# Negatives and Meaning: Social Setting and Pragmatic Effects

This volume deals with the pragmatic dimension of negations and is oriented towards empirical studies of negatives' meanings and functions in media and public discourses.

By bringing together scholars from different countries, with studies on different languages this volume aims to shed light and contribute to new knowledge about the forms and functionality of negation as a universal phenomenon. Linguists within Pragmatics generally agree that the use of negatives escapes logic and pure semantic description and is therefore best analysed with tools Negatives and Meaning: Social Setting and Pragmatic Effects Roitman (ed.)

from cognitive and pragmatic theories.

Based on hypotheses within pragmatics, semantics and discourse analysis, the main assumption is here that forms of expressing negatives emerge and adjust constantly and in accordance with the cultural domain and the social setting of their appearance. This is why this volume focuses on the functions of negative expressions in specifi c domains

and types of discourses.

Using Negatives in Political Discourse, Social Media and Oral Interaction

> Edited by Malin Roitman

## **Negatives and Meaning: Social Setting and Pragmatic Effects**

## **Using Negatives in Political Discourse, Social Media and Oral Interaction**

Edited by

*Malin Roitman*

Published by Stockholm University Press Stockholm University SE-106 91 Stockholm Sweden www.stockholmuniversitypress.se

Text © The Author(s) 2023 License CC BY 4.0

Supporting Agency (funding): Department of English, Stockholm University

First published 2023 Cover designed by Petter Blomqvist, Stockholm University Press

Stockholm Studies in Romance Languages (SSIRL) ISSN: 2002-0724

ISBN (Paperback): 978-91-7635-202-1 ISBN (PDF): 978-91-7635-203-8 ISBN (EPUB): 978-91-7635-204-5 ISBN (Mobi): 978-91-7635-205-2

DOI: https://doi.org/10.16993/bcd

This work is licensed under the Creative Commons Attribution 4.0 Unported License. To view a copy of this license, visit creativecommons.org/licenses/by/4.0/ or send a letter to Creative Commons, 444 Castro Street, Suite 900, Mountain View, California, 94041, USA. This license allows for copying any part of the work for personal and commercial use, providing author attribution is clearly stated.

Suggested citation:

Roitman, M. (ed.) 2023. *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*. Stockholm: Stockholm University Press. DOI: https://doi.org/10.16993/bcd. License: CC BY 4.0.

To read the free, open access version of this book online, visit https://doi.org/10.16993/bcd or scan this QR code with your mobile device.

## **Stockholm Studies in Romance Languages**

Stockholm Studies in Romance Languages (SSIRL) is a peer-reviewed series of monographs and edited volumes published by Stockholm University Press. SSIRL strives to provide a broad forum for research on Romance Languages of all periods, including both linguistics and literature.

In terms of subjects and methods, the journal covers language structure, variation and meaning, spoken and written genres, as well as literary scholarship in a broad sense.

It is the ambition of SSIRL to place equally high demands on the academic quality of the manuscripts it accepts as those applied by refereed international journals and academic publishers of a similar orientation.

## **Editorial Board**


## **Titles in the Series**


## **Peer Review Policies**

Stockholm University Press ensures that all book publications are peer-reviewed in two stages. Each book proposal submitted to the Press will be sent to a dedicated Editorial Board of experts in the subject area as well as two independent experts. The full manuscript will be peerreviewed by chapter or as a whole by two independent experts.

A full description of Stockholm University Press' peer-review policies can be found on the website: https://www.stockholmuniversitypress.se /site/peer-review-policies/

### **Recognition for Reviewers**

The Editorial Board of Stockholm Studies in Romance Languages applies single-blind review during proposal and manuscript assessment. We would like to thank all reviewers involved in this process.

Special thanks to the reviewers who have been doing the peer review of the manuscript of this book:


## **Contents**

Illustrations ix

## **PART ONE: INTRODUCTION**

1. Negative Form, Negative Meaning and the Impact of the Sociocultural Context 1 *Malin Roitman*

## **PART TWO: FORM-TO-MEANING NEGATIVES**

2. Negation and Climate Change in French Blog Posts 27 *Øyvind Gjerstad and Kjersti Fløttum*

3. Negative Campaigning: Communicating Negative Meanings in French Presidential Debates Over Time 43 *Malin Roitman and Bonnie Fonseca-Greber*

4. Metadiscursive Negation, Evidential Points of View and *Ethos* in Argentine Political Discourse 83 *María Marta García Negroni*

5. A Corpus-Pragmatic Account to Negation in Electoral Tweets 113 *Elena Albu and Francesca Capuano* 

6. A Corpus Study of Grammatical Negation in US Presidents' Inaugural Speeches 145 *José Manuel Durán*

## **PART THREE: MEANING-TO-FORM NEGATIVES**

7. Counterfactuality as Negative Meaning: A Case Study of *BE Supposed To* 179 *Anne-Laure Besnard*

8. The Meaning of Teachers' Negations in Hong Kong Classrooms Interpreted from their Co-Occurring Gestures 203 *Renia Lopez-Ozieblo*

9. Ontological Change Caused by Negation: The Case of Identity Statements 239 *Tomohiro Sakai* 

About the Authors 269

## **Tables and Figures**

## **Tables**


### **Figures**


## **PART ONE: INTRODUCTION**

## **1. Negative Form, Negative Meaning and the Impact of the Sociocultural Context**

*Malin Roitman*

## **1. Research field and aim of the book**

This is the second volume the editor will have assembled on negation studies. The present book, as well as the first (Roitman, 2017a), deals with the pragmatic dimension of negations. It originated from a 2017 conference at Stockholm University, *The Pragmatics of Negation: Aspects of Communication*, organised by the present book's editor. While the first book (Roitman ed., 2017a) covered negation studies on pragmatic matters from a wider range of linguistic fields, the present publication is more oriented towards empirical studies of negatives' meanings and functions in media and public discourses. The performed analyses are methodologically and theoretically oriented towards models in French pragma-semantics, enunciation, and cognitive theories.

Negation is one of our most central phenomena in human language and we use it daily for a vast range of different purposes: for rejection, denial and for expressing non-existence. Since ancient times it has captivated scholars, logicians and philosophers (Plato, Aristotle, Frege, Kant, Russell and Wittgenstein1 ), and the very last century linguists – syntactiens, semanticists, pragmatists, sociolinguists and psycholinguists (Carston 1996; Clark and Chase, 1972; Dahl, 1979; Ducrot, 1984; Givón, 1979; Horn, 1989; Katz, 1972; Klima, 1964; Labov 1972; Larrivée, 2004; Miestamo, 2005; Moeschler, 2006; Muller, 1991; Tottie, 1991) – have been intrigued by its evasive and versatile character. Being one of the so-called *semantic universals*, i.e. meaning components shared by all languages studied so far, reveals its deep importance in human expression (Wierzbicka, 1996). Negation is certainly one of the most multidimensional and complex units in

#### **How to cite this book chapter:**

Roitman, M. 2023. Negative Form, Negative Meaning and the Impact of the Sociocultural Context. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 1–24. Stockholm: Stockholm University Press. DOI: https://doi.org /10.16993/bcd.a. License: CC BY 4.0.

language, semantically, cognitively and syntactically, as well as from a functional, pragmatic, perspective.

Negations (negative morphemes) and negatives (here: sentences with negative meaning) have been analysed from an evolutionary perspective and synchronically, from a language internal or a language universal perspective. Depending of the theoretical framework, sentence negation in particular has been identified as a modal operator, a truth-value operator, a rhetoric device, a figure of thought, a polarity item and a marker of linguistic polyphony and as a linguistic unit with a variety of discursive and contextual meanings.

There remain, nevertheless, a large number of unsolved questions regarding negative forms of expressions and negative functions within specific languages, within different social settings and throughout the languages of the world. By bringing together scholars from different countries, with studies on different languages this volume aims to shed light and contribute to new knowledge about the forms and functionality of this universal phenomenon. Linguists and pragmaticiens generally agree that the *use* of negatives escapes logic and pure semantic description and is therefore best analysed with tools from cognitive and pragmatic theories. The variety of languages and different approaches in the book is by no means a disadvantage. Since the common denominator is analysing the functions and meaning of negatives, the different language specific parameters are pretty much an advantage when looking at the volume as a whole. Similar themes connected to negatives approached from different perspectives and examined in different languages offer a contrastive reading that actually enlarges the spectra of new knowledge presented in the books's chapters.

Based on hypotheses within pragmatics and discourse analysis, the main assumption is here that forms of expressing negatives (along with other forms of expressions) emerge and adjust constantly and in accordance with the cultural domain and the social setting of their appearance. This brings us to the second important and common denominator of the book's chapters, which is to study the functions of negative expressions in specific domains and types of discourses. The term *negatives* will here be used to cover *negative meaning* in general, from sentences containing negative morphemes – markers – to any sentence interpreted as bearer of negative meaning. I will come back to this dichotomy later on in this introductory chapter. Before continuing to discuss *negatives*, thus a *negative-meaning category*, we will try to clarify some formal issues necessary for understanding this complex phenomenon.

### **2. The relation between negative and declarative sentences**

One of the first issues when it comes to analysing negatives – our general category for negative valuated meaning and function – is to problematise and clarify their status towards negative clauses2 and negative sentences. Let's however leave the semantic category (negatives) for the moment and first straighten out some aspects regarding the relation between negative clauses and negatives sentences on the one hand – and then the relation – distinction –between negative sentences and declarative sentences on the other. The *negative clause* is first of all the simple morpho-syntactical structure of a subject and negated lexical predicate, and the *negative sentence* is basically a negative clause that may also contain a more complex structure including one or more negative morphemes and other syntactic operators (quantifiers, modalities, etc.) with different scope etc.; the negative sentence is to put in other words a well formed group of words including at least one negative marker, starting with a capital letter and ending with a full stop, an exclamation or an interrogative mark. Henceforward we will use *negative sentence* or *sentential negation* to cover all types of negative clauses3 .

Earlier comparisons between negative sentences and declarative sentences have exposed many complex linguistic, cognitive and functional features of negative clauses in natural languages that distinguish them from declaratives. This appears in particular in the last decades' pragmatic studies where the relation between form and literal meaning on the one hand and language function and pragmatic meaning on the other has been highlighted. Syntactically, negative morphemes generally bring forth a more elaborate distributional syntactic pattern; cognitively, negative sentences are proven to require more time for the understanding process, and pragmatically negative morpho-syntactic structures can be used for a varieties of meaning in different contexts (See for instance Clark and Chase, 1972; Dahl, 1979; Horn, 1989, Kaup et al., 2006; Lee, 2017; Roitman, 2017a; Tian et al., 2016).

These features characterising negative sentences are interrelated in various ways and seem to explain some aspects of its usage. In order to obtain a communicative and cognitive flow in discourse, and in the light of general information structure, negative sentences seem actually to require an 'alert' to be fully interpreted; language users in general and independently of context generally communicate how things *are*, and since negative sentences communicate how things *are not*, the latter may need reinforcement in discourse (Haspelmath, 2006; Miestamo, 2005). This communicative 'break' and the apparent need to highlight negative content in discourse may therefore be a plausible explanation to why negative sentences across languages to a high extent engender distributional patterns different from corresponding affirmatives and why negative morphemes have a tendency to appear early in the sentence such that it has been shown by Horn (1989) and Jespersen (1917). This distributional pattern seems to enhance a good communicative flow for the reason just mentioned. It was early suggested by Meillet (1912) and recently by Larrivée (2011) and others that negative morphemes' distribution over time may be motivated by pragmatic needs, rather than being solely a result of the phonetic evolution, as it was suggested by Jespersen. Mosegaard-Hansen (2009, 2021) has also shown in diachronic studies on the French bipartite negation *ne…pas* that the evolution of standard negation is ruled by discursive related principles of communicative flow in the information structure. It seems as if unexpected information, such as the rejection and denial of the state of affairs needs to be signalled.

The distinctions versus similarities between negative sentences and declaratives across languages have been widely studied within the language typology framework in terms of the *asymmetry* versus *symmetry* of standard negation, in relation to a declarative clause (van der Auwera and Krasnoukhova, 2020; Deprez, 2000). Our interest here lies however in the *functional* asymmetry of negative sentences (Givón, 1979; Miestamo, 2000, 2005). The functional asymmetric relation between standard negation and declaratives is in essence a rather uncontroversial postulate in modern linguistics. It is an almost indisputable fact that simple propositional logic cannot fully explain the function of negation in natural languages; thus the logic of the negative operator in *¬P is true if and only if P is false* (and vice versa *¬P is false if and only is P is true*) is not enough to explain the semantic complexity of a negative sentence. A negative sentence is normally a much more complex semantic phenomenon than a simple reversed affirmative (above), due to the way the negative morpheme interacts and creates meaning with other language items such as modality markers and quantifiers, but also in the way the sentence is used and interpreted in authentic contexts.

The longer process time for negative sentences, than for affirmatives, is actually related to the fact that a negative sentence involves more intricate semantic features – which allows it being used for different purposes – and therefore demands more time for the interpretation of it (Carston, 1996; Kaup and Zwaan, 2003; Kaup and Dudschig, 2020). The sentence negation (English *not* or French *ne…pas*) that is generally used for denials (of some issue) actually triggers a more or less manifest *activation* of the positive counterpart that underpins the negative sentence (See below and 3.2 on *Enunciation theories* etc.). The degree and impact of this activation is due to the quality of triggering elements in the surrounding context. This indicates that this *stratifying* of a negative sentence in two layers, a positive and a negative, that takes place being both a semantic language inherent phenomenon and a pragmatic ditto. Catching negative meaning thus demands – as it has been mentioned earlier – to higher extent an interpretation, which goes along with the longer process time for sentences containing a negative morpheme. Experimental research on the process time of negative sentences has however shown that this processing difficulty is mitigated with contextual support (Nordmeyer and Frank, 2014). It can also be added that negative sentences constitute a late acquired feature in first and second language leaning. Studies of children's language show evidence of larger efforts involved in their interpretation of negative sentences (Bardell, 2000; Leech, 1983).

This global idea of the activation of the positive counterpart in negative sentences to explain the meaning and functions of negative sentences is actually framed in several theories, using different terminology and models for explanation that will be further explained below and in the books' chapters (Fauconnier, 1994; Culioli, 1990; Martin, 1983; Givón, 1979, etc. to mention some). This idea is also what outlines the explanation of negation in the theory of linguistic polyphony (Ducrot, 1984; Nølke, 2017) repeatedly referred to in this book. Since our approach is a pragmatic one, the question of mapping affirmatives and negative sentences to the logic of true and false statement is for reasons mentioned earlier less important than the studying of their potential meanings, in a variety of contexts.

## **3. Fundaments and theoretical orientations of this book**

### 3.1. The pragmatics of negative meaning

This book focuses henceforth mainly on the emergence of *negative meaning* such as it is engendered in different social domains: political and media discourses and social interaction. The studies deal with questions regarding negative meaning and function in a broad sense, which implies interpreting negative utterances based on text type, genre, or sociocultural factors. The linguistic analysis will not be abandoned but the sociocultural aspect will be an important parameter for the interpretation of negative meaning.

The chapters cover thus analyses of *negatives* (negative meaning) from two perspectives: *form-to-meaning/function* and *meaning/function-to-form*. Thus, some of the studies are carried out on negatives sentences carrying negative morphemes, (quantifiers and adverbs: *no, nobody, not, n't, never*, etc.) while others chapters deal with negative meanings conveyed through other units or linguistic phenomena: counterfactual elements, prosody, stress, gestures, etc. involved in expressions and contexts producing negative meaning. Negatives shall thus, to put it differently, here be approached and considered from semasiological as well as from onomasiological standpoints; the differentiation between negative form and negative meaning is omnipresent in the book since there is no automatic mapping between them: affirmatives can produce negative meaning (irony, as for example "I'm excellent at predicting the weather!" declared on a rainy day at the beach) and sentences with negative morphemes can produce positive meaning such as in: *That's not too bad!* (litote) or *I didn't lose him* (double negation). When it comes to studying negative meaning in discourse (as opposed to positive meaning) there is consequently more to it than to differentiate standard negation *Paul didn't eat the apple* – negation of a main lexical verb – from an affirmative clause *Paul ate the apple* although this and other syntactic and semantic categories (mentioned earlier in the chapter) will be referred to whenever there is a need for it.

The focus lies here, once again, on determining what is negative meaning from a pragmatic point of view4 . The noun *negative* is hereby defined as a category of sentences that express negative meaning, will it hold negative morphemes or not. More precisely *negative* is a statement indicating or expressing a contradiction, denial, non-existence or refusal. *Negativity* will be used to design the outcome of a negative (the noun in the sense above). The adjective *negative* will refer to designating a proposition that somehow denies agreement between a subject and its predicate or to design a linguistic element as in "negative polarity" or "the French negative adverb *pas*".

Negative meanings (or negatives) emerge thus from interpretations grounded in the communicative situation where a particular sentence is used. Every chapter will present their methods and criteria for interpreting negative meaning in their corpuses. In accordance with the purpose, the studies in this book are thus framed in theories and methods within pragmatics, in a broad sense; these are French pragmatic theories on argumentation, enunciation, presuppositions, and polyphony (Benveniste, 1966; Anscombre and Ducrot, 1983; Ducrot, 1984; Carel and Ducrot, 2005; Culioli, 1990; Nølke, 2017), speech acts theory (Searle, 1969), systemic functional theory (Halliday and Hasan, 2000; Halliday and Matthiessen, 2014), cognitive pragmatics and the model of mental spaces (Fauconnier, 1994), sociolinguistics, in particular variation aspects and discourse analysis. Earlier studies on the pragmatics of negation also play an essential role in the chapters' theoretical framework, such as Horn (1989), Larrivée (2004, 2018), Moeshler (2016), Muller (1991), and Miestamo (2005).

Besides pragmatic theories, the questions dealt with in the different chapters are related to methods, models and theoretical postures in other linguistic fields, notably semantics, rhetoric, cognitive linguistics, syntax, language acquisition and paralinguistics. Some studies are primarily theory-based while others are more empirically oriented, though all are methodically oriented towards demonstration and reasoning through authentic corpora.

#### 3.2. Enunciation theories and the polyphony of negation

Various chapters deal with enunciation, in Ducrot's meaning. According to this view, the meaning of a sentence is a result of its own production – its enunciation, and language units thus hold *traces* of the speaking subject(s), the interlocutors, space and time in terms of personal pronouns, tense and modality markers (epistemic and axiological) and deictic expressions. This is the fundamental pillar of the Anscombre and Ducrot's (1983) framework *Argumentation dans la langue* which outlines *la pragmatique intégrée*, the idea of the enunciation process being an integrated part of the meaning of specific language phenomena, such as scalar words, presuppositions, connectors and negations. Ducrot (1984) further develops this idea in his theory of linguistic polyphony, where sentence negation (among other items) is described as the polyphonic 'multi-voiced' marker par excellence. According to this view, sentence negation discloses a 'crystallised discourse' exposing two different enunciators or 'voices', one positive and one negative. The linguistic polyphony has been developed and practised in a number of studies, of which Nølke (2017) and Nølke *et al*. (2004) is the most influential among others (Bres *et al*., 2005a; Bres, 2005b ; Kronning, 2009; Nowakowska, 2005; Perrin, 2009; Therkelsen, 2009). As for the polyphony of negation, it has beeen researched by Birkelund (2009), Fløttum and Gjerstad (2017) and Roitman (2015, 2017b) and others. This subdivision of negative sentences in two 'voices' challenges the established idea of the *unity of the speaking subject*. As mentioned earlier in this introduction, the idea of stratifying negative sentences, the superposing of layers 'enunciators', is present in other theories besides the models inherited from the French enunciation theories (see below), although the latter are predominant in this book.

### 3.3. Negative sentences and the crossroads between the semantic and pragmatics

With regard to the dominant enunciative perspective on negation in this book, an important issue will be the studying of the crossroads between the semantic content and the pragmatic functions of sentence negation. The pragmatic distinction between the so-called descriptive, polemic, and metalinguistic negations is widely dealt with within pragmatic studies on negation (Ducrot, 1984; Horn, 1989; Larrivée, 2011; Moeschler, 2016 etc.) and implies queries on whether to conceive linguistic meaning and negative meaning in particular as truth conditional or non-truth conditional. Some pragmatic theories applied in the book argue in favour of an overlapping between these two conceptions of meaning (Austin, 1962; Fauconnier, 1994; Horn, 1989; Searle, 1969; Sperber and Wilson, 2004) and some other (Culioli, 1990; Ducrot, 1984; Nølke, 2017) stand for a radical non-truth conditional model for linguistic meaning. This fundamental question will among other questions be problematised and discussed in the books' chapters.

## **4. Theoretical frames of the chapters**

The theories applied in the books' chapters will now briefly be framed in order to position this volume in the field of the pragmatics of negation. The chapters are organised according to whether their studies have a form-to-meaning or a meaning-to-form approach on negatives.

## 4.1. Negative forms and negative meaning

Fløttum and Gjerstad aim to explore the polyphony of negation and its impact on argumentation in environmental discourses, notably French blog posts on climate change, a highly confrontational forum where opposing views on this issue is exposed. Framed in the theory of linguistic polyphony (Nølke et al., 2004 and Nølke, 2017) their analysis searches for a matching of the divergent views on climate change and the polemic negation in order to decide whether the polyphony of negation is a characteristic tool for argumentation in this type of discourse.

The theory of linguistic polyphony also frames Roitman and Fonseca Greber's study on the dichotomy of polemic-descriptive negation in relation to the *ne*-loss and *ne*-retention in modern French political discourse. The interest lies in the relation between the *ne*-retention and a communicative, pragmatic need to emphasising the negative content within a specific type of discourse, the highly confrontational presidential debates. The evolution of French negation from a preverbal unit to a two-folded negation (*ne…pas, ne…plus*), the grammaticalisation of the post verbal nominal units, and the loss and retention of *ne* has been studied from a chronological and syntactical perspective (Dahl, 1979; Jespersen, 1917) but less from a pragmatic point of view. One hypothesis for this study is that the social setting does actually have an impact on the *ne*-loss and *ne*-retention observed in the French presidential debates and in general. Based on sociolinguistic criteria for language variation (register, relation between the interlocutors, social setting, genre) comparisons of the *ne*-loss and the *ne*-retention are therefore made with spoken corpora that differs with respect to most variation parameters.

The interpretation of the negatives' context is also crucial in García Negroni's study. The author explores evidentiality (the marking of the source of information in the utterance and the relating of it to a referent in the world) in relation to metalinguistic negations in Argentinian politicians' speeches. The originality of her study is to elaborate the concept of evidentiality within French enunciation theory, thus considering external sources as utterance-internal dialogic and polyphonic phenomena in the light of this non-truth-conditional approach to meaning. Metalinguistic negations are in Ducrot's definition the kind of negations that rectify external enunciations (the act of saying) such as in "He didn't 'die in a car accident'. He is still alive!". The metalingustic negation disqualifies thus in this example the presupposition (that someone actually died), which would be the default instruction for this negative sentence outspoken, without the cue *He is still alive!* The metalinguistic negation disqualifies also the *decreasing principle*  of a canonical negative sentence that would allow a cue such as for example "The truth is he *sat* in his car when he had a heart attack (and died)". In García Negroni's view, metalinguistic negation creates its own discursive and interactive frames where evidential points of views are staged and rectified. There are interesting parallels between this study and Besnard's who discuss countrafactual meaning in relation to evidentiality.

Two of the chapters studying negation as a marker of linguistic polyphony (Roitman and Fonseca-Greber, and Lopez) relate the two negation types — polemic and descriptive — to extra-linguistic speech features (intonation and stress) and paralinguistic markers (gestures) and then analyse them in the light of sociolinguistic matters of variation. Roitman and Fonseca-Greber study hence (also) how prosodic patterns interfere with the argumentative reinforcements of negative sentences in political debates by comparing the differences in terms of social setting. Their study is however focused on negative morphemes. Lopez, on the other hand, primarily relates negative gestures to different types of negations and we have therefore chosen to associate her study to the section 4.2 below where it will be further described.

Negatives' function and meaning related to the social setting and genre are also particularly significant in two large corpus studies. Albu and Capuano examine the distribution and functions of the negative items, negative quantifiers (*no-negations*) and sentence negation (*not/ n't negations*) in English-language tweets from UK candidates running for the EU parlament. They problematise the classic spoken-written dichotomy in the light of the social setting in this specific hybrid type of discourse that is dialogic in nature. Different distributional and collocation patterns are found in the use of *not* and *n't* which can sometimes be related to the informal style of this genre where the limits between spoken and written is rather unsettled.

Within the functionalist 'choice-grammar' framework (Halliday and Hassan, 2000), Durán studies negative polarity in American presidential inaugural speeches from the president Washington to Trump. The functionalist top-down approach allows scanning the pattern of choices from the negative system and then a comparison of the frequency and nature of negative polarity items at a given time, between the different speeches over time, and with other types of discourses. This method of scrutinising elements with negative meaning and function – polarity – without primarily taking into account their syntactic structure and semantic denotation is efficient for uncovering the nature of negatives in specific domains, discourse type and genre (*field, tenor* and *mood* in the functionalist terminology). Durán shows that negative polarity is to a higher extent represented in the presidential inaugural speeches than in other English language domains, in general. The level of polarisation in these speeches reveals actually the global stand the presidents take in relation to their predecessors.

### 4.2. Negative function and negative meaning

Non-negative forms may engender negative meaning and ironic utterances are probably the example par excellence of this. Although the book's chapters don't deal directly (but only indirectly) with irony, it may serve as an illustrative example of implicit negative meaning. Irony reverses literal meaning and may therefore express negativity by means of an affirmative sentence, under certain circumstances. This delicate problem of irony and negativity has been studied within the linguistic polyphony framework (Birkelund and Nølke, 2013; Bres, 2010; Dendale, 2008; Ouaz 2010; Perrin 1996). Irony is there described as a polyphonic phenomenon, where what is explicitly enounced covers another opposite enunciation. Thus when the speaker holds the implicit (and opposite) enunciation as true, ironic utterances express negative meaning without necessarily containing negative linguistic forms.

From the negative meaning-to-form perspective, Besnard examines in her chapter the idea of counterfactual meaning as a built-in potential of certain expressions containing a meaning of 'not to be the case' such that the expression *be supposed to*. The framework is Culioli's (1990) analysis of negation within his theory of enunciative operations. From this viewpoint a counterfactual linguistic item may be considered as an implicit negation expressing at least two different values for a given predicative relation: *p* and *p'* (*non-p/ other than-p*). Culioli's semantic theory has a cognitive dimension where negative meaning (and other meanings) are achieved through predicative operations in different notional domains of possible representations, supported by enunciative operations that locate the situation and the speaker's position. From this perspective the triggering of negative meaning of 'be supposed to' can be described as a result from the interaction between the counterfactual expression's *p'* value (*non-p/ other than-p*), other reinforcing facts in the context, and the act of enunciation, i.e. the contextual linking of the predicative operation to the speaking subject and place. This calculation of meaning is pragmatic since it reveals a constant dynamic relation between mental potential representations and the enunciative condition of an utterance.

Lopez explores the correlation between negative gestures and negative utterances performed in teachers' classroom discourse and if so whether there are correlations between different types of gestures and the three types of sentence negation described within French enunciation theories: descriptive, polemic and metalinguistic negations. Some of the gestures used for expressing negative meaning in this corpus appear to be language specific although some match universal known gestures for negations. Earlier studies within the cognitive field (Kaup and Dudschig. 2020; Giora, 2006) have explained the complexity and the longer process time for interpreting negative sentences in speech (See also introduction). Inspired by these studies Lopez finds a correlation between different gestures and the pragmatic functions of the negative sentences performed by the teachers. The polemic negation is accompanied by typical negation gestures but as for the descriptive negation the accompanying gestures seem to be motivated by a volatile act of supporting the audience to process negation. It is of course interesting to consider the results in relation to the social setting – classroom – and the type and function of the discourse (pedagogical).

Sakai's chapter represent a somewhat different conception of the pragmatics of negation. This study deals with the act of reference and how it differs between affirmatives and negatives, according to the *ontology* referred to. This study has a conceptual-cognitive approach to negative meaning and deals with the choices of the adequate referent to capture a plausible meaning in a given context. The occurring ontology change in the reference act, pragmatic in nature, in certain negated identity statements is here explained within Fauconnier's theory on mental spaces. This model explains the stratifying and the duality of sentence negation from a cognitive point of view, where different "*univers de croyance*"5 mentally overlap and interact and help to explain the interpretability of apparently illogical relations of certain utterances. Sakai's analysis shows that sentence negation – when it comes to certain identity statements – operates not on the truth conditional content of the proposition but on the *modes of representation* of the items denoted in it. This is an example of an analysis where there is an overlapping of truth-conditional and non-truth conditional – pragmatic – perspectives on linguistic meaning.

## **Abstracts**

## A corpus-pragmatic account to negation in electoral tweets *Elena Albu and Francesca Capuano, University of Tübingen*

This paper aims at discussing the constructional strategies and pragmatic uses of *no*-negation, *not*-negation and *n't*-negation in the political tweets sent by the UK candidates at the time of the 2014 European Parliamentary Elections. Using the tools and methods of corpus pragmatics, this is an exploratory study meant to cast light on the on-going debates about the oral vs. written features present in tweets and about the colloquialisation of political discourse on social media. The analysis revealed that in terms of overall frequency, negation is not extensively used (17.18%), while the tripartite division showed that *not*-negation and *n't*-negation are prevalent (72.44%) in comparison with *no*-negation (19%). Although numerically similar, the analyses of the first ten token collocates and of the first four most used parts of speech indicated that *not*-negation and *n't*-negation are not used interchangeably, and instead form distinct patterns and have different combinatorial preferences. To illustrate, *not* is generally part of non-verbal clausal negation, being mainly found in the [(X') not (X)] construction. It is also used in elliptical structures and followed by full stops, features which point rather towards non-standard values of the negative particle. In contrast, the bound inflectional form *n't* showed a strong preference for the auxiliary do, while no was found in weak recurrent patterns as a result of the great variety of items it combines with. Overall, in line with Wikström (2017), the electoral tweets in our dataset appear to be neither a form of spoken language nor written language but rather a hybrid form that extends beyond a mix of linguistic features. Additionally, the tweets present features that indicate a shift from the formality imposed by traditional political discourse to a more flexible and colloquial type of political discourse.

## Counterfactuality as negative meaning. A case study of 'BE supposed to' *Anne-Laure Besnard, University of Rennes*

The aim of this paper is to investigate how markers that are not typically negative may generate negative interpretations in context. More specifically, it focuses on the counterfactual, which can be considered a type of implicit negation insofar as it involves the expression of a state of affairs that is understood not to be the case. The issue of counterfactuality as negative meaning is approached via a case study of the quasi-modal marker *BE supposed to* within the framework of the Theory of Predicative and Enunciative Operations (Culioli, 1990). Drawing from a 40-million-word newspaper corpus (*The Independent* 2009), the study shows that this structure is more likely to give rise to a counterfactual interpretation than other apparently similar periphrastic expressions such as *BE expected to* or *BE believed to*, which suggests that what might look at first sight like a purely pragmatic phenomenon is actually rooted in semantics.

### A corpus study of grammatical negation in US presidents' inaugural speeches

### *José Manuel Durán, Universidad de Belgrano, Buenos Aires*

Systemic Functional Linguistics (SFL) describes the system of grammatical negation as realised through negative polarity items such as *not, never, no*. This paper analyses the frequency of grammatical negation per clause in a corpus of political speeches from an SFL perspective. The corpus is made up of the 45 inaugural speeches (122,848 words, 10,498 clauses) delivered by US presidents, from Washington 1789 to Trump 2017. The corpus was semi-automatically tagged with the aid of WodsmithTool and UAM CorpusTool. The study aims at discovering the patterns of the most frequent polarity items in the corpus and contrasting them with those in the overall pattern of English. Additionally, the chapter analyses the most frequent collocations and colligations of the most pervasive polarity items. Results show that polarity items in my corpus are much higher than those found in the literature (Halliday and James, 1993/2005, Matthiessen 2006). Besides, grammatical negation is found to be twice as frequent at clause level than at the level of the noun group.

### Metadiscursive negation, evidential points of view and ethos in Argentine political discourse

### *María Marta García Negroni, Universidad de San Andrés—Universidad de Buenos Aires*

Most often approached from referential or cognitive perspectives, evidentiality is usually understood as the semantic domain marking the existence of the source of information in the utterance and specifying what type of source—direct or indirect—it involves (Aikhenvald, 2004). The source is said to be direct when the knowledge the speaker refers to has been acquired by means of a perception arising from one of their senses, while it is called indirect when such knowledge derives from an inference or a quotation of somebody else's discourse (Anderson, 1986; Willet, 1988).

This chapter will focus on the analysis of the *evidential meaning of metadiscursive negation*. However, to account for such meaning, this study will drift apart from many of the assumptions on which most studies on evidentiality rest. On the research paths paved by the theories of polyphony (Ducrot, 1984, 2001), dialogism (Bakhtin, 1981, 1984), and argumentative semantics (Ducrot, 2004; Carel and Ducrot, 2005), the *dialogic approach to argumentation and polyphony*, within which this chapter is framed, advocates a non-truth-value, non-referential characterisation of meaning (*i.e.*, there is no meaning component that can be considered purely objective).

## Negation and climate change in French blog posts *Øyvind Gjerstad and Kjersti Fløttum, Universitetet i Bergen*

Can the use of negation be seen as a metric for the contentiousness of an issue? That is the basic assumption that forms the point of departure for the present chapter. By expressing a diametric opposition to another point of view (Ducrot, 1984, Nølke et al., 2004), negation has the potential to crystallise and reproduce two fronts of a given issue, and few issues are as societally important as climate change (CC). During the last few decades, the public prominence of different aspects of CC—the prognoses offered by science, the necessity of mitigation and adaptation, the division of respective responsibilities of various nationstates—have ebbed and flowed as the result of political and natural events such as the Kyoto summit in 1997 and the California wildfires in 2018. The year 2007 stands out as particularly important in this regard as the Intergovernmental Panel on Climate Change (IPCC) and Al Gore were jointly awarded the Nobel Peace Prize, a recognition of their efforts to inform the global population of the risks of CC. The award also had the added effect of amplifying this very message. However, the following years were characterised by both disappointment (e.g., the CC summit in Copenhagen in 2009) and controversy (e.g., the release of internal e-mails from the University of East Anglia, also referred to as 'climategate', in 2009). All the while, the IPCC's prognosis of CC grew more dire, as laid out in its 5th Assessment Report published in 2013. While the panel's mandate is only to give a summary of the available science, it did provide a subtle rebuttal to one specific argument advanced by CC sceptics, in the form of negation.

### The meaning of teachers' negations in Hong Kong classrooms interpreted from their co-occurring gestures

### *Renia Lopez-Ozieblo, Hong Kong Polytechnic University*

In an Asian pedagogical context, teachers' negations need to take into account Asian politeness attitudes (Gao and Ting-Toomey, 1998; Cheng and Tsui, 2009) as well as the pedagogical objective of the negation (Rees-Miller, 2000; Seedhouse 1997) and the difficulties inherent in its processing (Tian *et al*., 2016). Despite these issues, teachers do use negative particles in the classroom when answering students' questions or delivering content. This paper focuses on two Hong Kong tertiary teachers' explicit negations and the hand gestures that co-occur with them. I investigate how the gesture mitigates or accentuates the negation and the possible reasons for these actions. It would seem that negating gestures are more likely to co-occur with polemic negations while stressing and referential gestures are linked to descriptive negations.

### Negative campaigning: communicating negative meanings in French presidential debates over time

### *Malin Roitman, Stockholm University and Bonnie Fonseca-Greber, University of Louisville*

Given the well-documented, ongoing loss of *ne* in real-time during the last half-century (Armstrong and Smith, 2002; Ashby, 1976, 2001; Martineau and Mougeon, 2003; Hansen and Malderez, 2004), it is hypothesised that the remaining Spoken French negator, that is, *pas* [pa(:)] carries more negative meaning than it did 50 years ago and, therefore, it has become prosodically more prominent (e.g., through increased focal stress and/or vowel lengthening) in contexts where its negative meaning is paramount to the communicative/pragmatic intent of the utterance. The proposed chapter then explores, in realtime, the intersection of *ne loss* (and its concurrent reanalysis-in-progress for pragmatic emphasis (Fonseca-Greber, 2007, 2017; van Compernolle, 2009; Donaldson, 2017) and *pas prominence* in the Roitman corpus of televised French presidential debates (Roitman, 2009, 2015, 2017), that is, the 1974 and 2012 debates. Debates provide an ideal interactional counterpoint to friendly conversation, which seems to abide by the social agreement principle (Yaeger-Dror, 2002 ; Fonseca-Greber, 2017) because here, the candidates often argue and interact aggressively with each other, as if following a social *dis*agreement principle, instead.

## Ontological change caused by negation: the case of identity statements *Tomohiro Sakai, Waseda University*

This paper shows that the negation of (1) entails a change of ontology, which is pragmatic in nature.

(1) Clark Kent is Superman (1) is a covert existential, in that it is ontology-preserving just like an overt existential such as (2).


A difference emerges when the negation comes into the picture. If you accept the negation of (2), you can no longer hold (4) to be true. This is not the case for (1)/(3). Whether you assent to (1) or not has no bearing on the truth-conditional content of (3), affecting only the modes of presentation of the objects referred by 'Superman' and 'Clark Kent'.

## **Endnotes**

1. Regarding Aristotle's conception of negation, turn for example to Izgin (2020) and regarding Kant's, see Newton (2021). Plato examines the nature of negation in *The Sophist*. As for Frege, reference is made to *La Pensée, La Négation, La Composition des pensées* in *Écrits logiques et philosophiques*. (1918) and in the case of Russell reference is made to *On Denoting* (1905). Wittgenstein deals with negation in his *Philosophical Investigations* (1953).

2. The term *standard negation* is commonly used in the same way as negative clause i.e. a negation of a simple lexical predicate, although some researchers separate them.

3. Our focus lies on negative meaning and therefore this clause-sentence distinction is not necessary. See Miestamo 2005, Muller 1991 or Horn 1989 for further reading on these issues.

4. Negative meaning without negative markers has for example been studied within language typology by Miestamo (2000) from a semantic-universal perspective.

5. The term "univers de croyance" is borrowed from Robert Martin (1981) who developed a similar semantic model.

## **References**

Aikhenvaid, A. (2004). *Evidentiality*. Oxford: Oxford University Press.

Anderson, L. (1986). Evidencials, Paths of Change, and Mental Maps: Typologically Regular Asymmetries. In: W. Chafe and J. Nichols, eds., *Evidentiality: The Linguistic Coding of Epistemology*. Norwood: Ablex Publishing Corporation, pp. 273–312, https://doi.org/10.2307/415546


*Positions théoriques en confrontation*. Presses Sorbonne Nouvelle, Retrieved from http://books.openedition.org/psn/756


Leech, G. N. (1983). *Principles of pragmatics*. London: Longman.


## **PART TWO: FORM-TO-MEANING NEGATIVES**

## **2. Negation and Climate Change in French Blog Posts**

*Øyvind Gjerstad and Kjersti Fløttum*

## **1. Introduction**

Can the use of negation be seen as a metric for the contentiousness of an issue? This is the basic assumption that forms the point of departure for the present study. By expressing an opposition to another point of view (Ducrot, 1984, Nølke et al., 2004), negation has the potential to crystallise and reproduce two fronts of a given issue. Few issues are as societally important as climate change (CC). During the last few decades, the public prominence of different aspects of CC – the prognoses offered by science, the necessity of mitigation and adaptation, the division of respective responsibilities of various nation states – have ebbed and flowed as a result of political and natural events such as the Kyoto summit in 1997 and the California wildfires in 2018. The year 2007 stands out as particularly important in this regard, as the Intergovernmental Panel on Climate Change (IPCC) and Al Gore were jointly awarded the Nobel Peace Prize, in recognition of their efforts to inform the global population of the risks of CC. This recognition had the added effect of amplifying this very message. However, the following years were characterised by both disappointment (e.g. the CC summit in Copenhagen in 2009) and controversy (e.g. the release of internal e-mails from the University of East Anglia, also referred to as 'climategate', in 2009). All the while, the IPCC's prognosis of CC grew more dire, as laid out in its 5th Assessment Report published in 2013. While the Panel's mandate is only to give a summary of the available science, it did provide a subtle rebuttal to one specific argument

**How to cite this book chapter:**

Gjerstad, Ø. and Fløttum, K. 2023. Negation and Climate Change in French Blog Posts. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 27–41. Stockholm: Stockholm University Press. DOI: https://doi.org/10.16993/bcd.b. License: CC BY 4.0.

advanced by CC sceptics, in the form of negation (see Gjerstad and Fløttum, 2017):

(1) In addition to robust multi-decadal warming, global mean surface temperature exhibits substantial decadal and interannual variability (see Figure SPM.1). Due to natural variability, trends based on short records are very sensitive to the beginning and end dates and do not in general reflect long-term climate trends.

At first glance, the negation in (1) appears to correct a possible misrepresentation by the reader. However, it reads as far more polemic considering the context at the time. Sceptics had presented a global cooling between 1998 and 2012 as evidence that CC was not occurring, thereby seeking to undermine the IPCC's very mission, as well as the rationale for any global political agreement within the UN framework (Gjerstad and Fløttum, 2017). When read in this light, the IPCC's remark on how to read statistical data reads as a clear rebuttal to those voices.

While the IPCC's assessment reports set the premise both for media coverage and diplomatic negotiations on CC, it is difficult to discern any broader linguistic or discursive tendencies on the basis of a single text. The question is thus how, if at all, the use of negation in CC discourse at a societal level reflects the evolving stakes of the issue. To explore this question, the present study will analyse blog posts from the NTAP French blog corpus (see Section 2). Our research question is formulated as follows: to what extent does the polemical nature of CC discourse and its societal implications transpire through the use of negation? We hypothesise that negations in relation to CC are more frequent in the time period 2013–2014 than in the period 2007–2008, because of the increasing politicisation of the issue in the intervening years. Our findings will also be qualitatively compared to a sample of negations found in the corpus French Web 2017 (frTenTen17), serving as a reference corpus for the present study (see section 2). But first and foremost, this corpus will be used to explore two central questions that form the basis for this study: 1. Does negation correlate more frequently with lexical items associated with controversial topics than with those that are considered less controversial? 2. If such a metric proves to be fruitful, does climate change prove to be a controversial issue relative to others? This comparison will subsequently allow us to explore whether we can measure variation in the degree of contentiousness associated with climate change through the quantitative study of negation in the NTAP French blog corpus.

The study of climate change discourse in the present chapter is one among a series of analyses undertaken on the role of language in this field since the beginning of the 2000s (see Gjerstad and Fløttum, 2017). An agenda was set by the seminal paper published by Brigitte Nerlich and colleagues in 2010 which affirmed that: "Investigations of climate change communication cannot avoid attending to the role of language" (Nerlich et al., 2010, pp. 103). Another relevant reference is the book *Why we disagree about climate change*, by Mike Hulme, published in 2009, which provides an overview of the multi-faceted context in which linguistic representations of CC appear. One reason for the growing interest in language analyses of climate change discourse is the fact that CC has moved from being predominantly a physical phenomenon to being simultaneously social, political, economic, cultural, ethical, and, most importantly for this study, communicational. Research has shown that the meaning people ascribe to CC is closely related to how it is portrayed during communication. Thus, studies of words, of combinations of words, and of entire texts taken from different contexts, such as scientific reports, political documents, mainstream media, and now social media have been addressed in different quantitative and qualitative analyses, or in a mix of the two (for an overview, see Fløttum, 2016; see also Fløttum (ed.), 2017). The present paper adds to this research.

Our approach is based on the Scandinavian Theory of Linguistic Polyphony, or ScaPoLine (Nølke et al., 2004, Nølke, 2017b), which is heavily inspired by Ducrot (1984). According to this theory, there are linguistic markers that signal the presence of 'voices' other than that of the speaker at the moment of utterance. Among such markers are reported speech, argumentative connectives, and negation. Our choice of a polyphonic approach is justified by the fact that the climate change discourse is particularly "multi-voiced", and includes both explicit and implicit (or 'hidden') voices representing different actors and interests. In the present paper we focus on one polyphonic marker (Nølke et al., 2004), namely polemic negation, a choice that is based on a twofold justification. First, with some exceptions (e.g. Gjerstad and Fløttum, 2017), the potential of polemic negation as a marker of contentiousness in climate change discourse (about both causes of the phenomenon and solutions to tackle it) has not been fully explored. Second, information-wise, polemic negation seems to be the most efficient way of refuting an opinion; a single grammatical element is capable of invalidating an entire proposition. With regard to the choice of blogs as material for our study, we believe that this genre is particularly appropriate to our objective due to its discursive heterogeneity and large degree of interactivity (Gjesdal and Gjerstad, 2015, Fløttum, et al., 2019). A societal rationale for undertaking this research is that insights from linguistic studies contribute to an improved knowledge base, which allows for social and political actions to be undertaken in order to limit the dangerous consequences of climate change.

In the following section we will briefly present our material (table 2.1), before laying out and analysing our quantitative results (table 2.2). Section 3 will concern the difference between polemic and descriptive negations, and will ask the question of whether this difference is important to the reading of our quantitative results.

## **2. Material and quantitative analysis**

Our material is taken from the NTAP French blog corpus, accessible via the platform Corpuscle/Clarino at the University of Bergen (Meurer, 2012). This corpus consists of posts from 2,033 French blogs, related to the topic of climate change, and covers the years 1974–2014. The total word count is 1,506,074,082 words. It has been built through a topically-focused approach, which means that the blogs in this study have been selected according to the presence of lexical words broadly related to climate change. Our reference corpus is the French Web 2017 corpus (frTenTen17), comprising a total of 5.7 billion words collected from internet-based texts. By including texts from a number of internet based text genres, this corpus is well suited to explore our initial research questions: 1. Does negation correlate more frequently with lexical items associated with controversial topics than with those that are considered less controversial? 2. If such a metric appears to be fruitful, does climate change prove to be a controversial issue relative to others? Through simple frequency searches, we explored the correlation between negation and different words associated with varying degree of controversy. The search stipulated up to 10 words between 'pas' and the search words, yielding the results in table 2.1.

These results do come with an important caveat: The lexical items have not been chosen according to any objective criteria, meaning that the results must be considered as tentative. Nevertheless, the tendency seems clear: when talking about contentious topics, speakers or writers appear more likely to use negation than when they talk about less controversial topics. The results for 'climatique' and 'changement climatique' are quite surprising in this regard, as they indicate that climate change is a relatively uncontroversial topic on Francophone websites.


**Table 2.1.** Lexical items correlating with negation.

Moving on to the NTAP corpus, we first identified sentences containing the negation 'pas' ('not') in three combinations: with 'climat', 'climatique' and 'réchauffement', from all the blogs of the time periods 2007–2008 and 2013–2014. As in the reference corpus search, we allowed a text string of up to 10 words between 'pas' and the other search words. For the searches we used the statistics tool integrated in Corpuscle. This involved, first, eliminating all double occurrences (for example when both 'réchauffement' and 'climatique' appeared in the same sentence, this sentence entered in both the 'réchauffement' search and the 'climatique' search), and second, removing all occurrences of 'climat' and 'réchauffement' that were unrelated to the issue of CC, such as in 'le climat politique' or 'réchauffement de relations diplomatiques'. 'Climatique' did not require such a review, as it is almost exclusively used to denote CC. The remaining occurrences of the three words represent a wide variety of sub-topics, from the attribution of responsibility for CC to political measures and international negotiations:

(2) Ban Ki-moon a rappelé que, selon les derniers rapports du Groupe d'experts intergouvernemental sur l' évolution du climat (GIEC), "le réchauffement de la planète est sans équivoque, son impact est manifeste et il ne fait pas de doute que les activités humaines y contribuent de façon considérable" (2007\_climat)

 Ban Ki-moon has pointed out that, according to the latest reports of the Intergovernmental Panel on Climate Change (IPCC), "the warming of the planet is unequivocal, its impact is manifest and there is no doubt that human activities contribute to it to a considerable extent" (2007\_climat)

(3) "Il va nous falloir faire plus, et cela ne devrait pas être une question de politique partisane", a affirmé M. Obama, dont les ambitieuses promesses en matière de lutte contre le changement climatique se sont brisées depuis 2009 sur l'intransigeance des élus du Congrès. (2014\_climatique)

 "We will need to do more, and it shouldn't be a question of partisan politics", Mr. Obama stated. His ambitious promises regarding the fight against climate change have since 2009 been broken on the intransigence of members of Congress. (2014\_climatique)

The data provided by the machine searches and the manual analyses are presented in Table 2.2:


**Table 2.2.** Relative frequency of negation by year.

Contrary to our expectations, the relative frequency of negations related to CC is higher for the period 2007–2008 than for 2013–2014. This is despite a global increase in controversy during the intervening years. However, these numbers do not take into account the number of negations relative to the absolute frequency of 'climat', 'climatique' and 'réchauffement'. In other words, they do not account for the likelihood that a word pertaining to CC co-occurs with negation. If this likelihood correlates with the increase in the polemical nature of CC discourse, then our hypothesis is strengthened. However, as stated above, the corpus includes text material which has nothing to do with CC, such as the political situation in various Francophone countries. In order to remove this irrelevant data we decided to leave aside the words 'climat' and 'réchauffement' from our analysis, as they are frequently used in other contexts. Instead, we looked at the number of negations co-occurring with 'climatique', relative to its absolute frequency during each of the four years. The advantage of selecting only 'climatique' is that it is almost exclusively used in reference to the topic of CC.


