Applications of Relevance Theory : From Discourse to Morphemes [1 ed.] 9781443891684, 9781443872805

Citation preview

Applications of Relevance Theory

Applications of Relevance Theory: From Discourse to Morphemes Edited by

Agnieszka Piskorska and Ewa Wałaszewska

Applications of Relevance Theory: From Discourse to Morphemes Series: Advances in Pragmatics and Discourse Analysis Edited by Agnieszka Piskorska and Ewa Wałaszewska This book first published 2017 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2017 by Agnieszka Piskorska, Ewa Wałaszewska and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-7280-6 ISBN (13): 978-1-4438-7280-5

TABLE OF CONTENTS

Introduction ................................................................................................. 1 Applications of Relevance Theory: From Discourse to Morphemes Ewa Waáaszewska and Agnieszka Piskorska Part I: Discourse and Procedural Meaning Chapter One ............................................................................................... 16 Cognitive Environment and Information Structure Jadwiga Linde-Usiekniewicz Chapter Two .............................................................................................. 44 Towards a Relevance-Theoretic Account of the Iñupiaq Hearsay Evidential Guuq Signe Rix Berthelin Chapter Three ............................................................................................ 64 Instead Thornstein Fretheim Part II: Specialized Discourses and Relevance Theory Chapter Four .............................................................................................. 86 Putting Relevance at Centre Stage in Research on Human Activity on the Internet Francisco Yus Chapter Five ............................................................................................ 103 The Guru Effect in Blind People’s Comprehension Jolanta Sak-Wernicka Chapter Six .............................................................................................. 117 Relevance Theory and Cognitive Behavioral Therapy Theory in the Analysis of Psychotherapeutic Discourse Elwira Szehidewicz

vi

Table of Contents

Chapter Seven.......................................................................................... 132 The Translator’s Intentions – the Same as or Different from the Author’s? Cases of Manipulation in Polish Translations of British Press Articles in 1965-1989 Edyta ħraáka Part III: Figures of Speech in Literary Discourse Chapter Eight ........................................................................................... 152 Towards a Relevance Theory Account of Allegory Christoph Unger Chapter Nine............................................................................................ 175 Code and Inference in the Expression of Irony in Orwell’s Animal Farm and its Translation into Spanish Maria Angeles Ruiz Moneva Part IV: Humorous Discourse Chapter Ten ............................................................................................. 198 Degrees of ‘Punniness’? A Relevance-Theoretic Account of Puns and Pun-Like Utterances Agnieszka Solska Chapter Eleven ........................................................................................ 223 ‘To Classify or not to Classify?’ On a Relevance-Theoretic Classification of Jokes: A Critical Survey Magdalena Biegajáo Part V: Morphological Issues and Lexical Pragmatics Chapter Twelve ....................................................................................... 246 Broadening the Scope of Lexical Pragmatics: The Creation of Neologisms in Toposa Martin Schröder Chapter Thirteen ...................................................................................... 277 Contrastive Reduplication and Relevance Theory Ewa Waáaszewska

Applications of Relevance Theory: From Discourse to Morphemes

vii

Chapter Fourteen ..................................................................................... 299 On the Origin and Meaning of Secondary Interjections: A RelevanceTheoretic Proposal Manuel Padilla Cruz Contributors ............................................................................................. 327 Index ........................................................................................................ 329

INTRODUCTION APPLICATIONS OF RELEVANCE THEORY: FROM DISCOURSE TO MORPHEMES EWA WAàASZEWSKA AND AGNIESZKA PISKORSKA, UNIVERSITY OF WARSAW, POLAND

When Paul Grice put forward his theory of implicature in his 1967 William James Lectures, one of the crucial distinctions underlying his model was that between ‘saying’ and ‘implicating’. The ‘saying’ side of the dichotomy seemed to be relatively well understood and it was the implicit side of communication that needed explaining. The explanation offered included a set of rules governing the recovery of implicatures, i.e. the Co-operative Principle (CP) and its attendant maxims, as well as inference as the mechanism taking the hearer from ‘what is said’ to ‘what is implicated’. The expectations laid on the role of inference in interpreting ‘what is said’ were rather modest and included a few specific tasks, such as assigning reference to referring expressions, deciding on a sense of an ambiguous expression, and fixing some variables associated with the deictic parameters of an utterance. The role of the CP and the maxims in interpreting ‘what is said’ was not addressed, which can be taken as an indication that, indeed, the CP and maxims were seen as functional only in interpreting implicatures, or, at best, also in the ‘said’ layer of communication to the small extent to which inference played a role in it (cf. Grice 1989, 45 on the meaning of the verb say in the relevant sense). As the history of science, linguistics included, shows, pioneering ideas have to be not only developed, but also modified, often substantially. Relevance Theory (RT) adopted Grice’s general views that communication involves inference and that communicated meaning can be explained in terms of speaker’s intentions, but vastly redefined the role of inference and the status of principles governing inferential processes. This was necessitated by the change of perspective: although Relevance Theory continues the

2

Introduction

philosophical-linguistic tradition inspired by Grice, it is committed to the scientific rigour of a cognitive science at the same time. The latter means that information processing by the human mind in real time has to be treated as a crucial factor in constructing the theory. Therefore, it has to be taken into account that inference is spontaneously and instantenously performed on linguistic material, that the principle governing inferential processes has to be cognitive (rather than based on philosophy or rules of social conduct), and that the evolutionary advantage of the communication mode actually employed by humans over alternative models has to be explained too. Such requirements are met by the Communicative Principle of Relevance, which supports the relevance-based model of human communication on the strength of efficiency of information processing in real time. The principle states that every act of communication conveys the presumption of its own optimal relevance (Sperber and Wilson [1986] 1995, 158; Sperber and Wilson 1995, 260-261), which means that every utterance is interpreted in such a way as to maximize cognitive benefits and minimize processing effort. Thus, of many interpretations possible, the hearer will automatically and spontaneously choose the one that meets the two conditions: (1) it brings about large cognitive gains, which can be generally characterized as improvements in the representation of the world and (2) it requires such amount of effort that can be justified by these gains. When applied to the study of discourse, the cognitive perspective inherent in RT means that the object of study is not discourse per se, but the understanding of discourse by human beings, as Diane Blakemore (2001, 100-101) put it. Rather than define abstract relations, such as discourse coherence, RT thus focuses on how we see various elements of a text as coherent when we process them in search for optimal relevance. It goes without saying that constructing discourse relations is facilitated by what has been called in the literature ‘discourse markers’, such as but or so, which constrain the hearer’s search for relevance. Making no contribution to the truth-conditional or representational meaning of utterances, discourse markers are treated as encoding ‘procedural meaning’, i.e. instructions how to manipulate conceptual representations in order to optimize relevance. It needs to be mentioned that since the notion of procedural meaning was first introduced as an account of discourse markers by Blakemore (1987), it has undergone a notable development. For example, Wilson and Sperber (1993) put forward the claim that some procedural items, such as pronouns, contribute to truth-conditions. With the emergence of the idea that evaluating propositions communicated as

Applications of Relevance Theory: From Discourse to Morphemes

3

believable or not is served by a distinct mental module, the so-called ‘argumentation module’ (e.g. Mercier and Sperber 2009), the role of discourse markers has been re-analyzed with respect to whether they fall within the operation of the comprehension module or the argumentation module. The issue of procedural meaning is raised in Part I of the present volume, called Discourse and Procedural Meaning, which includes three contributions. Two of them focus on specific linguistic items and one puts forward a more general proposal about reformulating the traditional theme/rheme distinction in terms of procedural meaning. The opening chapter “Cognitive Environment and Information Structure” by Jadwiga Linde-Usiekniewicz explores the applicability of one of the fundamental concepts of Relevance Theory, that of ‘cognitive environment’, to developing a re-definition of the basic categories used in describing the information structure of sentences and/or utterances, namely the theme/rheme distinction. Linde-Usiekniewicz draws upon her previous work devising an Encoding Grammar (Linde-Usiekniewicz 2012), in which the claim is put forward that different components of the language system, such as syntax, semantics, and discourse organization could be at odds with one another, as well as with the speaker’s communicative needs. The formulation eventually chosen by the speaker could be thus seen as an instance of conflict resolution or a compromise. Citing arguments in favour of maintaining information structure as a crucial part of linguistic analysis, and in favour of the level of ‘utterancetype’, the paper argues that the theme-rheme articulation can be constructively viewed in terms of procedural meaning, on the level of ‘utterance-types’ Signe Rix Berthelin’s contribution “Towards a Relevance-Theoretic Account of the Iñupiaq Hearsay Evidential Guuq” dissects a single expression in an endangered language of Alaska on the basis of data collected during interviews with native speakers. The analyzed enclitic seems to mark utterances in which it occurs as providing hearsay information. By arguing that guuq may be successfully described within a relevance-theoretic approach to modality, Berthelin throws some light on the notions of evidentiality and modality. First, the author reveals certain inconsistencies in previous descriptions and classifications of guup and similar expressions in Alaskan languages in terms of evidentiality. Guuq is nevertheless classified as an optional evidential expression indicating that the speaker is not the source of evidence presented in the utterance. This indication does not necessarily make the evidence less reliable. Notably, in the context of storytelling guuq may increase the speaker’s responsibility

4

Introduction

for the truth of the presented information. To account for these findings, the author employs a relevance-theoretic framework developed previously by Papafragou (2000) to analyze the meaning of modal verbs. Interestingly, she also provides sound arguments for rejecting an analysis in terms of attributive or echoic use. In his paper, succinctly entitled “Instead”, Thornstein Fretheim provides a detailed analysis of the meaning and uses of the English expression. Noticing that instead in its two syntactic patterns encodes a contrary relation between a factual and a non-factual entity, the author focuses on the construction with a zero complement, or the “bare instead”. This is because the “zero anaphor” necessarily involves a lot of inference, which makes this pattern more interesting to pragmatics. Taking into consideration semantic, etymological and cross-linguistic data, on the basis of actual examples from translation corpora, Fretheim describes three distinct ways in which the procedural meaning of the analysed expression may contribute to relevance. This contribution may involve activation or provision of a contextual assumption, a bridging implicature or conceptual content. The analysis sheds an interesting light on the nature of procedural meaning and its interaction with conceptual, truth-conditional and implicit meaning. Although understanding all types of discourse involves the same process of searching for relevance, not all of them are equally amenable to interpretation, since access to contextual knowledge varies among individuals. The four chapters included in Part II, Specialized Discourses and Relevance Theory, address problems of communication taking place in circumstances removed from the prototypical situations of face-to-face interactions in which interlocutors share the same perceptible cognitive environment. One such case is discussed by Francisco Yus in “Putting Relevance at Centre Stage in Research on Human Activity on the Internet.” Yus shows that Relevance Theory can be fruitfully applied to ‘cyberpragmatics’ and account for all kinds of human activities undertaken on the Internet, including interactions between users and computer systems. Over the last few years, the development of Internet use has increased so radically that it has become not only a powerful tool in the hands of human beings but also an equally powerful factor affecting what human beings are. Thus, cyberpragmatics has become a valid and fascinating area of research on communication (see Yus 2011). As Yus shows, RT can be applied to this field partly on the strength of the Communicative Principle of Relevance, which governs all the communicative exchanges between users, and partly on the strength of the Cognitive Principle of Relevance, which governs processing of all stimuli

Applications of Relevance Theory: From Discourse to Morphemes

5

accessible to the human mind. In this way, four areas of relevance-driven processing are covered by Yus’s analysis: (a) relevance sought by the system for the user, discussing not only benefits but also limitations of applying relevance-based predictions of human needs by systems like search engines; (b) relevance sought by the user in the system, offering an explanation why and how the Internet exerts a strong and possibly harmful effect on human minds (cf. Carr 2008, 2010); (c) relevance sought by the user in another user’s coded input (e.g. an utterance), analysing how conditions specific to the Internet communication affect the effort-effect balance; and (d) relevance sought by the user in a group of users, observing that the possibility of staying in touch with several other users simultaneously offsets the considerable mental effort of split attention by benefits of social nature, such as shaping one’s identity, or sense of belonging to a group. Generally, as Yus observes, non-cognitive rewards are obtained in a number of Internet activities. Some groups of communicators may lack access to contextual resources available to others due to perceptual limitations. Such a case is discussed in Jolanta Sak-Wernicka’s paper “The Guru Effect in Blind People’s Comprehension,” putting forward the idea that sighted individuals play the role of guru (in the sense of Sperber 2010) in front of the blind, who are aware of the cognitive advantage afforded by the sense of sight but who are often unaware of the exact character and magnitude of this advantage. Sak-Wernicka takes as a departure point Sperber’s (2010) observation that when a speaker produces an obscure utterance, this does not necessarily lead to the hearer’s negative assessment of the speaker’s rhetorical abilities, but on the contrary, it may give rise to the ‘guru effect’: the hearer will ascribe to the speaker a level of knowledge and sophistication exceeding his own interpretive abilities. Sak-Wernicka goes on to note that blind people will inevitably ascribe the role of the guru to sighted people as those who have unlimited access to visual information. Consequently, the visually impaired will tend to blame their own limitations for not understanding sighted communicators, rather than seek the causes of such a situation in external factors. They also tend to trust sighted informants and take importance of their explanations for granted. Besides, they are naturally eager to invest a great amount of cognitive effort in the process of interpretation. These observations are corroborated by a number of comprehension task experiments, testing both spontaneous understanding and more conscious reasoning strategies applied by the blind. A different situation in which the roles of participants in communication are not symmetrical is dealt with by Elwira Szehidewicz in the chapter

6

Introduction

“Relevance Theory and Cognitive Behavioral Therapy Theory in the Analysis of Psychotherapeutic Discourse,” which applies RT to the analysis of a therapist-client interaction. For therapy to be effective, it is crucial that the two parties reach understanding and build a mutual cognitive environment. To reach this aim, the therapist has to engage actively in helping the patient formulate a verbal representation of her thoughts and emotions. Sometimes the only possible formulation involves a figure of speech, and as predicted by RT, the more original it is, the richer and subtler the range of communicated cognitive effects. By analyzing an example of a metaphor used by the therapist and accepted by the patient as a faithful representation of the onset of her panic attack, the author attempts to indicate strong and weak points of the cognitive routes postulated for the understanding of metaphor within RT, i.e. the route leading via the ad hoc concept formation (e.g. Carston 2002; Wilson and Carston 2008) and the route via the working out of weak implicatures (Wilson and Sperber 2002). Szehidewicz concludes that the most fruitful line of metaphor interpretation admits activating images and using them as input to deriving the final effect. As a case of interlingual and intercultural communication, translation naturally involves a shift from the context of the original text to the one in which the translation will be interpreted. Despite difficulties resulting from this, translators typically do their best to convey the meaning of the original to target text readers. But as Edyta ħraáka shows in the chapter “The Translator’s Intentions – the Same as or Different from the Author’s? Cases of Manipulation in Polish Translations of British Press Articles in 1965-1989,” this may not be so when political propaganda is involved. The problem touched upon in the paper, namely the inevitable interference of the translator’s communicative intention with that of the author’s, is vital for translation at large. Most translators who decide to tamper with the original author’s intentions by introducing additions, omissions, or some alterations to the text decide to do so to facilitate understanding or to conform to the target language and culture norms. Some translators, however, may be guided by ideological considerations, either their own, or imposed upon them by authorities, as was the case with translators creating Polish language versions of British press articles during the communist times in Poland, where the point was to conform to the regime’s expectations to produce texts showing the communist system in a more favourable light than that in which it was actually portrayed by the original authors. The paper discusses various perspectives on manipulation in translation, as well as strategies and techniques employed towards making the audience accept the translated

Applications of Relevance Theory: From Discourse to Morphemes

7

text without realizing that the author’s communicative intention has been overridden by the translator’s, or more precisely, the commissioner’s ideological purposes. Analyzing examples from articles published in the years 1965-1989, ħraáka observes that the most frequent manipulative techniques were extensive omissions and additions, ranging from single words to whole paragraphs. Part III called Figures of Speech in Literary Discourse contains two chapters addressing the issues of figurative language to be found in literary works. Generally, Relevance Theory treats metaphor and related figures of speech in a much deflationary way (Sperber and Wilson 2008), by positing that they do not involve a departure from any norm of literalness and that there is a continuum of cases ranging from literalness through loose uses of language to less and more creative metaphors. The chapter “Towards a Relevance Theory Account of Allegory” by Christoph Unger is an attempt to extend the RT analysis of non-literal uses of language to allegory. The phenomenon of allegory raises a number of theoretical issues, especially with regard to how it is related to metaphor. One of the problems stems from the fact that most accounts of allegory explain it in terms of its relationship with metaphor. For example, Crisp (2008) claims that allegory is based on the same processes as metaphor but differs from extended metaphor both quantitatively and qualitatively. The major difference between the two is that whereas metaphor involves both source and target domains, the language used in allegory relates exclusively to the source domain. According to Thagard (2011), allegory is similar to metaphor as it is also based on analogical mappings which are subject to three constraints concerning similarity, structure and purpose. However, to be successful, an allegory must not only satisfy the similarity constraint but also arouse emotions so as to bring about its purpose. Gibbs (2011) regards allegory as involving metaphorical interpretation of non-metaphorical language, with comprehension of both allegory and metaphor being based on the process of embodied cognitive simulation. In order to present a plausible relevance-theoretic account of allegory, Unger chooses to carry out a detailed analysis of the biblical “song of the vineyard” (Isaiah 5) and other data. It turns out that allegory may involve an extended metaphor but, more crucially, it relies on ambiguity between literal and figurative meaning, which is transparent for the audience. The difference between metaphor and allegory is thus substantial and theory-independent. The cognitive mechanisms behind the production and interpretation of irony have been widely discussed in RT since its origins (e.g. Sperber and Wilson 1981, [1986] 1995, 1998; Wilson 2006; Wilson and Sperber 2012). It is believed that the chief purpose of using irony is to tacitly

8

Introduction

communicate a dissociative attitude toward an assumption attributed to some source, be it a specific individual, a group of individuals, or people in general. The assumption alluded to typically refers to a state of affairs which has failed to materialise, so the need to express a negative attitude often results from a feeling of disappointment when events run contrary to expectations. That appears to be the main motivation behind the deployment of irony in Orwell’s Animal Farm, believed to mirror the Russian Revolution of 1917 and the disillusion that followed it. In the paper “Code and Inference in the Expression of Irony in Orwell’s Animal Farm and its Translation into Spanish,” Maria Angeles Ruiz Moneva observes that irony communicated by the narrator of the story is easy to grasp for the reader because he is aware of the discrepancy between the real outcome of the revolution and its original ideology and objectives. The idea put forward by Yus (2000, 2012) is that the realisation of the incompatibility between the literal content of an utterance and the contextual resources bearing on its interpretation is a trigger of an ironic interpretation. It thus follows that the easier the access to such contextual resources, the more accessible the ironic interpretation. In Animal Farm, the audience is put on the author-intended processing path by a frequent use of linguistic devices, such as understatement, syntactic parallelism, or the choice of words, which are therefore used as communicative clues (Gutt 2000), signalling the incompatibility of the statement expressed with the way things are in the reality of the plot. Since the use of language, or code, is conducive to the successful recovery of ironical attitude, Ruiz Moneva favourably assesses the strategy of direct translation (Gutt 2000) employed in the Spanish-language version of Animal Farm, preserving the original work’s linguistic form and providing the reader of the translated text with an opportunity to follow an inferential path analogous to that of the original audience. Humorous discourse appears to have its own characteristics: it requires the hearer to invest considerable processing effort, for which the pay-off comes in the form of amusement rather than purely cognitive gains. Part IV called Humorous Discourse consists of two chapters. In “Degrees of ‘Punniness’? A Relevance-Theoretic Account of Puns and Pun-Like Utterances,” Agnieszka Solska examines two types of utterances which exploit lexical ambiguity for the sake of a rhetorical effect in a pun-like manner and yet differ from puns proper in that they do not rely on representing the two meanings of a pivotal word simultaneously. One type involves metalingual cases of word play, such as Not all banks are river banks and the other type includes “pun-like comparisons”, such as Jane is as hard as nails. Rejecting the claim that puns and pun-like utterances

Applications of Relevance Theory: From Discourse to Morphemes

9

represent various manifestations or degrees of the same phenomenon, Solska postulates that distinct conceptual configurations are at play in metalingual word-play utterances and in pun-like comparisons: in the former, the two senses of a word are merged into a hybrid ad hoc concept, whereas in the latter only one of the senses genuinely contributes to the relevance of an utterance. Magdalena Biegajáo’s “‘To Classify or not to Classify?’ On a Relevance-Theoretic Classification of Jokes: A Critical Survey” starts with a critical overview of the three most widely known taxonomies of jokes developed within the relevance-theoretic framework, namely those by Jodáowiec (1991a, 1991b, 2008), Curcó (1995, 1996a, 1996b, 1997) and Yus (2003, 2004, 2008, 2012, 2013). The author tries to indicate their various methodological shortcomings by claiming that classes of jokes distinguished in each model are not clearly delineated. Since the criteria for teasing apart joke types are inferential mechanisms responsible for generating humour, Biegajáo argues that each class should include jokes exploiting only one such mechanism, and if more such mechanisms can be shown to be at play in jokes included in one class, this can be considered a flaw in the whole system of classification. The second part of the paper presents a quantitative analysis of a group of jokes employing the three typologies, with some modifications. As the author concludes, although it can be plausibly claimed that RT offers an exhaustive account of various inferential processes leading to humorous effects, it does not necessarily entail that they can be straightforwardly used as a basis for joke typology. With the rise of lexical pragmatics, relevance theorists have not only started but also intensified work on lexical modulation, understood as “different types of pragmatic effect on the meanings that lexical items are used to convey” (Allott 2010, 109). The main two types of such pragmatic effect are the broadening and narrowing of the lexically-encoded meaning of a word, or in other words, non-lexicalized ad hoc concepts which are broader or narrower than the lexicalized concepts from which they have been derived. What is interesting, it seems that the pragmatic processes yielding concept modulation may be related to the morphological process of neology, by means of which new terms are coined or existing words acquire new meanings, functions and uses (see Wilson and Carston 2007, 237). This idea is explored in Part V, called Morphological Issues and Lexical Pragmatics. The three chapters included there focus on the production and comprehension of neologisms, and explain them in terms of ad hoc concept formation. Martin Schröder’s contribution “Broadening the Scope of Lexical Pragmatics: The Creation of Neologisms in Toposa” offers a discussion

10

Introduction

on the formation of neologisms by an ethnic group that has undergone major changes in their language and culture over the last twenty-five years, the Toposa of South Sudan. An isolated tribe still in the ninetyeighties, the Toposa subsequently experienced a civil war which brought about the displacement of many ethnic groups and massive migration into Toposa areas, contact with warfare technology, and a flow of Western aid and development. All these factors, together with a Bible translation, language development and literacy projects, spurred a demand for new words to encode newly acquired concepts and set in motion a number of lexical processes leading to the creation of neologisms. Among the word-formation patterns operating in Toposa, the paper presents data exemplifying a number of derivational processes, including affixation, change of gender and change of noun class. Beside those, there are instances of compounding, meaning extension, the use of descriptive phrases, coining new collocations and loans. The analysis of the Toposa data provides the author with the opportunity to ask important questions about the nature of the meaning of root morphemes and of affixes, as well as their contribution to the meaning of the newly-created term. In relation to that, he proposes that the major component of the root morpheme meaning is conceptual and the minor one is procedural, whereas in the case of affixes, the reverse proportion holds. As to the process of combining the root meaning with an affix meaning, Schröder postulates that it largely resembles the process of utterance production and interpretation in that it is relevance-driven and involves the creation of ad hoc concepts. The analogy between neologisms and utterances is motivated by the fact that both are novel. Whether terms that originated as neologisms enter the lexicon, on the other hand, depends on a number of social and political factors. As one of his concluding remarks, Schröder puts forward the claim that the analysis of word formation processes that are taking place in a language of a society confronted with external changes could be beneficial for translators who often face the need to coin new terms. Ewa Waáaszewska’s paper “Contrastive Reduplication and Relevance Theory” provides a relevance-theoretic account of a word formation process, referred to as ‘contrastive reduplication’, which involves combining two identical words of which the first is contrastively stressed. The phenomenon is relatively widespread in spoken English, American English in particular, but it can also be observed in other languages. Perhaps, because of its rising popularity among language users, contrastive reduplication has attracted some attention in linguistics. The paper reviews the phenomenon by presenting existing definitions of contrastive reduplication and showing how it differs from other types of reduplication.

Applications of Relevance Theory: From Discourse to Morphemes

11

The most conspicuous difference seems to be that while reduplication has been studied mostly from a morphological perspective, contrastive reduplication turns out to be essentially pragmatic in nature. Previous analyses of this intriguing phenomenon have already indicated its contextdependence but they typically restrict the range of meanings conveyed by it to prototypical interpretations or a list of specific readings. Accounting for contrastive reduplication along the lines of Relevance Theory has the advantage of providing a unified explanation of how it works and what speakers may mean by it. The analysis put forward in the paper shows that contrastive reduplication is functionally similar to ‘hedges’ in that it is another linguistic device used to signal the need for concept adjustment, more specifically, for the narrowing of a lexicalized concept. Redescribing contrastive reduplication as involving procedural meaning helps explain in detail how it works and how it is different from similar and potentially confusable phenomena such as repetition (epizeuxis). In his chapter “On the Origin and Meaning of Secondary Interjections: A Relevance-Theoretic Proposal,” Manuel Padilla Cruz focuses on so-called ‘secondary’ interjections and suggests an explanation of their origin and expressive potential by resorting to the relevancetheoretic approach to lexical pragmatics. Secondary interjections (e.g. Hell! or Good!) have attracted much less attention than primary interjections (e.g. ugh! or phew!): researchers have mainly concentrated on their sociolinguistic distribution and language-dependent idiosyncratic properties as well as on their meaning in social contexts. The origin of secondary interjections is typically explained in terms of ‘grammaticalization’, a process which causes lexical items to develop certain grammatical functions, and ‘subjectification’, a process whereby words acquire new functions allowing for the recovery of the speaker’s internal state or subjective attitude. However, as insightfully observed by Padilla Cruz, this type of explanation does not make it clear what sets these two processes in operation, or in other words, what underlies such shifts of grammatical category. He goes on to suggest that the process of grammaticalization leading to the emergence of secondary interjections is tightly linked with the operation of lexical pragmatic processes of concept adjustment – more specifically, with the process of concept broadening. Due to this process, the concepts encoded by content words (nouns, verbs, adverbs, adjectives) acquire more general (abstract) meanings that will ultimately shift towards the emotions or attitudes associated with the initial lexically encoded concepts. Interestingly, the origin of secondary interjections is explained along the same lines as the origin of overextensions produced by children in early lexical development, discussed from a Relevance Theory

12

Introduction

perspective by Waáaszewska (2011). According to Padilla Cruz, what makes secondary interjections and children’s overextensions strikingly similar is that both are frequent in situations in which the right word is unavailable to speakers who, to overcome communicative difficulties, will use words encoding different though related concepts as pointers to the intended meanings. Moreover, the two types of broadenings found in children’s overextensions (over-inclusions and analogical extensions) appear to correspond to two types of broadenings in the case of secondary interjections: the broadening of adjectives and adverbs will yield ‘overinclusions’ and the broadening of nouns and verbs will result in ‘analogical extensions’.

References Allott, Nicholas. 2010. Key terms in pragmatics. London: Continuum. Blakemore, Diane. 1987. Semantic constraints on relevance. Oxford: Blackwell. —. 2001. Discourse and Relevance Theory. In Handbook of discourse analysis, edited by D. Schiffrin, D. Tannen and H. E. Hamilton. Oxford: Blackwell, 100-118. Carr, Nicholas. 2008. Is Google making us stupid? The Atlantic 07/082008. —. 2010. The shallows: What the Internet is doing to our brains. New York: Norton & Company. Carston, Robyn. 2002. Metaphor, ad hoc concepts and word meaning – more questions than answers. UCL Working Papers in Linguistics 14: 83-105. Crisp, Peter. 2008. Between extended metaphor and allegory: Is blending enough? Language and Literature 17, no. 4: 291-308. Curcó, Carmen. 1995. Some observations on the pragmatics of humorous interpretations. A relevance-theoretic approach. UCL Working Papers in Linguistics 7: 27- 47. —. 1996a. The implicit expression of attitudes, mutual manifestness, and verbal humour. UCL Working Papers in Linguistics 8: 89-99. —. 1996b. Relevance Theory and humorous interpretations. In Automatic interpretation and generation of verbal humor, edited by J. Hulstijn and A. Nijholt. Enschede: University of Twente. —. 1997. The pragmatics of humorous interpretations. A relevancetheoretic approach. Ph.D. diss., UCL Department of Phonetics and Linguistics, London.

Applications of Relevance Theory: From Discourse to Morphemes

13

Gibbs, Raymond W. 2011. The allegorical impulse. Metaphor and Symbol 26, no. 2: 121-130. Grice, Herbert Paul. 1989. Further notes on logic and conversation. In Studies in the way of words. Cambridge, Massachusetts: Harvard University Press, 41-57. Gutt, Ernst-August. 2000. Translation and relevance: Cognition and context. 2nd ed. Manchester: St. Jerome. Jodáowiec, Maria. 1991a. The role of relevance in the interpretation of verbal jokes: A pragmatic analysis. Ph.D. diss., Jagiellonian University. —. 1991b. What makes jokes tick. UCL Working Papers in Linguistics 3: 241-253. —. 2008. What‘s in the punchline? In Relevant worlds: Current perspectives on language, translation and Relevance Theory, edited by E. Waáaszewska, M. Kisielewska-Krysiuk, A. Korzeniowska and M. Grzegorzewska. Newcastle: Cambridge Scholars Publishing, 67-86. Linde-Usiekniewicz, Jadwiga. 2012. From conflict through compromise to collaboration: Semantics, syntax and information structure in natural languages. Warszawa: Nakáadem Wydziaáu Polonistyki UW. Mercier, Hugo and Dan Sperber. 2009. Intuitive and reflective inferences. In In two minds: Dual processes and beyond, edited by J. Evans and K. Frankish. Oxford: Oxford University Press, 149-170. Papafragou, Anna. 2000. Modality: Issues in the semantics-pragmatics interface. Amsterdam/New York: Elsevier Science. Sperber, Dan. 2010. The guru effect. Review of Philosophy and Psychology 1, no. 4: 583-592. Sperber, Dan and Deirdre Wilson. 1981. Irony and the use-mention distinction. In Radical pragmatics, edited by P. Cole. New York: Academic Press, 295-318. —. [1986] 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. —. 1995. Postface. In Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell, 255-279. —. 2008. A deflationary account of metaphors. In: The Cambridge handbook of metaphor and thought, edited by R. W Gibbs. Cambridge: Cambridge University Press, 84-105. Thagard, Paul. 2011. The brain is wider than the sky: Analogy, emotion, and allegory. Metaphor and symbol 26, no. 2: 131-142. Waáaszewska, Ewa. 2011. Broadening and narrowing in lexical development: How Relevance Theory can account for children’s overextensions and underextensions. Journal of Pragmatics 43: 314326.

14

Introduction

Wilson, Deirdre. 2006. The pragmatics of verbal irony: Echo or pretence? Lingua 116: 1722-1743. Wilson, Deirdre and Robyn Carston. 2007. A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In Pragmatics, edited by N. Burton-Roberts. Houndmills, Basingstoke: Palgrave Macmillan, 230-259. —. 2008. Metaphor and the ‘emergent property’ problem: A relevancetheoretic treatment. The Baltic International Yearbook of Cognition, Logic and Communication, vol. 3. New Prairie Press. (November 2012). Wilson, Deirdre and Dan Sperber. 1993. Linguistic form and relevance. Lingua 90: 1-25. —. 2002. Relevance Theory. UCL Working Papers in Linguistics 14:249290. —. 2012. Meaning and relevance. Cambridge: Cambridge University Press. Yus, Francisco. 2000. On reaching the intended ironic interpretation. International Journal of Communication 10, nos. 1 & 2: 27-78. —. 2003. Humour and the search for relevance. Journal of Pragmatics 35, no. 9: 1295-1331. —. 2004. Pragmatics of humorous strategies in El club de la comedia. In Current trends in the pragmatics of Spanish, edited by R. MárquezReiter and M. E. Placencia. Amsterdam: John Benjamins, 320-344. —. 2008. A relevance-theoretic classification of jokes. Lodz Papers in Pragmatics 4, no. 1: 131-157. —. 2011. Cyberpragmatics: Internet-mediated communication in context. Amsterdam: John Benjamins. —. 2012. Relevance, humour and translation. In Relevance Theory: More than understanding, edited by E. Waáaszewska and A. Piskorska. Newcastle upon Tyne: Cambridge Scholars Publishing, 117-145. —. 2013. An inference-centered analysis of jokes: The Intersecting Circles Model of humorous communication. In Irony and humor: From pragmatics to discourse, edited by L. Ruiz Gurillo and B. Alvarado. Amsterdam and Philadelphia: John Benjamins, 59-82.

PART I: DISCOURSE AND PROCEDURAL MEANING

CHAPTER ONE COGNITIVE ENVIRONMENT AND INFORMATION STRUCTURE JADWIGA LINDE-USIEKNIEWICZ, UNIVERSITY OF WARSAW, POLAND

1. Introduction This paper explores the applicability of one of the fundamental concepts of Relevance Theory – that of cognitive environment, particularly the notion that “when you communicate, your intention is to alter the cognitive environment of your addressees” (Sperber and Wilson 1986, 46) – to a field that may be seen as far afield from the central concerns of Relevance Theory: developing a re-definition of the basic categories used in describing the information structure of sentences and/or utterances.1 The re-definition in question was first conceived as a part of a larger project of devising some kind of framework within which “[s]ome surface phenomena, mainly considered peripheral ... can be explained within the language system as the result of conflicts, compromises and sometimes collaboration between syntax, semantics, and information structure” (Linde-Usiekniewicz 2012a, 28). There is no reasonable doubt that syntax and semantics (whichever way they are understood) should be seen as somehow working together to form sentences, or rather jointly contributing to both the surface form and to the meaning of a sentence or utterance. Similarly, information structure is often seen as a kind of interface between pure syntax and discourse.2 Yet little attention has been paid to the idea that different modules of a language could actually be at odds with one another and the eventual surface form of an utterance could

1 2

The issue of sentences vs. utterances will be discussed below, in the section 2.4. For example such suggestion is explicit in the very title of (Erteschik-Shir 2007).

Cognitive Environment and Information Structure

17

be seen as an instance of conflict resolution or a compromise.3 However, such a vision of language, or rather of language use, has provided some interesting insights into several syntactic patterns (Linde-Usiekniewicz 2012a, 2012b). These earlier attempts at exploring the notion of cognitive environment to account for information structure simply lifted the concept out of its original framework; subjected it to some modifications while disregarding whether such modifications had any validity outside Relevance Theory, and introduced it into an account that did not seek any resemblance to the theory of origin (Linde-Usiekniewicz 2008, 2012a). The present paper seeks to reformulate some of this proposal in terms that would be more compatible with Relevance Theory terminology and also to substantially refine it in the light of RT tenets. The resulting proposal can be seen as an attempt to bridge the divide between functional and formal approaches to information structure.4 On the one hand functional approaches rely too heavily on the actual communication situation, including previous discourse and even subsequent discourse, actual communication participants and other extralinguistic issues (Linde-Usiekniewicz 2012a, 89-100, 110-118), even leading to situations in which foci get confused with topics, as happens in Fortescue (2004, 159). On the other hand, purely formal approaches tend to disregard functional counterparts of formal markings – cf. ErteschikShir (2007, 12), where she rightly points out that fronting cannot be used as universal topic markers, since some languages allow only ‘old topics’ to be fronted while others both ‘old’ and ‘new’ topics. The presented proposal attempts to analyse information structure as a phenomenon closely associated with the Relevance Theory concept of procedural meaning (cf. Blakemore 1987 and subsequent refinements of this notion: Wilson and Sperber 1993; Bezuidenhout 2004; Wilson 2011). The paper is structured as follows: section 2 presents a brief and informal outline of the speaker-oriented framework (called Encoding Grammar in Linde-Usiekniewicz 2012a) within which information structure is being defined. Some of the issues addressed are the speakeroriented character of the proposed framework, and – as a result of its speaker-orientedness – the fact that what it seeks to describe is neither highly abstract sentences nor actual utterances, but patterns construed at a certain intermediate level of abstraction, i.e. utterance-types. Section 3 contains some arguments in favor of including information structure in 3

This kind of thinking about language in general resembles, albeit very broadly, the basic tenets of Optimality Theory; however, the resemblance ends here. 4 The distinction between formal and functional approaches is taken from Erteschik-Shir 2007, 72ff.

18

Chapter One

linguistic description, in spite of Sperber and Wilson’s (1986, 216-217) opinion that the “theme-rheme distinction [and other distinctions discussed by them, like topic, comment, background, presupposition, etc.] has no place in technical descriptive vocabulary of either linguistics or pragmatics.” Section 4 presents the proposed re-definition of information structure, while section 5 deals with two apparent fallacies of this proposal, ones indeed brought to light by the very attempt to reformulate the original proposal in RT terms. Section 6 briefly summarizes the departures from Relevance Theory tenets necessary to refine the presented framework and points out some of its advantages.

2. A Brief Outline of the Framework 2.1 The Starting Point The Encoding Grammar framework presented in Linde-Usiekniewicz (2012a) – devised to discuss the division of labor between semantics, syntax, and information structure as a putative conflict and eventual compromise between them – is a multilayered one, thus resembling in that aspect the Meaning Ù Text Model developed by Igor Mel’þuk (cf. Mel’þuk 2012, 85-161 for a recent survey) upon which it heavily draws. However, in contrast to MTM, which models mapping both meanings into corresponding (surface) texts and (surface) texts into their corresponding meanings, the framework presented here is one-dimensional and concerns itself only with how the meaning is delivered by the speaker to her audience.5 Assuming that there are conflicts between semantics, syntax and information structure in a language (or in languages in general), it is not the language itself that is resolving these conflicts, but the speaker, who works out a judicious compromise between what she wants to explicitly deliver to her audience and what the language she speaks allows her 5

Some terminological clarifications are in order here. First, the term ‘speaker’ is used in this paper in the RT sense, i.e. referring to the communicator, and should not be confused with the notion of ‘speaker’ as a language user. Similarly, the set of potential addressees and/or hearers are referred to as ‘audience’. Following RT’s long established convention, the speaker is invariably referred to as she, while for audience the pronoun he is applied. Secondly, the notion of ‘meaning’ used in a vague way will be elucidated in the following subsection, though it roughly covers what within RT is split into conceptual and procedural meaning. Thirdly, it is claimed within the framework that both kinds of meaning are in fact encoded in order to be delivered to the audience.

Cognitive Environment and Information Structure

19

actually to encode. Secondly, it is the speaker who actually makes the choices about what to encode and what to leave for her audience to infer, moreover, it is she who has all the available linguistic means to hand and makes appropriate choices about how to encode her meaning. The Encoding Grammar framework is therefore explicitly processoriented, though it falls short of being an actual model of utterance production. Here the framework differs radically not only from MTM, but also from Functional Discourse Grammar, which “is a pattern model that is inspired by process without seeking to model the latter” (Hengeveld and Mackenzie 2008, 24). By the same token, and again in contrast to both MTM and FDG, the potential process of utterance understanding cannot be held as a simple reversal of encoding procedures. The speaker modeled within Encoding Grammar shares some of the characteristics of the speaker represented within Relevance Theory, inasmuch that she is a rational being, concerned about effort involved in verbal communication, but, in contrast to RT speaker’s considerateness for her audience, in the model she is construed as if she were an out and out egotist, whose ultimate and actually only goal were to deliver her meaning to the best of her and the language’s abilities. However, the notion of such an ‘idealized’ speaker is restricted only to encoding phenomena as described within the framework. The considerate behavior of the RT speaker is seen as corresponding to choices about what to encode and deliver explicitly to the audience and what to (possibly) leave to inference. These choices in real communication may in fact be geared to the speaker’s assumptions about her audience’s cognitive environment, but the model focuses on how they may depend directly on the encoding means a particular language offers.6 At the same time, she is construed as relying on being able to deliver to her audience a linguistic entity (comprising a complete string of sounds/gestures or characters together with associated sense7) at a lump. Thus it is partly immaterial to her where in the sequence of linguistic units some part of meaning is delivered and what initial hypotheses her audience draws from partial apprehending of what she says. This element of the framework is postulated in conscious disregard of the claim, made among others by Hengeveld and Mackenzie (2008, 24), that language production is incremental. Another important restriction is that the framework deals with 6

As one of the anonymous reviewers has pointed out, there is considerable psycholinguistic evidence about even very young children adapting the level of informativness of their utterances to their audience’s cognitive environments. 7 Thus the speaker can in fact deliver to her audience a spoken, written, or signed version of her message.

Chapter One

20

units that would roughly correspond to sentences (or even truncated sentences) as to their size, and not to larger chunks of discourse. All these restrictions on the Encoding Grammar model are self-imposed with conscious disregard of the various complexities of actual communication. The way Encoding Grammar framework is devised could be seen as an attempt at reconciliation between the ‘principle of effability’ of Katz (1981, 226) and its rejection by Relevance Theory (Sperber and Wilson 1996, 191-193). Here the “possibility [or rather implied impossibility – JLU] of two people having exactly the same thought” (Sperber and Wilson 1996, 92) is immaterial, since the framework is meant only to show how the same speaker may sometimes be able to ‘say’ (understood here in its informal sense) or explicitly deliver the same ‘message’ or ‘information’ (again informally understood) in different languages, and sometimes she may not. As many bilingual individuals know and have commented, there are some things that are more easily sayable in one language than in another. To account for this, Encoding Grammar has the idea of conflicts built in into the very notion of semantics. On the one hand a speaker has some information to deliver, and on the other the language she will speak may not offer her an easy way of saying it or even no way at all. To illustrate this point let us imagine a female speaker who wants to inform her audience that she has called her female lawyer.8 Should the speaker be talking in Polish, she would say: (1)

Zadzwoniáam do mojej prawniczki. Lit. ‘I called-FEMININE to my-FEMININE lawyer-FEMININE’

Here the gender of the original speaker is encoded in the verb-form, while the gender of the lawyer is encoded as part of meaning of the noun prawniczka. (This distinction corresponds to the distinction between procedural and conceptual meaning, which is crucial to RT, cf. Blakemore 1987; Wilson and Sperber 1993; Bezuidenhout 2004; Wilson 2011). However, should the same speaker be talking in English she would say (2)

I called my lawyer.

But in (2) she cannot easily include the gender of the lawyer in the encoded conceptual meaning, because English lacks a single lexeme to that effect. One option would be to encode the lawyer's gender separately 8

The example is based on an adapted quote from Amanda Cross, Sweet Death, Kind Death, 1984, New York; Ballantine, p. 88.

Cognitive Environment and Information Structure

21

and explicitly, as female lawyer, but this may likely invite unwanted effects in English (possibly the speaker being perceived as pedantic or intentionally and militantly anti-sexist). The other is to leave (2) as it stands, thus risking that her audience may entertain some wrong (possibly sexist) assumptions as to the actual gender of the lawyer. The difference between (1) and (2) illustrates how a particular language restricts possibilities of delivering certain information. It also shows the difference between the ‘thought’ an individual may have (and even want to convey to her audience) and the meaning encoded in the string actually produced, referred to as ‘semantic representation’ of that string. Even though for the sake of our example we assumed it was the same speaker saying (1) in Polish and (2) in English, while the gender of the speaker and of the lawyer are elements of the semantic representation of (1) they cannot be construed as being elements of the semantic representation of (2). However, at the level of semantic representation the distinction between ‘conceptual’ and ‘procedural’ is immaterial: the semantic representation of (1) is construed to contain information about the gender of the two individuals. Whether some elements of the semantic representation would be eventually delivered as encoded conceptual meaning or procedurally encoded meaning will depend on many factors, including the amount of information to be delivered within a sentence-like string and the place some of the linguistic elements would occupy within that string. To illustrate these points let us go back to the original novel on which (1) and (2) have been based. In the story, the speaking protagonist9 actually informs the detective that the said lawyer has given her some advice. Thus the original quote is: (3)

I called my lawyer, in one of those midtown firms, and she said...10

Here the speaker can rely on the procedurally encoded meaning of the pronoun she, together with the procedural contribution of the intonation pattern and particularly the absence of prominent stress on she, to give rise to a procedurally derived co-reference of my lawyer and she. Thus Encoding Grammar claims that the gender of the lawyer is encoded within (3) and is a part of its semantic representation. It is however the presence 9

Who is in fact male, but here presented as female, to allow us to refer to all speakers as she without creating an unnecessary confusion. 10 The example is truncated because what the lawyer said is immaterial to our analysis.

22

Chapter One

of the second clause in (3) that allows the speaker to encode the gender of the lawyer. It is therefore possible that the analysis of (3) within some RTinspired accounts of utterance processing (e.g. Wedgwood 2005; Sax 2011) would most likely see (3) as carrying no reference to the lawyer’s gender in the first clause – since it cannot be decoded from the first clause, just as it could not be decoded from (1) – and having it delivered to the audience only in the second clause. Yet, for Encoding Grammar in (3) the gender of the lawyer is a part of its semantic representation as a whole: it is the gender of the lawyer that allows the speaker to choose the feminine pronoun to make the co-referentiality explicit. On the other hand, it is also the fact that in (3) the word lawyer precedes the word she and not viceversa, that allows the speaker to deliver the gender of the lawyer as procedurally encoded meaning. As already mentioned, at the level of the semantic representation, Encoding Grammar draws no distinction between conceptual meanings and procedural meanings, though, as it has been shown in (3) and will be discussed in more detail in the following sections, the semantic representation is delivered to the audience through both kinds of linguistic items identified within RT: those associated with encoded conceptual meaning and those associated with procedurally encoded meaning. The nature of the semantic representation will be discussed in more detail in the following section. The story of how semantic representations are turned into sentence-like strings runs in this framework as follows. The deep syntactic module matches appropriate chunks of meaning with signifieds of linguistic signs11 of a given language as precisely as possible. Thus in (3) the chunk ‘female lawyer’ was matched with the sense of the word lawyer, with the bit ‘female’ left unencoded. At this stage of encoding the second mention of ‘female lawyer’ remains unencoded (i.e. not matched with the pronoun she, but left as ‘female lawyer’, because the choice which of the two instances of reference to the same individual will be encoded nominally and which will be encoded pronominally depends on the eventual linearization of all constituents, since swapping the places of my lawyer and she, as in (4), would encode something totally different (4)

11

I called her, and my lawyer, in one of those midtown firms, said...

Within this framework what has previously been referred to as words and morphemes are seen as linguistic signs in the Saussurean sense: their surface appearance (previously called ‘strings’) corresponds to ‘signifiers’ and their meaning to ‘signifieds’.

