Alemannic transition zone and Central Bavarian, this affects lexemes containing /o͡a/ (< MHG ô) forming minimal pairs with lexemes traced back to MHG ei.

In condition 1 the minimal pair /ro͡asn̩ /–/ro͡usn̩ / is examined. While /ro͡asn̩ / means 'roses' and 'journeys' in the Bavarian-Alemannic transition zone, it only signifies 'journeys' in Central Bavarian and the variant for 'roses' is /ro͡usn̩ /.

This condition is separated into two sub-conditions, since a phonemic contrast is investigated in both directions in an oddball design typically. Thus, in condition 1a /ro͡usn̩ / serves as the standard and /ro͡asn̩ / as the deviant and in condition 1b vice versa. Note that only condition 1a reflects real dialect contact, while condition 1b is inserted due to methodological reasons.

For condition 1, the hypothesis is that the deviants /ro͡asn̩ / and /ro͡usn̩ / elicit an N400 since they evoke a semantic mismatch between standard and deviant. Alternatively, an N200 (MMN) can be expected due to the oddball paradigm as in the comparable study of Bendixen et al. (2014).

#### Condition 2: Incomprehension

Regarding cross-dialectal comprehension, incomprehension occurs when a lexeme only exists in one of two dialects. When applied to the dialect contact between the Bavarian-Alemannic transition zone and Central Bavarian this affects lexemes which include /o͡a/ (< MHG ô) but do not form a minimal pair with lexemes containing MHG ei.

In condition 2, the minimal pair /lo͡as/–/lo͡us/ is investigated. Both lexemes are heteronyms for the meaning 'sow', namely /lo͡as/ in the Bavarian-Alemannic transition zone and /lo͡us/ in Central Bavarian. In contrast to condition 1, the meaning does not differ between the two dialect regions. In contrast, /lo͡as/ is not part of the Central Bavarians' competence and thus is meaningless.

Since /lo͡as/ does not carry lexical meaning in Central Bavarian, it is most likely perceived as a well-formed non-word (pseudo-word). Previous EEG studies have demonstrated that non-existing words are processed differently depending on their phonotactic structure. Non-words violate the phonotactic rules of a given language, but pseudo-words are well-formed in this regard. In contrast to pseudo-words, non-words can be recognized as being non-existing very early at a pre-attentive level and thus do not function as potential candidates for semantic integration (cf. Domahs et al. 2009: 418). Studies like Bentin et al. (1999) and Domahs et al. (2009) indicate enhanced processing costs (N400) for pseudowords in a lexical decision task, that is, during classification of pseudo-words as not-existing. Thus, the N400 also reflects processes of word detection. While nonwords can be rejected very early, the classification of pseudo-words needs further processing resources in order to determine whether these words actually do not exist and thus do not fit the context (cf. Kutas / Van Petten 1994: 104–105).

The hypothesis for condition 2 is that the deviant /lo͡as/ elicits an N400 since it is processed as a pseudo-word. Otherwise, again an N200 (MMN) can be evoked due to the experimental oddball design.


Table 1: Experimental conditions

# 5.3 Procedure

In preparation for the experiment, context sentences with the same global structure were designed which prime the meaning 'roses', 'journeys' or 'sow'. Furthermore, for 'sow' neutral sentences were created. A pretest ensured that all sentences prime the respective meaning very well or are neutral in this regard (see Lanwermeyer et al. 2016: 5; Lanwermeyer 2019: 186–188 for a precise description of the pretests). For all conditions, 35 sentences were chosen which then were recorded by a male dialect speaker from the Bavarian-Alemannic transition zone (Merching). He adapted the sentences to his dialect phonologically and lexically. The best 30 sentences as well as ten critical items were chosen for each condition. The phonetic variability of the critical items is important since this leads to a more reliably and natural representation of the standard (cf. Phillips et al. 2000; Scharinger et al. 2011). Subsequently, all sentences were cross spliced.