**Table 2.3.** Frequency of negation and the word *climatique* by year.

The numbers in Table 2.3 indicate that during the time period 2013– 2014, there is a slightly higher likelihood of negation when talking about climate issues than during the period 2007–2008. When looking at individual years, the differences are more noticeable. The peak year is 2013, which is the same year as the publication of the IPCC's 5th Synthesis Report. This seems to support our initial hypothesis that the use of negation increases with the degree of controversy surrounding a given issue.

## **3. Qualitative analysis**

While the negation is *a priori* a polyphonic marker by virtue of refuting an underlying point of view (pov), it can also be interpreted descriptively, i.e. as a manner to characterise an individual or a state of affairs (Birkelund 2017). In such uses, the polemical interpretation of the negation is neutralised as a result of linguistic or contextual factors. An oft-cited example is *There's not a cloud in the sky*, which does not signal any disagreement regarding the weather, but is interpreted as a description akin to *The sky is blue* (Nølke 2017a, pp. 152)1 . This means that it is not sufficient to merely count the occurrences of negation to trace the changes in the construction of CC as a controversial issue. It is also necessary to qualitatively analyse a selection of occurrences from each time period, to see whether there is a change in the distribution of polemic and descriptive negations. There is an underlying assumption behind this question, namely that the share of polemic negations increases with the degree to which a given topic is contested. This is by no means a given, but it must be taken into account as a possibility2 .

The qualitative analysis consists of a random selection of 10 examples from each of the searches of 'climat', 'climatique' and 'réchauffement', from all four years (2007, 2008, 2013 and 2014), totalling 120 examples. The analysis follows mainly Nølke's work (1994, 2017a) on the linguistic and contextual factors which guide the interpretation of negation as either polemic or descriptive (see also Gjerstad and Fløttum, 2017).

Among the factors which reinforce the polemic interpretation are polyphonic and argumentative contexts, such as the concessive structure formed by *mais* (*but*) in (4):

(4) Le prix nobel de la paix a été remis conjointement, cette année, à AL GORE et aux experts du Groupement d'experts Intergouvernementaux sur l'évolution du climat (G.I.E.C.). Ainsi, l'écologie et le développement durable sont mis en avant, comme deux éléments nécessaires (mais malheureusement pas suffisants) pour garantir la paix à travers le monde. (climat\_2007)

 The Nobel Peace Prize has been given conjointly this year to Al Gore and to the experts of the Intergovernmental Panel on Climate Change (IPCC). In this way, ecology and sustainable development are put forth as two necessary (but unfortunately not sufficient) elements to guarantee peace throughout the world. (climat\_2007)

The counter-argumentative context creates several possible polyphonic interpretations. The negation could be interpreted as an attempt to avoid giving the reader the impression that the two elements are sufficient for world peace, an impression that is grounded in the fixed phrase 'une condition necessaire et suffisante'. Alternatively, the author could be attempting to avoid creating an impression of naiveté by inserting a caveat in parentheses, thus pre-empting a negative reaction on the part of the reader3 .

According to Nølke (2017a, pp. 158–159), the polemic interpretation imposes itself in utterances that form a contrast. Such a contrast is made explicit in example (5) :

(5) La production de co2 serait un phénomène naturel, et une conséquence du climat, pas la cause. (climat\_2008).

 The production of CO2 would be a natural phenomenon, and a consequence of climate, not the cause. (climat\_2008)

In this instance, the polemic interpretation is further reinforced by strong a priori expectation that CO2 emissions are the cause of CC.

In the absence of such an explicit contrast, non-gradable predicates can serve to create an implicit contrast, thereby blocking (or reducing the likelihood of) a descriptive interpretation (ibid.):

(6) Le nucleaire (sic) ne sauvera pas le climat. (climat\_2007)

Nuclear power will not save the climate. (climat\_2007)

The predicate creates a binary opposition between *save the climate* and *not save the climate*, which creates the basis for the polemic opposition of two points of view. In contrast, gradable predicates lend themselves to descriptive interpretations:

(7) L'idée de la géoingénierie, c'est-à-dire l'action à grande échelle pour modifier le climat, n'est d'ailleurs pas nouvelle. (climat\_2008)

 The idea of geoengineering, meaning large scale action to modify the climate, is not new, for that matter. (climat\_2008).

In example (7), the negation does not seem to refute an underlying pov, but to situate the predicate on a scale ranging from new to old. In other words, the use of negation in this case serves to create a description which would not be possible by using a positive predicate (i.e. 'not new' is slightly different from 'old'). In other terms, while the previous three examples are cases of contradiction, where either the positive or negative alternative must be true, example (7) presents a case of contrariety, where the two alternatives are mutually exclusive without being mutually exhaustive, leaving space for a number of other alternatives (see also Nølke 2017a, p. 161). However, it is important to note that extralinguistic factors, such as the debate around geoengineering, could intervene to impose a polemic interpretation of example (7). Determining whether a particular use of negation is polemic or descriptive is thus a question which, in many cases, depends not only on the linguistic features of the utterance, but also on the co(n)text in which it is embedded.

Additional factors which indicate a descriptive interpretation are predicational elements with no focus:

(8) Dire qu'il faut essayer de limiter les émissions de gaz à effet de serre de 40 à 70% d' ici le milieu du siècle pour ne pas dépasser les 2°C de réchauffement revient à fixer une tâche considérable. (réchauffement\_2014)

 To say that we need to limit GHG emissions between 40 and 70% from now until the middle of the century in order not to surpass 2°C warming, amounts to a considerable task. (réchauffement\_2014)

The scope of the negation is an infinitival clause ("not to surpass 2°C warming") in which there is no focus and thus no potential for contrast. In other words, the content of the infinitival clause is not asserted, which means that it cannot by itself form a pov, much less a contrast between several pov.

Other cases are less clear-cut:

	- Plus globalement, la question du lien entre activité cyclonique dans le monde et réchauffement du climat fait l'objet d'un vif débat dans la communauté scientifique internationale et il ne se dégage pas, pour l'heure, de consensus dans un sens ou dans l'autre. (climat\_2008)
	- Could this increase in cyclone activity in the North Atlantic be tied to to climate change?
	- More globally, the question of the link between cyclone activity and climate change is the object of fierce debate within the international scientific community, and there is at this moment no consensus in one direction or the other. (climat\_2008).

On one hand, the negation in example (9) reads as a simple reformulation of the preceding assertion regarding the 'fierce debate', thereby lending itself to a descriptive interpretation. On the other hand, the negation seems to contradict a commonly held view that there is an established link between warming and cyclone activity. In light of this contextual information, the negation could be interpreted as polemic. When faced with cases such as example (9) in this study, we systematically analysed the negations as polemic, because the fundamental polyphony of the negation is susceptible to being triggered by contextual factors. However, there are also cases where we were unable to identify such factors, which means that several occurrences of negation may have been categorised as descriptive, while authentic readers of the blogs would interpret them differently.

With this uncertainty in mind, the results of the analysis of the 120 examples are summarised in Table 2.4:


**Table 2.4.** Polemic and descriptive negations.

The numbers are remarkably stable across all the subcorpora and, more importantly, across the periods 2007–2008 (71.7% polemic negations) and 2013–2014 (66.7% polemic negations). While taking into account that the partially subjective nature of the analysis (see example (6) above) may influence these results to some extent, there appears to be very little variation in the distribution of descriptive and polemic negations over time. This means that we can rule this out as an influencing factor when looking at negation as a measure of controversy, at least in this instance. The question if whether a qualitative sample analysis of the reference corpus can produce a different result. Table 2.5 shows the distribution of a random selection of 100 negations from an open search of 'pas':

**Table 2.5.** Polemic and descriptive negations in the reference corpus.


Comparing tables 2.4 and 2.5, there appears to be no difference in the distribution of polemic and descriptive negation between utterances on climate change and those that concern other topics. This seems to fit well with the quantitative results in table 2.1, where climate change did not appear to be a particularly controversial topic on French language websites. The question is whether negation correlated with an undeniably controversial subject might give us a different distribution, i.e. a larger majority of polemic negations. Let us look more closely at a word that was clearly associated with negation in table 2.1, namely 'Trump'. A qualitative analysis of a random sample of 100 negations appearing +/– 5 words from 'Trump' gave the following distribution (table 2.6):

**Table 2.6.** Polemic and descriptive negations correlating with 'Trump'.


While these numbers indicate that the likelihood of polemic negation increases with the contentiousness of a topic, the uncertainty of many cases due to contextual factors prevents us from drawing any clear conclusions regarding this relation.

## **4. Concluding remarks**

In this study we set out to track the contentiousness of CC by studying negations in blog posts from two different time periods: 2007–2008 and 2013–2014. The underlying assumption was that negation is uniquely placed to perform a refutative function. Even without any other syntactic or lexical marker, negation can turn a proposition on its head, creating a polemic opposition between two points of view. Seeking to explore the extent to which the polemical nature of CC and its societal implications transpire through the use of negation, we formulated the following hypothesis: negations in relation to climate change are more frequent in the time period 2013–2014 than 2007–2008, because of the increasing politicisation of the issue in the intervening years.

Our material consisted of the NTAP French blog corpus, containing posts from 2,033 French blogs, in addition to the French Web 2017 corpus, serving as a reference corpus for our study. The latter was first used to explore two initial research questions serving as a basis for our study: 1. Does negation correlate more frequently with lexical items associated with controversial topics than with those that are considered less controversial? 2. If such a metric appears to be fruitful, does climate change prove to be a controversial issue relative to others? By correlating negation with different words associated with varying degrees of controversy, including 'tourism' and 'Trump', we found a consistent pattern of increasing likelihood of negation with an increasing degree of controversy. This would seem to indicate a correlation between the contentiousness of an issue and the frequency of negations being used when talking about it. However, the same test also revealed a low frequency of negation associated with climate change, indicating that this issue is relatively uncontroversial in French language texts online. Moving on to our study of the NTAP blog corpus, we searched for the word 'pas', in combination with 'climat', 'climatique' and 'réchauffement', which yielded 3,022 occurrences of 'pas' for the years 2007– 2008, and 7,768 occurrences for the years 2013–2014. As many of the blog posts in the NTAP corpus have little to do with the climate issue, the relative frequency of negation gave us little indication of the degree to which controversies surrounding CC were reflected in the use of this linguistic marker. However, we were able to isolate the climate issue in the corpus by looking solely at the frequency of negations co-occurring with 'climatique' relative to its total frequency during each time period. The results showed that there was 11.2 per cent likelihood of a negation co-occurring with 'climatique' in 2007–2008, and 13.7 per cent in 2013–2014. Perhaps more interestingly, the number for 2013, the year of the IPCC's 5th Assessment Report, was as high as 14.6 per cent, which would seem to indicate that the frequency of negation follows the topic's degree of contentiousness.

The qualitative analysis set out to correct any erroneous conclusions possibly arising from the quantitative analysis. As negation can have both a polemic and descriptive interpretation, it was important to map out any discrepancies in their distribution across each time period. The analysis of 120 random occurrences of negation in the material revealed a remarkably stable distribution of around 70 per cent polemic negations, thus removing the descriptive negation as a relevant variable in the comparison. Two control analyses produced similar results. First, an analysis of random occurrences of the negation 'pas' in the reference corpus gave 63 polemic negations out of 100, while the proportion of polemic negations associated with 'Trump' was 74 out of 100. The difference between these two samples could be explained by the many controversies surrounding Donald Trump, although they could also be the product of chance, coupled with the inherent uncertainties of many instances where the larger context comes into play.

In closing this study has opened up two main avenues for further research. First is the hypothesis that the proportion of descriptive negation increases as the contentiousness of an issue decreases. Our analyses may have revealed such a tendency but a study exploring more data in a combined quantitative and qualitative perspective could expose such a pattern more clearly. Secondly, we looked at negation related to the topic of CC without considering the large array of subtopics represented in the material, such as mitigation, adaptation, international negotiations, technological innovation, and climate science. A computer-generated topic modelling analysis (see Tvinnereim et al., 2017) could reveal interesting correlations in the use of negation on various subtopics in the material.

### **Endnotes**

1. The metalinguistic negation also opposes two pov. However the second pov does not refute the content of pov1, but the choice of a lexical item, e.g. 'Paul is not big, he's huge'.We will not discuss this use of negation in the present chapter.

2. The current study is not suited to test such a hypothesis, as the two subcorpora treat the same topic.

3. The latter of the two interpretations could also be analysed as a case of *interlocutive dialogism* (Bres and Nowakowska 2008).

### **References**


Bres, J. and Nowakowska, A. (2008). *J'exagère? …* Du dialogisme interlocutif. In : M. Birkelund, M.-B. Mosegaard Hansen and C. Norén, eds., *L'énonciation dans tous ses états. Mélanges offerts à Henning Nølke à l'occasion de ses soixante ans*, Bern: Peter Lang, pp. 1–27, Retrieved from https://hal.archives-ouvertes.fr/hal-00333034/document

Ducrot, O. (1984). *Le dire et le dit*. Paris: Minuit.


## **3. Negative Campaigning: Communicating Negative Meanings in French Presidential Debates Over Time**

*Malin Roitman and Bonnie Fonseca-Greber*

## **1. Introduction**

This chapter digs into new terrain in the well-studied field of French negation, exploring the discourse-pragmatic intersections of the loss and repurposing of *ne* (the original French negative 'not' out of Latin *ne* 'not') and its impact on *pas* (the now-basic French negative 'not,' originally and still, in other contexts, meaning 'step'. Here, we dig into the Roitman corpus of French presidential debates to uncover how negative meanings are conveyed in argumentative discourse, in particular in the televised French presidential debates from 1974, 2012, and 2017.

Given that *ne*-loss in real time is already convincingly documented in French in a variety of European and Canadian Frenches, by over a generation of Spoken French corpus linguists (Ashby 2001, Armstrong & Smith 2002, Hansen & Malderez 2004), presumably *pas* 'not' would carry more negative meaning than it would have half a century ago, now that it has become the basic negation of Spoken French.

Meanwhile though, *ne*—still alive and well in the scripted norm and therefore accessible to members of the speech community—seems to have been acquiring a new role in conversational discourse, reasserting its negative meaning to emphasise the speaker's negative view of the situation (Ashby 1976, Sankoff & Vincent 1980, Fonseca-Greber 2007, 2017, Poplack & St. Amand 2007, van Compernolle 2009, 2010, Donaldson 2017, French & Beaulieu 2020). Congenial conversation has been found to operate according to the social agreement principle (Yaeger-Dror 2002). So if the interlocutors share the same negative view of the matter, all is well, and social agreement is

#### **How to cite this book chapter:**

Roitman, M. and Fonseca-Greber, B. 2023. Negative Campaigning: Communicating Negative Meanings in French Presidential Debates Over Time. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 43–81. Stockholm: Stockholm University Press. DOI: https://doi.org/10.16993/bcd.c. License: CC BY 4.0. maintained. Conversation and arguing though are two different matters. Arguing—for example over public policy stands, as is the nature of presidential debating—may operate according to what we could call a social disagreement principle instead. Hence, Roitman's corpus of televised French presidential debates (Roitman 2015, 2017b) now provides fruitful ground for exploring how *ne* is used to communicate negative meanings in candidate interaction in French presidential debates over the years, as well as how it intersects with *pas*.

In this study then, we have two hypotheses

Hypothesis 1—*ne*:

Candidates will produce some *ne* tokens in the debates to reinforce their refutation of their opponent's position.

Hypothesis 2—*pas* 'not':

In these same utterances, where forcefully asserting negative meaning is essential for the discourse-pragmatic impact of the candidate's position, the candidate will produce *pas* 'not' tokens that are prosodically more salient (i.e., focal stress and vowel lengthening).

In other words, united under a single umbrella hypothesis: Will we see a correlation between the co-occurrence of the two unmarked variants in the unmarked, neutral negative utterances (i.e., *c'est pas* 'it isn't') and the co-occurrence of the two marked variants in the marked, refutational negative utterances (i.e., *ce n'est PAS* 'it is not')?

Anticipating the findings presented in our results section, and against the well-known backdrop of and ongoing language change, we do see a correlation between the presence/absence of *ne* and the prosodic qualities of *pas* 'not' in the construction we have chosen to examine—*c'est pas* vs. *ce n'est pas*—given its pragmatic importance in political debate:


The rest of this chapter is organised as follows. In section 2, we review what is already known about French negation relative to our goals for the present study. In section 3, we present the corpus and method we apply here. In section 4, we present our results. In section 5, we discuss our findings, advancing preliminary conclusions, situating them within the larger context of what is currently known about contemporary French language use, and avenues for future research. Finally, in section 6, we present a brief summary of our chapter.

## **2. Context and goals of the study**

2.1. The pragmatics of negation: negation as an argumentative strategy In the French enunciation and argumentation theories the negation has been explained as one type of polyphonic phenomena, i.e. an operator that enhances different "voices" in an apparent monological utterance (*énoncé*). These voices are footprints of the enunciation act (*énonciation*) defined as a unique process that produces an utterance, a historical event constituted by the appearance of a statement (Anscombre and Ducrot 1983; Ducrot 1984); this theory therefore rejects the idea of a unitary speaker of individual utterances. The French enunciation theories were inspired by speech act theories (Austin 1962 and Searle 1979) and philosophers of language within pragmatics (Grice 1975 and 1981) who explained meaning as the effective usage of language in different communicative situations. The originality in their approach is the disclosure of argumentation structure (traces of the enunciation act) within the denotative meaning of the linguistic units. Sentence negation – among other polyphonic markers – denotes thus a doubleness that can be exploited on a discursive level. Inspired by Ducrot (1984) and Nølke *et al*.'s (2004) and Nølke's (2017) adaptation and development of the polyphony theory model, the polyphonic structure of negation may be outlined as follows, applied on a translated example from our corpus:

The working time reduction has not been successful in other countries (Sarkozy debate 02/05/2007)

Point of view 1 'The working time reduction has been successful in other countries'

Point of view 2 'The working time reduction has not been successful in other countries'

Sentence negation *ne…pas* stratifies the utterance in two hierarchically organised points of view, one subjacent and implicit (1) and one explicit (2). These two layers, the activation of two points of views, are instructions1 in every negative sentence indicating that the default negation is a polemic negation. The descriptive negation is, on the other hand, considered a derivation of the polemic negation, which means the point of view 1 of a descriptive negation is barely activated. The derivation means the semantic instructions of the negative utterance and of its context do not indicate any opposing items that would favour a polemic reading of it. The polyphony of negation is, however, considered a continuum where different contextual aspects, non-linguistic and linguistic, govern the activation of the subjacent point of view. To fully understand Sarkozy's utterance above in this particular context, the point of view 1 needs to be activated. The interpretation of the inherent polyphonic structure of negation through the activation of the underlying point of view connects to Givón's (1979) discussion of the pragmatic presupposition and also to what in general linguistics is referred to as "common ground" (Stalnaker 1974, 1999) although linguistic polyphony illustrates the making use of linguistic units in order to exploit and create fictive voices used for argumentation. In this specific example, Sarkozy makes this negative statement in order to refute the idea that the working time reduction has been successful elsewhere; the opposite candidate Ségolène Royal had actually just promoted this point of view.

The polemic negation and emphatic negation, or emphatic function, will here refer to the same concept: the reinforcing the negative content and the activation of a pragmatic presupposition. From a polyphony theory point of view, a negation is an instruction to search for a possible subjacent point of view, a "voice", a pragmatic presupposition. We will stick to the two terms (polemic and emphatic) since we work with two methodological and theoretical frameworks in our analyses regarding whether the polemic-emphatic negation correlates with the return of a new *ne* and with a stressed *pas*. That is to say we are interested in whether the polemicemphatic negation (the *function* not the form) coincide with what must in contemporary spoken French be considered an emphatic marker *ne* (the form) from a Jespersen's cycle (1917) perspective. In other words:


### 2.2. French negation over time

Thanks to the depth of the written (and now spoken) record from Latin through to current 21st century French, and to the interest of Jespersen (1917) and his successors in historical linguistics and grammaticalisation studies in cyclical language change, French negation is one of most thoroughly studied negation systems of the world's languages. So, to briefly summarise this well-trod terrain, French negation has passed through a number of stages over the centuries. Emerging out of Latin, French negation was *ne*(*n*) 'not', placed before the verb. During the middle ages, a variety of emphatics were optional and placed after the verb to strengthen the pragmatic impact of the speaker's negative intent. Ultimately *pas* 'step' (e.g., I couldn't take a single step more) won out over other contenders (e.g., *mie* 'crumb' and *goutte* 'drop,' as in not being able to eat another crumb or drink another drop more), spreading from verbs of 'going' to all verbs. Not only did *pas* become the preferred post-verbal optional emphatic, but over time, it lost its emphatic quality and became an obligatory second half of French negation.2 Over more time, *pas* began to become perceived as such an integral part of French negation that the original negator *ne* began falling into disuse. Over even more time, *pas* came to be understood as the real negator, 'not,' while *ne*, the original 'not,' continued to wither away in naturalistically acquired first language French. This brings us to the current state of affairs where fewer than 10% of negative utterances in French conversation today continue to contain *ne* relying instead solely on *pas* to communicate negative intent.

### 2.3. Emphatic negative evaluation and the reinforcing role of *ne*

But *ne* is not gone completely—not in writing, not in prescripted/presentational speech, not even in everyday conversation. So, since there always seems to be something new to discover about French negation, despite the already vast literature on the Jespersen Cycle and the ensuing pragmatics of negation in French and other langauges (Jespersen 1917, Horn 1989, Schwenter 2006, Larrivée 2010, 2020, Mosegaard Hansen 2011, Mosegaard Hansen & Visconti 2014, Breitbarth 2020), what is the function of *ne* today and how does that interact with the current Spoken French negator *pas*?

A well-documented body of literature has appeared documenting the emergence of a new, related function of *ne* to convey emphatic negative evaluation (Ashby 1976, Sankoff & Vincent 1980, Fonseca-Greber 2007, 2017, Poplack & St. Amand 2007, van Compernolle 2009, 2010, Donaldson 2017) in a variety of Canadian and European Frenches, in a variety of synchronous contexts (face-to-face conversation, sociolinguistic interviews, and synchronous chat). This would suggest—despite the preponderance of *pas*-only negatives (and the likelihood that a learner's utterance with *ne* but without *pas* would cause communicative confusion)—that not all negative meaning has been bleached from *ne*, and that it is taking on new life as *ne*. The goal of this chapter then is to explore how *ne* is used, not in everyday conversation but in a new interactional format: the interactive segments of televised French presidential debates—argumentative by nature, and where candidates attempt to refute the position of their opponent in order to win election—and how it intersects with *pas*.

### 2.4. The 'Norm,' register, genre, and their reciprocal influence on *ne* use

Some scholars argue that France is currently characterised by diglossia (Lodge, 1993, Jakubowicz and Rigaut 1997, Fonseca-Greber 2000, 2011, 2018, Fonseca-Greber and Waugh 2003a, 2003b, Zribi-Hertz, 2011, 2013, 2019; Massot & Rowlett 2013, Palasis 2013, Barra-Jover 2013), or a unified speech community where two languages (or radically different forms of the same language) co-exist, each used according to the functional division of communicative labor between the two within the speech community.3 In France, this plays out as follows. The mythical French 'Norm' — or rules of 'good usage' for writing (or otherwise presenting4 ) the language — constitute Ferguson's so-called 'High' language, whereas the language acquired naturalistically at mother's knee—pro-drop, prefixally-inflected, Spoken French5 —would be the so-called 'Low' language, unfortunate labels, but representative of the linguistic prejudices of the speech community, as Ferguson clarifies in adopting the High/Low terminology (Ferguson 1959).

In contrast with the Arabic-speaking world where diglossia is widely acknowledged, in France, where national identity and national unity have traditionally been closely tied to a unifying—and unified—French language (von Wartburg 1946, Walter 1988, Lodge 1993), diglossia is an uncomfortable topic, even among the country's leading linguists (Blanche-Benveniste 2010 and associates), and recognition of a certain diglossia—as predicted by Ferguson's model—has been slow in coming,6 but see Zribi-Hertz (2019) for recent confirmation of French diglossia.

This is not to deny the role that register and genre play, but as Zribi-Hertz (2019) point out, this is a false debate—one, in fact, predicted by diglossia. In the face of the mounting evidence that we are dealing with two typologically distinct grammatical systems, used for two distinct communicative functions within the speech community (at least by those priviliged enough to have acquired access to both varieties), this is exactly what makes Roitman's corpus of French presidential debates so intriguing and such a rich corpus of data to study to deepen our understanding of *ne* use in contemporary French. On the one hand, presidential debates are a prototypically 'High' language communicative event (e.g., largely prepared/pre-scripted, public address), in contrast with a prototypically 'Low' language communicative event (e.g., everyday conversational interaction among family and friends). On the other hand, over the years, the debates increasingly contain unscripted, interactional exchanges, similar to conversational give-and-take. Yet in contrast with conversational interaction, where the social agreement principle (Yaeger-Dror 2002) tends to prevail in this phatic function of language (Jakobson 1990), in the interactional segments of the presidential debates, rather than exchanging pleasantries, the candidates argue with each other—at times vehemently—whether defending themselves or their proposed policies or refuting the opposing candidate and his/her proposed policies, as if the candidates were adhering to a social *dis*agreement principle instead.

It is in these segments where, over time, we see a re-emergence of *ne*-use, as if speakers/language-users in a media-age may be being influenced by the *ne* they continue to see and hear in the 'H/presentational' language around them, and—it not being part of their 'L/interpersonal' language—seek to interpret/imbue it with new, contextually plausible meaning, *à la* Andersen's (1973) abductive (or 'just-off') model of language change. In this case, the 'just-off' interpretation that some younger speakers might reach of *ne* is that of an emphatic of negative evaluation. This enlarges the communicative palette of 21st century French speakers, affording them the latitude to distinguish between their 'new' neutral negatives (*pas*, and their newer negative of emphatic rebuttal. We will see how this plays out in the Roitman corpus, in the Results section below.

But first, an overview of the corpus and how it allows us to address our research questions.

## **3. Corpus and method**

### 3.1. The Roitman Corpus of French presidential debates

The Roitman Corpus of French Presidential Debates is a diachronic corpus of televised French presidential debates presenting over 40 years of the language and culture of France. It contains 180 000 number of words and 17 hours of talk, and spans seven electoral cycles:

1974: **Valéry Giscard d'Estaing**<sup>7</sup> /François Mitterrand 1981: **François Mitterrand**/Valéry Giscard d'Estaing

1988: **François Mitterrand**/Jacques Chirac 1995: **Jacques Chirac**/Lionel Jospin 2007: **Nicolas Sarkozy**/Ségolène Royal 2012: **François Hollande**/Nicolas Sarkozy 2017: **Emmanuel Macron**/Marine Le Pen

As a collection of televised political debates, the corpus is primarily a speech sample of what the American Council on the Teaching of Foreign Languages (ACTFL) categorise as the *presentational* communicative mode (ACTFL 2012): the candidates present pre- or semi*scripted* position statements *unidirectionally* to the voters viewing from home. Over the years, however, as societal norms evolve, the debates begin acquiring a more interactive quality, as candidates argue, refute, and otherwise *spontaneously negotiate meaning with each other*, in, by contrast, the *interpersonal* communicate mode (ACTFL 2012), embedded within the overarching *presentational* format of presenting one's positions and qualifications to a national (in the internet age, now international) audience of viewers. Other than Armstrong & Smith's (2002) diachronic study of radio French, this hybrid *presentational-cum-interpersonal* mode has been an understudied communicative event, relative to the amount of work previously done on the purely interpersonal mode in French, i.e., private, face-to-face conversation, where intended/ unintended meanings and misunderstandings can be negotiated with one's interlocutor(s).

It is these more interactive segments of the presidential debates that are of most interest to us here, especially as the main construction we study, *ce n'est pas* 'it is not,' is also used in conversation, (1), to refute a previously held presupposition.

(1) Rebutting One's Own Presupposition (Fonseca-Greber 2007: 266) S13: ah! parce que c'était un coin de buissons…**ce n'est PAS** un coin d'herbe!

> 'Oh! because it was meant to be [drought-resistant] bushes…it isn't meant to be lawn!'

S1: ouais-ouais ouais-ouais… 'Yeah-yeah. Yeah-yeah…'

In the conversational exchange in (1), the speaker realises his own mistaken presumption and corrects it forcefully, using *ce n'est pas* 'It is not…' whereas, as we will see in the debates, the candidates usually use *ce n'est pas* to refute their opponent or their opponent's policy or proposals.

Forcefully refuting one's interlocutors is not a winning communicative strategy, if one wants to maintain friendly conversational interaction, however, especially if one were the hostess, as in (2), and took to refuting one's guests. Instead, S5, the gracious hostess, gently corrects her guest's presupposition (valid in her home canton, but not that of her hostess) by using the unmarked negation *c'est pas* 'it isn't' and social agreement is maintained in (2).

	- S4: non, mais chez nous, chez nous à G., quand tu le font au vin cuit.. tu mets du vin dedans
	- S5: ah, c'est un gateau au vin, non, mais **c'est pas** même chose un gateau au vin. Gateau au vin oui, mais le vin cuit c'est autre chose 'Oh, that's a wine pie, no, but it's not the same thing a wine pie. Wine pie, yes, but the cooked wine one is something different.'

When rebutting one's political opponent in televised debate, however, candidates may not seek social agreement or be attentive to the needs of their debate partner's 'face' as they argue policy points. Conversely, if a candidate needs to back-peddle and self-correct (rebut) his/her own erroneous presupposition on (inter)nationally televised political debate, the candidate may not want to draw needless attention to the fact and opt for the attenuated, unmarked *c'est pas* 'it isn't' negation instead, to save face.

### 3.2. Method

Through transcription of the debates, the presence versus the absence of *ne* is notified and quantified in the three debates we have selected diachronically from the corpus (i.e., the oldest and the two most recent). Other characteristics are also noted, such as the place in the sequence of the negation, the nature of the negative adverb, the nature of the subject, of the negative clause and of the predicate, etc. From these tags we have been able to categorise and list what seems to enhance the *ne*-dropping versus *ne*-retention.

The prosodic and phonetic quality of *pas* has been analysed with the assistance of *Voice Analyst*, and the data has been quantified and categorised. We have analysed the pitch and length of *pas* in relation to each candidate's average pitch from the actual negative sequence. The volume and the frequency of speech differ from one person to another and in particular between men and women and this has also been taken into account. Some few examples appear in interwoven speech and the prosody can therefore not be fully analysed.

We use the following abbreviations when presenting the examples:

Valérie Giscard d'Estaing = VGE (1974; M48) Nicolas Sarkozy = NS (2007, 2012; M52,57) François Hollande = FH (2012; M58) Marine Le Pen = MLP (2017; F49) Emmanuel Macron= EM (2017; M40)

Gender presents a methodological confound in the 2007 and the 2017 debates (with the women losing in both years) and therefore might better be explored elsewhere, to avoid skewing the 2007 and/or 2017 results.

Political party may present another methodological confound. Leftleaning, progressive politicians and their families may not automatically adopt the most innovative forms to avoid accusations of laxism and be perceived instead as 'upholding standards.' For example, see Ball (1997: 188–193) for a discussion of the 1989 French spelling reform and the opposition to it by Danielle Mitterrand, wife of Socialist then-president François Mitterrand. This could be a second methodological confound in the 2017 data. In contrast with these two language external methodological confounds, the next two are language internal.

The third methodological confound in the 2017 data is the most interesting, however, from the perspective of language change and language use. Given that numerous variationist sociolinguistic studies have repeatedly and convincingly shown that *ne*-loss is a case of real-time language change, with age being the decisive factor (younger speakers use it less) overriding gender and socio-educational standing (Ashby 2001), two overlapping changes—the tail-end of *ne*-loss and the leading edge of a pragmatically-conditioned emphatic negative verb, arising in phonetically favourable environment which had been on the leading edge of the first change, *ne*-loss with *c'est pas* 'it isn't' (see VGE 1974)—may account for the apparent up-tick in *ne* use in 2017. Methodologically, therefore, it is important to keep these two *ne* changes distinct from each other to avoid blurring the results by lumping *ne* use/deletion rates together, and prosodically neutral or prominent *pas* 'not' may be helpful in teasing this apart.

A fourth and final methodological confound in the 2017 data also relates to what constitutes a token (vs. a type) and how it is counted. Now, however, it is not old *ne* vs. emphatic *ne*, but rather the morphosyntactic difference between the free morpheme *ça* 'that' and its corresponding bound inflectional prefix *ce-/c*-. Given that types and tokens often display different distributional behaviour and that it is therefore typically recommended in corpus linguistics to count tokens separately from types (McCarthy et. al 2010), we probably do not want to skew our results by conflating (3):

```
(3) Tout ça n'est pas (EM)
   'All that is not'
```
with (4):

```
(4) ce n'est pas
    'it is not'
```
especially because the speaker could have uttered (5) in place of (3):

(5) Tout **ça**, **ce n'est pas** 'All that is not'

Here, we will consider (3), produced under prescriptive pressure of the nationally televised presidential debate, to be an inflectionless variant of (5), and therefore a variant of *ce n'est pas*.

## **4. Results**

4.1. Change 1—a new unmarked negation: loss of old ne '(old) not' + replacement by pas '(new) not'

Change 1—or the establishment of a new basic or neutral negation in French as a result of the ongoing loss of *ne* in real-time language change, documented repeatedly over the last 50 years in corpora of French conversation, sociolinguistic interviews, and broadcast radio-journalism—is also apparent in the Roitman corpus of televised French Presidential debates.

Table 3.1 presents the global diachronic decrease in *ne* use in the French presidential debates from the last quarter of the 20th century to the first quarter of the 21st century—specifically, from the first televised debate in 1974 (between Mitterrand and Giscard d'Estaing) to the decade of the 2010s (between Sarkozy and Hollande in 2012 and between Le Pen and Macron in 2017). While the uptick between 2012 and 2017 could be due to individual candidate differences and/or interactional differences between the debating pairs,8 it could also be due to an emergent reanalysis of a new form-meaning pairing for emphatic or refutative negation: *ne*-full negation + prosodically salient *pas* (*PAS*) (Section 4.2). Table 3.2 situates the televised presidential debate data relative to their respective sources of comparable chronological data.


**Table 3.1.** Diachronic decrease in *ne* use in the French presidential debates.

**Table 3.2.** This table situates the televised presidential debate data relative to their respective sources of comparable chronological data.


*(Continued)*


**Table 3.2.** *(Continued).*

Many of the examples of change 1 may be described as "chunks" i.e. frequent collocations where there is an apparent tendency to drop the *ne* according to our general observations above: *je veux pas* 'I don't want'*/je parle pas* 'I don't speak'*/je sais pas* 'I don't know'*/il y a pas* 'there isn't/there aren't':

(6) FH: **Je veux pas** citer les noms, vous les connaissez, ce sont vos proches. Donc, il y a eu des chèques du Trésor public qui ont été adressés aux plus grandes fortunes de notre pays.

 'I don't want to mention names, you know them, they're people you're close to. So, there were checks from the public treasury written to the richest people in our country.'

NS: Quels proches, monsieur Hollande ? (2012) 'Who do you have in mind, Mr. Hollande?'


Other factors that seem to correlate with the *ne* dropping in the debates – first person pronoun, negatives appearing early in the turn-taking, interactive conversational-like sequences – coincide with factors described in earlier studies although these, with their main focus on ne-dropping, present more precise data and a larger range of decisive factors.

Against the backdrop of on-going language change, *ne* loss in this corpus follows the collection of constraints well-documented in other corpora of journalistic and conversational language use. It still shows drop in Roitman corpora but not as strikingly and blurs Macron as an outlier and therefore change 2 is a more interesting explanation.

#### 4.2. Change 2—a new marked negation: emergence of new emphatic *ne* + prosodically salient *pas* (PAS)

Change 2—or the emergence of a new marked or emphatic negation in French as a result of the ongoing loss and repurposing of old *ne* into new, emphatic *ne*, through abduction via its ongoing presence in the 'High' ('presentational') language—is also apparent in the Roitman corpus of televised French Presidential debates, where it is used to refute the presuppositions of one's opponent in highly charged political debate.

Table 3.3 presents how there is a correlation and maybe even a cause – effect relation between the *ne*-retention and the presumably new emphatic *ne* used to refute the pragmatic presupposition of the opponent. The polemic function of negation in these debates has been analysed in Roitman (2015, 2017b) but is here related to the presence versus non-presence of new, emphatic *ne*. Whether the *ne*-retention is really evidence or not for the new *ne* remains although at this stage of the study a hypothesis. Since the *ne*-dropping is very low in general in the corpus as a whole and since there are many factors, which seemingly influence the retention and the dropping of *ne* (genre, style, lexicon. etc.) here and in earlier studies (see above), studying the negatives in this corpus as a whole is not sufficient to confirm this relation. Looking at table 3.2 it is easy to draw the conclusion that genre and style are the paramount decisive factors even though table 3.3 tempts us to interpret the *ne*-retention in favour of our hypothesis. This is also how we – with reserve – interpret our findings in this section of the study, before taking on the second step of the analyses. Still at this stage we need to accept that the various layers do not permit us to affirm anything.


**Table 3.3.** The *ne*-dropping and *ne*-retention in relation to non-emphatic versus emphatic function.

Table 3.4 below presents that there is also a correlation between the old *ne* becoming the new emphatic *ne* combined with the stressed high pitch *pas* in negatives used refute the pragmatic presupposition of the opponent. The high-pitch and elongated *pas* appears in the majority of the negatives in the debates that emphasise the negative content and might thus in general count as criteria for this kind function. The stressed *pas* is, *per se*, an emphatic element and therefore somewhat reinforces our hypothesis of the new emphatic *ne* and the emphatic function of negation. However, as has already been mentioned, this *ne*-retention may also be due to other factors such as style, presentative mode, etc. These factors make it difficult to draw any strong conclusion whether this is a "return" of the *ne* in order to reinforce the negative meaning. First, the two processes of *ne*-dropping and *ne*-retention are interwoven and second, what motivates the use of *ne* in the case where the negative sentence present a high pitch *pas* is uncertain although we might assume the correlation reveal a cause-effect relation.

**Table 3.4.** High and low pitch *pas* in *ne*-retention and *ne*-dropping negatives.


These tables thus show on the one hand the distribution of emphatic *ne* (polemic) vs. non- emphatic negation and its correlation with the *ne*-dropping and the *ne*-retention (table 3.3), and on the other the matching between the *ne*-retention and the high pitch *pas* (*PAS*).

The correlation between emphatic negations and the *ne*-retention in the debates is clear, although quite a few of the *ne*-dropping negatives are also emphatic-polemic (39% of the *ne*-dropping in 2012 are emphatic). What we cannot be sure of at this stage is the cause of the *ne*-retention. In other words, are these *ne* in emphatic-polemic negative sentences the new emphatic *ne*?

Since the starting point for our calculations is the concrete *ne*-dropping and *ne*-retention, the matching between the *ne*-retention, the emphatic-polemic negation and high pitch *pas* has been calculated by an equation resulting in a 70% matching between these categories. Does this mean the new *ne* negatives and the high pitch *pas* are predominant in the pragmatic reinforcement of negative meaning in these around 400 examples? At least we can see that the *ne*-dropping negatives correlate to a certain degree to the low-pitch *pas* and that these negatives are interpreted as non-emphatic (descriptive). Although these examples are quite few. Due to the many factors that might be involved in *ne*-retention mentioned above: style (presentative mode), degre of interactivity, type of pronoun, text type etc. we are unable to draw any sharp and general conclusions about the *ne*-retention in the debates. There are however tendencies that reinforce our hypothesis and that will be illustrated below.

### *4.2.1. Examples of change 2: new* ne *(*ne *retention with emphasis) + stressed* pas (PAS)

The following examples performed by NS, EM and FH expose what is presumably a new emphatic *ne* appearing with a high pitch *pas*. The pragmatic presupposition – the underlying point of view – is refuted, these sentences (9–11) are clear examples of emphatic-polemic negations. What makes us draw the conclusion that the *ne*-retention here is an example of new-*ne* is the nature of the negatives sentences where various factors normally, from what was mentioned earlier, enhance *ne*-dropping: first person pronoun, highly interactive dialogic sequences and common verbs that appear in chunks, lexicalised sentences, such as: *je parle pas* "I don't speak".

(9) NS: Juste un mot sur le rassemblement. Le rassemblement, c'est un très beau mot, une très belle idée, mais il faut y mettre des faits. Le rassemblement c'est quand on parle au peuple de France, à tous les Français. **Je ne suis pas** l'homme d'un parti, **je ne parle pas** à la gauche. Hier, je me suis adressé à tous les Français. (2012)

 'Just a note about the assembly. Assembly—it's a lovely word, a lovely idea, but we need to add a few facts. It's when we talk to the French people, about bringing all French men and women together. **I'm not** a party man. **I'm not speaking** to the Left. Yesterday, I spoke to *all* French.'

(10) EM : Non, moi j'ai pas envie d'essayer du tout. Et je crois que les Français non plus. Pas du tout envie d'essayer avec vous. 'No, I'm not interested in trying at all. And I think the French aren't either. Not at all interested in trying with you.' MLP : Pour faire en sorte que les Français… ne parlez pas à leur place. 'So that the French…don't go speaking for them.' EM : Non, **je ne parle pas** à leur place, mais **ils n'ont pas** envie. (2017) 'No, **I'm not speaking** for them, but **they don't want** to.'

(11) EM: Non, Madame Le Pen, **je ne vous laisserai pas** dire ça. Je les ai vus, les uns et les autres et qui ont fait cette guerre d'Algérie et qui aujourd'hui divisent notre pays. Moi, je veux passer à une autre étape justement. **Je ne veux pas** rester dans cette guerre des mémoires […] 'No, Madame Le Pen, **I won't let you** say that. I've seen them. One and another, the ones who fought in the Algerian war and who are now dividing our country. I want to move beyond that. **I don't want** to remain stuck in that war of memories…'

Negatives (9)–(11) may be contrasted with examples of non-emphatic-descriptive negation where – in a majority of the cases – the old *ne* (*ne*-dropping) appears with un-stressed *pas*. (There are however also non-emphatic-descriptive negations where the *ne* is retained.) In the following examples there are no indications of an underlying opposing point of view and the *ne*-dropping may probably be enhanced by the above-mentioned factor (lexicalised negatives, first person pronoun, interactive sequences).

(12) FH: **Je veux pas** citer les noms, vous les connaissez, ce sont vos proches. Donc, il y a eu des chèques du Trésor public qui ont été adressés aux plus grandes fortunes de notre pays.

 '**I don't want** to mention names, you know them, they're people you're close to. So, there were checks from the public treasury written to the richest people in our country.'

(13) NS: Quels proches, monsieur Hollande ? (2012) 'Who do you have in mind, Mr. Hollande?'

NS : **Je parle pas** des 75%, vous avez dit.. 'I'm not talking about the 75%, you said…' MLP: **On ne sait pas** trop ce qu'il y a dedans. Je suppose qu'il y aura la disparition, **je sais pas**, du CDI. '**Not much is known** about what's in it. I suppose things will be lost, **I don't know**, unlimited contracts'

### 4.3. A case study in refutation and presuppositions: c'est pas '(it) isn't' / ce n'est pas '(it) is not': preliminary comments

Before systematically looking at the results for the *ce* + *être* + negation examples we will problematise some examples illustrating the complexity of interpreting the *ne*-retention and the *ne*-dropping. In the following three examples (14–16) there are two consecutive negations, an obvious repetition on the same content and of which the first negation comes without *ne* and the second comes with *ne*. The *ne*-dropping of the first negation may be explained by one or several of the criteria described earlier: this negation may have a corrective function as it appears early in the phrase; the "*c'est pas*" has furthermore become a formulaic nearly lexicalised expression, etc. which may be what primarily induces the ne-dropping. The *ne*-retention in the second clause may either be analysed as a mere grammatical correction of the non-normative construction "*c'est pas*" that was just performed, or otherwise as a reinforced negation, where the negative content is emphasised through repetition in order to counter argue the underlying statement. Since we don't see "grammatical corrections" elsewhere in the corpus, besides these *ne*-dropping cases, we are inclined to believe this is a reinforced negation. According to our pitch-analysis, we notice that the first *pas* is unstressed while the second comes with high pitch and is elongated in all three cases (*PAS*), which also backup our interpretation of the *ne*–retention as being a new, emphatic, *ne*:


 MLP: **C'est pas** le sujet, **ça n'est pas** le sujet.9 Le sujet est toujours le même.

'**That's not** the topic. **That is not** the topic. The topic is still the same.'

(16) MLP: c'est la raison pour laquelle d'ailleurs vous voulez supprimer en réalité limiter l'indemnisation du chômage en expliquant que eh bien on leur fera deux offres. **On sait pas** où. **On ne sait pas** de quoi. Si ça se trouve à 200 kilomètres ou à 300 kilomètres…

 'That's the reason why you want to eliminate, well, limit, unemployment insurance besides, explaining all's well, they'll make'm two job offers. Who knows where. Who knows what. And if it's 200 or 300 kilometers away…'

### *4.3.1. Old* ne *versus new* ne

Choosing to closer study the *ce* + *être* + negation is due to the relative high frequency of this sequence in the debates and to its high rate of *ne*-dropping compared to other structures. It pairs with other formulaic sequences in the debates and as such it is susceptible for *ne*-drop. These conditions make the *ce* + *être* + negation cases more solid to study when it comes to *ne*-retention. The questions we asked are in a more elaborate version the following:

1. May the non-salience of the negative content be observed in the *ne*-dropping *c'est pas* phrases, besides a prosodic unmarked *pas*? In other words, is there a correlation between the non-emphatic, descriptive negation (non-salience of the negative content), i.e., *ne*-deletion and the prosodic unmarked *pas*?

2. May the salience of the negative content be observed in the *ne*-retention *ce n'est pas* phrases, besides a prosodic marked *pas*? In other words, is there a correlation between the emphatic, polemic negation (salience of the negative content), i.e., *ne*-retention and the prosodic marked *pas* (*PAS*)?

The results show differences between *c'est pas* and *ce n'est pas* that confirm to a great extent our hypothesis, although it will be important to develop the qualitative analyses.

The *ne*-dropping cases *c'est pas* are generally used in the contexts that do not emphasise the negative content, i.e. where nothing indicates a pragmatic presupposition of the contrary. This was first observed when we looked at all *ne*-dropping in general in the debates (see above). Using the Ducrot dichotomy descriptive versus polemic negation, we tend to qualify them as descriptive negations in many senses. There are of course many elements involved in the interpretation but the level of argumentative reinforcement is definitely low for the *ne*-dropping in the *ce* + *être* + negation cases.