Cognitive Environment and Information Structure

23

When chunks of meaning are matched with signifiers of verbs, semantic roles of arguments are matched to deep-syntactic arguments, as in the MTM model. Thus the identity of the caller is mapped into the first argument of the verb ‘call’ while the receiver of the call is mapped to the second deep-syntactic argument. Some chunks of meaning are not matched with lexemes, but with syntactic patterns. Such is the case of different, often contradictory, meanings encoded in English by auxiliary inversion, e.g. (5)

a. For no reason would Harry beat his wife. b. For no reason, Harry would beat his wife. (taken from Lakoff and Brugman 1986)

The deep-syntactic representation, together with items left unencoded, is in this framework then sent to the surface syntactic module, in which, among other things, passivization may occur, final word order in freeword-order languages is imposed, previously unencoded co-referential arguments may be given their phonetic form, etc. Finally, the surfacesyntactic representation is sent to the morphological module, where grammatical morphemes if appropriate are inserted (also to encode some bits of meaning left unencoded by previously applied modules). The morphological form is then converted to a spoken, signed or written message. A small sample of items encoded by non-lexical means (or not exclusively lexical means) will be discussed in sections 2.3 and 2.4.

2.2 What Exactly Is Encoded As indicated above, the semantic representation within Encoding Grammar framework is not something that the audience will eventually decode, but rather the meaning the speaker wants to convey exclusively through use of linguistic means. This meaning is fairly independent of the language it will be eventually encoded in, though some degree of ‘thinking for speaking’ (Slobin 1996) may be involved, both in terms of choosing what to include in the semantic representation and encode, and what to leave to inference, and, at a later stage, in dividing up chunks of meaning among appropriate signifiers. Thus, the speaker we have envisioned for (2) opted for leaving the gender of her lawyer outside the semantic representation instead of saying female lawyer,12 while the speaker construed for (3) can be seen as 12 The fact that further references to the lawyer, including the information as to her gender, may occur in further discourse, is immaterial here.

24

Chapter One

having decided to deliver the gender of her lawyer as procedurally encoded meaning. On the other hand the semantic representation is somehow restricted by the speaker’s knowledge of what linguistic means she has to hand. Hence the speaker of (2) could not have said lawyeress, as there is no such word in English. The encoded meaning should be recognized and apprehended in its entirety independently of actual communicative setting, and even of its actual truth value. To illustrate this point, let’s consider: (6)

Among the novels by Sir Walter Scott, I liked best the one that the author of The Last of the Mohicans chose to set in Palestine.

adapted from Linde-Usiekniewicz (2012a, 63). Besides being a very clumsy sentence it contains a blatant factual error: Sir Walter Scott has not written, as far as anybody knows, The Last of Mohicans. Nevertheless, that is what (6) would normally be understood to mean. If that were not what it asserts, there would be no error in it for cautious readers who know their Sir Walter Scott from their James Fenimore Cooper to spot. Encoding Grammar adopts the postulate that the semantic representation of a sentence-like string is what should be apprehended out of context; this corresponds to an attempt to draw some kind of division between semantics (as seen from the speaker’s perspective) and pragmatics. However, the line is shifted far into the pragmatic area. Specifically, as has already been mentioned, the semantic representation contains chunks of meaning that will eventually be delivered by procedural and/or conceptual means, without distinguishing at that stage which means would be used (cf. Wilson and Sperber 1993). Actually, the mistaken assumption encoded in (6) was encoded mainly through procedural means. It should be noted that there is no exact match between the semantic representation discussed here and any of the RT basic concepts, such as ‘decoded’, ‘inferred’, ‘truth-conditional’, ‘propositional’, ‘explicit’, ‘explicature’ etc., though some of the RT terms, such as ‘encoded’, ‘conceptual’, ‘procedural’ etc. have been adopted here when their use suggested itself. In addition, Encoding Grammar has a mismatch between the encoded semantic representation and the apprehended meaning built into its framework. The semantic representation discussed here cannot be considered ambiguous, even when the expression eventually used to deliver it is. The only exception to this rule is the speaker wanting it to be so, when indulging in all kind of language games. On the other hand, if the delivered structure is in fact ambiguous, it may be apprehended as such, or even apprehended in a sense that was not intended by the speaker.

Cognitive Environment and Information Structure

25

It should be borne in mind that the semantic representation the speaker encodes is not identical with her communicative intention. Thus within Encoding Grammar the semantic representation of (7)

Coffee would keep me awake.

discussed amply by Sperber and Wilson (1986, 11, 16, 34, 35, 56, 167) contains only the information about the causal relation between coffee(drinking) and staying awake as it would affect the speaker, and not the acceptance or refusal of the offer. (It will be seen in section 2.4 that for Encoding Grammar it is immaterial whether (7) was used when coffee was being offered or not.)

2.3 How It Is Encoded As has been already mentioned, the speaker has her semantic representation to encode in a given language on the one hand, and on the other, all the possibilities a particular language offers and constraints it imposes. The first set of such possibilities and constraints is constituted by the lexicon, with both its ‘lexical’ words and functional words. Lexical units are equipped not only with their appropriate signifiers (phonology and sometimes intonation pattern) and signifieds (meanings) but also their syntactic properties and grammatical categories, such as tense, mood, case, gender, number etc. Not only grammatical categories, but also syntactic properties may differ from one language to another even if the signifieds are the same. Besides, a language may possess what can be informally called ‘dedicated syntactic patterns’, which are also means of encoding, e.g. English auxiliary inversion, mentioned in section 2.1 and some language specific constraints on their structures (Linde-Usiekniewicz 2006); and – last but not least – the two types of passive voice in English, to mention just a few. Some of the encoding means may actually involve a judicious mixture of available function words and the rules governing their use. One such instance can be illustrated by the well-known behavior of Polish reflexive vs. non-reflexive possessive adjectives: (8)

a. Ø1 Powiedziaá ojcu2, Īe Ø1zepsuá swój1 zegarek. ‘He1 told [his] father2 that he1 had broken his1 watch.’ b. Ø1 Powiedziaá ojcu2, Īe Ø1zepsuá jego2 zegarek. ‘He1 told [his] father2 that he1 had broken his2 watch.’

26

Chapter One

where the reflexive possessive swój encodes co-reference to surface subject, while ordinary possessive jego excludes it. Thus (8a) encodes the information that is the speaker’s watch that got broken, while (8b) encodes the information that it was the speaker’s father’s watch. More examples are provided by some pronoun-related phenomena. A significant amount of research has been done on pronouns within the RT literature (cf. Blakemore 1987; Wilson and Sperber 1993; Cram and Hedley 2005; Hedley 2005; Wilson 2011), convincingly showing how pronouns could be seen as comprising both conceptual and procedural meaning. In terms of Encoding Grammar this means that they possess this specific complex meaning as their signifieds and thus are used to deliver it to the audience – as could be seen in (3) as opposed to (4). However, it appears that less attention has been paid to the absence of pronouns as means of delivering some meaning (although cf. Leonetti 2007a, 2007b, 2011 on Romance (reduplicated) clitics as significant exceptions). Here some more examples are given, drawn from Linde-Usiekniewicz (2011). They concern the distinction between pro-drop and subject pronoun retention in Polish, as illustrated by modified Polish and English versions of the much-cited ‘donkey sentences’. Because Polish personal pronouns do not encode the distinction between humans and non-humans, but do encode gender (either real-life gender of the referent or the grammatical gender of the antecedent word), the Polish version of (9)

If Pedro has a donkey, he is happy.

has to be encoded without the subject pronoun in the second clause: (10)

JeĪeli Pedro ma osáa, jest szczĊĞliwy. ‘If Pedro-MASC has donkey-MASC, Ø is happy-MASC.’

In fact, when the subject-pronoun is retained, as in (11)

JeĪeli Pedro ma osáa, on jest szczĊĞliwy. ‘If Pedro-MASC has donkey-MASC, he-MASC is happy-MASC.’

it delivers the meaning that it is the donkey that is happy, and that the donkey’s happiness is somehow derived from its being owned by Pedro, thus the English equivalent of (11) would be something like: (12)

If Pedro has a donkey, that donkey is a happy one.

Cognitive Environment and Information Structure

27

Besides ‘dedicated syntactic patterns’ and combinations of words and patterns, languages may offer other syntactic means of encoding some chunks of sense, for example a so-called free-word-order language may, in some instances, use order together with intonation to actually encode the definite vs. indefinite character of nominals. Thus while (13)

Bank jest koáo poczty. ‘bank is next-to post-office’

when pronounced with a neutral stress pattern, corresponds to: (14)

The bank is next to the post-office.

If the order in (13) is reversed and neutral intonation preserved to produce (15)

Koáo poczty jest bank. ‘next-to post-office is bank’

the meaning encoded would be that which is encoded in (16)

Next to the post-office there is a bank.13

While some of the differences encoded through different syntactic patterns are strikingly clear, with the two patterns being actually contradictory, as is the case of English auxiliary inversion seen above in example (5), others may seem quite subtle, as is the case of (13) and (15), and tend to be unjustly relegated to stylistic differences.

2.4 Resultant Representations It will most likely have become fairly obvious by now that Encoding Grammar cannot invoke sentences (understood as purely grammatical structures, conceived in terms of subjects, predicates, objects, whichever 13

English language examples come from (Erteschik-Shir 2007, 77), while their Polish counterparts are drawn from Linde-Usiekniewicz 2012a, 90-91). The claim about word order and intonation encoding ‘definiteness’, in (13) as opposed to (15) is not to be generalized to mean that word order (even together with intonation patterns) always encodes definiteness vs. indefiniteness in Polish. In LindeUsiekniewcz (2012a) Polish examples quoted here as (13) and (15) were used as evidence that the original claim that “actual sentence division” (Mathesius 1939) did not affect the meaning of a sentence was an overgeneralization.

28

Chapter One

way they are defined) as surface representations of encoded meanings. On the other hand, since it is concerned uniquely with the language system (i.e. Saussurean ‘langue’), neither can it model utterances, i.e. sentences occurring in real use of language. Instead, an intermediate level is demanded, for which the name ‘utterance-type’ is proposed.14 An utterance-type is characterized by the lexical items featuring in it, and by its surface syntax (if any, because there are utterances, and hence utterance-types, that contain only one word). The surface syntax is the syntactic structure (conceived in terms of either constituent structure or dependency structure), the linear order of elements, and intonation pattern. Any of the distinctions listed above is sufficient to distinguish two different utterance-types (though in writing the intonation distinction is by and large lost, save for isolated typographic traces such as italics). Each and every example listed above in fact represents a different utterancetype. The validity of this ‘utterance-type’ concept (as opposed to ‘utterance’) even outside Encoding Grammar should be self-evident to anyone accustomed to reading linguistic literature: for instance, for each and every example quoted above in a language the reader is familiar with (English, perhaps also Polish), its encoded meaning was apprehended by the reader (who would otherwise have been unable to follow the argument presented), even if no context was provided. Within Encoding Grammar it is postulated that the eventual strings that result from encoding procedures briefly described in 2.1 and previously referred to as being ‘sentence-like’ correspond in fact to utterance-types and not to utterances. Further evidence for the validity of this level of abstraction having some possible psychological reality comes from the fact that even real-life speakers (as opposed to Encoding Grammar’s model of the speaker) actually rely on their real-life audiences as being able to recognize an utterance-type and process it adequately. Actually, it is utterance-types and not utterances that appear in linguistic papers (not only the present one), and even in those concerned with actual communication. In addition, writers of narratives 14

The notion of ‘utterance-type’ introduced here does not correspond exactly to the same term and level of abstraction introduced in Levinson (2000). Here it is a technical term coined to avoid the misleading use of either ‘sentence’ or ‘utterance’ in the RT senses of the terms: as can be seen from the following discussion, utterance-types invoked first of all are defined in surface terms (linearization, stress pattern). For the lack of space I cannot discuss at length how my use of the term differs from the original Levinsonian postulation. One of the important differences is framework related: the way I use the term, the meaning of an utterance-type is conventional, and a part of what it encodes.

Cognitive Environment and Information Structure

29

successfully rely on their readers to be able to recognize utterance-types represented by bits of narrative and dialogue they put into their prose. Identifying the surface structures emerging as the result of encoding procedures as utterance-types and not sentences (and thus dispensing with the very notion of sentences) brings yet another advantage to the proposed framework. It is no longer necessary to be able to distinguish between instances where different surface strings represent (variants of) the same sentence and instances where different surface strings represent different sentences. While in some cases there may be no doubt as to two different linear surface representations corresponding to different sentences, e.g. (3) and (4), (5a) and (5b), by contrast (13) and (15) have been treated as representing two linear (and possibly stylistic) variants of the same sentence. Yet, if one of the criteria adopted for distinguishing different sentences is that of difference in their semantic contents, it can be convincingly shown that even in so-called free-order languages different linearizations may correspond to different semantic contents (as could be seen in (13) and (15)). Other instances of such subtle differences – as opposed to obvious ones exemplified by (5a, b) – involve the meaning of quotative inversion with emotion verbs introducing direct speech in Polish (Linde-Usiekniewicz 2012a, 206-213) and some instances of relative order of adjectives and nouns in Spanish nominal phrases (Demonte 1999, 2008). The same difficulty arises when intonation pattern is taken into account. There is a general tendency to treat intonation as a property of particular utterances and not sentences or utterance-types. Yet, the intonation pattern is of tantamount importance as a means of distinguishing utterance-types that deliver different meaning, as it may affect the propositional form decoded from each (Fretheim 2002). One such example is provided by the distinction between defining and non-defining relative clauses. Some languages do not have separate sets of relative pronouns appearing in each type of the clause, as in English, but instead rely on intonation (mirrored by punctuation or not) to distinguish between the two. Another example comes from Danielewiczowa (2002, 57-58), who convincingly shows that in Polish sentences with certain epistemic predicates different sentence stress positions give rise to two very different propositional contents. While (17)

Jan PODEJRZEWA, Īe ma raka. ‘John SUSPECTS that he has cancer’

30

Chapter One

would have as a part of its explicature the true statement that John in fact has cancer, (18)

Jan podejrzewa, Īe ma RAKA. ‘John suspects that he has CANCER’

would not have the proposition about John’s in fact having cancer as one of its entailments.15 Almost all examples illustrating the notion of utterance-types discussed here could be rightly recognized as somehow involving differences in information structure. It would be nevertheless wrong to assume that different utterance-types simply correspond to different information structure features. Examples of utterance-types can be drawn from literature involving other issues. It can be argued that (19)

Could you open the window?

is in fact a written version of at least two different utterance-types: a straightforward query and a request, that also differ in their intonation pattern. To sum it up: utterance-types are bilateral complex signs that, by the same token, possess both formal and semantic features. The formal features comprise their syntactic features (corresponding structurally to sentences or sentence parts; corresponding to single-clause or complex sentences; having all the necessary grammatical markers; being active, passive or intransitive; having a particular set of arguments (expressed or tacit); having a particular set of adjuncts, etc.); their lexical features, i.e. all the words appearing in them, including ‘function words’; the linear order of all the elements present and the intonation pattern. Their semantic features correspond to all the meaningful elements (be they conceptual or procedural in RT terms) that can be apprehended by an audience completely out of any type of context, actual speaking situation and particular speakers. Seen within the Encoding Grammar framework, an utterance-type constitutes the final outcome of the encoding procedure: thus its formal features should match exactly the semantic representation it encodes. 15

She actually considers stressed and unstressed podejrzewaü as representing different lexical units (she lists them separately and shows, rightly, that not all Polish epistemic verbs exhibit a similar semantic change under stress). Irrespectively of whether different lexical units are in fact involved here, or just a stress shift, it is the stress pattern that allows us to recognize each utterance-type and apprehend its meaning.

Cognitive Environment and Information Structure

31

Obviously this is not true. Although utterance-types are less ambiguous than the bare sentences they would correspond to, substituting utterancetypes for sentences does not allow ambiguity to be completely dispensed with if formal features of utterance-types are viewed from the audience’s perspective. Examples of this may be found, for instance, in English sentences that support readings with either argument focus or sentence focus if the sentence final stress is absent (cf. Lambrecht 2000, 631, among others), or sentences that support readings with either argument focus or predicate focus if the sentence-final stress is maintained (cf. Nikolaeva 2001, 3, among others). Nevertheless, for the purpose of further discussion it is assumed that utterance-types are unambiguous, since they are so if viewed from the speaker’s perspective (as already mentioned in 2.2). As already mentioned, many utterance-types differ as to the information structure they encode and possibly some of the semantic differences between utterance-types are in fact consequences of information structure encoded in them and delivered to the audience. However, the validity of the very concept of information structure has to be examined in the light of its partial rejection within Relevance Theory.

3. The Case for Information Structure Sperber and Wilson (1986, 216), in an investigation of how hearers might process utterances, conclude that the “theme-rheme distinction has no place in the technical descriptive vocabulary of either linguistics or pragmatics” and provide an analysis of some of the phenomena associated with information structure in terms of the “interaction between syntax, stress assignment and the principle of relevance” (1986, 216). They do not list information structure as such among the concepts they condemn with faint praise, but only certain specific information-structure phenomena, in particular theme-rheme articulation and the notion of topic (or theme) being useful as a descriptive term. Nevertheless, it is striking that there does seem to be a sizeable body of work within Relevance Theory, or at least seeking to be broadly compatible with it, that does invoke distinctions associated with information structure (cf. Breheny 1998; Fretheim 2002; and House 2005 on stress; and Leonetti 2007b, 2011 on Spanish clitics, among others), apparently in contradiction with Sperber and Wilson’s conclusion cited above. Evidently it has not been so easy for

32

Chapter One

RT-based authors to dispense with information structure categories after all.16 Here some specific arguments will be presented in support of maintaining the notion of information structure, and specifically the theme-rheme distinction, as a part of linguistic analyses. First of all, while the interaction of syntax, stress assignment and the principle of relevance may be enough to account for the appropriate ‘intuitions’ (Sperber and Wilson 1986, 216) about how utterances get interpreted, the framework presented here tries to grapple with the speaker’s perspective: it needs to model the speaker as someone who decides upon a certain syntax (among many available patterns), assigns stress (on some constituent and not on the other), and relies on the principle of relevance to get the desired effects across. This, however, is arguably a highly framework-dependent argument in favor of information structure categories, and may be easily discarded as such. Yet there is other, cross-linguistic and theory-independent (as opposed to theory-driven) evidence for the validity of the concepts of theme and rheme. First of all there are languages that employ segmental markers for either themes or rhemes. One such well-known language is Japanese, which has the theme marker wa that is substituted for the nominative marker ga on grammatical subjects when themes, and added to object and oblique markers (Romuald Huszcza, pc.) when other syntactic arguments are encoded as themes. However, apparently the use of ga vs. wa tends to be misanalyzed in the information structure literature where it is generally claimed that ga is in fact a focus particle that also appears in thetic sentences (cf. Lambrecht 2000, 628; Ishikawa 2004, among others). This issue cannot be addressed here at length, but conceptually it is much more convincing to postulate a set of case markers (ga for Subject, o for Object, ni for adverbial of place etc.) and a thematic marker wa and to account for the fact that wa substitutes ga and follows particles marking internal syntactic arguments and adjuncts by saying that generally subjects and themes tend to coincide or rather that themes tend to be encoded as subjects whenever possible. What is more, many languages possess straightforward ‘focus particles’ that can be attached to nominal and not to verbal rhemes. Moreover, languages have syntactic patterns appropriate for marking either themes or rhemes, particularly complex themes: cleft and possibly 16

Some RT-based work does admittedly try to follow through with Sperber and Wilson’s proposed hearer-centered approach to (the nonexistence of) information structure categories – for instance Sax (2011) tries to argue this point further for sentence stress in English.

Cognitive Environment and Information Structure

33

pseudocleft sentences that have the cleft part marked as rheme, and the relative clause as theme (cf. Lambrecht 2001, among others). Romance ‘left dislocation’ occurs with clitic resumptions if the dislocated objects are themes and without if they are rhemes (cf. seminal work by Rizzi 1997). In all, some languages possess specific, syntactic or morphological means to explicitly encode either theme or rheme, not to mention languages that use special markers for subjects as rhemes and objects as themes (Aikhenvald 2003, 139ff for Tariana, where -nuku/-naku marks topicalized non-subjects (as opposed to an ordinary non-subject) and -nhe/-ne marks rhematic subjects). Obviously, one cannot reasonably claim that all these explicit segmental or syntactic means mark themes and rhemes in different languages unless an opportune definition of the terms in question is provided. And that, in turn, cannot be done at the same time in formal and functional terms, to cover all the diverse situations which the labels ‘theme’, ‘rheme’, ‘topic’, ‘focus’, etc. have been haphazardly slapped onto (cf. Linde-Usiekniewicz 2012a, 95-96, 99, 117, 120). Even in functional terms most definitions rely too strongly on actual communicative situations and contexts to be applicable to some parts of the procedural meaning that is encoded in individual utterance-types. Specifically the ‘new’ vs. ‘old’ distinction cannot be applied out of a particular communicative situation, while the distinction between ‘context bound’ and ‘context unbound’ cannot be applied to utterance-types (or out-ofcontext utterances, as discussed above). Therefore we can conclude that the rejection of the theme-rheme articulation within Relevance Theory is valid, inasmuch as any functional definition of that articulation will be incompatible with the theory.

4. Theme and Rheme Redefined The discourse-independent definition of theme and rheme proposed in Encoding Grammar, though initially conceived outside the Relevance Theory framework (Linde-Usiekniewicz 2008, 2012a) can be reformulated in its terms in such a way that it would appear to be at least partly consistent and compatible with some of Relevance Theory’s tenets. The major departures, presented above, consist of postulating an exclusively speaker oriented approach (2.1), with semantic representations containing both conceptual and procedural elements being encoded by the speaker into an utterance-type (2.2, 2.4), through available linguistic means that include syntactic patterns, function words and particles, word order and intonation (2.3, 2.4).

34

Chapter One

The theme-rheme articulation is, in the framework presented here, one of the properties of the semantic representation, couched in terms of changes in the cognitive environment of the audience. This property is one of the many facets of meaning eventually encoded in a particular utterance-type. Making reference to changes in cognitive environment is how the proposed definition falls in line in one of the basic tenets of Relevance Theory. Yet, since whatever is communicated is taken to alter the cognitive environment of the addressee, a distinction between two basic types of such changes is introduced into the framework: quantitative changes and qualitative changes. Quantitative changes correspond to a new assumption being made manifest within the cognitive environment, whereas qualitative changes correspond to an assumption already present in the cognitive environment of the addressee being brought to the fore. The distinction affects chunks of mainly conceptual meaning to be apprehended by the audience, but is encoded in terms of the kind of changes to the cognitive environment of the addressee (quantitative vs. qualitative) they should participate in. The distinction between the two types of changes (qualitative vs. quantitative) is a part of the semantic representation of an utterance-type and is delivered to the audience by encoding the appropriate parts of the semantic representation either as theme or as rheme. In consequence, some parts of utterance-type, both in terms of their meaning and in terms of how they are encoded, are thematic and some are rhematic. The proposed definitions run as follows: (20)

A thematic part in an utterance-type is any part of its meaning encoded by the speaker as intended to introduce a temporarily qualitative change to the cognitive environment of the addressee.

(21)

A rhematic part in an utterance-type is any part of its meaning encoded by the speaker as intended to introduce a temporarily quantitative change to the cognitive environment of the addressee.17

It should be noted that in both proposed definitions, the changes are presented only as intended and as temporary. Temporary is taken to mean ‘for the time necessary to apprehend the current utterance(-type)’. This is 17

This is not how the original definitions appear in Linde-Usiekniewicz 2012a, 121. Here they are presented in a revised form, which owes much to helpful queries and challenges that the initial formulations received from the Relevance Theory community.

Cognitive Environment and Information Structure

35

introduced to make the theme-rheme articulation discourse independent. Therefore whatever type of changes might have been introduced in previous discourse may be countermanded and subsequent utterances may countermand changes produced by the current utterance. Actually a single utterance is envisaged here as corresponding to just a small part of the shared cognitive environment of the speaker and her audience, and contains nothing but assumptions that mirror what is actually encoded within the utterance, unless there is an explicit procedurally encoded meaning in it that would instruct the audience to process it together with assumptions derived from a previous one.18 Therefore encoding something as theme instructs the audience to bring the assumption(s) in question from the complete set of assumptions available to him, i.e. his extended cognitive environment (metaphorically, his mental attic), to the utterancerelated scope of the shared cognitive environment, i.e. the utterancerelated immediate cognitive environment. This situation can be illustrated by chunks of meaning encoded as a rheme in a preceding utterance and as theme in the subsequent one (Daneš 1974), illustrated here by an example adapted from Sperber and Wilson 1986, 213): (22)

Bill has a twin sister. She lives in Berlin.

where she is the theme of the second utterance, though the co-referential ‘twin sister’ has been rheme in the previous one. On the other hand, the presence of Bill, encoded as theme in the first utterance, instructs the audience to bring in his assumptions about some individual thus named to his immediate cognitive environment. Conversely, utterance-internal rhemes can actually correspond, at least partly, to assumptions already present in the extended cognitive environment of the audience. This possibility will be further discussed in the next section. The notion of a portion of meaning ‘encoded as intended by the speaker’ to have a certain effect upon her audience is introduced here first to make the eventual correctness or incorrectness of the speaker’s assumptions and the eventual changes being made as intended, or not, or even not made at all, completely immaterial for the appropriateness of the definitions. Secondly, it is necessary to explain why a speaker consciously encodes parts of her meaning as intended to produce a qualitative change even when she assumes that the corresponding assumptions are absent 18

Through so called ‘discourse markers’ or even as would have happened if the meaning of (3) were represented as a series of utterances corresponding to two independent sentences.

36

Chapter One

from the cognitive environment of her audience, and conversely, encodes parts of her meaning as intended to produce quantitative changes when at the same time she encodes apparently the same part of her meaning as corresponding to something already accessible within the cognitive environment of her audience. This account actually fits valid observations about the role of focus in utterance processing made by Sperber and Wilson (1986, 209), that “[t]he processing of each implication can contribute to the overall relevance of the utterance in two ways, either by reducing the effort needed to process or by increasing its contextual effects.” Encoding something as a theme reduces the effort needed to process it, if the intended rendering of an assumption or a set of assumptions more accessible than before actually occurs. Encoding something as rheme increases the contextual effects—a new assumption is made accessible. Completely new themes, i.e. those that in fact introduce new assumptions, actually both increase contextual effects (like rhemes) and reduce the effort needed to process, as will be shown in the next section. Moreover, the distinction between an item being relevant on its own (giving an answer on its own) or raising a relevant question (Sperber and Wilson 1986, 208) actually corresponds to theme-rheme articulation: A theme raises a relevant question but never answers one, while a rheme supplies an answer to a relevant question.

5. Two Apparent Fallacies: ‘New’ Themes and ‘Old’ Rhemes 5.1 New Themes or Nothing Succeeds as Planned The ‘encoded as intended’ part of the definitions proposed above liberates the notions of theme and rheme from actual changes occurring in the cognitive environment of the audience. Since the speaker is not omniscient, some of her assumptions about the cognitive environment of her audience may be wrong. Let’s examine the situation when the speaker produces (23)

Bill’s twin sister lives in BERLIN. (Sperber and Wilson 1986, 213, original emphasis)

(therefore under the current analysis encoding ‘Bill’s twin sister’ as theme), to an audience completely unaware of the existence of Bill and of his having any sister at all, let alone a twin one. Obviously, the piece of information ‘there being a Bill who has a twin sister’ when apprehended,

Cognitive Environment and Information Structure

37

results in a quantitative change in the audience’s environment, namely introduces this assumption. However, the audience also apprehends the encoded intention of this assumption being meant to introduce a qualitative change and would retain it in his immediate cognitive environment for the duration of processing the current utterance. Thus the speaker managed to successfully deliver to her audience what she wanted to deliver in spite of her mistaken assumptions about assumptions available to her audience, though possibly the proportion between processing effort on the part of the audience and the resultant cognitive effect in him within the overall relevance of Bill’s twin sister may be different in situations when the speaker’s assumptions about her audience’s assumptions are correct and when they are not. In either case the actual communication should be “unproblematic” (Sperber and Wilson 1986, 208), since the assumption of there being a Bill who has a twin sister would be relevant to the audience and in fact necessary to process the remainder of the utterance (her living in Berlin). It is not therefore surprising that a speaker may rely on the same mechanisms when quite wittingly encoding some assumptions she assumes to be absent in the extended cognitive environment of her audience as themes. That is what actually happens in the initial sentences (or utterances) of narratives (examples abound) and when linguists present utterances to their public (as already mentioned). Thus Sperber and Wilson implicitly trusted the very same mechanism when giving the example quoted in (23), and that is what is very often done when writing or when experienced speakers talk to experienced audiences, relying on their being able to process the utterance at relatively low cost. Moreover, the necessity of encoding all assumptions known as ‘new’ as rhemes would result in a speaker who wants to deliver the meaning of (23) encoding it somehow along the lines of (24)

There is this Bill, ya know. And he has a twin sister, ya know. And this twin sister lives in Berlin.

which would really make processing effort soar sky-high, greatly trying the audience’s patience.

5.2 ‘Old’ Rhemes Are Not as Old as They Seem The problem of ‘old rhemes’ is posed by definite nominals encoded as rhemes or as a part of the semantic representation thus encoded. Let us consider one example. If a speaker delivers something like:

38

(25)

Chapter One

Maxwell killed the judge with a HAMMER. (Erteschik-Shir 2007, 30-31, original emphasis)

with the entire verb phrase encoded as rheme (i.e. in Lambrecht 2000’s terminology as predicate focus, and not as argument focus), it might appear that she is contradicting herself as far as the procedural meaning is concerned. She must have assumed that concept of ‘the judge’, is already present in the cognitive environment of the addressee or at least she has encoded it as such through the use of the definite article. At the same time, encoding it as rheme she is seen as intending to bring about some quantitative change in the set of assumptions concerning this very judge. The apparent conundrum can be solved by going back to the notion, presented in 2.1, that the encoding means the speaker has to hand include not only words, but also syntactic relations between them, and to the fact that the definition of theme and rheme makes explicit reference to ‘parts of meaning’ and not words or constituents. Thus in (25) the NP the judge as it is used, encodes not only the reference to ‘the judge’ as a part of its meaning but also the assumption as to the judge being hit: this is encoded by the simple expedient of the judge being the direct object of the verb hit, and the entire transitive construction actually encodes its semantic role of the Patient. Again, ‘the judge’ can be seen as being brought down from the attic and put to a new use: as a ‘hitee’ in the assault by Maxwell. By contrast with a hammer within the rhematic part of the utterance encodes both the concept of hammer as an object and tool with its potential lethal function, and its role of the actual weapon as intended to produce a quantitative change in the cognitive environment of the audience.

6. Conclusions A model of language use couched in terms of semantic representation to be encoded by the speaker and delivered to her audience through available linguistic means needs to see the language not only as a system of bilateral signs, but also as a set of syntactic patterns dedicated to encode some parts of meaning. Thus syntax cannot be seen as a set of abstract rules operating on equally abstract features, including those labeled Focus or Topic. On the other hand, syntactic patterns do encode not only the obvious semantic relations (Patient, Instrument, etc.), but also the information structure within an utterance-type. Moreover, information structure, and particularly the status of some utterance-type parts as themes or rhemes, affects the way utterances are apprehended or understood and therefore needs to be

Cognitive Environment and Information Structure

39

seen as encoded in them.19 As a consequence of this claim, the articulation into themes and rhemes has to be a part of the semantic representation of the utterance-type. At the same time, there is a significant difference between meaning encoded by lexical words and syntactic relations between them, and meaning related to some item being either theme or rheme. As the latter is actually defined as instructions delivered within an utterance of how the audience is expected to re-arrange his immediate cognitive environment, it falls within the scope of procedural meaning, defined in contrast with conceptual meaning within Relevance Theory. On the other hand, if theme-rheme articulation, seen as a part of meaning, is taken to be a property of an utterance-type, and not of an utterance produced in a specific context, it cannot rely on any of the discourserelated or communicative situation-related concepts that have been offered within functionally oriented information structure literature. The definitions of theme and rheme proposed here employ one of the fundamental concepts of Relevance Theory – that of ‘cognitive environment’ and changes to it – but attribute both the environment and the changes to the audience only, since speaking (as opposed to apprehending utterances) does nothing to alter the cognitive environment of the speaker and the presumed cognitive environment of the speaker is deemed immaterial, if utterance-types and not specific utterances are examined. Moreover, it goes on to propose a distinction between the possible cognitive environment of a particular audience (extended cognitive environment) and the immediate cognitive environment, which is geared to semantic elements contained within a single utterance-type. The theme-rheme distinction is defined in terms of either quantitative or qualitative changes to this environment; the former consist in bringing some assumption already present in the extended cognitive environment of the audience to the fore and is a property of themes, and the latter consists in introducing a new assumption to it, and is a property of rhemes. Paradoxically, the assumptions brought to the fore correspond to background implications of Relevance Theory, as, when the utterance is apprehended, they are immediately foreshadowed by new assumptions (rhemes) that are foregrounded.

19

As seen in contrast between (13) and (15). Actually, it is the thematic character of some utterance elements, and particularly the thematic character of datum quaestionis in both independent and embedded questions, that tricks the audience of ‘garden path sentences’ of the kind: How many of each kind of animal did Moses take on the ark? or The authorities were trying to decide where to bury the survivors.

40

Chapter One

It has also been shown that the proposed definitions are not just the ‘given-new’ distinction disguised under Relevance Theory terms, but rather conceptual devices capable of adequately accounting for completely new assumptions being themes and assumptions encoded as already present in the cognitive environment of the audience being rhemes. As the presented framework is also capable of explaining why thematic parts of utterance-types correspond to what is presupposed in them, it may be applied to lexical presuppositions as well, providing a uniform account of the intersection between theme-rheme articulation and the ‘what is presupposed’ vs. ‘what is posed’ distinction. Finally, it should be noted that the notion of theme-rheme articulation, as presented here, even in its original, unrefined form, has been successfully applied to some phenomena, which, though far from principal concerns of Relevance Theory, are nevertheless of interest to linguistics as a whole.

Acknowledgements I would like to thank Daniel J. Sax for his valuable comments on the draft version of this text and for his help in preparing the final version. I am grateful to the two anonymous reviewers for their helpful suggestions.

References Aikhenvald, Aleksandra Y. 2003. A grammar of Tariana, from Northwest Amazonia. Cambridge: Cambridge University Press. Bezuidenhout, Anne. 2004. Procedural meaning and the semantics/pragmatics interface. In The semantics/pragmatics distinction, edited by C. Bianchi. Stanford: CSLI Publications, 101-103. Blakemore, Diane. 1987. Semantic constraints on relevance. Oxford: Blackwell. Breheny, Richard. 1998. Interface economy and focus. In Current issues in Relevance Theory, edited by V. Rouchota and A. H. Jucker. Amsterdam: John Benjamins, 105-139. Cram, David and Paul Hedley. 2005. Pronouns and procedural meaning: The relevance of spaghetti code and paranoid delusion. Oxford University Working Papers in Linguistics, Philology and Phonetics 10: 187-210. Daneš, František. 1974. Functional sentence perspective and the organization of the text. In Papers on Functional Sentence Perspective, edited by F. Daneš. Prague: Akademia, 106-128.

Cognitive Environment and Information Structure

41

Danielewiczowa, Magdalena. 2002. Wiedza i niewiedza. Studium polskich czasowników epistemicznych. Warsaw: Katedra Lingwistyki Formalnej. Demonte, Violetta. 1999. El adjetivo: Clases y usos. La posición del adjetivo en el sintagma nominal. In Gramática descriptiva de la lengua española, edited by I. Bosque and V. Demonte. Madrid: Espasa Calpe, 129-217. —. 2008. Meaning-form correlations and adjective position in Spanish. In Adjectives and adverbs: Syntax, semantics, and discourse, edited by Ch. Kennedy and L. McNally. Oxford: Oxford University Press, 71100. Erteschik-Shir, Nomi. 2007. Information structure. The syntax-discourse interface. Oxford: Oxford University Press. Fortescue, Michael. 2004. The complementarity of the process and pattern interpretations of Functional Grammar. In A new architecture for Functional Grammar, edited by M. J. Lachlan and M. GómezGonzález. Berlin: Mouton de Gruyter, 151-78. Fretheim, Thorstein. 2002. Intonation as a constraint on inferential processing. In Proceedings of the speech prosody conference, edited by B. Bel and I. Marlien. Aix en Provence: Laboratoire Parole et Langage, 307-310. Hedley, Paul. 2005. Pronouns, procedures and Relevance Theory. Durham Working Papers in Linguistics, vol. 11:41-55. Hengeveld, Kees and J. Lachlan Mackenzie. 2008. Functional Discourse Grammar. A typologically-based theory of language structure. Oxford: Oxford University Press. House, Jill. 2006. Constructing a context with intonation. Journal of Pragmatics 38, no. 10: 1542-1558. Ishikawa, Kuniyoshi. 2004. Overriding effects of focus on weak NPs in thetic sentences. Proceedings from the Annual Meeting of the Chicago Linguistic Society 40, no 1:127-138. Katz, Jerrold J. 1981. Language and other abstract objects. Oxford: Blackwell. Lakoff, George and Claudia Brugman. 1987. The semantics of auxinversion and anaphora constraints. Paper presented at the Annual Meeting of the Linguistic Society of America, San Francisco. http://georgelakoff.files.wordpress.com/2011/04/the-semantics-of-auxinversion-and-anaphora-constraints-lakoff-and-brugman-1986.pdf (13 December 2011) Lambrecht, Knud. 2000. When subjects behave like objects: An analysis of the merging of S and O in sentence focus constructions across languages. Studies in Language 24, no. 3: 611-682.

42

Chapter One

—. 2001. A framework for the analysis of cleft constructions. Linguistics 39 no. 3: 463-516. Leonetti, Manuel. 2007a. Clitics do not encode specificity. In Proceedings of the Workshop Definiteness, Specificity and Animacy in IberoRomance Languages, edited by G. Kaiser and M. Leonetti. Constanza Arbeitspapier Fachbereich Sprachwissenschaft der Universität Konstanz. —. 2007b. Sobre la relación entre doblado de clíticos y movimiento de objetos. Cuadernos de Lingüística (Instituto Universitario Ortega y Gasset) 14: 135-172. —. 2011. Indefinidos, nombres escuetos y clíticos en las dislocaciones en español. Cuadernos de la ALFAL 3: 100-123. Levinson, Stephen. 2000. Presumptive meanings: The theory of generalised conversational implicature. Cambridge MA: MIT Press. Linde-Usiekniewicz, Jadwiga. 2006. On the relative depth of cleaving. In La focalization dans les langues, edited by H. Wlodarczyk and A. Wlodarczyk. Paris: L’Harmattan, 81-94. —. 2008. On some controversial issues in the description of information structure in language and utterances. Etudes Cognitives 8: 129-138. —. 2011. OĞle zdania i struktura tematyczno-rematyczna. In RóĪne formy, róĪne treĞci, edited by M. BaĔko and D. KopciĔska. Warsaw: Nakáadem Wydziaáu Polonistyki UW, 127-137. —. 2012a. From conflict through compromise to collaboration: Semantics, syntax and information structure in natural languages. Warszawa: Nakáadem Wydziaáu Polonistyki UW. —. 2012b. Acerca de las oraciones pseudorrelativas y oraciones relativas de varios tipos. Neophilologica 24: 151-166. Mathesius, Vilém. 1939. O tak zvaném aktuálním þlenČní vČty. Slovo a slovesnost 5: 171-174. Mel’þuk, Igor A. 2012. Semantics. From meaning to text. London: John Benjamins Publishing. Nikolaeva, Irina. 2001. Secondary topics as a relation in information structure. Linguistics 39, no. 1: 1-49. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Elements of grammar, edited by L. Haegeman. Amsterdam: Kluwer Academic Publishers, 281-337. Sax, Daniel J. 2011. Sentence stress and the procedures of comprehension. In Procedural meaning: Problems and perspectives, edited by V. Escandell-Vidal, M. Leonetti and A. Ahern. Bingley, UK: Emerald Group Publishing Limited, 349-381.

Cognitive Environment and Information Structure

43

Slobin, Dan. 1996. From ‘thought and language’ to ‘thinking for speaking’. In Rethinking linguistic relativity, edited by J. J. Gumperz and S. C. Levinson. Cambridge: Cambridge University Press, 70-96. Sperber, Dan and Deidre Wilson. 1986. Relevance: Communication and cognition. Oxford: Oxford University Press. Wedgwood, Daniel. 2005. Shifting the focus: From static structures to the dynamics of interpretation. Amsterdam: Elsevier. Wilson, Deidre. 2011. The conceptual-procedural distinction: Past, present and future. In Procedural meaning: Problems and perspectives, edited by V. Escandell-Vidal, M. Leonetti and A. Ahern. Bingley, UK: Emerald Group Publishing Limited, 3-31. Wilson, Deidre and Dan Speiber. 1993. Linguistic form and relevance. Lingua 90: 1-25.

CHAPTER TWO TOWARDS A RELEVANCE-THEORETIC ACCOUNT OF THE IÑUPIAQ HEARSAY EVIDENTIAL GUUQ1 SIGNE RIX BERTHELIN NORWEGIAN UNIVERSITY OF SCIENCE AND TECHNOLOGY, TRONDHEIM, NORWAY

1. Introduction This paper proposes a modal analysis of the North Slope Iñupiaq hearsay evidential guuq in a relevance-theoretic framework. The analysis is motivated by findings in data obtained through interviews with native speakers of North Slope Iñupiaq as it is spoken in Barrow, Alaska. The Inuit in Alaska call themselves ‘Iñupiat’ (MacLean 1986a), and the term ‘Iñupiaq’ is used to refer to the Inuit dialect spoken in Alaska. Iñupiaq is subdivided into ‘Seward Peninsula Inupiaq’2 and ‘North Alaskan Iñupiaq’ (Nagai 2006; Dorais 2010). North Slope Iñupiaq (alongside Malimiut Iñupiaq) is a sub-dialect of the latter.3 Besides 1

Many warm thanks to Etta Frounier, Dr. Edna Ahgeak MacLean, Uumiñaq, Beverly Faye Aqaƾƾiq Hugo, Dick Weyiouanna and Ronald Aniqsuaq Brower Sr. for sharing your reflections and knowledge of the expression guuq with me. Also warm thanks to Kaja Borthen, Thorstein Fretheim, and Daniel Sax for valuable comments on various stages of the paper. Any errors and misconceptions are entirely my own. 2 The sound represented in the orthography as ‘ñ’ is found in North Alaskan Iñupiaq, but not in the Seward Peninsula Inupiaq (Kaplan 1981, cited in Nagai, 2006). 3 ’Inupiatun’ is an alternate name for Iñupiaq (UNESCO, 1995-2010). North Alaskan Iñupiaq may sometimes be referred to as ‘North Alaskan Inupiatun’ and North Slope Iñupiaq as ‘North Slope Inupiatun’ (Lewis, Simons, and Fennig, 2013).

The Iñupiaq Hearsay Evidential Guuq

45

Iñupiaq in Alaska and a few parts of Canada, the Inuit dialect continuum consists of ‘Inuktun’ in western Canada, ‘Inuktitut’ in eastern Canada and ‘Kalaallisut’ or ‘Greenlandic’ in Greenland (Dorais 2010); it is not agreed upon to which degree the dialects located at the two extremes of this continuum are mutually intelligible. A history of cultural and social oppression (see e.g. Alton 1998) has left all native languages of Alaska endangered, Iñupiaq included (see e.g. UNESCO 1995-2010). Iñupiaq was spoken by 2,144 people in 2007 (Krauss 2007), and most of these fluent speakers are above the age of 604. Besides contributing to the description and analysis of Iñupiaq, the present paper exercises the explanatory power of Papafragou’s (2000) relevance-theoretic framework – originally developed to account for English modal verbs – by using it to account for findings in qualitative Iñupiaq data. The data consists of native speakers’ elaborated reflections and judgments of sentences with guuq in various contexts. As we shall see, the data contains varying indications with regard to the epistemic status of the proposition expressed by the material in guuq’s scope. The main question is therefore how to account for the epistemic status of the expressed proposition in utterances of the form guuq(p), given that guuq can hardly be said to encode a certain speaker attitude towards the epistemic status of the proposition expressed. Through the application of a relevance-theoretic modal analysis to an expression which restricts the information source, the present paper has two further intentions: to explore Relevance Theory as a tool for language description, while at the same time contributing to the understanding of evidentiality and modality inside a cognitively plausible theory. The paper is organized as follows: section 2 sums up existing descriptions of the expression guuq in three Inuit dialects: North Slope Iñupiaq (MacLean 1995; forthcoming), Malimiut Iñupiaq (Nagai 2006) and West Greenlandic (Fortescue 2003). The section also gives a brief introduction to approaches to evidentiality and epistemic properties of hearsay expressions in the literature (Aikhenvald 2003; 2004; Palmer [2001] 2006; Öhlschläger 1989). Section 3 presents the North Slope Iñupiaq data which motivates the proposed modal account of guuq, which is then given in section 4. The motivations for the proposed analysis are outlined in 4.1, whereas 4.2 presents Papafragou’s (2000) framework and applies it to guuq in accordance with the data. Before the summary in section 5, subsection 4.3 addresses the possibility of choosing an alternative 4

Iñupiaq is, however, taught as a second language at the schools throughout the North Slope and North West Alaska.

Chapter Two

46

description of guuq, namely by means of the label ‘attributive use’. I shall conclude, however, that given the available data on guuq, the modal analysis best captures the varying epistemic status of the proposition when guuq is used in an utterance.

2. Approaching the Iñupiaq evidential expression guuq Like other members of the Eskimo-Aleut language family, Iñupiaq is agglutinative and has a very rich morphology. As for verbs, the only obligatory inflection is an ending which specifies mood, person and number, whereas nouns are inflected for case and number. Any other affixes, including enclitics, are optional (Fortescue 2003; Lanz 2010; Nagai 2006). Guuq is an enclitic, and thus is not part of an inflectional paradigm (Nagai 2006; Lanz 2010). The Inuit expression guuq seems to give rise to meanings conveyed in English by means of phrases such as ‘it’s been told’, ‘it is said’ and ‘I’ve heard’. For instance; an utterance of the sentence in (1) below, Sigmiaq kukiurugguuq ‘Sigmiaq is cookingguuq’ is appropriate when Sigmiaq himself has told the speaker that he is cooking as well as when somebody else has told the speaker that Sigmiaq is cooking (field notes)5. However, the speaker is not obliged to use guuq when the utterance represents information obtained through linguistic report: (1)

Sigmiaq kukiu-ruq-guuq Sigmiaq cook-3.SG.IND-guuq ‘Sigmiaq is cooking-guuq.’