The investigationwas conducted in Isen, a small market community in Central Bavaria (see Figure 4). During the experiment, the Bavarian-Alemannic sentences were presented auditorily to 20 Central Bavarian participants (13 women; mean age 44.5 (SD 4.87)) in separated blocks. For condition 1, the meaning 'roses' or 'journeys' was primed in two blocks of 90 sentences in each case. For condition 2, priming sentences for 'sow' and neutral sentences were mixed up in four blocks of 90 sentences. In both conditions, 2/3 of the sentences ended with the standard, 1/3 with the deviant. The participants sat in front of a computer screen and had to judge on a four-point scale (very well – very badly) how well the sentence-final lexeme fits the sentence context while the EEG was recorded.

Figure 4: Recording location (Merching) and experimentation location (Isen)

### 6 Results and Neurolinguistic Interpretation

#### 6.1 Condition 1: Misunderstanding

In condition 1a, the meaning 'roses' was primed. The Central Bavarian variant /ro͡usn̩ / served as the standard, the Bavarian-Alemannic variant /ro͡asn̩ / as the deviant. The Central Bavarian listeners judged the standard /ro͡usn̩ / as fitting the context very well, demonstrating that the sentence-final lexeme fit the expectation built up by the preceding sentence context. On a scale from 1–4 (1 = very well – 4 = very badly), participants rated the standard /ro͡usn̩ / with 1.12 (SD 0.20). In contrast, the deviant /ro͡asn̩ / was evaluated as fitting the sentence context badly (mean 3.72 (SD 0.79)). Thus, /ro͡asn̩ / is rejected consciously since it violates the expectation of the standard /ro͡usn̩ /.

The behavioral data of condition 1b present similar results. In contrast to condition 1a, this condition does not display real dialect contact, butwas inserted for methodological reasons. The sentences presented prime for the meaning 'journeys'. While /ro͡asn̩ / served as the standard, the Central-Bavarian variant /ro͡usn̩ / 'roses' was the deviant. Again, the standard was judged as fitting the sentence context very well (mean 1.15 (SD 0.17)), while the deviant was rejected (mean 3.90 (SD 0.08)).

Interestingly, the neural data show different results in the two conditions. Subsequently, first the negativity effects are presented and thereafter the positivity effects.<sup>6</sup>

Negativity effects (N200, N400)

In condition 1a, the neural data demonstrate an N200 between 100 and 200 ms (see Figure 5). The N200-family consists of different components, for example the N2a (MMN) and N2b. While the MMN displays an automatic pre-attentive discrimination process, the N2b is elicited when participants consciously perceive a deviation, which is relevant for the experimentation task (e. g., oddball detection task). Thus, both components display different stages of mismatch detection (cf. Folstein / Van Petten 2008: 153; Patel / Azzam 2005: 147; Pritchard et al. 1991: 76– 77). In this experiment, the participants' attention was directed consciously towards the critical items due to the evaluation task. Thus, the early negativity elicited for /ro͡asn̩ / can be classified as an N2b indicating a mismatch between the built-up memory trace through priming and the unexpected deviant. It displays an active discrimination process between predicted standard and actually per-

<sup>6</sup> The statistical results are not described here in detail (see Lanwermeyer et al. 2016: 7–8 for an exact description).

ceived deviant, since an active perception of the discrepancy between standard and deviant is necessary to fulfill the experimentation task. Thus, the N2b elicited for /ro͡asn̩ / displays an active mismatch detection. This interpretation is supported by the comparable study of Bendixen et al. (cf. 2014: 20), who explain the elicited N200 by active listening necessary for the experimentation task. The N2b shows that the Bavarian-Alemannic variant /ro͡asn̩ / is perceived as being different from the activated memory trace built up by the standard /ro͡usn̩ / at a conscious level.

Figure 5: Event-related potentials for the deviant /ro͡asn̩ / in contrast to the standard /ro͡usn̩ / measured from 100 ms prior the word onset up to 1000 ms (Lanwermeyer et al. 2016: 7)

Surprisingly, in condition 1b, a relatively late negativity between 300 and 500 ms is elicited for the deviant /ro͡usn̩ / (see Figure 6). Due to its latency and its parietal distribution this effect can be interpreted as an N400 indicating the semantic mismatch between the expected continuation of the sentence and the perceived input, namely the deviant. Due to the semantic priming and the repetitive standard /ro͡asn̩ /, the expectation for the Central Bavarian variant /ro͡usn̩ / is preactivated. The semantic incongruous deviant leads to a semantic mismatch between the information in short-term memory and the unfitting item. Thus, the N400 indexes integration difficulties of the incongruent deviant in the preceding sentence context (cf. Brown / Hagoort 1993; Kutas / Federmeier 2000; Kutas / Hillyard 1980; Lau et al. 2009).