**Table 3.5.** All *ce* +*être* + negation sentences.

The *ne*-retention *ce n'est pas* are generally used in the contexts that emphasise the negative content, i.e., where there are indications for a pragmatic presupposition of the contrary; this opposing view is at the same time refuted. Using the Ducrot dichotomy, we tend to qualify them as polemic negations in many senses. There are of course many elements involved in the interpretation of the *ne*-retention but the level of emphasising the negative content in these sequences seems to be quite an important parameter. There are in a majority of cases contrastive elements in the context – semantic instructions – that reinforce the emphatic negation and thus refutative function of these negatives.

## *4.3.2. The ce + être +* negation *sentences and pitch and quality of pas*

Regarding the stressed *pas* we have found the following, which is also exposed in table 3.6.


We will show how the *ne*-retention is a stronger marker for reinforced negation than the stressed *pas* but that there is also an important correlation and supposedly a cause-effect relation between the emphatic *ne* negatives and the stressed *pas* when it comes to the *ce + être + negation sentences*.


**Table 3.6.** The pitch and length of the *pas* for the two types of negatives.

### *4.3.3. Old* ne *non-emphatic negation + unstressed pas « c'est pas »*

First, we present some examples of the old *ne* + unstressed *pas* « *c'est pas* » for comparison. These negatives are mainly corrective and nonemphatic, that is they are not used for refuting the pragmatic presupposition but to dismiss a comment or an earlier statement that has no importance for the argumentation. They do, as a matter of fact, mostly appear in strongly interactive sequences which supports our hypotheses on different factors increasing the *ne*-dropping in the introductive chapter:


Still there are cases of *ne*-dropping negatives that are emphatic, and there are *ne*-retention negatives that are non-emphatic as it is exposed in table 3.6. Here, we follow the criteria characterising emphatic *ne* presented in Fonseca-Greber (2007), statistically confirmed by Donaldson (2017). Intersecting with this is Yaeger-Dror's Social Agreement Principle (2002). To integrate emphatic *ne* and the Social Agreement Principle, we propose a typology of face in Table 3.7. Fine-tuning our typology to account for how issues of language, gender, power and powerlessness play out may help account for the difference between (21) and (25), but a full investigation of this issue is beyond the scope of the present chapter. Here, suffice it to say, that Marine Le Pen, in (25), may have to assert herself more forcefully to be taken seriously than Emmanuel Macron in (21), while at the same time opening herself up to the gendered criticism of being 'outspoken' instead of 'demure.'

#### *4.3.4. New* ne *emphatic negation + stressed pas « ce n'est PAS »*

These negatives including the *ne* and an elongated *pas* are used to refute the pragmatic presupposition emphasising the negative content. It is often a question of an argument they want to "bring up in order to knock down". In the specific context of these negative sentences there are often contrastive elements such as *On a le droit de le dire* 'A person's allowed to say it' in (22), indicating and vitalising the underlying pragmatic presupposition – of the negative sentence "*ce n'est pas un insulte extraordinaire*" – that reinforces the refutation of the same. In (23), FH is explicitly ironic and echoes NS refusing to take responsibility for the shortcomings under his presidency. The underlying presupposition is associated to FH and the irony consists of the meaning of these negatives being opposite to what FH really intends to say. In (24), the syntactic emphasis on *eux* 'they' works as a contrastive element indicating and evoking the subjacent point of view: *il s'agit des Français et Françaises qui ont voté pour Marine Le Pen* 'the French men and women who voted for you,' a point of view which is at the same time refuted. The contrastive element in (25) is *particulièrement* 'particularly,' an element that reinforces the irony in Le Pen's utterance while referring to her previous *vous essayez de jouer avec moi à l'élève et au professeur* 'you're trying to play student-teacher with me.' This is one of several examples of them using the straw man device as a rhetorical strategy through negation and that has been studied in these debates earlier (Roitman 2017b). Using the strawman figure means here that the candidates exploit the polyphonic structure of negation to put words and expressions into their opponents' mouths, ideas that they have never expressed or that are distorted or strongly exaggerated versions of their opinions, in order to discredit them.

(22) NS : Dire que vos propositions ne sont pas bonnes **ce n'est PAS** une insulte extraordinaire. On a le droit de le dire.

 'Saying your suggestions aren't the best **is** *not* particularly insulting. Let's acknowledge it.'

(23) FH: On parlera de tout cela. On est sur le thème du pouvoir d'achat. Avec vous, c'est très simple, ce n'est jamais de votre faute. Vous avez toujours un bouc émissaire. Là, vous dites « **ce n'est PAS** moi, ce sont les régions, la formation, je n'y peux rien ». Sur l'Allemagne, " qu'est-ce que vous voulez, j'ai mis cinq ans avant de comprendre quel était le modèle allemand. Avant, j'avais le modèle anglo-saxon à l'esprit ". Ce n'est jamais de votre faute. Vous aviez dit 5% de chômage, c'est 10% de taux de chômage. **Ce n'est PAS** de votre faute

 'We'll talk about all that. For now, we're talking about buying power. With you, it's very easy, it's never your fault. You always have a scapegoat. You say, "**It's not** me, it's the regions, the training, I can't do anything about it." About Germany, it's "What do you expect? It took five years to figure out the German model. Before that, I had the Anglo-Saxon model in mind." It's never your fault. You said 5% unemployment, it's 10% unemployment. **It's not** your fault.'

(24) EM : Madame Le Pen, les Françaises et les Français qui ont voté pour vous, comme pour Monsieur Dupont Aignan, je les ai toujours respectés. Je n'ai jamais fait de leçon de morale, mais je les connais aussi. Dans ma région…

 'Madame Le Pen, the French men and women who voted for you, like Mr. Dupont Aignan, I've always respected them. I never gave them a lesson on morals, but I know them too. In my region…'

MLP : **Ce n'est PAS** eux que vous visiez quand vous disiez…

 '*They* were not the ones you were pointing the finger at when you said…'

(25) EM : Ҫa avait été créé … ne importe quoi … C'est un fond souverain. 'It had been created…whatever…it'a a sovereign fund.'

 MLP : M. Macron, ne jouez pas avec moi… Je vois que vous essayez de jouer avec moi à l'élève et au professeur. **Ce n'est PAS** particulièrement mon truc…

 'Mr. Macron, don't toy with me. …I see you're trying to play student-teacher with me. **That's not** exactly my cup of tea…'

### 4.4. The meaning and function of c'est pas and of ce n'est pas

The *c'est pas* negatives come generally as the numbers show with an unstressed *pas*. Other characteristics also follow these negatives. As noticed, most of these negations come early in the shift of turn-taking, these negations correct the counter candidate and do not appear in phrases with heavy argumentative impact. They have the character of "rowdy-negations" appearing in highly interactive sequences. Although some appear in monologues and carry some of the features of *ne*-retention sentences with an essential difference: they almost never reinforce the negative content of the sentence used for refuting the other candidate. Here is however an example of two of these exceptions, from the 2017 debate:

(26) MLP: La France que vous défendez, **c'est pas** la France. C'est une salle de marché dans lequel, encore une fois, c'est la guerre de tous contre tous, dans lequel les salariés devront se battre pour préserver leurs emplois, face aux travailleurs détachés ; dans lequel les entreprises entre elles, dans la même branche, devront se battre pour avoir les salaires les plus bas ou la durée de travail la plus longue pour essayer de conserver les marchés entre elles. **C'est absolument pas** la vision qui est la mienne. Moi, je crois encore une fois, à la solidarité.

 'The France you're defending **isn't** France. It's a marketplace where, once again, it's a free-for-all—salaried workers against free-lancers to protect their jobs, businesses in the same field against each other for the lowest salaries or longest hours to protect their profit margins. **That's absolutely not** my vision. Once again, I believe in solidarity.'

In this context, there are indications of pragmatic presuppositions and emphatic meanings even though there is a *ne*-drop. The hyperbolic phrases *salle de marché* 'marketplace,' *la guerre de tous contre tous* 'a free-for-all,' *se battre pour préserver leurs emplois* 'to protect their jobs' that the Le Pen associates with Macron, the opposite candidate, presuppose her having the exact opposite visions of *la France* 'France.' These (26) are examples of the straw man device, which means associating an exaggerated, distorted and even false statement to your opponent, an "argument" then may then argue against and refute. As we also just saw (examples 22–24), straw man arguments involved with negatives do effectively vitalise the opposing pragmatic presupposition of the negation, and thus emphatic negation.

The *ne*-retaining phrases *ce* + *être* + negation come, in the majority of cases, with a stressed *pas* (*PAS*). As noticed, these negations often appear in a longer argumentative sequence, or at least not at the beginning of the turn takings. These utterances (*ne*-retention, *ce* + *être* + negation + stressed *PAS*) appear in phrases with heavier argumentative impact refuting an idea of the opposite side. (ex. 22–25) reinforcing the negative content. This is to say that there is a clear correlation between the *ne*-retention – the reinforced function (emphatic negation) and the elongated *pas* (*PAS*). In the cases of emphatic negation where the *ne* is retained but the *pas* is not stressed there is often another item in the predication that is elongated. In the example below, it is the word "euro" that is focalised and stressed:

(27) EM : Mais on a besoin de l'Europe dans la mondialisation

'But we need Europe to be part of globalisation'

 MLP : Monsieur Macron, ça fait 25 ans que vous promettez l'Europe sociale, vous et vos amis socialistes. 25 ans !

 'Mr. Macron, you and your socialist friends have been promising social reforms for Europe for 25 years!'

 EM : Madame Le Pen, ça fait pas 25 ans que je suis dans la politique..

'Madame Le Pen, I haven't even been in politics 25 years…'

 MLP : La désindustrialisation massive ! Nous avons subi l'effondrement de nos emplois et les délocalisations massives. Et aujourd'hui, Monsieur Macron, l'épargne des Français

 'A massive shift away from an industrial economy! We've undergone the collapse of our job market and massive outsourcing. And now, Mr. Macron, the French people are faced with the loss of their savings accounts.'

EM : **Ce n'est pas** l'euro cela. Les Français le savent…

 'That's not because of the Euro…and the French people know that…'

This focalisation enhance however the emphasising of the negative content and the pragmatic function of negation: the opposite candidate is clearly refuted.

## **5. Discussion**

The results presented in this study have shed light on ever-evolving ways in which negative meanings are successfully communicated when refuting presuppositions in the argumentative discourse of televised French presidential debates over the decades. While individual speaker idiosyncracies remain:

• Macron rarely drops *ne* in this type of utterance and is the candidate who obeys the norm most closely

• François Hollande, Marine Le Pen, and Nicolas Sarkozy are, as a group, more informal than Emmanuel Macron, and they also drop *ne* in a wider range of utterances than Macron does, who drops *ne* mostly in semi-lexicalised forms like *c'est pas* 'it isn't'

The Roitman corpus of French presidential debates presents additional diachronic documentation of *ne*-loss between the last quarter of the 20th century to the second decade of the 21st century, in tandem with an emergent new use of *ne* for negative emphasis, in this case, refutation of a political opponent's presupposition.

As mentioned early in this study, it is difficult to prove anything on the debates as a whole regarding the return of the *ne* – the new *ne* – as a result of the emphasised-polemic negation, due to the general limited *ne*-drop in the actual political debates, a rather normative, elevated type of presentational discourse close to the written form. Whereas the *ne* is dropped in 80%–95% of the cases in conversational mode the *ne* drops in the two last debates only add up to 16%. Even though most negations are emphatic-polemic in these corpora, and the majority of the *pas* are stressed (*PAS*) there is no evidence to say the type *ne* (re) appears for pragmatic reasons to reinforce the negative content of the sentence, but may solely expose the elevated language style of this particular mediatised political event. Our general impressions will however be summarised here.

Regarding the criteria for *ne*-dropping, *ne* drops, to begin with, more with certain forms and contexts than with others: when the clitic pronoun is *ce* 'it,' *je* 'I,' and when it has scope over frequent verbs like *savoir* 'to know,' *vouloir* 'to want,' and *parler* 'to speak.' We have seen that *ne* drops in formulaic expressions (or chunks) such as *je sais pas* 'I don't know' and *je parle pas* 'I don't speak' and to a rather high extent in *c'est pas* 'it isn't' that we have studied more closely in this chapter. *Ne*-dropping also occurs more frequently when the negatives are involved in highly interactive sequences, when they have a corrective function. Furthermore *ne* is more often maintained when the negative content is emphasised, which is obtained through indications in the context such as semantic and syntactic contrastive element, irony, hyperbolic expressions, argumentative connectors, among others. This leads us to the idea that the emphasised function, the polemic negation seems to bring back the *ne*, which is a result that correlates with those of Fonseca-Greber (2007, 2017) among others (Ashby 1976, Sankoff and Vincent, 1980, van Compernolle, 2009, 2010, Donaldson 2017, French and Beaulieu, 2020). The *ne*-retention thus becomes—through its correlation with the stressed and elongated *pas* (*PAS*)—the pragmatically salient feature of negation emphasising the negative content through indications of the pragmatic presupposition. This reinforcing of the negative content works rhetorically in the political debates, refuting the visions and the political ideas of the adversary as it has been shown in Roitman (2015). Accompanied by the stressed *pas* (*PAS*) these negatives become thus a marker of argumentation discourse. Our results also coincide with the analyses on negations' refutative function carried out in Roitman (2017b). The differences found between the *ne*-dropping and the *ne*-retention examples led us early in the study to the pairing of the first with the descriptive negation and the second with the polemic negation. We have seen that the reinforced negation and its refutative function of polemic negation – the idea of a subjacent counterpart that is refuted – of these *ne*-retention examples are enhanced by other semantic and syntactic contrasting elements when closely examined in context.

Regarding the *ce* + *être* + negation that has been studied more in detail the tendency is that the *ne*-retention sentences *ce n'est pas* and the stressed, elongated *PAS* do influence and show a correlation with the emphasising of the negative content and the refutation of the other candidate's arguments, real or false (straw man arguments). In these sentences, *ne*-retention *per se* though is more decisive for this function than is the stressed *PAS*; in fact, sometimes other predicative items in the negative sentence are also stressed, as first observed by Ashby (1976). So, emphatic-polemic negation + stressed *PAS* seem to enhance *ne*-retention, the new emphatic *ne*. This may indicate what has been shown in earlier studies, i.e., that the evolution of French negation seems not merely be a result of the phonetic evolution suggested by Jespersen but may be reinforced through communicative pragmatic needs. The new *ne* would be a result of such a communicative pragmatic need.

In the *ne*-dropping examples of the *ce* + *être* + negation – *pas* sentences *c'est pas*, the *pas* is never stressed even though some of them are emphatic as we have seen in some examples. Our impression is that the sentences' speech rate is slower than normal in this specific discourse where *ne* is maintained although that has not been measured, and remains thus a hypothesis. Indeed, slowed speech rate is one of the correlates of emphatic *ne* (Ashby, 1976, Fonseca-Greber, 2007, Donaldson, 2017).

### 5.1. Conversational discourse versus argumentative discourse

Compared to what has been found in conversational discourse (Fonseca Greber 2007 and 2017) the *ne*-dropping is as it has been mentioned very reduced in the presidential debates. The differences between the retention and deletion of *ne* are enormous between the two very different types of discourses—one vernacular, conversational, agreeable and friendly, and the other formal, confrontational, disagreeable, and bellicose. Here then, we have broadened the scope of inquiry from conversational-interpersonal discourse to argumentative-presentational discourse. While similarities emerge between the two, differences also emerge, perhaps specifically with regards to interlocutor 'face' and the pragmatics of politeness. While the emphatic negations *ce* + *être* + negation *ce n'est pas* in the debates refute the presupposition, the point of view belonging to the other candidate and enhance thus a face threatening act, the emphatic negations *ce* + *être* + negation *ce n'est pas* in the Fonseca Greber conversational corpus are self-directed and do not constitute a face threatening act, other than towards the speaker himself:

(28) S13: Ah! parce que c'était un coin de buissons…**ce n'est PAS** un coin d'herbe! Oh! 'because it was meant to be [drought-resistant] bushes…it isn't meant to be lawn!'

S1: ouais-ouais ouais-ouais…

'Yeah-yeah. Yeah-yeah…'

Conversely the non-emphatic negations in the debates are often not directly towards the opposite candidate but are used as general corrections of erroneous ideas without sender, and non-emphatic negations in the daily conversation are directed towards the interlocutor, attenuating mistakes of the other participant:

(29) S4: non, mais chez nous, chez nous à G., quand tu le font au vin cuit.. tu mets du vin dedans

 'No, but for us, for us in G, when you make a cooked-wine pie, you put wine in it'

 S5: ah, c'est un gâteau au vin, non, mais **c'est pas** même chose un gâteau au vin. Gâteau au vin oui, mais le vin cuit c'est autre chose.

 'Oh, that's a wine pie, no, but it's not the same thing a wine pie. Wine pie, yes, but the cooked wine one is something different.'

Thus, a contrastive typology of self- vs. other-directed rebuttal/correction of presuppositions, based on face, do exist between conversational and argumentative discourse, as outlined in Table 3.7 below.



### **6. Concluding remarks**

The overall purpose of this study relies on its interest for the pragmatics of negation, how negation is used and for what purpose. We have tried to problematise the question of *ne*-retention and *ne*-dropping in French negatives by focusing more specifically on the conditions regulating the emphasising of negative content. In these political debates most of the sentence negations do have a refutative function; the negative content of the phrase is emphasised in the sense that the pragmatic underlying presupposition or point of view is refuted. We have been able to show that these emphatic negatives in one clause-type (*ce* + *être* + negation + *pas*) correlate with the retention of *ne* and to a certain extent also to the stressed *pas* (*PAS*). Thus, a certain need for reinforcement of the negative content do seem to enhance the return of *ne*. What we see is a pragmatic use of negation where an old form is used in order to create specific meanings, i.e. the emphatic negation. This has been shown in earlier studies on grammaticalisation in general and in particular studies on the evolution of French negation. Speaking of return and of a new *ne* must of course be problematised in more than one way. First, as mentioned earlier, this particular corpus with low rate of *ne*-dropping compared to conversational discourse types cannot disclose the pull and push factors behind the appearance of *ne* in the corpus as a whole, although supposedly on specific clause-types. Furthermore, there are apparently many cycles of negation going on simultaneously depending on all aspects of the communicative situation, as Mosegaard Hansen (2009, 2011, 2014 *et al*.) among others also have shown. However, analyzing specific negatives in a specific context and comparing them to earlier studies with other corpora still show the tendency to re-use older forms in new ways in order to satisfy pragmatic needs and communicate specific meanings.

## **Endnotes**

1. Within this framework, the concept of meaning (French: *signification*) "contains above all, according to our view, *instructions* given to those who will have to interpret an utterance of the sentence, inciting them to look in the communicative context for such or such type of information and to use it in such or such a way in order to reconstruct the meaning intended by the speaker " (Ducrot 1980a, 12 Own translation). Analyzing language units such as refutation-statements thus implies not the description of the meaning but the search for indications of the argumentative status (and therefore the argumentative function) of the utterance via the marks of the utterance process (French: *énonciation*) such as pronouns indicating interlocuteurs: connectors, negation, certains tense-forms, scalar words. These units, traces of the énonciation expose different and often opposing "voices" in the utterance, which has been described as polyphony. The presence of sentence negation in a statement indicates, for example, an instruction that the speaker must look for contradictory arguments in the context. This contains the core of argumentation theory in language (Anscombre and Ducrot 1983).

2. This '*ne* Verb *pas*' of Classical French remains in use in the written language today, because the French writing system was codified during this same time period.

3. Outside of France and writing earlier—both as predicted by Ferguson's (1959) model of diglossia—for earlier proponents of French diglossia, see also Lodge (1993), Fonseca-Greber (2000), and Fonseca-Greber & Waugh (2003).

4. For example, prepared speeches, telescripted radio and television reporting, and the like.

5. Ashby (1977), Miller (1991), Pierce (1992), Roberge (1986, 1990), Jakubowicz & Rigaut (1997), Fonseca-Greber (2000, 2009, 2018), Fonseca-Greber & Waugh, 2003a,b), Füß (2005), van Gelderen (2011), Zribi-Hertz (2011), Bahtchevanova & van Gelderen (2016).

6. And even slower in reaching (let alone gaining traction) among the language teaching establishment—whether French-as-a-foreign language overseas (Walz 1986, Joseph 1988, Waugh & Fonseca-Greber 2002, Durán & McCool 2003, Fonseca-Greber 2013, Grangier & O'Connor DiVito 2018) or even French-as-a-second-language within the French-speaking world itself (Giroud & Surcouf 2016).

7. The name of the winning candidate appears in bold.

8. The 2017 data includes a lot of 'noise': *gender* (the only M-F, not M-M, debate); *age* (a 'young' candidate who may have acquired or speechaccommodate the grammar of his elders' generation and/or who may be

on the front lines of a new change through abductive reanalysis (Anderson 1973)); *political party* (the first time a far-right Le Pen-family political party has reached the presidential run-off). Despite this, taken as a whole, the 2010-decade shows a distinct drop from forty years earlier.

9. An anonymous reviewer inquires if the shift from *ce* to *ça* plays a role. Although this is beyond the scope of this article, it is worth noting that *ça*  itself is the so-called stressed pronoun, and therefore, emphatic.

10. An anonymous reviewer inquires if the cleft structure plays a role. Although this is beyond the scope of this article, it is worth noting that clefting serves to focalise the noun, and is therefore a form of emphasis itself.

## **References**


*French Language Studies*, Retrieved from https://hal.archives-ouvertes.fr /hal-01940633

Benveniste, É. (1966). *Problèmes de linguistique générale*. Paris : Gallimard.


Amsterdam: John Benjamins Publishing Company, https://doi.org/10.1075 /pbns.283


## **4. Metadiscursive Negation, Evidential Points of View and** *Ethos* **in Argentine Political Discourse**

*María Marta García Negroni*

## **1. Introduction**

As is widely known, Ducrot (1984) has defined three types of negation: descriptive, polemic and metalinguistic. The descriptive type is regarded as a delocutionary derivative of the polemic type and is said to represent a certain state of affairs which happens to be negative. By contrast, the polemic negation always implies that there exist two antagonistic viewpoints that, within the framework of the theory of polyphony, must be attributed to different discursive beings: enunciators E1 and E2. The former is held responsible for the underlying positive point of view while the latter comprises of an objection to the former and represents the point of view to which locutor L adheres. According to Ducrot, polemic negation functions as an assertive representation of a given situation, and it always has a diminishing or decreasing effect (*i.e.*, its interpretation always implies *less than*) and assumes the presuppositions of the underlying positive utterance.

As opposed to polemic negation, Ducrot depicts metalinguistic negation as a type of negation that contradicts the semantic elements that are comprised within a given utterance that is aimed at being rejected. In this sense, this negation type does not oppose the points of view of two enunciators; rather it has in its scope a different locutor who uttered its positive counterpart. According to the French linguist, it is exactly within the framework of this refusal to an antagonistic locutor that this negation type retains a series of distinctive features. Firstly, this type can bring about 'a majoring or augmentative effect [of the argumentative force] instead of its normal diminishing or weakening effect' (1984, p. 217) [translation]. In Ducrot's view, 'one can say "Peter isn't intelligent, he's brilliant", only as a response to another locutor who

#### **How to cite this book chapter:**

García Negroni, M.M. 2023. Metadiscursive Negation, Evidential Points of View and *Ethos* in Argentine Political Discourse. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 83–111. Stockholm: Stockholm University Press. DOI: https://doi.org/10.16993/bcd.d. License: CC BY 4.0.

has effectively qualified Peter as intelligent' (1984, p. 217) [translation]. Secondly, this negation type may – though it is not necessarily a compulsory feature – reject the presuppositions assumed to be the case in the prior positive utterance. Thus, if in (1) the negative utterance only affects the explicit semantic content (*i.e.*, the presupposition that *Peter used to smoke* remains unaltered), (2) shows a different scenario as is seen in the rectifying utterance following it. In this case, the negative utterance precisely contradicts the presupposition.


To sum up, according to Ducrot, the distinctive feature of this negation type is the fact that it requires a prior enunciation coming from another locutor whose words locutor L considers improper either because they presented an erroneous point of view or because in their utterance, they expressed a presupposition or a certain degree that cannot be accepted, or even

'because there is something that, due to the mode of expression, may be considered inadequate (such as an element of speech independent of the content, a word which might be considered as 'out of place', too colloquial or too distant, a pronunciation or grammar mistake): all these aspects of speech can be rejected by means of a syntactically negative statement' (Ducrot, 2001, p. 30). [translation]

However, as I discussed in previous papers (García Negroni, 2009, 2017), a reply in a dialogue is not a necessary condition for metalinguistic negation to appear. In fact, this is only the case in its metalinguistic use, properly speaking, in which the negative utterance objects to the use of a given word or a group of words uttered by the interlocutor in light of a syntactic or morphological or social rule that is put to light in the utterance that follows – the one which introduces the rectification. However, the scenario will not necessarily be the same when it comes to what I refer to as a metadiscursive use of metalinguistic negation or, to be brief, *metadiscursive negation*. As I will argue, this novel type of negation characteristically rejects a *quotative discourse frame* –a discourse frame related to others' discourse which is dialogically evoked in the very same negative enunciation to favour a particular subjective positioning or stance. In other words, metadiscursive negation always conveys – I will posit – an *evidential quotative meaning*.

Most often approached from referential or cognitive perspectives, evidentiality is usually understood as the semantic domain marking the existence of the source of information in the utterance and specifying what type of source–whether direct or indirect–it involves (Aikhenvald, 2004). The source is said to be direct when the knowledge at stake has been acquired by means of a perception arising from one of the speaker's senses, and is said to be indirect when such knowledge derives from an inference or from a quotation of somebody else's discourse (Anderson, 1986; Willet, 1988).

Depending on whether it is direct or indirect, some researchers that hold a 'broad' conception of evidentiality posit that the source implies different degrees of reliability, which in turn impact on the speaker's epistemic attitude towards the message conveyed (Chafe, 1986; RAE, 2009). This relation of inclusion in which one of the terms is understood within the scope of the other is not, however, the only position regarding the relationships that can be established between evidentiality and modality. Indeed, as indicated in the bibliography on the subject (Dendale & Tasmowksi, 2001), other authors claim that the relationship is one of disjunction or one of overlap. In the case of disjunction, evidentiality and epistemic modality are conceived as independent categories (De Haan, 1999; Aikhenvald, 2004; Cornillie, 2007, among others), given that –as stated– a constant biunivocal correlation cannot always be established between the type of evidentiality and the degree of epistemic commitment. In the case of overlap, although it is admitted that evidentiality and modality constitute independent categories, the existence of a convergence between the two is insisted upon when it comes to inferential evidentiality (Van der Auwera & Plungian, 1998).

Languages vary in terms of the manifestation of evidentiality. Some, such as Tuyuca or Quechua, have different morphemes that specify whether the speaker was a direct witness to the events they are narrating, whether they inferred them or whether they heard about them from a third person. Others, like Spanish, do not compulsorily grammaticalise evidential meanings, but instead have resources that allow the deployment, in certain specific contexts, of 'evidential strategies' (Aikhenvald, 2004).

As previously stated, in this paper, I will focus on the analysis of the evidential meaning of metadiscursive negation in Spanish that, like certain syntactic structures, certain adverbial constructions, some discourse markers, certain uses of verb tenses, etc., can convey this type of meaning1 . However, with a view to accounting for such meaning, I will drift from many of the assumptions upon which most studies on evidentiality rest. In fact, on the research paths paved by the theories of polyphony (Ducrot, 1984, 2001), dialogism (Bakhtin, 1981, 1982) and argumentative semantics (Carel, 2011; Carel & Ducrot, 2005; Ducrot, 2004), the *dialogic approach to argumentation and polyphony* (Caldiz, 2019; García Negroni, 2017, 2018, 2019, 2021; García Negroni & Hall, 2020, 2022; García Negroni & Libenson, 2020a, 2020b, 2020c, 2021; Zucchi, 2020), within which this study is framed, advocates a non-truth-value and non-referential characterisation of meaning (*i.e.*, there is no meaning component that can actually be considered purely objective). Furthermore, this perspective drifts from the principle of the uniqueness of an intentional subject in discourse (*i.e.*, the subjective points of view posed in a given utterance cannot necessarily be attributed to the same discursive being) while also focusing on the functioning of signs in the language system and in discourse.

I will delve into what I have described as quotative evidential points of view (García Negroni, 2018, 2019, 2021) and into how such viewpoints are displayed in instances of metadiscursive negation. All the cases analysed here belong to a corpus collected from a series of speeches delivered by different Argentine political figures: Juan Domingo Perón (three times President of Argentina: 1946–1952, 1952–1955 and 1973– 1974), Cristina Fernández de Kirchner (twice President of Argentina: 2007–2011 and 2011–2015) and Julio Cobos (Vice President of Argentina from 2007 to 2011, during Cristina Kirchner's first administration). I will seek to attain a two-fold aim. First, I will bring to light the fact that, in these cases, negative enunciation is presented as emerging from and responding to a discourse frame related to other voices –which are dialogically evoked by the enunciation itself. I will then seek to show that in dialogic response to such discourse frame, a given subjective responsive positioning arises along with a given *ethos*  of the *locuteur* (Ducrot, 1984; Amossy, 1999). Associated with different scenes (Maingueneau, 1999, 2002) typical of political enunciation and moulded in light of the different addressees at stake, the variety of *ethos* identified in the analysis will be as follows: the *ethos* of confrontation and discredit; the pedagogical one; the defensive *ethos* of someone who cares for its own public image; the one showing a condition of symmetry; and the cautious, fearful one facing an extreme situation.

The present work will be organised as follows. In section § **2.**, I will discuss the theoretical and methodological foundations of the dialogic approach to argumentation and polyphony (§ **2.1**.). Within this framework, I propose an approximation to the concept of evidential meaning. More specifically, I explore the dynamics of evidential points of view and how these contribute to sense-making in discourse and the dialogic architecture of (inter)subjectivity (§ **2.2**.). In this light, I characterise the *quotative discourse frame* that metadiscursive negation urges the interpreter to identify and retrieve as its cause (§ **2.3**.). In the following sections, I will focus on the quotative evidential points of view materialised in metadiscursive negation, and the identification of the different *ethos* arising from political enunciation (§ **3.**). Finally, I will draw some concluding remarks on this concern (§ **4.**).

## **2. The dialogic approach to argumentation and polyphony**

## 2.1. Theoretical and methodological foundations

Deeply rooted in Ducrot's theory of polyphony (1984) and the theory of argumentative semantics (Anscombre & Ducrot, 1983; Anscombre, 1995; Ducrot, 2004; Carel & Ducrot, 2005), the *dialogic approach to argumentation and polyphony* (hereafter referred to as DAAP), adheres to the principle that the sense of a given utterance must be understood as the description that the very same utterance makes of the enunciation from which it emerged (Ducrot, 1984)2 . In this sense, DAAP recognises the existence of four types of semantic instruction to account for the meaning of an utterance, namely: illocutionary, argumentative, polyphonic and causal (Ducrot, 1984). These last ones, concerning speech characterisation according to 'cause', explain the difference in meaning between a declarative utterance such as *Peter is very intelligent* and an exclamative one as in *Peter is so intelligent!* According to Ducrot, in declarative utterances, enunciation appears 'as if it were the result of a choice, that is to say, as the result of the decision to provide certain information about an object in particular' (1984, p. 186) [translation]. In the exclamation, however, enunciation is presented as triggered by 'the representation of such object: it is Peter's intelligence itself that seems to have forced the enunciation *Peter is so intelligent!*' (1984, p. 186) [translation]. Following this line of thought, DAAP expands on the notion of causal instructions and characterises them from a dialogical perspective. DAAP proposes that every utterance displays as part of its meaning an image of the dialogic 'cause' that motivated its occurrence in discourse (García Negroni, 2019, 2021; García Negroni & Libenson, 2020a, 2021). It is precisely to that image – which I will call a *discourse frame* – that enunciation dialogically and argumentatively responds with a particular stance or subjective positioning (Bakhtin, 1981, 1982). In other words, the subjective stances3 reflected and imprinted in the utterance are to be seen as dialogic responses to a particular discourse frame that each enunciation brings forth.

In accordance with the principles of argumentative semantics (Carel, 2011; Carel & Ducrot, 2005), DAAP's perspective affirms that the discourse frame, *i.e.*, the image that the utterance provides of its 'cause', should be described in terms of argumentative chains. It should be remembered here that, according to Carel & Ducrot (2005), the argumentative sequences that display the meaning of an expression or an utterance are composed by two segments articulated through either a normative or a transgressive relation. In the first case, the segments of the chain are connected by means of the prototypical conclusive connector *therefore* (abbr. *THF*); in the second case, the two segments are articulated by means of the prototypical concessive connector *however* (abbr. *HW*)4 .

Another essential methodological distinction that DAAP takes from Ducrot (1984, 2001) is the distinction that refers to the existence of two different discursive subjects which should not, by any means, be mistaken for the speaking subject: locutor L and locutor l. Locutor L is the discursive character that, within the sense of the utterance itself, is held responsible for the enunciation at stake. In turn, locutor l is the being to whom all first-person indexicals are assigned and about whom something is said in the utterance. Nonetheless, DAAP departs from Ducrot's characterisation of enunciative polyphony in three main aspects: it does not conceive the locutor L as a theatrical *metteur en scène* who–deliberately and consciously–puts a range of *enunciators* on stage5 . Furthermore, DAAP does not insist on maintaining that the enunciator is the source of the semantic content presented in any given utterance, or that L would necessarily take different stances when dealing with diverse enunciators. With a view to avoiding any psychological bias derived from such characterisations and aiming to account for all the semantic perspectives rooted in discourse, DAAP embraces the concept of *point of view* (hereafter PoV) to account for the semantic values imprinted in the utterance. To add to this, it also advocates that all dialogic features (Bakhtin, 1981, 1982) of sense-making be added to the polyphonic and argumentative depiction of the concept of sense referred to above. In other words, when it comes to sense, all the features related to the engagement of any utterance within the discourse chain, along with the subsequent subjective positioning of response to as well as of anticipation of other discourses, should be considered. In light of these claims, DAAP delves into the different subjective stances presented within argumentative sequences in discourse. Such sequences always occur as dialogic responses regardless of the speaker's rhetorical or strategic intentions (García Negroni, 2021; García Negroni & Hall, 2022; García Negroni & Libenson, 2020a, 2021).

In view of this, by articulating principles rooted in the theories of polyphony and argumentative semantics with those of dialogism, DAAP aims to account for all polyphonic and argumentative features of sense from a dialogic perspective and formalise all dialogic relations within a polyphonic and argumentative framework. It is precisely from this viewpoint that DAAP advocates that the characterisation of the PoVs, staging the dialogic relations that a given utterance establishes with the previous and subsequent utterances within the argumentative sequence should be included in the semantic description. Among these PoVs, the central focus of the following section will be *evidential PoVs*.

### 2.2. DAAP and evidential PoV

From a DAAP perspective, a PoV encoding an evidential meaning involves a set of dialogic-argumentative instructions (García Negroni, 2019, 2021). Such instructions call for the identification of the 'cause' of the enunciation within an *evidential discourse frame*; a frame that has motivated the enunciation in which such PoV emerges. In other words, whether materialised in certain syntactic structures, driven by certain verb tenses or by certain discourse markers, evidential PoVs systematically display an image of the enunciation in which they are expressed as caused by a specific discourse frame (hereafter referred to as DF) which the interpreter must identify and retrieve to access the sense of the utterance at stake. Shown but not uttered, such DF is constituted by argumentative sequences which are related to perceptions or acts of saying that are normatively (*i.e.*, in *THF*) or transgressively (*i.e.*, in *HW*) articulated with different types of epistemic statements about l with respect to them. In other words, the dialogic argumentative instructions embedded in evidential PoVs display the enunciation as one motivated by argumentative sequences referring to perceptions verifying it;

[I have been a perceptive witness to X *THF* I (λ) can assert X]

as arising from a conjectural DF;

[I can see/ It is said/ it is known Y *THF* I (λ) can infer X]

or as a result of a quotative DF in which a series of different relations are established between λ and the quoted or evoked discourses

[They say/They said X *HW* I (λ) cannot confirm X] [They say/They said X *THF* I (λ) think that X is possible] [You said/They said X *HW* I (λ) have realised that it is not X] [They say/They may say X *HW* I (λ) don't think X is true]

among other possible instances6 . In all these cases, in light of the different types of DF giving rise to the enunciation, a given subjective, responsive positioning is conveyed and must be interpreted as the indicator of the dialogic response to such DF. Therefore, if the evidential PoV is direct, the subjective stance that emerges in response will involve a strong commitment to the words uttered. Instead, if it is indirect, enunciation will eventually indicate precaution, detachment, concession, reproach or even refusal according to the evoked DF to which the enunciation replies.

**Figure 4.1.** Direct and indirect evidential PoVs, evoked DF and subjective positioning of response.

Otherwise stated, the analysis of evidential meaning proposed by DAAP does not adhere either to the category of the speaker as the source of sense or as the idea that information is encoded in discourse. Nor does it consider the speaker the source or the empirical origin of the information supposedly conveyed in discourse, as is usually stated in studies conducted about evidentiality (refer to, among other authors, Willet, 1988; Anderson, 1986; Aikhenvald, 2004; Bermúdez, 2006, 2016; Cornillie, 2007; Rodríguez Ramalle, 2008, 2014). Within the DAAP framework, what is central is to account for the way in which the evidential PoVs displayed in an utterance contribute to evince the dialogic, argumentative and polyphonic 'causes' that the utterance offers of its own enunciation. In other words, the DFs – for which evidential PoVs urge identification – account for the reason why the enunciation, in which such PoVs are presented, entails a series of dialogic and argumentative features.

In light of the notions discussed above, I will now focus my analysis on instances of metadiscursive negation. As I will try to prove, in these cases, the negative enunciation urges the interpreter to identify and retrieve as its 'cause' a quotative DF.

### 2.3. Metadiscursive negation and quotative DF

Unlike the metalinguistic use, properly speaking, of negation (Ducrot, 1984), the metadiscursive type does not reject the interlocutor's previous utterance but a DF that must be retrieved as the argumentative representation of the 'cause' of that negative enunciation. Therefore, it is precisely this DF, which is related to external voices to be recovered from the 'discursive memory' (Courtine, 1981) and which are considered inappropriate or incorrect, that the enunciation responds to with a vigorous rejecting positioning. Thus, for instance, in (3), the refusal at stake (cf. *In fact, there are no new labels to depict our doctrine or our ideology. It is not by chanting 'I'd give my life for Perón' that we'll honour our country, our homeland*) is not a reply to a previous utterance of a real flesh-and-bone interlocutor. It involves the representation of a dialogically evoked discourse (Bakhtin, 1981, 1982) embedded in the very same negative discrediting enunciation. In this example, it is the discourse of the revolutionary Peronist youth movement that is being discredited:

3. No hay nuevos rótulos que califiquen nuestra doctrina ni a nuestra ideología. Somos lo que las veinte verdades peronistas dicen. No es gritando la vida por Perón que se hace Patria, sino manteniendo el credo por el cual luchamos.

> (Fragmento del discurso de Juan D. Perón, 21/6/1973. Disponible en http://www.historiadelperonismo.com /pensamientos.php).

 In fact, there are no new labels to depict our doctrine or our ideology. We are what the twenty Peronist principles state. It is not by chanting 'I'd give my life for Perón' that we'll honour our country, our homeland but by abiding by the doctrine we are struggling for.

> (Excerpt taken from Juan D. Perón's speech delivered on 21/6/1973. Available at http://www.historiadelperonismo .com/pensamientos.php).

In short, as the materialisation of a quotative evidential PoV, metadiscursive negation is endowed with semantic instructions which call for the retrieval of a given DF – a DF related to others' discourse which is dialogically evoked in the very same negative enunciation – which *is shown* as the main cause that has given way to the refusal (L's particular subjective positioning or stance) expressed in the enunciation. In the following schema, I represent the DF between brackets, the subjective positioning in italics and the link between the DF triggering the enunciation and the subjective stance of response by means of a dialogic connector, HENCE:

[They say X *HW* I (l)don't think X is true] HENCE *L's refutative (and rectifying) enunciation*

So then, in the case of (3), the evidential PoV embedded in the negative utterance prompts the interpreter to identify the following DFs as a constitutive element of sense in the utterance at stake:

[They say X (*Peronism is the historical label* HW *it can change into Socialism*) HW I (l) don't think X is true] [They say X (*chanting 'I'd give my life for Perón'* THF *honouring our country, our homeland*) HW I (l) don't think X is true]

It should be noted that, against Perón's will, the ultimate aim of the revolutionary Peronist youth movement was to transform Peronism into socialism and, what is more, die for it. That is why in (3), variable X (the discourse evoked in the DF) is expressed by means of the argumentative sequences *<Peronism is the historical label* HW *it can change into*  *Socialism>* and *<chanting 'I'd give my life for Perón'* THF *honouring our country, our homeland>*<sup>7</sup> .

Thus, it is this type of DF that accounts for L's subjective stance of rejection and rectification.

[as what they say is wrong/ illegitimate/inappropriate to me (l)], *I (L) strongly reject it and, in turn, I (L) suggest the argumentative sequence which I consider suitable for the situation at stake*.

To put it briefly, as metadiscursive negation materialises quotative evidential PoVs, the negative enunciation always urges the interpreter to retrieve the DF – shown by the enunciation– which is linked to voices considered erroneous or inappropriate and which must necessarily be interdiscursively recovered. In view of such DF, a certain subjective stance of response arises, along with the advent of a given *ethos* in the enunciative scene (Maingueneau, 1999, 2002). In the following section I will exemplify and analyse the different types of *ethos* that emerge from the enunciation of metadiscursive negations extracted from speeches delivered by contemporary Argentine politicians.

## **3. Metadiscursive negations, political discourse and** *ethos*

3.1. Metadiscursive negation and ethos of confrontation and discredit Shown but not uttered, the source of the evoked discourse in the DF may turn out to be 'more or less' distant, or 'more or less' identifiable in the discursive memory. In (3), paraphrase (3a) makes it clear who is to be held responsible for the origin of the discredited discourse by means of an evidential marker–in this case a reduced adverbial clause–of the type *as (the + an) Z + verb of saying* (Anscombre, 2011).

3. a. In fact, there are no new labels to depict our doctrine or our ideology, as the revolutionary Peronist youth movement claims/ as Montoneros think. We are what the twenty Peronist principles state. It is not by chanting 'I'd give my life for Perón' that we'll honour our country, our homeland, as they insist, but by abiding by the doctrine we are struggling for.

This is the scenario that gives rise to the emergence of the indirect, negative addressee in political discourse8 , along with the rise of a given L's *ethos* (Ducrot, 1984; Amossy, 1999; Maingueneau, 1999, 2002). L, holding the subjective position of a political leader, appears to be endowed with the suitable qualities to discredit and reject a controversial act of saying. To put it simply, retrieving the sense of an excerpt like (3) involves identifying in it – by means of the metadiscursive negative utterances – the quotative evidential PoV evoking a DF related to external voices strongly rejected by L.

In scenes of strong confrontation, in which L discredits and undermines the indirect antagonistic addressee's voice (García Negroni, 2016), the enunciations involving instances of metadiscursive negation do not simply show the emergence of an opposed discourse within L's own discourse. In light of their occurrence, generally followed (less frequently preceded) by a rectifying utterance, metadiscursive negative utterances will also introduce the PoV to which L will adhere. In (3), the rectifying utterances following the converse sequences of those rejected9 evoke the following argumentative sequences: *<Peronism* THF *twenty Peronist principles>, <abiding by the Peronist doctrine* THF *honouring our country, our homeland>*, which, in L's view, correspond to the depiction of the situation at stake. Thus, (3)'s overall structure can be outlined as follows:

[They (the revolutionary Peronist youth movement) say X (Peronism is the historical label HW it can change into Socialism / chanting 'I'd give my life for Perón' THF honouring our country, our homeland) *HW* I (l) don't think X is true]

HENCE

*L's refutative enunciation* (Peronism is the historical label THF NEG. change it into Socialism / chant-

ing 'I'd give my life for Perón' HW NEG. honouring our country, our homeland)

*L's rectification* (Peronism THF twenty Peronist principles / abiding by the Peronist doctrine THF honouring our country, our homeland)

In this sense, in (3) a scene of controversy emerges and in it the indirect antagonistic addressee's discourse is discredited and refuted by a contentious L, whose words are presented as the only legitimate utterances. Yet other instances might appear. In fact, the quotative evidential PoV – embedded in instances of metadiscursive negation – may give rise to other scenes, and eventually other *ethos*.

#### 3.2. Metadiscursive negation and pedagogical ethos

Though also related to previous discourses, the DF that the negative enunciation replies to may also evoke other's voices. Therefore, let us consider (4):

4. Apelo una vez más a mis hermanas de género, nosotras ciudadanas de dos mundos, como digo yo, siempre en el mundo de lo privado, para el cual fuimos educadas, la familia, la protección, los hijos, y en el mundo de lo público, al cual hemos decidido acceder para trabajar, para representar o para dirigir, pero siempre con un pie en un mundo y con un pie en el otro, nos da una visión, no solamente de lo grande que es lo público sino de lo pequeño, del detalle que es la familia, lo privado. Por eso vemos cosas que ellos no ven, por eso podemos distinguir y percibir no porque seamos mejores sino porque tenemos ese mundo dual que debemos compartir y articular.

> (Fragmento del discurso de Cristina Kirchner, 14/8/07. Disponible en: http://www.impulsobaires.com.ar)

I appeal to my fellow women. As I always say, we are citizens of two worlds. We were brought up to live in the private world of family matters, protection and kids. But we were also educated to deal with the public sphere. And we have decided to face it to have a job, to represent others or even to run organisations. Yet we are always split into two worlds, which gives us the chance to appreciate how large the public universe is and how small and delicate family and private life is. That's why we can see details men can't, that's why we can distinguish and perceive veiled aspects of life, not because we are better but because we have this two-world view that we must share and put together.

> (This is an excerpt taken from Cristina Kirchner's speech delivered on 14/8/07. Available at: http://www.impulsobaires.com.ar)

In light of the relationship with the figure of the positive addressee (in (4) the female collective: *my fellow women, we are citizens of two worlds*), the type of *ethos* that emerges from this excerpt can be depicted as a didactic or pedagogical *ethos* and the instances of metadiscursive negation embedded in it are not extrinsic to such emergence. In fact, along with other discursive procedures which are typical of expository/explanatory discourses (definitions, examples, causative relations, to name a few), (4)'s metadiscursive negation gives rise to a teaching scene (Maizels, 2010) in which an asymmetric relationship is established between L–in the position of an expert entitled to refer to a given state of things– and its addressees, discursively constructed as disciples who must be provided with certain knowledge. As can be stated, this type of scene gives rise to an intelligible representation of 'the real', thus the instances of metadiscursive negation do not aim to reject a political adversary's saying, as is the case in (3). The DF, which must be retrieved as the 'cause' of the refutative enunciation and to which the enunciation dialogically replies, –due to the position of power granted by knowledge– now alludes to indefinite voices which, as part of the interdiscursive arena, are held responsible for certain social representations that must be discarded.

[They (indefinite voices/ certain feminist groups) say X (*women can distinguish and perceive veiled aspects of life* THF *they are better than men*) HW I (l) don't think X is true]

Again, paraphrase (4a) containing an evidential marker of the type *as (the + an) Z + verb of saying* enables the interpreter to retrieve the collective voice to which the origin of the rejected discourse is assigned10.

4. a. We can see details men can't, but not because we are better, as certain collectives of women claim.

As opposed to l's own discursive positioning, these representations are rejected and corrected by L by means of the following rectifying utterance, which introduces the 'true/real' sense that L–from a position of power derived from knowledge–reveals to the addressees involved:

[As I (l) consider that what is sometimes said /certain feminist groups state is incorrect],

*I (L) discard it and, in turn, I (L) suggest* the argumentative sequences corresponding to the situation at stake.