The sentence in (1) is in North Slope Iñupiaq, and the judgments of the suitable contexts are from interviews with native speakers of this dialect. Judging from Lanz (2010) and Nagai (2006) there is no reason to hypothesize that guuq should be much different in the neighboring subdialect Malimiut Iñupiaq. The expression guuq is described as a marker of hearsay information in Malimiut Iñuiaq (Nagai 2006) and in North Slope Iñupiaq (MacLean forthcoming). MacLean (1995) labels North Slope Iñupiaq guuq in the following ways: ‘realization marker’, ‘evidential’, and ‘reportative’. We know from MacLean (1995) that at least in North Slope 5

In the examples used in this paper, guuq attaches to the verbal word. However, guuq can also attach to the nominal. In interviews with consultants I did not have the impression that sentences with guuq were interpreted differently according to whether guuq attached to the nominal or the verbal word, other things being equal.

The Iñupiaq Hearsay Evidential Guuq

47

Iñupiaq, guuq may be used in questions, such as in (2), “... when the speaker knows that the person s/he is addressing does not have first-hand knowledge of the situation in question.” (MacLean 1995, §5.3): (2)

Su-va-gguuq? what-3.SG.INT-EVID ‘What do they say s/he is doing?’ (MacLean 1995, §5.3)

For the West Greenlandic dialect of Inuit, Fortescue (2003, 295) reports that the use of the expression guuq moves the responsibility for the validity of the proposition expressed away from the speaker, although without implicated decrease in reliability; guuq merely suggests displaced responsibility for the factual status (veridity) of the proposition without further specification of the information source. Among the existing descriptions of the Inuit expression guuq, only Fortescue (2003) directly addresses the effect of guuq on the reliability or epistemic status of the proposition, and West Greenlandic guuq appears as the kind of hearsay evidential which does not encode decreased epistemic status of the propositional content. It follows, though, from MacLean’s (1995; forthcoming) choice of the labels ‘reportative’ and ‘hearsay marker’ that also North Slope Iñupiaq guuq signals displaced responsibility of the proposition’s truth, similar to what we saw for West Greenlandic (Fortescue 2003). However, this inference in itself does not lead to a valid hypothesis about the (absence of) effects North Slope Iñupiaq guuq may have on the epistemic status of the proposition. As we shall see in section 3, the data collected for the present study shows that the use of guuq in North Slope Iñupiaq may in fact increase the epistemic status of the proposition in some contexts! Judging from the data, it thus seems that North Slope Iñupiaq guuq differs from West Greenlandic guuq in that the former’s encoded displacement of responsibility has different contextdependent effects on the epistemic status of the proposition. The question is then how to give a semantic and pragmatic analysis of North Slope Iñupiaq guuq which accounts for these varying effects in terms of epistemic status. Preferably, the analysis should moreover account for MacLean’s (1995) motivation for using the label ‘realization marker’ which we shall discuss later. Note that the objective of the present paper is only to expand the understanding and analysis of the North Slope Iñupiaq expression guuq, and as such leaves further explorations of guuq in other Inuit dialects for later research. But before looking more closely at the data

48

Chapter Two

on North Slope Iñupiaq guuq, let us take a look at the notion of evidentiality and epistemic properties of hearsay expressions. There is no consensus in the linguistics literature as to whether or not evidentiality is a kind of modality. Aikhenvald (2004) argues that evidentiality is a grammatical category in its own right, which has nothing to do with (epistemic) modality, and views non-obligatory marking of evidentiality as an evidential strategy. Aikhenvald’s (ibid.) account thus deals with grammaticalized evidential expressions, which participate in inflectional paradigms, e.g. on a par with gender marking in other languages. It is therefore not obvious to me why a chapter on evidentiality in West Greenlandic was included in Aikhenvald (2003), given that evidentiality marking in West Greenlandic comes as optional postbases and enclitics. In Aikhenvald (2003, 2004), lexical expressions disqualify as evidentials because of their optionality, while the evidential expressions in ‘Eskimo languages’ are seen as evidentials in spite of their optionality. It seems clear from the literature, as well as from the collected data, that guuq signals that the source of evidence is an individual other than the speaker. Although passing on information acquired through other individuals does not require the use of guuq, this enclitic marker seems to require that the information source is an individual other than the speaker. Because of this restriction on the information channel, I shall view guuq as an evidential expression. In Palmer ([2001] 2006), evidential meaning is treated as a kind of modality; together with epistemic modality, evidential modality forms the semantic space of propositional modality. Epistemic modals primarily indicate speaker judgment of the proposition’s truth. Evidential modals indicate the evidence for the proposition’s truth. Thus, both kinds of propositional modals convey information about the factual status of the proposition they modify (ibid.). The notion of ‘evidential hierarchy’ is sometimes used in the literature to explain why the use of expressions encoding certain information sources communicates a higher epistemic status of the proposition in their scope, as compared to other evidential expressions (see e.g. Barnes 1984, cited in Palmer [2001] 2006). Reported information is here seen as less reliable evidence than visual or other firsthand evidence (ibid.). A hierarchy indicating that hearsay evidence is less reliable will not, however, account for the properties of the marker guuq in West Greenlandic; as we saw above the use of West Greenlandic guuq does not necessarily render the proposition less reliable. When presenting the statements from speakers of Iñupiaq in the next section, it will become clear that an evidential hierarchy is inappropriate for Iñupiaq guuq for similar reasons. Aikhenvald (2004) mentions that some evidentials

The Iñupiaq Hearsay Evidential Guuq

49

may have what she calls ‘epistemic overtones’. In the Samoyedic language Nganasan, use of the hearsay marker renders the proposition highly reliable, whereas the Estonian hearsay marker indicates speaker distance towards the truth of the proposition. Other hearsay evidentials, e.g. the German modal auxiliary sollen (Öhlschläger 1989), are described as neutral in terms of speaker attitude. According to Kratzer (in Matthewson in press, 12-13), epistemic modals may have a realistic or an informational conversational background. When epistemic modals have an informational conversational background, the speaker is not committed to the truth of the proposition, whereas a realistic conversational background includes commitment to the proposition. German epistemic sollen has an informational conversational background, and hence the speaker is not committed to the truth of a proposition modified by the meaning of sollen (ibid). Due to the conceptually close affiliations between evidential meaning and epistemic modal meaning, testing a modal analysis is worthwhile in attempting to account for the epistemic status of the propositions modified by guuq in Iñupiaq utterances. The present paper views the relation between evidential and modal meaning in the following way: the meaning of an expression may lend itself to a modal description which includes restrictions on information source – i.e. some modals may have evidential properties – and expressions with such meaning properties may be described in the model used in Papafragou (2000).

3. The data The study of evidential expressions in languages without obligatory encoding of information source seems especially interesting from a semantics-pragmatics perspective. The use of guuq is not a result of how states of affairs are grammatically encoded – recall that the speaker is not obliged to use guuq when the proposition conveys hearsay information. The data displayed in this section was collected with the purpose of understanding the speaker attitude towards the epistemic status of the proposition when guuq is used in the utterance. The method followed builds on Matthewson (2004) and Peterson (2010), who show how data collection for the purpose of establishing facts about linguistic meaning is appropriately done by approaching linguistic meaning in relation to contexts. The method from Matthewson (2004) and Peterson (2010) used for the present purpose can be summarized as follows: the researcher describes an imaginary scenario, and asks the consultant about the appropriateness of uttering a certain sentence which contains the expression under investigation. Such questions also spark further discussions and elaborations

Chapter Two

50

on the meaning of sentences containing the expression under investigation. The present data thus consists of native speakers’ statements and reflections6 concerning the suitability of uttering a sentence with guuq in various scenarios, as well as their elaborations on the meaning of sentences with guuq7. I also take into account Iñupiaq texts published with English translations (MacLean 1995), and MacLean’s (1995) description of the use of guuq in Iñupiaq narratives. As will become clear, divergent speaker attitudes are reported for various uses of guuq. The presence of guuq sometimes seems to decrease the speaker’s responsibility for the truth of the proposition expressed. This is indicated by the consultant’s response to the question concerning the sentence in (3). The consultant was asked whether a speaker could be considered a liar if the state of affairs represented by the proposition in the sentence in (3) turns out not to be the case. Her reply seems to indicate that guuq may move responsibility away from the speaker: (3)

Kukiu-rug-guuq cook-3.SG.IND-guuq ‘He is cooking-guuq.’

Question: “if I say kukiurugguuq and it turns out later that he is not cooking…, was I then lying?” Answer: “kukiurugguuq, that he is cooking I heard. So I cannot call you a liar.” (fieldnotes) Also the statement concerning (4) below indicates that Iñupiaq guuq may decrease the speaker’s responsibility for the truth of the proposition8; it 6

The interviews took place in the fall 2011 in Barrow, Anchorage and Fairbanks, Alaska. The consultants who gave the data presented in this paper speak the North Slope Iñupiaq dialect. 7 Future data collection and analysis may shed further light on aspects of the communicative intention behind the choice to utter a sentence with guuq instead of a corresponding simple sentence. 8 I am aware that an utterance of Aalaak sanaturugguuq ‘Aalaak is a good carverguuq’ conveys a subjective evaluation. It may therefore be questioned if it is indeed the presence of guuq which frees the speaker from the ultimate responsibility for the truth of the proposition; whether someone is a good carver is a matter of taste, and judgments based on taste can thus hardly be considered untrue in the strict sense by another individual with a different taste. However, the consultant’s statement points in the direction that it is the reportative meaning aspect of the sentence which causes the inappropriateness of viewing the speaker as ultimately committed to the truth of the proposition. The same consultant was

The Iñupiaq Hearsay Evidential Guuq

51

seems that a speaker who has uttered Aalaak sanaturugguuq (‘Aalaak is a good carver-guuq’) cannot be held accountable by the hearer, should he later on find Aaalaak’s carvings and reach the conclusion that they are ‘not good at all’: (4)

Aalaak sanatu-rug-guuq Alaak good.at.carvning-3.SG.IND-guuq ‘Aalaak is a good carver-guuq.’

Question: If you say to me, Aalaak sanaturugguuq. And I find Aalaak’s carvings and I don’t think they are good at all. Can I then come back and say hey you told me Aalaak was a good carver! Answer: Huh-uh, cause I just told you I’d heard he was a good carver. That’s what I heard. I did not tell you he was, I heard he was. (fieldnotes) Judging from data (3-4), Iñupiaq guuq thus seems to signal a slight speaker distance towards the truth of the proposition. But other data indicate that an utterance with guuq does not decrease the epistemic status of the proposition. According to consultants’ statements, it appears that utterances of the sentences in (5-6) would communicate the respective propositions as true: (5)

a. Sigmiaq kukiu-ruq b. Sigmiaq kukiu-rug-guuq Sigmiaq cook-3.SG.IND Sigmiaq cook-3.SG.IND-guuq ‘Sigmiaq is cooking.’ ‘Sigmiaq is cooking-guuq.’

Question: [The consultant was asked to elaborate on the meaning of Sigmiaq kukiurugguuq in comparison to with Sigmiaq kukiuruq. The following statement concerns Sigmiaq kukiurugguuq:] Answer: Somebody told you, so it is positive that he IS cooking. (fieldnotes)

also asked about the utterance of sanatuniqsuq ‘he is a good carver-niq’ (niq is a marker of narrow information focus (Berthelin 2012)). Here she indicated the subjectivity of taste as a reason for the lack of speaker commitment to the propositional content. Datum (4) is hence a reasonable indicator that the use of guuq in an utterance of sanaturugguuq frees the speaker from an ultimate commitment to the truth of the proposition.

Chapter Two

52

(6)

Pilaiƾa-rug-guuq tired-3.SG.IND-guuq ‘He is tired-guuq.’

Question: Imagine a situation where somebody tells me that Peter is tired. I don’t believe that it is true. Then I meet another person who asks about Peter. Would it then be appropriate of me to say pilaiƾarugguuq? Even though I don’t believe it? Answer: No. You will be saying, he is tired. (fieldnotes) A comparison of datum (4) and datum (5) is interesting, since both data demonstrate guuq’s ability to convey that the proposition represents hearsay information. However, in (4) the use of guuq is interpreted in such a way that the speaker in the scenario is less responsible for the truth of the proposition, whereas the consultant’s explanation of the sentence in (5) seems to indicate that an utterance of that sentence would communicate the propositional content as true. Data (5-6) indicate that guuq does not always yield an interpretation of speaker distance towards the truth of the proposition. Furthermore, judging from the statement rendered in (7) it seems that guuq can even encourage contextual assumptions about a high epistemic status of the proposition9: (7)

Statement given by a consultant during our discussions of sentences with guuq: ... sometimes it is not important to know [who said p], but if it’s someone you trust or know, you take their word for the “gospel truth”. ... I’m speaking the “gospel truth”, guuq. Affirmation. (fieldnotes)

In storytelling, guuq seems to be used to strengthen the assumed probability of the proposition expressed. MacLean writes that the use of guuq “… expresses the teller's commitment of truth for the story she or he is telling. The events in the story are believed to have happened and the 9

Such potential to strengthen epistemic status is also attested among similar expressions in other languages. French puisque constrains causation, and thus it is of course the causal relation and not merely the proposition which is affected with respect to epistemic status. Similar to guuq, the meaning encoded by puisque is such that it “... enables the speaker to refer to an external authority ...” (Zufferey 2012, 143) and thereby create the assumption of indisputability, while it may also be used to convey lack of speaker endorsement.

The Iñupiaq Hearsay Evidential Guuq

53

audience accepts them as true” (1995, §5.6). Datum (8) supports this description of guuq in storytelling: (8)

Statement given by a consultant during our discussions of sentences with guuq: ... it has been reported by so many people, it’s... you might say hearsay, but it’s also been reported many times. In the context of storytelling, it’s a bit more established10.

It should be noted that the audience do not necessarily believe in every part of a story as a literal and true description of the world. However, it seems that the use of guuq is used to encourage the hearer to entertain the proposition as true inside the context of storytelling, as (9) indicates: (9)

Statement given by a consultant during our discussions of the use of guuq in storytelling: ... if you are in a storytelling mood, you have to believe it to enjoy it, otherwise you will never get into that. You just have to go along with it.

In the examples presented so far, the meaning of guuq may be paraphrased as something like ‘it is said that p’, and guuq in those examples thus seems to convey that the proposition represents hearsay information. Data (10) and (11) are taken from a story in MacLean (1995). Here the presence of guuq appears to give rise to meanings which are in English expressed by means of the phrase she realizes: (10)

(11)

10

Sinik-tuar-guuq makua igiƥmi-mi. sleep-3.PL.I D-guuq DEM.PROX.3.PL sleeping.platform-LOC ‘These are sleeping ones, she realizes, on the sleeping platform!’ (MacLean 1995)11 Uvanitchiƥ-mig-guuq uuru-t puggutta-mi. DEM-LOC-guuq boiled.meat-PL platter- LOC ‘Here on this side, she realizes, is boiled meat on a platter’ (MacLean 1995)12

I was also told that the use of guuq makes the story more dramatic and exciting. This could well be due to guuq’s ability to signal that the content was passed on by elders and hence give it an authentic feeling; Iñupiaq elders are viewed as a source of knowledge. 11 The data and translation are MacLean’s (1995). The glossing is my own responsibility.

54

Chapter Two

MacLean (1995) at some points describes guuq as a ‘realization marker’. In relevance-theoretic terms, the function of guuq in (10-11) could be one of marking attributive use; the thought represented by the proposition13 is not to be entertained as the speaker’s description of the world, but rather as a thought which is attributed to someone other than the speaker (see Sperber and Wilson [1986] 1995). The individual to whom the linguistically represented thought is attributed in (10-11) seems to be the character in the story; it appears that character is the one who realizes that there are sleeping ones on the platform in (10) and that there is boiled meat on a platter in (11). In the interviews carried out with native speakers of Iñupiaq, no one indicated that guuq could mark attributive use. When elaborating on the meaning of sentences with guuq, guuq was translated by the phrases ‘it is said/I’ve heard’ in the interviews. This could indicate that guuq may only be used to mark attributive use in narratives. Such a hypothesis, however, is not well founded, as guuq’s full range of uses in discourse genres apart from storytelling remains to be surveyed in detail. The step taken towards a semantic and pragmatic description of guuq in the present paper should therefore take into account this expression’s ability to mark attributive use. To sum up the data: guuq may be used to encourage the entertainment of the proposition as established knowledge, as indicated in data (7-9). Data (5-6) indicate that the propositional content is communicated as true when guuq is used in an utterance. It seems, however, that the presence of guuq sometimes frees the speaker from commitment to the truth of the proposition, as data (3-4) indicate. In all the data (3-9), the use of guuq is interpreted as indicating that the propositional content has been linguistically represented to the speaker by another individual, and in (1011) guuq may be interpreted as signaling attributive use of the proposition, as the thought represented by the expressed proposition is attributed to a character in the story.

4. The semantics and pragmatics of guuq 4.1 Towards a description of guuq Looking at (3-4), guuq would appear to be similar to the hearsay marker in Estonian which indicates speaker distance towards the truth of the proposition (Aikhenvald 2004). Such description does not, however, apply 12

See previous footnote. i.e. [there are sleeping ones on the platform] in (10) and [here on this side is boiled meat on a platter] in (11) 13

The Iñupiaq Hearsay Evidential Guuq

55

to guuq; data like (5-6) indicate that guuq does not encode speaker skepticism towards the truth of the proposition. Data (7-9) are further indications that guuq in Iñupiaq does not encode a weaker epistemic stance towards the proposition. It is therefore problematic to specify guuq’s ‘epistemic overtones’ in the lexicon. Also the notion of evidential hierarchy, where hearsay evidence ranges lower than firsthand evidence, would not capture guuq, due to data like (5-9). The question is how these varying indicators of the communicated epistemic status of a proposition modified by guuq are accounted for in a description of the semantics and pragmatics of this expression. It is reasonable to assume that the context in which the utterance is produced and interpreted plays a crucial role: why else is guuq associated with establishing the truth of the proposition in the context of storytelling? It is worth noting that stories in Iñupiaq society are usually passed down by elders, who are a source of wisdom. I shall return to this in subsection 4.2. A speaker who utters a sentence of the form guuq(p) is not strictly committed to the truth of p (data (3-4)), though he still seems to communicate p as true (data (5-8)). Judging from these findings, it seems to me that the speaker is committed to the link between the very experience which gave him a reason to attribute the thought that p to someone else, and the thought represented by the material in guuq’s scope. This also captures hearsay meaning, as reports may constitute an experience which gives an individual a reason to attribute the thoughts represented by that communicated proposition to the reporter. In this way, guuq shares properties with modals like English can and must; when a speaker uses the English modal must with an epistemic reading, he is also not ultimately committed to the truth of the modified proposition. Rather, he coveys that he has an experience which he sees as a good reason for believing that the state of affairs represented by the proposition in must’s scope is the case. As for guuq, the speaker has heard that p or has another reason for believing that some individual entertains the thought represented by p, and guuq conveys that it is such kind of experience which makes the speaker convey that p. Guuq is thus different from must in that the former restricts the type of information source, in that the source must be another individual. The next subsection shows how Papafragou’s (2000) framework for capturing the semantics and pragmatics of English modals may be used to account of Iñupiaq guuq.

56

Chapter Two

4.2 Accounting for guuq – a modal analysis Papafragou (2000) gives a relevance-theoretic account of the English modals must, may, can, should and ought to. The model for modal meaning is a modal relation of entailment or compatibility holding between the linguistically realized proposition in the scope of the modal and a domain of propositions. The propositions in this domain may be entertained publicly or mentally, i.e. they may or may not be realized linguistically, and the latter is usually the case (ibid.). Papafragou (ibid.) views the modal source as a domain of propositions, which may or may not be restricted, depending on the given modal expression14. Following Sperber and Wilson ([1986] 1995), Papafragou (2000) assumes that a thought, or a proposition (which is an interpretation of a thought) may be entertained and stored in memory in several different ways, e.g. as a representation of a factual state of affairs or a desired state of affairs. The modal domain is thus a set of propositions entertained in a certain way, and the modal relation holds between the domain and the proposition expressed. For instance, can conveys that the truth of the modified proposition is compatible with a set of contextually relevant propositions in the ‘factual’ domain. Papafragou (2000, 43) models the semantics of can in the following way, where D stands for domain, ‘factual’ is the domain restrictor, and ‘compatible’ is the kind of modal relation: can : p is compatible with D-factual This means that when can is used in an utterance, the speaker conveys that the modified proposition is compatible with (the truth of) propositions entertained as factual. The conceptual search-space offered by the semantically specified domain restrictor may be narrowed down pragmatically (Papafragou 2000, 43). Thus, when interpreting an utterance of the sentence in (12), the factual domain referred to by the expression can is narrowed down to include propositions representing facts about Dad and his plans for today, e.g. [Dad is sober], [Dad is not too busy to take us], etc:

14

Must is underspecified in terms of domain restriction, and requires domain selection. That is, when must is used in an utterance its domain is pragmatically saturated (Papafragou, 2000). Since guuq does seem to restrict the modal domain, I shall not go further into the notion of pragmatic saturation and underspecified modal domains.

The Iñupiaq Hearsay Evidential Guuq

(12)

57

Dad can take us.

I propose that the propositions in guuq’s modal domain are entertained as attributed (to X). The context may or may not make assumptions available concerning the individual – or set of individuals – to whom the proposition is attributed. I shall return to this at the end of the section. In other words, guuq encodes that the proposition is entailed by a proposition entertained as attributed to someone other than the speaker. In this way, guuq conveys that in view of the experience making the speaker entertain [p’ as being attributed (to x)] it follows that p is true. This gives us the following denotation for guuq: Denotation for guuq guuq : p is entailed by D-attributed According to the suggested denotation, guuq thus encodes a modal entailment relation, and the modal domain is restricted to propositions which are entertained as ‘attributed’. Guuq also encodes an aspect of procedural meaning in the shape of an instruction to place a proposition in a modal domain which logically resembles the proposition modified by guuq in the utterance. Following Sperber and Wilson’s ([1986] 1995, 232) illustration of aspects of verbal communication, the proposition in guuq’s modal domain is an interpretation of a representation attributed to an individual other than the speaker. In the denotation suggested for guuq above, the truth of a proposition modified by guuq in an utterance follows from a resembling proposition entertained as attributed (to someone). The use of epistemic must conveys that the speaker has an experience giving him a reason to entertain that p is the case. In the case of guuq, this experience is restricted in such a way that it has given the speaker a reason to attribute a resembling p to another individual. A working understanding of attribution could be as follows: The speaker has a reason to believe that someone (other than the speaker) entertains p as a description of an actual state of affairs. The speaker attributes the thought that p to another individual, or set of individuals, either because they have conveyed a resembling proposition, or because they have provided other indications leading the speaker to believe that they entertain p as true15. In this way, the speaker who utters a sentence of the form guuq(p) communicates the assumption that the truth of p follows from an experience giving him a 15

It remains to be explored exactly which kinds of experiences may constitute a reason to attribute a thought to another individual. Verbal reports are likely to count, as guuq seems so closely associated with hearsay evidentiality.

58

Chapter Two

reason to attribute p’ to another individual – i.e. the speaker expresses commitment to the link of entailment between the proposition expressed and the modal source, where the modal source is a domain of propositions entertained as attributed to someone else. In this way, the speaker is not to be considered a liar should the state of affairs represented by the propositional content turn out not to be the case, as the speaker is only committed to the entailment relation between the proposition expressed and the modal source. This accounts for data like (3-4). The suggested analysis also captures data like (5-7): An utterance of the form guuq(p) does suggest entertaining p as true, since p is not only compatible with the modal domain16 but rather entailed by it. The analysis developed so far may explain the varying epistemic status of the proposition when guuq is used in an utterance. The analysis does not, however, at this point offer any explication of why the epistemic status of the proposition seems to be increased when guuq is used in storytelling. Such an explanation shall now be attempted. Similar to what Papafragou (2000) assumes for English modals like can and should, I assume that the context in which the utterance with guuq is interpreted makes assumptions available which lead the conceptual search-space provided by the domain restrictor to be narrowed down. As for guuq, the encoded modal domain restriction provides a quite narrow search-space, since it instructs the hearer to place in a modal domain a proposition p’ which logically resembles the proposition p expressed by the utterance. In this sense guuq is different from e.g. English can; as we saw above, the specific facts constituting the modal domain in (12) are less restricted by the denotation of can and more dependent on contextual assumptions. That is, an utterance with can requires enrichment in terms of relevant assumptions about which propositions constitute the modal domain in relation to which the expressed proposition is interpreted. But there is no reason to assume that pragmatic enrichment of the meaning contributed by a modal in an utterance should only pertain to assumptions about which propositions go in the modal domain. I suggest that pragmatic enrichment of a modal may also pertain to the domain restrictor – i.e. how the propositions in the modal domain are entertained – and that such enrichment takes place when utterances with guuq are interpreted. The pragmatic enrichment of utterances with guuq may thus pertain to the whole modal domain, i.e. the propositions plus the way these propositions 16

Recall that the modal domain is a set of propositions plus the way they are entertained. There is hence no danger of the logically erroneous assumption that the truth of the embedded proposition is entailed by itself; rather, it is entailed by a resembling proposition entertained as attributed.

The Iñupiaq Hearsay Evidential Guuq

59

are entertained. Guuq’s domain is restricted to attributed propositions, and thus the whole idea of [attribution of p’] is conceptually enriched in accordance with the principle of relevance: the denotation of guuq together with the context of the utterance may make accessible certain contextual assumptions about the set of individuals to whom the resembling proposition is attributed. In a storytelling context, there are available contextual assumptions that concern the set of individuals to whom the speaker attributes the thought, namely the elders. Elders pass down stories, and elders are a source of knowledge in Iñupiaq society. It is therefore reasonable to believe that if an elder has given the speaker (or the person who told the story to the speaker) a reason to assume that she believes that p’, then p is entailed by a particularly “strong” modal source. That is, if p’ is attributed to an elder, the modal domain from which the truth of p entails is so strong that any assumption that is entailed by this domain deserves to be entertained as true. In this way, access to such contextual assumptions pertaining to the modal domain may strengthen the epistemic status of a proposition modified by guuq, because this proposition is entailed by the modal domain.

4.3 Alternative account – attributive use? Before summing up I wish to address the question whether we could have done without the modal analysis and simply called guuq a marker of attributive use or a marker of echoic use. I shall first argue why ‘echoic use’ is too restrictive for guuq. In Sissala (Blass 1989) the expression ré is – like Iñupiaq guuq – used for hearsay information and attributed thoughts, and Blass (ibid.) analyzes ré as marking echoic use. Echoic use is understood in Relevance Theory as the use of a representation (mental or public) to attribute another representation (mental or public) to someone else (or to oneself at some other time) and, crucially, to express an attitude towards it. ... the relation between the two representations is one of resemblance. (Carston 2002, 377)

The notion of echoic use seems to match the function of guuq in that guuq signals that the linguistic material in its scope represents an attributed representation. However, the propositional attitude expressed by echoic use markers seems to be either of an endorsing or of a dissociative nature (see Carston 2002, 298). Given the available data on guuq, there is no indicator that guuq is used to convey a certain attitude to the echoed thought. The speaker is not responsible for the truth of the proposition (see

60

Chapter Two

data (3-4)), while on the other hand the use of guuq may indeed strengthen the probability of the truth of the proposition in certain contexts, such as storytelling. And judging from data (5-7) it seems that guuq(p) makes the hearer entertain p and assume that p is the case, although the speaker is not ultimately accountable. In stories, guuq establishes/backs up p. Because of those findings, guuq does not seem to encode either endorsement or distance in Iñupiaq. Rather, the speaker may use guuq to leave the actual epistemic status of the proposition to the contextual assumptions. While the label ‘echoic use’ seems too restrictive for guuq, we may try for the weaker notion of ‘marker of attributive use’, which may be understood as marking that the proposition represents an attributed thought and nothing more. The actual epistemic status of the proposition when guuq is used in an utterance would then – like in the modal account – be left to contextual assumptions. The only problem would be how to capture the finding that an utterance with guuq seems to encourage the entertainment of the proposition as true (data (5-7)) while at the same time the speaker is not ultimately committed to it (data (3-4)). This may be the main difference between the ‘marker of attributive use’ account and the modal account; the latter allows us to address the finding that propositions modified by guuq seem to be entertained as true, although with a lesser degree of speaker commitment than e.g. simple statements. This aspect of guuq is accounted for by the modal relation of entailment between the proposition expressed and the modal domain. The inclusion of a modal relation in the semantic denotation of guuq accounts for the finding that the speaker is not directly responsible for the truth of the proposition expressed. And the modal relation of entailment accounts for the finding that propositions modified by guuq seem to be entertained as more than just compatible with a certain experience; the truth of the proposition is not only compatible with the experience leading the speaker to attribute the thought to a given individual or set of individuals – the truth of the proposition is entailed by this experience. I conclude that an utterance of the form guuq(p) achieves relevance not by merely signaling that the proposition represents an attributed thought, because guuq encodes that the truth of the proposition follows from the speaker’s experience leading her to attribute a resembling proposition to someone. That is, the truth of the proposition expressed follows from the speaker’s [experience of someone linguistically representing p’] or from her experience of [someone giving her reason to attribute the thought represented by p’] to them. The suggested analysis thus captures the varying epistemic status that a proposition in the scope of guuq may achieve in an utterance. Due to the modal meaning aspect, the speaker is

The Iñupiaq Hearsay Evidential Guuq

61

only committed to the link between experience and proposition. The modal relation of entailment also captures the finding that utterances with guuq may encourage the entertainment of p as true, and as established knowledge in the context of storytelling. Here the availability of contextual assumptions concerning the set of individuals to whom the thought represented by the resembling proposition is attributed yields an assumption of an especially reliable source – i.e. the elders – which then strengthens the entertainment of the proposition as true, since its truth is entailed by an especially reliable source.

5. Summary The Iñupiaq enclitic guuq is described as a hearsay marker in the literature. In other languages, the presence of a hearsay evidential may render the proposition highly reliable or convey speaker distance towards the proposition expressed (Aikhenvald 2004). Iñupiaq guuq, however, does not seem to encode any particular speaker attitude, judging from native speakers’ reflections and elaborations on the meaning of sentences with guuq. Some data indicate that the use of guuq may free the speaker from the responsibility for the truth of the proposition, while other data indicate that the propositional content is communicated as true when guuq is used in an utterance. Moreover, we have seen data indicating that the use of guuq may encourage the entertainment of the proposition as established knowledge, as well as data where guuq seems to mark attributive use of the proposition expressed. These findings motivated a modal analysis of the semantics and pragmatics of guuq along the lines of Papafragou’s (2000) account of English modals. On this analysis, guuq encodes that the truth of the proposition is entailed by propositions in the domain D-attributed. The modal semantics predicts that the speaker who uses guuq communicates commitment to the entailment relation between the modal domain and the linguistically realized proposition. The modal meaning aspect in the denotation thus accounts for the finding that the use of guuq may decrease the speaker responsibility for the truth of the proposition although an utterance with guuq seems to encourage the entertainment of p as true. The domain specification accounts for the evidential restrictions of guuq. An utterance of the form guuq(p) conveys that the truth of p follows from an experience of a certain kind, namely an experience giving the speaker a reason to attribute a thought represented by p’ to another individual. When available, contextual assumptions concerning this individual may be accessed, and because the truth of p is

62

Chapter Two

entailed by the modal domain, this explains why the use of guuq in story telling encourages the entertainment of p as true. Further research could well include a survey of how guuq is used in discourse genres apart from storytelling. It also remains to be explored what kind of experiences (besides reports) may motivate the attribution of a thought to another individual.

References Aikhenvald, Alexandra Yurievna. 2003. Evidentiality in typological perspective. In Studies in evidentiality, edited by A. Aikhenvald and R. M. W. Dixon. Amsterdam and Philadelphia, PA: John Benjamins, 131. —. 2004. Evidentiality. Oxford: Oxford University Press. Alton, Thomas. 1998. Federal policy and Alaska native languages since 1867. Ph.D. diss., University of Alaska, Fairbanks. Barnes, Janet. 1984. Evidentials in the Tuyuca verb. International Journal of American Linguistics 50: 255-71. Berthelin, Signe Rix. 2012. The semantics and pragmatics of the North Slope Iñupiaq postbase niq. MA thesis, Norwegian University of Science and Technology. Blass, Regina. 1989. Grammaticalisation of interpretive use: The case of ré in Sissala. Lingua 79: 299-326. Carston, Robyn. 2002. Thoughts and utterances. The pragmatics of explicit communication. Oxford: Blackwell. Dorais, Louis-Jacque. 2010. The language of the Inuit – syntax, semantics, and society in the Arctic. McGill-Queen’s University Press. Fortescue, Michael. 2003. Evidentiality in West Greenlandic: A case of scattered coding. In Studies in evidentiality, edited by A. Aikhenvald and R. M. W. Dixon. Amsterdam and Philadelphia, PA: John Benjamins, 1-31. Kaplan, Lawrence D. 1981. Phonological issues in North Alaskan Inupiaq. Fairbanks, AK: Alaska Native Language Center. Krauss, Michael. 2007. Native languages of Alaska. In The vanishing voices of the Pacific Rim, edited by O. Miyaoko, O. Sakiyama, and M. Krauss. Oxford: Oxford University Press. Lanz, Linda Ann. 2010. A grammar of Iñupiaq morphosyntax. Ph.D. diss., Rice University. Lewis, M. Paul, Gary F. Simons, and Charles D. Fennig, eds. 2013. Ethnologue: Languages of the World, Seventeenth edition. Dallas, Texas: SIL International.

The Iñupiaq Hearsay Evidential Guuq

63

(15 November 2013) MacLean, Edna Ahgeak. 1986. North Slope Iñupiaq grammar: First year. 3rd ed. Fairbanks, AK: Alaska Native Language Center University of Alaska, Fairbanks. —. 1995. Inupiaq narratives: Interaction of demonstratives, aspect, and tense, vols. I and II. Ph.D. diss., Stanford University. —. forthcoming. Dictionary of North Slope Iñupiaq. http://www.uaf.edu/anla/collections/search/resultDetail.xml?id=IN%28 N%29971M2011 (10 October 2012) Matthewson, Lisa. 2004. On the methodology of semantic fieldwork. International Journal of American Linguistics 70, no. 4: 369-415.

(2 July 2012) —. in press. When fieldwork meets theory: Evidence about evidentials. To appear in Proceedings of linguistic evidence, edited by B. Stolterfoht and S. Featherston.

Nagai, Tadaka. 2006. Agentive and patientive verb bases in North Alaskan Iñupiaq. Ph.D. diss., University of Alaska, Fairbanks, Alaska. Öhlschläger, Günther. 1989. Zur Syntax und Semantik der Modalverben des Deutschen. Tübingen: Niemeyer Verlag. Palmer, Frank Robert. [2001] 2006. Mood and modality. 2nd ed. Cambridge: Cambridge University Press. Papafragou, Anna. 2000. Modality: Issues in the semantics-pragmatics interface. Amsterdam/New York: Elsevier Science. Peterson, Tyler Roy Gösta. 2010. Epistemic modality and evidentiality in Gitksan at the semantics-pragmatics interface. Ph.D. diss., University of British Columbia. Sperber, Dan, and Deirdre Wilson. [1986] 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. UNESCO, 1995-2010. UNESCO Atlas of the world’s languages in danger. (24 June 2012). Zufferey, Sandrine. 2012. “Car, parce que, puisque” revisited. Three empirical studies on French causal connectives. Journal of Pragmatics 44, no. 2: 138-153.

CHAPTER THREE INSTEAD THORNSTEIN FRETHEIM, NORWEGIAN UNIVERSITY OF SCIENCE AND TECHNOLOGY, TRONDHEIM, NORWAY

1. Introduction English instead occurs on the one hand in a full prepositional phrase with the complex preposition instead of and an overt prepositional object complement and on the other hand in a zero complement, an unarticulated anaphor that needs to be truth-conditionally saturated in a context. The construction X instead of Y is used to denote an entity X that the speaker or writer contrasts with another entity Y, and the connecting item instead instructs the addressee to form the thought that X, but not Y, is a true proposition or a component of one. As a worker in the field of pragmatics, I am more interested in the bare adverbial instead that lacks a prepositional object than in instead of with its mandatory overt complement. Comprehension of utterances with a bare instead involves more inference, but there is no reason to believe that the hearer’s context-driven selection of the antecedent of the zero anaphor Y requires more processing effort than resolution of zero pronominal arguments in languages that have them. My impression is that linguists who are native speakers of English are not necessarily conscious of the existence of a zero anaphor of the sort described here and the fact that processing of a bare instead includes the addressee’s recognition of the contextual value to be associated with its zero complement. Often there is no accessible overt antecedent in the preceding discourse at all, in which case resolution of the reference of the zero anaphor Y involves activation of a bridging implicature (Clark 1977; Matsui 2000). Syntactic constructions with instead, whether the word occurs alone or with an overt prepositional object, express what looks like a relation of contrariety between two contrasted entities, which may be entire

Instead

65

propositions, or singular or general terms as constituents of propositions. Instead encodes the information that Y is non-factual but also that the states of affairs represented by the contrasted propositions are to be construed as contemporaneous. The word informs us that only one of the contrasted propositions is true at a given temporal point. Miltsakaki et al. (2003) characterize what I call ‘entity Y’ as a salient but unchosen or unrealized alternative in English syntactic constructions with instead. This would seem to be a fair description, because logical contrariety may be too strong a condition. Still, the contemporaneity condition encoded by instead seems to call for a stronger predicate than ‘unchosen’, which suggests that both propositions could have been true. At the same time, because there does not have to be an agent that selects X at the expense of Y, the predicate ‘unchosen’ is also too strong. Let me offer some English illustrations, with and without an agent who is making a choice. If I say Instead I hailed a cab, the communicated contrast is between my having hailed a cab, i.e. the asserted explicature of the declarative sentence, and an eventuality described prior to this utterance, which is the rejected alternative. In Leo drove us to the harbor instead of Carl there is an opposition between Leo’s driving the people to the harbor and the overtly expressed but unreal alternative scenario in which Carl, not Leo, was the driver. The two contrasted scenarios are mutually exclusive. Because the complement of instead of is a linguistically expressed phrase, there is no antecedent–anaphor relation here similar to what is illustrated by Leo drove us to the harbor instead. In the shorter alternative where instead has no overt complement, the noun harbor needs to be de-accented if the intended contrast is meant to be between Leo (X) and some person antecedently mentioned (Y), but harbor needs to be contrastively accented if driving to the harbor (X) is meant to exclude an act of driving somewhere else during the same time period (Y). In contradistinction to these examples, a sentence like Instead the coldest month this year was March shows that no conscious selection of X at the expense of Y is necessary when instead occurs as connective. Suppose that the month of January is a referent already activated in the discourse. The proper name January is then likely to be the intended antecedent for the anaphoric zero complement of the sentence-initial bare adverbial instead. In anaphora relations, the normal thing is for the anaphor’s antecedent to provide input to a pragmatic process of saturation that furnishes the anaphor with a contextual value that contributes a truth-condition (e.g. Recanati 2010). I am tempted to use the terms saturation and saturate more widely, so that they cover even the inferential process of replacing the zero-anaphor complement of instead with context-dependent conceptual

66

Chapter Three

information, in spite of the fact that this paper will demonstrate that the concepts associated with those complements very often fail to impact on the truth-conditional content of utterances that include them. When they do supply a truth-condition to the content, the data I have examined shows that X instead of Y is interpreted as X as a substitute for Y/X as a replacement of Y. The English connective instead and corresponding terms in other languages – written as a variable ‘instead’ hereafter, unless specified as English instead, Norwegian i stedet/isteden or German stattdessen/statt dessen, ‘instead (of that)’ – encode a procedure in the sense of Relevance Theory (Blakemore 1987, 2002; Wilson and Sperber 1993; EscandellVidal et al. 2011; Fretheim 1998, 2000a, 2000b, 2010; Fretheim, Amfo and Vaskó 2011), that is, an instruction to the addressee to pursue a specific inferential route in the pragmatic processing of the utterance. Even if the meaning of English instead seems to be decomposable as something like ‘in place of’, the conceptual value of -stead is nothing that addressees will worry about or even be aware of in their processing of English utterances with instead, and I believe the same to apply to ‘instead’ in all grammars that include syntactic constructions of the type ‘X instead of Y’. The multifunctional preposition of serves as a required link between English instead and its overt complement Y. It is required because instead is historically derived from the preposition in plus a noun stead (‘place’; cf. Norwegian and Danish sted) that has otherwise survived in the adjectives steady and steadfast, qualifying adjectives that have come to be associated with favorable concepts such as UNWAVERING, UNSHAKABLE, FIRM and with a noun like homestead. Old English stede (‘place’, ‘position’, ‘standing’) is related to the Old English verb standan (‘to stand’). In the course of its historical development, what instead of acquired as one of its senses in Present-day English is a non-locational, abstract meaning based on the assumption that someone or something replaces or substitutes for someone or something else. Latin substitnjtus is the past participle of a verb meaning ‘substitute’, ‘replace’ (the latter English verb describing literally a change of place). While the original, literal meaning of this Latin participle was ‘put under another’, it acquired ‘put in place of another’ as its dominant conceptual meaning even in Classical Latin. The morpheme -sti- of substitute derives from the Latin verb stǀ (‘I stand’). It appears in English institute and constitute as well, and one of the meanings of the Latin verb restituǀ with the same morpheme – sti – is ‘replace’. Thus, there are historical traces of a link between X instead of Y and X substitutes for Y. I believe that such

Instead

67

diachronic facts may partially account for what may look like an ad hoc replacement or substitution reading of instead. A noun meaning ‘place’ appears quite frequently in expressions meaning ‘instead of’ in a number of languages, including French (à la place de), Dutch (in plaats van)1, Spanish (en lugar de), and Italian (in luogo di), as well as English in place of, but we should not therefore conclude that the concept PLACE is a constituent of the proposition expressed. Examining English, Norwegian and to some extent German data, mostly taken from a bi-directional translation corpus, the Oslo Multilingual Corpus (OMC), I have observed that there are roughly three distinct ways in which a bare ‘instead’ can contribute positively to the relevance of an utterance. It can facilitate activation of a contextual assumption that directs the addressee to a specific inferential step, it can induce a bridging implicature which is needed for the process of saturation of the zeroanaphoric complement of ‘instead’, and it can cause the addressee to develop ‘X instead of Y’ pragmatically as ‘X replaces Y’/‘X substitutes for Y’, a pragmatic modulation of the encoded meaning of ‘instead’ which affects the explicated content of the utterance. Section 2 looks at the connective ‘instead’ when it occurs with an overt complement Y, a complement that occasionally adds a truth-condition to the explicature of the utterance. Sections 3–5 will be devoted to analyses of how occurrences of the bare ‘instead’ contribute to relevance under varying contextual conditions.2 Miltsakaki et al. (2003) is an empirical study of the discourse context of instead and the properties of what, in that context, do and do not serve as an antecedent for its anaphoric argument in the authors’ corpus. They asked corpus annotators to identify the antecedents of tokens of instead used as a bare adverbial and examined the degrees of (dis)agreement among them about what counted as the discourse antecedent that licensed a zero complement after instead. Section 3 of the present paper shows that identification of an antecedent is not always a straightforward task, because sometimes there is no unique linguistic expression in the immediately preceding discourse which plays 1

Dutch in plaats van is used when there is an overt nominal complement. There is no bare in plaats that equals English bare instead (Wim van Dommelen and Jacques Koreman, p.c.). For the anaphoric kind of complement a pronominal form is needed: in plaats daarvan (literally: in place thereof). A similar formal constraint applies in German grammar (cf. footnote 3). 2 I am grateful to Signe Rix Berthelin and Terje Lohndal for comments on the first version of this paper, and to the editors Ewa Waáaszewska and Agnieszka Piskorska for their good advice at a later stage in the writing process.

68

Chapter Three

that role. Section 4 shows that an annotator, and an addressee, will sometimes search in vain for a linguistic antecedent because the entity represented by the zero anaphor (Y) is only accessible via a bridging explicature.

2. ‘Instead of’ and its overt complement The invented example (1) illustrates a communicated contrary opposition between two linguistically represented events: ‘Joan’s visiting Berlin’ and ‘Joan’s painting the ceiling in her kitchen’. (1)

Joan went to Berlin instead of painting the ceiling in her kitchen.

The two events are understood to be mutually exclusive because the procedural meaning of instead (of) triggers the inference that the time spent in Berlin overlaps with the time she could have spent painting the ceiling at home. Joan decided to go to Berlin, although she had intended to finish that which is described in the overt gerundial complement of instead (of). Is the assumption that the two contrasted events overlap in time, which is automatically derived from a syntactic construction like (1), communicated as an implicature or as a proper part of the explicated meaning of (1)? The utterance of (1) tells us not only that a unique individual named Joan went to Berlin at some time in the past, it also tells us that she did not paint the ceiling in her kitchen and the reason why this could not happen: JOAN DID NOT PAINT THE CEILING IN HER KITCHEN AT ti-n, BECAUSE SHE WAS IN BERLIN AT ti-n, where the exact period ti-n remains unspecified, though it might be known to the addressee in advance. This enrichment is a consequence of the addressee’s following up the procedural meaning that the two events coincide. If you are in Berlin you cannot at the same time be occupied with an interior painting job in, say, Norwich. Even if neither the positive explicature nor the negative explicature and the causal link between them includes a specification of the dates when Joan was in Berlin, the encoded synchronicity entails – as opposed to implicates – that Joan’s being in Berlin made it impossible for her to paint the ceiling in her kitchen during the same period. On the other hand, the communicated assumption that it was expected that she would paint the ceiling at that time is presumably due to an implicature read into (1). The speaker implicates that he, or his source of information, maybe Joan herself, had expected Joan to paint the ceiling of her kitchen in the course of the period that she (instead) spent in Berlin. Because of the encoded temporal

Instead

69

overlap, what Joan did excludes her having a chance to do what she was expected or supposed to do. The fact that the addressee will associate instead of with a situation where the event X replaces an expected event Y, means that the proposition expressed prior to instead of and the proposition expressed in the gerundial complement of instead of cannot both be true. The addressee of (1) is told that Joan went to Berlin (X), so the adjunct starting with the connective instead of must represent an event that did not come about. The communicated fact that one event replaces another more expected event in (1) does not in my opinion justify the conclusion that the concept REPLACE is added as a constituent of an explicature communicated by the utterance of (1). Things look different when we turn to (2). Like example (1), example (2) expresses first a positive and then a negative proposition but it conveys more than an utterance of ‘Joan went to Berlin but George did not’. (2)

Joan went to Berlin instead of George.