Figure 6: Event-related potentials for the deviant /ro͡usn̩ / in contrast to the standard /ro͡asn̩ / measured from 100 ms prior to the word onset up to 1000 ms (Lanwermeyer et al. 2016: 8)

The different latencies of the effects are rather surprising, as the same linguistic material and the same experimental design was used for both conditions. There are several approaches to explain the different latency of the negativity effects. For instance, the frequency of the lexemes and intrinsic characteristics of the diphthongs /o͡a/ and /o͡u/ (for a discussion see Lanwermeyer et al. 2016: 12–14; Lanwermeyer 2019: 199–206). The most promising factor for the explanation of the asymmetry is dialect contact, which is discussed in detail in Section 7.

#### Positivity effects (LPC (P300))

In both conditions, a late positive component (LPC) is elicited: in condition 1a between 400 and 900 ms (see Figure 5) and in condition 1b between 550 and 1000 ms (see Figure 6). The LPC can be interpreted as a member of the P300 family. The P300 is a positivity, which peaks around 300 ms (for an overview see Picton et al. 2000; Polich 2003; Polich 2007). However, the latency varies relatively strong. For instance, it may follow an N400 and thus can also be elicited up from 500 ms (cf. Roehm et al. 2007: 1260). The P300 family consists of different sub-components. The positivity found in this study can be interpreted as a P3b. This component is a parietal pronounced positivity, which is elicited when oddballs are interspersed in a group of homogeneous stimuli on which participants must draw attention to (cf. Gerloff 2005: 514). It displays the task relevant evaluation and categorization of the presented stimuli (cf. Bentin et al. 1999: 251; Bohn et al. 2013: 769; Knaus et al. 2007: 712; Kutas / Van Petten 1994: 112).

The P3b is closely linked with the refutation of expectations. For example, the amplitude of the P3b depends on the probability of occurrence of a stimulus category. Stimuli from a rare category lead to a more pronounced amplitude than those from a frequent category. Furthermore, the P3b is influenced by the assigned task, namely the manner how stimuli must be categorized (cf. Van Petten / Luka 2012: 181; Folstein / Van Petten 2011: 825).

In this experiment, participants had to judge how well the sentence-final item fits the sentence context. Thus, their attention was directed consciously to the linguistic material. In order to fulfill the task, participants had to compare and categorize the expected standards and the perceived deviant stimuli. Since the categorization of the critical stimuli was relevant for the task, the LPC elicited for /ro͡asn̩ / and /ro͡usn̩ / can be interpreted as a P3b. The interpretation is furthermore supported by the centro-parietal pronounced distribution of the component. The P3b displays the task relevant evaluation and categorization of the previously discovered mismatch (N2b). According to Domahs et al. (cf. 2008: 27–28) and Henrich et al. (cf. 2014) the P300 also reflects a reanalysis process. The amplitude of the LPC depends on the degree of the reanalysis process, since a sentence-final lexeme, which does not fit the expectation, requires a reanalysis of the previously built up structure.

In total, the P3b elicited for /ro͡asn̩ / and /ro͡usn̩ / displays a conscious and taskrelated evaluation and categorization of the actively perceived deviant in contrast to the standard as well as the reanalysis process, which is triggered by the unexpected deviant, which does not fit the context.

#### 6.2 Condition 2: Incomprehension

In condition 2, the meaning 'sow' was primed in half of the sentences, the other half was neutral. Condition 2 differs from condition 1a in that /lo͡as/ only carries lexical meaning in the Bavarian-Alemannic transition zone, but not in Central Bavarian, where this word-form does not exist, but is comparable to a pseudoword.