In (4), the new sense is expressed by means of the following sequence: *women have a two-world view* THF *they can distinguish and perceive veiled aspects of life*. Thus, (4)'s overall structure can be outlined as follows:

[They (indefinite voices/certain feminist groups) say X (*broader understanding of the state of things* THF *being better than men*) HW I (l) don't think X is true] HENCE *L's refutative enunciation* (*broader understanding of the state of things* HW neg. *being better than men*) *L's rectification* (*women's participation in two worlds* THF *broader understanding of the state of things*)

### 3.3. Metadiscursive negation and self-defensive ethos

However, if metadiscursive negation conveys a quotative evidential PoV, the other different voices evoked by the DF to which the enunciation replies may be either previous or prefigured. As Bakhtin states (1981, 1982), any utterance is a link in the discourse chain. As such, it involves not only a response to previous utterances but also, in an anticipatory manner, a reply to those utterances that might eventually follow it. This can be seen in (5):

5. ¡Qué nos pasa! Y acá quiero hablarnos –*y no es una cuestión de clase, por favor, soy peronista*– a nosotros mismos, a esta clase media tan volátil, a esta clase media como yo, universitaria, a la clase media que muchas veces no entiende y cree que, separándose de los laburantes, de los morochos, le va a ir mejor. Le pasó a todos, es como el sino de las grandes frustraciones en la historia argentina.

> (Fragmento del discurso de Cristina Kirchner, 14/10/2010. Disponible en http://www.lacampora.org/wp-content /uploads/2011/08/CFK-Discurso-acto-de-la-Juventud -Peronista-en-el-Luna-Park-14-09-20101.pdf)

What's going on to us! I'd like to speak to you Argentines –*and, please, it is not a question of class; I'm Peronist*– I'd like to address our own collective, such a volatile middle class I belong to. We belong to a social class that has been able to reach higher education and we sometimes think that life will become rosier for us if we drift apart from the darkskinned working classes. This has happened to us all. This has involved one of the greatest frustrations in Argentine history.

(This is an excerpt taken from Cristina Kirchner's speech delivered on14/10/2010. Available at http://www.lacampora.org/wp -content/uploads/2011/08/CFK-Discurso-acto-de-la-Juventud -Peronista-en-el-Luna-Park-14-09-20101.pdf)

At times introduced by a negative expression of the type *this doesn't mean/doesn't imply* (Sp. *no es que + polemic subjunctive*, RAE, 2009)11, or at times introduced by negative forms presented in explanatory clauses as seen in (5) (*please, it is not a question of class*), the quotative evidential PoV encoded in this type of negation leads the interpreter to identify a DF related to voices of likelihood as the dialogic 'cause' of the enunciation:

[They may say X (*CFK speaking to the middle class* THF *CFK being classist*) HW I (l) don't think X is true]

This teaching scene, in which the refutative enunciation appears to have been caused by a DF related to possible objections to the speech at stake, paves the way for a self-defensive, suspicious *ethos* of someone who is concerned about its own public image. While on the alert for any possible misinterpretations, we can see the rise of the discursive image of a person seeking to control sense, thus cancelling any possible criticism or negative judgement. Let us focus on the adverbial locution *por favor* (Eng., *please*), which can function as a rejection marker that either discards previous discourse as far-fetched or inadmissible (Santos Río, 2003) or, as in the case of (5), disregards what the interlocutor or somebody else might possibly say. Once rejected by means of the converse argumentation (Carel & Ducrot, 2005), the argumentative sequence is rectified on the grounds of a different argumentation. Thus, *<CFK speaking to the middle class* THF *CFK being classist>* is corrected by *<CFK being Peronist* THF *CFK Neg. being classist>*

> [As I (l) know that what somebody might say is wrong], *I (L) fully reject it and, in turn, I (L)suggest* the argumentative sequence that corresponds to the situation at stake.

Thus, (5) should be outlined as follows:

[They may say X (*CFK speaking to the middle class* THF *CFK being classist*) HW I (l) don't think X is true]

HENCE *L's refutative enunciation* (*CFK speaking to the middle class* HW *CFK* neg*. being classist*) *and L's rectification* (*CFK being Peronist* THF *CFK* NEG. *being classist*)

### 3.4. Metadiscursive negation and symmetric ethos

In certain cases, the possible critical utterances being cancelled in an anticipatory manner might be attributed to the adversary or negative addressee:

[They'll say X *HW* I (l) don't think X is true]

As can be seen in the example below, the self-defensive *ethos* merges with the confronting *ethos*:

6. Compañeros y compañeras: quisiera poder contarles de la mejor manera, lo que siento en este momento. Verlos a ustedes me hace recordar parte de la historia de mi propia vida y también la de nuestro país. Déjenme decirles que siento una sana envidia por todos ustedes. *Pero no por lo que seguramente alguno mañana va a decir "claro, querría tener 20 años, por eso se hace la nena". No, qué va*. ¿Saben por qué les tengo envidia? Porque cuando yo fui joven como ustedes, cuando junto a miles y millones de argentinos apostábamos a un país diferente, no tuvimos la suerte que tienen ustedes hoy de vivir en un país con todas las libertades.

(Fragmento del discurso de la Cristina Kirchner, 14/12/2010. http://www.casarosada.gob.ar/informacion /archivo/22619-blank-77976707)

 My dearest fellow citizens, I'd like to describe my feeling to you as clearly as possible. When I see you, I can remember my own personal life as well as the history of our own country. Let me tell you that I can feel healthy envious of you now. *But not because of what some will probably say tomorrow: "I understand, she'd like to be a twenty-year-old, that's why, she pretends to be younger than she really is." By no means*. Do you know why I feel envious? Because when I was young like you, we used to struggle for a different country –we were thousands, millions of us in those days– but unfortunately, we were not as lucky as you are now. Today you can enjoy a country that ensures freedoms of all kinds.

> (Excerpt taken from Cristina Kirchner's speech delivered on 14/12/2010. Available at http://www.casarosada .gob.ar/informacion/archivo/22619-blank-77976707)

Several features in the excerpt above contribute to the emergence in (6) of a conversational scene of convergence in which L addresses its supporters and followers (positive addressees) as if L were a friend of theirs speaking about its most private, intimate feelings; in this case, memories of political militancy in a youth movement. In this intimate scene, we can also see an explanatory clause (*but not because of what some will probably say tomorrow*) containing a quotative evidential PoV embedded in the negation. This example shows that this PoV leads us to interpret the refutative enunciation containing it as one resulting from a DF related to possible voices assigned to the adversary.

However, the negative enunciation in (6) does not just behave as an anticipatory reply which gives rise to a figure confronting an enemy whose voice is being discredited in a mocking tone. By stating some will probably say tomorrow, L alludes to a 'media corporation', which has been one of the favourite targets in former president Kirchner's speeches, particularly in the wake of the conflict with the farming sector in 2008. In light of this scenario, L refers to some knowledge (i.e., the troubled relationship between the former president and the press) that is shared with its positive addressees. This common ground gives way to a strong bond of support between L and its positive addressees, who bring that event to their minds as part of the 'shared memory' in Kirchnerist discourse. In other words, the segment *but not because of what some will probably say tomorrow* acts as a memory of previous discourses, while the sense of those utterances is being updated in the present act of enunciation. This gives rise to a bond of mutual, intersubjective understanding between L and its positive addressees. Therefore, since the addressees are constructed as members of the same collective who share certain knowledge about what is being alluded to12, the symbolic roles L assigns to its addressees and to itself reinforce a scene of a relaxed, friendly talk of mutual understanding. This raises an inherent symmetric ethos.

> [as I (l) know –and we all know– that what some (members of the media corporation) will say tomorrow is a lie],

> *I deny (L) it and, in turn, I suggest (L) the argumentative sequence corresponding to the situation at stake*.

Negation in (6) can be outlined as follows:

[Some (the media corporation) will say X (neg. *being young* THF *being envious of their young age*) *HW* I (l) don't think X is true] HENCE *L's refutative enunciation* (*CFK* neg. *being young* HW neg. *CFK being envious of their young age*)

*and L's rectification* (*CFK's young age without freedom* THF *CFK 's positive envy of young age with freedoms of all kinds*)

### 3.5. Metadiscursive negation and cautious, fearful ethos

A last example will be analysed to illustrate the relationship that can be established between metadiscursive negation and the category of *ethos*. In this case, we can see negative utterances embedded with evidential PoVs urging the interpreter to identify DF related to critical voices or instructions of what should be done or said. Such is the case of the following excerpt taken from former vice-president Cobos in the so-called 'non-positive vote' event, which takes place within the framework of the heated confrontation between the national government and the farming industry over the export tariffs imposed on agricultural goods. It should be noted that this excerpt comes from the final speech of the parliamentary debate on the 125 Bill in which, as President of the Senate Chamber, J. Cobos had to cast a vote to resolve the deadlock. However, his position opposed that of the national government to which he belonged.

7. Yo sé que me cabe una responsabilidad histórica en esto, hay quienes desde lo político dicen que tengo que acompañar por la institucionalidad, por el riesgo que esto implica. Mi corazón dice otra cosa y no creo que esto sea el motivo para poner en riesgo el país, la gobernabilidad, la paz social. Quiero seguir siendo el vicepresidente de todos los argentinos, el compañero de fórmula hasta el 2011 con la actual presidenta de los argentinos. Vuelvo a decir que es uno de los momentos más difíciles de mi vida. *No persigo ningún interés*. Estoy diciendo, o expresando, tratando de expresarlo, que mis convicciones, mis sentimientos empujan la decisión. Muy difícil, seguramente. Yo creo que la presidenta de los argentinos nos va a entender, me va a entender. Porque no creo que sirva una ley que no es la solución a este conflicto. La historia me juzgará, no sé cómo. Pero espero que esto se entienda. Soy un hombre de familia como todos ustedes, con una responsabilidad en este caso. *No puedo acompañar, y esto no significa que estoy traicionando a nadie*. Estoy actuando conforme a mis convicciones. Así que yo le pido a la presidenta de los argentinos que tiene la oportunidad de enviar un nuevo proyecto que contemple todo lo que se ha dicho, todos los aportes que se han brindado, gente de afuera o aquí mismo. Que la historia me juzgue. Pido perdón si me equivoco. *Mi voto, mi voto no es positivo*, mi voto es en contra.

(Fragmento final del discurso de J. Cobos, 17/7/2008. Disponible en https://www.youtube.com/watch?v=PDHWP9\_XWhE)

I know I am responsible for a historic decision. On the one hand, some people state I should vote in favour to ensure institutionality, as otherwise might be risky. On the other hand, my heart says something different and I don't think this might eventually put institutions, governance or social peace at risk. I wish to continue holding this position as Argentina's vice-president and Argentine President's running mate until 2011. Once again, this is one of the most difficult moments I have been through in my life. *I don't advocate any vested interests*. What I mean to say is that my decision is driven by my convictions and feelings. This is very difficult, surely. I think that Argentina's president will understand us all, she will understand me. As I don't think that if this law is passed, it will imply a solution to the conflict. History will judge me. I don't know which judgements will be made. I hope this can be understood. I am a family man, like all of you. The only difference is that I bear institutional responsibility over this issue. *I can't vote in favour and this doesn't mean that I'm a betrayer*. I stick to my convictions. Thus, I ask Argentina's president to take this golden opportunity to submit a new bill considering all the aspects dealt with and all the contributions given in this parliamentary debate, whether by parliamentary members or by people from other social sectors. History will judge me. I apologise if I am mistaken. *My vote, my vote is not positive*. My vote is against the law.

> (Excerpt taken from J. Cobos' speech delivered on 17/7/2008. Available at https://www.youtube .com/watch?v=PDHWP9\_XWhE)

Showing signs of nervousness (the speaker moves the microphone constantly, removes signs of sweating, his breathing is shallow) and at a very slow pace (it takes the speaker six minutes to utter his words), his speech puts on stage the figure of an ordinary man (*I am a family*  *man, like all of you*) in a difficult situation (*the only difference is that I bear institutional responsibility over this issue*): either casting a vote in light of his feelings or being accused of betrayal due to his vote because it differs from the ideas supported by the government he is a member of. This excerpt contains several instances of polemic negation (*I don't think this might eventually put institutions, governance or social peace at risk, I don't think that if this law is passed, it will imply a solution to the conflict*) giving rise to dynamic PoVs following the same argumentative orientation (Anscombre, 1990; Roitman, 2009). Such orientation ensures argumentative coherence in the text and is made explicit by some instances of metadiscursive negation such as *I don't advocate any vested interests, I can't vote in favour and this doesn't mean that I'm a betrayer, my vote is not positive*, which clearly reject DFs related to previous or prefigured sayings regarding any vice-president's duties and responsibilities. Therefore, such sayings would condemn his dissenting vote. In view of such voices, attributed to the government of which l is a member as the vice-president (cf. *some people state I should vote in favour to ensure institutionality; I wish to continue holding this position as Argentina's vice-president*):

[You (government l is a member of) say X (*against the law* THF *advocating vested interests*) *HW* I (l) don't think X is true] [You (government <sup>l</sup> is a member of) say X (*negative vote* THF *betrayal*) *HW* I (l) don't think X is true] [You (government <sup>l</sup> is a member of) say X (*being a member of the government* THF *voting in favour*) *HW* I (l) don't think X is true]

the enunciation replies in disagreement. Nonetheless, it is not until it comes to an end that a quasi-confessional scene showing a cautious, fearful *ethos* of someone who is concerned and tentative about its words and just acts in line with its heart, its convictions and feelings that the rectification presented by its own PoV emerges: *my vote is against the bill*.

> [as following my convictions (l) what is said (by the government) is false],

> *I (L) reject it and, in turn, I (L) suggest the argumentative sequence corresponding to the situation at stake*.

The instances of metadiscursive negation should be outlined as follows:

[You say X (*against the law* THF *advocating vested interests*) *HW* I (l) don't think X is true] HENCE *L's refutative enunciation* (*against the law* HW neg*. advocating vested interests*)

[You say X (*negative vote* THF *betrayal*) *HW* I (l) don't think X is true] HENCE *L's refutative enunciation* (*negative vote* HW neg*. betrayal*) [You say X (*being a member of the government* THF *vote must* be positive) *HW* I (l) don't think X is true] HENCE *L's refutative enunciation* (*being a member of the government* HW neg*. positive vote*) *L's rectifying enunciation* (*in the light of my own convictions* THF *negative vote*)

### **4. Concluding remarks**

As has been seen above, DAAP's conception of sense opposes the view that states that linguistic meaning is constituted by informative or cognitive aspects. In view of this conception, DAAP rejects the hypothesis that the study of language implies evaluating propositions in terms of truth values or analysing information content in light of a speaker's underlying intention. In fact, framed within the ground-breaking theories of polyphony, argumentation and dialogism, DAAP drifts from any type of perspective considering that the semantic value of an utterance results from an intentional subject's willingness. Thus, this approach does not focus on the mental activity of a real, intentional subject who would eventually let us know how they acquired the information content embedded in the utterance they assert. Rather, the main interest of this approach lies in the depiction any utterance makes of its own enunciation. In light of these fundamentals, this study has sought to contribute to a dialogic, polyphonic and argumentative depiction of the evidential meaning of metadiscursive negation.

In the analysis of this particular issue, it is held that meaning lies in the identification and retrieval of a DF which, shown by the enunciation, gives rise to a given argumentative depiction of other – previous or prefigured – voices. The DF that gives rise to this enunciation, and in which the evidential PoV is expressed through a metadiscursive negation, is constituted, depending on each case by argumentative chains like the following:

[They say/ They will say/They may say/ You / Some say X HW I (λ) don't think X is true]

Given that λ (the individual L was and is beyond the enunciation) does not agree on X, L (the discourse being who is held responsible for the enunciation) rejects and cancels such saying, which is dialogically evoked in the negative enunciation and then introduces its own rectifying PoV:

[as what they say/ they will say/ they may say/ you say/ some say is unreasonable/illegitimate/inappropriate to me (λ)],

I (L) reject it/ discard it/ cancel it/ deny it and, in turn, I (L) suggest the argumentative sequence corresponding to the situation at stake.

Regarded as the depiction any utterance makes of its own enunciation, the sense of an utterance displays a figure of L that emerges from the dialogic relations established by the utterance as a link in the discourse chain. Thus, L is not in fact an entity that is external, extrinsic to language functioning (*i.e.*, the speaker); rather, it is an internal, language-intrinsic figure that responds in a dialogical way to a DF upon which enunciation is founded. In view of what has hopefully been revealed in this study, it can be held that it is this dialogic stance that gives rise to the emergence of different *ethos* in political discourse in the specific case of metadiscursive negation.

## **Endnotes**

1. Cf. in this regard, among many others, the works by Reyes (1990, 1994), Bermúdez (2006, 2016), Cornillie (2007), Escandell Vidal (2010), Rodríguez Ramalle (2008, 2014), Cornillie & Gras Manzano (2015), González Ruiz, R. *et al*. (2016), García Negroni (2018, 2021), García Negroni y Libenson (2020a, 2020b y 2020c), Maldonado y de la Mora (2020).

2. In French *théorie de l'énonciation*, there is not only that which is said, the utterance (Fr. *le dit*), but also the fact of saying it, the *enunciation* (Fr. *le dire*) that reflects itself in the structure of the utterance. In terms of Ducrot (1984), the enunciation is 'the event, the fact constituted by the appearance of an utterance'. Therefore, to describe the meaning of an utterance is no other thing than to describe its enunciation.

3. From DAAP's view, stance does not refer to the epistemological positioning of the speaking subject but to the positioning which is manifested, within enunciation, as a dialogic-argumentative response towards a discourse frame that unchains it (García Negroni, 2021; García Negroni & Libenson, 2020a, 2021) The present dialogic notion of stance is philosophically anchored in the bakhtinian perspective of discourse, according to which subjectivity is set up as an act of self's response towards otherness.

4. From the very beginning, argumentative semantics (cf. Anscombre & Ducrot, 1983; Anscombre, 1995; Carel & Ducrot, 2005; Carel, 2011; García Negroni, 2017, 2018, among others) has claimed that meaning should be described in terms of argumentative chains and not in terms of reference to reality or to previous cognitive categories. Initially conceived as a sequence 'Argument-Conclusion' connected by means of the prototypical conclusive connector *therefore*, the notion of argumentative chain has been redefined in terms of semantic interdependence between the two segments of the chain (Carel & Ducrot, 2005; Carel, 2011). This semantic interdependence can be expressed not only in terms of a conclusive or normative argumentation (i.e., A *therefore* B), but also in terms of a transgressive relation (i.e., A *however Neg*. B) by means of the prototypical concessive connector *however*.

5. In Ducrot's theory of polyphony (1984), the enunciator (Fr. *énonciateur*) is the discursive character to whom the origin of semantic content of the utterance is attributed.

6. In order to facilitate the interpretation of the DF evoked by the evidential PoVs referred to in the examples, I have resorted to a variable *X* as a simplifying form of the argumentative sequence embedded in its sense in each case.

7. Example (3) involves an excerpt taken from a famous speech delivered by J. Domingo Perón as soon as he returned to Argentina after being in exile for eighteen years. The day before his return, millions of people went to Ezeiza International Airport to offer him a warm welcome. Yet, the event ended up in a very serious armed clash between Peronist union activists and the revolutionary Peronist youth movement (*Montoneros*). Slogans such as *Perón, Evita, glory to the socialist people!* and *Here is the youth movement. Perón again, or rather dead* to which the youth movement adhered are precisely the ones dialogically evoked by the DF which gives rise to (3).

8. According to Verón (1987), as opposed to other discourse types, political discourse intrinsically implies the simultaneous construction of a positive addressee and a negative one. Verón adds that in contemporary democratic systems there is a third type of addressee that is to be persuaded. This figure is the 'indecisive addressee'. For a further characterisation of the different types of negative addressees (veiled, indirect, direct, third-person negative addressees) in political discourse, refer to García Negroni, 1988 & 2016.

9. Let us consider that in terms of argumentative semantics (Carel & Ducrot, 2005), converse argumentation keeps the two segments of the sequence together, but it alters the connector and the negation type. Therefore, for example, the converse sequences X THF Y and X THF *Neg*. Y involve X HW *Neg*.Y and X HW Y respectively.

10. It should be noted that the negative utterance *not because we are better* cannot be analyzed as a polemic negation. In the first place, because it is followed by a rectification utterance introduced by *but*, characteristic of metalinguistic/metadiscursive negation. In the second place, because the positive position (i.e., *we are better*) to which the locutor L opposes is not internal to the discourse in which it is questioned. As shown in paraphrase (8a), the utterance involves a plausible discourse attributed to other locutor(s) in the enunciative scene. Hence the metadiscursive value of the negation at stake is confirmed.

11. In order to provide an example of this type of negative expression with a polemic *subjunctive* (RAE, 2009) in Spanish, we should consider the following excerpt taken from C. Kirchner's speech on 3/12/2016: '*No es que esté descreyendo –en absoluto–* de los partidos políticos y de las organizaciones partidarias, nada más alejado de mí, soy profundamente democrática, pero entiendo que no es suficiente'. *This doesn't mean I don't trust in political parties. I'm far from that idea. On the contrary. I deeply advocate democracy, though I understand it is not enough*. According to RAE (2009, p. 1945), this polemic *subjunctive* (i.e. *'No es que esté descreyendo', This doesn't mean I don't trust*) appears to contradict or reject a previous affirmation.

12. In García Negroni (2019), I deal with *alluding PoVs*. Strongly connected to what Authier-Revuz (1984, 1992, 2020) refers to as marked or unmarked forms of heterogeneity, these PoVs involve saying forms in which something said in previous events is alluded, transformed or framed. In fact, as they do not involve the explicit object of the act of enunciation but imply alluded sense, these PoVs are likely to be overlooked. However, and in line with Authier, if they are identified by the interpreter they act as a memory of previous discourses and they enable the interpreter to have access to them.

## **References**

Aikhenvald, A. (2004). *Evidentiality*. Oxford: Oxford University Press.


Anscombre, J.-C. (1995). *Théorie des Topoi*. Paris: Kimé.


and the expression of surprise, *Pragmatics*, https://doi.org/10.1075/prag .20015.gar


## **5. A Corpus-Pragmatic Approach to Negation in Electoral Tweets**

*Elena Albu and Francesca Capuano*

## **1. Introduction**

The extensive use of Twitter to serve political agendas has led to the emergence of a new type of political discourse in the form of political tweets. Characterised by salient textual and discursive features, political tweets have led to a reassessment of the traditional division between spoken and written language, and a re-evaluation of political discourse. The analysis of the constructional strategies and interactional functions of negation may indicate whether the balance is tipped towards the prevalence of oral or written features in political tweets and whether the formal register of political discourse has generally shifted to a more flexible and colloquial form, particularly during electoral campaigns.

This paper aims to discuss the way in which negation is used in the political tweets sent by the UK candidates during the European Parliamentary Elections held in May 2014. Although negation in English is usually represented by *not*-negation and *no*-negation (Biber et al., 1999; Tottie, 1991), we distinguish between *no*-negation, *not*negation and *n't*-negation as separate negative categories (for a similar approach, see Xiao & McEnery, 2010). This tripartite division is ascribed to the textual and discursive properties of tweets, thus better serving the purposes of our investigation. This study will help identify whether the negative constructions in the dataset are illustrative of spoken or written language and will point out how certain features of these constructions contribute to shaping the individuality of political tweets. These research questions are intended to cast light on the on-going debates about the oral vs. written features present in

**How to cite this book chapter:**

Albu, E. and Capuano, F. 2023. A Corpus-Pragmatic Approach to Negation in Electoral Tweets. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 113–144. Stockholm: Stockholm University Press. DOI: https://doi.org /10.16993/bcd.e. License: CC BY 4.0.

tweets, and about the colloquialisation of political discourse in the digital environment1 .

This paper is an exploratory study and uses the tools and methods provided by corpus pragmatics (Aijmer & Rühlemann, 2015; Jucker, 2018; Romero-Trillo, 2008; Rühlemann & Aijmer, 2015). The first part of the analysis will take a contrastive quantitative approach, complemented in the second part by qualitative analyses of the collocates found in the immediate surrounding context. Particular attention will be paid to the prevalence and distribution of *no*-negation, *not*-negation and *n't*-negation. By means of syntagmatic patterning, we will carry out two types of collocational analyses: we will analyse the top ten token collocations and the top four part of speech collocations for the negators *no* and *not* with a span of one word to the right and to the left. These will be complemented by contextual analyses, which are meant to disambiguate between the multiple values of the negators. By contrast, the contracted form *n't* cannot be used independently. We will, therefore, present the overall distribution of the auxiliary and modal verbs to which *n't* is attached and carry out contextual analyses of the lexical verbs which combine with the auxiliaries, modals and *n't*.

The paper is structured as follows. In Section (2), we will briefly present the main features of tweets, in general, and on political tweets sent during electoral campaigns, in particular. Section (3) is dedicated to the discussion of the three categories of negation: *no*-negation, *not*negation and *n't*-negation. In Section (4), particular attention is paid to the data analysis, with an emphasis on how the relevant data have been retrieved. For *no*-negation and *not*-negation, we will present the overall frequency of the negation categories and we will analyze the top ten token collocates and the four most used parts of speech, with a span of one word to the right and to the left. For *n't*-negation, we will present the overall distribution and analyse the auxiliaries and modal verbs *n't* attaches to. Section (5) concludes the paper.

### **2. The linguistics of tweets**

The extensive use of the Internet and computer technology has led to the emergence of more types of digital communication and discourse genres (Davis & Brewer, 1997; Panckhurst, 2006; Thurlow & Mroczek, 2011; Bou-Franch & Garcés-Conejos Blitvich, 2019). The messages sent on the Twitter platform, coined as tweets, are one such example. In comparison to other forms of digital communication, tweets display important changes and significant variation in the way in which language is structured and used. In what follows, we will briefly present the formal and discursive features of tweets and then indicate how they shape the electoral political discourse in the digital environment.

Tweets are a form of multimodal communication, where typed, oral and visual elements combine in novel forms. Although tweets are mainly text-based, images, audio elements and videos can also be used, hotlinked or embedded and displayed inline (Wikström, 2017). Generally limited to no more than 140 characters, the body of a tweet may also contain communicative operators: @-mentions, retweets (RT), hashtags (#) and hyperlinks (http://). Having developed an important number of pragmatic functions, the operators compensate for the lack of any visual or auditory information, such as body movement, facial expressions, eye gaze, or intonation and pitch of the voice (Scott, 2015; Einspanner-Pflock et al., 2016). In addition, other Twitter-specific strategies have been developed: non-standard spelling conventions, alternate spellings of words, special orthography and the creative use of punctuation, acronyms, emoticons and emojis (Androutsopoulos, 2015; Knobel & Lankshear, 2008). Other techniques, such as abbreviations, clippings, orthographic reduction, shortenings ellipsis and deletion of pronouns, are used as a means of linguistic economy (Ferrara, Brunner, & Whittemore, 1991; Werry, 1996). However, not all forms are necessarily intended, as some spelling errors or the utilisation of non-standard punctuation and upper case may also be the result of the spontaneous nature of digital communication (Sims, 1996).

Tweets are often considered a form of written discourse, which contains oral features or have an oral discursive style (Soffer, 2016; 2019). In this regard, different labels have been suggested to describe the language on Twitter: 'spokenlike', a 'chatty' writing environment or a part of 'Netspeak', which relies on characteristics of both speech and writing (Crystal, 2006; Kern, 1995; Wikström, 2017). Some of the features that have been attributed to orality are: lexical substitutions, where the phonetic sound of a single letter or digit may replace entire words or repetitions of letters, filled pauses (uhm, uh..), onomatopoeic signs, the strategic uses of punctuation, and the use of emoticons and emoji (Soffer, 2019; Yus, 2011; 2014). Typographic playfulness, flexibility and multimodality are usually employed to establish a dialogical atmosphere (Jovanovic & Van Leeuwen, 2018). In comparison with face-toface conversation, there are no turn transitions, silences, overlappings, interruptions, or interactional combinations such as adjacency pairs and latched turns on Twitter. The number of digressions and hesitations is usually reduced, and if they are used, they are pragmatically loaded. In addition, despite their written format, tweets are said to be registered psychologically as having the temporal immediacy of oral exchanges, which has led to their labelling as 'textual verbal exchanges' (W. Ong in Kleine & Gale, 1996).

The interaction on Twitter can be described as a form of conversation (Zappavigna, 2012), but compared to face-to-face conversation, online conversations often shift from the prototypical dyadic structure to a multi-party configuration, allowing the user to participate in an open-ended, multi-user, one-to-many conversation (Wikström, 2017). The tweets are usually public, the messages becoming instantly visible to other users, especially the sender's followers. Even if the messages are sometimes directly addressed to a user, they can be seen by others who can also take part in the interaction by means of a relevant hashtag. If different forms of Internet communication have been initially described as either being synchronous or asynchronous (Hirotani, 2009; Munneke et al., 2007), tweets are characterised by a mixed temporality. Herring (2008) suggests the term of 'semisynchronous', while Wikström (2017) uses the notion of 'heterochronicity' to talk about the simultaneous presence of several different temporal logics, i.e., they are both interactions in the present and archived artefacts.

Political talk has quickly adapted to the digital environment and to the novel production conditions and, as a result, a new form of political communication has emerged in the form of political tweets. Political tweets are powerful communicative tools, especially during electoral campaigns, and have been greatly used during parliamentary, presidential, congressional, federal and local elections in numerous countries (Conway et al., 2012; Gaffney, 2010; Golbeck et al., 2010; Larsson, 2012). For instance, the 2014 European elections are considered to be 'the Twitter elections' due to the extensive use of the online platform in all 28 member states (Smyrnaios, 2014).

Political discourse is going through a process of 'hybridisation', as a result of the digital environment in which it is used (Moschini, 2010). The structure of the electoral talk has undergone a major change: the traditional one-way oriented electoral speech has shifted to a multi-party configuration and a conversational structure, which invites both the senders and the receivers to interact. The digital medium has shortened the distance from the audience, granting them the possibility to interact with and get immersed in the campaign. In this regard, the political tweets sent during electoral campaigns have been labelled as 'electoral' or 'collective' conversations (Moschini, 2010), where both the public sphere and the private dimension are combined.

## **3.** *No***-negation,** *not***-negation and** *n't***-negation**

Negation is a complex linguistic phenomenon, with multiple forms of expression and various discursive functions. It can be expressed by employing negative markers, *n*-words, negative affixes and lexical words. It is also associated with discourse functions, such as denial, contradiction, rejection, refutation, disagreement and irony, to name a few. Although negation is often associated with negativity, the two notions do not overlap: not every negative form leads to a negatively connoted structure and not every negatively connoted structure is expressed by means of a negative form (for a comprehensive overview, see Horn & Wansing, 2017; Horn, 2001).

The typical ways of expressing negation in English are the negative adverb *not* (including the inflected bound morpheme *n't*) and the particle *no* (including the *n*-words *neither, never, nobody, none, no one, nothing, nor, nowhere*). This formal distinction has led to the classification of negation in *not*-negation and *no*-negation (Biber et al., 1999; Tottie, 1991). The negator *not* and the contracted form *n't* are usually used in the verbal phrase. However, *not* can also be part of non-verbal negative structures, i.e., when it is associated with quantifiers, adverbs, determinatives, degree expressions, prepositional phrases or coordinating structures. There are cases when the structures with *no* and the structures with *not* are used in alternation or have little difference in meaning. If they are grammatically similar, they display pragmatic and stylistic differences: for instance, *no* followed by a noun makes the negation stronger in comparison with *not* followed by the indefinite *a* or *any*.

Following Xiao and McEnery (2010), we distinguish between *no*-negation, *not*-negation, and *n't*-negation as separate categories. We consider that the tripartite division captures better the uses of negation, in line with the textual and discursive features of tweets. From a usage-based perspective, negation is globally more frequent in spoken language than in written texts (Tottie, 1991), and more prevalent in conversation when compared to fiction, news and academic texts (Biber et al., 1999). Additionally, different negative constructions are considered representative of either spoken or written language, formal or informal register, and dialogical or monological structure (Palacios Martinez, 1995; Roitman, 2017; Silvennoinen, 2017; Xiao & McEnery, 2010). Usage differences are found between the full form *not* and the contracted form *n't*: for instance, there is a preference for *not* in writing and for *n't* in spoken language (Xiao & McEnery, 2010). Furthermore, the contracted form *n't* is preferred in conversational contexts, unless there is a clear reason (such as strong emphasis) to use the full form. The contracted form is a marker of informality, and is not preferred in writing or in solemn contexts. When it is used, it increases the level of familiarity and accessibility of the information conveyed. With these findings in mind, before embarking on the analysis of *no*-negation, *not*-negation and *n't* negation in our corpus, we formulate the following assumptions:


## **4. Data analysis**

## 4.1. Methodological aspects

This investigation is part of the international project "Twitter at the European Elections: A Comparative International Study of the Use of Twitter by Candidates at the European Parliamentary Elections in May 2014". The project aimed to compare Twitter use during the European Elections in several countries: Belgium, France, Germany, Italy, Spain and the UK. All the tweets sent by the candidates in these countries, along with the messages addressed publicly to them and major hashtagrelated conversations associated with the elections in each country were gathered (see Brachotte & Frame, 2015; Frame et al., 2016). Over 50 million tweets were collected in total within a time span of one month: three weeks prior to the elections and one week after the election date. The corpus under analysis comprises of the tweets sent by the UK candidates running for the European Parliamentary Elections in May 2014. A total number of 72,859 tweets (8,346,373 words in total, including all the retweets) were collected from all the 309 Twitter accounts identified as belonging to the UK candidates from 32 national political parties.

The hybridity and medium-specific properties of tweets (the nonconventional use of punctuation, the lack of capitalisation, the alternate ways of spelling the negative particles and the use of ungrammatical structures), the size of the corpus under investigation, and the exploratory nature of the study have posed challenges to the retrieval of the relevant data, rendering the simple automatic searches based on criteria of form more difficult. To retrieve as many instances of *no, not* and *n't* as possible, we first automatically split all the tweets by whitespaces and then defined category-specific steps within these substrings. For *no*-negation, we assumed that, once lowercased and deprived of any non-alphanumeric word character, the corresponding strings would be proper instances of *no* (except when it was part of the expression *no one* or part of a nickname). The remaining instances were inspected by hand, and in case of uncertainty, we referred back to the whole tweet. For instance, we eliminated the instances when *no* was used as an abbreviation for number one (e.g., *no.1*). The possible misspelling or alternate expression of the negative particle were also manually checked and included in the final count (e.g., *nooooooooooo, NO2EU*). The same procedure was followed for retrieving the instances of *not*negation. The presence of *nt* and *not* with additional *o* characters (e.g., *noot, nooot*, etc.) as possible misspellings or alternate expressions of *not* were also inspected manually and included in the final count.

In contrast, the contracted form *n't* is always attached to an auxiliary or a modal verb. We retrieved all the unique substrings that, once lowercased, contained *n't*, or the potential alternatives *n´t, n`t* or *nt*. Due to the big number of English words that include the substring *nt*, we restricted the search to the substrings that only ended in *nt* or contained at least one of the auxiliary or modal verb + *n't* tokens found in the previous step as well as any hashtag containing *nt* in any position in the tweet. Again, we always referred back to the whole tweet if needed (e.g., for *ant* we kept only the instances where it was used as an alternative for *ain't*).

#### 4.2. Overall frequency

The overall frequency comprises all the tokens that belong to one of three negation categories. Overall, 17.18% of the tweets in the corpus contain at least one negative instance, as shown in Figure 5.1. Regarding the overall counts per category, there are 5,230 total instances of *not*, 5,234 total instances of *n't* and 3,986 total instances of *no*, as shown in Figure 5.2.

**Figure 5.1.** The overall distribution of negation.

**Figure 5.2.** The distribution of *no*-negation, *not*-negation and *n't*-negation.

From a contrastive perspective, *not*-negation and *n't*-negation seem to have similar distributions (36.23% vs. 36.21%), while *no*-negation is less represented (*no* alone represents approximately 19%, while *n*-words represent approximately 9%). If we compare the instances of *not*-negation and *n't*-negation (72.44% in total) with *no*-negation, the latter is less numerous in our corpus. The similar distributions of *not* and *n't* do not allow us to draw any conclusions regarding the first assumption about the spoken vs. written features of tweets. Nevertheless, the overall frequency does not imply that the negative constructions have the same function. The contextual analysis will reveal whether *not*-negation and *n't*-negation are used interchangeably or whether there are finer-grained differences that can be drawn between them.

Given the textual and discursive features of tweets, we have added to the general counts all the instances of the negative markers, including their presence in longer strings and hashtags, and their combinations with both conventional and non-conventional punctuation marks. The non-conventional use of punctuation marks may be a typing mistake, but it could also be a strategy used in the attempt to gain more space, given the limitation of 140 characters. Table 5.1 shows the overall counts for *no*-negation and *not*-negation.

Non-conventional spellings, non-standard uses of punctuation marks and inconsistency with standard capitalisation have led to variation among the forms of the negative markers. To illustrate, the standalone counts of *no* (1,338 instances of *no*, 615 instances of *No* and 91 instances of *NO*) are not very faithful to the traditional punctuation and spelling conventions: *no* is sometimes used in initial position, whereas *No* is sometimes found in the body of the tweet. The particle *no* is multifunctional, and only after analysing the contexts in which the particle is used, will we be able to accurately tell when *no* is used as a response particle or as a determiner (expressing both negation and quantification in a noun phrase structure).


**Table 5.1.** Overall counts for *no*-negation and *not*-negation.

Similar to *no*, there is also variation among the instances of *not*: the instances of stand-alone *not* consist of 3,786 instances of *not*, 568 instances of *Not* and 251 instances of *NOT*. We could also identify one instance of *nt* used as *not*, as shown in example (1):

(1) Hd 5k more voted in 2009 we wld nt hv hd 2 endure 5 yrs of bnp hate n portrayal of UK as intolrnt anti-islamic soc Pls dnt let UK dwn again

As previously mentioned, the contracted form *n't* was found to have alternate spellings: either different symbols are used for the apostrophe or the characters are written together (*nt*). In comparison with *no* and *not*, the bound inflectional form *n't* can only be part of longer strings, represented by the prototypical grammatical instances (*don't, can't*, etc.), non-standard instances (*dont, cant*, etc.) or hashtags *(#itslikehedoesntevenknowme*, *#DontVoteUKIP, #cantbeliveaword*). In sum, there are 4,985 instances of *n't*-negation and 249 instance of *nt*-negation. In line with the conventions of use, the hashtags can only contain the form *nt* (31 such instances were identified). A total number of 4,903 tweets have been retrieved.

### 4.3. Collocations

This subsection discusses the token collocations of the negative markers *no, not* and *n't*. For simplicity reasons, we adopt a general view of collocation by which we intend any corresponding token identified by the tagger, be it a grammatical unit, a hashtag or punctuation, with a span size of one to the left and to the right. Punctuation marks are not usually found among the inventory of collocates but we have deemed their inclusion necessary because they are usually employed intentionally and strategically to compensate for the absence of other features. Based on the top neighbouring elements to their right and to their left, it is possible to formulate general assumptions about their behaviour and combinatorial preferences. This is the first step in discriminating between their distinct grammatical properties and multiple pragmatic values. However, only by means of a through contextual analysis can we tell with precision how these negators are grammatically employed and what their pragmatic values are. This is, however, beyond the aims of this paper.

*No***-negation.** We have made an inventory of the first ten collocates and analysed the recurrent combinations with a span size of one word to the left and to the right. Table 5.2 shows the top ten collocates of *no*, together with the corresponding counts and percentages:


**Table 5.2.** Counts and percentages of the top ten collocates of *no*.

From a comparative perspective, the top ten collocates represent only 25% of all the instances of *no* to the right, whereas to the left they represent 40%. This shows that there is more variation to the right, indicating that *no* is being used in combination with more elements to the right than to the left. To the right, *no* is more frequently followed by punctuation marks, either commas or full stops, which suggests that the particle is most likely used as a response particle, i.e., being part of an on-going interaction between more users. It may be used to answer a question or to correct a mistaken belief or a previous statement. *No* is usually used in initial position, directly addressing previous tweets and users, as illustrated in (2), but it can also be used at the end of the tweet to refute a specific statement, as shown in (3):


The imperative value of *no* in combination with the preposition *to*, used as an incentive to take some action, is also among the top collocates, as illustrated in examples (4) and (5). Other patterns we could identify are the comparative constructions *no longer* and *no more* and the idiomatic phrases and formulaic expressions *no doubt, no wonder*, and *no idea*.


To the left, *no* collocates even more often with punctuation marks, but the values are different: after a full stop, a colon or a dash, *no* is most likely a quantifying determiner in a nominal phrase, used in negative explanations or additional comments, as shown in (6) and (7). The verbal tokens *have* and *is*, followed by the coordinating conjunction *and* and *but* and the preposition *with* are also among the top collocates. Although both verbal forms *have* and *is* are in the present tense, they show grammatical differences in the choice of the grammatical subject. Restricted only to the third person singular, *is no* combines mostly with the pronoun *there*, being part of the negative existential expression *there is no*. In contrast, as a result of the great variability of the items used to the right of the collocation, no particular pattern seems to stand out for *have no*. Taken together, the particle *no* has weak recurrent patterns as a result of the great variability among the elements used.

*Not***-negation.** Looking at the top combinatorial preferences of *not* in Table 5.3, to the left, the collocates are divided mainly into two types: punctuation marks and primary and modal auxiliaries. This leads to a first division between the cases when *not* is part of the verbphrase negation, and when it is part of non-verbal negation. Using *not* after commas, full stops and colons is an indication that the negative adverb is part of non-verbal negations, a different constituent being focused in order to give prominence to the associated piece of information.

The examples below illustrate a complex negative structure, consisting of a rejection segment (X) and a correction segment (X'). In examples (8), (9), (10) and (11), the negative segment (X) is preceded by the corrective segment (X') as part of the negative structure [(X') not (X)], whereas in (12), the order of the segments is reversed, being part of the structure [not (X) but (X')].


**Table 5.3.** Counts and percentages of the top ten collocates of *not*.


These are constructional strategies of contrastive negation (McCawley, 1991; Silvennoinen, 2017), which are a type of metarepresentational negation (Albu, 2012a; 2012b). The constituents in the scope of negation represent discourse-old information, while the corrected information represents discourse-new information. The negative segments (*not a politicians answer, not a twisted one, Not individual muslims, Not union or business funded, Not as a default choice*) reject some statements, which were either previously mentioned or implicitly recovered. In other words, the negative construction renders the negated information discourse-old information. Taken together, [(X') not (X)] and [not (X) but (X')] accommodate the negated information, which was probably not part of the hearer's knowledge before. Even though

both structures have the same structure – rejection of some propositional content and correction – the order in which the segments are used may have different motivations. Although the typical construction is [not (X) but (X')] with the negative segment on the first position, the second pattern with negation on the second position [(X') not (X)] is very well represented in our dataset. It appears to be an efficient and more economical construction with a strong argumentative force. It may, however, be more costly, demanding more cognitive effort in retrieving what the negated information represents and to whom it may be attributed. These constructional forms are employed more frequently in speech than in writing, as conversations are usually dominated by asyndetic clause combinations (Silvennoinen, 2017). Similarly, no corrective conjunctions are used in the corpus, and additionally, there does not seem to be much consistency at the level of the punctuation mark used, as the negative segment can be delimited by commas, full stops and dashes. It can, therefore, be speculated that the extensive use of these negative constructions represents a mark of spoken language and conversational style.

When preceded by punctuation marks, *not* can also be used in a less conventional way, as part of an elliptical construction in which the main verb is omitted: *not sure about* standing for *I am not sure about, not cool* standing for *it is not cool, not the way* standing for *this is not the way*, and *not interested* standing for *I am not interested*, as shown in (13–16). This telegraphic way of expression in which the verbs, the first-person pronouns and the discourse markers are omitted may be a strategy employed in the attempt to recreate spoken language and to shift away from the formality imposed by the traditional political discourse:


The negator is also associated with different inflections of the verb *to be* (*is not, are not, I'm not*), the modal *will* and the auxiliary *do*. Looking at the combinatorial patterns to the right, the verb *to be* is used as a copula verb, as shown in (17) and (18). In contrast, *will not* and *do not* usually require a lexical verb to the right, except the cases when they are used at the end of the utterance. *Will not* is used in combination with more elements, but the recurrent patterns contain the verbs *to be* and *stop*, as illustrated in (19) and (20). *Do not* can be used in the indicative mood, situation in which it collocates with lexical verbs such as *want, have, understand, think*, or in the imperative mood, meant to convey various requests or commands, as shown in (21). *Does not* forms weak collocations patterns with *have, mean* and *get*.


When looking at the distribution to the right, the first positions are occupied by the indefinite and definite articles, indicating a nominal preference, to the expense of verbs, which are less frequently used. *Be* and *have* are among the top collocates, being most likely preceded by modal verbs (*will not, may not, should not, could not, would not*). An interesting collocational pattern is represented by *not* followed by full stops. In addition to its common value, i.e., used in combination with the subject and the auxiliary verb to reinforce negation, *not* can also be used as an anaphoric pro-form for a negative clausal complement, marking non-verbal negation in structures like *I'm afraid not, I suspect not, Thought not*, as illustrated in (22) and (23):


All in all, there is great variation among the elements used, which leads to a lack of recurrent expressions and repetitive constructions. The use of the present tense is predominant, while past tense seems to be deficient. Negative descriptions and negative evaluations with mainly mental verbs anchored in the present appear to be preponderant. This is not a surprising finding if we consider the context in which the tweets are sent. Unexpectedly, the distribution of *not* in verbal (approximately 52%) and non-verbal negations (approximately 48%) appears to be almost equal among the top ten collocates.

*N't***-negation.** In comparison with the negative markers *not* and *no*, the bound inflectional form *n't* is always attached to primary or modal auxiliaries. Instead of the top ten collocates to the right and to the left, we present the distribution of the contracted form *n't*, including the alternative spelling *nt* in combination with the verbal base, as shown in Table 5.4. The merged lemmas stand for all the instances of both *n't* and *nt* tokens, while the other columns (\**n't* tokens and \**nt* tokens) show their frequencies separately. The lemmas are illustrated in the table with capitalised letters, but the counts include tokens with both upper and lowercase letters. To show their general prevalence, the percentages include all the *n't* and *nt* instances (5,234 in total).

The data show that there is a strong tendency to use *n't* to the expense of non-conventional forms (which represent only 3.98%). We could also identify alternate spellings of some auxiliary verbs, but they are isolated occurrences: *dn't, havn't, din't, sin't, dsn't, musn't, arn't, cudn't, cdn't, wldn't, wdn't*. Overall, the most used auxiliaries are *don't, didn't*, and *doesn't*, followed by *isn't, haven't, aren't*, and *wasn't*, whereas the most used modal verbs are *can't, won't, couldn't*, and *wouldn't*. From a comparative perspective, the auxiliary *do*, including all its inflected forms, represents approximately 55% of the data, the modal verbs altogether represent approximately 27%, and the auxiliaries *be* and *have* represent approximately 13% and 6%, respectively. If one looks at the stems to which *n't* attaches, *don't* is the most used form, representing 38.8% of the entire category of *n't* negation. The second most used verbal base is the modal verb *can't*, but the latter is considerably less used. *Don't* is widely employed, either as an auxiliary in the indicative mood, as shown in (24) and (25), or as part of an imperative structure, as illustrated in (26) and (27). When used in the indicative mood, it further correlates to the left with subjects such as: *I* (282 instances), *you* (178 instances, including u), *we* (141 instances), *they* (82 instances), and *people* (49 instances, including *ppl*). There are also cases, when the subject is omitted (*I* in particular), contrary to the grammatical rules in English. In line with the findings in Werry (1996), the deletion of pronouns may be a strategy of linguistic economy.





As expected, to the right, *don't* is followed by lexical verbs: *forget* (192 instances), *want* (111 instances), *have* (110 instances), *know* (101 instances), *think* (87 instances), *let* (80 instances) and *vote* (71 instances). *Don't forget* further collocates with the preposition *to* (139 instances), followed either by the verb *vote* (52 instances) or similar expressions, e.g., *go and vote, use your vote, post your postal votes, get out and vote, cast your vote, register to vote, have your say and vote*. There does not seem to be a preference for the subject with which *don't want* combines, because the distribution of the pronouns is similar. In contrast, the verb *think* shows a preference for the first-person singular pronoun (*I don't think*), 63 instances being identified in total. Our observations are in line with the findings in Iyeiri, Yaguchi and Baba (2015), who found that the mental verbs *know, think* and *want* were frequently used in the press conferences from the White House. Biber et al., (1999) further highlighted that mental verbs (e.g., *forget, know, mind, remember, think, want*, and *worry*) are more likely to collocate with negation in conversations. If we compare the same mental state verbs used in combination with the full form *not*, the recurrent patterns change. We could find only one instance of *do not forget*, 12 instances of *do not want*, 11 instances of *do not have*, two instances of *do not know* and three instances of *do not think* and *do not let*. These findings endorse the observation that the association with mental verbs is especially strong with the contracted form *n't*  (Biber et al. 1999).