Not only do we understand that the period when Joan was in Berlin coincided with a period when George was somewhere else; we also read into (2) the assumption that Joan went to Berlin as a substitute for George, who was supposed to go to Berlin but was not able to go there. Joan was entrusted the job or role in Berlin that was originally meant for George. I have already said that there is a diachronic semantic tie between ‘X instead of/in place of Y’ and the semantically strengthened interpretation ‘X as a substitute for Y’. Synchronically our interpretation of instead of George in (2) as ‘as a substitute for George’ is due to a lexical modulation or ad hoc pragmatic development of the meaning of instead (of) (Wilson 2004; Wilson and Carston 2007). Adding a couple of stipulated contextual premises, I would say that the explicature of (2) is JOAN REPLACED GEORGE AS A REPRESENTATIVE AT THAT MEETING IN BERLIN IN NOVEMBER, which entails both JOAN WENT TO BERLIN FOR THE NOVEMBER MEETING THERE and GEORGE DID NOT GO TO BERLIN FOR THE NOVEMBER MEETING THERE. The impact of ‘instead’ on the explicitly communicated thought impresses me as being much less questionable in (2) than in (1).

3. The bare ‘instead’ and mutually exclusive propositions The sentences in (3a–c) display some superficial similarities between the bare instead and adversative coordinating connectives like but (e.g. Iten 2005) and however.

70

(3)

Chapter Three

a. Eve wasn’t there, but there was a written note on the table. b. Eve wasn’t there. However, there was a written note on the table. c. Eve wasn’t there. Instead there was a written note on the table.

There is a concessive relation in (3a) and (3b) between the negative proposition of the first conjunct and the affirmative proposition of the second conjunct, but the communicated opposition, due to instead, between Eve’s absence and the presence of the note in (3c) is similar to what is indicated by German sondern (cf. ‘not p but rather q’/‘q, (and) not p’), as opposed to German concessive aber (cf. ‘not p but nevertheless q’/‘q but not p’). Depending on context, an utterance of (3c) may also convey as an implicature that Eve’s physical presence (‘Y’) was expected. Interchanging the order of the sentences in (3), we find that (4a) and (4b) are both possible concessive constructions, albeit the consequence connective so would definitely make more sense here than but and however if the speaker’s intention is to express much the same as in (3a– b). The final sentence in (4c), however, is completely incoherent after the preceding sentence. (4) a. There was a written note on the table, but Eve wasn’t there. b. There was a written note on the table. However, Eve wasn’t there. c. There was a written note on the table. #Instead Eve wasn’t there. In order to try to make a linguistic structure reminiscent of (3c) above relevant even if the sentences are presented in the reversed order, we would have to furnish the first sentence with a negation marker and remove the negation from the second sentence, as shown in (5) where the content of the negative proposition expressed in the first sentence saturates the zero anaphor after instead. (5)

There was no written note on the table. Instead Eve was there.

(3c) and (5) share no reading, of course, but the assumption that Eve’s presence obviates information written on a note left on the table, in (5), and the assumption that the note was left there because Eve could not be there herself, in (3c), are due to the same procedural constraint encoded by instead. These mutual exclusiveness interpretations are triggered by the procedural semantics of instead and its anaphoric zero complement, in (5) as in (3c). Still, I would not be comfortable with a paraphrase to the effect that the note on the table replaced Eve in (3c) or that Eve replaced the note in (5). The modulation proposed for (2) above, which has the

Instead

71

consequence that the concept REPLACE is entered into the proposition expressed, does not apply here. (6) is taken from the OMC corpus (http://www.hf.uio.no/forskningsprosjekter/sprik/): (6) I’ve been urging Mother to file for divorce, but so far she’s refused. Maybe I can persuade him instead. The infinitival direct object complement to file for divorce of the verb urge in the first sentence must be pragmatically developed as ‘Mother files for divorce from her husband’, and the same development is required for the zero complement of refused and the zero complement of instead. The added connective instead in the final sentence of (6) contributes positively to relevance, provided the pronoun him in the last sentence is produced with a pitch accent for contrast, a signaled contrast between what he might do and what she could have done but did not want to do. Thanks to the interplay of instead and an accented him the contrast between her filing for divorce from him and him filing for divorce from her is made more manifest. Since (6) is written, not spoken data, the reader of (6) has to avoid reading that final sentence with an unstressed him in mind. We see that the combination of instead and a contrastive him offers the reader procedural information that facilitates the mandatory truth-conditional saturation of the zero complement of persuade him. The two arguments, referring to the narrator’s mother and her husband (the narrator’s stepfather), respectively, are to be interchanged in the process of contextdriven saturation of the complement of persuade him. While utterances modified by instead often convey an implicature related to the communicator’s opinion that what did not happen was something one would have expected to happen, the narrator’s implicated propositional attitude in example (6) is one of hope rather than expectation. The modal auxiliary can and the modal embedding due to maybe are responsible for this. The antecedent of the bare instead in the English text fragment of (7) contains the verb ignore, a verb with a lexically incorporated negation component. That Prince Philip ignored an offer entails that he did not accept the offer. The zero complement of the bare instead has to be free of negation, so the zero anaphor complement of instead must be construed as a conceptual structure that contradicts the conceptual meaning of the antecedent verb phrase ignored her kind offer, namely ‘(instead of) accepting her kind offer’.

72

(7)

Chapter Three

But Prince Philip ignored her kind offer. Instead, he turned to Tony and complained, “Can’t get the hot water on; need a shave. See to it, will you?” (OMC)

Thus the bare instead in (7) supplies no truth-condition to the proposition expressed in the linguistic complement of instead but if it were omitted, a reader would be likely to infer a temporal succession and nothing more: ‘Prince Philip ignored the lady’s kind offer and turned to Tony with a complaint’. Instead causes the reader to focus on the inference that Prince Philip ignored the lady’s offer (whatever it was) because he felt that a solution to his problem with the lack of hot water from the tap was more important than any other business at the time. The reader will infer that there is a causal link between his imminent need to address Tony with the request described in (7) and his failure to accept what the lady offered him. This inference is arguably facilitated by the presence of instead. The Medieval seafarers referred to in (8) were Scandinavians whose voyages along the East coast of North America happened in the wake of the adventurous sailings of the Viking explorer Leif Ericsson. This is a passage from the English translation of a Norwegian source text. (8)

Under such circumstances it would have been sensible to establish a base by the mouth of the river, the way all the world’s settlements have begun. But at this point they made a choice that the saga does not explain. They sailed up the river instead, and passed through a quiet stretch where the water flowed calmly. (OMC)

The group of people denoted by the pronoun they reached the mouth of the river but decided not to settle there, contrary to what the first sentence in (8) describes as a sensible thing to do and contrary to “the way all the world’s settlements have begun”. They decided to sail into the river, a much less predictable continuation of their voyage. Instead is probably not a vitally important trigger of intended pragmatic processes in (8) but it does highlight the contrast between what was the expected thing to do, namely to refrain from sailing any further when they came to the mouth of the river, and what they actually did. Apart from that, the token of instead in (8) does not seem to contribute much in terms of increased cognitive effects. The middle sentence already suggests that one might have expected a documentation of why they decided not to settle by the mouth of the river. In the excerpt of (9) the contextualizing role of instead is quite clear. The word reminds the reader of the context in which the final sentence introduced by instead should be processed.

Instead

73

(9) I stood there and watched Alvin Limardo move off down the hall. I should have called him back, folks. I should have known right then that it wasn’t going to turn out well. Instead I closed the office door and returned to my desk. (OMC) The two instances of the expression I should have + Past Participle reveal the counterfactuality of the respective thoughts I CALLED ALVIN LIMARDO BACK and I KNEW AT THAT TIME THAT WHAT ALVIN LIMARDO WAS ABOUT TO DO WAS NOT GOING TO TURN OUT WELL. Instead in (9) is relevant to the reader only if it is interpreted as ‘Instead of calling him back’. Calling someone back and closing the door on that person would both be actions performed by an agent, so in the pursuit of the right antecedent for the zero complement of instead in (9), the reader must bypass the immediately preceding sentence and choose as antecedent the sentence I should have called him back, folks. Instead with its inferred complement does not give the reader any new information. The negative proposition I DID NOT CALL ALVIN LIMARDO BACK resulting from pragmatic processing of instead and its zero complement in the final sentence in (9) is implied by the earlier counterfactual information presented in the sentence I should have called him back, folks. Instead contributes to relevance by causing the reader to focus on the contrast between calling the man back and closing the door on him, and on the narrator’s implicated remorse because he did not act differently. If instead were left out, the resulting sentence I closed the office door and returned to my desk would presumably be read simply as a description of what happened when the door had been closed on Alvin Limardo, rather than as a comment on the narrator’s failure to act as he would in hindsight have preferred to do. My next example of a somewhat subtle effect of the presence of the bare ‘instead’, this time not English instead but the corresponding Norwegian connective isteden, is found in the Norwegian translation of the Turkish author Orhan Pamuk’s novel Snow (Norwegian Snø, original Turkish title Kar). The English translation is mine. (10) Etter at øpek hadde forlatt ham foran Halitpaúa-bygningen og vendt tilbake til hotellet, drøyde Ka en stund før han gikk opp de to etasjene til Velferdspartiets lokallag. Isteden streifet han en stund omkring blant læregutter, arbeidsledige og dagdrivere som drev omkring i korridorene i første etasje. (‘After øpek had left him in front of the Halitpaúa building and returned to the hotel, Ka let some time pass before he walked up the two floors to the office of the Welfare Party’s local division. Instead

74

Chapter Three

he was mingling with apprentices, job seekers and idlers who were loitering in the ground floor corridors.’) The intended reading of isteden in (10) is ‘instead of walking straight up to the Welfare Party’s office on the second floor’. This enrichment of instead leaves no impact on the reader’s comprehension of the explicature conveyed by means of the declarative starting with this word but it underlines the fact that the character Ka did something the reader has learnt that he was not expected to do. The reader has already been made familiar with Ka’s immediate plan to visit the Welfare Party’s local division. Isteden (‘instead’) seems to emphasize his reluctance to go ahead and do what he had meant to do. (11) is the English translation of a sentence appearing in a German text. It is rendered here together with the corresponding source text, which is in turn accompanied by a more literal translation into English. Instead seems to have been added as a result of the translator’s interpretation of the German sentence. There is no equivalent item in the source text. (11) If some sweetener is needed, refined sugar should be avoided, and honey, malt extract, syrup or fruit juice concentrates used instead. (OMC) ĸ

Wenn es trotzdem noch nicht süß genug ist, vermeiden wir den raffinierten Zucker und süßen mit Honig, Malzextract, Sirup oder Frucht-Dicksätten. (OMC) (‘If it nevertheless isn’t sweet enough, we avoid the refined sugar and sweeten with honey, malt extract, syrup or fruit juice concentrates.’)

The German preposition statt corresponds to English instead (of), and this preposition governs the genitive case.3 In the German source text there is no statt and no Prepositional Phrase statt dessen (‘instead of this’) with the genitive case dessen of the demonstrative prepositional object. Nevertheless the presence of instead in the target text is fully justified. The 3 The German preposition statt is a cognate of the noun Statt, which, like the noun Stelle, means ‘place’. This preposition governs the genitive case. For instance, German an meiner Stelle (‘in my place’, ‘in my stead’) equals statt meiner (‘instead of me-Gen’). The preposition statt always occurs with an overt prepositional object, but statt dessen (‘instead of this’) with a genitival demonstrative dessen is equivalent to the English bare instead. Cf. footnote 1 on the Dutch expression in plaats daarvan.

Instead

75

meaning is that refined sugar should be replaced by one of the alternative sweeteners mentioned. Recollect that replacement of one thing by another is often a salient element of the communicated meaning of syntactic constructions with instead. The assumption that ‘X replaces Y’ is a possible result of a pragmatic development of ‘X instead of Y’, a development with potential truth-conditional consequences. Is there a similar effect of instead in the English translation in (11)? In the German source text the verb vermeiden (‘avoid’) takes care of the substitution aspect of the reader’s interpretation; in the English translation the lexical meaning of the verb avoid interacts with the bare instead, to be read as ‘instead of refined sugar’. My tentative analysis is that the token of instead in (11) introduces no new truth-condition independently of the lexical semantics of avoid with its inherent negation. However, there seems to be a cline between the obvious cases where ‘X instead of Y’ needs to be enriched at the explicit level of communication by pragmatic introduction of a two-place predicate REPLACE and the most obvious cases of ‘X instead of Y’ that do not prompt activation of this predicate (cf. my comments on example (16) in section 4).

4. A bridging implicature can substitute for a linguistic antecedent in the pragmatic process of saturating a zeroanaphoric complement of ‘instead’ The illustration in (12) appeared in a text translated into English from Norwegian. The King referred to is the former Norwegian King Olav V. (12) When the King stepped onto the dock, handsome and elegant in his admiral’s uniform, the people of Mosterøy were on hand to welcome him. But a little boy, when he had gotten a look at the King, looked up at his mother and said with great seriousness, “Why didn’t you marry the King instead, Mom?” (OMC) The bare instead is here an adjunct added to a negative interrogative. Interrogatives do not communicate an explicature (Wilson and Sperber 1998), so the impact of instead and its zero anaphor on the relevance of this interrogative must be felt at the implicit level of communication. There is no linguistic antecedent in (12) that the reader can draw on in order to fill in the conceptual material to be associated with the zero anaphor. Prior to the quoted question there is no negation. There is also no overt reference to the boy’s father but the boy was making a comparison

76

Chapter Three

between King Olav and his own Dad, concluding that his mother might have done better, both for herself and for him, if she had married the King rather than the man she did marry. The implicature to be derived from the ‘wh’-interrogative with instead is obviously not the indicative negative proposition MOM DID NOT MARRY DAD but the non-indicative negative proposition MOM SHOULD NOT HAVE MARRIED DAD, given the existence of a dashing candidate husband like King Olav. This implicature takes over for the missing antecedent before the interrogative. Incidentally, the Norwegian source text did not contain the marker isteden/i stedet but heller, meaning ‘rather’, the literal meaning of the interrogative in the source text being ‘Why didn’t you rather take the King, Mom?’ (13) is a statement made by the former Norwegian prime minister Gro Harlem Brundtland on December 10, 2012, the day when the European Union was awarded the Nobel Peace Prize in Oslo. The English translation is mine. (13) Jeg synes EU er en verdig vinner. Det at europeiske land er i krise er ikke noe argument mot selve tildelingen. Her må vi isteden finne løsningene på tvers av landegrensene. (‘I think the EU is a worthy winner. That European countries are in crisis is no argument against the award itself. Here we must instead find the solutions across the national borders.’) There may be not just one but two bridging implicatures in (13) which are triggered bottom-up by two different anaphors, her (‘here’) and the zero anaphor associated with the bare isteden (‘instead’). The bridging implicature that the latter gives rise to is THE FACT THAT THE EUROPEAN UNION WAS AWARDED THE NOBEL PRIZE FOR PEACE HAS BEEN CRITICIZED.

The English text fragment in (14) translates the German text underneath. (14) One evening in April 1943 a German soldier was shot dead in the street. The alarm was raised: SS and Polish police officers in civilian clothes searched the nearby houses for hidden weapons. Instead they found Simon Wiesenthal. He was marched off for the third time to, as he believed, his certain execution. (OMC) ĸ

German source text: An einem Abend im April 1943 wurde in der Straße, in der sich das Haus befand, ein deutscher Soldat erschossen. Es gab Alarm: SSLeute und polnische Kriminalbeamte in Zivil durchkämmten die

Instead

77

umliegenden Häuser nach versteckten Waffen. Statt dessen fanden sie Simon Wiesenthal. Er wurde zum drittenmal zu seiner, wie er glaubte, sicheren Erschießung abgeführt. (OMC) Instead prompts the reader’s search for a negated antecedent, but no negation appears in the sentences prior to Instead they found Simon Wiesenthal in (14) or the source text’s Statt dessen fanden sie Simon Wiesenthal. The reader must therefore construe a negative antecedent on the basis of the combination of instead and the information about the search for hidden weapons. What actually happened was that members of SS together with Polish police officers found Simon Wiesenthal, the famous post-war hunter of Nazi war criminals, in one of the houses that they entered. Consequently a bridging implicature that represents something they failed to find needs to be activated in the reader’s mind. NO HIDDEN WEAPONS WERE FOUND is this negative bridging implicature, and the interpretation of the next-to-final sentence in (14) that makes instead relevant is therefore ‘Instead of finding any hidden weapons they found Simon Wiesenthal’. One would not normally say that the disclosure of Simon Wiesenthal’s hiding place replaced discovery of the hidden weapons they were looking for. The English verb search and the corresponding German verb durchkämmen (literally: comb through) do not tell us whether the objects sought were found, but instead and statt dessen buttress the reader’s inference that no hidden weapons were found. They make the sentences that they introduce relevant in the context made manifest by the meaning of the previous sentence, via the mentioned bridging implicature. These connectives cause the reader to infer that the only discovery of any interest was Simon Wiesenthal. (15) is a caption that accompanied footage showing some US soldiers in Afghanistan and what looks like a gigantic explosion in the background, caused by an ambushing enemy. (15) Marines encountering an I.E.D. Had the bomber waited a few more seconds, the patrol would have been on top of the explosive. Instead, nobody was seriously wounded. (International Herald Tribune, October 2009) The zero anaphor after instead cannot be ‘(instead of) being on top of the explosive’ extracted from the apodosis of the counterfactual conditional. Readers are expected to be cognizant of the fact that if one is on top of an explosive – in this case a detonated roadside bomb (I.E.D.) – there is a strong risk that one will be killed instantly. (15) reminds the reader of this

78

Chapter Three

via a bridging negative implicature – THE PATROL WAS NOT BLOWN UP – derived from the counterfactual conditional, and this implicature serves as antecedent information that helps the reader saturate the zero anaphor as something with the meaning of ‘(instead of) experiencing that the patrol was blown up’. The rest of the final sentence of (15) will then be enriched as ‘the marines represented in the picture found that no one among them was seriously wounded’. This in turn yields the strong scalar implicature that no one was killed. (16) is my final example of how the anaphoric zero complement of instead can be saturated by means of an implicature that takes on the role otherwise assumed by a linguistic antecedent. It was found in a Norwegian daily newspaper. The interpreter referred to is a native speaker of Somali living in Trondheim who works as an interpreter in encounters between Somalis and Norwegian officials, including health personnel. (16) Tolken uteble. Møtet ble isteden gjennomført med en stor dose kroppsspråk og peking på de kroppsdelene som gjorde mest vondt. (Adresseavisen, November 2012) (‘The interpreter failed to appear. The meeting was instead carried out with a large dose of body language and pointing at those parts of the body that hurt the most.’) The bare isteden in (16) will not be read as ‘instead of the interpreter’. Simultaneous interpretation in both directions, from Somali into Norwegian and from Norwegian into Somali, is the interpreter’s job and because he was not there, they were forced to improvise. The negative bridging implicature that the reader of (16) will access is something like this: THE CONVERSATION WAS NOT CONDUCTED IN SPOKEN LANGUAGE. This inferred assumption substitutes for a possible linguistic structure that might have been a reasonable candidate antecedent for the zero complement of isteden. The content of this implicature is derived on the basis of information expressed in the first sentence of (16) plus the later information that non-verbal language – body language and pointing – was the medium of exchange of information between a Somali woman identified earlier in the article and local Norwegian health personnel. Communication with the help of verbal discourse (Y) is replaced by nonverbal communication (X), so the example in (16) could arguably have been put in the same category as the examples offered in the next section, which deals with uses of English instead that have consequences for the way that the proposition associated with X is represented in the addressee’s mind.

Instead

79

5. A bare ‘instead’ can trigger modulation at the explicit level of communication Provided the reference is to the same Joan and the same George, the second sentence Joan went instead in (17) means the same as sentence (2) in section 2: Joan went instead of George. (17) George couldn’t go to Berlin. Joan went instead. Specifically, (17) communicates more than the minimally different sentence sequence presented in (18). (18) George couldn’t go to Berlin. Joan went. Joan went in (18) means that Joan did go to Berlin, unlike George, who did not go. In the absence of contextual guidance this sequence of sentences does not necessarily cause us to infer that Joan was a substitute or stand-in for George. Joan went may be interpreted concessively in the same way as the alternative Joan DID go. George’s planned trip to Berlin and Joan’s actual trip to Berlin referred to in (18) may even have been at different times; they are not necessarily related in any way. The bare instead in (17), however, does for the addressee what the PP instead of George did in example (2) in section 2. If Joan went to Berlin instead of George, she replaced him. The exchange between A and B in (20) below is another example of how instead and its zero complement triggers a context-driven modulation of the encoded meaning of ‘X instead of Y’ in the direction of ‘X replaces Y’. Let us first consider the corresponding conversation in (19) where there is no token of instead in B’s interrogative. (19) A: We could buy strawberries. B: What about raspberries? Speaker B in (19) may or may not be interested in strawberries. B’s suggestion could be that it might be a good idea to buy raspberries as well, or else that B’s preference is raspberries. Instead in (20), however, constrains the interlocutor’s pragmatic search for the intended content of B’s utterance. The meaning may be paraphrased as I SUGGEST THAT WE REPLACE YOUR PROPOSAL TO BUY STRAWBERRIES BY MY PROPOSAL TO BUY RASPBERRIES.

80

Chapter Three

(20) A: We could buy strawberries. B: What about raspberries instead? For B in (20), one type of berry excludes the other type of berry. B could have responded And maybe raspberries? but if we let the coordinating connective and co-occur with instead, the result is confusing: ??And maybe raspberries instead?. Attested examples of a bare ‘instead’ that affects the speaker’s explicit content are truly hard to come by. I consider (21) to be one such example. (21) Meat-eaters can do without meat and also fish for a while. Instead there will be plenty of raw vegetables on the menu, perhaps exclusively on some days, accompanied by soured milk products. (OMC) The presence of instead is crucial in (21). The first sentence expresses a general statement but the next declarative starting with instead informs us of the policy at the institution described prior to this text fragment. One learns that certain kinds of food replace the regular menu on certain days. Raw vegetables and soured milk products occasionally substitute for meat and fish. In addition to the explicated affirmative proposition RAW VEGETABLES AND SOURED MILK PRODUCTS ARE SOMETIMES SERVED IN THE INSTITUTION REFERRED TO and the explicated negative proposition MEAT AND FISH PRODUCTS ARE SOMETIMES NOT SERVED IN THE INSTITUTION REFERRED TO there is a communicated explicature

that entails these, which may be paraphrased in the following way: SOMETIMES RAW VEGETABLES AND SOURED MILK PRODUCTS REPLACE MEAT AND FISH AT MEALS SERVED IN THE INSTITUTION REFERRED TO. This semantically strengthened interpretation is much

easier for the reader to access when instead is present.

6. Conclusion The connective ‘instead’ occurs in two types of syntactic construction in a number of languages. Its complement is either an overt nominal constituent (‘instead of NP’) or ‘instead’ is not accompanied by a linguistically filled complement (a bare ‘instead’). Some languages, including Dutch and German (footnotes 1 and 3), do not permit a bare ‘instead’ but require a pronoun that carries no descriptive meaning in corresponding contexts. The lexical meaning of this connective is the same whether or not there is an overt complement there. Due to its procedural meaning, ‘instead’ instructs the addressee to construe X of the relation ‘X

Instead

81

instead-of Y’ as factual and Y as non-factual. When ‘instead’ is not accompanied by an overt complement, the value of the zero complement of the connective is determined contextually, in a manner similar to the relation between a zero anaphor and its linguistic antecedent. The illustrations presented in section 4 demonstrate that an overt linguistic antecedent for a zero-anaphoric complement Y of ‘instead’ may not always be accessible to the addressee. Our pragmatic interpretation of the entity associated with X is factual and the entity associated with Y non-factual (because ‘instead’ instructs the hearer to make it so). Hence the contrasted propositions that are the members of the ‘instead’ relation have opposite polarity values and the linguistic antecedent of a zeroanaphoric Y will therefore normally be an antecedent that expresses negative polarity. If no negatively marked linguistic antecedent is available, the addressee has to activate an implicature with a negative polarity whose function is equivalent to a proposition that could have been expressed by an overt negative antecedent located in the immediately preceding discourse. This implicature has the role of a contextual assumption that saturates the zero-anaphoric complement Y in such a way that the addressee’s expectation of a relevant utterance is satisfied. Example (2) in section 2 illustrates how instead can trigger pragmatic inference that results in a semantic strengthening of ‘X instead of Y’ to ‘X replaces Y’ when Y is an overt prepositional object. Section 5 demonstrates that the English bare instead (and its zero complement) sometimes brings about a context-driven modulation that has a similar impact on the addressee’s mental representation of the explicature of the utterance. What regularly happens in such cases is that a predicate whose semantic value is equivalent to the concept REPLACE is understood to be a constituent of the proposition expressed. This context-driven modulation of the grammatically encoded meaning of sentences modified by instead is far from surprising, given the intimate historical ties between the English lexical item instead and its counterparts in other languages on the one hand and the concept REPLACE on the other hand. In place of is a fairly common English alternative to the more frequently occurring expression instead of. We all know the meaning of the noun place in this expression, and we appreciate the idea that something may metaphorically be put in a place occupied by something else, thus ousting the latter from its position. This does not imply that the concept associated with the noun place in this metaphorical expression is entered into the proposition expressed. In place of is a relational expression that connects two entities supplied to two contrary propositions. Its encoded meaning is of a procedural rather than conceptual sort, despite the fact that the noun place is its central element.

82

Chapter Three

With the more frequently used expression instead (of), the absence of a conceptual meaning supplied to the proposition expressed is even more obvious. Whether or not a given speaker associates the morpheme -stead with a specific concept, instead does not supply a concept to the proposition expressed, it instructs the addressee to draw inferences about the communicator’s informative intention in such a way that the two propositions connected by means of instead are understood to be mutually exclusive.

References Blakemore, Diane. 1987. Semantic constraints on relevance. Oxford: Blackwell. —. 2002. Relevance and linguistic meaning: The semantics and pragmatics of discourse connectives. Cambridge: Cambridge University Press. Carston, Robyn. 2002. Thoughts and utterances: The pragmatics of explicit communication. Oxford: Blackwell. Clark, Herbert H. 1977. Bridging. In Thinking: Readings in cognitive science, edited by P. N. Johnson-Laird and P. C. Wason. Cambridge: Cambridge University Press, 411-420. Escandell-Vidal, Victoria, Manuel Leonetti and Aoife Ahern, eds. 2011. Procedural meaning: Problems and perspectives. Bingley, UK: Emerald Group. Fretheim, Thorstein. 1998. Intonation and the procedural encoding of attributed thoughts: The case of Norwegian negative interrogatives. In Current issues in Relevance Theory, edited by V. Rouchota and A. H. Jucker. Amsterdam: John Benjamins, 205-236. —. 2000a. Procedural encoding of propositional attitude in Norwegian conditional clauses. In Pragmatic markers and propositional attitude, edited by Gisle Andersen and Thorstein Fretheim. Amsterdam: John Benjamins, 53-84. —. 2000b. Constraining explicit and implicit content by means of a Norwegian scalar particle. Nordic Journal of Linguistics 23, no. 2: 115-162. —. 2010. Anaphoric and non-anaphoric uses of the Norwegian adverb først (‘first’): A pragmatic analysis based on a univocal lexical meaning. In Memory, mind and language, edited by H. Götzsche. Newcastle: Cambridge Scholars Publishing, 163-178. Fretheim, Thorstein, Nana Aba Appiah Amfo and Ildikó Vaskó. 2011. Token-reflexive, anaphoric and deictic functions of ‘here’. Nordic Journal of Linguistics 34, no. 3: 239-294.

Instead

83

Iten, Corinne. 2005. Linguistic meaning, truth conditions and relevance: The case of concessives. Basingstoke: Palgrave. Matsui, Tomoko. 2000. Bridging and relevance. Amsterdam: John Benjamins. Miltsakaki, Eleni, Cassandre Creswell, Katherine Forbes, Aravind Joshi and Bonnie Webber. 2003. Anaphoric arguments of discourse connectives: Semantic properties of antecedents versus non-antecedents. Proceedings of the EACL Workshop on Computational Treatment of Anaphora, 53-59. Recanati, François. 2010. Truth-conditional pragmatics. Oxford: Clarendon Press. Wilson, Deirdre. 2004. Relevance and lexical pragmatics. UCL Working Papers in Linguistics 16: 343-360. Wilson, Deirdre and Dan Sperber. 1998. Mood and the analysis of nondeclarative sentences. In Pragmatics: Critical concepts, vol. II, edited by A. Kasher. London: Routledge, 262-289. Wilson, Deirdre and Robyn Carston. 2007. A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In Pragmatics, edited by N. Burton-Roberts. Basingstoke: Palgrave, 230-259.

PART II: SPECIALIZED DISCOURSES AND RELEVANCE THEORY

CHAPTER FOUR PUTTING RELEVANCE AT CENTRE STAGE IN RESEARCH ON HUMAN ACTIVITY ON THE INTERNET FRANCISCO YUS, UNIVERSITY OF ALICANTE, SPAIN

1. Introduction Years ago, the Internet was a mere complement to our physical activities, a limited text-based environment we had to log onto, a poor cues-filtered scenario for human communication. Nowadays, by contrast, the Internet is hybridised and grafted into our daily lives. We do not log onto the Net since we never really get disconnected from it. Instead of paying to get an Internet connection, today many users would rather pay to get disconnected from it. Besides, we no longer need computers to access the Net; we carry it with us on our mobile phones. As a consequence, we now devote an enormous amount of mental activity and time to obtaining information and communicating with others online, either through text or with the aid of pictures, video and multimodal combinations of verbal and visual inputs. The Internet has seen a radically accelerating pace of development and nowadays, with the aid of smart phones and tablets, it is accessible everywhere. What is especially interesting in the context of this relevance-centred book is the increasing intermeshing of our physical and virtual lives with the aid of technology, how we combine traditional physical settings and virtual spaces to manage our networks of friends and relatives (Yus 2005b, 2007), how our identities are increasingly shaped in virtual scenarios (Yus 2002, 2012b, 2013). As Turkle (2011) summarizes things, nowadays technology is “always on, always on us.” According to Relevance Theory (Sperber and Wilson 1995), we always aim to obtain rewards (positive cognitive effects) in exchange for an

Research on Human Activity on the Internet

87

adequate amount of mental effort (in theory, the less effort the higher relevance) when we process inputs (mental, verbal, visual, multimodal, etc.). This search for relevance is an evolved capacity, rooted in human psychology, and hence it is at work in any human processing of information, regardless of whether it takes place in physical or virtual scenarios. The aim of this chapter is to show that Relevance Theory not only explains human-to-human communication on the Internet, but can be extended to all human searching for and exchanging of information on the Net, including interactions that take place between users and computer systems. Four areas of relevance-driven processing will be studied in the next headings: (a) relevance sought by the system for the user; (b) relevance sought by the user in the system; (c) relevance sought by the user in another user’s coded input (e.g. an utterance); and (d) relevance sought by the user in a group of users.

2. Cognition and the search for relevance Much of relevance-theoretic research has focussed on how addressees turn the speaker’s schematic logical form of his/her utterance into (explicit and/or implicated) meaningful interpretations with the aid of context. Much less attention has been paid to the fact that all cognitive activity of the human mind is relevance-oriented, beyond the processing of verbal stimuli. All inputs for mental processing are subject to what is called the cognitive principle of relevance: “Human cognition tends to be geared to the maximisation of relevance.” In general, human cognition engages is typical activities such as the ones listed in (1a-e): (1)

a. b. c. d. e.

To filter out information from outside the individual that hypothetically is not relevant. To update the general picture of the world that the individual possesses. To identify intentions and attitudes that underlie the interlocutors’ communicative activity. To combine new information that reaches the individual’s mind with information which is accessible from context. To select from context only the information that is going to be useful for the derivation of relevant conclusions.

According to Wilson and Sperber (2002, 251), “relevance is a potential property not only of utterances and other observable phenomena, but of

88

Chapter Four

thoughts, memories and conclusions of inferences.” For example, we filter out potentially irrelevant visual phenomena from the outside world. We are unable to process the whole barrage of in-coming information from the environment and we have developed an ability to focus only on potentially interesting phenomena. The same applies to our thoughts, some of which are more likely to be entertained than others in a particular situation. Consider the listed possible thoughts (3a-g) in situation (2): (2) (3)

The bell in my house rings. a. Someone has rung the bell. b. The bell in my house has rung. c. The person ringing is not a dwarf (he or she can reach the bell). d. The electricity supply has not been cut. e. The electricity company has not gone bankrupt. f. Nobody has stolen my doorbell. g. I’ve paid my last electricity bill.

It is obvious that (3a, b) are the most likely thoughts for the stimulus in (2), even though all of (3a-g) are, in theory, possible (manifest, in relevance-theoretic terms). However, thoughts are ranked in terms of likelihood in a context, and other thoughts might be more relevant in a different context. Imagine, for example, that I have been suffering an electricity cut that has lasted several days. Upon hearing the ring, (3d) would be very likely to be entertained in that context, but perhaps not in a normal context of non-stop flow of electricity. In general, any external stimulus or internal representation registered by our cognitive system may be relevant to us at some time and we have a rooted ability to pick up potentially relevant stimuli. As Wilson and Sperber (2002, 254) stress, As a result of constant selection pressure towards increasing efficiency, the human cognitive system has developed in such a way that our perceptual mechanisms tend automatically to pick out potentially relevant stimuli, our memory retrieval mechanisms tend automatically to activate potentially relevant assumptions, and our inferential mechanisms tend spontaneously to process them in the most productive way.

For an input to be relevant, it has to combine fruitfully with context to yield positive cognitive effects (“a positive cognitive effect is a worthwhile difference to the individual’s representation of the world – a true conclusion, for example. False conclusions are not worth having. They are cognitive effects, but not positive ones,” Wilson and Sperber

Research on Human Activity on the Internet

89

2002b). Indeed, human cognition is constantly gathering information and combining it with contextual information to yield relevant conclusions. In Yus (2011a, 11-12) the following example is provided: (4) (5)

(6)

New information (visual input): A yellow Mercedes is parked near our Department. Information already available (from encyclopaedic knowledge): a. Professor Smith, who supervises my thesis, owns a yellow Mercedes. b. Professor Smith usually takes the bus to the university. c. Only when he intends to stay at university till late in the evening does he drive his car to university (since there are no late buses returning to where he lives). (Relevant) conclusion (inferred by combining (4) and (5)): This evening I will be able to discuss with my supervisor at length how my thesis is progressing.

In this example, a student is walking on campus, sees a yellow Mercedes and concludes (6). Relevance Theory claims that “what makes an input worth picking out from the mass of competing stimuli is not just that it is relevant, but that it is more relevant than any alternative input available to us at that time” (Wilson and Sperber 2002, 252) and in the environment there is no other visual input more relevant than that car at that time. However, there is nothing in the Mercedes that leads to this conclusion, unless the student combines the visual input with the background information (context) in (5). Verbal communication is a very sophisticated human tool for conveying relevant interpretations out of coded thoughts. However, in essence it does not differ radically from the overall procedure of the human mind, always geared to the maximisation of relevance. In communication, we also combine contextual information with verbal inputs to yield relevant conclusions which can only be drawn from the combination of these informative sources and not from these taken separately. Consider the example in (7): (7)

(8)

New information (verbal input from Ann): Tom: Are you going to next Saturday’s party? Ann: My parents are away this weekend. Information already available (from encyclopaedic knowledge): a. Ann lives with her parents and her grandmother. b. When Ann’s parents are away she has to look after her grandmother.

Chapter Four

90

c. (9)

Someone who has to look after a relative cannot normally go to parties. (Relevant) conclusion (inferred by combining (7) and (8)): Ann cannot go to the party next Saturday.

Again, there is nothing in Ann’s utterance in (7) that immediately leads to (9) unless it is combined with contextual information (8). In this example, Tom accesses just the right type and amount of contextual information to yield this conclusion and, to do so, he also relies on his rooted tendency to maximise relevance.

3. Relevance, cognition and cyberpragmatics Our mental activity for communication on the Net, even if sometimes less contextualised than face-to-face communication, should follow the same inferential strategies as oral conversations and these should also follow the overall relevance procedures of human cognition in general. This is the interest of what is called cyberpragmatics (Yus 2001, 2010, 2011a, 2011b, 2012a). But on the Net the placement of discourses in the oral-written, visual-verbal and synchronous-asynchronous axes influences enormously the users’ search for relevance and the interpretive outcomes obtained from processing these online discourses. Specifically, several aspects are stressed and addressed by cyberpragmatics: (a) There is no difference in how users predict relevance and look for relevance in online discourses when compared to offline ones. In other words, when interpreting messages on the Net Internet users resort to inferential strategies that do not differ from the ones that they use in the interpretation of utterances in communication framed by physical copresence. There is only one biologically rooted ability to turn verbal stimuli into relevant interpretations, irrespective of the type of utterance, the channel used, and the richness of contextual information. (b) Addresser users expect their virtual interlocutors to access some specific contextual information that enables them to reach the intended interpretation of their messages. Similarly, addressee users invariably access contextual information as an inherent part of their relevanceseeking inferential activity. However, context accessibility is often constrained by the cues-filtered quality of online discourses. Certainly, the design of the computer applications for Internet communication (chat rooms, Messenger, e-mail, web pages...) has an impact on the balance of cognitive effects (i.e., interest) and mental effort obtained during the relevance-seeking interpretation of the information contained therein.

Research on Human Activity on the Internet

91

(c) There is a link between relevance and the command of the interfaces used for Internet-mediated communication. Non-expert users face supplementary mental effort even before interactions with other users or the processing of documents from web pages start. Expert users, on the contrary, will obtain the highest reward from the Internet without the burden of technology-related effort. (d) All Internet-mediated communication can be arranged on a scale of contextualization ranging from those discourses with a high level of contextual support (videoconferencing, online telephone conversations, web cam-mediated chat rooms...) at one end of the scale and plain textbased communication (plain text-based chat rooms, Internet forums, email...) at the other end. In any case, there are always two informational gaps: one between what is coded (i.e., typed, spoken) and what is intended and one between what is coded and what is interpreted, regardless of the level of contextual support. In other words, what is coded and what is intended and interpreted are related in terms of resemblance, rather than identity. Needless to say, this filling of the gaps is made easier if the channel of communication is capable of providing a lot of contextual support, especially in terms of the visual and vocal nonverbal information typically communicated in parallel to verbal utterances in situations of physical co-presence. In sum, cyberpragmatics claims that Relevance Theory can provide a valid explanation to all human communication, including the human inferential activity on the Net. So far, this activity has been studied within the user-to-user communicative framework. In this scenario, the task of the addresser user, just like any addresser’s, is to design the utterance (or nonverbal behaviour, if allowed by the system) in such a way that the interpretation intended is finally selected. The addressee user’s task, just like any addressee’s, is to build up an interpretive hypothesis (that is, “an interpretation”) about the addresser user’s intentions when he/she produces an utterance (or nonverbal behaviour) on the Net. This chapter aims to expand this communicative scenario to cover all human activity on the Internet, including the interaction that takes place between computer systems and users, to the extent that the underlying claim of this chapter is that all human inferential activity on the Net can be accounted for by cognitive pragmatics and, specifically, by Relevance Theory. Four main areas of interest will be commented upon in the next headings: (a) relevance sought by the system for the user; (b) relevance sought by the user in the system; (c) relevance sought by the user in another user’s utterance or intentional nonverbal behaviour; and (d) relevance sought by the user in a group of users.

92

Chapter Four

4. Areas or human activity on the Net. A relevance-theoretic account 4.1 System to user Relevance Theory (Sperber and Wilson 1995) claims that inputs for human processing are relevant depending on the fulfilment of two conditions: (a)

(b)

Other things being equal, the greater the positive cognitive effects achieved by processing an input, the greater the relevance of the input to the individual at that time. Other things being equal, the greater the processing effort expended, the lower the relevance of the input to the individual at that time.

The computer systems that control the interfaces for human activity on the Net learn from this activity, store a personal profile of the user’s likes, dislikes, preferences, etc. and automatically propose specific sources of information when they are potentially relevant to the user. The default criterion is that a user who has accessed a certain kind of website, bought certain items on the Net or surfed through certain web pages will quite likely find it relevant to be offered information of a similar quality. Although this is not human-to-human communication, and hence not a proper object of study for pragmatics, this proposal of sources of interest by the computer system resembles normal relevance-seeking coding/inference of information between people. Indeed, the system calls the user’s attention and alerts him/her that there is certain information that is worth attending to (the system’s algorithm automatically computes that it is likely to be worth attending to regardless of the effort invested). This calling for attention plays a similar role to the communicative intention underlying the communicative principle of relevance in interactions between humans: “Every ostensive stimulus conveys a presumption of its own optimal relevance.” In this case it is the system that plays the role of an individual calling the addressee’s attention. Relevance-oriented messages automatically selected by the system for the user include: (1) Emails sent by Facebook informing users that a photo has been commented upon, that someone likes one’s post, that someone left a message for the user, that a user wants the user to add him/her as a friend. (2) RSS feeds. These are links to pages in other newspapers where the same event is addressed, links to multimedia content, etc. Specifically,

Research on Human Activity on the Internet

93

RSS feeds are an interesting way of accessing information that inevitably alters the final relevance of the processing of the news article. Instead of clicking on potentially relevant links, it is the site that feeds the user with pre-established topics of interest. The RSS feed is re-written automatically whenever some updating takes place in the content of a website. In this way, the RSS file makes it possible to know whether the website has added new content or texts, but without having the need to go to the actual website unless one wants to read the “extended version” of the potentially interesting content. (3) Pop-up Messenger or email windows that appear at the bottom of the screen and warn users of an attempt at communication by another user. A similar warning is provided by the new “notification center” on devices such as the iPad. In all of these cases, the system assumes full responsibility for what is bound to be relevant to the user, to the extent that the system even stops users from accessing certain information if the computer algorithm decides that this information should not be relevant enough for them to process. In other words, the algorithm decides which information is bound to have an adequate balance of cognitive effects and mental effort and hides potentially non-relevant information. Take Facebook, for instance. It automatically filters information in three different ways (Pariser 2011a, 37-38): (1) By affinity. The more you contact somebody and the more someone visits your profile and engages in conversations with you, the more likely that Facebook will show updates from that user. (2) By the relative value of content. For example, updates on user status, e.g. on the user being no longer married etc., are more valuable and the computer system emphasizes them, but other types of content are also underlined by the system if it detects the user’s tendency to see that kind of content. (3) By time. Obviously, the most recently published entries have prominence over the oldest. Other automatic filtering of information by the system includes online bookshops’ personalized suggestions depending on the user’s previous purchasing activity on the site. And of course, we can also include here the output results by search engines such as Google. Why is Google so popular? Because it uses algorithms to (apparently) yield the most relevant results. Google learns from users and yields personalized results for each user, to the extent that specific users with specific search histories will get radically different output results. Singer (2011) exemplifies this web personalisation like this: If you type “bank” into Google, the search engine recognizes your general location, sending results like “Bank of America” to users in the United

94

Chapter Four States or “Bank of Canada” to those north of the border. If you choose to share more data, by logging into Gmail and enabling a function called Web history, Google records the sites you visit and the links you click. Now if you search for “apple,” it learns and remembers whether you are looking for an iPad or a Cox’s Orange Pippin.

Pariser (2011a, 2011b) is very critical of the “relevance criteria” that sites use to narrow output results, to offer ads by placing them on webs of related content and to offer personalised suggestions. For him, personalization prevents citizens from engaging with multiple viewpoints, and the Internet limits such engagement when it offers only information that reflects users’ already established points of view and preferences. If the computer system picks up and stores the user’s likes and preferences upon previous activity on the Net, then it is easier to bombard the user with personalised ads and suggested options for purchasing. But the main problem, according to Pariser, is that we no longer choose what discourses we interpret; rather, we are offered previously filtered discourses according to an automatic criterion of relevance. We click on what Google proposes that we should click on, a mass of personalised options for each individual user. We think that output results on a query are objective but surprisingly relevant to us, when in reality they are based upon a non-stop monitoring of our activity on the Net. Pariser underlines the fact that personalisation is gratifying, an environment filled with discourses that are relevant to us, a mirror for our likes and preferences. However, at the same time personalisation generates a vicious circle in which Google provides us with what we are expecting and we expect what Google yields as results, preventing the user’s processing of other kinds of information, the serendipity of finding surprising outcomes by chance. Creativity is filled with unpredicted processing of information that surprisingly combines with background knowledge to yield relevant conclusions. Search engines and their algorithms prevent that. Pariser even claims that automatic personalisation of information alters the natural way of interacting with the world, making us over-confident in our ideas and points of view. To navigate through an environment filled with discourses that we like is pleasant. To question pre-conceived ideas and to contrast them with new discourses is often frustrating. As Vaidhyanathan (2011) stresses, by definition learning entails facing the unknown, with what one has not thought about before. The kind of filter that Google interposes between the users and the output results isolates the users from this kind of encounter with the unknown.

Research on Human Activity on the Internet

95

4.2 User to system Apart from search engines such as Google, we often search for information on the Net, read online documents, choose reading paths by clicking on the links that might lead to the most relevant information. We also access web pages expecting to get relevant information with as little effort as possible, that is, we expect a high level of web page usability (effort-relieving web page design). Again, for these tasks we cannot avoid following our rooted search for relevance. Our tendency to maximise relevance is also at work when we combine information from visual, verbal and multimodal sources on the page to get relevant outcomes, when we engage in multi-tasking trying to get to grips with the increasing demands of Internet use, and when we read texts online with the pressure of ever-pushing competing sources of interest on the Net. All this human activity has a clear explanation in relevance-theoretic terms. Information in general has to be compensated for with positive cognitive effects. In Yus (2011a, 69) a rewriting of the conditions of relevance was proposed for the specific quality of information on web pages, specifically the condition related to processing effort, which was divided into two possible sub-conditions, one quantitative (number of clicks, condition b1) and one qualitative (inter-link coherence, condition b2): Condition a. The information in link-mediated content of web pages is relevant to an individual to the extent that the cognitive effects achieved when it is optimally processed are large. Condition b1. This information is relevant to the individual to the extent that the number of clicks that the user has to make in order to obtain these effects is small. Condition b2. This information is relevant to the individual to the extent that the level of coherence obtained from linking different texts is optimal despite the non-linear arrangement of the linkmediated texts. Finally, Carr’s (2008, 2010) claims on the harmful effect of the Internet on human minds can also be reinterpreted in relevance-theoretic terms. He claims that our brains are now addicted to the barrage of microstimuli and we are eager to receive inputs all the time. Our brains are now restless and impatient, in need of constant cognitive gratification. One of the reasons for this addiction and mental restlessness is the increasing amount of micro-messages, flashes of immediate relevance to the user but

96

Chapter Four

which demand little processing effort in exchange (e.g. SMS and WhatsApp messages, Facebook posts and Messenger utterances). Therefore, we have ended up addicted to a non-stop processing of micromessages whose balance of effects and effort is turning our minds into lazy cognitive mechanisms, to the extent that many intense Internet users are no longer able to devote the effort required to process a long text such as a novel, since its relevance is not immediate but deferred. We are too impatient. Our minds long for immediate relevant outcomes from processing effort-relieving micro-discourses, and also long for immediate gratification from non-stop flashes of information.