In the priming condition, the standard /lo͡us/ was evaluated as fitting better to the sentence context (1.74 (SD 0.94)) than the deviant /lo͡as/ (2.72 (SD 1.3)), but this difference does not reach statistical significance possibly due to the high standard deviation of the ratings. The behavioral data underline the difficulties of classification and integration of the pseudo-word /lo͡as/. The lexical status of pseudo-words is difficult to evaluate, since the evaluation of phonotactic wellformed pseudo-words is more costly than the evaluation of existing lexemes or non-words, which do not follow the phonotactic rules. Thus, the data support that the participants were unsure about the classification of the pseudo-word. Since all participants recognized that the speaker comes from another dialect region, they might have weight whether /lo͡as/ is an old unknown dialect lexeme or a dialect lexeme from another region. In neutral sentences, the deviant /lo͡as/ was evaluated significantly worse (3.07 (SD 0.91)) than the standard /lo͡us/ (2.41 (SD 0.74)) showing that the deviant /lo͡as/ was rejected consciously.

#### Negativity effect (N200)

The neural data present an early negativity between 100 and 200 ms (N200) for priming sentences (see Figure 7), which is absent for neutral sentences (see Figure 8). The N200 is not consistent with previous studies examining pseudoword. The latency can be explained by the experimental oddball design and the experimental task. In contrast, in the study of Domahs et al. (2009) different nonexisting words were integrated in the experiment. Thus, participants had to process the whole lexeme in order to solve the lexical decision task. In this study, only two altering phonological forms were presented. To solve the experiment task, participants did not have to process the whole lexeme since the lexemes differed up from the beginning of the diphthong. Thus, participants were able to identify the respective lexeme very early. This interpretation is supported by Connolly / Phillips (1994) who examine the processing of sentence-final lexemes which either fit to the sentence context or not and which either exhibits a predicted initial phoneme or not. For the condition containing an initial unpredictable phoneme, an early negativity is elicited. Although the lexeme cannot be identified completely, participants recognize due to context information that it does not fit the predicted lexeme. Since the lexeme already differs in the onset, it influences the selection process of the right candidate. Thus, when the expectations evoked by the context do not match the characters of the initial phoneme, an early negativity is elicited (cf. Connolly / Phillips 1994: 263).

The N200 for the deviant /lo͡as/ in the priming condition can be interpreted similarly as in condition 1a. It reflects an active early error detection mechanism. The priming context and the repetitive presented standard /lo͡us/ build up a memory trace and an expectation for the Central Bavarian variant /lo͡us/. This is violated when the deviant /lo͡as/ is perceived instead. Thus, the deviant /lo͡as/ is identified as being different than the activated memory trace.

Figure 7: Event-related potentials for the deviant /lo͡as/ in contrast to the standard /lo͡us/ in priming sentences measured from 100 ms prior to the word onset up to 900 ms (Lanwermeyer et al. 2016: 10)

In contrast to priming sentences, no N200 could be elicited in neutral sentences. This shows the influence of semantic priming on perception. Since the context does not deliver any hints about the continuation of the sentences, no expectations were built up and violated. The frequent presentation of the standard /lo͡us/ is not enough to build up a strong memory trace against which the deviant can be compared.

Figure 8: Event-related potentials for the deviant /lo͡as/ in contrast to the standard /lo͡us/ in neutral sentences measured from 100 ms prior to theword onset up to 900 ms (Lanwermeyer et al. 2016: 10)

#### Positivity effects (LPC (P300))

In priming and neutral sentences, an LPC between 350 and 600 mswas elicited for /lo͡as/ (see Figure 7 and Figure 8). This LPC can be interpreted in the same way as in condition 1. The P300 reflects the categorization and evaluation process of the previously detected error. This relates to the semantic evaluation task and the reanalysis process, which is necessary due to the unfitting deviant /lo͡as/ in sentence-final position. In neutral sentences, the LPC is less pronounced. This can be explained by the fact that the amplitude also is modulated by the degree of complexity and difficulty of the experimental task. A more pronounced effect shows that it is easy for the participants to evaluate the sentences (cf. Bohn et al. 2013: 769; Domahs et al. 2009: 428). The enhanced amplitude for priming sentences suggests that the evaluation is easier when an expectation is built up through priming. This differs in neutral sentences where the context does not give any hints about the continuation of the sentences. The reduced LPC component indicates that it is easier for the participants to categorize and reject the pseudo-word /lo͡as/ in neutral sentences. Furthermore, the reduced amplitude can be explained by the reanalysis process. In priming sentences, the semantic cues enhance the expectation for the standard /lo͡us/. Thus, the mismatch between expected and perceived variant is stronger than in neutral sentences. The more pronounced amplitude for /lo͡as/ in priming sentences results from a stronger degree of deviation in comparison to neutral sentences, which do not give any hint about the critical item.