The first ten collocates of *n't* show a preference for the auxiliary *do* compared to the collocates of *not*: *don't* represents approximately 39% of the data, whereas *do not* represents only 2.3%. This preference is contrary to the collocational tendencies of the token collocates of *not*, which showed a greater preference for the verb *be*. Another difference is found in the use of the modal verbs. There are four modal verbs among the first ten collocates of *n't* (*can't, won't, couldn't*, and *wouldn't* which represent almost 25%), whereas only the modal verb *will* is found among the top ten collocates of *not* (2.33%). There is also a preference for past tense among the *n't*-negation (*didn't*), tense which was less employed with the full form *not*.

In sum, if initially *not*-negation and *n't*-negation showed similar distributions, the analysis has revealed that *n't*-negation is the predominant form of verbal negation. This further indicates that the full form *not* and the contracted form *n't* shows that they are not used interchangeably, but instead they form different patterns and have distinct combinatorial preferences.

### 4.4. Part-of-speech collocations

This subsection is dedicated to the analysis of the tokens that belong to the same part of speech. We used the off-the-shelf English Part-Of-Speech (POS) tagger, especially created for Twitter data (Gimple et al., 2011), and retrieved the four most frequent parts of speech employed to the right and to the left of the negative markers.

*No***-negation.** Table 5.5 illustrates the main tagsets for *no*-negation in the corpus. The distribution to the right reveals novel patterns. In contrast to the information in Table 5.2, where the top collocates were mainly idiomatic expressions, the part of speech with which *no* collocates the most is represented by nouns. Within the noun category, the idiomatic phrases and formulaic expressions (*no doubt, no idea, no way, no point, no problem, no time, no thanks*) represent approximately 20%. This means that the nouns are more numerous, but they are also very dispersed and thematically widespread, leading to a few recurrent patterns. In addition, there is a big numeric difference between the parts of speech ranked first and second, which indicates that there is a strong tendency towards employing *no* in nominal structures.


**Table 5.5.** Distribution of the main parts of speech with *no*.

To the left, the first two parts of speech that collocate with *no*, punctuation marks and verbs, corroborate the previous findings. The contextual analysis has revealed that the most frequent verb forms used in combination with *no* are the verbs *be* and *have*, followed by the verbs *say* and *vote*. Although they are not a traditional part of speech, @-mentions are also greatly used. This highlights the dialogical and conversational structure of tweets, indicating that the tweet is usually sent in response to previous messages and directly addressing particular users.

Regarding punctuation, if we compare the distribution of the punctuation marks to the right and to the left, *no* is more often preceded than followed by punctuation and the punctuation marks are more diversified to the left. The distribution of the punctuation marks to the left is presented in Table 5.6.


**Table 5.6.** Punctuation marks used to the left and to the right of *no*.

Using *no* after full stops indicates that it is used in initial position and that the tweet is made up of more utterances. As previously anticipated, the contextual analysis has revealed that *no* after a full stop is mostly used as a quantifying determiner in order to negate the noun phrase. Rarely does *no* appear in a fully formed utterance. Instead, it is frequently used in utterances where the predicate is omitted: in (28) and (29) the existential *there* is missing, in (30) the lexical verb and the subject are missing (*I do not have a tv*), while in (31) the elliptical predicate can be represented by either the existential expression *there is* or the lexical verb *have* (*I do not have any*). The corresponding utterances may also have different syntactic structures: for instance, in (33) the noun phrase stands for the entire utterance and seems to be part of an *if-then* structure, whereas in (30) and (32) *no* is used in relation to the conjunction *so*, highlighting the consequences of the information expressed by the negative noun phrase:


It can also be used as a directive speech act, aimed at provoking an action in the addressee, as shown in (34), or it can be part of a negative question, as illustrated in (35). These values are not numerous, but they reflect the multifunctionality of this negative particle.


When it is preceded by commas, *no* can be part of both positive and negative enumeration constructions, as illustrated in (36), (37) and (38), may be used to indicate repetition and emphasis, as shown in (39) and (40), or may be used to delimit markers of personal stance (*no surprise, no problem, no question, no doubt*) after addressing a particular person, as indicated in (41), (42), (43) and (44):


In comparison to the distribution to the left, when *no* is followed by a comma to the right, it is most likely used as a response particle. It may be surprising at first sight to find full stops after *no*, but the analysis revealed that they function like commas, *no* being used as a response particle. There are, however, some pragmatic differences. The use of a full stop after *no* as a direct reply to a previous question or statement is meant to give more assertive strength to the answer and to dissociate it from the rectification introduced afterwards by means of *but* and *and*. The full stop is used similarly to a pause in spoken language, as illustrated in (45) and (46). The users are usually mentioned by means of @-mentions but when a name is employed, it is pragmatically loaded, as shown in (47):


We could also identify instances of *no* followed by exclamation marks, which are thought to put greater emphasis on the negative answer. It can be used as a single answer, as illustrated in (48) and (49), or it can be followed by further comments, as shown in (50) and (51). The emphasis can be additionally marked by multiplying the exclamation marks, as in (50).


In sum, when used in initial position, *no* is very often part of elliptical constructions, where the verb and the subject are omitted. The negative nominal phrase usually stands for the entire utterance, which highlights a non-standard use of the particle *no*. In contrast, when it is followed by punctuation marks, *no* is usually employed as a response particle. Accordingly, the tweets can be characterised as informal, resembling spoken language: they are short, simple and have a telegraphic style. Although some patterns can be identified overall, a thorough contextual analysis of all the instances of *no* is necessary in order to identify all its pragmatic uses in the digital environment.

*Not***-negation.** The four most used parts of speech to the left and to the right of the *not* are illustrated in Table 5.7:


**Table 5.7.** Distribution of the main parts of speech with *not*.

It is not surprising to find that the top positions to the left and to right are represented by verbs. To the left, there is a tripartite division: auxiliary verbs, modal verbs and a smaller number of lexical verbs, as indicated in Tables 5.8, 5.9 and 5.10. The most represented verb is *to be* with a percentage of approximately 42%, followed by the auxiliaries *do* with approximately 19% and *have* with approximately 3%. The modal verbs represent approximately 20%, whereas the lexical items that are used before *not* represent approximately 15%.

If we take a closer look at Table 5.9, *hope* is highly prominent when compared to the other lexical items. Moreover, it is part of a larger collocational structure [(X') not (X)]: *hope not fear* or *hope not hate*, as illustrated in examples (52) and (53). The negative segment is associated with negative emotions, e.g., *fear* and *hate*, while the affirmative segment is associated with *hope*, which is positively connoted (Albu, 2018). The tweets in which the structures are integrated represent illustrations of 'call for action' tweets (Albu, 2016), i.e., they represent a form of strategic and goal-oriented discourse meant to persuade the electorate to vote for them.


**Table 5.8.** Distribution of the modal verbs to the left.

**Table 5.9.** Distribution of the lexical items to the left.


**Table 5.10.** Distribution of the auxiliaries to the left.



The forms counted under *can* include 153 instances of *cannot* and 12 instances of *can not*. In comparison with the *n't* forms, although numerically different, the first positions are occupied by the same modal verbs: *can* and *will*. As for the instances of the verb *to be*, approximately 91% of these (and approximately 38% from the total) are used in the present tense (*is, are, am*). Among the lexical items used before *not*, there are 24 instances of -*ing* forms, 39 instances of past participle in -*ed* and 11 instances of irregular past participles.

To the right, after checking all the corresponding inflected forms, *be* (9.68%) is the most common, followed by *vote* (5.98%), *have* (4.46%) and *hate, get*, and *fear* with smaller percentages. There are 459 instances of the progressive aspectual form -*ing*, which encompasses 24.82% of the total number of verbs used after the full form *not*.

The third category is labelled *nominal + verbal*, and stands for structures where the operator is contracted and attached to the subject, i.e., followed by the full form of the negator *not*. Most of the occurrences contain personal pronouns in the subject position and the most frequently used operator is *be* (approximately 92%). As Table 5.11 shows, there is a preference for the present tense compared to past tense and other aspectual forms. In terms of subject forms, the first three positions are occupied by the impersonal pronoun *it*, followed by the first-person singular pronoun *I* and the second-person pronoun *you*. At the opposite end of the spectrum, the contracted forms of *have* are very rare, representing only 4.5% of this category. This contrasts with the findings of negative declarative sentences in spoken and written varieties of English in Varela Perez (2013).

By comparing the overall prevalence of the full form not when preceded by verbs, the operator contraction and the *n't* occurrences, it can be concluded that *n't* is the most prevalent form of verbal negation, followed by the negations with the full form *not* and the operator contraction. If we compare the operator contraction occurrences with those with *n't*, the former shows a preference for *be*, whereas the latter is much more often used with different inflected forms of *do* (*don't, didn't, doesn't*). Another difference is that there is a preference for *I'm* in comparison with *I am*. There are less uses of *aren't* compared with *haven't*, which is contrary to the preference shown for the verb *be* in the case of operator contractions.


**Table 5.11.** Overall distribution of the operator contraction.

## **5. Concluding remarks**

Based on the assumption that negation may be a useful strategy in the investigation of electoral tweets, we discussed the constructional strategies of *no*-negation, *not*-negation and *n't*-negation. Due to the novel production conditions and particular interactional features of tweets, negation was divided into three heterogeneous categories, including standalone occurrences of the negative markers, conventional and non-conventional punctuation use, alternative spellings, and longer strings in which the negators were included. The results showed that *not*-negation and *n't*-negation are more prevalent (72.44%) compared to *no*-negation (approximately 27.57% of the relevant data, including *n*-words). Even though *not*-negation and *n't*-negation seem to have similar distributions, the collocational analyses revealed that these two negation categories are not used interchangeably, and their uses do not overlap. Instead, they have distinct patterns and combinatorial preferences. For instance, *not* contributes almost equally to verbal and non-verbal clausal negation. The latter type is mainly expressed by means of the construction [(X') not (X)], which highlights the use of negation in combination with correction. Another pattern of *not* is represented by elliptical structures, where the subject and the predicate are omitted, resulting in a telegraphic way of expression similar to oral conversation. The use of *not* followed by full stops highlights some non-standard values that *not* can have in final position: as an anaphoric pro-form for a negative clausal complement (*I'm afraid not, I suspect not, Thought not*).

In contrast, *n't*-negation is the most frequently used verbal negation in this dataset. If we compare the first ten collocates of *n't* and *not*, the bound inflectional form *n't* shows a preference for the auxiliary *do*. In contrast, the token collocates of *not* showed a tendency to combine with *be*. Another difference is found in the use of the modal verbs: whereas the contracted form is mainly used in combination with *can't, won't, couldn't*, and *wouldn't*, the negative adverb *not* is mostly used with *will*. The two negators also show dissimilarities in terms of the tenses with which they combine: *not* is associated most of the time with present, predominantly in negative descriptions and negative evaluations, while *n't* is used with both present and past tense. We could also identify instances of *not* preceded by operator contraction forms, but they are less frequent. The predominant operator in this case is *be*.

The analysis of the top ten token collocates of *no* has indicated that it forms weak recurrent patterns, as a result of the great variety of items with which it combines. The frequent combination of *no* with punctuation marks is an indication of the fact that the particle is most likely used as a response particle when it is followed by commas and full stops, and as a quantifying determiner when it is preceded by them. Nevertheless, the particle *no* is multifunctional, and has a multifarious discursive behaviour, as illustrated by the selective contextual analysis. Further research with thorough contextual analyses is necessary in order to disambiguate between the multiple values that *no* can have.

The division between spoken and written language is considered obscured in digital communication. Based on previous research, we have made some initial assumptions meant to help disentangle these issues. First, we assumed that the higher the frequency of negation overall, the more the tweets show similarities with spoken language. The results showed that negation is not massively employed in our dataset, as only 17.18% of the overall tweets contain at least one instance of negation. Therefore, the similarity with spoken language cannot be confirmed based on this criterion. The second assumption was about the correlation between *not*-negation and written language, whereas the third assumption correlated *n't*-negation with spoken language. The overall frequency showed a similar distribution of *not*-negation and *n't*-negation, which renders the predicted assumptions between spoken and written language difficult to account for. Taken together, the general distribution of these negation categories in the present dataset do not show strong similarities with either the spoken or the written language. However, the patterns generated by *not* are usually illustrations of spoken language, showing a tendency towards the prevalence of oral features in comparison with written features.

Regarding the questions of whether the formal register of political discourse has shifted towards a more flexible and colloquial form, the short format and the interactive structure of tweets speak in favour of a new type of political discourse with distinctive properties. Although divergent from casual conversation, the political tweets in our dataset present features that point towards a shift from the formality imposed by the traditional political discourse. These include the use of elliptical structures, the use of non-verbal negation with the full form *not*, the telegraphic way of expression in which the verbs, the first-person pronouns and discourse markers are omitted, the typographic playfulness, the use of non-conventional punctuation with negative markers and the non-standard negative structures.

## **Acknowledgements**

We would like to thank Olli Silvennoinen, Baiyao Zuo, Pierre Larrivée, and the anonymous reviewers for their constructive suggestions and valuable comments on previous drafts of the manuscript. The remaining shortcomings are entirely our responsibility.

## **Endnote**

1. This investigation is part of the international project Twitter at the European Elections: A Comparative International Study of the Use of Twitter by Candidates at the European Parliamentary Elections in May 2014. http:// iutdijon.u-bourgogne.fr/pedago/src/politicsmediastic/en/tee2014-2/

## **References**


Thimm, C. (eds.) *Tweets from the Campaign Trail: Researching the Use of Twitter during the European Parliamentary Elections*, Peter Lang, pp. 145–169.


## **6. A Corpus Study of Grammatical Negation in US Presidents' Inaugural Speeches**

*José Manuel Durán*

### **1. Introduction**

Patterns of grammatical negation are found to be quite frequent in portions of an inaugural address by Argentine President Alfonsín (Lavandera and Pardo, 1987). However, as the authors argue, there is a need for larger studies that analyse negation in corpora of inaugural addresses so as to gain deeper insights into the use of this strategic device by presidents. This chapter aims to fill in this gap through a study of grammatical negation in the framework of Systemic Functional Linguistics and Corpus Linguistics. I scrutinise the inaugural addresses of all 44 US Presidents in search of all instances of grammatical negation1 and their patterns of collocation and colligation.

Patterns of negation in political discourse have been explored from different approaches. For example, negation has been categorised from a polyphonic discourse analysis standpoint (Ducrot, 1984; García Negroni, 2009, 2016). However, the quantification of the linguistic strategies that politicians use in their discourse has still not attracted enough attention within this perspective (Cfr. Roitman, 2014, 2017). This is where Corpus Linguistics comes into play, as it contributes to understanding the realisations of systemic features in a corpus of texts.

Additionally, register analysis has contributed to characterising the sources of the texts that make up a corpus. Registerial analysis of US inaugural addresses reveals their shift from the veneration of the past to the enunciation of political principles (Campbell and Jamieson, 1990). While their focus on policies has given way to a focus on values (Chester, 1980), their purpose has shifted from personal beliefs to popular values

#### **How to cite this book chapter:**

Durán, J. M. 2023. A Corpus Study of Grammatical Negation in US Presidents' Inaugural Speeches. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 145–176. Stockholm: Stockholm University Press. DOI: https://doi.org /10.16993/bcd.f. License: CC BY 4.0.

(Windt, 1984). Still, it is necessary to try and find the frequencies of patterns2 in order to obtain more objective results (Denton and Hahn, 1986). Such is the quantitative analysis of political speech from Truman to Reagan (Hart, 1984), which covers only 14% of the 228-year span of American presidency. This proves the need for a large-scale study on inaugural addresses.

Such a large-scale study can fruitfully be carried out in the theoretical framework of Systemic Functional Linguistics (SFL) for a number of reasons. First, SFL establishes and defines the systems of language. Second, it allows for the quantification process of language features in a system with the aid of Corpus Linguistics. Additionally, SFL pays great heed to the registers that contextualise the texts constitutive of the corpus.

Previous SFL studies focus on the contrasts between a few presidential speeches (Durán, 2008; Krizsán, 2011) or the analysis of the speeches of a single world leader such as Obama (Kazemian and Hashemi, 2014), Mandela (Martínez Lirola, 2012; Nur, 2015), or Buhari (Koutchadé, 2015). However, so far there seems to be no large-scale studies that cover all the speeches that comprise a closed-set corpus. Such is my aim in this work on the study of inaugural addresses by all US Presidents. Naturally, such an all-encompassing task requires a focus on one particular system of the system networks that constitute language (Halliday and Matthiessen, 2014). My choice here is the system of negation, after Lavandera and Pardo (1987).

My hypothesis is that since inaugural addresses mark a pivotal point in history (Schlesinger and Israel, 2009), language features such as negation must play a crucial role in US Presidents' discourse. As US inaugural addresses shape new presidential terms, the lexicogrammatical choices made are carefully selected and are expected to exhibit a characteristic pattern. One linguistic device that is highly exploited in political discourse is the pattern of negation in an inaugural address (Lavandera and Pardo, 1987). If this pattern is recurrent in US inaugurals, the frequency of negative polarity items is likely to reach high levels.

The study is organised as follows. Section 2 summarises the theoretical framework of SFL. Section 3 gives details of the corpus and methodology used. Section 4 deals with the results of the study and is organised into four subsections devoted to the overall results, the frequencies of the most pervasive negative polarity items, and *not*-negation and *no*-negation, respectively. The chapter closes with some concluding remarks.

## **2. Theoretical framework**

Systemic Functional Linguistics is a social theory of language that focuses on meaning potential. In this Section, I will briefly summarise some of the most important guiding principles of this theoretical framework, namely context, function and system. Additionally, I will outline the system of negative polarity in English from the SFL approach and present the contextual features of field, tenor and mode in inaugural presidential addresses. Finally, I will schematise the contribution made by Corpus Linguistics.

Within SFL, language is deemed a resource for making meaning that speakers or writers use in specific social contexts. Thus, the language used by doctors in the institutional context when they perform the medical examination of a patient is different from that used by professors in an end-of-term university class when they interact with their students. There are many variables involved in the study of language in context. In the case of political discourse in particular, considered in its narrow sense—i.e., that produced by political actors to achieve political goals in (in)formal contexts (Graber, 1981) —some of the variables at play are the role of the speaker/writer, the interlocutor(s) or intended audience, their parliamentary party, the topic of the speech, and the time of government at which it is pronounced, among many others.

Apart from the social contexts that constrain language use, one of the fundamental tenets of SFL is its functional approach. SFL places a great deal of emphasis on the functional characteristic of language. The functions of language are theoretically unlimited (Thompson, 2014, p. 46). Speakers of a language use it for a range of things: to ask, request or suggest; to complain, deny or promise; to greet, thank or apologise; to inform, report or explain; and to perform many other functions. However, within SFL these functions have been encapsulated in 'the four primary speech functions of offer, command, statement and question' (Halliday and Matthiessen, 2014, p. 135). These functions3 are defined in terms of two basic variables, the role in exchange and the commodity exchanged, as summarised in Table 6.1. Examples (1) to (4), taken from my corpus, illustrate instances of an offer, a statement, a command and a question, respectively.


A further precept of SFL is its systemic view of language, which entails that meanings in SFL are expressed paradigmatically more than syntagmatically. The main focus of the theory is the system4 , rather than the structure. When speakers/writers use language, they make choices within the systems available in that language. One of the aims of systemic linguistics is to try and account for the description of the choices made by the users of the language when they perform a specific function in a specific context. The kind of choices that these users of the language make are realised in the lexicogrammar and lexicogrammatical structures available in the system of the language (Halliday and Hasan, 2000; Hasan, 1985).



SFL is organised in a very intricate network of systems that are highly dependent on the language or variety of language under study. Systemic grammars of languages have approximately between 700 and 1000 systems (Halliday and James, 1993, p. 95). I will focus here on the system of negation. Every clause in a text reflects either one of the two choices in the system of POLARITY5 . In this respect, every clause is either positive or negative in value. It has been proven that in the overall picture of the English language, positively polarised clauses are chosen 90% of the time while negatively polarised clauses are used only 10% of the time (Halliday and James, 1993). This is expressed graphically in Figure 6.1 with the respective probabilities attached to the two terms: 'positive, 0.9; negative, 0.1'. This constitutes a highly skewed system wherein the least frequent alternative is said to be the marked choice (Halliday, 1991a).

**Figure 6.1.** The system of POLARITY in English.

Source: Halliday and Matthiessen (2014, p. 23). Copyright 2014. From *Halliday's Introduction to Functional Grammar* by Michael Halliday and Christian Matthiessen. Reproduced by permission of Taylor and Francis Group, LLC, a division of Informa plc.

At a more delicate level, negative polarity can be subdivided into two types, so that every negatively polarised clause is either generalised or specialised. The former is realised in English through the full lexical item *not* or its contracted version *n't*. The latter is realised through a number of lexical items that can be classified in parallel from two functional points of view. Thus, each of these items simultaneously realise a function in the nominal group and one in the clause. While the nominal group function can be either that of a Deictic6 or a Thing, the function of the negative polarity item in the clause is realised by either a participant or a circumstance. The intersection of these function types renders four possible options as a result, namely a Deictic in participation, a Deictic in circumstance, a Thing in participation and a Thing in circumstance. The whole array of possibilities offers five different choices, which are illustrated in examples (5) to (9), all taken from my corpus, wherein the negative polarity items are highlighted. Negative polarity items *not, none, no, nothing* and *never* respectively illustrate instances of generalised negation (5) and specialised negation (6–9). *None* in (6) is a case of Deictic in participation, *no* in (7) is a case of Deictic in circumstance, *nothing* in (8) is a case of Thing in participation and *never* in (9) is a case of Thing in circumstance.

(5) For the first time in this century, for the first time in perhaps all history, man does *not* have to invent a system by which to live. [Reagan 1989]


The typical test that can be applied in order to check the negative value of a clause is to add a question tag of reversed polarity. The corresponding question tags that could have been added in (5) to (9) are *does he?, can they?, will I?, will it?* and *did it?*, respectively.

As was said before, the probabilities of each of the alternative choices in a specific system are contextually constrained. One of the most important variables that influence lexigrammatical selections is that of register (Halliday and Matthiessen, 1999; Matthiessen, 2015). Register is defined as 'the configuration of semantic resources that a member of a culture typically associates with a situation type' and involves the contextual elements of field, tenor and mode (Halliday, 1975, p. 182). Field is concerned with what texts are about, tenor refers to the social roles of the users of language and mode denotes the channel of the communicative event. The field of the texts that constitute my corpus is the 44 inaugural speeches delivered by US Presidents. In them, each president informs American citizens–and more recently, a worldwide audience–of his priorities and goals while in office. The tenor is a monologic one through which a president-elect is addressing his audience. The mode is pre-planned oral communication.

SFL studies can be complemented with the aid of Corpus Linguistics, which is an empirical approach to the study of language. The principle behind Corpus Linguistics is that 'grammatical systems are probabilistic in nature' (Halliday and Matthiessen, 2014, p. 52). For example, in the system of clause types in (at least) western languages, clauses in the active voice are far more likely to occur than those in the passive voice. By the same token, finite and declarative clauses are more frequent than non-finite and interrogative ones, respectively. The frequencies with which these alternatives in a system occur in a register of a language can easily be measured with the aid of Corpus Linguistics. This renders the results obtained more objective and statistically validated (Bod, Hay and Jannedy, 2003).

This phenomenon was observed by Halliday some sixty years ago. Even before the early days of SFL, Halliday in his studies on Chinese made use of Corpus Linguistics (Halliday, 1959). What is more, the early theoretical underpinnings of the theory, which deal with the intonation of English, are based on the analyses of large corpora of authentic text (Halliday, 1963, 1967). Although some advances have taken place in the description and explanation of the systems of language (see e.g., Thompson and Hunston, 2006), there is still a wide gap to be filled. To begin with, Corpus Linguistics makes it easier to explore the lexical pole of the lexicogrammatical cline. For example, Stubbs (2006, p. 29) explores the psychological speaker's interpretation and speech acts encoded in Searle's terms (Searle, 1969) of lexical units such as *the naked eye* or *reach a ripe old age*.

However, the exploration of the grammatical pole of the lexicogrammatical cline proves a much more time-consuming task and requires a high level of manual analysis. A possible solution to this problem is to sacrifice the length of the corpus in order to obtain a complex, detailed grammatical analysis. This is the methodology used by Nesbitt and Plum (1988), wherein they intersect the systems of TAXIS and LOGICO-SEMANTIC RELATIONS in a corpus of 2,733 clause nexuses taken from interviews. An alternative solution is to concentrate on a theoretically limited system in terms of the array of choices available and investigate their frequencies in a large corpus. This is what Halliday and James (1993) opt for in their exploration of the systems of POLARITY and PRIMARY TENSE in the English finite clause in a corpus of 18 million words. A third solution is to resort to an archive of texts that belong in different registers and focus on a system that may be difficult to interpret. This is what Matthiessen (1999, 2006) performs in his study of the system of TRANSITIVITY in a corpus of 1.5 million words. The following Section explains the corpus and methodology I have used in this chapter.

## **3. Corpus and methodology**

My corpus is made up of 447 inaugural addresses delivered by US Presidents (122,848 words). This corpus is closed in character in the sense that it comprises all inaugural presidential speeches pronounced so far, from Washington 1789 to Trump 2017. Besides, while 14 presidents were re-elected, I have considered only the inaugural speech of their first term. The whole list of speeches with details of party, period, date, and word count is given in Table 6.2. Only the first president, George Washington, was unaffiliated to a political party. By the mid-nineteenth century the bipartisan system of government that currently prevails was well established. While 13 of the last 30 presidents succeeded a president of the same party, there have been nine alternations from Democratic to Republican presidents and eight changes from Republican to Democratic ones.

As can be seen, speeches are very dissimilar in length. While the shortest speech is 433 words long (Arthur 1881), the longest is 9231 words long (Johnson 1865). The mean of the whole group of data is 2792 words and the standard deviation is 1978 words. If the sample is reduced to the thirty addresses whose length lies around the mean, the mean and standard deviation of this sample is 1878 and 585 words, respectively. This renders the speeches more even in length and the results observed more comparable. Still, in order to make the speeches fully comparable, I have reduced the raw numbers of the results obtained to their frequencies, as is explained below.

I semi-automatically tagged all instances of negative polarity items with the aid of Wordsmith Tool (Scott 1998) and UAM CorpusTool (O'Donnell 2008). The former allowed me to identify all instances of negative polarity items such as: *not, no, none, neither, never, seldom*, and so on in the corresponding clauses in which they appear and to quantify the results obtained. The latter helped me identify and quantify functional and class features of the clauses and groups in which the abovementioned items appeared. I pasted all negative polarity items onto a spreadsheet file, wherein I added a number of traditional and functional features of these items and the number of finite clauses in each speech.

For the quantification process and for the sake of comparability between speeches, I obtained the frequencies of each negative polarity item found per 100 clauses. This was carried out by dividing the actual occurrences of negative polarity items by the number of finite clauses found in each speech. Later, I applied a number of filters in order to focus on specific features. Finally, I applied a chi-square test to verify the strong dependence of some of the different functional subsystems of the system network in this particular register. Figure 6.2 exhibits a small sample of the general results of five of the negative polarity items that are present in the first thirteen US Presidents' speeches, namely *no, not, n't, cannot* and *never*. Figure 6.3 illustrates a portion of the codification of every instance of three related lemmas found in the corpus, namely *not, n't* and *cannot*. This kind of codification allowed me to manually check the features of some specific items. The following Section analyses the results obtained.


**Table 6.2.** List of speeches.

*(Continued)*



† Died of a natural cause while in office.

†† Assassinated while in office.


**Figure 6.2.** Sample of general results.


**Figure 6.3.** Sample of codification of specific lemmas.

## **4. Results and discussion**

### 4.1. Overall results

Results show that in the specific register of American inaugural addresses under study, the frequency of negative polarity is much higher than that reported in the literature (Halliday and James, 1993; Matthiessen, 2006). I have found a mean of 25.8% of negatively polarised clauses, as shown in Figure 6.4. Thus, on average, US Presidents choose more than 1 out of 4 clauses in their inaugural speeches to be negatively polarised clauses. By contrast, Halliday and James (1993) find only 10% of clauses in an 18-million-word corpus of written text are negative polarity clauses, and Matthiessen (2006) obtains a frequency of 8.5% of negative polarity in interviews. A slightly higher frequency of negation is found in a corpus of 50,000-word written text (Tottie and Paradis, 1982; Tottie, 1991). Their findings reach a total of 12.8 negative items per 1000 words, which is equivalent to approximately 15 negative items per 100 clauses, i.e., 15% of clauses are negatively polarised clauses. However, they include affixal forms such as prefixes *in*- or *un*-, which constitute instances of morphological negation and are not considered in the sources abovementioned or in my study.

**Figure 6.4.** Frequencies of negative items per 100 clauses in speeches.

As is shown by the linear trend line in Figure 6.4, negative polarity increases chronologically over the 228-year time span covered in my corpus. Thus, later US Presidents tend to select polarised clauses in their inaugural addresses more overtly than their former counterparts. This discoursal strategy allows them to scaffold a more polarised discourse to craft their future policies in opposition to those of their predecessors in office (Hetherington and Weiler, 2009). This increasing level of polarisation may be the result of the characteristic weakness of bipartisan presidential regimes (Mainwaring, 1993).

However, once the results are reorganised from the lowest to the highest frequency rather than chronologically, as shown in Figure 6.5, further interesting trends are revealed. To begin with, four of the speeches that appear toward the lowest end in Figure 6.5 were delivered by presidents Truman, Tyler, L. B. Johnson and Taylor, who accessed the presidency after the death of a previous leader8 . Their inaugural addresses are not pronounced in opposition to the policies of their predecessors (Campbell and Jamieson, 2008). Truman's inaugural address, for example, occurs after the end of World War II and is thus mostly devoted to his foreign policies rather than his local ones. He pronounces the lemma *world* 24 times in his speech and 6 out of his 7 instances of *not* co-occur with this lemma in the same sentence (see example (10)). Thus, his address has a prospective character driven by the fear of a new world order based on communism rather than a retrospective one.

(10) Hundreds of millions of people all over the world now agree with us, that we need not have war – that we can have peace. [Truman 1949]

A Corpus Study of Grammatical Negation in US Presidents' Inaugural Speeches 157

**Figure 6.5.** Redistribution of frequencies of negative items per 100 clauses.

Tyler's speech, in turn, can fairly be considered an inaugural address, as President Tyler, rather than formulate his future policy, elaborates on the contingencies that surround his access to office and he pledges to follow the guiding principles established by the late President Harrison. A. Johnson and Fillmore, who succeed the assassinated President Lincoln and President Taylor respectively, pronounce inaugural addresses with a word count of 9231 and 8322 words, respectively, which makes them the longest in American history. Each of these two speeches triples the length of the average inaugural address9 , which lowers the frequency of negative polarity items in each considerably.

Additionally, within the least negatively polarised inaugural addresses are speeches by presidents affiliated to the same party as their predecessors and lie therefore as an endorsement to–rather than an opposition to–the policies of the previous term10. President Taft, for example, mentions *my (distinguished) predecessor* 7 times in his speech and 6 out of the 36 instances wherein he uses the word *not*, he does so in the context of a counterfactual condition that makes reference to his party affiliation, as in example (11).

(11) I should be untrue to myself, to my promises, and to the declarations of the party platform upon which I was elected to office, if I did not make the maintenance and enforcement of those reforms a most important feature of my administration. [Taft 1909]

In the same line, Republican President Hayes' words do not run counter to those of his previous leader, Republican President Grant. Rather, President Hayes devotes most of his speech to outlining his principles of reconstruction in the aftermath of the Civil War. It is precisely in this portion of his address that he mostly uses negation in the crafting of equality between rival factions, as is shown in example (12).

(12) […] my best efforts in behalf of a civil policy which will forever wipe out in our political affairs the color line and the distinction between North and South, to the end that we may have not merely a united North or a united South, but a united country. [Hayes 1877]

Finally, President Monroe also cherishes his immediate predecessor and his policy. So much so that 67% (28 out of 42) of the instances of grammatical negation that he exploits co-occur with other forms of negation, which renders this double negation a positive value (Osmankadić, 2015). For example, in (13), *we cannot fail* can be interpreted as *we will succeed*.

(13) If we persevere in the career in which we have advanced so far and in the path already traced, we cannot fail, under the favor of a gracious Providence, to attain the high destiny which seems to await us. [Monroe 1817]

Conversely, at the other end of the scale, the presidents who exploit a high degree of negative polarity clauses in their inaugural addresses frequently resort to this strategy in order to orchestrate a discourse that runs counter to that of their former leaders or to the prevailing principles of the opposing party11. Six out of ten of the presidents whose inaugural speeches exhibit the highest frequency level of negative polarity also signal an alternation with the party of the immediately previous president12. Thus, Presidents Bush Jr., Kennedy, Lincoln, Carter, L. B Johnson and Pierce, whose speeches reach the negative polarity levels of 47%, 43.7%, 40.9%, 39%, 37.3% and 35.5%, respectively, belong to the opposing party as their immediate predecessor.

For example, while President Bush Jr. overtly thanks both his adversary Al Gore and former President Clinton, he also makes a call for unification of the country (see example (14)) in an attempt to leave behind the long-disputed events of the recount of the Florida votes which won him the presidency. This rhetorical device of appealing to soften the effect of highly contested campaigns in the fight for the presidency has become an obligatory initial stage in the register of inaugural addresses (Campbell and Jamieson, 1990). After the seminal inaugural speech by President Kennedy (example (15)), all US Presidents appeal to a unified country13 that leaves aside party differences. However, only half choose negative polarity in this portion of their first speech as presidents.


By the same token, President Lincoln, whose presidency marks the institutionalisation of the Republican Party (see Table 6.2 above), revolutionises the American political scene with new anti-secessionist schemes. His inaugural address, which is regarded as probably one of the best ever delivered and is evoked by later presidents throughout American history, is mostly devoted to this new guiding principle. Negative polarity and contrasts are among the linguistic resources that make this speech so memorable. In it, *no*-negation is very frequently used to the point of reaching 10% of all negative polarity items in this speech, as is shown in example (16).

(16) I have no purpose, directly or indirectly, to interfere with the institution of slavery in the States where it exists. I believe I have no lawful right to do so, and I have no inclination to do so. [Lincoln 1861]

Additionally, negative polarity is masterly exploited in subjunctive and interrogative clauses to the highest level, as example (17) illustrates. Moreover, interrogative clauses are used 22 times in this speech, 8 of which are negatively polarised.

(17) In any law upon this subject ought not all the safeguards of liberty […] to be introduced, so that a free man be not in any case surrendered as a slave? And might it not be well at the same time to provide by law […]? [Lincoln 1861]

However, as Chilton (2004) argues, polarisation in political discourse operates in deictic domains that extend in more global rather than local time-spatial levels. Negative polarity is brought into discourse to construe experiential meanings and to enact interpersonal ones (Halliday, 1998, p. 27), which evoke the word of past leaders. In this way, US Presidents use this strategy both to portray themselves as the champions of nationally cherished values inherited from their honoured forbears and also to imply that their adversaries stand on the opposite side. Thus, the grammatical choice used by President Bush Jr. in example (18) brings to mind the frequently quoted words of President Kennedy shown in example (19).


Still, the aim of this work is not to trace the discourses that later speeches evoke or to find the sources of the metadiscursive instances14 present in them. Rather, my aim is to present the recurrent patterns characteristic of the register of inaugural speeches, which are an outstanding feature of this particular register. In the following subsection, I delineate my findings in the whole corpus and attempt an explanation of the results obtained.

### 4.2. Frequencies of negative polarity items

The occurrences of the most frequent negative polarity items in my corpus are shown in Figure 6.6. Other grammatical items such as *none, nobody, nowhere* and *seldom* are very infrequent and constitute less than 10% of cases when taken together. As is deduced from the results, *not*–negation, including the variants through the attached morpheme in *cannot* and the reduced form *n't*, reaches 56.25% of all instances. The second most recurrent negative polarity item, the negative determiner *no*, appears in all inaugurals with a frequency of only 26.42%. Thus negation at clause level through the adverb *not* in all its variants doubles negation at the level of the noun group through the determiner *no*.

This is in line with the findings in Biber et al. (1999, p. 170), wherein the ratio of *not*-negation to *no*-negation ranges from 2.1, 3.1 to 9.1 in the registers of news, academic/fiction, and conversation, respectively15. Thus, if I disregard conversation, which is comparatively different from all other registers (Biber et al., 1999, p. 12), it can be said that the register of inaugural political speeches can be regarded as similar to the other three registers in terms of the variable *not*/*no* ratio. This suggests that this feature is not necessarily characteristic of the register under study but a feature of the English language as a whole.

Still, there is great variation in the frequencies of *no*-negation that US Presidents use in their inaugural addresses, as is shown in Figure 6.7.

While in President Harding's speech this frequency reaches a peak of 14.8%, in Kennedy's inaugural there is not a single instance of *no*-negation. It is striking that whereas the former is considered one of the worst speeches in American history, the latter is regarded as probably one of the best crafted inaugural addresses. However, it is not necessarily the high or low frequency of determiner *no* that makes a speech a memorable text per se. Yet, while Harding's inaugural makes use of long sequences of negation in a row, as shown in example (20), example (21) illustrates a well crafted extract in which oppositions16 are intelligently exploited in Kennedy's inaugural.

A Corpus Study of Grammatical Negation in US Presidents' Inaugural Speeches 161

**Figure 6.6.** Occurrences of most frequent negative polarity items.

**Figure 6.7.** Frequency of no-negation in inaugural addresses.


As the frequencies of *not*-negation and *no*-negation taken together cover 83% of all instances in my corpus, I turn to a more thorough analysis of their uses in the following subsections.

### 4.3. Not-negation

The adverb *not* can have a wide or a narrow scope. In the former, considered in SFL an instance of clausal negation, the particle *not* affects the whole clause and can be paraphrased as *it is not the case that X*. In the latter, considered in SFL an instance of group negation, the lexeme *not* affects only a constituent of the clause. Example (22) exhibits an instance of clausal negation, while examples (23), (24), (25), (26) and (27) illustrate cases of group negation, wherein the particle *not* has scope over a prepositional phrase, an adverbial group, a noun group, an adjectival group and a pronominal group, respectively.


The distribution of the tokens of scope of negation is shown in Figure 6.8. As expected, the particle *not* has a wide scope over the whole clause more recurrently–681 out of 888 instances–than a narrow scope over all other constituents taken together.

**Figure 6.8.** Scope of negation.

I now turn to the intersection of clausal *not* with further grammatical features in the system of the clause. Results are exhibited in Figure 6.9, wherein negatively polarised clauses are intersected with six clausal systems in a parallel fashion.

**Figure 6.9.** Intersection of negative polarity and clausal systems.

Thus, clauses are tagged and quantified into paratactic or hypotactic, finite or non-finite, active or passive, non-modalised or modalised, declarative or interrogative, and in the present, past or future tense.

My findings show that most of these systems of the clause are unaffected by the choice of negative polarity. Thus declarative clauses (96%) prevail over interrogative ones (4%), finite (94%) over non-finite ones (6%) and active (83%) over passive ones (17%). These patterns of negatively polarised clauses are fairly similar to those in the overall pattern of English (Matthiessen, 2006), which suggests that the systems of MOOD, FINITENESS and VOICE are independent of POLARITY. Along with these highly skewed systems of the clause, the systems of TAXIS and TENSE are also unaffected by negative polarity, as paratactic clauses prevail over hypotactic ones (62% vs. 38%) and present over past and future (68%, 19% and 13%, respectively).

However, there is one clausal system that is noticeably affected by the choice of the negative alternative in the system of POLARITY, namely the system of MODALITY. That is to say, once negative polarity is chosen, the distribution of modalised clauses reaches an equiprobable level (Halliday, 1991b). This contrasts with the results in Biber et al. (1999, p. 486), who find that English clauses as a whole exhibit a highly skewed distribution towards non-modalised ones (83% non-modalised vs. 17% modalised). In a more specific 4,429,976-word corpus of speeches by US Presidents, non-modalised clauses are found to be the unmarked choice while modalised clauses reach the scant level of 5% (Ahrens, 1995).

Still, it is not only the choice of modalised clauses that is favoured by negatively polarised clauses but it is also the relative distribution of the modals chosen that is altered. Figure 6.10 illustrates a comparison of the frequencies of the eight most recurrent modals in the negatively polarised clauses with those in all clauses – negatively polarised or not – in my corpus.

**Figure 6.10.** Comparison of relative frequencies of 8 modals in negatively polarised clauses and total corpus.

As is shown in Figure 6.10, whereas the most recurrent modal in all clauses is *will* (29% of occurrences), the most frequent modal in negatively polarised clauses is *can* (38% of occurrences).

The distribution of modalised clauses in all my corpus is fairly similar to that reported in Biber et al. (1999, p. 486), wherein modal *will* is the most frequent (26% of occurrences). One of the meanings of this polysemous modal is to indicate the speaker's/writer's intention. This meaning is highly exploited by US Presidents in their inaugural addresses. Yet, as *cannot* frequently appears in the context of a positively polarised modal *will* in my corpus, it can be argued that there is strong interplay between these two modals, whereby politicians promise what they will do on the grounds of what cannot be the case, as is shown in example (28).

(28) It cannot be doubted that the proposed reductions will for the present diminish the revenues of the Department. [Fillmore 1850]

The strong interplay between the systems of POLARITY and MODALITY is statistically confirmed with a chi-square test. Once the findings of the two most frequent modals in my corpus, namely *can* and *will*, are intersected with those in the system of polarity, as depicted in Table 6.3, the value obtained for χ<sup>2</sup> is 59.46 at a level of significance 0.001. This entails that while the choice of negative polarity triggers the choice of modal *can* and precludes the choice of modal *will*, the choice of positive polarity triggers the choice of modal *will* and precludes the choice of modal *can*.


**Table 6.3.** Intersection of POLARITY and 2 modals.

A further intersection of the two most frequent alternatives of negation, through *not* and *no* with the system of PROCESS TYPE, is dealt with in the following subsection.

### 4.4. No-negation

As opposed to the type of negation analysed in the previous subsection, which affects primarily the whole clause, *no*-negation has always a narrow scope and is thus called local negation. In my corpus, the negative determiner *no* modifies fundamentally a noun with a frequency of 89%. The other uses of *no*-negation are as a modifier of an adverb 7% of the time and a modifier of a pronoun in 4% of the cases. The 10 most frequent nouns that collocate with *no* in my corpus are depicted in Figure 6.11, among which political organisations such as *nation, people, power* and *government* stand out. Although none of these appear as typical collocates in more general corpora (Biber et al., 1999, p. 173), their higher occurrence in my corpus of political speeches is not unexpected.

**Figure 6.11.** Most frequent noun collocations with *no*.

Additionally, within the most frequent collocates of *no* are the nouns *doubt* and *appeal*, which constitute instances of grammatical metaphor (Halliday, 1998). This linguistic strategy consists of a transcategorisation scheme more complex though not less frequent than the rhetorical device of lexical metaphor, which is so pervasive in political discourse (Chilton, 2004; Charteris-Black, 2005). Grammatical metaphor and lexical metaphor operate at the two distant poles of the cline of lexicogrammar. Therefore, there is a clear parallelism between Chilton's three strategic functions of language in politics—coercion, (de)ligitimisation and (mis)representation—and Halliday's three metafunctions of language – experiential, interpersonal and textual (Semino, 2008, p. 86).

Grammatical metaphor allows users of a language to reconstrue experiences through incongruent linguistic realisations. For example, an event, which is congruently expressed through a clause in language, can have a more incongruent realisation through a noun. This is what happens in example (29a), whereby President A. Johnson depersonifies the *appeal to force* so as to avoid taking or assigning responsibility for an action that is controversial. The unpacking of the grammatical metaphor (Halliday, 1991b) through a more congruent realisation of (29a) is offered in (29b).

	- b. […] what happened in the last four years The Civil War, we hope that we will never appeal to force again. / we hope that we will never go to war again.

As *no*-negation frequently appears in the context of grammatical metaphors, this particular register is likely to exhibit a higher level of the verbal processes that are characteristic of expounding registers such as academic discourse, in which grammatical metaphor is very frequent (Matthiessen, 2015). This renders the contrast between the processes used in *no*-negation and those in *not*-negation worthy of analysis. Results are shown in Figure 6.12.

**Figure 6.12.** Comparison between relative frequencies of processes under notnegation and under *no*-negation.

A number of points are worth mentioning here. To begin with, in the particular register under study, material processes are by far the most frequent processes in negatively polarised clauses, either through *not*-negation (365 tokens = 54%) or through *no*-negation (167 tokens = 39%). While these figures are not fully consistent with those found in the literature for the overall system of English, in the register of political discourse, results tend to converge more closely. For example, it has been found that material processes are relatively less frequent in negatively polarised clauses than in their positively polarised counterparts (26% vs. 34%, respectively) (Matthiessen, 2006, pp. 126–128). Besides, the frequency of material processes has been found to be very sensitive to the register type, ranging from 32% to 48% across different registers (Matthiessen, 2015).

More specifically, in the study of political speeches, material processes seem to be more recurrent than in other registers (Durán, 2008; Kazemian and Hashemi, 2014; Adjei, Ewusi-Mensah and Okoh, 2015; Adjei and Ewusi-Mensah, 2016), wherein material processes range from 39% to 59%. US Presidents seem to choose this higher frequency of material processes because they feel the need to portray themselves as leaders of action, particularly in their early stages in office (Durán, 2008; Wang, 2010). Moreover, it is through the excessive use of material processes that US Presidents project an image of themselves as both national and world leaders (Kuosmanen, 2015).

As for the comparison between the frequencies of PROCESS TYPES under *not*-negation and *no*-negation, the most striking difference is the higher level of existential processes that co-occur with *no*-negation. While the former reaches the scant level of 1%, which is in line with the figures in Halliday and Matthiessen (2014, p. 308), the latter considerably increases up to 13%. This difference is proved to be statistically significant through a chi-square test (χ<sup>2</sup> = 86.65). This implies that at the level of significance 0.001, while the choice of *not*-negation favours material processes over existential ones, the choice of *no*-negation favours existential processes and disfavours material ones.

Relational processes also increase, though more slightly, from 30% in the scope of *not*-negation to 34% in the scope of *no*-negation. These two types of processes allow for a higher degree of grammatical nominalisation, as is shown in example (30a), which exhibits an instance of an existential process and (31a), which illustrates the use of a relational one. Both of them have more congruent realisations in their (b) counterparts.


The increase of the relative frequencies of both existential and relational processes in the case of *no*-negation is carried along with a decrease of the frequency of material processes. It is worth noting that the more congruent realisations expressed in the (b) counterparts above involve the use of the material processes *invade* and *use force* in (30) and *intend* and *expand* in (31).

The use of *no*-negation with existential and material processes allows US Presidents to exploit higher levels of grammatical metaphor. With this grammatical device, users of the language resort to a higher level of deagentivisation and to a very dense packing of verbal complements into Things that can be modified in a complex manner (Halliday, 1991b). Thus, for example in (30a), President Lincoln does not express *who will not invade who* or *who* the Agent of *use the force* is. Similarly, in (31a), President Hoover does not make it overt *who will not expand their territories*, as the more congruent realisation in (31b) shows through the unpacking of the grammatical metaphor used in its original counterpart in (31a). Additionally, the Complements of the processes *there be* in (30) and *have* in (31) are very dense nominals that are treated as abstract entities whose Heads are *using* and *desire*, respectively, which are postmodified by long prepositional phrases in both cases. These nominals are deemed objective entities in the world of reality as presidents construct their discourse persuasively.