4.3 User to user Nowadays we engage in many Internet-mediated interactions, either synchronous conversations (chat rooms, instant messaging, 3D-avatarmediated interactions, Skype…) or asynchronous interactions (posts on Facebook, e-mail, Twitter, blog posts…). This area is, no doubt, where Relevance Theory can be more directly applied to the Internet, especially through the communicative principle of relevance, and it is also the main research area within cyberpragmatics. Indeed, we aim at maximising relevance in daily conversations with other users, in the social uses of the net (web 2.0) and in identity shaping through daily communication with other users. As claimed above, communication on the Net follows the same inferential patterns as face-to-face interactions. Senders code their messages predicting the interlocutor’s accessibility to contextual information so as to be able to select the intended interpretation and addressees metarepresent the sender’s informative intention, attitudes and emotions. Crucially, the quality of the text typed (or spoken) in the visualverbal, oral-written and synchronous-asynchronous axes, and the limitation of contextual information on the Net influences or affects these inferential activities of metarepresentation and prediction. That is, the quality of the text or utterance may influence this communicative activity, affecting balances of effects and effort or making it hard to metarepresent the sender’s intentions, attitudes and emotions, as happens for example in plain-text chat room interactions (see Yus 2005a). Much of cyberpragmatic research focuses on analysing how Internet users compensate for the cues-filtered quality of typed texts so that they end up being optimally relevant and communicate their thoughts, feelings and emotions in an appropriate way. An example is what I call “text deformation,” normal orthography altered by users with emoticons,

Research on Human Activity on the Internet

97

repetition of letters and punctuation marks, written simulation of intonational patterns, etc. In these cases, to the traditional gaps of interpretive resemblance between what the user intends to communicate and what he/she codes, and between what is read by the addressee user and what is interpreted by him/her, new gaps are created: between what the addresser user could have said in a face-to-face situation and what the user actually types, and between what the addressee user reads and what the user could have listened to in a face-to-face situation. How much these new gaps actually alter eventual relevance depends on how familiar the user is with this kind of textual alteration. Actually, familiarity with the special ways in which utterances are constructed on the Net is such an important factor that it should be incorporated to the general formula of relevance and its conditions of cognitive effects and mental effort (the same applies to familiarity with the interfaces for human interaction). And other factors also influence communication on the Net in a much more radical way than in normal face-to-face interactions. Therefore, I think we can fine-grain the conditions for relevance with aspects that on the Net turn out to alter, affect of influence eventual relevance. This fine-graining would explain why the Internet generates balances of effects and effort that defy normal relevance assessment in physical scenarios. On the Net, there are several alternative sources of user satisfaction that may add to (or overlap with) the actual information exchanged on the Net (the “cognitive rewards” suggested in Yus 2011b). Among them, the entertainment of feelings and emotions, the fostering of empathy and phatic connotations, the feelings of group, network or community membership and the reward arising from being connected and interacting with others. These “rewards” are also found in physical scenarios, but on the Net they acquire such an important role that they end up altering the eventual balance of cognitive effects and mental effort. Similarly, there are potential sources of user (dis)satisfaction on the Internet that may reduce or increase the eventual relevance of the information being processed (these were called “environmental constraints” in Yus 2001b). I have already mentioned the importance of familiarity with the rules of the medium as one potential source of added (dis)satisfaction related to higher/lesser levels of mental effort and to a certain (in)ability to reach the expected information and its intended interpretation. Other constraints include: (a) The type of search or interaction. In relevance-theoretic terms, it is not the same to engage in aimless surfing (browsing) as to try to find some specific information (focussed search). Similarly, an interaction with someone on the Net when there is an important topic to address demands

98

Chapter Four

different amounts of effort and produces different effects than casual conversations just to kill time. (b) The quality of the medium in the oral/written, visual/verbal and synchronous/asynchronous scales. The extent to which the medium exhibits a cues-filtered quality affects the effort to pick up the intended interpretation and derive positive cognitive effects. At the beginning of Internet-mediated communication, most communication was text-based and hence it lacked the contextual support of nonverbal behaviour. Nowadays, even though text still plays an essential role in virtual communication, the increased bandwidth and developments in software has made it possible to interact with the visual support of the interlocutors, normally decreasing processing effort. (c) The quality of the interface. As I mentioned above, in Yus (2011a) a rewriting of the conditions for relevance was proposed in which the effort-related condition was related to both a quantitative notion of relevance on the web (number of clicks needed to reach the intended information) and to a qualitative one (inter-link coherence, that is, the extent to which the chunk of text just accessed after clicking on a link relates coherently to the previously processed chunk of text). Therefore, web usability (whether or not the design of the web page allows – or not – for easy access to the information that is interpreted) is an important constraint in the eventual balance of cognitive effects and mental effort. A poorly designed page, with links that do not clearly lead to the expected information, will increase mental effort gratuitously. Hence, relevance equations of Internet communication must be complemented with these cognitive rewards and environmental constraints (Yus 2011b). The general relevance-related equation of Internet-mediated communication would be as follows: the cognitive effects obtained from information processing plus cognitive rewards plus positive environmental constraints [should exceed…] the effort demanded for the derivation of these cognitive effects plus negative environmental constraints

Research on Human Activity on the Internet

99

Two examples will illustrate this equation. Firstly, the situation in which a user engages in multiple conversations in parallel Messenger windows. In theory, the mental effort required to pay attention to multiple interactions in multiple windows should increase compared to a normal one-to-one conversation, since managing multiple windows is in itself quite a challenge for human cognition. If the user is a non-expert and has no familiarity with the interface, supplementary effort will be added in order to manage the conversation adequately. However, this effort can be compensated for with an offset of cognitive rewards, for example with what Thompson (2008) called “ambient awareness,” the feeling of being physically near the other users and connected to them with the aid of this multiplicity of synchronous conversations, or also feelings of group or community membership. Secondly, the situation in which a user enters the 3D environment Second Life and interacts with other avatars there. As commented upon in Yus (2011b, 76-77), interactions in virtual worlds such as Second Life using a 3D avatar are interesting for pragmatics because they offer the possibility of mixing verbal and nonverbal communication (of both intentional and exuded quality) in the interaction. The users can find many “cognitive rewards” from these 3D interactions, for instance from the satisfaction of being able to sustain text-based and nonverbal communication simultaneously, or from the attention-drawing potential of the avatar designed by the user. However, these interactions also pose a challenge for effective interactions through a number of “environmental constraints,” among them the different options for verbal communication (in Second Life there are three options: chat room, messenger and real voice) that intertwine with the avatar’s intentional nonverbal behaviour (as intended by the user) and the avatar’s exuded nonverbal behaviour (as generated by the software beyond the user’s control). Besides, the user has to type his/her messages while paying attention, at the same time, at how the avatar is behaving nonverbally (walk, look, grab things...). Hence, for this relevance equation to be effective, the user needs to find cognitive rewards that make up for the environmental constraints of the 3D scenario that increase mental effort. For example, the user might find a reward when succeeding in engaging in profitable verbal-nonverbal interactions, plus feelings of empathy, phatic connotations, the possibility of having multiple identities with parallel discursive manifestations, etc.

100

Chapter Four

4.4 User to group of users Finally, there are potential sources of relevance not in the actual content of the interactions or the discourse typed on the keyboard (or spoken in voice-enabling software), but in the offset of group-connoted effects that can be obtained from these interactions, even if the information is, in itself, utterly irrelevant. In general, the most popular sites for social interaction are intended to generate relevance by strengthening the individual’s identity shaping and presence in the group. This is also at work in physical settings, but on the Net it is a central source of relevance. For example, profiles in social networking sites such as Facebook clearly exhibit the user’s identities, in the varieties of self-identity, interactive identity, and social identity (a three-fold proposal in Yus 2012b). Undoubtedly, it is inside the minds of the members of the community where we have to look for the features of group-connoted relevance, for example feelings of community, both in online and offline scenarios. Those features such as norms, commitment, values, interests which make community different from a group of people simply brought together. As claimed in Yus (2005b), if we were able to look inside the minds of the people supposedly sharing a community, we would see differing mental representations which, through steady and sustained Internet-mediated interactions, end up intersecting one another, resembling one another, forming a stable but continuously re-negotiated set of communal representations. This mental storage of social information is especially relevant when dealing with computer-based communities which lack an embracing, as it were, physical location.

5. Concluding remarks Relevance Theory can provide valid explanations to all Internet communication involving, one way or another, a human being interacting with another individual or a system on the Net. This is because the inherent human search for relevance is found in all aspects of human cognition, including human activities on the Internet. We do not know which interfaces, programs, sites or virtual channels will be popular on the Internet in the future. But whichever they are, Relevance Theory will be able to provide a scientific explanation to human interactions with, within or through them.

Research on Human Activity on the Internet

101

References Carr, Nicholas. 2008. Is Google making us stupid? The Atlantic 07/082008. —. 2010. The shallows: What the Internet is doing to our brains. New York: Norton & Company. Pariser, Eli. 2011a. The filter bubble: What the Internet is hiding from you. London: Viking. —. 2011b. When the Internet thinks it knows you. The New York Times, May 22, 2011. Singer, Natasha. 2011. The trouble with the echo chamber online. The New York Times, May 28, 2011, Technology. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and cognition. Oxford: Blackwell Thompson, Clive. 2008. Brave new world of digital intimacy. The New York Times, September 5, 2008, Magazine. Turkle, Sherry. 2011. Alone together: Why we expect more from technology and less from each other. New York: Basic Books. Vaidhyanathan, Siva. 2011. The googlization of everything (and why we should worry). Berkeley: University of California Press. Wilson, Deirdre and Dan Sperber. 2002a. Relevance Theory. UCL Working Papers in Linguistics 14: 249-290. —. 2002b. Relevance Theory: A tutorial. In Proceedings of the Third Tokyo Conference on Psycholinguistics, edited by Y. Otsu. Tokyo: Hituzi Shobo, 45-70. Accessed online at

Yus, Francisco. 2001. Ciberpragmática. El uso del lenguaje en Internet. Barcelona: Ariel. —. 2002. Discourse and identity. In International encyclopedia of the social & behavioral sciences, edited by N. J. Smelser and P. B. Baltes. London: Pergamon Press, 3728-3732. —. 2005a. Attitudes and emotions through written text: The case of textual deformation in Internet chat rooms. Pragmalingüística 13: 147-174. —. 2005b. The linguistic-cognitive essence of virtual community. Ibérica 9: 79-102. —. 2007. Virtualidades reales. Nuevas formas de comunidad en la era de Internet. Alicante: University of Alicante, Servicio de Publicaciones. —. 2010. Ciberpragmática 2.0: Nuevos usos del lenguaje en Internet. Barcelona: Ariel.

102

Chapter Four

—. 2011a. Cyberpragmatics: Internet-mediated communication in context. Amsterdam: John Benjamins. —. 2011b. Relevance equations of effective Internet communication. In Interdisciplinarity and languages: Current issues in research, teaching, professional applications and ICT, edited by B. Pennock and T. Suau. Berlin: Peter Lang, 65-86. —. 2012a. Cyberpragmatics. In The encyclopedia of applied linguistics, edited by C. A. Chapelle. London: Wiley-Blackwell. —. 2012b. Identidades físicas e identidades en línea: Atracciones, desencuentros y solapamientos. Paper presented at Identitats Perverses, Identitats en Conflicte, Lléida (Spain), November. —. 2013. Una aproximación discursiva a las identidades en línea. Paper presented at Language and Identity in the Spanish-Speaking World (6th International Conference of Hispanic Linguistics and 4th Biennial Meeting of the International Association for the Study of Spanish in Society), London, July.

CHAPTER FIVE THE GURU EFFECT IN BLIND PEOPLE’S COMPREHENSION JOLANTA SAK-WERNICKA, JOHN PAUL II CATHOLIC UNIVERSITY OF LUBLIN, POLAND

1. Introduction When we meet a knowledgeable and gifted individual, such as an upstart professor, renowned writer or experienced practitioner, we admire that person’s wisdom and intellectual prowess. We know that the knowledge and abilities of this person make her very special and superior, therefore we do not dare to question the validity of her statements and we are eager to accept whatever she says as true even if it seem abstruse. Although it taxes our intelligence and knowledge, we judge as profound the statements we cannot comprehend, believing that our intellectual limitations prevent us from fully appreciating the words uttered by this ‘intellectual guru’ (Sperber 2010, 583). This is very unlikely to happen if the statement is communicated by someone with no reputation for deep thought, in which case an obscure utterance is quickly rejected. This mechanism, described by Dan Sperber as ‘the guru effect’ (1994, 2010), can be observed to operate in a great variety of social contexts, including education, business and healthcare. However, as a relatively new and poorly investigated concept, the guru effect continually raises many questions and controversies. First of all, it is not clear what exactly makes a particular person an authority. In particular, which aspects of personality or social factors lead others to admire and follow someone as an authority? Is it universal acclaim or rather admiration of individuals which decides that some people are seen as intellectually gifted, if a bit eccentric, while others are viewed as loonies? Secondly, are we equally eager to accept statements which do not concern or fall beyond the area of a guru’s expertise, and what happens if we know that such a person makes an

104

Chapter Five

elementary mistake in his or her utterance? Does it change our attitude towards this person? Finally, is this psycholinguistic phenomenon seen as bringing about positive or negative consequences in communication? Although there are hardly any studies on this subject (see Eriksson 2012), it has been suggested that there is plenty of room for manipulation and unethical practices in this area if such a source turns out to be dishonest. Intriguing as it is, the guru effect remains a little explored concept and its understanding still remains more based on theoretical considerations than on well-grounded findings. The aim of this article is to discuss the guru effect in blind people’s comprehension. In the realm of cognitive science and pragmatics, the guru effect has been mainly discussed in relation to the interpretation strategies of sighted people (Sperber 2010; Mioduszewska 2006, 2008; KosiĔska 2008; Jezierski 2011), and no special reference to people with impairments or disorders has so far been made. The article is intended to demonstrate that blind people, due to their perceptual limitations, take for granted the authority of sighted people (as those who have unlimited access to visual information). They believe that what sighted people communicate is true and relevant, even if not entirely comprehensible. This is because they do not have adequate perceptual evidence and knowledge based on previous visual experience to hold the opposite view. Blind individuals are not only ready to accept sighted people’s explanations which they do not entirely comprehend, but they are also eager to invest a great amount of cognitive effort into the process of interpretation. To demonstrate this we will try to analyse interpretation strategies employed by the blind, on the basis of empirical data. However, before we proceed to analysing this phenomenon in the context of blind people, a brief introduction to the guru effect will be presented.

2. The mechanism of the guru effect ‘The guru effect’ has been used in a variety of different contexts. The way we understand this term in the article was first described by Dan Sperber. In his seminal article “The Guru Effect” (2010) he presents the reasons why obscurity of expressions is not always considered to be a flaw. More than that, he argues convincingly that this obscurity is often taken as evidence of intellectual superiority of the people who produce such expressions.

The Guru Effect in Blind People’s Comprehension

105

2.1 Authority as social relationship The guru effect leads people to invest a great deal of cognitive effort into processing complex and obscure utterances. Even if individuals fail to arrive at logical interpretations of the utterances, they will still remain convinced of the profoundness and relevance of these incomprehensible utterances. This is because the statements are uttered by a reliable and competent source, also called an ‘authority’. In the field of psychology, authority is defined as the power to affect thoughts, attitudes and behaviour of others. Usually we associate it with the role of a master, leader or person having a considerable knowledge and/or practical abilities in a given discipline, someone we respect and wish to follow. An alternative definition is proposed by Sperber (ibid., 591) who explains, Authority is social relationship that involves at least two individuals, and typically many more. Authority in a group goes with reputation. The reputation of a person is the more or less consensual view of her competence and reliability that spreads through repeated acts of communication across a social group.

It is easy to notice that authority is not actually a position that someone holds. It is rather a connection which regulates social contacts between people where one of them has the reputation of being reliable and competent because of his previous achievements. This ‘social relationship’ is based on the subjective perception of group members who are convinced of the superiority of the source. Therefore, what the person communicates is treated as relevant and worthy of greater cognitive effort – even more so if it appears to be mysterious, complex or obscure. Although the guru effect might take place between two individuals, it usually involves more than two people. The number of individuals further strengthens the source’s reputation and is likely to attract other participants. This is not only because people feel an individual needs to be in touch with those from whom they can acquire knowledge and experience. It is also because of the social need of belonging to a group, which secures the comfort of being perceived as intelligent. By accepting an obscure statement we prove our own competence and become part of the authority. By rejecting it, we point out our own intellectual inadequacy. As noted by Sperber (2010), the reputation of the source is often driven by what in psychology is known as a ‘confirmation bias’ (Watson 1960). This makes others look for evidence to support benevolence and competence of the source, and perhaps to overlook the facts which might

106

Chapter Five

deny his or her reliability. Therefore, when we trust someone, each time this person was right strengthens this feeling of trust and confidence in his or her trustworthiness. Bearing all this in mind, the question is who meets the criteria of being an authority. Can it be anybody who knows more and better than others? Sperber (2010) and Mioduszewska (2006) in their analyses of the guru effect refer to quotes of great thinkers, showing that obscure statements made by such individuals are often taken as evidence of their genius. In this article we seek to demonstrate that discussion of the guru effect should not be limited to the obscure prose of philosophers, but should be continued by considering other and perhaps more natural types of communication. For instance, the words of a doctor commenting on a medical condition in (1) produce very similar effect: (1)

Septo-optic dysplasia, also known as de Morsier syndrome, is a congenital malformation syndrome made manifest by hypoplasia of the optic nerve and absence of the septum pellucidum.

For an average person, this comment appears very complex and difficult to comprehend by referring to the knowledge which people other than students of medicine or optometrists do not normally have. Therefore, people other than experts need to believe that the doctor saying (1) above is competent and his words are relevant and true. Similarly, a blind person who has no access to visual information and whose knowledge of the world is somewhat limited has to rely on the knowledge and competence of sighted people.

2.2 The guru effect and comprehension strategies Since early childhood we are confronted with intellectually challenging statements which are difficult to comprehend. As children, we have very limited knowledge and underdeveloped interpretative abilities which prevent us from understanding explanations of the properties of the physical world and rules of nature. Then we learn to accept the unknown and not to question the competence of parents, teachers and older individuals who are said to know better. At this stage we uncritically assume that the speakers are all competent and their utterances always true and relevant. With time, however, this naïve optimism in comprehension turns into cautious optimism and sophisticated understanding. We begin to take into consideration the fact that a speaker might be benevolent, but not

The Guru Effect in Blind People’s Comprehension

107

competent, or that he or she only may seem to be benevolent or competent (Mioduszewska 2006). Therefore, we are ready to reject the statement which does not meet our expectations of relevance (Wilson and Sperber 2004). In other words, the interpretation process might be abandoned, if the expectation of relevance is not satisfied. However, if we trust in the benevolence and competence of the speaker (authority), we will put extra cognitive effort into understanding the statement. In some cases we will resort to its partial interpretation or accept our failure, taking for granted the relevance of this statement. The reason why we accept the statement as true and relevant, even though we do not understand it, is that we are convinced that an authoritative source could not find a simpler way of conveying his or her thoughts. The sole presence of the authoritative source is a sufficient proof of the validity of his or her statement. Therefore it does not require further examination and double-checking.

3. The guru effect and blindness Having outlined how the guru effect works, in this section we will show that this mechanism can be observed to a much greater extent in the comprehension of blind people, with sighted people assuming the role of authority. In section 2.1 we argued that an authority is well-known for being a competent and reliable person among a group of people. Sighted people have unlimited access to visual information which allows them to obtain in-depth knowledge about the world. Deprived of such information and unable to entirely compensate for its absence through the other senses, blind people have to rely on the competence and reliability of sighted individuals in explaining visual concepts and eliminating gaps in their knowledge and misunderstandings. Therefore the perceptual abilities of sighted people are treated by the blind as a considerable advantage of the former. The following words of a young congenitally blind girl, explaining what ‘to see’ means to her, provides confirmation for this assumption: It is like telling the future – because you know now that there will be a tree, and I will know later, when I come up to it and touch it. (Marek 2000)

Sighted people’s ability to see, treated by some children as the equivalent of clairvoyance, allows them to provide answers to many intriguing questions they often ask themselves, such as the following (2ac), taken from Blagden and Everett (1992) and Marek (ibid.):

Chapter Five

108

(2)

a. Does a stone look the way it feels? b. What colour is the wind? c. How can you see a big mountain through a small window?

Many of these questions involve concepts which are very difficult to understand if no access to visual information is granted (for the discussion on the questions, see also Sak-Wernicka 2012). Therefore the explanations offered by sighted people might sometimes turn out to be incomprehensible and obscure. However, the children accept and remember these explanations without fully understanding them. The children believe that their visual impairments hinder proper understanding of certain concepts which sighted people are perfectly able to comprehend and fully experience. This guarantees that what they say is both true and relevant. These observations have been also confirmed by two studies performed on two groups of blind adults which are separately described and discussed below (Sak-Wernicka 2011). Study 1 A group of 19 congenitally and early blind individuals (8 women and 11 men) aged between 19 and 67 (M = 38.32, SD = 19.00) participated in the experiment. The subjects were recruited from the Polish Association of the Blind, Blind Co-operative Society and Occupational Therapy Workshops. 20 sighted participants (10 men and 10 women) aged between 19 and 67 (M = 38.55, SD = 15.60) were recruited from the John Paul II Catholic University of Lublin (Poland) and also participated in the experiment. The participants were students and administrative/technical workers at the university. In order to investigate the guru effect in blind individuals’ comprehension, the blind participants were asked to interpret 12 short dialogues provided as a recording. All dialogues were based on real life conversations chosen to refer to visual experience or to include visual concepts. The participants were tested during individual sessions and asked to answer questions examining their understanding of the dialogues, such as the following presented in (3): (3)

Tom: What did you do on Saturday? Ben: I went to the zoo with my kids. Tom: Did you enjoy it? Ben: The twins didn’t give a damn about animals and were playing caps. Alice wanted to see a baby ostrich, but it refused to come out of the enclosure. In the end all we saw was a cleaning lady.

The Guru Effect in Blind People’s Comprehension

109

After listening to the recording, the participants were asked to specify how many children the man had taken to the zoo. In order to provide the correct answer, it was necessary to infer that he must have taken three children (the twins and Alice). Otherwise, it seems that the girl was uninterested in watching the animals (hence the reason why she was playing caps), and wanted to see the baby ostrich. As we can see, instead of analysing individual expressions in which the context was not specified (and therefore open to numerous ides, associations and thought), the participants were presented with dialogues embedded in a well-specified context. All the participants’ responses were recorded, analysed and compared to those of the sighted participants. The latter were provided with the same dialogues and questions, but the dialogues were presented as film clips. In our analysis, we measured the number of times each participant was unable to provide a correct answer and needed the experimenter’s assistance. For example, in the above-mention case, many blind participants asked if the word ‘twins’ always describes two people (or perhaps more). Many of them responded tentatively that the man took three children, but only if there can be only two twins. In the same dialogue, an additional complication for the blind was that they did not know that ‘playing caps’ is a visually engaing game which involves throwing bottle caps around in a certain way. As a result of this, they might have been unaware that it was unlikely for someone who was playing caps to be also interested in “seeking a baby ostrich”. That inference appears to be a crucial piece of information for reaching the conclusion that there were three children. Similar tendencies were observed in the blind participants’ responses to dialogue (4) used in the same study: (4)

Wife: We need a new bedspread… and carpet… Husband: If a carpet, then chairs, if chairs, then a mirror, wardrobe and then our kids will hate us. Wife: Why? Husband: Because we will get a divorce.

After listening to this dialogue, the participants were asked the question if the man would divorce his wife. They could hear that the people in the dialogue talked calmly, but they could not observe them hugging or kissing. Although the reason mentioned by the man (i.e. choosing new furniture to the house) should have appeared too trivial a reason for divorce, it was really difficult for the participants to recognise the man’s

110

Chapter Five

true intentions and the fact that he was only joking. Therefore a number of the participants were confused and made up stories of what had happened previously which might have had an effect on the man’s decision. They also claimed that it was hard to say how the situation would end because evidently something wrong was happening between the spouses, or they said that the man did not know what he wanted and that he was pretending that everything was fine but deep down he was considering a divorce. They also commented that it would depend on if the wife bought the new furniture or not and referred to personal experience saying that their friends or relatives had had a similar argument and they got divorced. At the same time they did not offer one clear answer but gave the experimenter a choice. This showed not only their problems with understanding but also their reliance on the knowledge of the experimenter. They assumed that she had to know what really happened because she could see the situation. Overall, only 55% of the blind subjects taking part in the study provided correct answers, as compared to 70% of correct responses in the group of sighted individuals. The remaining participants either provided answers other than the intended ones or no answer at all claiming that they had an insufficient amount of information to answer the above-mentioned question. In order to ascertain whether the two groups differed in how often they were unable to provide any answer and relied on the experimenter’s assistance, an independent samples Mann-Whitney U test was conducted. The analysis revealed significant differences between the blind and sighted participants (Mann-Whitney U= 95, p = .04). The individuals who were blind (Mdn = 6) found it more difficult to interpret the dialogues without the experimenter’s help than sighted participants (Mdn = 3). At the same time, the statistical analysis indicated no differences in how often the blind group (Mdn = 10) and the sighted group (Mdn = 6) provided incorrect answers (Mann-Whitney U = 174, p = .645). The only difference was that the sighted subjects gave one (incorrect) interpretation, while a great majority of blind individuals offered many possible interpretations in hope one of them would turn out to be right. By doing this, they wanted the experimenter to choose the correct one on her own. Because they had no access to visual information, which significantly reduced the number of possible interpretations, the blind participants were unable to reject some interpretative hypotheses and found the dialogues more ambiguous than the participants with normal vision. It is important to observe that, despite these difficulties, they put a lot of effort in the interpretation of the dialogues. This was because they were convinced of the relevance of the dialogues.

The Guru Effect in Blind People’s Comprehension

111

Study 2 In the second study a group of four congenitally blind students of the John Paul II Catholic University of Lublin (aged 19 to 25) were asked to solve a crossword puzzle consisting of 20 puns. During a group session the participants were encouraged to contribute, share their ideas and freely express their thoughts. The primary aim of this study was to observe the guru effect as social relationship involving more than two individuals. In other words, we wanted to examine whether the presence of other people in a group may give rise to the guru effect in the case of blind people. For this reason, in this study we concentrated on the blind participants’ comments, which in this case gave us more insight into the phenomenon described in this article than analysing how many puns individual subjects managed to solve. Out of all 20 puns, the group provided 5 correct answers. In the other cases, they offered their incorrect associations or provided no answer at all. Both such situations are discussed below. One of the clues used in this crossword, particularly problematic for blind participants, was the one presented in (5) below, to which they could not find an answer, despite numerous attempts. (5)

Clue: wąĪ z piórami ‘a snake with feathers’ Intended answer: boa

When the answer was finally revealed to them and the participants were asked to make connection between the clue and the answer, they could not find a good explanation. Although they knew that there is a snake called boa, surprisingly they did not realise that the same word can be used to refer to a scarf. However, they appreciated the experimenter’s explanations, saying that this was something they had not known previously. The example shows that blind participants were convinced of the relevance and profundity of this expression, otherwise they would not have put so much effort into looking for a logical explanation. They must have been also convinced that the examiner knew the answer to this clue, which confirmed them in believing that some logical explanation existed. They also knew that if they gave up or said that the clue was absurd, it could be taken as a sign of their intellectual inferiority. Apparently, the presence of one participant appreciating the profundity of the explanation (even if he could not entirely understand it), exerted an influence on the other participants, who also eagerly accepted it.

112

Chapter Five

These tendencies were observed in other challenging clues in this study, usually when they referred to visual concepts. The attempts of blind people to arrive at logical interpretations of the clues sometimes led to unexpected and quite amusing associations. A good illustration of such ideas was the answer offered by a blind participant to the clue in (6), which was ‘a Chinese person’. (6)

Clue: Īóáty egoista ‘a yellow egoist’ Intended answer: narcyz ‘a narcissus/narcisist’

Providing this answer, the participant failed to make a necessary connection between a yellow flower (narcissus) and a self-centred man (narcissist), which in the Polish language are represented by a single word (narcyz). The difficulties of blind people with understanding visual concepts probably contributed to the occurrence of this connotation. After detailed explanation offered by the experimenter, it turned out that the participants were perfectly aware of the fact that there is a flower called narcissus, and that we may call a self-centred man ‘a narcissist’. What they did not know was that this flower is yellow, and thus were unable to draw a connection between the two seemingly unconnected concepts in this way. It appears that even if what sighted people communicate seems incomprehensible to blind individuals, they tend to take it as true and relevant, knowing that it might be connected with information which remains inaccessible to them. They are also ready to put a great deal of effort into the comprehension of sighted people’s utterances, offering many different interpretations and seeking fragmentary or tentative interpretative hypotheses. The studies presented in this article demonstrate some of the strategies blind people use in their interpretations. The analyses showed that in their comprehension they do not uncritically accept other people’s statements. If they detect any dissonance between their previous assumptions and the statement, they look for possible explanations (utilising their general knowledge and personal experience). They engage sighted people in discussions in order to obtain more data and let the sighted people help them choose the correct interpretations. This evidences their intelligence, logical thinking and sophistication. At the same time they are very careful about making assumptions, because they are aware that what they know might be insufficient, partially understood or misunderstood.

The Guru Effect in Blind People’s Comprehension

113

4. Conclusions The guru effect has been only scarcely discussed in the literature and hardly any experimental studies investigating this phenomenon have so far been performed. As a result, there are more questions than answers concerning what the guru effect really involves. Because it can be analysed as a linguistic, psychological and social phenomenon, it appears very complex and difficult to explain. The aim of this article was to continue the discussion on the subject started by Dan Sperber (2010) and to encourage researchers of various disciplines to further contribute to exploring this intriguing concept. By presenting some experimental studies, we intended to show that the guru effect can be observed in a great variety of social contexts, one of which is communication between blind and sighted people. In this article we argued that the role of authority is not always attributed to great philosophers, scientists and mentors. This ‘social relationship’ may involve any person who is respected, admired or appreciated by others for his or her knowledge or experience. The cases presented show that blind people readily learn from the experience and knowledge of sighted people, who often help them bridge gaps in their knowledge and overcome their limitations. Verbal exchanges with sighted people are, for people who are blind, opportunities to discover new things and verify previous assumptions. In this analysis we also tried to show that the guru effect leads blind people to accept sighted people’s explanations and statements which they might not understand. This is because visually impaired idnividuals are aware of the fact that sighted people, as fully equipped participants of communication, might perceive certain concepts differently and perhaps more accurately. Vision may also provide them with information that blind individuals are not normally aware of. This, however, does not indicate their naïve optimism in comprehension. We would rather say that blind people are willing to trust knowledgeable sighted individuals (educators, therapists, advisors, doctors) who work with them. We can expect that they will be much more cautious talking to people they do not know and who might not be honest or competent. The presented empirical material also demonstrates that obscurity of expression is very subjective; there are few statements, including the quotes of great thinkers, which would be equally abstruse for all readers or hearers. This does not depend only on the intellectual abilities of individuals, their social status or educational attainment. As in the case of blind people, an insufficient amount of information or misunderstanding of

114

Chapter Five

certain concepts may affect the interpretation of utterances in some people. This does not have to occur in others, however. The analyses of blind participants’ responses presented above also shows that the guru effect might involve an intellectual and social benefit. This is in accordance to Sperber’s predictions. The second experiment shows that participants engaged in a collective process of interpretation are likely to judge an explanation true, relevant and profound, if at least one of the subjects present can understand it or is able to appreciate it. By accepting this explanation, they may be underscoring their own intelligence and appreciation of its profoundness. To sum up, due to their visual limitations, blind individuals might not be able to understand certain concepts, if these concepts are difficult to explore. Aware that sighted people have access to information they cannot access, the blind are willing to accept the explanations and comments of the sighted, even if these sometimes seem obscure or ambiguous. From this perspective, sighted people might be treated as an authority for the blind. In the same way, visually impaired people might share their knowledge and experience about blindness and respond to the questions which many sighted people tend to ask, resulting from a lack of information about the everyday world and lives of blind individuals, such as (6a-d): (6)

a. Do blind people dream in colours? b. Can blind people see only darkness? c. Can all blind people read Braille? d. Can blind people live alone?

References Blagden, Sue and John Everett. 1992. What colour is the wind? Bristol: NSEAD. Bordens, Kenneth and Irwin Horowitz. 2002. Social psychology. Mahwah: Lawrence Erlbaum Associates. Eriksson, Kimmo. 2012. The nonsense math effect. Judgement and Decision Making 7, no. 6: 746-749. Hall, Amanda. 1982. Teaching specific concepts to visually handicapped students. In A teacher’s guide to the special education needs of blind and visually handicapped children, edited by S. Mangold. New York: AFB. Jezierski, Grzegorz. 2011. Guru effect and the translator. Paper presented at Relevance Round Table Meeting 3, Warsaw, Poland.

The Guru Effect in Blind People’s Comprehension

115

Kantowitz, Barry, Harry Roediger and David Elmes. 2005. Experimental psychology. Belmont: Wadsworth. KosiĔska, Katarzyna. 2008. Conversational humour as a power game. In Relevant worlds: Current perspectives on language, translation and Relevance Theory, edited by E. Waáaszewska, M. Kisielewska-Krysiuk, A. Korzeniowska and M. Grzegorzewska. Newcastle: Cambridge Scholars Publishing. Landau, Barbara and Lila R. Gleitman. 1985. Language and experience: Evidence from the blind. London: Harvard University Press. Marek, Bogusáaw. 2000 Does a stone look the way it feels? Paper presented at the Fifth European ICEVI Conference, Cracow. Mioduszewska, Ewa. 2006 The guru effect and sophisticated understanding. Kwartalnik Pedagogiczny 4: 181-191. —. 2008 On the relevance of non-communicative stimuli: the case of unintentional obscurity. In Relevance Round Table I, edited by E. Mioduszewska and A. Piskorska. Warsaw: WUP. Ohlson, Kjell. 1986. Compensation as skill. In Communication and handicap, edited by E. Hjelmquist and L.-G. Nilsson. Amsterdam: Elsevier Science Publishers. Perez-Pereira, Miguel and Gina Conti-Ramsden. 1999. Language development and social interaction in blind children. Hove: Psychology Press Ltd. Piskorska, Agnieszka, Tomasz Krzeszowski, and Bogusáaw Marek. 2008. UczeĔ z dysfunkcją wzroku na lekcji angielskiego, Wskazówki metodyczne dla nauczycieli. Warszawa: Uniwersytet Warszawski. Sak-Wernicka, Jolanta. 2011. Context and understanding: evidence from the blind. PhD diss. The John Paul II Catholic University of Lublin. —. 2012. Gaps in blind learners’ knowledge – can Relevance Theory account for them? In Relevance Studies in Poland, vol.4. Essays on language and communication, edited by A. Piskorska. Warsaw: Warsaw University Press. Sperber, Dan. 1996. Understanding verbal understanding. In What is intelligence?, edited by J. Khalfa. Cambridge: CUP. —. 2010. The guru effect. Review of Philosophy and Psychology 1, no. 4: 583-592. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. Watson, Peter. 1960. On the failure to eliminate hypotheses in a conceptual task. Quarterly Journal of Experimental Psychology 12: 129-140.

116

Chapter Five

Wilson, Deirdre and Dan Sperber. 2004. Relevance Theory. In The handbook of pragmatics, edited by L. Horn and G. Ward. Oxford: Blackwell.

CHAPTER SIX RELEVANCE THEORY AND COGNITIVE BEHAVIORAL THERAPY THEORY IN THE ANALYSIS OF PSYCHOTHERAPEUTIC DISCOURSE ELWIRA SZEHIDEWICZ, UNIVERSITY OF WARSAW, POLAND

1. Introduction The main goal of this article is to discuss the applicability of Relevance Theory (RT) to the analysis of psychotherapeutic discourse. Relevance Theory was originally designed to account for the analysis of communication and cognition, but had a different focus from classical discourse analysis (Blakemore 2002). Nevertheless, RT linguistic studies are increasingly making use of contextual information provided by discourse (Wilson and Carston 2008). In addition, this article offers an overview of the most influential works on psychotherapeutic discourse and indicates areas where Relevance Theory, cognitive behavioral therapy theory and discourse analysis may converge. Moreover, after offering some justification for using RT to study psychotherapeutic discourse, an attempt is made at analyzing a piece of a therapy session involving metaphorical language (in Polish).

2. Why a study on Relevance Theory and psychotherapeuticdiscourse analysis? A study that attempts to combine a linguistic theory of communication with a therapeutic theory and discourse studies may become a target of criticism. On the one hand, this is a danger that all multidisciplinary studies should face at some point or another (Pawelczyk 2010), while, on

118

Chapter Six

the other hand, linguistic analysis of psychotherapeutic discourse has become one of my methods of monitoring and seeking greater insight into my job as a therapist. Moreover, it appears that there are obvious links between psychotherapy and language as well as points of intersection between cognitive behavioral psychotherapy and the Relevance Theory approach to communication. This issue is elaborated further upon in this article. My earlier analysis of psychotherapeutic sessions (Szehidewicz 2011) found that in the process of psychotherapeutic conversations both the therapist and the client put substantial effort into the comprehension process and into therapy in order to achieve a certain level of cognitive gains. The question still remains if the effort may be compensated by cognitive effects for both therapy interlocutors.1

2.1 Cognitive-behavioral therapy theory Cognitive-behavioral therapy theory (Beck 2011) is based on the assumption that in every situation people have thoughts whose consequences are threefold: emotional, physiological and behavioral (Reinecke and Clark 2003). One of the many characteristics of certain thoughts (called Negative Automatic Thoughts) is that they can produce emotions which may be much stronger and may last much longer than the initial thought. Apart from negative automatic thoughts, our cognitive appraisals are built on the basis of our experience about the world, about ourselves and others in the form of many functional and dysfunctional assumptions and beliefs which, in turn, direct the way we think we should act and live (Clark and Fairburn 1997; Reinecke and Clark 2003). The therapeutic process should be based on a partnership-based relationship and a good understanding of the patient’s core problem. The process of therapy incorporates many cognitive techniques (working with thoughts – e.g. reformulating, recognizing the relationship between thoughts, emotions, bodily experience and behavior, cognitive distortions; Socratic dialogue; arrow-down technique; metaphors; etc.) (Leahy 2003) as well as behavioral techniques involving various kinds of behavioral experiments (Bennett-Levy et al. 2004). Metaphor is considered to be a therapeutic tool. It can be used in either a directive or non-directive way, it can be structured by the therapist or not. If used by a patient, it is usually further elaborated on in the course of the therapy (Stott et al. 2010). 1

Cognitive effort and cognitive effect are used here in the sense of RT (Sperber and Wilson [1986] 1995, 123-132).

RT and CBTT in the Analysis of Psychotherapeutic Discourse

119

2.2 Relevance Theory In a nutshell, the relevance-theoretic approach to communication and cognition assumes that all human behavior is geared to the maximization of relevance (Sperber and Wilson [1986] 1995, 158; Wilson and Sperber 2002, 256) and that each communicative act carries a presumption of its own optimal relevance (Sperber and Wilson [1986] 1995, 159; Wilson and Sperber 2002, 258); these principles are called the Cognitive and Communicative Principle of Relevance, respectively. The process of communication involves combining different kinds of information, seeking to produce maximum cognitive effects (strengthening or contradicting old assumptions or forming new contextual implications) for the smallest processing effort. In the process of comprehension, the hearer arrives at the informative intention of the speaker by testing interpretive hypotheses in order of accessibility. This task consists of forming appropriate explicatures, retrieving intended contextual assumptions (implicated premises) and intended contextual implications (implicated conclusions) (Wilson and Sperber 2002, 262).Understanding metaphor, which typically relies on the simultaneous broadening and narrowing of a category, is on this approach not qualitatively distinct from understanding literal talk (Carston 2002, 89; Wilson 2003, 272). In this sense, the RT approach to metaphor is deflationary (Sperber and Wilson 2008, 85). Besides, as pointed out by Wilson and Carston (2008, 1): metaphorical language use is taken to lie on a continuum with other cases of loose use, including hyperbole [and] metaphor interpretation is a wholly inferential process, which does not require associative mappings from one domain … to another …

The interpretation of metaphor is guided by the search for relevance (Wilson and Carston 2008, 5). The hearer is invited to explore the assumptions stored as encyclopaedic information attached to the vehicle concept and select the portion of these assumptions intended by the speaker to be used for understanding the tenor concept. For instance, by uttering “Fiona is a princess” the speaker intends to activate in the hearer’s mind some stereotypical assumptions associated with princesses, like “princesses are delicate, vulnerable, spoilt, etc.” Other assumptions forming the encyclopaedic content of the concept ‘princess’, such as “lives in a palace”, “participates in official ceremonies” will not be activated as they do not contribute to the relevance of the utterance (Wilson and Carston 2008).

120

Chapter Six

2.3 Points of convergence between RT and CBTT What seems to link Relevance Theory and Cognitive-Behavioral Therapy Theory is their core subject – cognition and communication. Both theories deal with the process of comprehension of linguistic input and the consequences of that process, i.e. mutual understanding between the speaker and hearer. Relevance Theory looks at what is on the explicit and the implicit side of communication, which necessarily involves taking into consideration possibly important background information (e.g. contextual assumptions) (Sperber and Wilson 2002), while cognitive-behavioral therapy searches for what is in the mind of the speaker at the level of thoughts and possible basic assumptions about oneself, the world and other people (Beck 2011). Additionally, mental processes and representations seem to be core issues in both RT and in CBTT. The difference is that Relevance Theory proposes a division of layers of mental representations connected with different levels of metarepresentational abilities (Sperber and Wilson 2002) while cognitive-behavioral therapy theory does not take human metarepresentational capacity under consideration. What CBTT is concerned with is the division of mental representations into thoughts, assumptions and core beliefs, each of which may be in the linguistic or visual/sensual form. Moreover, the strength, importance and frequency of occurrence of each of these representations may be put on a continuum and/or a scale, which is directly connected with the fact that in CBTT there are no clear-cut boundaries between concepts, representations, categories, cognitive appraisals, emotions, etc. On the other hand, in RT the boundaries between concepts, their encyclopedic, lexical and logical entries seem to be clearly defined (Sperber and Wilson [1986] 1995, 86). Yet another similarity between RT and CBTT is that they underscore the importance of the functionality and/or relevance of human actions, be they thoughts, speech or other phenomena. However, there are areas in which the two theories display different foci: CBT aims to create models and tools for the meta-goal of relieving people in their psychological and/or life problems, while RT aims to explain what is happening in human interaction at the level of verbal and non-verbal communication (Sperber and Wilson [1986] 1995). The difference can be traced back to the fundamental division of linguistic studies of discourse into functional and structuralist in their theoretical outlook (Schiffrin 1994). As Blakemore (2002) says, Relevance Theory is not functional, but structuralist in nature as it does not perceive language to be a societal phenomenon. Unlike Relevance Theory, cognitive-behavioral studies are virtually entirely functional. All in all, it may be said that despite some essential differences between Relevance Theory and Cognitive Behavioral Therapy Theory, it

RT and CBTT in the Analysis of Psychotherapeutic Discourse

121

may turn out that some new interesting and profitable ideas may arise from doing a study that links the two approaches together (Žegarac 2012).

2.4 Discourse and RT Another important question that needs to be taken under consideration is whether it is possible or not to use Relevance Theory to analyze traditionally understood discourse in the first place. Blakemore (2002, 150) say that “a study of theory of verbal communication should not be construed as the study of discourse at all”. However, these authors also say that relevance theory does something impossible, because it embeds the study of discourse within the study of human cognition … [and] it is true that relevance theorists speak of understanding utterances (and sometimes of understanding discourse) and of deriving interpretations of utterances (and sometimes discourse). However, there is a fundamental difference between the two approaches. For relevance theorists, utterances (and discourse) are simply public phenomena which are produced as representations of private phenomena (thoughts). … [T]he point of processing is not to achieve an acceptable representation of the discourse, but to enlarge the mutual cognitive environment of speaker and hearer ... (2002, 156-157)

To sum up, it seems reasonable to try to find possible areas or interrelationships between cognitive theories like CBTT and RT with the use of psychotherapeutic discourse (Ritchie 2004)

3. Psychotherapeutic discourse There have been many successful attempts made at systematizing what is taking place in the field of discourse analysis (Schiffrin 1994; Alba-Juez 2009). Unfortunately, psychotherapeutic discourse turns out to be problematic as far as schematic typologies go. For this reason, I will limit the discussion to a short presentation of the most influential studies conducted on this kind of discourse. William Labov and David Fanshel (1977), in their Therapeutic Discourse: Psychotherapy as Conversation, advocated the study of therapeutic discourse. They used a variationist approach (Alba-Juez 2009, 156) in the thorough analysis of a 15-minute psychoanalytic therapy extract of a female patient suffering from anorexia nervosa. The greatest part of their study focused on speech acts performed by the patient and the rules of interpretation and production that relate speech to actions

122

Chapter Six

performed (Labov and Fanshel 1977, 6). Each piece of conversation was looked at from four different perspectives: the text itself, paralinguistic cues, expansion/analysis and interaction. It is in the interaction part that they studied actions performed by speakers. The authors found that in therapy “the crucial actions in establishing coherence … are not such speech acts as requests and assertions, but rather challenges, defenses and retreats, which have to do with the status of the participants, their rights and obligations and their changing relationships in terms of social organization” (Labov and Fanshel 1977, 58). Among other aspects, they devoted a lot of attention to the analysis of requests in discourse such as requests for actions, information, attention or approval. They claim that at a deeper level requests may be understood as challenges aiming at throwing a shadow of doubt on something that the patient believes to be true (Labov and Fanshel 1977, 63), but may as well be understood as criticisms, attacks, praise or reinforcement. Overall, the study showed that anorexia nervosa (ICD-10 2010) is characterized by an indirect kind of communication and the aim of therapy should be to enable the patient to use more direct ways of communication, which would let the patient be in touch with their own cognitive and emotional selves as well as with relevant family others (Labov and Fanshel 1977, 68). It is within the ethnography of communication (Schiffrin 1994) that a major study on psychotherapeutic conversations was conducted by the linguist Kathleen Warden Ferrara (1994) – she considered a psychotherapeutic session to be a speech event and people taking part in the session to be an ‘invisible speech community’ (Ferrara 1994, 15), whose participants need to socialize into becoming psychotherapists and ‘psychotherapy-users’ (Ferrara 1994, 15). Her extensive research on 48 hours of recordings of psychotherapeutic sessions of 10 clients and their therapists was made from the outsiders’ view (she did not participate in the recorded sessions). The clients suffered from a number of psychological problems, such as suicidal thoughts, alcoholism or marital problems. The therapists belonged to different schools of psychotherapy. Her study is influential for a variety of reasons. First of all, the author identifies differences between regular conversations and psychotherapeutic conversations on a number of dimensions: parity, reciprocity, routine recurrence, bounded time, restricted topic, remuneration and regulatory responsibility (Ferrara 1994, 39-42). According to the author, unlike in unspecialized conversations, therapeutic conversations happen to be partial in the sense that the agreement to share power and responsibility equally between speakers does not obtain here. The relationship is not equal. She also claims that the relationship between a therapist and a client is not reciprocal, because the participants are

RT and CBTT in the Analysis of Psychotherapeutic Discourse

123

somehow pre-taught as to whom, how long and when each person should talk, which highly depends on the theoretical background of the therapist. Among the issues Ferrara discusses are dreams and metaphors occurring in therapies. Peräkylä, Antaki, Vehviläinen and Leudar (2008) in their Conversation Analysis and Psychotherapy provide CA (conversation analysis) terminology that they prove to be useful in analyzing discourse. For example, Charles Antaki (2008) studied formulations put forward by therapists in psychotherapeutic sessions. He defined formulations as therapist’s summaries of the client’s input and specified their function. On the one hand, formulations show the attentiveness of the therapist, but on the other hand they let the clients draw some new meaning or understanding for themselves. Another study conducted on about a hundred therapeutic sessions analyzed clients’ responses to therapists’ reinterpretations (Bercelli, Rossano and Viaro 2008). The analysis has shown that reinterpretations may provoke a number of different responses on the part of the client. For example, acknowledgement, mere agreement, agreement followed by a description/narration (extended agreement), disagreements followed by a description/narration as well as disjunctively introduced autobiographical material (Bercelli, Rossano and Viaro 2008, 51). These researchers have managed to show that reinterpretations conducted by the therapist play a major role in creating a possibility for the client to extend his/her life story, and through this the clients can obtain new perspectives on their lives. In his article Psychotherapeutic Discourse Analysis (1995), Bradley Lewis uses interactional sociolinguistics (Tannen 1989) in an attempt to work with such communicative issues as connectedness, mutual understanding, coherence and involvement (also emotional involvement) (Lewis 1995). The author analyzed an extract from his therapy with a female patient using Tannen’s (1989) approach to discourse analysis and showed that therapeutic discourse is often like cross-cultural or crosssubcultural communication (Lewis 1995, 6) with the therapist often talking in high-considerateness style and patient talking in high-involvement style. High-considerateness style involves restrained enthusiasm, little overlap, slower pace and low mutual revelation. High-involvement style consists in high enthusiasm, frequent overlap, faster pace, greater persistence and high mutual revelation (Lewis 1995, 6). His analysis also showed that it is critical for therapeutic sessions to achieve a feeling of involvement and coherence. Coherence means mutual understanding that creates a feeling of coherence in the world. The therapeutic process really ‘clicks’ (Lewis 1995, 3) if the participants share assumptions (largely unconscious ones)

Chapter Six

124

about how they proceed in discourse. “The assumptions that a participant brings to the conversation are his or her ‘conversational style’. When conversational styles are matched and there is involvement, coherence is achieved” (Lewis 1995, 3). The critical discourse analysis issues of power and authority have been studied in a long-term psychoanalytic therapy by Nye (1998). This study shows that therapists and clients engage in issues connected with power and authority when they are working on the reconstruction of meaning. With an improving therapeutic relationship, these issues help in establishing a sense of alliance. We should underline that I have not yet found any complete studies conducted on psychotherapeutic discourse from the Relevance Theory perspective, except for a project on metaphorical language in psychotherapy currently being conducted by Isabelle Needham-Didsbury at University College London (Needham-Didsbury 2012).