#### 7 Conclusion and Research Desiderata

The results of the EEG-study demonstrate, that the deviants /ro͡asn̩ / and /lo͡as/ lead to enhanced neural costs during cross-dialectal comprehension. The elicited N200 and LPC indicate the active detection and evaluation of an acoustic deviation as well as a mismatch between expected and encountered word form. Thus, the thesis of cross-dialectal comprehension difficulties between Bavarian-Alemannic speakers and Central Bavarian listeners concerning the usage of /o͡a/ variants is confirmed. In accordance with the linguistic dynamics approach of Schmidt / Herrgen (2011), Central Bavarian listeners compare the Bavarian-Alemannic sentences with their competence and recognize that the non-native dialect variants deviate from their own variants. When the meaning 'roses' is primed, Central Bavarian listeners expect their own variant /ro͡usn̩ /. When they perceive /ro͡asn̩ / instead, they interpret it as 'journeys' as the behavioral data suggest: The deviant /ro͡asn̩ / is evaluated as not fitting to the sentence context due to a discrepancy between meant and understood content of the utterance (misunderstanding). Following Schmidt / Herrgen (2011), these misunderstandings lead to negative feedback from Central Bavarians during interaction. As a result, Bavarian-Alemannic speakers modify their pronunciation in direction to the prestigious Central Bavarian dialect. Interestingly, the deviant /lo͡as/ could not be rejected that easy. The behavioral data suggest that the participants were uncertain about how to categorize the lexeme. Since there is no concept for /lo͡as/ in Central Bavarian, the listeners were not able to interpret the lexeme (incomprehension). Thus, a difference between both conditions can be found in the behavioral data.

Typically, semantic mismatches (misunderstandings) are indexed by a rather late component, namely the N400. The reason why no N400 is elicited for /ro͡asn̩ / and /lo͡as/ is likely due to dialect contact. The results suggest that close contact to non-native dialect variants influences speech processing. This assumption is supported by the results of another EEG study, which was conducted in Rhine-Franconia. In this study, the same design was used, however the sentences were adapted to the Rhine-Franconian dialect. In contrast to the present study, amongst others, a real pseudo-word, which does not occur in the adjacent dialect region served as the deviant (/ma͡uk/ 'sow'). Furthermore, in another condition, the minimal pair /loːs/ 'winning ticket' vs. /la͡us/ 'louse' was examined, which does not showa different phoneme-to-lexeme assignment in the different dialect regions, but is a normal minimal pair like in Standard German. Interestingly, in both conditions N400 effects are elicited (see Lanwermeyer 2019: 215–225; Lanwermeyer et al. submitted). Therefore, it seems clear that dialect contact influences the latency of the effects. However, it is uncertain so far what this influence looks like. Studies like Brunellière et al. (2009), Brunellière et al. (2011) and Conrey et al. (2005) provide first evidence that contact to non-native dialect variants affects even the ability to discriminate native phonemes.

From a theoretical point of view, it can be assumed that, based on the priming sentence context, listeners make predictions about the continuation of the sentences. Due to Wolpert et al. (2003) people decode actions by activating their own production system. The production system of interaction partners is activated during the whole dialogue since people produce and perceive speech simultaneously. In accordance with Pickering / Garrod (2013) listeners covertly imitate perceived utterances based on their own speech production system. Subsequently, they use forward models in order to predict the ongoing utterance resulting in a predicted utterance percept, which is compared with the actual utterance percept on different linguistic levels. In the case of discrepancies, a prediction error is generated.

Applied to this study, Central Bavarian listeners predict the sentence final lexemes based on the sentence context and draw on experiences during their own speech production (simulation route). Thus, they expect their own variants /ro͡usn̩ / 'Roses' or /lo͡us/ 'sow'. The early negativities (N200) elicited by the deviants /ro͡asn̩ / or /lo͡as/ indicate prediction errors due to a discrepancy between predicted and perceived utterance percept. This interpretation is supported by the fact that no N200 is elicited for /lo͡as/ in neutral sentences. Thus, priming clearly influences the amplitude of the component.