## **5. Conclusion**

As Charteris-Black (2005) argues, the more democratic a society, the more effective its politicians' persuasive strategies need to be. US Presidents, who champion themselves as democratic leaders, find the need to exploit linguistic strategies in their inaugural addresses to mark a memorable new beginning (Atkinson, 1984). While this is carried out through the careful selection of lexical items, it is even more successfully achieved through the exploitation of the grammatical pole of the lexicogrammatical cline (Halliday and Matthiessen, 2014, p. 64). One such linguistic strategy, as I have argued in this chapter, is the increasing choice of grammatical negation in their speeches.

This chapter has analysed the frequency of negative polarity clauses in a closed set of inaugural addresses by US Presidents. It has been shown that US Presidents choose a comparatively higher degree of negative polarity than that found in the overall pattern of English. What is more, not only is the selection of negative polarity a systematic feature of their inaugurals but this register also exhibits an increasing trend in the chronology of US inaugural speeches. Thus, later presidents find a higher need for a choice of negation in their inaugurals than their earlier counterparts.

Additionally, we have seen that presidents that assume office after the death of an immediate leader and those who succeed a former co-partisan president tend to choose a lower level of negation in their speeches as they do not need to distance themselves from the policy of the previous term. On the other hand, the presidents who alternate party with their predecessor tend to express themselves in a more polar way as a token of the veering course they intend to apply to American history. Thus, the use of negative polarity in inaugural addresses tends to express the speaker's promise of distance from a previous course of action.

A further result has been the fact that clausal negation through the adverb *not* is twice as frequent as the more local type of negation realised by negative determiner *no*. This latter type of negation allows politicians to fully exploit the linguistic device of grammatical metaphor, through which they can avoid the mention of the Agent under crucial circumstances. Besides, with this strategy, politicians produce a discourse with an objective rendering typical of academic registers. Finally, the higher degree of *not*-negation in this register favours the use of modal *can* and disfavours the use of modal *will*. By the same token, *no*-negation increases the frequency of existential processes to the detriment of material ones.

From the methodological point of view, my intention in this chapter has been to strike a balance between linguistic analysis and an explanation of the grammatical choices in context (Chilton, 2003, p. 411). While it is a time-consuming task to analyse the context in which every instance of negative polarity is produced in a 228-year corpus, in this chapter I have reported tendencies of the negation frequencies found. This is what Matthiessen (2006) calls to distance from the pole of instantiation in registerial analysis and advance towards the systemic end of the cline of language. By reporting frequencies of negation in US presidential inaugural addresses, I have offered a deeper insight into the characterisation of this register. Further lines of exploration can be the analysis of other systems in the register of inaugural addresses or the comparison of the system of negation in other registers.

## **Endnotes**

1. Grammatical negation is what Tottie (1991) calls non-affixal negation as in *This is not possible* or *There is no possibility*, as opposed to the affixal type of negation as in *This is impossible*. Alternative labels are, respectively syntactic and morphological negation (Hamawand 2009). The latter type is also known as nexal negation (Jespersen 1917). I have also considered in this study the types of incomplete negation such as *hardly, barely* and *seldom* (Jespersen 1917), although they are very infrequent in my corpus.

2. Whereas patterns of binary oppositions are mentioned in the literature (Atkinson 1984, Chilton 2004), frequencies of these patterns are still not reported.

3. The notion of function in SFL is further elaborated in Halliday (1984).

4. Two opposing but complementary viewpoints of language in SFL are language as system and language as instance (Halliday and Matthiessen 2014, 27).

5. Here I follow the convention within systemic functional linguistics, according to which lexicogrammatical systems are identified in capital letters and functions are symbolised with their initial in capitals.

6. In SFL, functions are capitalised by convention. For example, for the definition of Deictic in SFL, see Halliday and Matthiessen (2014, p. 368).

7. As can be seen in Table 6.2, the actual number of speeches is 44, as President Cleveland assumes the presidency in two non-consecutive periods.

8. None of the 8 presidents that accessed the presidency after the death of their predecessor delivered an actual inaugural address although they did address the Congress thereafter. That is why four of the speeches considered for the analysis here – those by T. Roosevelt, Coolidge, Truman and L. B. Johnson – are those given at the point of accessing the presidency through election, which in actual fact is considered their second inaugural (see Table 6.2).

9. For example, Pres. Lincoln includes in his inaugural address a long quote of an article of the US Constitution on slavery.

10. This is probably the reason why the 13 presidents who serve a second term resort to a comparatively lower frequency of negative polarity in their second inaugural address. However, for lack of room, I have left 2nd inaugural addresses unanalysed here.

11. See for example, some examples of contestive inaugural addresses in Ryan (1993, p. xviii).

12. The frequency of negation is even higher in presidential campaigns than that in inaugural addresses (Lau and Rovner 2009).

13. This appeal to unifying the country appears at least since T. Roosevelt delivers his inaugural in 1905. However, there is no mention of the opposing party until the speech by Kennedy.

14. The literature in this regard is abundant. See, for example, Clarke (2004) for the metadiscursive references in Kennedy's inaugural speech. The construction of rhetorical shifts and their historical connections are explored in detail in Widmaier (2015).

15. See also the results in Tottie (1991).

16. Well-crafted oppositions are among the most effective linguistic resources in a politician's speech that arise a resounding 'claptrap' in their audiences (Atkinson 1984).

## **References**


## **PART THREE: MEANING-TO-FORM NEGATIVES**

## **7. Counterfactuality as Negative Meaning: A Case Study of** *BE Supposed To*

*Anne-Laure Besnard*

## **1. Introduction**

The aim of this paper is to contribute to the reflection on how markers that are not typically negative may generate negative interpretations in context. More specifically, it will focus on the counterfactual, which can be considered a type of negative meaning insofar as it involves the expression of something that is understood not to be the case. In other words, drawing from Culioli's analysis of negation (1990, pp.91–113), the counterfactual can be characterised as a form of implicit negation which relies on the representation of at least two possible values for a given predicative relation: the positive one (namely *p*), and its complementary *p'* (which can either be *non-p*, i.e. void of *p*, or *other-than-p*). In the absence of explicit cues such as the conditional marker *if*, the interpretation of an utterance as counterfactual appears to be the result of complex interactions. Crucially, a counterfactual reading arises when the value selected on the surface for the propositional content conflicts with facts established elsewhere in discourse. It thus often seems to be a matter of pragmatics rather than semantics, and yet some linguistic expressions seem to trigger, or at least to be associated with, counterfactual meaning more often than others.

To explore this issue, the present study will focus on the structure *BE supposed to* which is one such expression and contrasts in this respect with otherwise related structures like *BE expected to* or *BE believed to*. After providing an overview of the uses of the structure and of the ways it has been described in the literature, I will propose an alternative analysis within the framework of the Theory of Predicative and Enunciative Operations (Culioli) to try to explain why *BE supposed* 

**How to cite this book chapter:**

Besnard, A.-L. 2023. Counterfactuality as Negative Meaning: A Case Study of *BE Supposed To*. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 179–201. Stockholm: Stockholm University Press. DOI: https://doi.org /10.16993/bcd.g. License: CC BY 4.0.

*to* favours a counterfactual reading. I will conclude with an exploration of the contextual factors that play a role in the development or blocking of the negative or counterfactual reading. The corpus used for this study is the British newspaper *The Independent* (2009)1 . Representative examples were selected for qualitative analysis using the concordancer *CasualConc*<sup>2</sup> .

## **1. Uses of** *BE supposed to*

### 1.1. Overview of the literature

In the literature, including reference grammars of English such as Quirk et al. (1985) and Biber et al. (1999), *BE supposed to* is often identified as a quasi-modal3 , that is to say as a periphrastic expression that shares semantic characteristics with the modal auxiliaries (especially *should*), but lacks the syntactic properties typical of auxiliary verbs. Because of this intermediate, not quite grammatical status and the lack of straightforward relation between the (quasi)modal construction and the passive of the verb SUPPOSE, most studies4 have focused on the emergence of the different modal meanings of *BE supposed to* from a diachronic perspective. This is indeed a complex question, given that it poses a challenge to the grammaticalisation principle—according to which epistemic meanings normally derive from deontic meanings, and not the reverse. This is outside the scope of this paper so I will not comment on this particular issue, but I will present the different values of *BE supposed to* that linguists usually agree upon.

There is a general lack of consensus on how many values there are, how they should be categorised and how the terms often used to describe them—namely, *evidential, deontic* and *epistemic*—should be understood. However, previous studies agree on the fact that at least two main uses should be distinguished, which I will refer to as the evidential use and the deontic use, both illustrated below5 .

(1) Military Scenarios: The three options // […] 2. 'McChrystal-Lite': US sends in 30,000 // This **is supposed to** be the preferred option of the US Defence Secretary Robert Gates. Mr Gates is said to have had initial doubts about large-scale Afghan reinforcements but he has been won over by the military commanders. However Mr Gates is also said to feel that not enough has been done on the border to counter al-Qa'ida and that sending a larger number of troops would mean that the counter-terrorism aspects of the mission can be addressed as well. *(In2009)*

(2) Dean Whitehead's miscued header found Spurs substitute David Bentley and his prod sent Darren Bent down the Sunderland left. There was space there because, as aggrieved Sunderland manager Ricky Sbragia noted dryly, left-back George McCartney was in the Tottenham box trying to score. // 'I don't know why he's there, George **is supposed to** defend,' Sbragia said. *(In2009)*

In (1), the interpretation given to *BE supposed to* can be traced to the meaning of the verb SUPPOSE and is rather close to that of *BE believed to*, which is a common paraphrase for the construction in this type of use, or even *BE said to*, which occurs twice in the right co-text: it is used as a quotative evidential marker to indicate that the source of the propositional content of the utterance is not the speaker (the journalist), but someone else whose identity remains indeterminate. The utterance can thus be contrasted with (1') *this is the preferred option of the US Defence Secretary Robert Gates*, which provides similar information, but does not attribute it to a source other than the journalist. In (1'), the absence of evidential marking also makes the utterance more assertive, which is why *BE supposed to* is sometimes characterised as having epistemic meaning in utterances similar to (1). Nevertheless, primarily identifying such occurrences as *evidential*, i.e. as encoding information source as opposed to speaker attitude (Aikhenvald, 2006), seems more satisfactory since the possible epistemic interpretation is context-sensitive and, most importantly, stems from the evidential value of the construction6 .

In (2) on the other hand, *BE supposed to* has a deontic value. A possible paraphrase of the utterance is *George should defend*, meaning that George's role is to defend; it is what is expected of him, that is, he is under some form of obligation to do so. This type of interpretation is far less typical of what Noël (2008) refers to as 'NCI constructions' (from Latin *nominativus cum infinitivo*), i.e. constructions like *BE said to, BE believed to, BE thought to, BE reported to*, and is actually only shared by *BE expected to*, which could also serve as a paraphrase in (2) and has a lot in common with *BE supposed to* both from a present-day and diachronic point of view (see Noël and van der Auwera, 2009, pp.618–620).

There seems to be a clear distinction between these two uses, which lend themselves to different paraphrases—*is believed to* in (1), *is expected to* in (2)—but there is actually a degree of overlap between them. If we consider example (3):

(3) Thursday's elections may have become a de facto referendum on our imploding government but what people are actually doing is electing some local councillors and, much more substantially, a new European parliament. // Europe **is supposed to** be important to us. A lot of our exports go there. It imposes a lot of regulations on us. And we pay a lot of money to remain members of the club — we are the second largest net contributor after Germany. But we seem hardly to care at all. If you don't believe that, try these two questions. Who is the Europe minister and who replaced Peter Mandelson as the British EU commissioner in charge of trade? *(In2009)*

we can see that *BE supposed to* does not fall neatly into either one of these two types of uses. On the one hand, it carries the implication that *it is generally considered to be the case (that Europe is important to us)*, which gives it an evidential dimension. On the other hand, it implies that *it can't be any other way (given that a lot of our exports go there, etc.)*, so that there is a necessity for <Europe-be important to us> to be the case, which means that this use of the structure also has a lot in common with the deontic interpretation (although the intersubjective dimension is clearly missing).

Interestingly, in this type of context, *BE supposed to* cannot be replaced with *BE believed to*, or *BE expected to*. Because previous studies were more interested in the degree and paths of grammaticalisation of *BE supposed to*, they do not account for this lack of interchangeability with other NCI constructions. My hypothesis is that substitution is impossible because *BE supposed to* is in fact notionally7 very different from *BE expected to* and *BE believed to* in that it has a very strong counterfactual potential, which is needed here in light of the argumentative structure of the passage; that is, *BE supposed to* allows the speaker to signal the existence of conflicting representations and in doing so, it prefigures the adversative *but*, while *BE expected to* or *BE believed to* could do no such thing.

This hypothesis will be further investigated in Section 2. The rest of this Section will analyse the general pattern of use of *BE supposed to* in the *Independent* 2009 corpus to see how common the counterfactual interpretation actually is.

### 1.2. Distribution of *BE supposed to* and counterfactuality

In line with example (3), Visconti (2004, p.184) notes the "overwhelming presence of counterfactual signals in the context of use of *be supposed to* from Modern to Present Day English, such as *but, in fact, in reality*". According to Visconti (2004, p.185), "[t]hese contexts, in which *be supposed to* evokes a possible world, a state of affairs which would be expected to occur but does not, represent the most frequent ones in which the construction is used in Present Day English". Visconti continues by associating this counterfactual interpretation with "the overwhelming use of the past form of the construction in Present Day English".

Indeed, the following examples show that the association of the past tense with *BE supposed to* is likely to trigger a counterfactual interpretation.


This is especially noteworthy considering that past tense occurrences of *BE supposed to* represent about 48 % of all its uses in the *In2009* corpus, which is a remarkable proportion compared to other NCI or semantically-related *BE Adj TO* constructions (see Figure 7.1):

#### 184 Negatives and Meaning: Social Setting and Pragmatic Effects

**Figure 7.1.** Distribution of BE -EN/Adj TO structures in *The Independent* (2009).

Among the constructions listed above, *BE supposed to* is not the only construction prone to be interpreted as counterfactual in the past tense, as shown by examples (9) and (10):


However, the fact that it occurs so frequently in past tense contexts suggests a stronger counterfactual potential than for *BE expected to* or *BE due to*—which is in line with my earlier hypothesis. Moreover, with *BE supposed to*, the counterfactual interpretation is also very frequent in the present tense, which differentiates this construction from other structures.

What explains the tendency for the counterfactual reading to arise with *BE supposed to* as opposed to other quasi-modals? This is addressed in Section 2, which provides an enunciative analysis of the construction.

## **2. Schematic form of** *BE supposed to*

### 2.1. Pragmatics and semantics within Culioli's TPEO

Within Culioli's Theory of Predicative and Enunciative Operations, all linguistic expressions are considered to be markers of invariant cognitive operations with some in-built potential for variation that is activated by the interactions with the co(n)text within a particular discourse situation.

One very important difference between the theory and many other approaches is the idea that pragmatic potentials are in a large measure built into the semantics of linguistic items—provided one recognises that fully-fledged meaning emerges only at the end of a complex process of configuration relative to context and situation. (Ranger 2018, pp.v–vi)

From this perspective, *BE supposed to* (just like *BE expected to* or *BE believed to*) can be viewed as a complex marker whose counterfactual pragmatic potential is the result of interactions between the markers BE, SUPPOSE, -EN (past participle) and TO. The next part of this paper focuses more specifically on -EN and SUPPOSE, which are most central to the argument.

### 2.2. *BE supposed to* as a dissociative marker

### *2.2.1. Dissociation of the assertive source from the original speaker*

When combined with the identification operator BE8 , the past participle marker -EN entails a passive reading of the construction. As SUPPOSE is a cognition predicate, the passive allows the original speaker *S*<sup>0</sup> (the one producing the *BE supposed to* utterance) to dissociate him/herself from the assertive or modal source S1 , i.e. the source of the 'supposing'9 . This forms the basis of the quotative evidential reading, which can clearly be traced to the passive origin of the construction.

In line with grammaticalisation studies, however, one might object to the characterisation of some uses of *BE supposed to* as passive, such as the one given in (2) *George is supposed to defend*, since deontic occurrences cannot be rephrased in the active (\**X supposes George to defend*) or be specified by an agent *by*-phrase (\**George is supposed by X to defend*). Although they probably *were* passive originally (see Noël and van der Auwera, 2009), they can no longer be described as such in present-day English because the verb SUPPOSE is no longer found with the 'intend' meaning thought to have given rise to the deontic use of *BE supposed to*. Yet, just like evidential occurrences, they also imply an underlying evaluation process whose agent, or rather, experiencer, is not identified, thus creating distance between the speaker and the assertive source. This is why *BE supposed to* is typically characterised as 'objective', that is, as involving a discourse-external source—as opposed to *should* for instance (see Verhulst et al., 2013).

This can be shown to be the case in many occurrences, such as the following, where some assertive source other than the speaker can be retrieved from the context:


In (11), which can be characterised as an evidential use of *BE supposed to*, the 'rumours' mentioned in the right co-text suggest the speaker is not the source of the supposing. In (12), where the structure could be paraphrased as *BE intended to* with an interpretation verging on the deontic, the modal source is likely to be Google, the initiator of the project alluded to in the context. In (13), where we have a more clearly deontic meaning, the distancing effect is strengthened by the quotation marks in the left co-text, and the rhetorical question in the right co-text which suggests that the speaker disagrees with <we-applaud> being desirable.

More generally, the absence of identification between the speaker and the assertive source increases the distance between the speaker and the propositional content of the utterance, which the speaker does not take responsibility for, and thus, might be seen as a factor contributing to the counterfactual interpretation of *BE supposed to*. This is suggested by Visconti (2004, p.185):

The inference arises as follows: if the Speaker/Writer chooses to use the *be supposed to* construction (s)he evokes an unspecified source of belief/expectation, which is distinct from her/himself. This choice invites the inference that the Speaker does not identify with the source of evaluation, and also that the Speaker signals a distance between the expected world and the 'real' world[.]

However, this is not a specificity of the marker, since we also find this dissociation between speaker and assertive source with other pseudo-passive structures, such as the ones mentioned before. These constructions may lend themselves to an 'uncertain' epistemic reading (*BE believed to* in example (14)) or a deontic reading (*BE expected to* in example (15)) but do not carry any implication of counterfactuality:


As outlined in Section 3, context plays a role here since *BE supposed to* could have been used in place of *BE believed to* in (14), and *BE expected to* in (15), without triggering a counterfactual reading.


In other words, *BE supposed to* does not automatically imply the selection of *p'* over *p*—this is explained further in Section 3—but, as noted in Section 1.2, *BE supposed to* tends to appear in counterfactual contexts much more frequently than *BE believed to* or, perhaps more surprisingly, *BE expected to*. This means that the counterfactual interpretation cannot solely be accounted for by the distancing effect resulting from the dissociation between speaker and assertive source.

### *2.2.2. Dissociation of the propositional content from the reference situation*

To better understand what is special about *BE supposed to*, we need to go over the notional properties of SUPPOSE, from which it is derived. An examination of the varied uses of the verb recorded by the *Oxford English Dictionary* (*OED*) in present-day English suggests a core theoretical or hypothetical dimension10, also found in now obsolete uses. The etymology of SUPPOSE provides at least a partial explanation: according to the *OED*, SUPPOSE comes from the Anglo-Norman *su(p)poser*, which is inherited from the Latin *suppōnere* meaning 'set/put below'. This spatial origin suggests that the 'supposed' state of affairs (SoA) belongs to another reality plane than the 'actual' state of affairs (Figure 7.2):

**Figure 7.2.** Spatial representation of SUPPOSE.

In linguistic terms, this means that the propositional content associated with SUPPOSE is not held to be valid in the reference situation where the supposing takes place but is the object of a fictive assertion, insofar as that propositional content is presented as independent from the reference situation and its associated plane of events and merely serves as a premise (i.e., as the basis of a line of argument or reasoning). This analysis is supported by the fact that the verb SUPPOSE has a few 'special grammatical uses' (*OED*)—for instance in the imperative—where it serves to introduce a hypothesis:

**(16) Suppose** a dozen qualified electricians had warned you that the wiring in your house was dangerously faulty. Would you listen instead to some bloke down the pub who told you that it was all a scam, and that house fires weren't caused by faulty wiring but by sunspots? // So why do so many people ignore the experts when it comes to climate change? *(In2009)*

In (16), *suppose* constructs an imaginary world in which the predicative relation <a dozen qualified electricians-warn you that the wiring in your house is faulty> is the case. The speaker asks the co-speaker to select the *p* value for that predicative relation in a fictive situation of events and uses this as a starting point for a hypothetical question (*would you…*). The answer to this hypothetical question is meant to sustain the argument that follows: that people should really listen to the experts on climate change. By creating space between the situation of utterance (which serves as the reference situation here) and the state of affairs evoked in the complement clause it introduces, *suppose* allows the speaker to make a point in the abstract.

This 'in the abstract' component appears to be fundamental for *BE supposed to* which can often be rephrased as, or seen to function harmonically with, *in theory*:

(2') *in theory, George is supposed to defend / defends / plays defense*

(3') *in theory, Europe is supposed to be / is important to us*

(12') *in theory, the service is supposed to / will/shall make it easier for readers to follow evolving news stories*

See also the following authentic example:

(17) In theory, the training is reasonably thorough. After an initial 12 week basic course in Kabul, recruits **are supposed to** receive an intense three week session before they deploy to the front. But the realities of combat mean that this has been reduced to three days before the troops are sent out to fight. *(In2009)*

EXPECT, on the other hand, does not evoke an alternative—hypothetical or theoretical—state of affairs. As shown in Besnard (2017), it is positively-oriented and involves a projection of the validation of the predicative relation awaiting confirmation. In other words, with EXPECT (from *ex(s)pectāre* 'look forward to, wait for', *OED*) and *BE expected to, p* is selected on the same plane as the reference situation in anticipation of the actual validation of the predicative relation. This is why the predicate introduced by *BE expected to* often has future-time reference and there is typically no implication that the actualisation of the state of affairs is doubtful—see again (15) *Between now and Christmas, I am expected to do two essays, two short presentations and a book review*.

To summarise, the consequence of the hypothetical or theoretical character of SUPPOSE and the lack of speaker-commitment attributable to the passive origin of *BE supposed to* is that there is a double distance between the (speaker-related) situation of utterance and the state of affairs evoked by the propositional content of the *BE supposed to* utterance. This appears to be a strong basis for the development of the counterfactual reading. To illustrate with example (3) *Europe is supposed to be important to us*, which is quite typical of the use of *BE supposed to* in journalistic texts: the selection of the positive value *p* for <Europe-be important to us> is carried out on a theoretical (i.e., not empirical) plane and is attributed to an assertive source which is not identified to the speaker. Thus, in the end, there is no commitment to the validity of the propositional content by any subjective source whatsoever, even though, because the proposition functions as a premise, the positive value is normally represented as desirable. Furthermore, by explicitly marking the validity of the propositional content as theoretical, *BE supposed to* suggests that the reality (which is usually what a non-conditional utterance is about) is either entirely or somewhat different. As a marker of non-conformity, *BE supposed to* thus sets up two independent levels of representations—an explicit theoretical one, and another one, more directly related to the situation of utterance, which can be inferred from the context. In doing so, *BE supposed to* opens up space for contradiction and is as a consequence particularly prone to appear in argumentative contexts.

As opposed to *BE expected to, BE supposed to* thus favours a counterfactual, i.e., negative, reading because there is a potential in-built gap between the theoretical situation of events set up by the structure and the actual situation of events. However, this gap is underspecified, only potential, so that contextual elements are needed to activate the counterfactual reading. This allows for a variety of uses in discourse. These are examined in the next Section, which focuses on present-tense occurrences of the quasi-modal.

## **3. Discourse functions of present-tense** *BE supposed to*

### 3.1. Highlighting contradictions and passing comment on the state of affairs

Very often, *BE supposed to* occurs in contexts where the selection of *p* for the proposition within its scope appears, on some level, contradictory to the actual situation of events referred to in the surrounding co-text. As suggested earlier, this is likely to give rise to a counterfactual interpretation, as in (18) and (19):


In these two examples, the contradiction is quite explicit and complete. In (18), the speaker rejects *the idea that I am supposed to be cringingly embarrassed* as nonsensical by means of the negative evaluative marker *(such) rubbish*, so that <I-be cringingly embarrassed> is negated, with the selection of the complementary value *p*', and more precisely *non-p*, instead of *p*: <I-be (cringingly) embarrassed> is *not* the case *at all* vs. complementary <I-be proud> (logical negation inferred primarily on the basis of the negative attitude to the proposition expressed by the speaker in the co-text). In (19), the number 20 is opposed to 22, with the adverb *actually* functioning as a direct counterpoint to the 'theoretical' character of *BE supposed to*. Again, *p*' is selected instead of *p* for the predicative relation at stake <20-be the number of countries taking part> (which is *not* the case), but this time it is contextually defined as *other-than-p*, with the following sentence providing the right value for the first argument (logical negation inferred on the basis of a contradictory proposition presented as fact in the co-text).

Most of the time, however, the contradiction is only partial. *BE supposed to* often occurs with generic subjects in argumentative sequences where the theory is contrasted with specific facts typically introduced by adversative markers like *but*. Examples (20) to (25) illustrate this type of use:


Apart from example (20) where *data protection regulation* is uncountable—so that it is the spatio-temporal verification of the predicative relation as a whole that is at stake (*seldom* = *not always* and even *not often*)—the theory is that the predication that falls under the scope of *BE supposed to* is valid for all *faces and car numberplates* (21) / *war veterans* (22) / *GPs* (23) / *managers* (24) / *books* (25), while the right co-text indicates that there is at least one occurrence of the notion corresponding to the subject that does not validate the predicate as it is *supposed* to:


This does not, however, necessarily mean that the resulting reading is counterfactual, i.e. that *p'*, be it *non-p* or *other-than-p*, is selected over *p*. More specifically, the point of (24) or (25) is not to deconstruct the *BE supposed to* statement—i.e. the statement that, in theory, the role of a manager is to encourage you (24) or that, in theory, the purpose of books is to inform or entertain (25)—but rather to point out that the particular occurrences of /manager/ (Flavio Briatore) and /book/ (*Fame*) mentioned in the co-text are not *good* occurrences, in that they do not have the expected properties normally associated with the corresponding notions. Thus, in these utterances, *BE supposed to* allows the speaker to make a negative evaluative comment. This is also the case in another common type of sequence illustrated below:


In (26) to (28), even with minimal context, it is quite clear that the validation of the proposition that is under *BE supposed to*'s scope is seen as desirable by the speaker, while being presented as counterfactual in the specific context of the events alluded to in the co-text—with the *not* complementary predication corresponding to *p'* being the actualised value.

What all the examples above have in common is that the theory is used as a starting point, to be contradicted only later by means of explicitly negative or adversative markers, so that *BE supposed to* can be seen to anticipate a problematic mismatch between theory and reality. The same type of mechanism seems to operate when the structure is used in subordinate relative clauses to attribute a property to the subject, which is then questioned in the main clause:


In (29), the notional contradiction between the idea of a *gold standard*  and *sub-standard care* is obvious. In (30), the use of an embedded *BE supposed to* predication instead of a simple noun phrase such as *his friends* shows that the attitude of the people referred to conflicts with the notional properties normally associated with the word *friend*, which makes the qualification of those people as friends doubtful. This does not exactly yield a counterfactual reading, but still orients towards the non-validation of the predicative relations <flagship institutions of NHS-set gold standard for medical treatment> in (29), and <they-be his friends> in (30). However, it is worth noting that the occurrence of *BE supposed to* in subordinate nominal or adjectival clauses does not always favour a negatively-oriented interpretation, partly because what is predicated within these types of subordinate clauses is often not at issue but pre-constructed which still shows the *a priori* status of the *BE supposed to* predication.

The following Sections explore unambiguously non-counterfactual uses of *BE supposed to*, starting with occurrences of the structure in subordinate clauses which are shown to match the reality in the immediate co-text.

### 3.2. Matching facts with theory

Example (31) below illustrates the compatibility between *BE supposed to* and an actualised proposition:

(31) 'I am ahead of schedule. The physios are really happy with me and how I feel at the moment is how I **am supposed to** be feeling. There is obviously a bit of muscle wastage in the legs so I am working on that but I am pleased.' *(In2009)*

Yet, it is precisely the role of *BE supposed to* to set up a theoretical level of representation so that facts can be shown to concord with the theory; it does not follow that *BE supposed to* can itself be factual plus, its occurring within a *wh*- clause means that there is no assertion of the theory, which is not then considered for itself as the basis of a line of reasoning but is simply used to characterise a state of affairs as consistent with what might be expected.

### 3.3. Providing justification for a preceding statement

This matching of theory and facts can also be seen to operate to some extent in another common use of the structure where no counterfactual value can be construed, illustrated in (32) and (33):

(32) Thousands of people lined up outside a planetarium in Patna on Tuesday to buy solar viewing goggles. The goggles, costing 20 rupees (40 cents), **are supposed to** act as filters and allow people to look at the sun without damaging their eyes. // But millions across India were shunning the sight and planned to stay indoors, gripped by fearful myths. *(In2009)*

(33) Dream trip? // I would love to go to the Galpagos Islands. The landscape **is supposed to** be quite incredible, and I'm fascinated by the wealth of nature. I can just imagine chartering a boat with a group of friends and sailing around the islands. *(In2009)*

These two occurrences show that *BE supposed to* utterances may fulfil an explanatory function, without any reference to the actualisation of the proposition under the scope of the structure in the surrounding co-text. In (32), where the *BE supposed to* predication specifies the intended use of the viewing goggles referred to in the previous sentence, it provides context for the preceding statement, thereby explaining the interest of *thousands of people* in said goggles. In (33), the evidential *BE supposed to* serves to justify the speaker's selection of the *Galapagos Islands* as his dream trip.

What differentiates these uses of *BE supposed to* from the counterfactual or negatively-oriented uses is the objective, or informative, rather than argumentative nature of the passage, as well as its thematic structure: in (31) there is an absence of negative or adversative markers from the right co-text, while in (32) the *but* sequence that follows does not contradict the proposition of the *BE supposed to* utterance. Moreover, in both examples there are strong cohesive ties to the left co-text, as the theme (or topic) of the *BE supposed to* sentence corresponds to the focus of the preceding sentence.

### 3.4. Defining roles in the abstract

When the theme of the *BE supposed to* utterance has a generic reference and does not take up the focus but the theme of the preceding sentence, the explanatory function is often lost:

(34) Once entrusted with a case, an investigating magistrate is independent and impartial. He or she **is supposed to** assemble all the evidence suggesting both innocence and guilt. When the investigation is complete, he or she recommends whether the suspect should be prosecuted or cleared. *(In2009)*

Thus, in example (34), *BE supposed to* helps to characterise the role of the subject by defining the properties theoretically associated with it. Compared to the present tense assertions found in the surrounding sentences, *BE supposed to* presents the validation of the predicative relation as desirable, but this is done in the abstract, without any reference to a specific situation, so that the verification of the supposed characteristics of *he or she* is not at stake.

## 3.5. Reporting the opinion or the words of others

In a number of cases, *BE supposed to* is used to refer to the opinion or words of others, which is typically characterised as the evidential use of the structure. There might be a degree of overlap between the type of use presented here and the preceding Sections since many of the occurrences presented in Sections 3.1 to 3.4 can be understood to have an evidential dimension. This is to be expected given that reference to an assertive source other than the speaker was shown to be one of the invariant features of *BE supposed to* (see Section 2.2.1). I would argue, however, that this evidential dimension is not enough for just any occurrence of the structure to be interpreted as fulfilling primarily a reporting function; such an interpretation arises when the information source is at issue, or when there is clear reference to a previously held discourse, as in examples (35) and (36):


Here, *BE supposed to* serves to index a proposition to another enunciative source, thus diluting the speaker's responsibility for the utterance, while signalling that the content of the proposition is not to be taken for granted. In (35), <turquoise-evoke calm and compassion> is not presented as a fact but as a theory held by *psychologists and brand experts*. In (36), <G K Chesterton-say that…> is a *maxim*, something that is generally held to be true but that cannot be checked in the absence of a definite source.

As seen here, there might be epistemic overtones to this type of use of the structure, which are incompatible with the development of a counterfactual interpretation, given that contradictory facts necessarily resolve all uncertainty.

### 3.6. Referring to an uncertain future

Finally, *BE supposed to* regularly occurs with future time reference very often with an 'expected' or 'intended' type of meaning—which blocks the development of the counterfactual reading, although the context might cast doubt on the realisation of the event:


In (37) to (39), because the event is located in the future—*next year* in (37), when the deal is approved in (38), later *today* in (39)—the predicative relation qualified by *BE supposed to* can neither be validated or non-validated in the situation of utterance. In this context, the role of the structure is to present the validation as theoretical, i.e. as planned but relatively uncertain, the uncertainty stemming from the nonfulfillment of the necessary conditions for the event to take place, at the time of utterance: the subject *I* does not have the proper motivation to realise the predicate in (37); the very existence of the subject *the deal* is uncertain in (38); the reason for the speaker's uncertainty regarding the validation of the predicative relation is unspecified in the immediate context in (39), but the uncertainty itself is explicitly stated.

As suggested above, it is worth noting that *BE expected to*, which often has a future orientation, would be acceptable in the context of (37), (38) and (39) but, lacking the theoretical dimension of *BE*  *supposed to*, would not carry the same implication of uncertainty that is echoed in the context of all three occurrences.

## **4. Conclusion**

This study has shown that hypothetical or distancing markers can generate negative meanings in certain contexts. The quasi-modal *BE supposed to* is one such structure as it possesses invariant notional or semantic properties which make it prone to express counterfactual meaning. However, we have also seen that co(n)textual or pragmatic factors play an important part in the activation or blocking of this counterfactual potential.

As a matter of fact, for the counterfactual interpretation to arise, the proposition under *BE supposed to*'s scope needs to be seen to conflict with the actual state of affairs established elsewhere in the discourse. One interesting fact about the counterfactual reading of the structure is that although it is negatively-oriented from an existential point of view, it is typically accompanied by desirability, and is as such positively-oriented from a subjective point of view. This points to the ambivalent nature of *BE supposed to*, which operates simultaneously on various levels of representation.

## **Endnotes**

1. This 40-million-word digital corpus (referred to as *In2009* in what follows) is part of a larger 620-million-word corpus comprising eighteen full years of publication of the digital edition of the *Independent* newspaper (from 1992 to 2009) collected by Catherine Collin (University of Nantes).

2. © 2008–2023 Yasu Imao (Osaka University).

3. See Collins (2009) for a discussion of the term 'quasi-modal'.

4. See for instance Westney, 1995; Visconti, 2004; Moore, 2007; Noël and van der Auwera, 2009; Verhulst et al., 2013; Agrafojo Blanco, 2014.

5. Double slashes in the examples stand for paragraph breaks in the original.

6. The relation between evidentiality and epistemic modality is a complex matter which has been much debated in the literature. The approach taken here follows Aikhenvald (2006, p.320): "Evidentiality is a verbal grammatical category in its own right, and it does not bear any straightforward relationship to truth, the validity of a statement, or the speaker's responsibility. Neither is evidentiality a subcategory of epistemic or any other modality".

7. Culioli (1990, p.69) defines the *notion* as "a complex bundle of structured physico-cultural properties". In this approach, "[n]otions are representations […]; they epitomize properties (the term is used here in a very extensive and loose way) derived from interaction between persons and persons, persons and objects, biological constraints, technical activity, etc." (*Ibid*.)

8. Within the TPEO, BE is considered as an instantiation of the locating operator epsilon (ϵ), meaning 'is located relative to'—identification being one of the possible values for the operation of location. The idea behind the concept of location is that "[n]o term is isolated; all terms can only acquire a referential value if they are part of a locating system" (Chuquet et al., 2010). See Besnard (2016) for more details about BE as a locating operator.

9. I will use the phrase *original speaker* (or simply the term *speaker*) to refer to the *énonciateur* (French for the 'uttering subject' serving as subjective origin within the situation of utterance and, consequently, as the ultimate locator for the construction of referential values), to be differentiated in principle from the *assertive* or *modal source* corresponding to the TPEO's *asserteur* (or *locuteur*) responsible for the propositional content of the utterance and its modal evaluation. See Chuquet et al. (2010) for more details on these aspects of the theory.

10. Relevant excerpts from the *OED* entry for *suppose, v*.:

I. Senses involving mental action.

\* General uses.

1. *trans*. To assume (without reference to truth or falsehood) as a basis of argument, or for the purpose of tracing consequences; to frame as a hypothesis; to put as an imaginary case; to posit.

2003 A. F. Alford *Pyramids of Secrets* iv. 147 Let us now take a different approach to the problem… Let us suppose, as some Egyptologists have suggested, that the Grotto was a sacred site for centuries before the Pyramid was built.

5. To form an idea of, conceive, imagine; to apprehend, guess.

2006 E. D. Stevens *Burnt Rec*. xiv. 135 Then he tried to suppose how she would feel.

7. *trans*.

a. Of a person, system, etc.: to lay down or assume as true, take for granted, accept without question, presuppose.

2003 T. Rockmore *Before & after Hegel* i. 26 Like the majority of philosophers since the ancient Greeks, Maimon simply supposes that it is necessary to avoid all circular reasoning.

b. Of an action, condition, fact: to involve as a ground or basis; to require as a precondition; to imply, presuppose.

2008 D. M. Harland *Exploring Moon* ii. 37 It was argued..that the crater marked the spot where a semi-molten 'volcanic bomb' fell after being ejected from an explosive vent, but this supposed that the Moon had recently been active and that there was a local vent.

9. To entertain as an idea or notion sufficiently probable to be practically assumed as true or to be at least admitted as possibly true, on account of consistency with known facts; to infer hypothetically; to incline to think, sometimes mistakenly.

2003 P. D. Smith *Seas that Mourn* xvii. 139 As Jimmy stood his lookout watch that night atop the flying bridge, he studied the area where he supposed the city to be.

## **References**


## **8. The Meaning of Teachers' Negations in Hong Kong Classrooms Interpreted from their Co-occurring Gestures**

*Renia Lopez-Ozieblo*

## **1. Introduction**

Pragmatic theories seek to explain how intended meanings are formed in social interactions. Rational speakers assume their interlocutors are trying to be truthful, informative and relevant, according to Grice's Cooperative Principle (1975). The Cooperative Principle maxims of quality, quantity, relevance or manner can be achieved not only through the use of verbal language but also through gestures. Gestures can highlight the relevance of certain parts of the utterance, such as negations, or help process them. Negations can be harder to process than positive utterances as they might pose cognitive difficulties, for example when the interlocutor has to establish logical connections based on the context (Tian and Breheny, 2015). We believe that gestures, already identified by Morris in 1938 as potential pragmatic elements, can be key to the processing of negations.

Negation is considered a universal and unique feature of human language (Dahl, 2010), albeit a highly complex one (Roitman, 2017). There is little variation in the functions of negative particles across languages, which all share the same basic linguistic meaning of non-existence, rejection and denial (Roitman, 2017, p.1). Negation can be achieved through morphological or negative affixes, negative particles or negating verbs (Dahl, 1979; Payne, 1985), or through combining with modality and quantity operators. The result is not only the straightforward linguistic non-truth of a proposition p and its rejection, denial or contradiction, but also additional information about the context (Roberts, 1996), metaphoric meaning (Giora, 2006), and other implicit meanings such as sarcasm (Giora, 2016) that can be difficult to interpret.

#### **How to cite this book chapter:**

Lopez-Ozieblo, R. 2023. The Meaning of Teachers' Negations in Hong Kong Classrooms Interpreted from their Co-occurring Gestures. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 203–238. Stockholm: Stockholm University Press. DOI: https://doi.org/10.16993/bcd.h. License: CC BY 4.0.

Non-existence, rejection and denial can also be indicated through hand gestures (as well as with other parts of the body, not covered in this paper), as can references to other content as well as the pragmatic function of the utterance. Although the exact relationship speech-gesture is still being debated, the widespread belief is that they are closely related at the conceptualisation level (McNeill, 2015). Gestures have both cognitive and communicative functions (Gullberg, 2010) including attracting the interlocutor's attention. Speakers often gesture when they speak, and with negative utterances it is possible to observe similar negating gestures across individuals of different cultural backgrounds, such as those shown in Figures 8.1 and 8.21 .

**Figure 8.1.** Vertical Open Hand Prone (OHP): As if pushing away content.

**Figure 8.2.** Horizontal Open Hand Prone: As if re-enacting the action of skimming off the top of something with an outwards wrist rotation.

A number of scholars have confirmed the use of these recurrent gestures in speakers of French, English and Italian (Calbris, 2011; Harrison, 2018; Kendon, 2004), suggesting that there is a strong correlation between the form and movement of the gesture and the semantic meaning of the negation. However, not much has been written about the pragmatic function of gestures co-occurring with negations, and whether the emphasis is on the interaction with the interlocutor, the content negated or the negating act itself. This study seeks to deepen the understanding of the relationships between the type of negations and the gestures co-occurring with them, in particular their pragmatic functions. We propose that linguistic negative utterances are likely to occur with gestures as these mark and clarify the function of the negation, aiding the interlocutor in the processing of its communicative intent.

In a previous study focusing on teachers' disagreements with students, carried out in a Hong Kong Higher Education context with English as the medium of instruction, it was found that teachers heavily mitigated the disagreement, not only linguistically but also by avoiding head and hand gestures that could convey rejection or dismissal (Lopez-Ozieblo, 2018). These observations led to further focus on the use of the negative particle *not* to explore whether there was a general aversion to negation in the discourse of these teachers and, if negation occurred with a gesture, to identify its function. From the existing corpus of ten hours of recorded classroom time, two hours were selected, corresponding to lectures delivered in English to a Cantonese or Mandarin native audience. These lectures were further analysed for negative utterances and the gestures co-occurring with them, excluding disagreements as those had been covered in Lopez-Ozieblo (2018). Only hand gestures considered to be an "integral part of language" (Müller, 2013a, p.2) were taken into account, excluding gestures with a "social-psychological dimension […] [that separate] the body from language" (idem).

This study found that aside from negating gestures, such as holding a palm open facing outwards as if to stop something (Figure 8.1), speakers also perform other types of non-negating recurrent gestures which seem to vary with the type of negation (Figures (8.3 and 8.4). The analysis confirms that the two modalities, gesture and speech, need to be considered together to really understand the communicative intent of the utterance.

**Figure 8.3.** Ring gesture: used to clarify or offer precise information (speaker's view).

**Figure 8.4.** Palm Up or Open Hand Supine (OHS): offering information to the interlocutor.

As this chapter is included in a volume on negation it will not provide an extensive introduction to negation, instead focusing on gestures. It will describe some of the basic concepts related to negation that have been considered when analysing the gestures. After detailing the methodology applied, some of the more illustrative gestures co-occurring with negations are discussed, in particular with negating particles *no, not* and contractions. The results suggest that there are at least three functions in gestures co-occurring with negations: emphasising the negation with a negating gesture, stressing the utterance with a non-negating gesture and focusing on the negated concept.

## **2. Negation**

A standard negation is defined as "the basic way(s) a language has for negating declarative verbal main clauses." (Miestamo, 2005, p.1). In English the scope of the standard negation is an entire declarative clause, formulated by following a general strategy through the addition of a negative particle (and an auxiliary when relevant) (Van der Auwera, 2006); these include *no, not*, and *never*. Negation can also be achieved through negative intensive negators that, together with the negative particle, can indicate the negation is complete, such as the Negative Polarity item *at all*, or can identify small quantities, such as *a bit* (Cruschina, Hartmann and Remberger, 2017). While not all negations are standard in every language, such as imperative, existential and nonverbal clauses, they can also be negated using standard negators (Miestamo, 2007).

Negation research, based on the works by Jespersen (1917), Klima (1964) and more recently Horn (1989), has led to several typologies that classify negation according to its syntax or pragmatic meaning (Dahl, 1979; Payne, 1985, Nølke, 2017). From a pragmatic approach (Ducrot, 1972; Nølke, 2017), negations can be interpreted as inhabiting a continuum from describing the state of the world (descriptive) to opposing a former assertion, which is not always explicit (polemic), including form-based responses to a speaker (metalinguistic). Descriptive negations emphasise their descriptive value – this being the reason for the negation, transforming negative content into a new assertion that could not, in the speaker's mind, be expressed in a more accurate manner, although this does not suggest that interlocutor believes the underlying positive proposition:

(1a) There is no cloud in the sky (Ducrot, 1972, p.38, cited in Nølke, 2017)

has the affirmative meaning of:

(1b) The sky is blue (Nølke, 2017, p.151)

Although, in this case, the statement 'The sky is blue' or 'The sky is clear' might describe a perceived reality more accurately and simply.

Polemic negations make implicit reference to former positive assertions and refute them by providing an alternative that actualises the context. Sometimes they might be addressing the expected beliefs of third parties. In these cases, there are two incompatible voices or points of view, the second rejecting the first:

(2a) This wall is not white (Ducrot, 1972, p.38, cited in Nølke, 2017)

Uttered in opposition to a previous thought (explicit or implicit) supported by an interlocutor meaning:

(2b) This wall is white (Nølke, 2017, p.153)

Some scholars include a metalinguistic variation to polemic and descriptive negations where the scope is the locution or the form. In metalinguistic negations the speaker utters an objection to the grammatical or phonetic form, to the register or to the possible implicatures of a previous utterance (Horn and Wansing, 2015). Metalinguistic negations also respond to a previous context but expand upon the presupposition:

(3) Paul is not **big**, he is gigantic (Nølke, 2017, p.152)

Positively worded alternatives are not always necessary with metalinguistic negations (see example (4)). Other scholars (Larrivée, 2018; Moeschler, 2015) defend an exclusive category for metalinguistic negations, as these have the specific function of correcting previously introduced content with a positive outcome (while polemic and descriptive negations have negative outcomes) (Larrivée, 2018).

One important element of metalinguistic negations is the use of the structure and intonation to stress specific elements (Cruschina et al., 2017). This type of negation is more prosodically marked than descriptive negations, at least in English (Bolinger, 1989). In example (3), the emphasis, marked in bold, lies on the adjective *big* which is stressed, rejected and corrected with *gigantic*. The stress can also appear on the negative particle itself:

(4) Paul has **not** beaten the dog with the stick (Nølke, 2017, p.155)

Here, a response to the interlocutor is given, correcting the description of the situation, although not providing a useful update (Nølke, 2017, p.155).

A functional approach has also been proposed by Miestamo (2005) based on the asymmetric features of negations compared to affirmations. He (2005) adds that the negative clause is stative, indicating that something is not happening or changing. This results in a prosodic conflict as the negated action loses strength, and so prosodic stress, while at the same time there is a need to emphasise the negation (Dahl, 2010; Horn, 1989).

#### 2.1. Teaching and negations

In some pedagogical contexts, such as language proficiency classrooms or with non-native audiences, speakers tend to be aware of their use of negations as their processing relies on an accurate understanding of the pragmatic context. In the classroom there are two factors that might influence how teachers use negations: they are harder to process (Kaup, 2001); or they are associated with a strategy for correcting (Givón, 2015) that might threaten the *face* of students, potentially damaging their self-image if they feel their contributions are put down or disagreed with (Kerssen-Griep, 2001).

When a speaker uses a negative sentence not *p*, this not only means that the speaker believes that *p* is false but also that she believes the interlocutor finds *p* to be true. In cases like this, where speaker and interlocutor are both aware of the others beliefs, the negation is considered pragmatically felicitous, within context, and thus easier to process. However, when one of the interlocutors has no reason to believe that the other believes *p* to be true, perhaps because there is no shared context (such as cases of cultural differences between interlocutors), then the negation might cause processing difficulties by being pragmatically infelicitous (Horn, 1989). Psycholinguistic studies confirm that negative sentences are harder to process than positive ones (Kaup, 2001). Following Wason and Evans's (1974) observation that there is an extra step in the processing of negations, studies confirm that negative utterances hinder sentence verification, memory recall and logical reasoning (Tian and Breheny, 2015). Horn (1989) suggests that this could be because affirmative sentences present facts about the world, while negative ones give facts about the affirmatives, operating on affirmative concepts by modifying them. In addition, the unmarked affirmative structure occurs more often than the negative and so it should be easier to process, if only by reason of frequency (Roitman, 2017). Nordmeyer and Frank (2015) further propose that difficulties in processing negation seem to occur particularly when the contextual information is missing, as might happen in a classroom with a nonnative audience. Teachers who are aware of the difficulties inherent in processing negations might try to avoid them, especially when addressing non-fluent students.