4. Sample analysis In the exchange we will now be discussing, the therapist and the client are analyzing in detail what happens during a panic attack. In order to do so, they need to identify both emotional and physiological symptoms of the attack. In this example they concentrate on specifying the kind of feeling that starts such an attack, and it is only through the use of a metaphor that they can describe those feelings most effectively (Reboul 2011). (1) T:

P: T:

Nie wiem, czy tak do koĔca to rozumiem, tak jakby Pani ktoĞ tutaj przydusiá jakiĞ gáaz? ‘I’m not sure if I can properly understand you. So it is as if somebody crushed a stone against you here?’2 tak, tak, tak, tak, tak, to nie byá ból, tylko po prostu takie klikniĊcie. ‘yes, yes, yes, yes, yes, it was not pain, only simply such a click.’ Aha, czyli takie jednorazowe poáoĪenie tego (...) Aha, so it is like a single placement of it (...)’

The mutual cognitive environment of the two participants consists of the context of therapy, all the (often unsaid or unconscious) assumptions 2

There is an ongoing discussion concerning simile and metaphor comprehension processes in RT (Carston 2010a, 255; Hernández Iglesias 2010, 174). According to some authors there is a crucial similarity between the process of metaphor and simile comprehension as both may involve ad hoc concept construction (cf. Waáaszewska 2013).

RT and CBTT in the Analysis of Psychotherapeutic Discourse

125

about the way the client and the therapist cooperate as well as all the information they have about each other. As a way to illustrate the possible links between CBT, RT and discourse analysis, we will now focus on the metaphoric meaning of having a stone crushed against one’s body. Although, strictly speaking, the therapist uses a simile (using the word jakby ‘as if’) to prompt the image of having a stone crushed against the patient’s body, the actual formulation seems to be irrelevant here. When asking whether the image is adequate to represent the patient’s emotions, the therapist incites the patient to imagine that her state of panic is an act of a stone being crushed against her body, so in fact, to think about it metaphorically. The metaphor is a live one, which explains the importance of the discourse context in the process of arriving at its interpretation. Understanding dead metaphors, on the other hand, would rely to a greater extent on identifying the conventional meaning of the metaphor (Reboul 2011). At this moment in the therapeutic process, with the participants dwelling upon the issue of how the client felt when she was caught in a panic attack, it seemed that there was no other satisfactory way to explain and understand the client’s feelings than by the use of a metaphor, because the information was too complex or difficult to communicate literally (Reboul 2011). An explanation may also be offered here by the relevancetheoretic account of metaphor, according to which a metaphor should bring about an array of weak implicatures (Wilson and Sperber 2002, 275). This means that the hearer has the possibility of deriving many different cognitive effects and is left with a choice concerning the strength and/or repercussions of the metaphoric expression. The therapist’s intended contextual assumption is that if there is a strong pressure on one’s chest, then it might be compared to a sensation that one may have when one has a stone crushed against one’s body. The problem with this metaphoric expression is that it is difficult to spell out its possible explicatures by specifying the meaning of the ad hoc concept. If this is done, the therapist is left with a feeling of the expression’s meaning being obscured. On the one hand, there is a robust intuition that “A STONE* being crushed against a body” may mean that something very heavy, hard, achromatic, causing pain is being pressed against the client’s body/chest. On the other hand, this seems to be an unsatisfactory route of interpretation because when initiating the metaphoric talk, the therapist intended a myriad of different cognitive effects of conceptual, perceptual, emotional and physical nature, all of which are induced by representing a panic attack as “a STONE* being crushed against a body” and all of which are relevant and more complex than the ad hoc concept STONE*. A

126

Chapter Six

theoretical analysis without an ad hoc concept would not comply with the standard version of metaphor interpretation in relevance-theoretic account (Wilson and Carston 2008), according to which what may be happening is that the lexical meaning of “stone” is extended to a broader meaning of the concept with all the consequent implicatures derived via the creation of this ad hoc concept. As we noted earlier, the ad hoc concept STONE* may be something very heavy, hard, achromatic with a texture of some kind, causing pain when crushed against one’s body. Theoretically, the ad hoc concept STONE* in the analysis of the patient’s interpretation of the therapist’s input is somewhere there in the process of finding the meaning of what the speaker said, but the problem is that it does not seem to correspond with what the therapist really felt and with what was activated in her mind. The problem is naturally connected with the imagery that was activated somewhere there: namely, by a movie once seen by the therapist, in which people were lying on the ground with a one big stone crushing against their chests for a long time, till death. This image which is stored in the encyclopaedic entry for “stone” in the therapist’s mind differs from the prototypical version of an image of people being stoned, because usually when we think about stoning somebody we see people throwing stones at a person. However, in this particular conversational context it was this first, non-prototypical and personal image that was activated. The client’s old assumptions might have been that the therapist has problems with understanding how she felt as well as she gets a panic attack, it starts with such a simple click in her chest. The new information is that the therapist says that when you feel pressure on your chest it may be like all the feelings you get when you have a stone crushed against your body. Consequently, a cognitive effect may be either in the form of strengthening the old assumptions of the patient that the feeling is like having a stone crushed against one’s chest or in the form of new contextual implications that the therapist now finally understands how she feels and that the feeling is like when one has something very heavy, hard, achromatic and pain-causing crushed against one’s chest. When the patient says yes, it is like a click, the therapist receives new information that the client agrees with what the therapist has said as well as information that the activity talked about is like the “clicking of a stone” which activates an ad hoc concept STONE**, corresponding to the feeling of having a stone crushed against one’s body in a single, quick, momentary, short, pain-causing touch. At this moment the meaning of “stone” is narrowed to a single momentary touch of something very heavy, hard, achromatic on the client’s body (Wilson and Carston 2008; Carston 2002).

RT and CBTT in the Analysis of Psychotherapeutic Discourse

127

However, returning to what was said about “having a STONE* crushed against one’s body” and “having a STONE** clicked against one’s body”, it seems that this kind of interpretation is limited in the sense that there is much more that is taking place in the interpretation process of the phrase “clicking of a stone on one’s body”. And much of it lies in the encyclopedic entry of the concept “stone”, because not only did my (the therapist’s) mind activate such assumptions as “if somebody puts a stone on my chest and then immediately takes it away it hurts and then it is over”, but also I had an image of a movie I once saw where people were stoned as a method of torturing them. I somehow even heard them screaming and the stone I envisioned was the one from a movie. What may now prove useful is Cognitive-Behavioral Therapy Theory and its approach to the cognitive appraisals we hold in our mind. The idea is that negative automatic thoughts, dysfunctional assumptions and basic beliefs (or schemes) may be stored in the human mind both in verbal and/or in visual form (Paivio 1986). The line of metaphor analysis suggested by Carston (2010b) and Pilkington (2010) looks very promising, especially the suggestions that ad hoc concepts may take an imagistic form and that encyclopedic information can be represented not only as assumptions, but also as visual images. On this line of thought, the approach to metaphor interpretation in RT would be more acceptable for my purposes (Carston 2002, 90-93; Carston 2010b; Pilkington 2010). The part of the encyclopedic entry for “stone” activated in the context of “stone crushed against one’s body” is an image from a movie in which I saw people suffering with the pain of being stoned, and it is this picture that leads my body to feel both physical and emotional pain that may be in some way comparable to what the client felt. According to CBTT bodily and emotional experiences are just consequences of mental (lexical and/or visual) inputs. The direction may be the following: speaker’s lexical input – hearer’s imagistic ad hoc concept – hearer’s emotional response (apprehension, surprise, pity, sadness etc.) – hearer’s physiological reaction (pain in the back) – various implicatures. The therapist’s implicated conclusions are that the feeling in the patient’s body that starts the panic attack is a short crushing of a stone against one’s body, which is very painful, akin to when the people in a certain movie were crushed to death by a stone. Additional cognitive effects may be connected with sadness and pain in the therapist’s back as well as with the feeling of heaviness in the entire body that leads to an understanding what the patient may be going through. There was also a sense that the emotional tension that had previously accumulated during the session evaporated after the exchange under analysis. For the patient, it

128

Chapter Six

seems that this release of tension, but also the feeling of being understood were the main cognitive and therapeutic effects.

5. Summary In this article I tried to show that it is possible to fruitfully draw upon three theories – Relevance Theory (a theory of communication), CognitiveBehavioral Therapy Theory and discourse analysis – in the search for a better understanding what happens in psychotherapeutic exchanges. The example used to illustrate these interconnections was a metaphorical expression “a stone being crushed against one’s body” as used in a therapy exchange. I hope to have clearly demonstrated the usefulness of the Relevance Theory approach to communication, as it is only through the use of RT methodological tools that the insights of the current discussion could be easily reached. Another issue that I would like to underscore is that according to RT a number of differing ad hoc concepts may be activated in metaphor interpretation, but only one is processed further on (Carston 2002, 96; Carston 2010a, 243). My reservation here is that the idea of ad hoc concepts presented in this way may turn out to be fruitless in some cases (as it seems to be in the example discussed herein) (Carston 2010a, 243). However, new proposals are now appearing within RT concerning a “second cognitive route” to metaphor comprehension (Carston and Wearing 2011, 285; Carston 2010b, 297). According to this proposal, in extended metaphors “the literal meaning (and crucially, the image it evokes) plays a more extensive role than just providing the materials for constructing the intended ad hoc concepts” (Carston and Wearing 2011, 305). I do not feel in a position to weigh in on such issues within Relevance Theory, but my intuitions certainly run closer to this newest account of metaphor, which would simplify matters to a satisfactory extent for my purposes (Wilson and Carston 2008). Additionally, a final word needs to be said about cognitive effort vs. effects. My earlier discussions of psychotherapeutic discourse showed that for both the client and the therapist cognitive effort was great and was not really offset by satisfactory cognitive effects (Szehidewicz 2011). This time, however, a metaphor was shown to achieve something extraordinary, because while requiring a certain effort, it provided cognitive effects that left both the therapist and the client satisfied, mainly because both participants gained a feeling of being understood by or of being able to understand another person.

RT and CBTT in the Analysis of Psychotherapeutic Discourse

129

References Alba-Juez, Laura. 2009. Perspectives on discourse analysis: Theory and practice. Cambridge: Cambridge Scholars Publishing. Antaki, Charles. 2008. Formulations in psychotherapy. In Conversation analysis and psychotherapy, edited by A. Peräkylä, C. Antaki, S. Vehviläinen and I. Leudar. Cambridge: Cambridge University Press, 26-42. Beck, Judith. 2011. Cognitive behavior therapy: Basics and beyond. 2nd ed. New York: The Guilford Press. Bennett-Levy, James, Gillian Butler, Melanie Fennell, Ann Hackman, Martina Mueller, and David Westbrook, eds. 2004. Oxford guide to behavioural experiments in cognitive therapy. Oxford: Oxford University Press. Bercelli, Fabrizio, Federico Rossano and Maurizio Viaro. 2008. Different place, different action: Clients’ personal narratives in psychotherapy. Text & Talk 28, no. 3: 283-305. Blakemore, Diane. 2002. Relevance and linguistic meaning: The semantics and pragmatics of discourse markers. Cambridge: Cambridge University Press. Carston, Robyn. 2002. Metaphor, ad hoc concepts and word meaning – more questions than answers. UCL Working Papers in Linguistics 14: 83-105. —. 2010a. Explicit communication and ‘free’ pragmatic enrichment. In Explicit communication: Robyn Carston’s pragmatics, edited by B. Soria and E. Romero. Houndmills, Basingstoke: Palgrave Macmillan, 217-285. —. 2010b. Metaphor: Ad hoc concepts, literal meaning and mental images. Proceedings of the Aristotelian Society cx, no. 3: 297-323. Carston, Robyn and Catherine Wearing. 2011. Metaphor, hyperbole and simile: A pragmatic approach. Language and Cognition 3, no. 2: 283312. Clark, David, and Christopher Fairburn, eds. 1997. Science and practice of cognitive behaviour therapy. Oxford: Oxford University Press. Ferrara, Kathleen Warden. 1994. Therapeutic ways with words. Oxford and New York: Oxford University Press. ICD-10. 2010. International statistical classification of diseases and related health problems (10th revision). (September 2012)

130

Chapter Six

Hernández Iglesias, Manuel. 2010. Ad hoc concepts and metaphor. In Explicit communication: Robyn Carston’s pragmatics, edited by B. Soria and E. Romero. Houndmills, Basingstoke: Palgrave Macmillan, 173-182. Labov, William and David Fanshel. 1977. Therapeutic discourse: Psychotherapy as conversation. New York: Academic Press. Leahy, Robert. 2003. Cognitive therapy techniques: A practitioner’s guide. New York: The Guilford Press. Lewis, Bradley. 1995. Psychotherapeutic discourse analysis. American Journal of Psychotherapy 49, no. 3: 371-385. Nye, Catherine. 1998. Power and authority in clinical practice: A discourse analysis approach to narrative process. Clinical Social Journal 26, no. 3: 271-280. Needham-Didsbury, Isabelle. 2012. Metaphors in psychotherapy. Paper presented at the conference: Interpreting for Relevance: Discourse and Translation (IFR), Warsaw, 25-26 September. Paivio, Allan. 1986. Mental representations: A dual coding approach. New York: Oxford University Press. Pawelczyk, Joanna. 2010. Talk as therapy. Linguistic investigations in psychotherapy. PoznaĔ: Pracownia Wydawnicza IFA UAM. Peräkylä, Anssi, Charles Antaki, Sanna Vehviläinen and Ivan Leudar. 2008. Conversation analysis and psychotherapy. Cambridge: Cambridge University Press. Pilkington, Adrian. 2010. Metaphor comprehension: Some questions for current accounts in Relevance Theory. In Explicit communication: Robyn Carston’s pragmatics, edited by B. Soria and E. Romero. Houndmills, Basingstoke: Palgrave Macmillan, 156-172. Reboul, Anne. 2011. Live metaphors. In Philosophical papers dedicated to Kevin Mulligan, edited by A. Reboul.

(August 2012) Reinecke, Mark and David Clark, eds. 2003. Cognitive therapy across the lifespan: Evidence and practice. Cambridge: Cambridge University Press. Richie, David. 2004. Metaphors in conversational context: Toward a connectivity theory of metaphor interpretation. Metaphor and Symbol 19: 265-287. Schiffrin, Deborah. 1994. Approaches to discourse. Oxford: Blackwell. Sperber, Dan and Deirdre Wilson. [1986] 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell.

RT and CBTT in the Analysis of Psychotherapeutic Discourse

131

—. 2008. A deflationary account of metaphor. In The Cambridge handbook of metaphor and thought, edited by R. W. Gibbs. Cambridge: Cambridge University Press, 84-105. Stott, Richard, Warren Mansell, Paul Salkovskis, Anna Lavender, and Sam Cartwright-Hatton, eds. 2010. Oxford guide to metaphors in CBT: Building cognitive bridges. Oxford: Oxford University Press. Szehidewicz, Elwira. 2011. Emotions in a cognitive-behavioral therapy session – a Relevance Theory perspective. In New pathways in linguistics, edited by S. Puppel and M. Bogusáawska-Tafelska. Olsztyn: UWM. Tannen, Deborah. 1989. Talking voices: Repetition, dialogue and imagery in conversational discourse. Cambridge: Cambridge University Press. Waáaszewska, Ewa. 2013. Like in similes – a relevance-theoretic view. Research in Language 11, no. 3: 323-334. Wilson, Deirdre. 2003. Relevance and lexical pragmatics. Italian Journal of Linguistics 15, no. 2: 273-291.Wilson, Deirdre and Robyn Carston. 2008. Metaphor and the ‘emergent property’ problem: A relevancetheoretic treatment. The Baltic International Yearbook of Cognition, Logic and Communication 3. (November 2012) Wilson, Deirdre and Dan Sperber. 2002. Relevance theory. UCL Working Papers in Linguistics 14: 249-290. Žegarac, Vladimir. 2012. U-pragmatics and e-pragmatics: Making a case for I-pragmatics. Paper presented at the conference: Interpreting for Relevance: Discourse and Translation (IFR), Warsaw, 25-26 September.

CHAPTER SEVEN THE TRANSLATOR’S INTENTIONS – THE SAME AS OR DIFFERENT FROM THE AUTHOR’S? CASES OF MANIPULATION IN POLISH TRANSLATIONS OF BRITISH PRESS ARTICLES IN 1965-1989 EDYTA ħRAàKA, UNIVERSITY OF SILESIA, KATOWICE, POLAND

1. Introduction How we understand the phenomenon of manipulation in translation is generally conditioned by two distinct meanings of the word ‘manipulation’, listed in most dictionaries and quoted by many authors dealing with the problem: one physically oriented towards objects, the other psychologically oriented towards people. As de Saussure defines the notion, based in part on the American Heritage Dictionary: “To manipulate is to use one’s hands to instrumentalize an object (to operate or control by skilled use of the hands …) and sometimes to change the object’s original shape” (de Saussure 2005, 117). The other definition, given by for instance Salich (2011, 29) drawing upon the Cambridge Advanced Learner’s Dictionary, includes an aspect of purposeful deception: “manipulation (INFLUENCE) (mainly disapproving) controlling someone or something to your own advantage, often unfairly or dishonestly.” Manipulation understood as any kind of intervention in the ST while seeking equivalent options for the TT would correspond more with the first definition, while manipulation understood as deliberately trying to change the ST for some ideological reasons would represent the idea behind the second definition.

The Translator’s Intentions

133

Consciousness and intention are key notions for deliberate manipulation and, at the same time, preconditions for achieving the translator’s manipulative aim, if any such aim exists. As Kizeweter writes: Translation may be seen as a process in which conscious and intentional reshaping of the source text leads to considerable changes in the target text in relation to its original, in extreme situations even to complete alterations of the source’s meaning. Here the idea of translation being simply a copy of the original in a different language is long abandoned. (Kizeweter 2011, 20)

In the author’s research project to be presented here, which is a comparative study of original articles from the British press concerning the situation in Poland and their translations into Polish published in Forum magazine in the years 1965-1989 (from the magazine’s inception to the fall of communist rule in Poland), such “complete alterations” are in fact quite common cases. Indeed, manipulation in the sense of distortion seems to have been a condition for an article to be accepted for publication. This is presumably due to the requirements of the censorship authorities in the People’s Republic of Poland, which will be discussed in detail below. Relevance Theory, a theory of communication and cognition in general, offers a solid theoretical framework for discussing the idea of manipulation in translation, in terms of the notions of informative and communicative intentions (introduced by Sperber and Wilson [1986] 1995, see below). The authors of press articles published under the propagandabased regime of communist-era Poland were to be tools in the hands of the communist authorities, wielding the power to create public opinion. The informative intention, which consists in making a set of assumptions manifest to the audience, thus primarily or sometimes exclusively fell within the authors’ (or even more, the translators’) scope of power. Whatever they were permitted to include in the Polish texts could attain the status of a guideline for common beliefs, especially thanks to the commonly recognized authority of Western media on the political situation in Poland. This is why the treatment of source texts by Forum translators seemed to offer such interesting and engaging material for research. This paper will discuss certain linguistic elements of such translated articles, the cultural stereotypes evident therein, and ways of attaining the aim of targeted communication. We will come to a conclusion concerning how much manipulation can be attributed to the research material and what type of attitude towards the ST rendering (purposeful distortion or objective equivalent seeking) is evidenced by the Forum translators. The

134

Chapter Seven

extent to which they manage to convey positive cognitive effects through their translations will be assessed.

2. Definitions and types of manipulation in translation One of the common opinions on the subject under discussion is that there are as many attitudes towards translational manipulation as there are scholars dealing with the problem. Dukate (2009) states that nobody has explicitly conceptualised and defined translational manipulation. Various scholars and practitioners speak and write about this phenomenon, but do not provide a concrete definition of manipulation, instead relying on a vague common understanding of it. Dukate states that there are generally two ways translational manipulation is understood: -

-

manipulation as handling (comprising normal translational strategies, which do not involve major changes but only result from the linguistic differences of the languages in contact, and are neither positive nor negative), manipulation as change, taking on a positive, adaptive character (e.g. in localization), or a negative one (e.g. in distortive ideological translation), which in practice means processing a text according to someone’s intention or cultural requirements and involves major changes (Dukate 2009, 84-85).

Dukate proposes a definition in which manipulation is considered to be the translator’s handling of a text which results in an adaptation of the text for the target audience, taking heed of the differences (cultural, ideological, linguistic and literary) between the cultures involved, which takes place in a particular cultural setting and is carried out by a human agent, the consequence of which is the possible influence of individual- or psychology-related factors upon the end product (Dukate 2009, 11-12, 84). In seeking to define manipulation, the author quotes Theo Hermans’ opinion that “from the point of view of the target literature all translation implies a degree of manipulation of the source text for a certain purpose” (1985, 11), which has become known as the manipulation hypothesis (Dukate 2009, 15). The Manipulation School, with which Hermans is associated, mainly interested in the translation of literary texts and cultural aspects of translation, perceives translation as the rewriting of texts for a specific target audience according to the TL norms and including various constraints (Dukate 2009, 43). The school was most active during the 1970s and 1980s. In the collection of articles Translation, History and

The Translator’s Intentions

135

Culture, edited by Bassnett and Lefevere (1990), translation is studied in a broader political and cultural context, characterised by power and manipulation, and influenced by institutional and ideological factors. According to this group's views, any translation might be characterised as manipulation because of the factors and processes in three translational stages, the pre-production stage, the production stage and the postproduction stage. The first stage relates to the selection of a text for translation. In the production stage the translator receives certain guidelines, indications and hints about the way a text should be translated. Also various objective (e.g. language-related) and subjective (e.g. ideological and psychological) factors are of considerable importance. The post-production stage is related to the way the text is presented and reflected in metatexts and discourse. All those activities, which could be termed ‘manipulation’, are mostly due to cultural, political and ideological considerations. According to Dukate (2009, 49), manipulation manifests itself in the text in the form of shifts – changes in translations as compared to the originals (after Catford 1965). Shifts may occur for various reasons, the most common of them being linguistic, cultural, psychological or ideological. Shifts may also result from ignorance. They can be realised in the form of different strategies, for example, omissions, additions, substitutions, attenuations or replacements, which are typical of ideological or cultureinduced manipulation (Dukate 2009, 82). According to Dukate: Grammar-related shifts could hardly be considered manipulative in the conventional sense of the word. Shifts, which are due to cultural or ideological considerations, could be labeled instances of conventional manipulation. However, the labels in this case depend on one’s perception and understanding of manipulation. (Dukate 2009, 55)

Dukate claims that manipulation can take the form of a translation strategy or method. It may be voluntary or mandatory. Voluntary manipulation refers to such cases where the translator might not have introduced a shift in the output, but nevertheless has done so. Mandatory manipulation usually results from the fact that translatorship plays a social role within a cultural context and certain existing standards of translation expressed in the form of rules, norms and conventions are observed. Dukate goes on to state that there are two major types of manipulation: text-external manipulation and text-internal manipulation. Under each of these, three further types can be differentiated: manipulation as improvement, manipulation as handling and manipulation as distortion, which can be either conscious or unconscious (Dukate 2009, 113-115). Conscious

136

Chapter Seven

manipulation is intentional and can be of two kinds: ideological and linguo-cultural, respectively aimed at following someone’s ideology and at overcoming linguistic and/or cultural differences. Unconscious manipulation is unintentional and a result of how the human mind functions, which manifests itself in normalization, explicitation, or digression, and on the other hand – as errors that are a consequence of ignorance (Dukate 2009, 87). Other scholars dealing with manipulation also mention two types of the phenomenon: conscious and unconscious, e.g. Zauberga (2004, 67) and Kramina (2004). According to Kramina’s definition: Manipulation arising due to ideological, economic, and cultural considerations proceeds consciously, and thus might be termed conscious manipulation. Manipulation ascribed to the features of human psychology and manipulation due to ignorance (lack of language or world knowledge) might be termed unconscious manipulation. (Kramina 2004)

As Dukate claims (2007), manipulation as handling is used in the sense of a change that cannot be labelled ‘improvement’ or ‘distortion’. Distortion is the conventional connotation for the word manipulation. Traditionally, only manipulation as distortion qualifies as manipulation proper, thus it is the type of our particular interest in the context of this research project. Distortive manipulation is defined as changes which misrepresent reality and are perceived as misrepresentation by the target culture (Dukate 2007). Dukate goes on to state that text-external manipulation as conscious distortion is a policy leading to the translation of carefully selected authors and texts only, thus leads to misrepresenting the source culture. The translated texts may also be accompanied by guidelines in the form of external guidance on how to perceive a particular text. Text-external manipulation as unconscious distortion consists in automatic processes that lead to a distorted perception of the culture(s) involved, or are consequences of a lack of knowledge on the part of the text handler (Dukate 2007, 2009, 101-102). Text-internal manipulation as conscious distortion is a type of manipulation, which usually is due to the dominant political ideology, and may take the form of omissions, additions, substitutions and attenuations. Text-internal manipulation as unconscious distortion is manipulation due to the translator’s lack of professionalism, and is manifested as errors that seriously mislead the reader and distort the original text (Dukate 2007).

The Translator’s Intentions

137

3. Manipulation in translation through the perspective of Relevance Theory Speaking in terms of Relevance Theory, de Saussure (2005, 118) claims that manipulative discourse is a type of usage of natural language in which the speaker’s aim is to make certain number of assumptions manifest to the hearer and have him consent to them, provided that they would be rejected under normal conditions. He writes: “… one of the core problems of manipulation in language resides in the identification by the hearer of the manipulative intention through formal and non-formal features; when this detection fails, manipulation is effective” (de Saussure 2005, 118). Applying the relevance-theoretic definitions of the speaker’s intentions to translation, it should be stated that the communicative intention pursued by the translator as communicator consists in making it mutually manifest to an audience and the translator that this particular informative intention is being aimed at.  o matter if through the explicit content of a message, with a set of decoded assumptions, or the implicit content, with inferred assumptions, a translator’s primacy over the author in the process of manifesting intentions and creating a desired cognitive effect on a reader remains an unquestionable fact. This idea is particularly reasonable when the translator is expected to serve some purposes, such as propaganda. Blass (2005, 176, after Taillard 2000, 169) argues that the two levels of speaker intention distinguished by Sperber and Wilson ([1986] 1995) – an informative intention, making certain assumptions manifest to the audience, and a communicative intention, involving only the recognition of the informative intention – should be extended to include a persuasive intention, which constitutes the essence of manipulation when the manipulative intention is covert and not part of the communicative intention. De Saussure sees the following basis for manipulation to reach its effectiveness: Thus to manipulate a human being may be about using a person, i.e. have that person adopt specific behaviours to fulfill the needs and interests of the manipulator, regardless of the ones of the manipulated. But an individual, contrarily to an object, has a cognition that enables him to pursue his own interests; therefore, our first step is to admit that a manipulation first of all manipulates some aspects of human cognition, notably reasoning, checking for likeliness, emotions, etc. (de Saussure 2005, 117)

138

Chapter Seven

Manipulation is related to applying constraints the manipulated individual is not aware of (de Saussure 2005, 117). The condition of the lack of awareness on the part of the manipulated as a success of manipulation is mentioned in all definitions. To add one more, Blass (2005, 170) quotes Puzynina (1992), paraphrased in Galasinski, according to which: Manipulation is an attempt to affect the target in such a way that his behavior/action is an instrument of attaining the goals of the manipulator, who acts without using force but in such a way that the target does not know the goal of the manipulator’s actions. (Blass 2005, 170)

There are certain conditions that facilitate manipulation. According to de Saussure: Totalitarian ideological manipulative discourse is most efficient in societies where some conditions are met, especially a state of crisis. For example, economic recession, war, and post-war situations are factors that favour less stable moral judgement for individuals and open the way for a change in moral values. (de Saussure 2005, 124)

This idea gives us motivation to seek manipulation in translations of Western press articles performed into Polish in communist-era Poland. When manipulation is intended, the first deception lies in creating the hearer’s false belief that the speaker is benevolent, cooperative or relevant – which is at the level of the communicative intention, not the level of information as such (de Saussure 2005, 137). In the case of the Forum articles to be studied here, such a belief, if possessed by readers, would be an effect of deception, as the censorship authorities tried not to allow full and credible information to be available in the Polish press.

4. Text typology and manipulation Even if some authors working in Relevance Theory (e.g. Gutt 1990) have a skeptical attitude towards the idea of text typology as such and question text classification based upon subject matter (e.g. legal or scientic texts), instead classifying texts on the basis of the main contextual focus (e.g. expository, argumentative or instructional texts), the idea of text types nevertheless still seems vital in the context of manipulation of particular texts (no matter if viewed contextually or thematically) and some authors still insist on referring to text types in their analyses. According to Dukate, articles in the press, operating instructions, reports and accounts, are a type

The Translator’s Intentions

139

of informative text, which together with vocative texts (advertisements, brochures, electoral and some political speeches) and legal texts, belong to a group called non-literary texts – one of three text-types, alongside literary and sacred texts, distinguished by the author as regards to manipulation and its different representations in translation (Dukate 2009, 28, 144, based on the classification of text types by Reiss 1977). As Dukate characterises press articles: In translation they are taken to be content-oriented texts, because the main purpose of these texts is to pass on information. Also readers of this type of text generally expect to receive information and are less critical as to the language quality of the text. The reader usually expects that texts of this type will not be translated word-for-word and sound more or less natural in the target language. (Dukate 2009, 29)

This definition, if understood literally, enables translators to adjust the ST content to their own needs and to the needs of the whole system in which they function – politics, their papers’ targets, their employers’ business, etc.  o doubt, in such a case manipulation can be taken for granted and surely characterised as being of the type which involves not the mere transfer of the ST content into the TL, but purposeful modifications. As concerns the motive for manipulation in the corpus analysed, it was mainly ideological and, we should state clearly, it was the censorship office that responsible for it to the largest extent. As Szarkowska (2011, 161) writes: Censorship in post-WW II Poland was legally regulated by a decree of 5th July 1946 issued by Krajowa Rada Narodowa i Rada Ministrów (‘ ational State Board and the Council of Ministers’), which established a censorship body known as Gáówny Urząd Kontroli Prasy i Widowisk (Pietrasik 2006, 443). According to Pietrasik (2006, 443) the censorship office was controlled by the propaganda division of KC PZPR (‘Central Committee of the Polish United Workers’ Party’).

As Elias claims, based on Romek (2000, 35-36), the censorship in Poland was abolished on the 29th of January 1990 when the PZPR, ‘the Polish United Workers’ Party’, to whose authority the Gáówny Urząd Kontroli Prasy, Publikacji i Widowisk (‘Main Office for Control of Press, Publications and Public Spectacles’) was subordinate, ceased to exist (Elias 2011, 138-139).

140

Chapter Seven

5. Manipulative strategies in translation De Saussure mentions two types of strategies used by the manipulator to “block truth, likeliness and acceptability checking, as well as consistency checking”: local and global strategies. Local strategies constrain the interpretation at the level of utterance processing. Global strategies are used to create adequate social and psychological conditions to obtain irrational consent. Both strategies can be of two types: linguistic (e.g. rhetorical questions) and non-linguistic, which accompany the discourse – for written discourse this may be the typeface, the organization of text on the page, images, etc., as is typical of journals and newspaper layouts (de Saussure 2005, 126-128). As de Saussure claims: “… manipulation in discourse is often discovered on the basis of inconsistencies and (formal and non-formal) fallacies” (de Saussure 2005, 114). In his view, manipulation can take place when some problematic inferences appear due to the inappropriate use of a word, changing the understanding of a concept, e.g. as in the case of a word “democracy” understood not as a political system enabling people to take part in political decisions, but as the freedom of investment (de Saussure 2005, 116). Zauberga (2004) claims that ideological manipulation can take a form of -

-

deletion – a frequently pursued translation strategy under Soviet rule and, possibly, the best example of open ideological manipulation together with feminist translation (a type classified by Dukate (2009, 75) as distortion), substitution, addition, attenuation (Dukate 2009, 58, 60).

Insertion of some material in place of omitted material constitutes additions and modifications in the translation (Dukate 2009, 62, after BenAri 2002, 294), and seems to be quite a common procedure in the corpus analysed. As Galasinski (2000, after Blass 2005, 170) states: “Withholding information, controlling it, is the essence of deception in general” and this is certainly evidenced as a way of manipulating the translations in the corpus analysed for this paper. Galasinski claims that deception manifests itself by omission and by commission (2000, 22). Blass defines the terms mentioned in the following way:

The Translator’s Intentions

141

Under ‘omission’ falls information that is withheld in spite of the fact that it could have been relevant to an addressee in a particular situation. Omission can be passive, in which case the speaker is simply withholding information […]. Manipulation by commission happens if a deceiver is active. Her goal is to make the addressee acquiring or continuing a belief that she intends the addressee to accept. This can be done either explicitly or implicitly. Under explicit commission fall lies, half-truths (distortions), evasions and equivocations. Under implicit commission fall misleading underspecified explicatures and false implicatures. (Blass 2005, 173)

6. Corpus analyses, strategies used and manipulative intention effects in the translation of Forum articles 1965-1989 The claims made by the scholars quoted above, concerning the essence of manipulation in translation, were borne out by the analyses performed of Forum articles. The sample analysed includes Polish articles carried in the magazine’s column Echa Polskie, as compared with their source English versions published in the British press (The Guardian, The Times, The Observer, Financial Times), accessed by the author in the British Library, London. The corpus consists of 20 articles, i.e. 10 original versions and 10 translations. The strategies found to be most commonly used in the corpus specifically aim to distort the ST content, rather than handling it in the most equivalent form acceptable in the TT. The application of such strategies as omissions, additions, or substitutions considerably alters the articles’ reception by the Polish readers. Such strategies as attenuations, or sometimes exaggerations, and reorderings also serve the manipulative aim understood as the conscious, or even planned ideological deformation of the ST content. Omissions were found to be the most frequently used strategy. There were 59 cases of omission observed in the corpus, usually being separate pieces of crucial information or even whole paragraphs present in the English articles but absent from their Polish translations. Such disappearing passages for instance include: (1)

“This means that besides the danger of spontaneous outbreaks of discontent there is a still greater danger that hardliners inside the apparatus, perhaps with Soviet support, will engineer provocative incidents in the hope of driving the Government into using force against its own people, thereby provoking civil war and Soviet intervention.”

142

(2)

(3)

Chapter Seven

“ o one here has any doubts that the Russians will not shrink from invading Poland if they felt that the party was disintegrating and the leadership was in danger of losing control.” [in this particular case the omitted passage is replaced with the elided-material marks “…”] “He added that he could not ‘over-emphasise’ how much damage the priest’s murder had done to Polish interests and angrily rejected allegations that he might have been involved in the conspiracy. He said he would ‘not even stoop to answering such charges.’ But he vowed there would be no cover-up in the case.”

There are also some minor omissions of irrelevant information or single words or expressions, for instance: (4)

“Pink-faced and gasping in the tremendous heat of a Rumanian summer, the Warsaw Pact leaders achieved this week in Bucharest a discreet but important adaptation to the changing climate in Europe”, translated without the introductory comment up to the comma, as follows: “Przywódcy Ukáadu Warszawskiego dyskretnie, ale w sposób nie pozbawiony znaczenia dostosowali siĊ do zmieniającego siĊ klimatu w Europie”.

Additions were found to be the second most prominent group of strategies used to manipulate the TT content, with 33 cases observed in the corpus, normally involving translators’ explicit comments on the ST content, sometimes also in tandem with pieces of specific information absent from the ST, e.g.: (5)

(6)

(7)

“Nie moĪna pozbyü siĊ obawy, Īe byü moĪe wolny związek zawodowy, podejmując swą akcjĊ nie w porĊ, zraniá siĊ Ğmiertelnie”, a possible translation of which would be: ‘We cannot get rid of the fear that, by taking action at the wrong time, the free trade union may have wounded itself fatally’. “NaáoĪywszy zamiast zwykle noszonych ciemnych okularów szkáa przejrzyste”, which would mean ‘Wearing normal glasses instead of his usual dark glasses…’, “Przedstawiá on pogląd, Īe sankcje zastosowane przez Zachód pomogą Polsce wáaĞciwie oceniü siáĊ jej sojuszu ze Związkiem Radzieckim i pozostaáymi krajami obozu socjalistycznego. «Są to partnerzy solidni…którzy szanują nasze suwerenne prawo, byĞmy…rozwiązywali nasze polskie problemy. Daje nam to oparcie

The Translator’s Intentions

(8)

143

…(i) poczucie pewnoĞci»”, the English equivalent of which could be: ‘He expressed his view that sanctions imposed by the West will help Poland properly assess the power of its alliance with the Soviet Union and the rest of the socialist countries. “They are reliable partners… who respect our sovereign law, so as to let us… solve our Polish problems. It gives us support …and confidence”’, “W tym wspaniaáym wnĊtrzu rozsadzono dziennikarzy wzdáuĪ czterech dáugich stoáów. Za stoáem ustawionym przy jednej ze Ğcian widniaáa niewielka sylwetka generaáa”, which might be translated as: ‘In this magnificent interior journalists were seated along four long tables. Behind the table situated near one of the walls there was a small silhouette of the General’.

There might even be an addition of a whole paragraph, with an idea completely absent from the ST, like in the following comments added at the end of a translated text: (9)

(10)

“Telewizja moskiewska dodaáa pewien historyczny – i na czasie – kontekst, oznajmiając tego samego dnia o publikacji tajnego przemówienia Chruszczowa z 1956 r. (na XX zjeĨdzie KPZR – przyp. FORUM). Przemówienie to byáo sygnaáem wielkich, lecz faászywych nadziei na nowy początek w Europie Wschodniej, jak fatalnie zakoĔczony dla WĊgrów i niemal tak samo dla Polaków. Gomuáka wykorzystaá strajkujących robotników, by powróciü do wáadzy, po czym uáoĪyá siĊ z Chruszczowem i woáanie o nowy system polityczny skierowaá w reformistyczne ksztaáty. BojowoĞü robotnicza odĪyáa w 1970 r., gdy polscy robotnicy z GdaĔska doprowadzili do upadku Gomuáki. W pewnym stopniu oddano sprawiedliwoĞü mĊczennikom poznaĔskim z czerwca 1956 r.” [briefly put, this is a reference to a speech by Khrushchev on Gomuáka’s policies and the Polish leader’s fall], “Rozmowy – dodaje siĊ w komunikacie – potwierdziáy solidarnoĞü Komunistycznej Partii Związku Radzieckiego z wysiákami Polskiej Zjednoczonej Partii Robotniczej. Pomaga to ustabilizowaü sytuacjĊ w Polsce i przyczynia siĊ do obrony «podstawowych wartoĞci socjalizmu».” [this is a description of solidarity between the Communist Party in the USSR and the Polish United Workers’ Party].

144

Chapter Seven

Sometimes single words or expressions are added for explicitation, especially with dates, e.g. “b.r.” (‘this year’), or as in the following case (shown in boldface): (11)

“Although production was 10 per cent down in the first quarter of 1982”, which has been translated as: “Choü w I kwartale br. Produkcja PRL zmalaáa o okoáo 10 proc. w stosunku do analogicznego okresu sprzed roku” (which may be back-translated as ‘receded by around 10 per cent in comparison with the analogous period last year.’)

One interesting method of manipulation involves using two strategies at the same time, as in the following example where the Polish translation has a summary added (an ostensive factor of manipulation), shown here in boldface: (12)

“Ostatecznie jest to w koĔcu historyczny kompromis, a nie zwyciĊstwo. Dopóki nie przywróci siĊ powszechnego zaufania, zarówno WaáĊsa jak i Jaruzelski bĊdą generaáami bez armii”, which in English would be: ‘In the end, this is in fact a historical compromise, not a victory. Until the common trust is reinstated, both WaáĊsa and Jaruzelski will be generals without armies.’

There are many situations in which instead of providing complete translations of the ST, the Polish articles in Forum merely invoke certain references to the original content. The only things taken from the ST are some facts described in the translators’ own way. Fourteen such cases were found in the corpus, including for instance the following (with such references shown in boldface): (13)

“Meanwhile the Government won sweeping powers to push through its economic reforms”, rendered as: “zanim wáadze uznaáy, Īe muszą, lub zanim miaáy czas zadziaáaü” (which may be backtranslated as: ‘before the authorities realised that they had to act, or before they had time to act.’)

In certain other examples where a closer connection between source and target texts does obtain, the translations can nevertheless still not be claimed to be formally equivalent (but must instead be recognized as functional equivalents). There were 10 such cases found in the corpus, e.g.:

The Translator’s Intentions

(14)

(15)

(16)

145

“…the trial of the alleged killers of Father Popieluszko would begin next month. He said he was determined to expose all the facts of the case”, translated as: “Generaá odpowiedziaá na wiele pytaĔ dotyczących m.in. morderstwa Popieáuszki. Zapowiedziaá, Īe proces trzech funkcjonariuszy sáuĪby bezpieczeĔstwa oskarĪonych o to morderstwo zacznie siĊ w grudniu”, “Discussing Poland’s future, General Jaruzelski said that the country was ready ‘to resume its traditional role in international diplomacy’”, the published Polish equivalent of which is: “Rząd Polski przywiązywaá do spotkania w Jabáonnie pewną wagĊ, poniewaĪ to tak szerokie miĊdzynarodowe zgromadzenie umocniáo Īywione przez PolskĊ ambicje powrotu do jej tradycyjnej roli miĊdzynarodowej”, “Referring to contacts the government has had with Poland’s Primate, Cardinal Jozef Glemp, General Jaruzelski said: ‘I assess them very positively and I appreciate them very highly’”, in the TL version translated as: “…i poáoĪyá nacisk na sprawĊ wspóápracy z KoĞcioáem, którego prymas jest «realistą i postĊpuje wáaĞciwie»”.

The attenuation of the ST content is also quite common.  ormally translators use a different grammar form – there are for example changes of mood, or some milder words to express the ST meanings. Seven interesting cases of this implicit manipulative strategy were found in the corpus, e.g. (17)

(18) (19)

(20)

(21)

“increased foreign exchange earnings from these sources would have to be diverted to”, translated as “wpáywy dewizowe z tego tytuáu są przeznaczane na” (modality changed into declarative mood), “this may be possible” ĺ “jest to moĪliwe” (modality changed into declarative mood), “the Soviet press has been extremely critical”, with its target equivalent: “radziecka prasa bardzo krytycznie wypowiada siĊ” (“extremely” translated as “bardzo” – ‘very’), “the Polish regime senior sources here believe”, rendered as: “w koáach póáoficjalnych panuje przekonanie” (‘quasi-official circles are certain’), or “the delay in carrying out target programmes”, translated as: “opóĨnienia w realizacji waĪnych programów” (“target” becomes “waĪny” – ‘important’).

146

Chapter Seven

Also noteworthy are cases of reordering (5 observed), where some changes in the paragraph or sentence organization result in an implicit manipulative effect on the reader. Hiding more important pieces of information at the end of a sentence seems to be quite a widespread strategy, as in the following examples: (22)

(23)

(24)

“The use of force in Poland, as western leaders are now pointing out, would destroy this hope”, translated as: “Jak podkreĞlają obecnie przywódcy Zachodu, uĪycie siáy w Polsce przekreĞlaáoby tĊ nadziejĊ”, “In a co-ordinated move, President Gustav Husak and President Nicolae Ceausescu, the Czechoslovak and Romanian party leaders, spoke out publicly in favour of a summit at the end of Mr Ceausescu’s three-day official visit to Czechoslovakia last week”, the Polish version of which is: “Pod koniec trzydniowej oficjalnej wizyty w Czechosáowacji prezydenta Rumunii Nicolae Ceausescu zarówno on, jak i prezydent Gustav Husak wypowiedzieli siĊ publicznie za zwoáaniem konferencji na szczycie”, “An end to NATO sanctions against Poland is in sight now”, from the very beginning of the article the part is moved to the paragraph’s end in the translation: “Teraz, kiedy prymas Polski w komunikacie wydanym wspólnie z generaáem Jaruzelskim uznaá publicznie, Īe sytuacja ustabilizowaáa siĊ i Īe utorowana zostaáa droga do wizyty papieĪa w czerwcu bieĪącego roku, pojawiáa siĊ perspektywa zniesienia przez NATO sankcji wobec Polski”.