Possibly, Central Bavarian listeners use a combination of simulation route and association route during speech perception. Due to Pickering / Garrod (2013) listeners use the association route when the interaction partners differ with regard to their social, cultural or language background. Interestingly, each participant was aware that the speaker comes from another dialect region (Augsburg, Swabian, Allgäu), as a short interview after the EEG study shows. Thus, it can be expected that the Central Bavarian listeners base their perception on former experiences with comprehending Bavarian-Alemannic speakers as well. They first generate what they would say under these circumstances, but consider the context (the interaction partner) as well, in order to predict what the speaker would say. That is to say, the context, which is defined by differences between the dialect systems of Central Bavarian and the Bavarian-Alemannic transition zone, has an influence on speech perception. Central Bavarian listeners predict the Bavarian-Alemannic utterances by combining experiences with their own and the Bavarian-Alemannic speech. Therefore, generally a combination of simulation route and association route can be assumed for speech perception in dialect contact situations.

This interpretation is supported by Sebastián-Gallés et al. (2005) and Sebastián-Gallés et al. (2008), who focus on Spanish-Catalan language contact concerning the Catalan phonemes /e/ and /ɛ/. In contrast to Catalan, the Spanish phoneme system only exhibits /e/. The authors conduct lexical decision tasks, in which participants had to judge Catalan lexemes, which either are real words or are non-words due the replacement of /e/ and /ɛ/. Interestingly, Catalan-dominants bilinguals and Catalan speakers judge stimuli, in which /e/ is replaced by /ɛ/ significantly better than vice versa. This can be explained by the increased contact with false pronunciations by Spanish speakers, who use their Spanish /e/ phoneme for the Catalan /ɛ/-category. Therefore, these false variants are accepted as existing lexemes by Catalans. In a further study Larsson et al. (2008) show that the acceptance of these false variants as real words is a phenomenon on the lexical level. According to the authors, the contact to the Spanish /e/-variants in false contexts results in the creation of new alternative lexical entries in Catalan speakers, which represent the dialect variation. This interpretation seems to be plausible for the dialect contact between Central Bavarian and the Bavarian-Alemannic transition zone, as it would explain the missing N400 for /ro͡asn̩ / in contrast to /ro͡usn̩ /. Variants occurring in dialect contact might be represented in the mental lexicon and therefore, the semantic mismatch is missing in the neural data.

To sum up, the present EEG study shows that the use of the /o͡a/ phoneme (< MHG ô) leads to enhanced neural costs during sentence processing indicating cross-dialectal misunderstanding and incomprehension. Furthermore, the results support the idea that dialect exposure generally influences speech perception. Unfortunately, recently there are few neurolinguistic studies on phoneme perception and even less that focus on sentence processing. Therefore, the interpretations offered for the asymmetric results of the negativity effects should be supported by further studies. A superordinate research desideratum is to examine the question of how dialect contact influences speech processing on different levels in more detail. For example, the expectation of different lexical entries for dialect variation should be investigated further. The present study is only a first approach in order to look at German dialect variation from a new, namely a neurolinguistic point of view.

# Bibliography


40jährigen Bestehens der Schwäbischen Forschungsgemeinschaft und zur 20-Jahrfeier der Universität Augsburg. Mit Berichten aus der landesgeschichtlichen Forschung in Augsburg. Sigmaringen: Thorbecke, 235–315.


Sebastián-Gallés, Núria / Echeverría, Sagrario/Bosch, Laura. 2005. The influence of initial exposure on lexical representation: Comparing early and simultaneous bilinguals. In: Journal of Memory and Language 52(2), 240–255.

Sebastián-Gallés, Núria / Vera-Constán, Fátima / Larsson, Johan P. / Costa, Albert / Deco, Gustavo. 2008. Lexical plasticity in early bilinguals does not alter phoneme categories: II. Experimental evidence. In: Journal of Cognitive Neuroscience 21(12), 2343–2357.

Trudgill, Peter. 1986. Dialects in Contact. Oxford: Blackwell.


Mag. Dr. Manuela Lanwermeyer University of Vienna Department of German Studies Universitätsring 1 1010 Wien Austria manuela.lanwermeyer@univie.ac.at

© 2020, Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen ISBN Print: 9783847111443 – ISBN E-Lib: 9783737011440