The second factor relates to potential disagreements. A recent study of teacher-student disagreement, in the same context as this study (Lopez-Ozieblo, 2018), found that the ten teachers under study avoided disagreements whenever possible and minimised the salience of the act by avoiding negative gestures or head movements. Disagreements were further mitigated through linguistic markers, and potential face threats to students were avoided through the use of nods during the disagreement, to encourage students' interaction. This study focuses on the first factor, exploring how teachers' might be using gestures to facilitate students' processing of negations.

## **3. Gestures**

Gestures, for the purposes of this paper, are defined as deliberate and conscious movements of the hands co-occurring with speech and are believed to be part of the speech act (Kendon, 2004). Gestures are not add-ons to speech or indicators of emotions (Müller, 2013a, p.2) but form a unit with speech, externalising the thought in both modalities (Lopez-Ozieblo and McNeill, 2017). Gestures, these deliberate hand movements co-occurring with speech, have been categorised on a continuum (McNeill, 1992) according to their form and relationship with speech. At one end of this continuum sit hand signs, used in signed languages, which do not require words to be understood. They are followed by mime and emblems which are codified gestures with a shared meaning within a specific social group, such as the OK sign, represented by a circle made with the index finger and thumb while the other fingers remain extended. Other hand movements used with speech can be generally representative of the content externalised in the speech, illustrating the content or alluding to the pragmatic intent of the communicative act (Kendon, 2004) -sometimes both. Representational or referential gestures can have either an iconic or metaphoric relationship with the content of the speech (McNeill, 1992); their meaning could be thought of as mostly semantic. They can be used to illustrate the concept by drawing its outline or indicating its shape, enacting or representing it (Müller, 1998) or to point at it or its position (deictic gestures, see Figure 8.5). Non-referential gestures, known as beats, are those used to keep the prosodic rhythm of the utterance or to stress parts of it, such as an index finger moving up and down marking the syllables. It is likely that metalinguistic negations are accompanied by these gestures to further mark their prosody.

**Figure 8.5.** Deictic gesture, pointing.

Both referential and non-referential gestures can have a pragmatic function when they relate to the "features of an utterance's meaning that are not a part of its referential meaning or propositional content" (Kendon, 2004, p.158). Pragmatic gestures have been ascribed three functions: organising the flow of the discourse, such as indicating a disfluency or repair, commenting on the utterance, or linking or stressing parts of the utterance (metadiscursive function); adding interaction with the interlocutor, such as managing the floor, offering or taking ideas, and evaluating or dismissing them (interactive function); and providing logical connections or inferences (cognitive function) (Lopez-Ozieblo, 2020).

Many pragmatic negating gestures exhibit form commonalities that also make them referential (semantic) as they provide a metaphorical illustration of an act commonly associated with a negation. Gesture form refers to its handshape, orientation, movement and location in space (Bressem, 2013). Recognisable gestures with a "stable form-meaning unit" (Ladewig, 2013, p.1559) that is partly conventionalised (close to being emblems but exhibiting nuances in meaning dependent on context) are referred to as recurrent gestures. Their meaning is not matched to a specific word, but is rather more schematic and somewhat related to its form, such as extending a hand with the palm facing up to present an idea to an interlocutor (Figure 8.4). Recurrent gestures sharing similar forms and functions have been grouped into families, such as those described by Kendon (2004) and Müller, Bressem and Ladewig (2013). One such family is that of the vertical Open Hand Prone (OHP), as if trying to stop the advancement of something or someone in front of the speaker, a common negating gesture (Figure 8.1). Other versions of these gestures are the index finger, or whole palm facing away from the speaker and oscillating horizontally left to right, observed in French (Calbris, 1990) and English speakers (Harrison, 2010), as if re-enacting the erasure of the concept. This gesture has been observed with apologies and when refusing offers, suggestions or implications (Harrison, 2018, pp. 95–100). A horizontal orientation of the same gesture form is the horizontal palm down (Figure 8.2), often observed with a horizontal lateral movement, as if skimming off something and pushing it away, indicating the ending or suspension of a line of action, usually interpreted as being outside the control of the speaker. It is also a gesture associated with universal statements with no exceptions or with extreme positive or negative assessments, indicating also that no other options exist. These gestures express denial, interruption or negation, stressing the impossibility of continuing with a specific line of action or discussion. This gesture can also be seen with an extreme positive evaluation that leaves no room for other evaluations or that uses a negative particle such as *never* to indicate a positive: *I have never seen anything so beautiful*. In many cases these gestures have a modal function and act specifically on the clause (the thought or action expressed by it) to negate it, just as verbal polemic negative particles often do. OHP gestures can also be used to push away any other options, not operating on the clause just uttered but negating a potential counter response from the interlocutor (Kendon, 2004, pp.248–264). Another variation can be observed when both hands come together by the midline of the body and move outwards and down, sweeping aside or clearing away. When the hands cross first and then move downwards they are illustrating the impossibility of carrying on with a line of action forcibly slicing through it like a pair of scissors (Harrison, 2018, pp.22–46).

Other gesture families, with meanings other than negating, include the R– family – ring shaped gestures (similar to the OK emblem but with a different orientation, Figure 8.3) used when clarifying or offering precise information; the palm up or Open Hand Supine (OHS) family, as if offering something to an interlocutor, often co-occurring with a new topic (Figure 8.4); or the G-family, where the fingers bunch together into a grip that indicates precision, also called the *Grappolo gesture* (Figure 8.6). These are gestures that make salient new information, focusing on content as descriptive negations do. Detailed information about these gestures is provided under the Discussion, when relevant.

**Figure 8.6.** Grappolo gesture: when adding precision to the information.

Each family is composed of a number of gestures that share the same form but might vary in orientation or movement with subtle changes to the meaning depending on these variations (for more details see Kendon, 2004, pp.225–283). In addition, gestures have phases within a *phrase* that refer to how they are organised. In many cases, the gestural phrase will start from the resting position of the hand. The hand then moves to come into position – the preparation phase – and might pause for a moment – the pre-stroke hold – before the stroke is performed. The *stroke* is the nucleus of the gesture, where key content, or the newsworthy element, will be delivered. After the stroke the hand might pause again – the post-stroke hold – and finally return to the resting position (Kendon, 2004, pp.110–114; McNeill, 2005, pp.31–33). Gestural phases might overlap or be linked as a chain in such rapid succession that not all the phases are performed or are obvious to the observer.

### **4. Methodology**

The objective of this study was to identify whether different types of negations co-occur with different types of recurrent gestures and how these might aid the processing of the negation. The qualitative analysis focused on two hours of lectures from two separate teachers, working in a Hong Kong Higher Education institution, where English was the medium of instruction and a second language to most students.

This corpus was a subset of the previously mentioned study that analysed teacher-student disagreements in ten lectures in the same institution (Lopez-Ozieblo, 2018). All ten lectures were initially quantitatively analysed to identify the number of negative particles used per word and the frequency of their occurrence with gestures. Having confirmed that all teachers used negations, two teachers were randomly selected from the ten, one female and one male. Their discourse was further analysed using a qualitative approach that selected only negations containing the markers *no* (these were noted but not analysed as they were covered by a previous study on disagreement), *not*, and their contractions, providing detailed descriptions of the gestures co-occurring with them. The analysis was based on a combination of Larrivée's (2018) categorisation of negations, including metalinguistic negations as a third category, and that of a polemic-descriptive continuum, based on Nølke's (2017) pragmatic categorisation of negations, as we believe that some negations are more polemic than others. The analysis of gestures was based on Bressem's (2013) gesture description and Kendon's (2004) families of gestures. The framework used for interpreting the gestures was based on the pragmatic functions identified by Lopez-Ozieblo (2020).

As the focus of polemic negations is the opposition itself, we expected to see more negating gestures in these types of utterances. Descriptive negations, on the other hand, focus on the content, therefore the expectation was to see either referential gestures, representing the content of the utterance, or recurrent gestures indicating new information, clarifications or precision. As metalinguistic negations are often marked prosodically (Bolinger, 1989), we predicted more beats with this specific type of negation.

### 4.1. Participants

From the ten recorded lectures, two sessions were randomly selected for this qualitative study, corresponding to one female European teacher and one male Southeast-Asian teacher. In both cases, the author and an assistant were present and carried out the recordings. Two cameras were used, usually one at the front of the classroom and one at the back, both pointing at the teacher in order to avoid recording students' faces. Classes were recorded after week 8 of term to ensure that the teacher/student immediacy bond had already developed.

Teachers were lecturing on Language Education topics at Masters level. One teacher was a monolingual English speaker (Teacher 1) and the other a bilingual Mandarin-English speaker (Teacher 2). They each had over five years teaching experience in the Hong Kong context. The average class size was 33 students, many of which were primary or secondary English teachers. The majority of students (90%) were from Hong Kong. The rest were from East Asian countries (Mainland China, Korea, Taiwan). A fifth (20%) were male.

Teachers were aware that we would be carrying out a multimodal analysis of their deliveries but were not told we would be focusing on negative utterances. To exclude the possibility of the recordings affecting teachers' behaviours they were asked whether that particular session differed from others where cameras had not been present. One admitted to having been nervous to start with and then forgetting about the cameras, while the other teacher video-recorded her own teaching sessions regularly and was used to the camera. We believe their deliveries, and thus their gestures, were spontaneous and natural.

### 4.2. Data analysis

The speech from the recordings of each session was transcribed by student helpers, using *Praat* (a free software for voice transcription), and checked by two research assistants and the author. The transcriptions were then imported into *ELAN*, a free software for multimedia analysis, where gesture transcriptions were added by the research assistants and the author (checked by both, obtaining intercoder reliability of 100% on 95% of the data after discussion –unclear events were excluded).

The team used a Corpus linguistics tool, *Wordsmith*, to identify all cases of the negative particle *not* and all contractions of auxiliary verbs and *not* including *haven't, hasn't, isn't, wasn't, don't, didn't, won't, wouldn't, can't* and c*annot* (cases of *no* were also noted). These results were corroborated with those found using the *ELAN* search function and the gesture information was added (i.e., whether a gesture co-occurred with the particle and its semantic relevance to the negative particle). All standard and non-standard negations containing negative particles were initially identified. A detailed analysis of the discourse, taking into account the context, eliminated tag questions such as *isn't it*?. The examples discussed below were chosen from all other remaining cases to illustrate how these two teachers employed negations.

## **5. Results and discussion**

This Section briefly details the quantitative results obtained from the analysis of the discourse of two teachers. It then focuses on the instances where negations co-occurred with gestures and explores the form and function of the gesture.

The study found that the bilingual teacher (Teacher 2) used almost twice as many negations per word as the English monolingual teacher (Teacher 1). In both cases, slightly over half of the negations co-occurred with gestures that were relevant to the negation, either to the particle itself or to the negated content. The qualitative analysis was based on five examples from each teacher, selected to illustrate the different functions of the gestures.

### 5.1. Quantitative results

The discourse of Teacher 1 (monolingual European female) contained 4570 words. We found 37 cases of basic clausal negations using the particle *no* or *not* in its contracted and non-contracted forms, which amounted to 0.81% of all words. The contracted forms included *isn't*, *won't*, *haven't* and *don't*. The 37 cases included 6 cases of the use of *no* as a response particle in answer to students' comments or questions. These have been excluded as they are considered disagreements. Twenty-one of the 31 negative markers (68%) co-occurred with a gesture – twice as many as without a gesture (10) – and just four of those gestures (13% of the 31) were unrelated to the negative utterance (three with the particle *not* and one with *don't*) (Table 8.1).

The discourse of Teacher 2 (bilingual Southeast-Asian male) contained 4755 words. We found 62 cases of basic negations including the contractions *can't*, *didn't, don't, haven't* and *won't*, and one case of *cannot*, amounting to 1.3% of all words, including 6 cases of the use of *no* not related to disagreement (excluded from the analysis). Thirtythree of the 56 negations co-occurred with gestures (59%) but in 8 of those cases (14% of the 56) these were not relevant to the negative utterance (Table 8.1).


**Table 8.1.** Corpus summary.

### 5.2. Qualitative analysis

The cases with gestures related to the negative meaning were further analysed, the examples below have been selected to provide an illustration of these gestures. Some of these are recurrent gestures, repeated in different speakers, with negating as well as specifying, showing, querying or thinking functions.

### Teacher 1 (T1) – English native female

Teacher 1's lecture focused on the use of oral discourse markers and their importance in everyday speech. The teacher indicated that language teachers should explicitly teach discourse markers, including fillers, as their use is seldom covered by textbooks on English as a foreign language. She provides a number of personal and general examples to illustrate the theoretical points being made. She stood in front of the classroom, often moving sideways with occasional walks up and down the central divide. The key points were presented in slide format projected onto a screen at the front of the classroom, controlled by a remote presenter which the teacher switched between both hands throughout the lecture.

### T1 – Example (1)

The teacher is explaining how journalists practice in their live reports in advance, saying: *they know what they gonna do*. She had previously mentioned a war context and is afraid that students might misunderstand by thinking that journalists are planning the war rather than planning the content of their speech. She clarifies that this is not the case. We observed a negating gesture co-occurring with the first negative particle *not*:


Gesture 1 (G1): Both hands close together by the center of the body, palms open, the left facing up and the right facing left, fingers separated chest height. The left hand (LH) is holding the remote presenter with the thumb. The right hand (RH), palm facing left moves sideways to the right at the elbow (fingers together) and back to meet the LH (Figure 8.7).

**Figure 8.7.** Gesture (G1): RH moves to the left and back.\* \*Note: All gestures performed by actors due to low video resolution.

In this example, the gesture is synchronous with the negative particle (often observed with negating gestures (Harrison, 2018)). It is a pragmatic gesture with a metadiscursive function, stressing a polemic negation, the marker itself, where unintended potential content from a previous utterance is denied (Ducrot, 1972; Nølke, 2017). The teacher makes implicit reference to previous content—the war—and actualises it. With the gesture she seems to remove the negated concept she had provided (the war). Kendon (2004) describes the function of similar gestures under the Vertical Open Hand Prone (OHP) family as pushing away content. OHP gestures include those where the palm (or palms) is open, fingers close to each other, and facing down or away from the speaker. They illustrate the act of stopping, rejecting or pushing away something (Figure 8.1). There are two variations: gestures with a vertical forearm, the palm facing away from the speaker – Vertical Palm gestures; or those with the forearm horizontal or at a slight angle, with the palm facing down – Horizontal Palm gestures. Both are commonly regarded as negating gestures.

Vertical Palm gestures locate a boundary in front of the speaker, suggesting the halting of an action or a concept. If this boundary is close to the speaker, it indicates the speaker's own actions or thoughts, while a boundary closer to the interlocutor would indicate the interlocutors' actions. When both hands are involved in the gesture, and the two palms are facing and moving away from each other, it is as if the speaker is moulding a barrier (Müller, Bressem and Ladewig, 2013; Harrison, 2018) to block a line of thought. We suggest Gesture (G1) in this example is yet another variation of the Vertical Palm subfamily where just one hand moves sideways (not both moving away from each other, as described by Müller et al., 2013), blocking the thought of war and pushing the idea away.

T1 – Example (2)

The teacher expands on the topic of oral casual conversation, insisting on the need to teach it as part of English as a foreign language:


G2.1: RH holding the remote presenter at rest by chest, LH forms into a ring gesture with the thumb and forefinger touching that comes down with the negative marker (Figure 8.8).

G2.2: The LH holds the ring gesture as the arm comes up and down to stress "why".

G2.3 and 2.4: LH holds the ring gesture and goes up and down twice, the last movement synchronous with the "not" is more forceful

G2.5: Representative gesture with the LH fingers opening up into a loose fist that circles as if moulding a circle.

**Figure 8.8.** Gesture (2) Ring gesture, LH moves up and down, movement repeated four times.

Gestures 2.1 to 2.5 are a chain used to stress various parts of the utterance, including two out of the three negations (other negations in this extract did not co-occur with gestures). As in Example (1), these are pragmatic gestures with metadiscursive functions, keeping the prosodic rhythm of the utterance and stressing the negation (and the "why", gesture 2.2). The two negations marked by Gestures 2.1 and 2.5 are considered descriptive, not referring to previously mentioned content.

The first *no* corresponds to the teacher's self-answer to her question; a strong negation that could be considered somewhat face threatening, as it may be construed as a criticism of students (who are English teachers themselves). In this case, we see a clear example of a ring gesture, with the index and thumb tips touching. The negation is mitigated through a non-negating gesture that is instead focusing on the delivery with a metadiscursive gesture that is commenting on the importance of the utterance. The depiction of ring gestures dates back to Greek amphora making times, maybe earlier, where gesturing figures decorated the vessels (de Jorio, [1832] 2000). Quintilian in the 1st c. AD (Butler, 1920) also described ring gestures as one of the strategies of good rhetoric, and they are still in use in various cultures – as can be attested when observing the speech of many politicians today. Some ring gestures are classified as emblems, codified within a society, which do not need speech to be understood, such as the OK-sign. However, what we observe here is a rhetorical gesture which is part of the multimodal utterance (Müller, 2013b). Kendon (2004, pp.240–245) divides ring gestures into three types: Ring-to-Open gestures are those where the ring opens up as something is being clarified or an exact piece of information given; Ring-display gestures are those where the hand is lifted and the ring formed, often when opposing a previous idea; and Ring-vertical gestures are those with the ring formed, palm towards the speaker and the forearm is moved up and down coinciding with the prosodic stress. With Ring-vertical gestures, the speaker tends to be very insistent on a specific point.

In the above gestures (G2.1 to 2.4) we observe a ring-vertical gesture – the ring is formed with the palm towards the mid-line of the speaker and the forearm is moved up and down. The point is repeated with *why isn't it taught*, this time stressing the *why* (G2.2) and *think* (G2.3). The potential face threat is further minimised by referring to *the people*, rather than *you, teachers*. A second negation (G2.4), also coinciding with the prosodic stress, repeats the gesture yet again.

In this extract there were three negations uttered, none marked with a negating gesture. Instead the teacher coordinated the gesture to highlight specific information which she seems to be quite passionate about. The ring gesture with the vertical downwards movement has also been associated with the idea of justice (de Jorio, [1832] 2000), from the iconic representation of a set of scales. The element of justice is also present in Example (2) as the teacher condemns the lack of oral English teaching.

T1 – Example (3)

The teacher is explaining various methods to evaluate students, explaining cloze (fill in the gap) questions and how these could be used to test students' knowledge, but stressing that they are not a teaching tool:


reason is a cloze exercise.] /// [cloze exercise] is where you just [tip (G3.2)

out- [tippex out [^words [… it's ^just /[ ^testing. /// [it is not (G3.3)

teaching it's [^simply ^testing / the students,]

G3.1: RH (holding the remote presenter) has been lifted to shoulder and held, it then comes down (Figure 8.9), together with a movement of the head that comes forward and down.

G3.2: The hands, originally mid-body, open outwards and back to mid point.

G3.3: RH moves to enact the tippexing-out or underlining of words, by acting as if holding a pen /tippex brush and selecting words, with quick horizontal left to right movements at shoulder level. This gesture, representing the correction of students' essays, is repeated seven times.

The first negation (*you should not do*), is another (somewhat) polemic negation that refers to the already introduced content of activities, and is already mitigated by the use of *should*, as the teacher avoids the imperative. As in Example (1), we observe the stress in the negative marker, emphasised by the negating gesture of a Vertical Open Hand Prone, as if trying to quell something. This is often observed with negative markers (Harrison, 2018), although in this case the movement is constrained by holding the remote presenter.

**Figure 8.9.** Gesture (3.1): Vertical Open Hand Prone (palm partly closed as it is holding the remote presenter), as if trying to quell something, often observed with negative markers.

The metalinguistic *it is not teaching* is uttered after the context—teaching methods and the use of cloze questions—has already been introduced in the discourse. This clause is suggesting that doing cloze exercises is not pedagogically sound. Using cloze tests might be a useful tool for testing students' knowledge, but the teacher does not consider it an effective teaching tool. The focus of the gesture is not so much the negation itself but the act of correcting students' work, which the teacher illustrates repeatedly with the referential gesture G3.3. Just crossing out or highlighting words cannot be considered as pedagogically instructive, and thus cannot be considered as teaching. The clause *it is not teaching* would seem to be an unmitigated denial where the gesture aids the interlocutor to focus neither on the negative particle (*not*) nor on the action being negated (*teaching*) but on a parallel action (*testing*) that is not a direct antonym of the negated action. The teacher repeats the idea that cloze questions are just for *testing*, before and after the negation.

T1 – Example (4)

The teacher is discussing the absence of fillers such as OK, listen, ehm, uhm and ahm in textbooks:


G4.1 LH palm by chest facing out waves L to R (Figure 8.10).

G4.2 to 4.3: Representative gestures not related to the negations

G4.4 Teacher is standing sideways, RH to board behind teacher and rests there, LH by side of the body, elbow bent, palm facing right, thumb up and fingers slightly apart pointing forward and moves slightly down at the elbow.

**Figure 8.10.** Gesture (4.1): Vertical OHP moving sideways.\* \*Note: Gesture performed by an actor as the resolution of the video was insufficient.

Although the existence of these fillers has not been previously confirmed, it could be argued that this teacher felt they were important enough to be included in the textbooks. Therefore, this study sees these as examples of somewhat polemic negations. In these examples, we observe gestures from the negating family of gestures of the Open Hand Prone family. The gesture in the first descriptive negation *they are not present*, does not just stress the negative marker, but the whole existential structure. The gesture is a standard negating one: Vertical Palm Open. The second negation also co-occurs with a gesture, also a Vertical Palm, although less marked than the first one, as if creating a barrier so that these oral fillers cannot go through (Müller, Bressem and Ladewig, 2013). The gestures have the metadiscursive function of stressing the negations.

### T1 – Example (5)

The teacher is discussing different pedagogical approaches to teaching, including movement-based instruction:


G5.1–5.3: Both hands are lifted chest high as if holding a large ball in front of the body (this refers to the context that is being discussed), RH is holding the remote presenter with the thumb and it rotates back and forth at the elbow three times, corresponding with the underlined text (Figure 8.11).

**Figure 8.11.** Gesture (5): Deictic gesture up and down repeated three times.

In this chain of gestures, we observe a deictic reference to the *people next door* which takes the form of a beat gesture coinciding with the rhythmic prosody of the utterance. Thus, although the gesture co-occurs with the descriptive negation *don't like*, it is hard to confirm that this is specifically stressing the negation. Hedberg and Sosa (2003, cited in Tian and Breheny, 2015) found that negative particles were usually marked with a high-pitched accent, except in structures with the contraction *don't*. It is interesting to observe a similar phenomenon in the example above, where there is no obvious prosodic stress in the negative utterance. The gesture occurring with *don't like* is of a referential deictic nature pointing to the *people next door*, thus not emphasising the nature of the negation.

Teacher 2 (T2)

Teacher 2's lecture focused on the devices used to create a good narration. He stood in front of the classroom, moving mostly up and down the central divide. As with Teacher 1, the key points were presented in slide format projected onto a screen at the front of the classroom, controlled by a remote presenter which the teacher switched between hands throughout the lecture.

### T2 – Example (6)

Teacher 2 is looking for a situation where children are likely to be disobedient. He begins a negative utterance but encounters some difficulties in finishing it, not having planned it fully, interrupts himself and asks what might be a self-directed question, and then completes the utterance:

T2 [if you want to persuade children to do something [like don't G6.1 [eh don't do what? [/// [don't eat your: ^finger for example.]]]] G6.2 G6.3

G6.1: RH is holding the remote presenter, arm bent at the elbow with the hand mid-body. LH also bent at the elbow but slightly higher, hand close to the L shoulder and the thumb and forefinger close into a ring facing the speaker which goes slightly down and up.

G6.2: LH releases ring gesture palm facing in, still by shoulder, fingers extended vertically upwards, slightly separated.

G6.3: LH fingers move back and forth quickly in a seeking gesture (Figure 8.12).

**Figure 8.12.** Gesture (6.3): Seeking gesture. Fingers move back and forth for the right searching word (interlocutors view).

In Example (6) the only gesture related to the negation is the first one – the descriptive negation (there is no previous referent) *like don't*, where the teacher brings his fingers together into a ring, as if holding up a concept for viewing and making it prominent. The concept is a prohibition, where the gesture, with a metadiscursive–modal–function, is directing students to focus on the negated action, not even the prosodic stress focuses on the negation. Unfortunately, the teacher encounters a planning difficulty and is not able to find a good example, repeating the negation and stalling for time. The second negation does not carry a semantic negative meaning but a pragmatic one, buying himself time and still holding the floor. This is reinforced in the gesture of the third *don't*, also descriptive and co-occurring with an interactive/metadiscursive pragmatic gesture to indicate the speaker is holding the floor while looking for a word. These seeking gestures differ from lexical ones in that they are not iconic, they tend to be repetitive small gestures that indicate that the speaker is thinking and keeping the floor. Lexical gestures illustrate the elusive words in an attempt to prime them or to ask the interlocutor for help (Gullberg, 2011). Despite the fact that the teacher is talking about fingers (*don't eat your finger*), the gesture does not seem to be referring to the action, which would have brought the hand closer to the mouth.

#### T. 2 Example (7)

The teacher continues to provide examples relating to prohibitions and children:

$$\begin{array}{ccc} \text{T2} \longrightarrow & [\text{ $ }\text{$ ^\circ $} \text{don't } \underline{\text{liter}} \text{ for } \text{exam}] \text{ple } \text{right?} \\ & \text{G7.:} \\ \longrightarrow & [\text{$  }\text{ $] $ ^\circ $ so } / \underline{\text{i}} \text{ you say} \text{) [$  \underline{\text{don't}} \text{ liter.} ] \text{they probably won't listen.} \\ & \text{G7.:} \\ & \text{G7.:} \end{array}$$

G7.1: LH moves to the left as palm facing right, as if re-enacting the action of throwing something away.

G7.2: Both hands together initially by mid-body line, open outwards and close in a quick movement.

G7.3: This is a similar gesture to the previous one; initially the hands are together, they separate with the palms facing each other, the left travelling further outwards than the right with a slight rotation of the wrist and then come back together (see Figure 8.13):

**Figure 8.13.** Gesture (7.3): As if re-enacting the action of throwing something away.

In gestures (7.1 and 7.3) the hand could be negating or representing the action of throwing something away; there is a small flick of the wrist at the end of the stroke. This *don't litter* is again a descriptive negation, not referring to a previous action. As the referential gesture co-occurs with the verb, rather than the negating particle, the salience is being placed on the negated action. The verb *to litter* already carries negative associations, explaining, perhaps, why the gesture occurs with the action and not the negation (an example of the potential conflict between the need to stress the action but also the negation (Dahl, 2010)). In this case, both the speaker and the interlocutors are more likely to first form a mental representation of the littering act, and later of its negated form. The second gesture (7.2) is metaphorically holding the concept between the two palms and as these open it is released for the audience. Finally, the third gesture (7.3), co-occurring with the negation, seems to be a combination of the two previous gestures where the concept is again held for the audience but then the action of throwing away is re-enacted. The gesture is presenting the whole of the negated utterance as an example, held between the two hands and offered to the students.

### T. 2 Example (8)

Teacher 2 is discussing what children can learn from a fairy tale and is focusing now on the vocabulary:

T2 so you can say there are some difficult verbs [even ^I ^I don't know] G8 them [very specific] words right↑

G8: Both hands start by mid line, under the chest, and open up to the sides palms facing up (RH is holding the remote presenter, so it is partly closed), and come back to the initial position (Figure 8.14).

**Figure 8.14.** Gesture (8): Open Hand Supine negative gesture with a shoulder shrug.

The negation seems to paraphrase the previous assertion of there being *difficult verbs*. This gesture is a negating recurrent gesture, although it is from the Open Hand Supine family of gestures (Figure 8.4). The gestures in this family share form and palm orientation; an open palm that faces upwards. It is a gesture associated with presenting or offering something, such as providing an explanation. If there is movement

towards something or someone it could be that another source of information is being acknowledged or being asked to acknowledge the item being presented. If the hand is retracted it might also indicate receiving something. A third possibility, the one we observed here, is a lateral movement away from the speaker, often starting with the palm down and with a half rotation at the elbow away from the speaker with a shrug of the shoulders, indicating a withdrawal from the idea or the situation. This gesture could be interpreted as having a metadiscursive modal function, indicating inability or unwillingness to further the idea. In this case, the teacher is indicating his difficulties with verbs, thus stressing that aspect of the semantic content of the utterance.

#### T. 2 Example (9)

The teacher continues to discuss the fairy tale as a tool to teach certain values to children:

$$\begin{array}{rcl} \text{T2} \longrightarrow & \text{so it [\underline{\text{is}} \text{\textquotedblleft not}] just} \\ & \text{G9} \end{array}$$

[an instructive story but the-the emotion / matters / <utterance continues.>]

G9: RH holding the remote presenter in front of the body (forearm distance), arm slightly bent at elbow, looking sideways to the screen and pointing at it. Left arm across the body, forefinger extended. The R arm is moved slightly up and down at the elbow at the same time as the body moves back and forth (Figure 8.15).

**Figure 8.15.** Gesture (9): Deictic gesture with a beat.

In this descriptive negation the negative marker is synchronous with the negative particle and the verb. It is a pragmatic gesture with a metadiscursive function, stressing the negation, but not a recurrent negating gesture.

### T. 2 Example (10)

The teacher has provided students with another story; this one is written down and projected for students to read. He looks for volunteers to read it:

### T2 eh I am [^not good at] reading stories. so anyone want to read it? G10

G10: Hands by chest and mid-line, RH holding the remote presenter and LH closed but with index finger extended, with the stroke they open slightly outwards and up then move downwards (index still extended tracing a small arc) (Figure 8.16).

**Figure 8.16.** Gesture (10): Stretched index finger.

In Example (10) we observe the use of a polemic negation; students probably expect the teacher to be able to read a story. The gesture and the prosody are both stressing the negation. This seems to be a strategy chosen by the teacher to make students more comfortable by denying his own ability to read. The held stretched index finger has been identified as a recurrent gesture that calls the attention to new content, or dismisses previous statements (Müller, Bressem, Ladewig, 2013), and is a gesture related to the OHP oscillating finger, only in this case the movement of the index is more subtle. In this case we might interpret it as dismissing one's own abilities and expressing the impossibility of continuing with a specific line of action – that of reading.

## **6. Summary**

This study focused on the negative markers and the gestures that co-occur with them of two teachers, working within the Hong Kong Higher Education context. From the observations of these two teachers, it is clear that both teachers use negation markers. Both were more likely to produce negative utterances with gestures (rather than without gestures), indicating their salience, confirmed by the observation that, in most cases, these gestures were related to the negation. These related gestures were further analysed and were observed to have three functions, sometimes combined: (1) Stressing the negative marker but without adding negating salience. Some of these gestures have been identified as being recurrent, such as the ring gesture, often used in rhetoric to clarify; however, others were small up and down movements used to stress an element (beat gestures), or to achieve immediacy with the audience by offering ideas to them. (2) Focusing on the negated concept. These gestures, mostly referential, illustrated the action being negated or an element of it. (3) Stressing the negative marker by adding a negating gesture. These recurrent gestures were from negating families of gestures, which are often associated with negative utterances (Harrison, 2018).

Out of the fifteen negations detailed above, one was classified as metalinguistic (7%), eight as descriptive (53%), and six as polemic (40%). In all six of these polemic negations, the gesture co-occurring with them was also a negating gesture. However, with the descriptive and metalinguistic negations, the gestures did not represent the negation itself; instead they stressed it, with a beat, deictic or a ring gesture, or they shifted the salience of the utterance to the negated concept by representing the action that was being negated. The (cautious) suggestion put forward here is that there might be a stronger relationship between polemic negations and negating gestures than with other types of negations. However, as this is just a case study, and the categorisation of both negations and gestures can be considered somewhat subjective, more research is needed to confirm these results.

Optimising interpersonal communication can help increase student confidence and develop their intrinsic motivation–achieved by satisfying learners' needs for competence, relatedness, and autonomy (Kerssen-Griep, 2001, p.257)–leading to better student performance. Teachers are able to build contexts that satisfy these needs through a number of communicative devices, including gestures. The concept is important and so it is often marked with a clarifying or beat gesture. Other interactive gestures are those that give and take (ideas) to and from students to build up the connections between teacher and students by encouraging students' autonomy, and their right to think for themselves, by offering new ideas to them.

Gestures emphasising not the negation itself but either the negated concept or an alternative one suggest that teachers either seek to mitigate the effect of the negation or are aware of the difficulties inherent in processing negation. This is likely to be a (perhaps unconscious) strategy devised to improve students' understanding in a context where English is a second language. Interlocutors update information on the context with every utterance and gesture. If one assumes that Grice's Maxims are being followed, the negation is processed within this context, including the gesture, and is understood as a signal to retrieve the relevant part of the utterance (Roberts, 1996). In these cases, where there is contextual support, the processing of the negation is considered to be facilitated (Giora, 2016; Nieuwland and Kuperberg, 2008). These gestures could also be seen as support to a two-phased processing mechanism, where first the negating action is activated, and then its rejection (Kaup, 2001).

Negating gestures, observed with polemic negations, are classified under various families of recurrent gestures, such as the Open Hand Prone. These gestures are thought to intensify and make the grammatical negation more explicit, supporting the results observed in speakers of Italian (Kendon, 2004), French (Calbris, 1990; Harrison and Larrivée, 2016) and English (Harrison, 2018). In most of the cases analysed, the stroke of the gesture co-occurred with the marker which also carried the prosodic stress, as noted by Harrison (2010) and Harrison and Larrivée (2016) in English and French speakers, respectively. This suggests that, when necessary, teachers are comfortable using negations, despite potential processing difficulties and face threats.

## **7. Conclusions**

Linguistic negations might all share a basic meaning of non-existence, rejection and denial. However, when considering the gestures co-occurring with them, additional contextual information becomes clearer, helping interlocutors focus on different aspects of the communicative act. Through the above examples, we have illustrated how these Hong Kong teachers communicate negative utterances, showing that they also use negations in their classroom discourse. Overall, there was a preference to use gestures with negations, the majority of these related to the negative utterance. Among the three types of negations observed (polemic, descriptive and metalinguistic), the first seemed to co-occur with negating gestures, while the latter two corresponded to gestures that stressed the negation or illustrated the negated concept. It would seem that our participants used gestures to mitigate the potential processing difficulty inherent in the negation, only using negating gestures when needing to negate previous content.

A limitation of this study is that it only considers hand gestures, following a common practice in the study of gestures, when other body parts are also recognised as playing a role in the communicative act (Müller, 2013a; Lopez-Ozieblo, 2018). Further multimodal research in this area is needed to confirm the relationships between negative markers and different functions of co-occurring gestures as well as facial expressions, gaze, head and body movements. The sample size is small and our results would be strengthened if similar results were found in a wider corpus. This study is linked to teaching, where gestures can be emphasized or added for pedagogical purposes. A comparison with a corpus where this parameter is absent would enhance our understanding of negations.

### **Acknowledgments**

Thank you to all the participants who have made this study possible and to Mr. Cyril Lam and R.A. I am also grateful to Prof. Matti Miestamo and Prof. Pierre Larrivée for their help on negations.

This study was funded by the Teaching and Learning Fund of the Department of English, Hong Kong Polytechnic University. This project was supported by the Research Centre for Professional Communication in English (RCPCE) of the Department of English of the Hong Kong Polytechnic University.

### **Endnote**

1. All images are the author's own collection and participants have given their consent for publication.

### **References**

Bolinger, D. (1989). *Intonation and its uses: Melody in grammar and discourse*. Stanford, MA: Stanford University Press.

Bressem, J. (2013). A linguistic perspective on the notation of form features in gestures. In: C. Müller, A. Cienki, E. Fricke, S. H. Ladewig, D. McNeill and S. Teβendorf, eds., *Body-Language-Communication*. Berlin: de Gruyter, pp. 1079–1098, https://doi.org/10.1515/9783110261318.1079


## **Appendix A**

Transcription Conventions (speech transcription adapted from Du Bois (1991) and gestures transcription from McNeill (2005))


## **9. Ontological Change Caused by Negation: The Case of Identity Statements**

*Tomohiro Sakai*

## **1. Introduction1**

The negation of sentence P generally has no effect on the ontological status of objects denoted by proper names occurring in P. Thus, 'François Hollande' denotes as much an individual in the negative sentence in (1b) as in the affirmative sentence in (1a).

(1) a. François Hollande is a candidate in the presidential election. b. François Hollande is not a candidate in the presidential election.

This comes as no surprise if we assume that the proposition that there is a person named 'François Hollande' is not part of the assertion made by the utterance of (1a), thus not falling within the scope of the negation in (1b). That proposition is no more the object of negation than it is the object of assertion. This line of thought dates back to Frege (1892a, p. 40; 1997, p. 163), who says "[t]hat the name 'Kepler' designates something is just as much a presupposition for the assertion 'Kepler died in misery' as for the contrary assertion".2 Following Frege, Strawson (1952, p. 213) maintains that the existential proposition corresponds to a contextual requirement whose satisfaction is "not a part of what is *asserted* by the use of a sentence"; "it is, rather, presupposed by the use of the expression" (emphasis in the original). On this view, there is an individual *h* such that (1a) is true if and only if *h* satisfies the predicate *C* ('a candidate in the presidential election'), and such that (1b) is true if and only if *h* does not satisfy *C*. (1a) and (1b) have the truth-conditions illustrated in (2a) and (2b), respectively.

(2) a. C (h) b. ¬ C (h)

#### **How to cite this book chapter:**

Sakai, T. 2023. Ontological Change Caused by Negation: The Case of Identity Statements. In: Roitman, M. (ed.) *Negatives and Meaning: Social Setting and Pragmatic Effects: Using Negatives in Political Discourse, Social Media and Oral Interaction*, pp. 239–268. Stockholm: Stockholm University Press. DOI: https://doi.org /10.16993/bcd.i. License: CC BY 4.0.

Neither (2a) nor (2b) contain the proposition that there is an individual *h*, thus fulfilling Recanati's (1993, p. 17) definition of singular proposition: "The truth-condition of an utterance *G (t)* is singular if and only if there is an object *x* such that the utterance is true if and only if *x* satisfies *G ( )*."

A major exception to this observation is existential statements.


Compare these statements to (5a) and (5b), below. It is not the case that (3a)–(4a) and (3b)–(4b) express the singular propositions illustrated in (5a) and (5b), respectively.

$$\begin{array}{c} \text{(\(\mathfrak{s}\)) a. E \ (\mathfrak{o})\\ \text{b. } \neg \to \text{ (o)} \end{array}$$

To say that there is an individual *o* such that (3a) is true if and only if *o* satisfies the predicate *E* ('exist') would leave the cognitive significance of (3a) unaccounted for. If, through the utterance of (3a), we could presuppose that there is an individual *o*, understanding what the predicate 'exist' contributes to the assertion becomes difficult (Kripke, 2013, p. 5). Even more obvious is the problem raised by the negative statements in (3b)–(4b). As Quine (1953, p. 2) says: "[i]f Pegasus were not, […] we should not be talking about anything when we use the word; therefore it would be nonsense to say even that Pegasus is not". When viewed through this lens, (5b) seems nonsensical. Nevertheless, (3b) and (4b) can be held to be true, hence meaningful, statements, thus suggesting that they do not express any singular proposition of the form (5b)3 . It seems that, contrary to the negation in non-existential statements such as (1b), negative existentials such as (3b) and (4b) are not ontology-preserving; (3b) and (4b) commit us to a poorer ontology than (3a) and (4a), respectively.

This chapter argues that this seemingly trivial observation provides an insight into the information conveyed by identity statements like (6a)–(6b).

(6) a. Émile Ajar is (identical with) Romain Gary. b. Émile Ajar is not (identical with) Romain Gary Identity statements can plausibly be analysed as covert existentials. The affirmative statement in (6a) is ontology-preserving, just like its overt counterparts in (3a)–(4a), while the negative statement in (6b) is no more ontology-preserving than its overt counterparts in (3b)–(4b). This chapter asserts, however, that there are two differences between overt and covert existentials:


These differences allow us to submit that the negation in covert existentials is pragmatic in character.

## **2. Overt existentials**

### 2.1. Descriptivism

Sentence (7) can hardly be considered to express any singular proposition, as it does not presuppose the existence of Pegasus, and is irreducible to the proposition that there is an individual *o* such that (7) is true if and only if *o* satisfies the predicate *E* ('exist').

(7) Pegasus exists (in reality).

A well-known solution to this problem is descriptivism (Russell, 1918/2010; Quine, 1953, Ch. 1), according to which a proper name N that can occur meaningfully in the sentence of the form 'N exists' "is not really a name, but a sort of truncated description" (Russell, 1918/2010, p. 79). Russell's position is eloquently articulated in the following passage:

If ["Romulus"] were really a name, the question of existence could not arise, because a name has got to name something or it is not a name, and if there is no such person as Romulus there cannot be a name for that person who is not there, so that this single word "Romulus" is really a sort of truncated or telescoped description, and if you think of it as a name you will get into logical errors. (Russell, 1918/2010, p. 79)

If (7) is a truncated version of (8), for example, then (7) expresses a general rather than singular proposition. Thus, as outlined in Russell's Theory of Descriptions (Russell, 1905), there exists one and only one x which fulfills the description "x is a winged horse".

(8) The winged horse exists (in reality).

There has been much debate about whether Frege was a descriptivist or not. As far as existential statements are concerned, there is some textual evidence in Frege's work that emanates an eminently descriptivist impression: "[t]he sentence 'There is Julius Caesar' is neither true nor false but senseless; the sentence 'There is a man whose name is Julius Caesar' has a sense, but here […] we have a concept […]" (Frege 1892b, p. 200/1997, p. 189).4 This remark is highly compatible with the idea that 'Pegasus' in (7) denotes a concept rather than an individual, and it is congenial to what Russell said more than ten years later:

["Romulus"] stands for a person who did such-and such things, who killed Remus, and founded Rome, and so on. It is short for that description; if you like, it is short for "the person who was called 'Romulus'". (Russell 1918/2010, p. 79)5

Notwithstanding this, however, the rest of this chapter assumes that descriptivism for proper names is not a viable option. There are two reasons for this. Firstly, 'Odysseus' in (3)–(4) and 'Pegasus' in (7) intuitively sound like proper names, and, secondly, these words exhibit the same grammatical properties as typical proper names: 'Odysseus' and 'Pegasus' are employed in the singular form without any determiner accompanying them, and capitalised in written English, on a par with 'François Hollande' and 'Bucephalus'. Furthermore, *Wiktionary* classifies both 'Odysseus' and 'Pegasus' as proper nouns.6 It is hardly justifiable to adhere to descriptivism at the sacrifice of these intuitive and grammatical considerations:7

Contemporary philosophers of language study language as it is rather than it ought to be: when it comes to proper names, they try to capture the characteristic features of those words which *are* called 'proper names' rather than the features of the words which *deserve* to be so-called.

(Recanati, 1993, p. 177, emphases in the original)

Likewise, the object of contemporary linguistics is the nature of knowledge of language, rather than what it ought to be. Throughout this chapter, we will therefore follow the view held by contemporary philosophers and linguists, assuming that what strikes us as proper names, really are proper names.

### 2.2. The Frege-Fauconnier solution

Once descriptivism is abandoned, there remain at least two more solutions to the problem raised by existential statements. The first solution (Solution 1) consists of viewing existentials as expressing propositions about names rather than objects. This view was once suggested by Frege: "People certainly say that Odysseus is not an historical person [= (4b)], and mean by this contradictory expression that the name 'Odysseus' designates nothing, has no meaning [= reference]" (Frege, 1969, p. 208/1979, p. 191).8 On this view, (7) expresses a proposition about the name 'Pegasus' rather than what the name denotes, meaning that the name 'Pegasus' has reference.

Another solution (Solution 2) to the puzzle comes from an entirely different discipline. From a cognitive linguistic perspective, Fauconnier (1985/1994) puts forward the view that existentials allude to objects found in domains—or, in his terminology, '(mental) spaces'—other than the speaker's reality. On this view, "to say [(7)] is to set up a counterpart of the mythical Pegasus in space M', 'in reality \_\_\_\_' (the speaker's 'reality')" (Fauconnier, 1985/1994, p. 149).

Even though these two solutions might appear to be quite different, in reality, the first solution is subsumed by the second. Adopting Fauconnier's (1985/1994) notation, we can illustrate the interpretation of (7), as outlined in Figure 9.1.

**Figure 9.1.** Pegasus exists (in reality).

According to Solution 2, defended by Fauconnier (1985/1994), (7) says that element *a* in M(myth) has a counterpart *a'* in M'(speaker's reality). According to Solution 1, suggested by Frege, (7) says that the name 'Pegasus' has reference. What should be noticed here is that 'reference' as defined by Frege can only be found in M' (reality).9 The principle of compositionality dictates that the reference of a sentence must be a function of those of its constituents and the way they are put together. Since, in the Fregean semantics, the reference of a sentence is a truthvalue and that of a proper name is an object, it follows that the sentence must be either true or false if all its constituents have references and they are correctly put together. Moreover, Frege says that sentences containing fictitious names are neither true nor false:

The sentence 'Scylla has six heads' is not true, but the sentence 'Scylla does not have six heads' is not true either; for it to be true the proper name 'Scylla' would have to designate something. […] Names that fail to fulfill the usual role of a proper name, which is to name something, may be called mock proper names. […] Instead of speaking of 'fiction', we could speak of 'mock thoughts'. Thus if the sense of an assertoric sentence is not true, it is either false or fictitious, and it will generally be the latter if it contains a mock proper name. […] Assertions in fiction are not to be taken seriously: they are only mock assertions. Even the thoughts are not to be taken seriously as in the science: they are only mock thoughts.

(Frege, 1969, pp. 141–142/1979, pp. 129–130)10

This remark entails that fictitious names (mock proper names) have no reference. As a consequence, nothing in M counts as reference in Figure 9.1, and reference, if any, can only be found in M'. Thus, element *a* in M(myth) has a counterpart *a'* in M'(speaker's reality) only if the name 'Pegasus' has reference in Frege's sense. In this way, Solution 1 is subsumed by Solution 2.

This relation equally holds when the statement in (7) is negated as in (9).

(9) Pegasus does not exist (in reality).

Under Fauconnier's analysis, (9) would construct a configuration in which element *a* in M has no counterpart in M', as illustrated in Figure 9.2.

Given the state of affairs represented by Figure 9.2, Frege would say that the name 'Pegasus' has no reference. Again, Fauconnier's analysis subsumes Frege's. It is worth noting that the way Frege employs the terms 'truth/true' and 'reference/refer' is widely—if not universally accepted in the literature. This is evidenced by the remark Burge (1973; p. 436) makes quite independently from the Fregean view of proper names: "the failure of 'Pegasus' to designate in my utterance of ['It is not the case that Pegasus exists'] follows from the fact that I referred to nothing that the proper name is true of". It is only in the space in which the utterer/thinker believes corresponds to reality that reference and truth can be fully grounded11.

**Figure 9.2.** Pegasus does not exist (in reality).

### 2.3. Generalisation on existential statements

From the observations above, we obtain the generalisations given in (10).

(10) a. Affirmative existentials are ontology-preserving. b. Negative existentials are ontology-impoverishing.

An affirmative existential constructs a configuration in which *a* in M(myth) has a counterpart *a'* in M'(speaker's reality). As a result, the ontology in M is preserved in M'. A negative existential, on the other hand, constructs a configuration in which *a* in M has no counterpart in M'. In this case, we can say that M' has a poorer ontology than M. So, if you approve the negative existential in (9), then you cannot employ 'Pegasus' to talk or think about M', hence to make any true or false assertions. Even if you accept (9), you can still employ 'Pegasus' to talk or think about M(myth), but if truth or falsity is defined in terms of the Fregean semantics, your utterance or thought about M is neither true nor false.12 Alternatively, in Frege's (1969, p. 142/1979, p. 130) and Austin's (1962, p. 22) terminology, the acceptance of (9) makes it impossible to use 'Pegasus' "seriously". As will be discussed in Section 3, this apparently trivial observation provides insight into the semantics and pragmatics of identity statements in their affirmative and negative forms.