Substitutions are another strategy used to avoid certain content inconvenient for propaganda reasons. Three such cases have been identified, in which translators replaced some information they did not decide to include in their translations in favour of some other facts. For example: (25)

“Whether Mr Suslov came here to prevent personnel changes at the top, demanded by the rank-and-file party members, is hard to tell. Despite the pressures from below, the leadership is seeking a solution which would allow it to enlarge the Politburo without giving in to demands for the removal of some members opposed to reforms”, replaced by “Pod tym wzglĊdem jest rzeczą znamienną, Īe proces oĞmiu dysydentów – przebywających w wiĊzieniu od wrzeĞnia ubiegáego roku – zostaá odáoĪony, a dwie osoby spoĞród zatrzymanych – uwolniono” (“The crucial thing in this respect is that the trial of eight dissidents, imprisoned since September last

The Translator’s Intentions

147

year, was postponed and two people among the ones in custody were released”). Many equivalents used seem to be of a manipulative character and more of the implicit than explicit kind, as for instance in the following examples: (26) (27) (28)

(29) (30)

“clear clues” is not the same as “jasnej oceny”, “to calm Soviet anxieties” does not exactly mean “uspokoiü stronĊ radziecką”, “Agonised reluctance to get involved, presumably inspired by the Soviet Union, is still apparent” is not quite “Widaü nadal wyraĨne wahania, inspirowane prawdopodobnie przez ZSRR, przed zaangaĪowaniem siĊ”, “and ambitions to gain control of nuclear weapons” does not exactly mean “i dąĪeniom do uzyskania broni nuklearnej”, “It has an immediate interest” is not really “Zachód jest przede wszystkim zainteresowany”,

Very often the equivalents chosen do not change the meaning of the ST, but are meant to serve propaganda purposes, such as by means of manipulative vocabulary or expressions: niepokoje robotnicze (‘unrest among workers’), wielka rewolta (‘great revolt’). There are also situations in which the strategies used in Polish translations do not directly serve the purpose of manipulation understood as ideological distortion, though they might be seen as such. There are individual cases of: (31)

(32)

(33) (34)

exaggerations: “and will probably be accompanied by”, which is a weaker assertion than “Najprawdopodobniej konferencja ta wyda teĪ zalecenie”, elisions: “in the Polish Communist Party”, rendered as: “wewnątrz partii”, which avoids describing the party’s character, or takes for granted that there can only be one party, superordinates: “both the Polish people and Moscow” ĺ “narodu polskiego i ZSRR”, transpositions: “a discreet but important adaptation” ĺ “dyskretnie, ale w sposób nie pozbawiony znaczenia dostosowali siĊ”, which stresses the action, or “The declarations they signed” ĺ “Podpisane deklaracje” – a diverted case (with a clause changed into a participial construction),

148

(35)

Chapter Seven

transformations (tense): “but the latest statements by Mr Husak and Mr Ceausescu indicate that the next meeting will involve party chiefs as well as Prime Ministers”, shifted into the past: “ale z tych najnowszych oĞwiadczeĔ Husaka i Ceausescu wynikaáo, Īe w najbliĪszej sesji wezmą udziaá – oprócz premierów – równieĪ przywódcy partii”.

Examples of non-manipulative strategies include explicitations, such as (36) (37)

(38)

“would begin next month” translated as: “zacznie siĊ w grudniu” (‘would begin in December’), “the trial of the alleged killers” ĺ “proces trzech funkcjonariuszy sáuĪby bezpieczeĔstwa oskarĪonych o to morderstwo” (‘the trial of the three Security Service officers charged with this murder’), “The 35th Comecon Council is likely to be held next month”, the Polish equivalent of which is: “35 sesja tej Rady ma siĊ odbyü w czerwcu” (‘The 35th Comecon Council is meant to be held in June’).

7. Conclusions The results of the corpus study reported here indicate that manipulation in the translation of the body of articles examined is not (or is only rarely) a simple kind of handling necessitated by the purposes of information exchange, but rather a conscious and planned process intended to serve the requirements of the censorship authorities. In most cases the translations are rife with distortions – ranging from some minor changes concerning the choice of equivalents for particular words or expressions up to substantial alterations consisting in omissions of crucial, but inconvenient information, manipulative additions or substitutions and all types of hiding the message expressed in the ST. The range of manipulative effects found to be applied in the translations examined here goes even as far as complete reformulations of the TT content, whereby the connection with the ST can often only be recognized based on certain corresponding pieces of information, whilst the rest represents the translator’s personal idea for how to express his own views on the ST content. Indeed, the degree of modification goes so far that the ST’s connection with its ‘translation’ can sometimes only be recognized by means of the ST publication date. Given these findings, we can conclude that the communicative and informative intentions of the authors of the articles and the translators very

The Translator’s Intentions

149

often (or even as a rule) do not coincide. Even if translators produce equivalent translations which in general could be considered genuine reproductions of the ST content in the TL, it is enough to spot one or a few cases of purposeful omissions, additions or substitutions to be able to say that the translator’s intention differs from the author’s and that a manipulative intention is incorporated. This is especially because the translations for Forum do cite their source articles – at least the titles of the original magazines or newspapers plus publication dates, sometimes also followed by the authors’ names. All the modified content is then ascribed to them, portraying them as being responsible for what the translators, whose names are not given, decided to include in the translated versions. At the same time, the cognitive effects intended by the translators are not the same as those intended by the original authors.

References Blass, Regina. 2005. Manipulation in the speeches and writings of Hitler and the NSDAP from a relevance theoretic point of view. In Manipulation and Ideologies in the twentieth century. Discourse. language, mind, edited by L. de Saussure and P. Schulz. Amsterdam and Philadelphia: John Benjamins. Bassnett, Susan and André Lefevere. 1990. Translation, history and culture. London and New York: Pinter Publishers. Catford, John C. 1965. A linguistic theory of translation: An essay in applied linguistics. Oxford: Oxford University Press. Dukate, Aiga. 2007. Manipulation as a specific phenomenon in translation and interpreting. (18 March 2012) —. 2009. Translation, manipulation and interpreting. Frankfurt am Mein: Peter Lang. Elias, Urszula. 2011. The family – manipulated manipulation manual. Censorship in modern Poland as exemplified by Mario Puzo’s last novel. In Manipulation in translation. Theory and application, edited by J. Esquibel and M. Kizeweter. Warszawa: Wydawnictwo Szkoáy WyĪszej Psychologii Spoáecznej. Galasinski, Dariusz. 2000. The language of deception: A discourse analytical study. London: Sage. Gutt, Ernst-August. 1990. A theoretical account of translation – without a translation theory. Target: International Journal of Translation Studies 2, no. 2: 135-164. Kizeweter, Magdalena. 2011. (Un)intentional manipulation in literary translation – an attempt at redefining the role of manipulation in

150

Chapter Seven

translation. In Manipulation in translation. Theory and application, edited by J. Esquibel and M. Kizeweter. Warszawa: Wydawnictwo Szkoáy WyĪszej Psychologii Spoáecznej. Kramina, Aiga. 2004. Translation as manipulation: causes and consequences, opinions and attitudes. In Studies about languages 6. (7 July, 2012) Puzynina, Janina. 1992. JĊzyk w Ğwiecie wartoĞci. Warszawa: PaĔstwowe Wydawnictwo Naukowe. Romek, Zbigniew. 2000. Káopoty z cenzurą. Kilka refleksji zamiast wstĊpu. In Cenzura w PRL: relacje historyków, edited by Z. Romek. Warszawa: Neriton, Instytut Historii PAN. Salich, Hanna. 2011. Manipulation in translation: manipulation or creativity. In Manipulation in translation. Theory and application, edited by J. Esquibel and M. Kizeweter. Warszawa: Wydawnictwo Szkoáy WyĪszej Psychologii Spoáecznej. Saussure, Louis de. 2005. Manipulation and cognitive pragmatics. In Manipulation and ideologies in the twentieth century. Discourse, language, mind, edited by L. de Saussure, P. Schulz. Amsterdam and Philadelphia: John Benjamins. Saussure, Louis de, and Peter Schulz 2005. Manipulation and ideologies in the twentieth century. Discourse, language, mind. Amsterdam and Philadelphia: John Benjamins. Szarkowska, Agnieszka. 2011. Manipulation in audiovisual translation: at the crosswords of language, culture and politics. In Manipulation in translation. Theory and application, edited by J. Esquibel and M. Kizeweter. Warszawa: Wydawnictwo Szkoáy WyĪszej Psychologii Spoáecznej. Sperber, Dan and Deirdre Wilson. [1986] 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. Taillard, Marie-Odile. 2000. Persuasive communication: The case of marketing. UCL Working Papers in Linguistics 12: 145-174.

PART III: FIGURES OF SPEECH IN LITERARY DISCOURSE

CHAPTER EIGHT TOWARDS A RELEVANCE-THEORETIC ACCOUNTOF ALLEGORY CHRISTOPH UNGER, SIL INTERNATIONAL AND JOHANNES GUTENBERG UNIVERSITY MAINZ, GERMANY

1. Introduction Wilson (2012) draws attention to the phenomenon of allegory, a narrative device where in talking about an object or event the communicator is really talking about another. Classical rhetoricians following Quintilian define it as saying one thing and meaning another. Examples of allegory include the following: (1) (2)

When you walk through a storm, hold your head up high (Wilson 2012) O ship, shall new waves bear thee back into the sea? O what art thou doing? Make resolutely for the harbor.1 (Horace, quoted by Quintilian in Butler (1922, 8.6.44))

Allegory often involves extended metaphor, as in (2). This, and the closeness of the classical definition of allegory to that of metaphor give rise to the idea that allegory may work like metaphor. For example, Gibbs (2011) treats both with the same analytical tools. And while Crisp (2008) argues that there is a qualitative difference between the two notions, he still argues that it is essentially the mechanisms of domain mapping that lie behind allegory as well as metaphor.

1

“O navis, referent in mare te novi fluctus; o quid agis? fortiter occupa portum.”

Towards a Relevance-Theoretic Account of Allegory

153

However, allegory does not always involve metaphor, as Quintilian shows with example (3) from Cicero. Quintilian claims that this example is unmixed with metaphor (Butler 1922, 8.6.47): (3)

What I marvel at and complain of is this, that there should exist any man so set on destroying his enemy as to scuttle the ship on which he himself is sailing.2

And even where allegory is built on (extended) metaphor, there is a crucial difference that virtually all modern scholars working on allegory point out: in allegory, both the ‘literal’ and the ‘allegorical’ meaning are transparent to the audience, so that the audience is consciously aware of them both. Kurz (1997, 37) expresses this concisely when he writes: “Allegory has a double meaning, but metaphor has not.”3 For similar comments, see Crisp (2008) and Werth (1994). Consequently, Wilson (2012) suggests analyzing allegory and extended metaphor differently from metaphor. In a nutshell, she suggests that allegory achieves relevance in the same way as fiction: allegory portrays a fictional situation that is relevant as an (often experiential) example for the idea that the speaker wants to communicate. For example, in (1) going through a storm with the head up high is an example for the way the speaker believes that one should live through difficult times. At this point Wilson (2012) raises the question: how do examples work? How do we find out what the cognitive mechanisms are that are involved in understanding examples? It seems that a crucial element in an explanatory account of allegory is missing. However, Unger (2006) discusses in a fairly detailed way an example of an allegorical text. The example follows the general pattern suggested by Wilson (2012). This analysis suggests that Relevance Theory already has explanatory accounts of all the cognitive mechanisms involved in the interpretation of allegories. My strategy in this paper is to start from this assumption, argue for its truth, and see how convincing this argument will turn out to be. The result will not be a comprehensive theory of allegory; certainly, much more needs to be said about allegory from a relevancetheoretic perspective than I can say here. This paper is organized as follows: section 2 surveys important approaches to allegory in literary theory and in pragmatics in order to pinpoint the central research questions to be addressed in this paper. In 2

“Hoc miror, hoc queror, quemquam hominem ita pessumdare alterum velle, ut etiam navem perforet id qua ipse naviget.” 3 “Die Allegorie ist doppeldeutig, die Metapher nicht.”

154

Chapter Eight

section 3 I outline the main tenets of a relevance-theoretic analysis of a complex allegorical text that was provided in Unger (2006). Section 4 argues that the allegory analysis in section 3 follows a pattern that is generalizable to all instances of allegory, and in section 5 I discuss other examples of allegory by applying this pattern. In section 6 I discuss how the proposed relevance-theoretic account of allegory relates to the treatment of extended metaphor, and especially to the dual-processing route hypothesis of Carston (2010) and Carston and Wearing (2011).

2. Allegory in literary theory and pragmatics In this section I present a brief survey of accounts of allegory in the rhetorical tradition, literary studies and pragmatic theory with the aim to identify the main research questions that an adequate pragmatic account of allegory must answer. For fuller surveys of the rhetorical and literary studies literature, see for example Fletcher (1964); Quilligan (1992) or Boys-Stones (2003).

2.1 The notion of allegory in the rhetorical tradition and literary studies The standard definition of allegory in the rhetorical tradition goes back to Quintilian (Butler 1922, 8.6.44): Allegory, which is translated in Latin by inversio, either presents one thing in words and another in meaning, or else something absolutely opposed to the meaning of the words.4

Here, allegory is understood as a trope involving substitution of thought, where what the speaker intends to convey differs from what the sentence means. This definition is reminiscent to that of metaphor, but metaphor is understood as a trope involving substitution of words, where the word employed is used in place of another one. As Kurz (1997) points out, this definition attempts to understand allegory in terms of metaphor. Notice in particular the suggestion that allegory mostly involves extended metaphors such as in example (2). Notice also that Quintilian’s definition includes irony as a subtype of allegory.

4

“Allegoria, quam inversionem interpretantur, aut aliud verbis aliud sensu ostendit aut etiam interim contrarium.”

Towards a Relevance-Theoretic Account of Allegory

155

The difference between allegory and metaphor in Quintilian’s definition consists in the entities which enter into the figurative expression: whereas metaphor is claimed to be a figure of speech where a word is used ‘improperly’ for another one, allegory is a figure of speech involving a substitution of the whole text. This distinction is based on the so-called substitution theory of metaphor (Kurz 1997, 7-11). As is well known, the substitution theory of metaphor is nowadays widely discarded in pragmatic theories as well as in literary studies. Kurz (1997), for instance, adopts what he calls an ‘interaction theory’ of metaphor which holds that metaphorical meaning is actively created in interaction with the context. Virtually all modern cognitive theories of metaphor, most notably those of Relevance Theory and cognitive linguistics, would count as instances of interaction theories of metaphor in this sense. This raises the question whether a distinction between allegory and metaphor can be drawn at all once the substitution theory of metaphor is rejected, and if so, how. Kurz (1997) meets this challenge by focusing on another salient property of allegory: that it communicates a double meaning. Allegory communicates the literal descriptive meaning (which Kurz calls the ‘initial meaning’) by way of a direct speech act, and simultaneously an allegorical meaning by way of an indirect speech act. These two levels of meaning of an allegory are coherent within themselves and are interconnected. Metaphor, in contrast, creates an identification between the vehicle and the topic of the metaphor. For example, in (4) the speaker conveys the idea that her lawyer has some properties that can also be attributed to sharks. (4)

My lawyer is a shark.

No such identification occurs in allegory. Thus, in (2), the point is not merely that properties of a ship drifting away from the harbour can be attributed also to the state; rather, the author encourages the reader to understand what is said about the ship as something that is said about the state. In short: allegory communicates a double meaning in the sense that both the literal meaning (‘initial meaning’) and the allegorical meaning are transparent to the audience, whereas metaphor does not communicate its literal meaning, only its metaphorical one. Quintilian noticed this property of allegory as well, in 9.2.46. Moreover, he also acknowledges cases of allegory that do not involve metaphor, such as (3). Thus, Quintilian provides two ways to characterize allegory: first, a figure that says one thing in words and another in sense, and second, a figure that simultaneously communicates a literal description and an allegorical meaning. This property of allegory is independent of

156

Chapter Eight

metaphor theories. So even if the substitution theory of metaphor is rejected, it is possible to distinguish between allegory and metaphor. This suggests that there may be a substantial difference between metaphor and allegory, not merely a theory-internal one that collapses when the definition of metaphor is changed.

2.2 Allegory in pragmatics The central concern of modern pragmatic theory is to explain how the gap between linguistic meaning and speaker's meaning is bridged. The fact that speakers often use figurative expressions rather than literal ones constitutes a major challenge for the development of an adequate pragmatic theory. For this reason, pragmaticists have studied figures of speech such as metaphor, irony and metonymy intensively. Much of this work converges on the idea that the rhetorical figures distinguished in the fine-grained taxonomies of the rhetoricians can ultimately be explained by principles of comprehension at work in processing literal expressions, metaphor and metonymy. This is perhaps most radically worked out in the work of Dan Sperber and Deirdre Wilson, who argue that metaphor and irony are based on pragmatic processes that are at work independently in the interpretation of everyday speech, and are not at all specific to tropes (Sperber and Wilson 1995, 1990, 2008; Wilson and Sperber 2012). Against this background it is not surprising that allegory has not so far been widely studied in pragmatics as a topic of enquiry of its own. An exception to this trend is the work of Peter Crisp (2001, 2005, 2008). In the wake of this research, other cognitive linguists have turned to study allegory as well. Much of this work has been published in a special issue of the journal Metaphor and Symbol (Harris and Tolmie 2011; Gibbs 2011; Thagard 2011; Kasten and Gruenler 2011; Oakley and Crisp 2011), and related resources are available on the website of the cognitive allegory workshop (Cognitive allegory workshop, 2009). In what follows I survey some of the most fundamental work in this framework. Crisp (2001, 2008) claims that allegory is based on the same cognitive processes and structures as metaphor: it involves a cognitive process of metaphorical mapping between a literal source domain and a figurative target domain. In allegory, these metaphors are ‘super-extended’ in the sense that the language used relates only to the source domain, whereas in metaphor, both expressions belonging to the source and target domains are used. This is also true of extended metaphor, which is “a linguistic metaphor extending over more than one clause whose language relates directly to both the metaphorical source and target” (Crisp 2008, 291).

Towards a Relevance-Theoretic Account of Allegory

157

Both allegory and extended metaphor crucially involve metaphorical mappings. But there are both quantitative and qualitative differences. From a quantitative perspective, allegory involves more elaborate metaphorical mappings. From a qualitative perspective, allegory and extended metaphor involve different types of mental representations with different logical properties. Allegories describe fictional situations, that is, fragments of possible worlds. Allegorical language refers to entities in fictional (that is, possible) situations. Extended metaphors, on the other hand, create blended mental spaces in the sense of Fauconnier and Turner (2002). A blended mental space is a unique conception of entities that can function as a means to refer to certain entities in the world and in doing so understand this entity in a new way. Crisp (2008, 292) provides the following example: “When Charles Causley in ‘A Ballad For Katherine Of Aragon’, in a linguistic metaphor extended over four lines and clauses, elaborates the idea of war as a casual mistress, the reader consciously experiences a strange, seemingly impossible, fusion of war and mistress, a metaphorical blend.” Whereas Crisp (2008) focuses on investigating the relation between allegory and extended metaphor, Thagard (2011) investigates how the analogical processes in allegories (and metaphor) work. He proposes that there are three constraints that direct the analogical mappings in allegory and metaphor: the similarity constraint, the structure constraint and the purpose constraint. The similarity constraint which says that source domain concepts may be mapped to target domain concepts that are similar in meaning or appearance. The structure constraint operates on relations that hold within the source domain and maps them to corresponding relations in the target domain. Finally, the purpose constraint holds that all analogical mappings made must serve the overall cognitive function or purpose that the allegory is intended to achieve, for example to inspire the audience, to teach, to explain how to avoid past mistakes, or to entertain. The purpose of allegories is often achieved by arousing emotional reactions in the audience. Emotions can be analyzed as appraisals, that is, representations of propositional attitudes, or as physical states. To achieve an integrated theory of emotions, Thagard proposes to understand emotions as brain processes. Conceptual representations, too, can be understood as brain processes, that is, as firing patterns of neurons that are caused by some object or group of objects in the world. Thagard’s (2011) main claim is that allegories are successful to the extent that the similarity constraints are satisfied and the emotions aroused by analogical structures in source and target contribute to fulfill the allegory’s purpose. He

158

Chapter Eight

supports this claim by showing how these conditions are fulfilled in Orwell’s Animal Farm (Orwell 2008), which is known to be a successful work of allegoric literature, but violated in José Saramago’s Blindness (Saramago 1999), which in Thagard’s opinion is not a successful work of allegoric literature. While Crisp and Thagard are mainly concerned with allegory as a trope, Gibbs (2011) deals with allegoresis, what he dubs “the allegorical impulse” and which is a desire to “...continually seek to connect, in diverse ways, the immediate here and now with more abstract, enduring symbolic themes” (Gibbs 2011, 122). He observes that processing allegory is different from processing metaphorical language. Allegory involves “metaphor processing as a more general strategy” (122). This is a general mode of understanding that can be applied to utterances even when they do not involve “easily identifiable instances of metaphorical language” or situations. Evidence for the empirical impulse is fourfold: first, people see allegorical meanings in poems that do not involve metaphorical language. Second, people see allegorical meanings in other non-metaphorical texts such as satire. Third, people may see allegorical meaning where none is intended. Finally, engaging with allegorical meaning is a cross-cultural phenomenon and seems to be a universal activity humans engage in. The allegorical impulse indicates that allegorical interpretation is an aspect of unconscious cognition. There are at least two ways to understand this: First, when we hear allegorical speech, we realize that the literal description activates cognitive metaphors such as LIFE IS A JOURNEY. This allows us to make metaphorical mappings between the source situation and the target domains given by this metaphorical structure. A second way to understand the cognitive nature of the allegorical impulse is this: when we hear allegorical speech, we imagine engaging in an activity literally described by the utterance and experience what it feels like and what thoughts it may arouse in us. This means, we engage in an embodied simulation of the eventuality or situation described. This experience makes us realize which source-to-target domain mappings may be made. Gibbs prefers this second way to understand the cognitive nature of allegory. As has been mentioned above, little work has been done on allegory in other pragmatic frameworks. Specifically, there is no published study in Relevance Theory directly dealing with allegory as such. However, Wilson (2012) draws attention to issues raised by the phenomenon of allegory. First, she points out that the relevance-theoretic account of metaphor does not seem to work well for examples such as (2) and (3). The relevance-theoretic account of metaphor argues that metaphors, like other cases of loose uses of language, communicate concepts that are not

Towards a Relevance-Theoretic Account of Allegory

159

encoded in the word used, and metaphor comprehension involves a process of ad hoc concept construction constrained by the Communicative Principle of Relevance (see Wilson and Carston 2007 and Sperber and Wilson 2008 for exposition). As a result, metaphors do not communicate thoughts about the source domain. For example, in (4), no thought about SHARKS are conveyed, only thoughts about SHARKS*, which refer to a certain kind of aggressive lawyers. In order to account for cases such as (2), this account would have to be extended in order to explain how the utterance could convey not only thoughts about a SHIP* conceptualizing the institution of the state, but also thoughts about a SHIP that is drifting back into the tossing sea. Moreover, the existence of non-metaphorical allegories such as (3) seriously calls into question the assumption that metaphorical processes should lie at the heart of allegory. While it may be possible to extend the ad hoc concept account to explain allegories, Wilson (2012) suggests that another approach may be more promising, and this is to understand allegories in the same way as fiction. According to Sperber and Wilson (1987), fiction involves layered communication acts: in the simplest case, one layer where thoughts about the fictional world are conveyed, and another one where the author conveys thoughts about the real world to the reader. Relevance is optimized at each of these layers, and Wilson (2012) speaks of internal and external relevance. It is easy to see how allegories may be regarded as mini-fictions that describe a situation exemplifying an abstract thought. For example, in (1), the speaker portrays the fictional situation where the addressee walks through a storm in an upright position as an example of the way she recommends her addressee to deal with difficult situations in life. At this point the question arises how the addressee works out the example: what are the thoughts that the literal meaning of an allegory exemplifies? Wilson (2012) points out that we often work out such examples quickly and reliably, but what are the processes involved? She leaves this question as a topic for further research. There are at least two other relevance-theoretic publications that are particularly relevant for investigating allegories, although they do not address the phenomenon of allegory as such: Unger (2006) and Carston (2010). Unger (2006) offers a detailed analysis of a literary text using the multi-layered approach of Sperber and Wilson (1987). This text happens to be an allegory, but because Unger (2006) focused on understanding issues of genre knowledge and global coherence in discourse, it does not specifically develop an account of allegory. I will discuss this work in more detail below, where I use it as the basis for an account of allegory in relevance-theoretic terms. Carston (2010) similarly deals with many

160

Chapter Eight

examples that appear to be allegories while her concern is primarily to expand the relevance-theoretic account of metaphors to cover extended metaphors and artistic metaphors where there do seem to be robust intuitions of continued awareness of the literal meaning of the metaphor even after resolving the metaphorical meaning. I will discuss this work, too, in more detail below.

2.3 Central questions for pragmatic accounts of allegory As this short survey of literary and pragmatic theories of allegory reveals, a running theme in theoretical accounts of allegory has been its relation to metaphor. From very early on, the close relation of allegory to metaphor has been fundamental to descriptions of this phenomenon, and many modern pragmatic accounts of allegory unreservedly assume that metaphorical processes are at the heart of allegory (e.g. Crisp 2005, 2008; Thagard 2011; Gibbs 2011). This does not mean that the analysts see no difference between allegory and metaphor, as Crisp’s (2008) investigation of qualitative difference between these two phenomena show. Nor is it the case that the existence of non-metaphorical allegories is ignored. Gibbs (2011, 121), for instance, writes: Allegory is a cognitive action in which people apply a metaphoric mode of understanding to situations and discourse that typically does not contain metaphorical language per se.

However, this raises the question how do people intuitively know when to apply a metaphorical mode of understanding to literal discourse and when not. Consider an example discussed by Gibbs, the poem “The Road not Taken” by Robert Frost (Frost 1969, quoted from Gibbs (2011, 123-124)): (5)

Two roads diverged in a yellow wood, And sorry I could not travel both And be one traveller, long I stood And looked down one as far as I could To where it bent in the undergrowth; Then took the other, as just as fair, And having perhaps the better claim, Because it was grassy and wanted wear; Though as for that, the passing there Had worn them really about the same,

Towards a Relevance-Theoretic Account of Allegory

161

And both that morning equally lay In leaves no step had trodden black. Oh, I kept the first for another day! Yet knowing how way leads on to way, I doubted if I should ever come back. I shall be telling this with a sigh Somewhere ages and ages hence; Two roads diverged in a wood, and I – I took the one less traveled by, And that has made all the difference. Gibbs comments that although this poem does not use metaphorical language per se, it seems to be clearly structured by the cognitive metaphor LIFE IS A JOURNEY. So perhaps we apply metaphorical mode of understanding when the discourse we process triggers cognitive metaphors deeply entrenched in our mind. But this cannot be the answer. Consider the case of two academics meeting at a conference and telling each other their stories of delayed planes and missed flight connections on their way to the conference venue. These narratives should also activate the cognitive metaphor LIFE IS A JOURNEY, among others. But intuitively, we do not apply a metaphorical mode of understanding to process such narratives. It appears that holding on to the assumption that allegory interpretation is based on metaphor comprehension processes is problematic in the face of non-metaphorical allegories. But the relation of allegory to metaphor is by no means the only salient property of the phenomenon. Another property recognized early on in the history of theoretical accounts of allegory relates to the intuition that in allegory a double meaning is conveyed in a way that is not conveyed in metaphor. Some analysts such as Kurz (1997) appear to treat this as the main property of allegory. Perhaps a more promising approach to allegory is to focus on the cognitive mechanisms that give rise to the intuition of the double layering of meaning and not assume that these mechanisms are related to metaphorical processes. Wilson’s (2012) remarks point in this direction and I will follow them in the account of allegory I propose in this paper. In any case, these considerations show that the relation of allegory to metaphor and metaphorical processes is an important topic of research.

162

Chapter Eight

3. A complex two level allegory In Unger (2006), I provided a detailed relevance-theoretic analysis of a rather complex allegorical text: Isaiah 5:1-7 of the Hebrew Bible, commonly known as ‘the song of the vineyard’. (6)

(1a) I will sing to my friend/lover a song of my friend to his vineyard. (1b) My friend had a vineyard on a fruitful hilltop (2a) and he dug it up and removed its stones and planted the best species of grapes in it (2b) and he built a tower in its midst, and also a winevat he dug into it. (2c) And he waited for the vineyard to yield good grapes, but it yielded wild (bitter) ones. (3) And now, inhabitants of Jerusalem and people of Judah, judge between me and my vineyard. (4a) What else was there to do for my vineyard which I haven't done, (4b) why did I expect good grapes and it brought forth sour ones? (5a) And now I will let you know what I will do to my vineyard. (5b) I will remove its hedge and it will become firewood. (5c) I will tear down its walls and it will become a trampled land. (6a) I will bring about its peril, it will not be pruned and not be dug, and in it shall grow thorns and thornbushes. (6b) And I will command the clouds to not let rain fall on it (or: . . . the clouds to refrain from letting it rain over it) (7) For the house Israel is the vineyard of the Lord of hosts, and the people of Judah are the planting of his delight, and he waited for justice, but there was only injustice (or bloodshed), for righteousness, but there were only cries. (Translation CU, quoted from Unger 2006, 123)

This text involves a two level allegory (see table 1 for an overview). A first level allegorical interpretation is shaped by conventional mappings used in love songs, where the lover is often portrayed as a Gardener and the (female) beloved as the vineyard or garden. On the first level, the gardener represents the lover, the vineyard the beloved. Several statements convey the idea that the gardener works hard and well in the vineyard; these statements correspond with the idea that the lover cares well for his beloved. But the gardener find sour grapes – the lover’s expectations are frustrated, and the gardener stops caring for the vineyard and destroys it can be allegorically understood as the lover withdrawing his friendship from his beloved and trying to punish her. However, in the context of the book of Isaiah as part of the Hebrew Scriptures, another level of allegorical meaning opens up: the lover who works hard and well in his vineyard represents God, who cares diligently

Towards a Relevance-Theoretic Account of Allegory

163

for his beloved, that is, the people of Israel. However, despite his good care, the beloved – the people of Judah – bring forth only sour grapes: godless behaviour, breaking the law, the covenant of God. As a result, God will punish the people of Israel as the Gardener has ‘punished’ his vineyard: by burning its hedges (letting the countryside be invaded by a foreign army that uses burning down villages and cities as a method of warfare), tearing down its walls (allowing the fortified cities in the country be captured by the foreign army) and withdraw any care. .

base level

first level allegory

second level allegory

gardener

lover

God

vineyard

beloved

people/nation of Judah

gardener works hard and lover cares well for well beloved

God provides land, state and protection

gardener expects good fruit

lover expects rewarding relationship

God expects Judah to keep Mosaic covenant

gardener find sour grapes

lover’s expectation frustrated

God sees Judah do injustice

gardener stops caring for the vineyard and destroys it

lover withdraws friendship and punishes the beloved

God allows enemies to destroy Judah by war

Table 1. Layers of allegory in the Song of the Vineyard How does the audience manage to comprehend this complex allegory? Unger (2006) provides a detailed analysis of the comprehension processes involved with an emphasis on showing how genre knowledge of Hebrew love songs and the writer’s strategies of fine-tuning relevance expectations induced in the reader facilitate comprehension of the text as a coherent whole. Here, I will only summarize those aspects of this analysis that are directly relevant to the comprehension of the allegorical meanings involved. First of all, the stimulus (the text) communicates the presumption of its own optimal relevance: (7)

This text is optimally relevant to the audience.

In order to test this claim, very early on in the interpretation process, this general presumption of optimal relevance must be inferentially

164

Chapter Eight

enriched into a concrete expectation of relevance for the audience. In my book I showed how such concrete relevance expectations could be raised in parallel with the task of resolving reference for the first person reference in I will sing to my beloved...: the specific word used for ‘friend’ in the phrase a song of my friend to his vineyard is one that is preferentially used to denote the male partner in a male-female relationship. This suggests that the first person reference in I will sing... should be resolved to that one of a fictitious female speaker and not to the prophet Isaiah. Moreover, mentioning the term song and vineyard gives access to the encyclopaedic knowledge that vineyard or garden imagery is conventionally used in love songs to speak figuratively about the beloved persons and their relationship. This easily accessible cultural knowledge (actually, genre knowledge in this case) makes it possible for the audience to infer expectations about what the text may convey, or in other words, more specific expectations of relevance: (8)

The fictitious female speaker intends to describe her relationship with her friend.

However, the text that follows does not talk about a love relationship, but about a gardener and a vineyard. This does not satisfy the specific relevance expectation created. This failure to satisfy relevance expectations is manifest, even mutually manifest. It follows that the very fact that the fictitious speaker produces an ostensive stimulus that fails to satisfy relevance expectations is in itself an ostensive stimulus. Thus, a second layer of communication is entered, and the expectation of relevance created at this level is (9)

The fictitious speaker intends the audience to find out in which way talk about vineyards may resemble things that could be said about the love relationship between the speaker and her friend.

This concrete expectation triggers the search for interpretive resemblance relations between what could be inferred by the description of the vineyard scenes and what one may want to say about the state of a love relationship. Once sufficient resemblances of this kind are found, relevance expectations are met. Recognizing such resemblances will make the audience understand in which way the allegorical situation may be an example for the reality that the speaker intends to talk about. Cultural knowledge about conventional imagery in love songs will help save processing effort for this process.

Towards a Relevance-Theoretic Account of Allegory

165

However, it is the prophet Isaiah speaking, and this raises the expectation that what he says is optimally relevant. It is obvious that when Isaiah quotes a fictitious female speaker singing a love song that this will not in itself satisfy the audience’s relevance expectations. One would expect a quite different message from a man that is known to talk about political and military doom and God’s punishment for the people of Judah. Thus, we have the situation that Isaiah is producing an ostensive stimulus whose straightforward interpretation does not manifestly satisfy the audience’s relevance expectations. In fact, this discrepancy is so great that it must be mutually relevant to speaker (Isaiah) and audience (the people) that this does not satisfy the relevance presumption carried by the prophet’s speech. This can only be explained by assuming that the speaker producing an ostensive stimulus that does not fully satisfy overall relevance presumptions is in itself an ostensive stimulus: by saying something unexpected, the speaker intends the audience to interpret what has been said in a deeper, possibly symbolic or allegorical, way. But in which direction should such an allegorical interpretation be sought? Easily available cultural assumptions about the people of Judah having a special relationship with God are easily accessible, and since it is manifest that a love relationship is also a special one, it is easy to see how the general presumption of relevance raised by this layer of communication together with these easily accessible contextual assumption will now raise a specific expectation of relevance: (10)

The third layer communication is relevant in terms of what the love song may reveal about God’s relationship to Judah

Again, this specific expectation of relevance (cognitive effects) triggers a search for interpretive resemblances between what could be inferred from what the fictitious speaker intended to convey about her relationship with her friend and what Isaiah may want to say about God's relationship to the people of Judah.

4. A general pattern for the interpretation of allegory Unger (2006) continues to specify in more detail how the interpretation process may unfold. For our present purposes this is not necessary. What is important to notice is the general pattern of the interpretation: first, specific expectations of relevance are raised so that the interpretation will follow an effect-based strategy, i.e. the interpretation looks to process contextual assumptions, inferential enrichments and implicatures that

166

Chapter Eight

support cognitive effects with the expected content. A first intuitive interpretation will make sense but fail to satisfy the expectations of relevance, and this failure is mutually manifest. The fact that an ostensive stimulus has been made that mutually manifestly does not satisfy relevance expectations is itself treated as an ostensive stimulus. Consequently, a second layer of communication is entered. The relevance expectations of this layer are satisfied by recognizing interpretive resemblances between inferences from the first communication layer and ideas of the sort that satisfy the concrete expectations about the content of cognitive effects. That ostensive communication may involve several layers of communication, and that overall expectations of relevance may be satisfied on several layers of communication was pointed out by Sperber and Wilson (1987, 751). What this layering means is that the ostensive stimulus produced (the text) is passed through the comprehension procedure twice, or even several times. Mercier and Sperber (2009) have recently looked at intuitive and reflective reasoning processes. They argue that reasoning processes that have a decidedly reflective, conscious, feel are actually reasoning processes where the argumentation module is passed through several times. While the processes of the argumentation module are intuitive (unconscious), we may become consciously aware of the output. Now when a reasoning process requires several passes through the argumentation module, then many intermediate outputs (steps in the reasoning process) may become conscious to us. I suggest that similar considerations apply to the comprehension module. The comprehension module’s process is intuitive. However, multi-layered ostensive communication involves at least two passes through the comprehension module, as described above. As a result, there is at least one intermediate output (the interpretation result of the first communication layer) that may be conscious to us. I propose that this is what explains the intuition that in allegory, the meaning of the source situation and the allegorical interpretation are both transparent, and even conscious to the audience. And the more layers of communication are involved, the more passes through the comprehension procedure are made, the more the interpretation will get a conscious feel.

5. Applying the pattern to other cases It seems to me that the general pattern of allegorical interpretation outlined above is applicable also to simple cases such as in (11) below, or (1) above:

Towards a Relevance-Theoretic Account of Allegory

(11)

167

You can’t put the toothpaste back in the tube. (Wilson, 2012)

When I say (11) to someone who has realized he has carelessly set a false rumor in motion and is worried about what he has done, then my utterance obviously fails to satisfy relevance expectations. For it will be expected that what I say will touch upon the other person’s worries, but instead I talk about toothpaste. However, this failure to satisfy relevance expectations must be mutually manifest, and my addressee will assume that my producing an utterance that mutually manifestly fails to satisfy relevance expectations will be treated as an ostensive stimulus in turn. Hence, the utterance passes through the comprehension process once again, as outlined above. This time the expectation is raised that one can find resemblances between what could be inferred from the situation where toothpaste is pushed out of the tube and someone tries to push it back to the sort of advice I might give to the person worried over his carelessness in spreading rumors. Here, only two layers of communication are involved, and so there is only one intermediate step in the comprehension process. This may account for the intuition that this example is simpler, more intuitive than the elaborate allegory of the song of the vineyard. Moreover, in these conventionalized sayings a process of pragmatic routinization of the associated interpretations has likely taken place, which will further reduce interpretation effort.

6. Remarks on metaphor, extended metaphor and allegory The suggested Relevance Theory account of allegory does not directly involve lexical pragmatic process in metaphor comprehension. Of course, this does not mean that such processes may not be involved in allegory. In fact, it seems plausible that since allegory interpretation crucially involves a strongly effect-based processing strategy guided by strong, determinate expectations about cognitive effects, it follows that if metaphors are involved in the allegory, the metaphorical interpretations will be strongly guided by these guiding expectations about the content of cognitive effects on the allegorical layer of communication. This will predict that extended metaphor will exploit a certain metaphor in related, similar directions in a subsequent sentence, or that the text will contain similar metaphors leading to related interpretations. Example (2) is a case in point, where the SHIP* concept (a ship that can be addressed as a person, and that can play a role in politics) sets the stage for the use of other metaphors from the domain of sea-faring: SEA* and HARBOR*.

168

Chapter Eight

But not all extended metaphors appear to be triggered by an effectbased strategy. Wilson and Sperber (2012, 121-122) discuss the poem “Fog” by Carl Sandburg (1992): (12)

the fog comes on little cat feet. It sits looking over harbour and city on silent haunches and then moves on.

While the title of the poem and the first line definitely raise the expectation that the following lines will be relevant by way of conveying some ideas about fog, these expectations are by no means definite in the sense that the content-based expectations in the allegory cases discussed above were. So while these expectations have the power to enable the audience construct an ad hoc concept ON-LITTLE-CAT-FEET* that denotes a property shared by the movement of fog and the movement of cats, it is not strong enough to trigger a multi-layered analysis as is required by allegories. If these observations are on the right track, then it must be concluded that the phenomenon of extended metaphor is not a unified phenomenon. Rather, some cases of extended metaphor may be embedded in an allegorical interpretation, guided by a strongly effect-based interpretive strategy. Others appear to involve a more effort-based strategy. Only the first type of cases is closely linked to allegory (and perhaps indistinguishable from it), but not the second one. Carston (2010) and Carston and Wearing (2011) argue for a different account of extended metaphor in Relevance Theory. These authors draw attention to extended metaphors such as (13) and (14), and literary metaphors with a strong imagistic feel such as (15) and (16), and argue that the standard relevance-theoretic account of metaphor as a special case of loose talk does not fully account for their properties. (13)

Life’s but a walking shadow, a poor player That struts and frets his hour upon the stage And then is heard no more: it is a tale Told by an idiot, full of sound and fury, Signifying nothing. (Shakespeare, Macbeth, V. v. 24-30, quoted from Carston (2010, 306))

Towards a Relevance-Theoretic Account of Allegory

169

(14)

Depression, in Karla’s experience, was a dull, inert thing–a toad that squatted wetly on your head until it finally gathered the energy to slither off. The unhappiness she had been living with for the last ten days was a quite different creature. It was frantic and aggressive. It had fists and fangs and hobnailed boots. It didn’t sit, it assailed. It hurt her. In the mornings, it slapped her so hard in the face that she reeled as she walked to the bathroom. (Zoë Heller, The Believers, 2008, 263, quoted from Carston (2010, 307))

(15)

Love is the lighthouse and the rescued mariners. (Oskar Daviþo, “Hana,” 1979, quoted from Carston (2010, 295))

(16)

My Life had stood – a Loaded Gun – In Corners – (Emily Dickinson 1863, in Franklin 2005, quoted from Carston 2010, 309)

Following the standard relevance-theoretic account of metaphor, virtually every word after the first one in extended metaphors such as (13) should trigger major contextual adjustments. Carston (2010) and Carston and Wearing (2011) argue that this would incur gratuitous processing effort. Moreover, they point out that recent studies of online interpretation of metaphors reveal that the literal meaning of metaphors remains accessible for a rather long period of time (Rubio Fernández 2007). This seems to be true especially of extended metaphors and novel metaphors with an imagistic feel to them such as (15) and (16). To explain these properties of metaphor, Carston and Wearing argue that there are two routes for the interpretation of metaphor: the first one involves the adjustment of ad hoc concepts in the search for a relevant interpretation. The second one is triggered when the first route would incur too much processing effort for contextual concept adjustment and leads to a literal interpretation of expressions used which is then metarepresented. This metarepresentation triggers slower, more conscious inference processes that establish the intended resemblance relations between the metarepresented literal expressions and the thoughts that the communicator intends to convey by means of these expressions. The main aim of Carston (2010) and Carston and Wearing (2011) is to account for the intuition that the literal meaning of metaphors plays a larger role in some metaphors and especially in extended metaphors than recognized in the standard Relevance Theory account of metaphor. My

170

Chapter Eight

account of allegory is similarly concerned about explaining the intuition that allegory communicates a “double meaning,” where the literal meaning is quite transparent to the audience. However, these respective accounts do so in significantly different ways: whereas Carston (2010) and Carston and Wearing (2011) account for the relative transparency of the literal meaning of metaphors by recourse to the notion of metarepresentation, my account developed here accounts for this phenomenon by appeal to the presence of several layers of communication. This difference may be important to account for the differences that do remain between metaphor and allegory with respect to the role (and perhaps degree) of the transparency of literal meaning. Thus, while Kurz (1997, 17-18) insists that the audience remains aware of the literal meaning of metaphors, he also points out that allegory communicates a double-layered meaning that metaphor does not: “metaphor fuses two meanings into one, allegory puts them side by side.”5 Carston and Wearing (2011) point out that they do not intend their account to cover allegory. A crucial difference between extended metaphor and allegory is that the latter, but not usually the former, achieves a certain ‘internal relevance’ in the sense that the literal description is coherent to the point that one could enjoy this description alone. In extended metaphor, on the other hand, the literal meaning and metaphorical meaning are often interwoven: for example, poor player in (13) has both a metaphorical meaning (denoting an attitude description that is attributed to “Life”) and a literal one on which it functions as the subject of the phrase ...that struts and frets his hour upon the stage and then is heard no more. Seen in this light, Carston and Wearing’s second route for metaphor interpretation and my multiple-communication-layers-based account of allegory complement each other by offering different explanations for the transparency of the literal in figurative utterances, highlighting different roles that the literal meaning plays in allegory and some types of metaphor respectively. However, adopting both accounts side by side also raises questions. Consider the on-line interpretation of (13) at the point when the audience has processed the opening words Life is... The expectation is raised that whatever follows will contribute to a description what life is in the character's opinion. Obviously, the literal expressions that follow do not satisfy these expectations. According to my account of allegory processing, this should trigger another pass through the comprehension procedure, another layer of communication raising its own expectations of relevance. In other words, the question arises whether the proposed 5

“Die Metapher verschmilzt zwei Bedeutungen zu einer, die Allegorie hält sie nebeneinander.” Kurz attributes this observation to Dumarsais and Fontanier (1818).

Towards a Relevance-Theoretic Account of Allegory

171

account of allegory comprehension might not also account for the processing of extended metaphors such as (13). Similar remarks can be made about the on-line processing of (14): after processing the sentence Depression, in Karla’s experience, was a..., the expectation is raised that the following will be relevant in describing Karla’s perception of depression. The literal descriptions that follow obviously do not satisfy this expectation – unless this failure is taken as an ostensive stimulus in its own right and another pass through the comprehension procedure is triggered. This raises several questions: can examples such as (13) and (14) be processed in various different ways? Or is Carston and Wearing’s second route for metaphor interpretation more restricted in its application, so that it applies mostly to cases such as (15) or (16)? Is the account of allegory processing developed in this paper too general? There is no space here to address these issues and I leave these questions open for future research.