## **3. Identity statements**

### 3.1. The morning star/evening star problem

Identity statements raise the classical Morning Star/Evening Star problem (Frege, 1892a; 1997). The problem can be formulated as follows: "how could someone possess knowledge sufficient to understand a sentence like 'Hesperus is Phosphorus' without thereby knowing, already, that the sentence is true?" (McDowell, 1977, p. 163) Suppose that the name 'Émile Ajar' denotes an individual *a*. Then the meaning of (11) might be represented as in (12).

(11) Émile Ajar is (identical with) Émile Ajar.

(12) a = a

The formula in (12) can accommodate the fact that (11) is *a priori*, in contrast to (13), which is *a posteriori* and can extend our knowledge.

(13) Émile Ajar is (identical with) Romain Gary.

How are we to account for the difference in cognitive value between (12) and (13)? The meaning of (13) can hardly be represented as in (14), where *b* purports to represent the individual denoted by the name 'Romain Gary'.

(14) a = b

Since Émile Ajar is identical with Romain Gary, *a* and *b* represent one and the same individual, making (14) equivalent to (12). Accordingly, (14) fails to account for the cognitive significance exhibited by (13). It would be useless to contend that (12) and (14) are not equivalent on the ground that (12) and (14) are of different forms, because this claim would imply that *a* and *b* represent different objects13. By definition, different objects cannot be identical with each other, as Russell and Wittgenstein remark:

Identity is a rather puzzling thing at first sight. When you say "Scott is the author of Waverley", you are half-tempted to think there are two people, one of whom is Scott and the other the author of Waverley, and they happen to be the same. That is obviously absurd, but that is the sort of way one is always tempted to deal with identity. (Russell, 1918/2010, p. 84)

Roughly speaking, to say of *two* things that they are identical is nonsense, and to say of *one* thing that it is identical with itself is to say nothing at all. (Wittgenstein, 1922, para. 5.5303, emphases in the original)14

This fact is so fundamental that it is equally assumed by more recent frameworks not necessarily inspired by analytic philosophy. Thus, within a cognitive linguistic framework, Fauconnier (1985/1994; pp. 154–155) maintains that "[i]n a single space, the identity relation expressed by *be* can never been satisfied by two distinct elements", or equivalently, that "one never establishes identity between two elements of a space". If the meaning of (13) cannot be properly represented by (14), the negative statement in (15) cannot be properly represented by (16) for exactly the same reason.


(16) would not make any sense if *a* and *b* represent one and the same individual, and it would be uninformative if *a* and *b* represent different individuals. Nevertheless, it may in principle be possible for someone to understand (15) without knowing it to be true or false; such is the gist of the classical Morning Star/Evening Star problem.

### 3.2. Solution based on sense and reference

Various attempts have been made to deal with the puzzle raised in Section 3.1. Arguably the best-known solution is the one Frege (1892a; 1997) proposes, drawing on the distinction between sense (Sinn) and reference (Bedeutung). This solution attributes the difference in cognitive value between (11) and (13) to their differing senses, traceable to the difference in sense between 'Émile Ajar' and 'Romain Gary'. These names have the same reference but have different senses, making it the case that, due to the principle of compositionality, the sentences in which they occur have difference senses.

The problem with this conception is that the notion of sense as used here has a strongly descriptivist impression. The Fregean solution at issue seems to force us to equate the cognitive significance of (13) with that of, say, the proposition that there is a man who wrote *La vie devant soi* and *Les racines du ciel*, where 'Émile Ajar' and 'Romain Gary' are considered respectively equivalent to the descriptions 'man who wrote *La vie devant soi*' and 'man who wrote *Les racines du ciel*'. Construed in this way, Frege's 'sense' is indistinguishable from Russell's 'description', as far as identity statements are concerned. Indeed, Russell (1918/2010) says:

When I say "Scott is the author of Waverley" and that "is" expresses identity, the reason that identity can be asserted there truly and without tautology is just the fact that the one is a name and the other a description. Or they might both be descriptions. (Russell, 1918/2010, p. 84)

Since, as in Section 2.1 above, we are not committed to descriptivism, we will not pursue this solution any further.

### 3.3. Solution based on concepts

Another solution suggested by Frege (1969, p. 131/1979, p. 120) says that "an object *a* is equal to an object *b* (in the sense of completely coinciding with it) if *a* falls under every concept under which *b* falls, and conversely".15 Given the identity between Émile Ajar and Romain Gary, for any predicate P, if Émile Ajar is P, then Romain Gary is also P, and vice versa. For instance, Émile Ajar was born in the Russian Empire just as Romain Gary was; Émile Ajar committed suicide in 1980 just as Romain Gary did; and so on and so forth. It is important to note that, for Frege, a concept (Begriff) is the referent of a predicate, not its sense (Morscher, 2001, pp. 236–237). This gives the solution based on concepts the advantage of not having to appeal to the sense of any expression occurring in the sentence.

Frege's condition, however, is too strict. Even when you accept (13) and (17), you do not have to accept (18a)–(18b).

	- b. #Romain Gary won the Prix Goncourt earlier than Romain Gary.

The identity statement in (13) can be true even though *Romain Gary* falls under the concept "won the Prix Goncourt earlier than Émile Ajar", but *Émile Ajar* does not.

### 3.4. The fauconnier-recanati analysis

Fauconnier (1985/1994) proposes a substantially different analysis, which rests on neither 'sense' nor 'concept'. Under this analysis, "the final effect of [an identity statement] is to replace two elements in the origin space R0 by a single one in the target space R" (Fauconnier, 1985/1994, p. 154). Applied to (19), it gives the configuration illustrated in Figure 9.3. 16

(19) Clark Kent is (identical with) Superman.

In R0 , *a* and *b* are distinct elements. Those who endorse the ontology represented by R0 believe the negative proposition in (20).

(20) Clark Kent is not (identical with) Superman.

**Figure 9.3.** Clark Kent is (identical with) Superman. (tentative)

In R, by contrast, *a* and *b* are merged into a single element *b'*. Those who endorse the ontology represented by R believe the affirmative proposition in (19). In Fauconnier's (1985/1994) framework, "negatives set up corresponding counterfactual spaces in which the positive version of the sentence is satisfied" (Fauconnier 1985/1994: 96). Since R0 is the negation of R, and vice versa, R0 is a counterfactual space if you endorse R, while R is a counterfactual space if you endorse R0 . Figure 9.3 illustrates the case where R is construed as the reality space and R0 as a counterfactual space.17

A similar analysis is proposed by Recanati (2012), who maintains that "[t]o accept the identity 'A = B' is to link the two [mental] files corresponding to the terms on each side of the equals sign" (ibid., p. 44), and that "their [= encyclopedia entries = mental files associated with proper names] multiplicity could only reflect the mistake of thinking that there are two objects where there is one" (ibid., p. 47). In this scenario, R0 in Figure 9.3 represents the mistake which the speaker or thinker once committed. At this stage, the speaker/thinker assents to the negative proposition in (20). When she recognises her mistake, she comes to entertain the ontology represented by R, dismissing R0 as a misrepresentation (cf. Bergson, 1907/1941, p. 287).

Although intuitively adequate, the Fauconnier-Recanati analysis as such is question-begging, because it presupposes the notion of identity. The replacement of *a* and *b* in R0 by *b'* in R is a result of our understanding the identity between *a* and *b*, rather than what constitutes the identity. Why do we not construct a configuration analogous to Figure 9.3 for, say, (21)?

(21) Donald Trump is (identical with) Emmanuel Macron.

The answer is obvious: we know (19) to be true and (21) to be false. But there is nothing in Figure 9.3 that prevents the configuration from applying to (21). The question to be asked here is what constitutes the notion of identity. Answering this question by appealing to the notion of replacement is, at best, question-begging.

Besides the theoretical difficulty just mentioned, the Fauconnier-Recanati analysis also raises an empirical problem. On their view, the negative identity statement in (20) holds in R0 , while the affirmative identity statement in (19) holds in R. By assenting to (19), *a* and *b*, distinct elements in R0 , are replaced by a single element in R. This observation would lead us to the generalisation in (22).

(22) [Tentative generalisation] Affirmative identity statements are ontologyimpoverishing.

As noted previously in (10a), affirmative existentials are ontologypreserving. Comparison of (22) and (10a) suggests that, on the Fauconnier-Recanati view, existential statements and identity statements are different with respect to the ontology which accepting them allows us to embrace. Their view predicts that accepting the existential statement in (7) (= 'Pegasus exists (in reality)') allows us to talk about 'Pegasus in reality', whereas accepting the identity statement in (19) prevents us from talking truthfully about Superman and Clark Kent at the same time. As said above, R0 is a counterfactual space if you endorse R, while R is a counterfactual space if you endorse R0 . Those who assent to (19) should then only be able to talk about Superman and Clark Kent counterfactually. This prediction is not borne out, however. Even if you accept (19), hence the alleged impoverished ontology represented by R, you can still hold (23) to be true.

(23) Superman leaps more tall buildings than Clark Kent. (Braun and Saul, 2002, p. 1)

Lois Lane does not have to discard the belief expressed by (23) when she becomes aware of the identity between Clark Kent and Superman. Put briefly, (23) can remain true regardless of whether the identity holds or not. Nevertheless, the configuration illustrated in Figure 9.3 does not allow us to talk truthfully about Superman leaping more tall buildings than Clark Kent, or about Superman being more popular among women than Clark Kent, since R, the only space in which the question of truth or falsity arises, contains only one element. All that the configuration allows us to assert is (24a) and (24b).

(24) a. Lois Lane once mistakenly believed that {Superman leaped more tall buildings than Clark Kent/Superman was more popular among women than Clark Kent}. b. It is not true that {Superman leaps more tall buildings than Clark Kent/Superman is more popular among women than Clark Kent} (, because they are one and the same person).

This runs counter to our intuition that (23) can remain true even after the truth of (19) is recognised. The compatibility between (19) and (23) is therefore at odds with the view that affirmative identity statements are ontology-impoverishing.

### 3.5. Identity statements as covert existentials

Given the apparent compatibility between (19) and (23), the truth condition of identity statements in (25) suggests itself.

(25) The identity statement 'a is (identical with) b' is true if and only if there is an individual c such that the entities denoted by 'a' and 'b' are (different) aspects of c.

(25) says that the entities denoted by 'Clark Kent' and 'Superman' are preserved even if (19) is embraced as true. What (19) contributes to the configuration is a new element *c*, which has those entities as aspects, as illustrated in Figure 9.4.

**Figure 9.4.** Clark Kent is (identical with) Superman. (revised)

Under the revised analysis, as well as under the original Fauconnier-Recanati analysis, the negative identity statement in (20) (= 'Clark Kent is not (identical with) Superman') holds in R0 , while the affirmative identity statement in (19) (= 'Clark Kent is (identical with) Superman') holds in R. Under the revised analysis, however, R has a richer ontology than R0 . Instead of having a single element as in Figure 9.3, R has a new element *c* in addition to *a'* and *b'*, inherited from R0 . The dashed lines linking *c*, *a'* and *b'* represent individual-aspect relations such that *a'* is an aspect of *c* and that *b'* is another aspect of *c*. Identity statements can now be viewed as covert existentials, in that accepting (19) amounts to claiming the existence of *c*, in addition to *a'* and *b'*.

If you endorse R0 as opposed to R, you can talk about *a* and *b*, but not about *c*. If, on the other hand, you endorse R, you can talk about *c* as well as *a'* and *b'*. This enables us to obtain the generalisation in (26).

(26) a. Affirmative identity statements are ontology-enriching. b. Negative identity statements are ontology-impoverishing.

This generalisation captures the fact that there exits an element which affirmative identity statements, but not negative identity statements, allow us to talk truthfully about. Otherwise, both can commit us to the same ontology. For this reason, while both allow us to talk truthfully about Clark Kent and Superman, it is only when we approve the affirmative identity statement in (19) that we can truthfully talk about that individual who is sometimes Clark Kent, and sometimes Superman.

It is important to note that (26) is subsumed by (10) above. On one hand, as we claim here, identity statements are covert existentials. Affirmative and negative identity statements are a particular kind of affirmative and negative existentials, respectively. On the other hand, being ontology-enriching is a special case of being ontology-preserving. In general, the ontology of space Mx is preserved in space My only if (i) every element of Mx has a counterpart in My ; and (ii) My contains at least as many elements as Mx . The conjunction of (i) and (ii) allows for the possibility that My has a richer ontology than Mx . There is thus no obstacle to saying that, in Figure 9.4, the ontology of R0 is preserved in R. This confirms that the generalisation in (10) applies to all existentials, whether they be overt or covert.

The analysis proposed here has several advantages. First, it accounts for the cognitive value of identity statements without being question-begging, as was the case with the original Fauconnier-Recanati analysis discussed in Section 3.4. In Figure 9.4, the discovery of an individual *c* such that *a* and *b* are different aspects of *c* does not require a prior understanding of *a = b*. To posit *c* is merely to suppose the identity between *a* and *b*, the former not being the result of the latter in any sense.

Second, the present analysis enables us to understand why Fauconnier (1985/1994, Ch. 5) treats both 'be' and 'exist' as trans-spatial operators. The trans-spatial character of these verbs is a consequence of them being existential predicates. In general, when P is an existential predicate, P introduces a new element into a space. This new element must be in a different space from the space in which the element denoted by the subject of P is found. If they were in the same space, predicating P of the element denoted by the subject would be either tautological or contradictory. If, on one hand, P reintroduced into the original space the element denoted by the subject, the predication would not contribute anything new. If, on the other hand, P introduced into the original space another element identical to the element denoted by the subject, the predication would amount to saying, in a contradictory manner, that there are two distinct elements which are identical with each other. It then follows that existential predicates are necessarily trans-spatial operators.

Third, the analysis proposed here allows us to account for the fact that the construal of (23) varies with the ontology one embraces. Those who believe (20) construe (23) as saying that individual X leaps more tall buildings than individual Y, while those who believe (19) construe it as saying that individual Z leaps more tall buildings when Z is X than Z does when Z is Y (Sakai, 2018, p. 216). Put differently, those who deny the identity between Clark Kent and Superman construe (20) as a singular proposition about two distinct individuals, whereas those who accept the identity at issue construe it as a proposition about two different aspects of one and the same individual. This difference in construal can be accommodated by Figure 9.4, but not by Figure 9.3.

Finally, this proposal captures the generalisation that 'a is identical with b' is true if and only if, in the singular construal of 'a' and 'b', 'b' can be substituted for 'a' and vice versa *salva veritate* (Wittgenstein, 1922, para. 6.24; Sakai, 2017, p. 21). Suppose Lois Lane believes (27).

(27) Superman wears a cape and leaps tall buildings.

When she becomes aware of the identity between Clark Kent and Superman, she utters (28).

(28) Wow, so sometimes Clark Kent wears a cape and leaps tall buildings! (Braun and Saul, 2002, p. 6)

As is clearly seen here, accepting (19) enables her to replace 'Superman' in (27) by 'Clark Kent', resulting in the utterance in (28). This relation between (27) and (28) stems from the possibility for both 'Superman' and 'Clark Kent' to identify *c* in Figure 9.4. 'Superman' denotes *a'*, and, by metonymy, serves to identify *c*. 'Clark Kent' denotes *b'*, and, by metonymy, serves to identify *c*. Consequently, the proposition that Superman is P is equivalent to the proposition that Clark Kent is P, insofar as both names are employed to identify the same individual *c*. This equivalence would not arise if you assented to the negative proposition in (20), hence disbelieved in the existence of *c*. We may note in this connection that the equivalence between (27) and (28) only holds when the subject nominals are interpreted as referring to an individual rather than aspects of an individual, i.e., the sentences are interpreted as expressing a singular proposition. (28) is obviously false if 'Clark Kent' is meant to refer to *b'* in Figure 9.4. By the same token, Lois Lane, who believed (23) above, would not utter (29) or (30), even after recognising the identity between Clark Kent and Superman.


Since the singular construal of (23) leads to the contradictory proposition "*c* leaps more tall buildings than *c*", (23) can only be interpreted as a proposition about aspects, namely as saying that *a'* leaps more tall buildings than *b'*. This is the reason why the substitution fails in (29)–(30).

In summary, unlike the original Fauconnier-Recanati analysis, the analysis defended here can account, among others, for the following three facts at the same time:


## **4. The pragmatics of negation**

### 4.1. The nature of terms occurring in existential statements

To summarise the argument so far: existential statements such as (31a) and (31b) exhibit the properties shown in (32).

	- b. Pegasus does not exist (in reality).

(31a) introduces into the reality space R a counterpart of mythical Pegasus, the ontology of the myth space R0 being preserved in the reality space. In contrast, (31b) denies the existence of such a counterpart element in the reality space R, so that the real world has a poorer ontology than the mythical world. In a nutshell, (31a) commits us to a richer ontology than (31b). This is what is meant by the generalisation in (32).

Also argued was that identity statements such as (33)–(34) are covert existentials, and no less exhibit the properties in (32) than overt ones.

	- b. Clark Kent is not (identical with) Superman.

(33a) introduces into the reality space R counterparts of Émile Ajar and Romain Gary, and in addition, an individual which has Émile Ajar and Romain Gary as aspects. In this case, the reality space R preserves the ontology of the past space R0 and adds another element. By contrast, (33b) denies the existence of such an individual in the reality space. Accordingly, (33a) commits us to a richer ontology than (33b). While both (33a) and (33b) allow us to talk truthfully about Émile Ajar and Romain Gary, it is only when we accept (33a) that we can truthfully talk about that individual whose name is Romain Gary and who sometimes writes under the pseudonym 'Émile Ajar'. The same holds for (34). (34a) claims that there is an individual *c* such that *c* has Clark Kent and Superman as different aspects. That individual is sometimes Clark Kent and sometimes Superman. (34b) denies the existence of such an individual and claims that Clark Kent and Superman are two distinct individuals.

These observations are unsurprising, since affirmative existentials can naturally be considered to serve to introduce a new element into reality, while negative existentials can plausibly be considered to prevent any new element from coming up. However, when regarding the status of the entity whose existence is asserted or denied, there is a crucial difference between overt and covert existentials. In overt existentials, this entity is explicitly named by a term occurring in the statements. For example, (31a) asserts the existence of an element named by the subject nominal 'Pegasus', while (31b) denies its existence. Accepting (31b) rather than (31a) would then make it impossible to talk truthfully about the entity named by 'Pegasus'. This is indeed the case, as confirmed by the fact that those who believe (31b) cannot judge (35) to be true.

(35) Pegasus leaps more tall buildings than Bucephalus.

For them, Bucephalus is the horse of Alexander the Great—hence a real horse—whereas Pegasus is not a genuine object. A statement like (35) can only be made by those who embrace the ontology expressed by (31a)18. One can say that the difference between the ontology represented by (31a) and that represented by (31b) is semantic in nature, in that (31b) serves to deprive 'Pegasus' of its reference. The negative existential of the form 'N does not exist' affects the semantic potential of N; if one assents to the negative existential, one cannot use N to make serious assertions. This observation is accommodated by Figure 9.2 above, where there is no element corresponding to N in M'.

The same is not true of covert existentials. In covert existentials, the entity whose existence is in question is not explicitly named by any term occurring in the statements. For example, (33a) asserts the existence of an element which is not named by any expression in the sentence, while (33b) denies its existence. Explicitly named by the terms in (33a)–(33b) are only Émile Ajar and Romain Gary. That individual that is supposed to have Émile Ajar and Romain Gary as aspects is not named by any term in (33a)–(33b). Accordingly, whether you believe (33a) or (33b), you can make serious assertions about Émile Ajar and Romain Gary, as in (36).

(36) [= (17)] Romain Gary won the Prix Goncourt earlier than Émile Ajar.

The same holds for (34). Whether you believe (34a) or (34b), you can talk truthfully about Clark Kent and Superman, as in (37).

(37) [= (23)] Superman leaps more tall buildings than Clark Kent.

In this sense, the negative existential of the form 'N1 is not N2 ' has no effect on the semantic potential of N1 and N2 ; even if one assents to the negative existential, one can continue to use N1 and N2 to make serious assertions. This observation is captured by Figure 9.4 above. In both R, which has a richer ontology, and R0 , which has a poorer ontology, there are elements corresponding to N1 (*a*) and N2 (*b*). The negation of a covert existential of the form 'N1 is not N2 ' only affects element *c*, which is named neither by N1 nor by N2 . Thus, there is a sense in which the difference between the ontology represented by (33a) and that represented by (33b) is pragmatic in nature.

One potential objection here is that the analysis developed in this section entails that pairs of terms such as Dr. Jekyll/Mr. Hyde or Lenin (Ленин)/Vladimir Ilyich Ulyanov (Владимир Ильич Ульянов) will serve merely to refer to aspects of individuals, who cannot be named at all as such, but only as conceived from a particular perspective. This is counterintuitive, and we might want to say that, unlike 'Superman', 'Dr. Jekyll' and 'Lenin' are names for individuals. It is important to note, however, that the argument put forward here does not entail that individuals can never be named as such. The gist of the argument is that we have equipped with the ability to think about the same object (taken in the widest sense of the term, as Frege (1892a, p. 27; 1997, p. 153) puts it) either under an 'individual-mode' or under an 'aspect-mode', to employ Crimmins's (1998) terminology. The availability of these two modes accounts for the possibility of communication between what Braun and Saul (2002) call 'enlightened' and 'unenlightened' subjects, namely between those who are aware that Dr. Jekyll is Mr. Hyde and those who are not. For example, no one would have difficulty taking (38) to be a statement about an individual.

(38) Dr. Jekyll lives in a well-appointed house in London.

This remains the case even when it turns out that Dr. Jekyll is Mr. Hyde. Therefore, 'Dr. Jekyll' in (38) can plausibly be considered to name an individual. In contrast to (38), (39) is interpreted differently for enlightened and unenlightened subjects.

(39) Dr. Jekyll is more attractive than Mr. Hyde.

Enlightened subjects can interpret (39) in a coherent manner only by activating the 'aspect-mode' and taking 'Dr. Jekyll' to name an aspect of one and the same individual. These observations establish that a name can name either an individual or an aspect, depending on the context19.

Another objection that might be raised is that the analysis developed here only applies to utterances which are objectively truth-evaluable, insofar as it rests on such notions as reference or truth/falsity – notions that play little part in contemporary pragmatics launched by Austin (1962). What has been said so far, however, can be extended to cover utterances which are not objectively truth-evaluable, such as (40).

(40) France is hexagonal.

Austin (1962, p. 142) maintains that (40) is a rough description rather than a true or false one. Particularly relevant to our discussion is an utterance like (41).

(41) Superman is more popular among women than Clark Kent.

'Popularity' is so subjective a notion as to be rigorously estimated. The definition of 'woman' is even more complicated and dependent upon the political, social or religious background against which it is placed. What is essential here is the fact that, regardless of whether (41) may be objectively truth-evaluable or not, one can believe/assert it on some occasion or other. In general, the act of asserting that P is not intrinsically tied to any felicity condition that refers to the verifiability of P, all that can be said being that the assertion that P implies the belief that P (Grant, 1958; Austin, 1962, p. 49). It is certainly obvious that our beliefs include ones that are hardly truth-evaluable, such as (40)–(41), but it is still the case that those who utter (40)–(41) are deemed to believe the propositions expressed by these sentences. Provided that these points are conceded, our argument raises no difficulty; whether or not Superman is identical with Clark Kent has no bearing upon the belief or assertion expressed by (41). The truth-evaluability of (41) is orthogonal to the pragmatic character of the negation in (34b) (= 'Clark Kent is not (identical with) Superman').

### 4.2. Modes of presentation of objects

We have said that the negation of a covert existential of the form 'N1 is not N2 ' only affects element *c*, which is named neither by N1 nor by N2 . There is more to the effect of the negation, however. As Figure 9.4 indicates, in R0 , *a* and *b* are individuals, whereas in R, *a'* and *b'* are aspects of *c*. Following Braun and Saul (2002), let us call those who believe (33a) (= 'Émile Ajar is (identical with) Roman Gary') and (34a) (= 'Clark Kent is (identical with) Superman') 'enlightened' subjects, and those who believe (33b) (= 'Émile Ajar is not (identical with) Roman Gary') and (34b) (= 'Clark Kent is not (identical with) Superman') 'unenlightened' subjects. The contrast between R0 and R in Figure 9.4 accommodates the fact that unenlightened subjects construe N1 and N2 as naming two distinct individuals, whereas enlightened ones construe the same terms as naming two distinct aspects of one and the same individual (Sakai, 2018). Following this, we must grant that the negation in 'N1 is not N2 ' not only denies the existence of *c*, but also affects the ontological status of *a*/*a'* and *b*/*b'*, elements explicitly named by N1 and N2 . This ontological change caused by the negation in 'N1 is not N2 ' might appear to conflict with the generalisation given in (32). Unenlightened subjects are committed to the existence of two individuals, while enlightened ones are committed to the existence of only one individual. In other words, if one accepts the affirmative identity statement, the ontology of individuals in the new space is impoverished by comparison to the original space. Conversely, if one accepts the negative identity statement, the ontology of individuals in the new space is thereby enriched.

At first blush, this difference between *a*/*b* (*qua* individuals) and *a'*/*b'* (*qua* aspects) may appear to threaten the generalisation in (32). The nature of the difference, however, deserves more careful examination. In fact, there is evidence that the ontology remains the same whether *a*/*b* and *a'*/*b'* may be individuals or aspects, and that the difference lies in the construal of the same ontology, not in the ontology *per se*. How many objects are in a space defines the ontology of that space, but whether those objects are individuals or aspects does not; rather, it constitutes the mode of presentation of the same ontology.

The first point to be noted is that enlightened subjects can talk about Superman and Clark Kent with unenlightened ones. For example, they both can fully agree that (37) is a true statement, without there being any misunderstanding between them. Their commitment to different ontologies does not hinder communication. It is often assumed that, when communication is successful, some thought (or proposition) must be shared between speaker and hearer, as illustrated by the fact that if John successfully communicates to Mary the thought expressed by the utterance 'It is raining in Stockholm', Mary comes to entertain the thought expressed by that utterance. Even if it is conceded that some thought (or proposition) must be shared between speaker and hearer in successful communication, it does not follow that the very thought expressed by the speaker must be shared by the hearer. Suppose that John says to Mary, "I am hungry". It is obvious that communication fails if Mary comes to entertain, on her part, the thought expressed by the utterance 'I am hungry', because this would make it the case that Mary now thinks that she rather than John is hungry (Recanati, 2016, p. 111). In that case, the shared belief must be 'John is hungry', not 'I am hungry'. In John's thought, John is presented as 'I', namely in the first-person. But this mode of presentation must not be shared by Mary, who is expected to understand John's first-person utterance 'I am hungry' as expressing the third-person thought that John (or he) is hungry. 'I am hungry' is a different thought from 'He is hungry' if thoughts contain modes of presentation of objects, as Frege (1892a/1960) extensively argues. Even in that case, however, something must be shared between speaker and hearer in order for communication to be successful. For some philosophers, this something is a singular proposition, namely a proposition about one or several individuals. Thus, Perry (1988, p. 5; 1993, p. 231) contends that "[o]ne reason we need singular propositions is to get at what we seek to preserve when we communicate with those who are in different contexts". This idea accounts for the fact that, when Mary understands John's utterance 'I am hungry', she entertains the singular thought that John is hungry, in which no mode of presentation of John figures.

In light of the possibility of communication between enlightened and unenlightened subjects, however, Perry's idea is too stringent. It is not simply the case that there is any singular thought that both enlightened and enlightened subjects come to entertain when they understand (36)–(37). Whereas these utterances express singular thoughts for unenlightened subjects, they express aspectual thoughts for enlightened subjects. As Wittgenstein (1922, para. 4.126) puts it, "[t]he name shows that it designates an object"20. But the name does not linguistically encode whether the object designated is an individual or an aspect. If it did, there would always be misunderstanding between enlightened and unenlightened subjects when they talk about Émile Ajar or Superman. Intuitively, however, a proper understanding of (36), for example, only requires that one entertain the 'objectual' thought that the object referred to by 'Romain Gary' won the Prix Goncourt earlier than the object referred to by 'Émile Ajar'. This is consonant with the view, advanced by Bezuidenhout (1997, p. 217), that "[t]he conception the speaker has of the name-bearer doesn't have to be shared by the audience to any great degree". As another example, someone who takes 'Émile Ajar' to be the name of a rock, or 'Superman' to be the name of an aspect of a planet, for instance, would not be deemed to have correctly understood (36)–(37). There is a minimum level of understanding, which the linguistic community expects competent speakers to have (Putnam, 1975, p. 168). This does not entail, however, that there needs be such a thing as *the* thought that all subjects, enlightened or not, must entertain in order to properly understand an utterance:

[I]t is not necessary in addition to the speaker-relative content and the listener-relative content to posit some non-relative notion of utterance content in order to account for successful communication. Successful communication requires only a contextually determined degree of similarity between speaker-relative and listener-relative content. (Bezuidenhout, 1997, p. 222; see also Recanati, 2016, p. 119)

Enlightened and unenlightened subjects stand in different epistemic relations (in a broad sense) to the objects, giving rise to their different construals of the objects, i.e., objects *qua* aspects and objects *qua* individuals, respectively. But this difference can safely be relegated to the sphere of modes of presentation, rather than the ontology *per se*. It is also worth noting that the modes of presentation at issue here are not Fregean descriptions, namely object-independent senses, as in Section 3.2 above, but rather what Recanati (2012; 2016) calls non-descriptive modes of presentation, based on the subject's contextual or epistemically rewarding relations (ER relations) to the reference – namely "relations to entities which make information flow possible between the subject and these entities" (Recanati, 2016, p. 71).

In conclusion, we can maintain the generalisation that affirmative existentials are ontology-preserving while negative existentials ontology-impoverishing. The negation in 'N1 is not N2 ' leaves the references of the terms N1 and N2 intact, but alters the way their references are presented or thought about. Again, the pragmatics of negation comes into picture.

## **5. Conclusion**

Identity statements such as 'Émile Ajar is (identical with) Romain Gary' are covert existentials in that—just like overt existentials such as 'Pegasus exists (in reality)' —they bring a new element into the representation of the real world. In the representation thus obtained, there is an element which only affirmative existentials, overt or covert, allow us to seriously consider. It is only through the acceptance of 'Pegasus exists (in reality)' that we can make serious assertions about Pegasus. The sentence 'Pegasus leaps more tall buildings than Bucephalus' is judged to be true only by those who accept 'Pegasus exists (in reality)'. Similarly, it is only through the acceptance of 'Émile Ajar is (identical with) Romain Gary' that we can talk seriously about that individual whose name is Romain Gary and who sometimes writes under the pseudonym 'Émile Ajar'. The sentence 'Romain Gary is the only person to have won the Prix Goncourt twice (under two names)' is judged to be true only by those who accept 'Émile Ajar is (identical with) Romain Gary'.

Although affirmative existentials, overt or covert, commit us to the existence of an element that negative existentials fail to refer to, there are two senses in which the opposition between affirmative and negative existentials is pragmatic in character when they are covert ones. First of all, in the case of overt existentials, the opposition between affirmative and negative statements concerns the existence of an entity named by a term occurring in the statements. Thus, the debate between 'Pegasus exists (in reality)' and 'Pegasus does not exist (in reality)' centres around the existence of Pegasus, an entity named by the subject nominal of each statement. This is not the case with covert existentials, where the opposition between affirmative and negative statements concerns the existence of an entity that is not named by any terms occurring in the statements. Thus, the debate between 'Émile Ajar is (identical with) Romain Gary' and 'Émile Ajar is not (identical with) Romain Gary' centres around the existence of the individual who wrote under these two names. Such an individual, if any, is not named by any terms occurring in the statements. Both sides agree that there is an entity named 'Émile Ajar' and that there is another entity named 'Romain Gary'. These two entities are explicitly named by the terms in both 'Émile Ajar is (identical with) Romain Gary' and 'Émile Ajar is not (identical with) Romain Gary'. The divergence resides in whether there is, in addition, an entity that subsumes Émile Ajar and Romain Gary. That entity falls outside what is said by, or the explicature of, both affirmative and negative statements.

There is another sense in which the opposition between affirmative and negative existentials is pragmatic when they are covert ones. Negation in overt existentials 'N does not exist (in reality)' deprives N of its reference. For this reason, if one accepts 'Pegasus does not exist (in reality)', one cannot use 'Pegasus' to make true assertions, insofar as truth is defined as in the Fregean framework. Such utterances as 'Pegasus flies' are viewed at most as fictitious.21 Negation in covert existentials in 'N1 is (not identical) with N2 ', on the other hand, has no effect on the references of N1 and N2 , merely affecting their modes of presentation. Those who approve "Émile Ajar is not (identical with) Romain Gary" can nevertheless use 'Émile Ajar' and 'Romain Gary' to make true assertions such as 'Romain Gary won the Prix Goncourt earlier than Émile Ajar'. The way one thinks about the references, however, is different than when one embraces the ontology represented by 'Émile Ajar is (identical with) Romain Gary'. In the negative, the references of the terms are presented as two distinct individuals, while in the affirmative, they are presented as two distinct aspects of one and the same individual. Whether one may assent to the affirmative or the negative, the truth-conditional content of 'Romain Gary won the Prix Goncourt earlier than Émile Ajar' remains the same – but the same content is presented in different manners, depending on the ontology one endorses.

## **Endnotes**

1. My work on this chapter was supported by JSPS KAKENHI Grant Number 22K00540.

2. "Dass der Name "Kepler" etwas bezeichne, ist vielmehr Voraussetzung ebenso für die Behauptung "Kepler starb im Elend" wie für die entgegengesetzte." (Frege, 1892a, p. 40)

3. In current philosophical logic, it is more common to make use of the formulas in (ia) and (ib), instead of (5a) and (5b), respectively (Morscher, 2001, pp. 240–242).

(i) a. ∃x (x = o) b. ¬∃x (x = o)

This move does not amend the situation, however. Insofar as 'o' is an individual constant, (ia) is always true, while (ib) is always false, given the standard semantics of predicate logic.

4. "Der Satz "es gibt Julius Cäsar" ist weder wahr noch falsch, sondern sinnlos, wiewohl der Satz "es gibt einen Mann mit Namen Julius Cäsar" einen Sinn hat; aber hier haben wir [...] einen Begriff […]." (Frege, 1892b, p. 200)

5. Bach (2010, p. 60) sees no fundamental difference between Frege's sense and Russell's description when he says: "[Some philosophers] suppose that we can think of things only under descriptions, only by entertaining general propositions, or via something like Fregean senses, which though objectdetermining are object-independent."

6. https://en.wiktionary.org/wiki/Odysseus (Last accessed 20/01/2023) https:// en.wiktionary.org/wiki/Pegasus (Last accessed 20/01/2023)

7. Another worry is that, philosophically, descriptivism seems to entail a highly qualitative view of the world. If what we take to be proper names were not really proper names, "we would never be related in thought to anything in particular" (Bach, 2010, p. 39).

8. "Man sagt wohl, Odysseus sei keine geschichtliche Person, und meint mit diesem widerspruchsvollen Ausdrucke, dass der Name "Odysseus" nichts bezeichnete, keine Bedeutung habe." (Frege, 1969, p. 208)

9. Fauconnier (1985/1994), *qua* linguist, talks about speaker's reality, whereas Frege, *qua* logician, talks about reality *tout court*. We will set aside this difference here.

10. "Der Satz "Die Skylla hat sechs Rachen" ist nicht wahr; aber auch der Satz "Die Skylla hat nicht sechs Rachen" ist nicht wahr; denn dazu wäre nötig, dass der Eigenname "Skylla" etwas bezeichnete. […] Namen, die den Zweck verfehlen, den ein Eigenname zu haben pflegt, nämlich etwas zu benennen, mögen Scheineigennamen heißen. […] Statt "Dichtung" könnten wir auch "Scheingedanke" sagen. Wenn der Sinn eines Behauptungssatzes also nicht wahr ist, so ist er entweder falsch oder Dichtung, und dies letzte ist er im Allgemeinen, wenn ein Scheineigenname darin vorkommt. […] Die Behauptungen sind in der Dichtung nicht ernst zu nehmen: es sind nur Scheinbehauptungen. Auch die Gedanken sind nicht ernst zu nehmen wie in der Wissenschaft: es sind nur Scheingedanken." (Frege, 1969, pp. 141–142)

11. We are not claiming here that truth and falsity are meaningless concepts in fiction. The concepts of truth and falsity obviously play a part in accounting for the difference between (i), which sounds true, and (ii), which seems to be false even in the Greek mythology.


It is important, however, to keep apart 'truth *tout court*' and 'truth in fiction' (Lewis 1978). (i) is certainly true in fiction, but not true in the strict sense of the word. (ii) seems to be false because the world it depicts is less similar to the real world than the one (i) depicts. Being a horse and having wings are both properties of an animal, while having propellers and an engine is not. This makes it more difficult to construe 'Pegasus' in (ii) as a fictitious horse.

12. Frege (1969, p. 211/1979, p. 194) says: "Thoughts in myth and fiction do not need to have truth-values. A sentence containing a meaningless proper name is neither true nor false; if it expresses a thought after all, then that thought belongs to fiction. In that case the sentence has no meaning [= reference]." ["Die Gedanken in Sage und Dichtung brauchen keinen Wahrheitswert zu haben. Ein Satz, der einen bedeutungslosen Eigennamen enthält, ist weder wahr noch falsch; der Gedanke, den er etwa ausdrückt, gehört der Dichtung an. Der Satz hat dann keine Bedeutung."] This classic conception of truth is challenged, among others, by Lewis (1978), an early

attempt to define 'truth in fiction'. Even if we accept Lewis's conception of truth, it is still the case that those who accept (9) cannot talk about 'Pegasus in reality', unlike whose who assent to (7). In this respect, (9) can be viewed as ontology-impoverishing.

13. One might say that 'Superman is Clark Kent' means that 'Superman' and 'Clark Kent' are two different referential processes and that the identity statement concerns the reference of signs rather than the ontology of objects. But this idea does not work for sentence (23), 'Superman leaps more tall buildings than Clark Kent', which will be discussed below. This sentence is presented as talking about two different entities, rather than two different referential processes. It would not make sense to say that 'referential process *a* leaps more tall buildings than referential process *b*'. What is crucial is that, for unenlightened speakers, who are not aware of the identity between Superman and Clark Kent, (23) is indistinguishable (in the relevant sense here) from sentences like 'Superman leaps more tall buildings than Emmanuel Macron'. When uttering (23), unenlightened speakers believe that there are two different individuals, just as when enlightened speakers utter 'Superman leaps more tall buildings than Emmanuel Macron'. For unenlightened speakers, it is not only that 'Superman' and 'Clark Kent' are two different signs, but it is also that Superman and Clark Kent are two different individuals. Mental space configurations given in the text should therefore be interpreted as representing objects, and not linguistic signs. This is not to say that the notion of referential processes is irrelevant to the discussion. In Section 4.2, it will be argued that Superman *qua* individual and Superman *qua* aspect correspond to different referential processes or modes of presentation.

14. "Beiläufig gesprochen: Von *zwei* Dingen zu sagen, sie seien identisch, ist ein Unsinn, und von *Einem* zu sagen, es sei identisch mit sich selbst, sagt gar nichts." (Wittgenstein, 1922, para. 5.5303)

15. "Wir sagen, ein Gegenstand *a* sei gleich einem Gegenstand *b* (im Sinne des völligen Zusammenfallens), wenn *a* unter jeden Begriff fällt, unter den *b* fällt, und umgekehrt." (Frege 1969, p. 131)

16. The following discussion proceeds as if Clark Kent and Superman were real people. We will assume that R is not an imaginary world, but the reality.

17. On the most natural interpretation, (19) asserts what is supposed to hold only in the Superman story. On this interpretation, (19) is taken to be equivalent to the 'paratextual' or 'parafictional' statement in (i) (García-Carpintero, 2014, p. 17; Recanati, 2018).

(i) In the Superman story, Clark Kent is (identical with) Superman. R is thereby included in a larger fiction space, together with R0. Even in that case, however, the assertion of (19) presents R, but not R0, as real, which

makes it adequate to distinguish 'truth in fiction' and 'falsity in fiction' (Lewis 1978). The structure of language by itself does not tell us whether a sentence talks about truth *tout court* or truth in fiction, or equivalently, whether a name occurring in a sentence is a genuine proper name or a fictitious name. This is, in our opinion, part of the reason why Frege (1969, pp. 141–142/ 1979, pp. 129–130) called fictitious names 'mock proper names', as we have seen in 2.2 above. As Perry (1977, p. 477/1993, p. 6) points out, Frege took the structure of language to be suggestive but not sure guide to the structure of thought. In this chapter, we will ignore the larger fiction space in which both R and R0 may be included, because its possible existence has no bearing upon the discussion. As Kripke (2013, p. 23) cautions, "[t]he existence of fiction is a powerful argument for absolutely nothing".

18. One might wonder how one can meaningfully utter a sentence like (i). (i) Among famous horses, I prefer Pegasus to Bucephalus, because it leaps more tall buildings.

Two remarks can be made here. On the one hand, it is fictitiously true, but not true *tout court*, that Pegasus leaps more tall buildings than Bucephalus. On the other hand, such verbs as 'prefer' or 'like' are intensional verbs whose arguments may have no extensions in the real world. Thus, there may be as many fans of Sherlock Holms as there are fans of Napoleon Bonaparte. These points suggest that the utterer of (i) is not truthfully talking about Pegasus leaping more tall buildings than Bucephalus.

19. One might say, following Crimmins (1998), that enlightened subjects are merely pretending that there are two distinct individuals, when they utter or interpret (ii). Further research is called for, however, in order to determine whether there is any substantial difference between the analysis developed here and the 'pretense' account offered by Crimmins (1998).

```
20. "Der Name zeigt, dass er einen Gegenstand bezeichnet […]" (Wittgenstein, 
1922, para. 4.126)
```
21. We set aside here statements called 'metatextual' (García-Carpintero, 2014, p. 17) or 'metafictional' (Recanati, 2018), illustrated by (i). (i) Pegasus is a mythical horse.

## **References**

Austin, J. L. (1962). *How to Do Things with Words*. Cambridge, MA: Harvard University Press.

Bach, K. (2010). Getting a Thing into a Thought. In: R. Jeshion, ed., *New Essays on Singular Thought*, Oxford: Oxford University Press, pp. 39–63.

Bergson, H. (1907/1941). *L'évolution créatrice*. Paris: PUF.


## **About the Authors**

### **Editor**

**Malin Roitman** is associate professor at the Department of Romance Studies and Classics, Stockholm University. Her research focuses on political and media discourse and the argumentation strategies that arise there. She is particularly interested in phenomena of pragmatics and enunciation, such as acts of disagreement and refutation, polyphony markers, and including rhetoric, expressions of ethos and axiology. Many studies focus on the function of negation in the debates of French presidential campaigns. Malin Roitman has edited the collective volume The Pragmatics of Negation: negative meanings, uses and discursive functions (2017) and co-edited three more books on the pragmatics of political discourses.

ORCID: https://orcid.org/0000-0001-7045-7557

### **Authors**

**Elena Albu** is a postdoctoral fellow at the University of Tübingen. Her research focuses mainly on understanding the discursive behaviour and the mental representation of various negative utterances in natural language use. She has published works on the pragmatics of negation in Romanian political debates and is currently dealing with negatives in social media discourses. She has published works on the pragmatics of negation in Romanian political debates and is currently interested in the processing of negative utterances.

ORCID: https://orcid.org/0000-0001-5988-0688

**Anne-Laure Besnard** is an associate professor of English linguistics at Rennes 2 University (Univ. Rennes) in France. She is affiliated with the research team *ACE (Anglophonie: communautés et écritures)* and specializes in the study of point of view in media discourse. More specifically, her work focuses on present-day English modal and evidential periphrastic expressions approached from an enunciative perspective (Culioli's Theory of Predicative and Enunciative Operations). ORCID: https://orcid.org/0000-0001-6171-6983

**Francesca Capuano** is a research associate at the Department of Psychology of the University of Tübingen. She has a background in linguistics and computational linguistics, and her current research focuses on the pragmatics of negation, as well as on embodied language processing.

ORCID: https://orcid.org/0000-0003-1898-3255

**José Manuel Durán** is an associate professor at the Universidad de Belgrano. His field of research is related to Systemic Functional Linguistics, corpus linguistics and negative polarity items such as *not*, *never*, or *no* and their collocations. Duran works currently on big data corpora from American presidential discourses to disclose patterns of negative polarity over time and in relation to other types of texts. ORCID: https://orcid.org/0000-0003-1353-8718

**Kjersti Fløttum** is a professor of French linguistics at the University of Bergen. Her research fields are related to text- and genre theory, semantics, pragmatics, linguistic polyphony, and discourse analysis. The major part of her empirical research is oriented towards enunciation, polyphonic, lexico-semantic and narrative analysis of academic, political and climate change discourse. She currently heads the project "Living with climate change: motivation and action for lifestyle change" (2020–2023), funded by the Research Council of Norway. ORCID: https://orcid.org/0000-0003-0834-824X

**Bonnie B. Fonseca-Greber** is associate professor of French in the Department of Classical and Modern Languages at the University of Louisville, where she is also affiliated faculty in the Linguistics Program (Department of Comparative Humanities) and its Discourse and Semiotics Workshop. She has accomplished research on spoken French and works on a corpus of conversational Swiss French (1997–1998) with special interest for how pragmatic factors appear to influence the use/non-use of *ne* as speakers communicate negation in face-to-face conversation.

ORCID: https://orcid.org/0000-0002-1020-9030

**María Marta García Negroni** is a professor of French Linguistics at the University of Buenos Aires and associate professor at the University of San Andrés where she in charge of the University Writing Center. Her research is related to French theories of enunciation, argumentation and polyphony with special interest on connectors and discursive markers, types of negation, attenuation, reinforcement and reformulation and in evidentiality. Her research has been published in several books and articles in Argentine, in Europe and other Latin America countries. ORCID: https://orcid.org/0000-0001-7204-6575

**Øyvind Gjerstad** is an associate professor of French linguistics at the University of Bergen, Norway. Among his main research interests are linguistic polyphony and climate change narratives in journalistic and political discourse, as well as in in survey discourse. He currently heads a project, funded by the Research council of Norway, that is part of the consortium "Overcoming Obstacles and Disincentives to Climate Change Mitigation: A cross-cutting approach by human and social sciences" (2020–2023), a collaboration between Belgian, French and Norwegian universities.

ORCID: https://orcid.org/0000-0002-6618-0783

**Renia Lopez-Ozieblo** is the coordinator of the Spanish and European studies programs in the English Department of the Polytechnic University of Hong Kong. Her main lines of research are gestures in language learners and language teaching from the perspective of cognitive linguistics. She directs a number of projects on these topics. She is co-founder of the Hong Kong chapter of the International Society for Gestures Studies and a member of the AILA ReN research group: Gestures in second language acquisition. She is also a member of the Research Center for Professional Communication in English (RCPCE) at Hong Kong Polytechnic University.

ORCID: https://orcid.org/0000-0003-1752-9392

**Tomohiro Sakai**, Ph.D in Linguistics, University of Paris VIII, 2004, Ph.D in Arts and Sciences, University of Tokyo, 2003, is a Professor for the Faculty of Letters, Arts and Sciences at Waseda University in Japan. His research interests include Philosophy of Language, Semantics, Pragmatics and Mental Space Theory.

ORCID: https://orcid.org/0000-0003-3806-8070

### This volume deals with the pragmatic dimension of negations and is oriented towards empirical studies of negatives' meanings and functions in media and public discourses.

By bringing together scholars from different countries, with studies on different languages this volume aims to shed light and contribute to new knowledge about the forms and functionality of negation as a universal phenomenon. Linguists within Pragmatics generally agree that the use of negatives escapes logic and pure semantic description and is therefore best analysed with tools from cognitive and pragmatic theories.

Based on hypotheses within pragmatics, semantics and discourse analysis, the main assumption is here that forms of expressing negatives emerge and adjust constantly and in accordance with the cultural domain and the social setting of their appearance. This is why this volume focuses on the functions of negative expressions in specifi c domains and types of discourses.

Negatives and Meaning: Social Setting and Pragmatic Effects Roitman (ed.)

Negatives and Meaning:

Social Setting and

Pragmatic Effects

Using Negatives in Political

Discourse, Social Media and

Oral Interaction

Edited by Malin Roitman