7. Conclusion I have argued that several cases of allegory can be accounted for in Relevance Theory without recourse to additional theoretical machinery. The interpretation of allegories does involve a characteristic pattern of processing for relevance: allegories occur in communication spots where strong, specific expectations of the content of cognitive effects are raised. These expectations are blatantly not satisfied by the ostensive stimulus produced, and this is mutually manifest. Hence this failure is another ostensive stimulus and a second pass through the comprehension procedure is required. But relevance expectations are finally met by finding resemblance relations between what can be inferred from the source situation and hypotheses about which ideas the communicator would need to convey to satisfy the relevance expectations. This account breaks with the view that allegory processing is so closely related to metaphor processing that any account of allegory must be based on metaphorical processes. Rather, the transparency of the literal meaning in allegory is taken to be the primary distinctive property whose explanation is foundational for understanding allegory. The allegory interpretation pattern described here is fully inferential in nature. This presents a real alternative to ‘cognitive linguistic’ approaches such as that of Gibbs (2011) who argues that embodied cognitive simulation process may underlie both allegory and metaphor comprehension, or that of Crisp (2008) which relies on the notion of conceptual blending. Further research will need to be done to determine

172

Chapter Eight

the adequacy of my inferential account of allegory as opposed to Gibb’s and Crisp’s non-inferential accounts. The inference route for allegory interpretation discussed can be seen as complementing Carston’s (2010) and Carston and Wearing’s (2011) account of the processing of extended metaphors. However, it also raises interesting questions for Carston and Wearing’s account.

Acknowledgements I warmly thank Robyn Carston and Deirdre Wilson for very helpful comments and discussions on an earlier version of this paper.

References Boys-Stones, G. R. (ed.) 2003. Metaphor, allegory, and the classical tradition: Ancient thought and modern revisions. Oxford: Oxford University Press. Butler, Harold Edgeworth. 1922. Quintilian. With an English translation. Cambridge, Mass.: Harvard University Press. Carston, Robyn. 2010. Metaphor: Ad hoc concepts, literal meaning and mental images. Proceedings of the Aristotelian Society cx, no. 3: 295321. Carston, Robyn and Catherine Wearing. 2011. Metaphor, hyperbole and simile: A pragmatic approach. Language and Cognition 3, no. 2: 283312. Cognitive allegory workshop. 2009. University of Waterloo, Ontario, Canada. http://www.arts.uwaterloo.ca/~raha/CognitiveAllegory/index.html> (14 December 2012). Crisp, Peter 2001. Allegory: Conceptual metaphor in history. Language and Literature 10, no. 1: 5-19. —. 2005. Allegory and symbol – a fundamental opposition? Language and Literature 14, no. 4: 323-338. —. 2008. Between extended metaphor and allegory: Is blending enough? Language and Literature 17, no. 4: 291-308. Daviþo, Oskar. 1979. Hana. Pesme: Hana. Belgrade: Prosveta. Dumarsais, César Chesnaut and M. Fontanier. 1818. Les Tropes de Dumarsais, avec un commentaire raisonné, par M. Fontanier. Paris: Belin-le-Prieur. Fauconnier, Gilles and Mark Turner. 2002. The way we think: Conceptual blending and the mind’s hidden complexities. New York: Basic Books.

Towards a Relevance-Theoretic Account of Allegory

173

Fletcher, Angus. 1964. Allegory: The theory of a symbolic mode. Ithaca, NY: Cornell University Press. Franklin, R. W., ed. 2005. The poems of Emily Dickinson. Reading edition. Harvard University Press. Frost, Robert. 1969. The poetry of Robert Frost: The collected poems, complete and unabridged, edited by E. C. Lathem. Henry Holt & Co. Gibbs, Raymond W. 2011. The allegorical impulse. Metaphor and Symbol 26, no. 2: 121-130. Harris, Randy Allen and Sarah Tolmie. 2011. Cognitive allegory: An introduction. Metaphor and Symbol 26, no. 2: 109-120. Heller, Zoë. 2008. The believers. Penguin. Kasten, Madeleine and Curtis Gruenler. 2011. The point of the plow: Conceptual integration in the allegory of Langland and Voltaire. Metaphor and Symbol 26, no. 2: 143-151. Kurz, Gerhard. 1997. Metapher, allegorie, symbol. 4th ed. Göttingen: Vandenhoeck und Ruprecht. Mercier, Hugo and Dan Sperber. 2009. Intuitive and reflective inferences. In In two minds: Dual processes and beyond, edited by J. Evans and K. Frankish. Oxford: Oxford University Press, 149-170. Oakley, Todd and Peter Crisp. 2011. Honeymoons and pilgrimages: Conceptual integration and allegory in old and new media. Metaphor and Symbol 26, no. 2: 152-159. Orwell, George. 2008. Animal farm: A fairy story. Penguin. Quilligan, Maureen. 1992. The language of allegory: Defining the genre. Ithaca, NY: Cornell University Press. Rubio Fernández, Paula. 2007. Suppression in metaphor interpretation: Differences between meaning selection and meaning construction. Journal of Semantics 24, no. 4: 345-371. Sandburg, Carl. 1992. Chicago poems. University of Illinois Press. Saramago, José. 1999. Blindness, translated by G. Pontiero. Harvest Books. Sperber, Dan and Deirdre Wilson. 1987. Presumptions of relevance. Behavioral and Brain Sciences 10, no. 4: 736-754. —. 1990. Rhetoric and relevance. In The ends of rhetoric: History, theory, practice, edited by D. Wellbery and J. Bender. Stanford: Stanford University Press, 140-156. —. 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. —. 2008. A deflationary account of metaphor. In The Cambridge handbook of metaphor and thought, edited by R. W. Gibbs. Cambridge: Cambridge University Press, 84-105.

174

Chapter Eight

Thagard, Paul. 2011. The brain is wider than the sky: Analogy, emotion, and allegory. Metaphor and Symbol 26, no. 2: 131-142. Unger, Christoph. 2006. Genre, relevance and global coherence: The pragmatics of discourse type. Basingstoke: Palgrave Macmillan. Werth, Paul. 1994. Extended metaphor – a text-world account. Language and Literature 3, no. 2: 79-103. Wilson, Deirdre. 2012. Relevance and the interpretation of literary works. Plenary lecture presented at the EPICS V conference, Sevilla, Spain. Wilson, Deirdre and Robyn Carston. 2006. Metaphor, relevance and the ‘emergent property’ issue. Mind and Language 21: 404-433. —. 2007. A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In Pragmatics, edited by N. Burton-Roberts. Basingstoke: Palgrave Macmillan, 230-259. Wilson, Deirdre and Dan Sperber. 2012. Meaning and relevance. Cambridge: Cambridge University Press.

CHAPTER NINE CODE AND INFERENCE IN THE EXPRESSION OF IRONY IN ORWELL’S ANIMAL FARM AND ITS TRANSLATION INTO SPANISH MARIA ANGELES RUIZ MONEVA, UNIVERSITY OF ZARAGOZA, SPAIN

1. Introduction Orwell’s Animal Farm has been understood as a political fable or allegory, which mirrors the Russian Revolution of the early twentieth century, in 1917, and the subsequent disillusion undergone by Orwell with the unfolding of events and also with the utopia of Communism that was to become a dystopia for the British author. It may be argued that one of the most important aspects of irony and satire in the work has to do with the turning down or disruption of expectations. This squares with the most recent approach to irony pointed at by Wilson and Sperber: “The most common use of irony is to point out that situations, events or performances do not live up to some norm-based expectation” (2012, 127). Indeed, one of the most significant aspects of Animal Farm has to do with the progressive breaking and overturning of the expectations that the animals might have harboured in the revolution pioneered by the pigs. This is certainly reflected in some of the best-known mottos of the revolution–such as the famous “All animals are equal but some animals are more equal than others”; or “Four legs good, two legs better” (1985, 114), but it a1so underlies many of the aspects of irony of the work, as we shall illustrate below. But first of all, we shall deal with the relevance-theoretic proposals on irony, particularly, the most recent views. Next, we shall focus on our proposals for the translation of irony from a relevance-theoretic standpoint, before turning to the analysis of the translation of irony in Orwell’s Animal Farm into Spanish.

176

Chapter Nine

1.1 The relevance-theoretic approach to irony The relevance-theoretic proposals on irony go back to Sperber and Wilson (1978, 1981), that is, they are previous to the explicit and detailed formulation of Relevance Theory as a whole. At the time, these authors already rejected the traditional rhetorical dichotomy between literal and figurative meaning, and also the view that irony was necessarily supposed to imply meaning either the opposite of or different from what has been said. Sperber and Wilson argued that irony consists in the speaker’s echoing or distancing herself from a thought attributed either to individuals or to people in general, so that she was therefore expressing a mocking, sceptical, critical or derogative attitude to such a proposition or thought. Irony was understood in terms of attitude; concretely, it is defined as communicating the speaker’s attitude either towards the proposition or towards those that might have held it. The theory was experimentally tested by Jorgensen, Miller and Sperber (1984). The proposals offered in Relevance: Communication and Cognition (Sperber and Wilson [1986] 1995) and also in their 1992 paper “On verbal irony” allow us to establish a certain connection between the relevancetheoretic approaches to irony and to translation. Thus, in Relevance irony is seen as a form of second-degree interpretation, when the speaker’s thought interpreted by the utterance is seen in itself as an interpretation ([1986] 1995, 238). Echoic utterances – and therefore, irony – achieve relevance by expressing the speaker’s attitude of dissociation towards the proposition expressed by the utterance. The analysis of irony within this framework is recast by Wilson and Sperber (1992), where the distinction between use and mention made in their 1981 paper is seen as a concrete instance of a more general notion of interpretive resemblance. Then they claim that irony may be approached as a particular instance of echoic interpretations of an attributed thought or utterance. In other words, irony becomes a variety of echoic interpretive use. In Relevance, a distinction had been drawn between descriptive and interpretive uses of language, so that whereas a descriptively used representation represents a certain state of affairs, an interpretively-used one represents another representation which it resembles in content, be it logical, semantic or conceptual. In turn, Blakemore (1992) draws on the indeterminacy of ironic utterances, which had not been catered for by traditional rhetorical approaches. She also rejects the traditional analysis of irony as an opposition between what is said and what is meant, insofar as it fails to account for many instances of irony. Irony is then seen as an interpretation

Code and Inference in the Expression of Irony

177

of a thought attributed to somebody other than the speaker at the time of the utterance. The debate on the necessarily echoic nature of verbal irony has also been recurrent within Relevance Theory. Thus, Sperber and Wilson (1998) reaffirm its necessarily echoic character, and draw a distinction between ostensive and non-ostensive forms of irony, which is based on the corresponding types of communication. If irony is ostensive, therefore, it exploits an echoic interpretive mechanism, in contrast to non-ostensive forms of irony (1998, 291). Research on irony within this theoretical framework has continued in very recent times. Wilson (2006) examines two pragmatic approaches to irony, namely, pretence (initially proposed by Clark and Gerrig 1984) and introduces the analysis of the relevance-theoretic account which approaches irony as an echoic use of language. Only the latter caters for essential aspects of irony, and can also account for irony in comprehensive terms, whereas pretence irony can only explain those varieties that involve some kind of impersonation. An essential trait of irony is that it involves “the attribution of a thought, a propositional or conceptual content or a meaning” (2006, 1740). Wilson (2009) retakes the contrasts between pretence and echoic approaches to irony, and draws on irony as a form of metarepresentation. A metaprepresentation (Wilson, 1997) has been defined as a representation of a representation, which, therefore, as Deirdre Wilson notes, “involves a higher-order representation with a lower-order representation embedded in it” (1997, 130). Wilson (2009), and also Curcó (2000), have claimed that the interpretation of irony requires higher orders of metarepresentation than other stylistic features, such as metaphor. Irony is defined as being tacitly attributive: that is, “the speaker is attributing a thought to some source other than herself at the current time” (Wilson 2009, 218). Yus Ramos (1997-98, 2000, 2000-01, 2009, 2012) has also offered valuable contributions with a view to coping with the analysis of irony within the relevancetheoretic framework, particularly as regards the role of context in the processing of these utterances. First of all, Yus (1997-98) had approached the hotly debated issue in pragmatic theory of whether the processing of ironical utterances (and ‘figurative’ language in general) was supposed to take longer than that of their literal counterparts. In Yus’s view, this must mostly depend on the context in which the utterance was to be processed: “... It all depends on how optimal is the listener’s activation of contextual sources, and how many simultaneous incompatibilities between these sources and the speaker’s ironic utterance are detected” (1997-1998, 396).

178

Chapter Nine

In his most recent papers on irony, Yus (2012, 2009) undertakes a thorough revision of the hypotheses put forward in previous investigations. Following in this respect other authors (such as Attardo 2001, 2000), it is claimed that it is contextual inappropriateness that triggers or raises the ironic interpretation of an utterance. Thus, irony entails the activation of certain contextual information, or certain contextual sources defined as inappropriate. These contextual sources may be of any of the following seven types (Yus, 2012): A) general encyclopaedic knowledge; B) specific encyclopaedic knowledge concerning the speaker (likes, dislikes, habits, beliefs, opinions, and the like); C) knowledge, still stored in the addressee's short term memory, of events or actions which have taken place only very recently; D) previous utterances or assumptions inferred from them; E) speaker’s non-verbal communication, be it vocal or visual; F) lexical or grammatical choices made by the speaker which may function as linguistic cues about the speaker's ironic intention; and, finally, G) that information springing from the physical area where the communicative interchange takes place. These contextual sources may be activated by the addressee, either simultaneously or sequentially, when he is interpreting the utterance. This gives way to what Yus has termed as contextual saturation, which basically means that the addressee may obtain information deriving from several of these contextual sources. Yus has also contributed other aspects to the relevance-theoretic approach to irony. Thus, in his view, the processing of irony will be governed by a balance between context accessibility and processing effort. Even though in his earlier proposals Yus had argued that irony mainly springs from a certain clash or incompatibility between some of these contextual sources,1 in Yus (2009, 320) it is seen that irony may be identified through access to the contextual information from just one of these contextual sources. Be it as it may, the thing is that in the earlier models proposed by this author, a distinction had already been drawn between the different contextual sources: these could be, on the one hand, the so-called leading contextual source, which is the one whose information is more accessible than the other possible contextual sources, and which may prove to be essential for the addressee's appropriate interpretation of the ironical utterance; on the other 1

This is Yus’s definition of incompatibility: “Throughout this article the term incompatibility will be used to refer to the clash arising between the proposition expressed by the utterance and the information conveyed by single or multiple contextual sources, a clash which triggers the identification of the echoic mention and the speaker’s attitude of dissociation, essential for an optimal access to an implicit ironic interpretation” (2000, 24).

Code and Inference in the Expression of Irony

179

hand, the supportive contextual sources, which tend to strengthen the evidence provided by the main source. In this paper, Yus (2009) re-examines the criterion of optimal accessibility to irony.2 The aspects that have been revised are the following. To begin with, the addressee need not activate the contextual sources simultaneously, but may do so sequentially. Second, it is argued that if the contextual information is highly accessible or saturated, it may not be necessary to identify the entirety of the de-contextualised logical form of the utterance. For these reasons, a new formulation of the Criterion is proposed: CRITERION OF OPTIMAL ACCESSIBILITY TO IRONY The processing effort required by the interpretation of the ironic meaning decreases in so far as the number and quality of contextual incompatibilities increase. These incompatibilities (traced by the addressee) are given, on the one hand, between the information contributed by the inferential integration of contextual sources activated (either sequentially or simultaneously), and on the other hand, the information supplied by the utterance, regardless of whether it has been processed totally or only partially. (2009, 322, my translation)3

This new formulation of the criterion of optimal accessibility to irony has certain consequences. First, it may not be necessary for the addresser to supply a whole utterance, so that the addressee may have access to the information springing from the different contextual sources and may infer 2

In earlier formulations, this criterion read as follows: CRITERION OF OPTIMAL ACCESSIBILITY TO IRONY: “The processing effort required for the interpretation of the intended ironic meaning of an utterance decreases in proportion to the increase in the number (and quality) of incompatibilities (detected by the addressee) between the information supplied by the inferential integration of simultaneously activated contextual sources (leading or leading plus supportive) and the information provided by the proposition expressed by the utterance” (Yus 2000, 54). 3 What follows stands for Yus’s (2009) original Spanish formulation of the CRITERION OF OPTIMAL ACCESSIBILITY TO IRONY – CRITERIO DE ACCESIBILIDAD ÓPTIMA A LA IRONÍA (VERSIÓN 2): “El esfuerzo de procesamiento que exige la interpretación del significado irónico decrece en proporción al aumento en el número y calidad de las incompatibilidades (detectadas por el destinatario) entre la información aportada por la integración inferencial de fuentes contextuales activadas (en secuencia o de forma simultánea) y la información aportada por el enunciado, haya sido éste parcial o totalmente procesado” (2009, 322).

180

Chapter Nine

the speaker’s underlying ironical intention. Secondly, therefore, the addressee may anticipate the speaker’s communicative intention before the whole of the utterance has been produced, and accordingly, irony may be spotted by him even before the grammatical structure of the utterance has been completely identified. Then, it may not be necessary for him to identify the logical form of the utterance. A third consequence springing from the new reformulation of the criterion of optimal accessibility is that it may not be necessary to completely process the explicit information of the utterance for the addressee to have access to the ironical information implicitly conveyed. This may be particularly the case if the addressee is able to activate a number of contextual sources, and there is therefore an important degree of saturation. Fourthly, and in contrast to the third case, Yus claims that if the supporting contextual evidence is poor, the addressee may have to process the whole of the explicit information of the utterance, and test different hypotheses until he may reasonably suppose that his interpretation of the utterance matches the meaning intended to be communicated by the speaker. The point arising from this discussion and the presentation of the former different cases to be found in the processing of an ironical utterance is that, contrarily to what had been maintained by Grice, the interpretation of the ironical utterance may not necessarily rely upon the previous processing of the literal utterance. Accordingly, ironical utterances may not necessarily take any longer or demand any further processing effort than the processing of their ‘literal’ counterparts. Likewise, in certain contexts, both the literal and the ironical interpretation of an utterance may be equally accessible and valid. This means that the speaker may be ironic and yet agree with the propositional content of her utterance. Therefore, this challenges the traditional view of irony as meaning either the opposite or something different from what has been said. An approach to irony in terms of attitude – as held by Relevance Theory, and indeed, most pragmatic and even rhetorical approaches nowadays – results in a much more appropriate way to cope with the complexity of irony. Still a fifth possible case to be found in the interpretation of ironical utterances is the situation that gives way to misunderstanding: that is, the addressee fails to have access to the context envisaged by the speaker and therefore is unable to identify her communicative intention. Whichever the case, the success or otherwise failure in the speaker’s communication of her ironical utterance will depend – as in any other communicative

Code and Inference in the Expression of Irony

181

situation – on her adequate assessment of the contextual resources available to the addressee. In the statement of their new proposals on irony from the relevancetheoretic framework and thus reformulating their previous views on the topic, Wilson and Sperber (2012) reject the traditional and Gricean assumption that irony should necessarily take any longer to process than their literal counterparts. More interestingly, they also point at what they regard as three puzzling features of irony: first, the fact that irony is characterised by a certain attitude – whereas, in contrast, this is not the case for metaphors; second, the fact that irony tends to show a certain breaking of expectations, and in what they term as a normative bias, or what may be regarded as an asymmetry in its use, irony shows a different behaviour depending on whether positive or negative comments are involved: thus, in the case of the latter, it appears that further or extra contextual information must be supplied; and thirdly, the so-called ironical tone of voice, the reasons for which are left unexplained by either traditional, Gricean or pretense theories. After reaffirming the echoic nature of irony, as a subtype of attributive use, and after examining alternative accounts of irony within pragmatics (most importantly, the pretense account, and several hybrid attributivepretence models), Wilson and Sperber set out to provide a relevancetheoretic explanation for the features of irony put forward above. Regarding the first aspect, namely, the connection between irony and attitude (so that the ironical attitude is seen as fully constitutive of it), their claims made in former works are confirmed: namely, it is stated that irony communicates the speaker’s attitude of dissociation towards her utterance, which she regards as ludicrously false, inappropriate, uninformative or irrelevant. As regards the second aspect, the so-called normative bias, that is, the fact that irony is most often used to blame or complain, rather than to praise, it is claimed by Wilson and Sperber that there are, on the one hand, aspects such as norms that are always available for becoming sources of echo; on the other hand, there are aspects – basically, negative utterances of the kind “What horrible weather!” or “She is so very impolite” – where, as Wilson and Sperber note, the thought must have been entertained or even expressed: “Only then is there some identifiable thought that can be ironically echoed” (2012, 142). With reference to the third aspect, the so-called ironical tone of voice – which has no counterpart in the case of metaphorical utterances, and which, in Wilson and Sperber’s view, is left unexplained by both traditional and Gricean accounts, these authors claim that it has to do with

182

Chapter Nine

“the tone of voice optionally used to convey the attitudes characteristic of irony” (2012, 143), and contrarily to what is claimed by pretence accounts, it has nothing to do with an exaggeration or an imitation of the person that the ironist might have pretended to be. The latter, for Wilson and Sperber, has to do with parody rather than with irony. In contrast, the tone characteristic of irony would be one of contempt, and, in any case, “the tone of voice expresses the relationship of the ironist to the thought she is echoing” (2012, 144). Wilson (2013) points at the most significant features of irony – namely, the expression of attitude, the normative bias or asymmetry in the use of irony (that is, the fact that it is negative irony that is most often used, in order to criticise or complain, rather than to praise) – and contrasts the ways how both the echoic and the pretense accounts of irony explain these features. These features also allow Wilson to conclude that it is necessary to distinguish irony from other rhetorical figures, such as hyperbole, jocularity, understatement or rhetorical questions.

1.2 A relevance-theoretic approach to the translation of irony In Ruiz Moneva (2001), I focused on the problems that the translation of irony might have raised, and put forward certain proposals on the basis of the application of the relevance-theoretic framework. As for the problems, they were traced to be the following: first, the possible translatability of irony; second, the importance of context, understood, within the relevancetheoretic framework, in cognitive terms, and bound to be modified as the communicative process unfolds; third, the different, specific and linguistically and culturally based (pragmatic) connotations that may be expressed in the source and in the target language to refer to what may be taken to be (semiotic) universal meanings; or fourth, the problems created by the different levels of explicitation that may be found in either the source or the target text. I then proposed the following possible solutions: first, the communication and translation of irony is expected to follow a balance between processing efforts and cognitive effects; second, it is worth exploring the relationship between code and inference in translation; third, as translation has been defined as ‘interligual interpretive use’ (Gutt 2000a), then the explicatures and the implicatures of the translated or target text can be expected to correspond to those of the source text; fourth, the role of context or cognitive environment, as understood in the relevance-theoretic framework, can be expected to have implications for

Code and Inference in the Expression of Irony

183

translation and for interlingual communication; and fifth, on account of the close connection between communication and the context, and translation being a form of communication, the contextual sources (as defined by Yus 2012, 2009, 2001-01, 1997-98) may be taken as communicative clues. We shall test these assumptions on the basis of a Spanish translation of Orwell’s work (1984). However, within the relevance-theoretic framework, Gutt’s most outstanding proposals (2000, 1991) point at a distinction between two main forms of translation, namely, direct and indirect translation, in analogy with direct and indirect speech quotations. I did not really take into account this distinction in my initial proposals on the translation of irony within the relevance-theoretic framework.4 Direct translation is meant to preserve all the communicative clues and all the stylistic features of the original. It must be noted, however, that for direct translation to be possible and feasible in practice, it is necessary to find a sufficient degree of overlap of the cognitive environments of both the intended audience of the source text, on the one hand, and the audience of the target or translated text. By communicative clues Gutt refers to “a subset of the textual properties that are significant for the intended meaning” (2000b, 153). Therefore, if there is quite a great difference between the cognitive environments of both the source text audience and the target text readership, direct translation may not be adequate. In contrast, indirect translation is approached as the most general case, and it sets out to interpretively resemble the original. It is in fact defined as interlingual interpretive use, which means that the target text shares the explicatures and the implicatures of the original (Gutt 1996). An implication of this approach is that it allows us to assess translations. This may be done on the basis of the comparative or extent-conditions definition of relevance (Guillén 1995-96). Let us note, in any case, that direct and indirect translations have different aims, so that an indirect translation is not intended to provide the TT audience with a ‘faithful’ representation of the original text, as it does not assume that they will be able to follow the inferential paths possibly undertaken by the ST readership. Rather, indirect translations are preferred whenever it may reasonably be assumed that the TT readers may actually be unable to do so, on account of the lack of relevant contextual resources for them. As a consequence, in cases of indirect translation the translator finds it appropriate to facilitate the 4

I wish to thank Profs. Deirdre Wilson and Robyn Carston for their helpful suggestions and comments in this respect, at the Conference Interpreting for Relevance, Discourse and Translation, held in Warsaw, 24-26 September 2012. I also want to express my gratitude to the organizers of this meeting.

184

Chapter Nine

intended target text readership’s access to the intended interpretation of the author, namely, by including contextual information in the text (which might not have been necessary for the original audience), or by making implicatures explicit (as predicted by the application of the relevancetheoretic notion of interpretive resemblance to translation), just to refer to some of the most widely used translational procedures in this case.

2. Analysis of the translation of irony into Spanish in Orwell’s Animal Farm Let us note that Relevance Theory is mostly concerned with a cognitivepragmatic analysis of ostensive, echoic irony within a general account of ostensive-inferential communication as it occurs in conversation. This may be different from the analysis of irony as a literary device (Wilson 2011, 2006). If irony is echoic, as relevance theoreticians claim, an outstanding instance may certainly be said to be Orwell’s Animal Farm, in the sense that events that happen after the revolution breaks out are ostensively distant from what may have been entertained before. Because this is an aspect thoroughly recurrent in the work, I have chosen to focus upon a significant aspect, that is the rewriting and contradiction of the so-called Seven Commandments. A problem that Orwell’s work presents us with is that irony relies heavily upon the words expressed – which will acquire new shades of meaning and will be endowed with senses other than the originally intended ones. Animal Farm is one of the works by Orwell in which he expresses his disillusion with the outcome of the Russian Revolution, and it may be read as a satire of some of the events that occurred after 1917. For the purposes of the analysis of irony in the work, this entails that the narrator may be said to express the author’s attitude towards the events that are allegorically reflected in this political fable. As a matter of fact, in my view, the work may be said to exemplify the ways in which the code model of communication both needs to be enriched by and is complementary to the ostensive-inferential model, as put forward by Sperber and Wilson ([1986] 1995). Much of the meanings which become important in the work are encoded, as this is a fundamental process in the institutionalisation of the new society of the animals, the work being a sort of political fable. This leads us to question how far the translation of the work may rely upon interpretive resemblance, which, as had been formulated by Gutt (1996), may have entailed the explicitation of implied meanings. However,

Code and Inference in the Expression of Irony

185

in the case of irony, there seems to be a basic tension between what is made explicit and what is left for the addressee to infer. Moreover, one of the points of Orwell’s book is precisely the way in which expectations, which rely strongly on the encoding of certain messages, are contradicted. As will be shown in the analysis, the initial and the final situation in each of the cases concerned, and hence irony, depend on the way in which the essential messages are encoded and verbalised. In turn, it seems to be the case that an analysis in terms of Yus’s contextual sources and their incompatibility may provide a more fruitful approach. The Seven Commandments express the process of institutionalisation of the new society that springs up in Animal Farm as a result of the revolution undertaken by the pigs, which are aiming to replace the power exerted over them by the human beings. Such institutionalisation is reflected by the fact that they set out what might have been regarded as Old Major’s wishes for the new society. Even though these assumptions are confirmed in the wording of the Commandments, yet at the same time we find the first aspects that will be contradicted by the future unfolding of events. Thus, Old Major predicts that Jones will kill the old Boxer, as soon as the animal is of no use for him: You, Boxer, the very day that those great muscles of yours lose their power, Jones will sell you to the knacker, who will cut your throat and boil you down for the fox hounds. (From Old Major’s speech; ch. l, p. 10) Tú, Boxer, el mismo día que tus grandes músculos pierdan su fuerza, Jones te venderá al descuartizador, quien te cortará el pescuezo y te cocerá para los perros de caza. (p. 53)

And indeed, the ending of Boxer could not be any crueller, even though it will not be men but rather his old “comrades”, the other pigs, that will eventually take him to the knacker, therefore partially breaking the expectations that might have been entertained. Besides, obviously, such an ending contradicts one of the basic Commandments, “No animal shall kill any other animal”; that is, “Ningún animal matará a otro animal”, where the use of the future form equally maintains the obligatory character of the action expressed in the original text: It had come to his knowledge, he said, that a foolish and wicked rumour had been circulated at the time of Boxer’s removal. Some of the animals had noticed that the van which took Boxer away was marked “Horse Slaughterer,” and had actually jumped to the conclusion that Boxer was being sent to the knacker’s. It was almost unbelievable, said Squealer, that any animal could be so stupid. Surely, he cried indignantly, whisking his

186

Chapter Nine tail and skipping from side to side, surely they knew their beloved Leader, Comrade Napoleon, better than that? But the explanation was really very simple. The van had previously been the property of the knacker, and had been bought by the veterinary surgeon, who had not yet painted the old name out. That was how the mistake had arisen. The animals were enormously relieved to hear this. (ch. IX, p. 106) Había llegado a su conocimiento–dijo– , que un rumor disparatado y malicioso circuló cuando se llevaron a Boxer. Algunos animales notaron que el furgón que trasladó a Boxer llevaba la inscripción: "Matarife de caballos", y sacaron precipitadamente la conclusión de que ése era en realidad el destino de Boxer. Resultaba casi increíble, dijo Squealer, que un animal pudiera ser tan estúpido. Seguramente, gritó indignado, agitando la cola y saltando de lado a lado, seguramente ellos conocían a su querido Líder, camarada Napoleón, mejor que nadie. Pero la explicación, en verdad, era muy sencilla. El furgón fue anteriormente propiedad del descuartizador y había sido comprado por el veterinario, que aún no había borrado el nombre anterior. Así fue como nació el error. Los animales quedaron muy aliviados al escuchar esto. (p. 164)

It is therefore Squealer who once more disseminates the propaganda that is found to be convenient by the established regime. Thus, the true fate of Boxer – which the reader is acquainted with, thanks to the fact that they have a much wider cognitive environment than that which can be accessed by the other animals – has become “a foolish and wicked rumour”, in the voice of the narrator, who is not only superior to the animals in his knowledge of what is actually happening, but who does not hesitate, moreover, to satirize over their limited capacity to grasp what is going on. In this case, the Spanish text has drawn on possible explicatures that can be derived from the English expression jumped to the conclusion, so that the adverb precipitadamente – that is, ‘hastily’ – has been added. The expressions of mood – and therefore, of attitude – employed by Napoleon, surely, are accurately conveyed in the Spanish translation seguramente, and the repetition of this key word in the context is also reproduced. However, in my view, it appears that a more emphatic version would have been preferred for the translation of the narrator’s omniscient reflection on the reaction of the animals towards Napoleon’s explanation: enormously relieved has been down-toned to just muy aliviados. In any case, the irony springs from the clash or incompatibility between the contextual sources of what has actually happened to Boxer, which can in fact be grasped by the external readership, on the one hand, and on the other hand, the sad unawareness among the animals.

Code and Inference in the Expression of Irony

187

This is a Commandment, the sixth one, which will also affect many other animals, when these dare to question and challenge the authority of the leaders, and the latter do not hesitate in killing them. In this case, the new formulation or rewriting of the Commandment contains a new explicature, which was not necessarily derivable from the original: “No animal shall kill any other animal WITHOUT CAUSE”; that is, “Ningún animal matará a otro animal sin motivo”. But the whole context where this commandment is rewritten shows most of the aspects of the uses of irony by Orwell in the book, which in my view maintain the same level of explicitness and of irony in the original text and also in the Spanish translation: A few days later, when the terror caused by the executions had died down, some of the animals remembered–or thought they remembered–that the Sixth Commandment decreed “No animal shall kill any other animal.” And though no one cared to mention it in the hearing of the pigs or the dogs, it was felt that the killings which had taken place did not square with this. Clover asked Benjamin to read her the Sixth Commandment, and when Benjamin, as usual, said that he refused to meddle in such matters, she fetched Muriel. Muriel read the Commandment for her. It ran: “No animal shall kill any other animal WITHOUT CAUSE.” Somehow or other, the last two words had slipped out of the animals’ memory. But they saw now that the Commandment had not been violated; for clearly there was good reason for killing the traitors who had leagued themselves with Snowball. (ch. VIII, p. 78) Días después, cuando ya había desaparecido el terror producido por las ejecuciones, algunos animales recordaron–o creyeron recordar–que el sexto mandamiento decretaba: “Ningún animal matará a otro animal”. Y aunque nadie quiso mencionarlo al oído de los cerdos o de los perros, se tenía la sensación de que las matanzas que habían tenido lugar no concordaban con aquello. Clover pidió a Benjamín que le leyera el sexto mandamiento, y cuando Benjamín, como de costumbre, dijo que se negaba a entrometerse en esos asuntos, se fue en busca de Muriel. Muriel le leyó el Mandamiento. Decía así: “Ningún animal matará a otro animal sin motivo”. Por una razón u otra, las dos últimas palabras se les habían ido de la memoria a los animales. Pero comprobaron que el Mandamiento no fue violado; porque, evidentemente, hubo motivo sobrado para matar a los traidores que se coaligaron con Snowball. (p. 133)

Thus, the omniscient voice of the narrator allows us to cast on the visions of reality that are hidden from most of the animals. Irony is here expressed similarly in the original and in the translation through a comment by the narrator in the form of an understatement: “or thought

188

Chapter Nine

they remembered” / “o creyeron recordar”. In our view, even if understatement has to be kept distinct from irony, as Wilson (2013) has argued, it can be a fruitful way to convey the speaker’s attitude, as it appears to be the case here. In fact, the alternation of impersonal and personal verbal forms is maintained in similar terms in the source and in the target text, when it comes to communicating the vaguer impression that the new wording of the commandment does not correspond to the old one: “it was felt” / “se tenía la sensación de que”. The omniscient narrator provides the interpretation of the situation, caused by the changes undergone in the wording of the commandments: “Somehow or other, the last two words had slipped out of the animals’ memory”, or in the Spanish translation: “Por una razón u otra, las dos últimas palabras se les habían ido de la memoria a los animals”. There is, therefore, a basic contradiction between what had been the shared knowledge of the characters of the book and the current state of affairs. Irony is also present in the interpretation supplied by the narrator, which corresponds to the reading that power in that society would disseminate, the latter being certainly contradicted by the actual situation. It is best interpreted, in my view, as the omniscient narrator’s expression of an attitude of distance, both because he does not share the standpoint adopted by the animals, and also because he has more information available than what the latter are actually able to master: “But they saw now that the Commandment had not been violated”, or in the Spanish edition: “Pero comprobaron que el Mandamiento no fue violado”. This openly contradicts the actual state of affairs, and may ultimately be interpreted as Orwell’s implicit intention to show the inconsistencies of such a society. But all the other Commandments will likewise be disrupted in similar ways, even if they had been intended to “form an unalterable law by which all the animals on Animal Farm must live for ever after” (ch. 2, p. 23); or, analogously, “... formarían una ley inalterable por la cual deberían regirse en adelante, todos los animales de la «Granja Animal»” (p. 68). This is certainly not ironic in itself, but it will indeed be so in the light of the outcome of events, where, as we shall see next, every other Commandment will be contradicted and abandoned. As we can see, once more, the propositional form of the original text is maintained in the translation. Thus, the First Commandment, “Whatever goes upon two legs is an enemy”, or in the Spanish version, “Todo lo que camina sobre dos pies es un enemigo”, expresses the common enmity established between all the

Code and Inference in the Expression of Irony

189

animals of the farm and the human beings. However, this principle will be gradually distorted: first, it seems to have arisen out of pure necessity: WHATEVER GOES UPON TWO LEGS IS AN ENEMY. WHATEVER GOES UPON FOUR LEGS, OR HAS WINGS, IS A FRIEND. One Sunday morning, when the animals assembled to receive their orders, Napoleon announced that he had decided upon a new policy. From now onwards Animal Farm would engage in trade with the neighbouring farms: not, of course, for any commercial purpose, but simply in order to obtain certain materials which were urgently necessary. (ch. VI, p. 56) Un domingo por la mañana, cuando los animales se reunieron para recibir órdenes, Napoleón anunció que había decidido adoptar un nuevo sistema. En adelante, "Granja Animal" iba a negociar con las granjas vecinas; y no por supuesto con algún propósito comercial, sino simplemente con el fin de obtener ciertos materiales que hacían falta con urgencia. (p. 106)

But the thing is certainly that this Commandment – and also the rest – does not seem to have provided for the possibility of any exception whatsoever. And, as time goes by, more resolutions like this are turned down. The omniscience of the narrator leads him to behave as a witness that knows much more than the characters involved, and who is able to metarepresent the innermost and deepest thoughts of these animals: “All the animals remembered passing such resolutions: or at least they thought that they remembered it”; “Todos los animals recordaron haber aprobado tales resoluciones o, por lo menos, creían recordarlo.” In due course, this clash between the perception of reality by the animals – which is always seen as imperfect and partial – on the one hand, and the way matters actually stand or at least the way they evolve, on the other hand, grows stronger. Such opposition is obviously mediated by the omniscient narrator, who is able to see things from above, and therefore, from a very different perspective from that of the animals. This contrasting situation may be lexically enhanced through parallelisms which highlight the formal similarity but the difference between conflicting ideas: “Again the animals seemed to remember that a resolution against this had been passed in the early days, and again Squealer was able to convince them that this was not the case.” A similar parallelism is reflected in the Spanish translation, thus showing what the translator has regarded as the invariant aspects – that is, the aspects to be preserved in the translation, and therefore, leading to relevance: “De nuevo los animals creyeron recordar que ..., y de nuevo Squealer hubo de convencerlos de que no era así.”

190

Chapter Nine

Persuasion also plays an essential role: thus, there are certain characters (mainly, the pigs and the characters that are at their disposal, precisely for purposes of manipulation and propaganda, such as the dogs, or Squealer) who play a major role in manipulating the cognitive environment that is accessed by the rest of the animals. Again, the assumptions entertained in the past fail to be confirmed, insofar as the strength with which the animals have access to them is indeed very poor. This being so, these animals end up by rejecting such assumptions and depriving them of any validity whatsoever: A few animals still felt faintly doubtful, but Squealer asked them shrewdly, “Are you certain that this is not something that you have dreamed, comrades? Have you any record of such a resolution? Is it written down anywhere?” And since it was certainly true that nothing of the kind existed in writing, the animals were satisfied that they had been mistaken (ch. VI, p. 57-58). Algunos animales aún tenían ciertas dudas, pero Squealer les preguntó astutamente: "¿Están seguros de que eso no es algo que han soñado, camaradas? ¿Tienen constancia de tal resolución? ¿Está anotado en alguna parte?". Y puesto que era cierto que nada de eso constaba por escrito, los animales quedaron convencidos de que estaban equivocados. (p. 108)

On the whole, therefore, the main type of contextual source (Yus 2012, 2009) at play in the work has to do with that knowledge which is still stored in the addressee’s short term memory, and the clash or incompatibility established between it, on the one hand, and on the other hand, the unfolding of events. The animals – except for some remarkable exceptions, such as Clover – fail to have any objective knowledge of the past, and therefore, they are easily manipulated. In this way, the omniscient narrator plays on the two perspectives which, in contrast to the animals, he is able to master. In an attitude that even strengthens such superiority, he tends to emphasise the antagonisms underlying the two situations at stake. Again, there is an obvious incompatibility between different contextual sources, namely, between the knowledge stored in the memory of the animals and the present situation. Such knowledge becomes purely subjective, and is contradicted by the current state of affairs, which denies the hypothetical situation of the recent past. There is, therefore, an obvious incompatibility or contradiction between the state of affairs and the way it is perceived:

Code and Inference in the Expression of Irony

191

But a few days later Muriel, reading over the Seven Commandments to herself, noticed that there was yet another of them which the animals had remembered wrong. They had thought the Fifth Commandment “No animal shall drink alcohol,” but there were two words that they had forgotten. Actually the Commandment read: “No animal shall drink alcohol TO EXCESS.” (ch. VIII, p. 93) Pasados unos cuantos días, cuando Muriel estaba leyendo los siete mandamientos, notó que había otro que los animales recordaban malamente. Ellos creían que el Quinto Mandamiento decía: «Ningún animal beberá alcohol», pero pasaron por alto dos palabras. Ahora el Mandamiento indicaba: «Ningún animal beberá alcohol en exceso». (p. 150)

As this quotation illustrates, the aspects that enhance irony – what may be regarded as the leading contextual sources – are even enhanced through typographic devices. It is interesting to note that a mistranslation into Spanish, indeed one that is very frequent among learners of English whose L1 is Spanish – actually translated as ahora – contributes to highlighting the mismatch between the two contexts: on the one hand, that of the early formulation of the Commandment, and on the other hand, the subversion of it by its being rewritten by the pigs who hold power. The subversion of the initially harboured hopes and expectations and also the rewriting of the past even end up by affecting the most basic mottos and indeed aspirations of the revolution, at least in the early days. Thus, the initial slogan, “all animals are equal”, now becomes “all animals are equal but some animals are more equal than others”. This is literally translated into Spanish, so that the target utterance shares all the analytical and contextual implications of the original. The core statement is then conveyed as “Todos los animales son iguales, pero algunos animales son más iguales que otros”. This is just another – but certainly a representative – instance that shows that it has been found important to preserve the literal words, and therefore, the logical and the propositional form of the original. In our view, this has a readily relevance-theoretic explanation: such procedure has enabled the target text readers to achieve similar cognitive and contextual effects, thus reaching similar conclusions to those of the source text readers, without being required to expend any further (but certainly neither any less) processing effort.

192

Chapter Nine

3. Conclusions This analysis has shown that the Spanish translation of irony and satire in works such as Orwell’s Animal Farm shows certain peculiar features. We have seen that despite the fact that it might have been expected to rely on indirect procedures, this was not found to be the case. Over all, the kind of translation employed has tended to be the direct form. Furthermore, the implicatures further made explicit in the target text have been rather scarce. In our view, this can be explained if we take into account that irony in Orwell’s work really relies upon the words expressed, and on the contradiction between the expectations that might have been initially entertained, on the one hand, and the actual outcomes reached towards the end of the work, on the other hand. But above all, the utterances or assumptions in themselves in the initial context are many times not ironic, and it is only through the clash of expectations that irony is to be felt. Moreover, the criterion of accessibility has been found to be important: on the one hand, the different animals – except for remarkable exceptions, such as Clover, and only to a certain extent – cannot access the context of the ironically entertained assumptions; on the other hand, irony is mainly to be traced at the level of communication established between the thirdperson heterodiegetic narrator and the target readership. Therefore, it must have been important to preserve the actual words used. This is so because irony in the work relies on the contradiction between what was pointed out at first, on the one hand, and on the other hand, how the propositions expressed in the end directly subvert these initially harboured expectations and objectives, and thus, deny or contradict those initial propositions. This aspect accounts for the fact that, contrarily to what can otherwise be commonly expected, there has been, in this way, an essential correspondence between the explicit and the implicit contents of irony in both the source and the target texts. This explains, in our view, why the direct forms of translation have been preferred. However, to a certain extent, it may be argued that, because of the fact that the target text has tended not to make any contents further explicit, the overall sum of explicatures and implicatures has been maintained. This would also fulfil the notion of interpretive resemblance in the translation. In my view, this also further means that, as Mateo (2001) had predicted, the notions of descriptive and interpretive forms of resemblance, respectively, are best approached in the form of a continuum. It may also be concluded that irony – as it is conceived of herein – is successful in the Spanish translation basically because it relies on the same kind of contextual sources pointed at by Yus and their corresponding

Code and Inference in the Expression of Irony

193

incompatibilities. In our view, this shows that they may be regarded as valuable tools for the analysis of irony and the assessment of the possible ways in which irony may be conveyed in the target language, so that it may achieve similar effects, without demanding or calling for any extra processing effort. In other words, they may be approached as invariant factors that the translator may analyse and assess when it comes to conveying the target text. They will undoubtedly help him to achieve in the target text the balance between effort and effects, between what is manifested explicitly and what is left implicit, between what is encoded and what is to be inferred, and between what descriptively and what interpretively resembles the original, respectively. These aspects reveal themselves as the key elements that the translator will have to bear in mind when he faces the source text and sets out to convey the target text. He must set out to maintain the same sort of balance between them as the one that is established in the source text. In other words, and just to conclude, the reader of the target text should be able to reach similar contextual and cognitive effects as the reader of the source text, without being required to expend any further processing effort for doing so.

References Attardo, Salvatore. 2000. Irony as relevant inappropriateness. Journal of Pragmatics 32: 793-826. —. 2001. Humorous texts: A semantic and pragmatic analysis. Berlin and New York: Mouton de Gruyter. Blakemore, Diane. 1992. Understanding utterances. London: Blackwell. Clark, Herbert H. and Richard J. Gerrig 1984. On the pretense theory of irony. Journal of Experimental Psychology: General 113, no. 1: 121126. Curcó Cobos, Carmen. 2000. Irony: Negation, echo and metarepresentation. Lingua 110: 257-280. Guillén Galve, Ignacio. 1995-96. Evaluating the appropriateness of a translation. A pragmatic application of Relevance Theory. Pragmalingüística 3-4: 27-51. Gutt, Ernst-August. 1996. Implicit information in literary translation: a relevance-theoretical perspective. Target 8: 239-256. —. 2000a. Translation and relevance: Cognition and context. 2nd ed. Manchester: St. Jerome Publishing. —. 2000b. Textual properties, communicative clues and the translator. In Transcultural communication: Pragmalinguistic aspects, edited by M. P. Navarro Errasti et al. Zaragoza: Anubar, 151-59.

194

Chapter Nine

Jorgensen, Julia, George A. Miller and Dan Sperber. 1984. Test of the mention theory of irony. Journal of Experimental Psychology: General 113, no. 1: 112-120. Mateo Martínez, José. 1998. Be relevant (Relevance, translation and crossculture). Revista Alicantina de Estudios Ingleses 11: 171-182. Orwell, George. [1945] 1985. Animal Farm. Harmondsworth: Penguin. —. 1984. Rebelión en la Granja, translated by R. Abella. Barcelona: Destinolibro. Ruiz Moneva, Maria Angeles. 2001. Searching for some relevance answers to the problems raised by the translation of irony. Revista Alicantina de Estudios Ingleses 14: 213-247. Sperber, Dan. 1984. Verbal irony: Pretense or echoic mention? Journal of Experimental Psychology: General 113: 130-136. Sperber, Dan and Deirdre Wilson. 1978. Les ironies comme mentions. Poétique 36: 399-412. —. 1981. Irony and the use-mention distinction. In Radical pragmatics, edited by P. Cole. New York: Academic Press, 295-318. —. [1986] 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. —. 1998. Irony and relevance: A reply to Seto, Hamamoto and Yamanashi. In Relevance Theory: Applications and implications, edited by R. Carston and S. Uchida. Amsterdam and Philadelphia: John Benjamins, 283-293. Wilson, Deirdre. 1999. Metarepresentation in linguistic communication. UCL Working Papers in Linguistics 11: 127-161. —. 2006. The pragmatics of verbal irony: Echo or pretence? Lingua 116: 1722-1743. —. 2009. Irony and metarepresentation. UCL Working Papers in Linguistics 21: 183-226. —. 2011. Relevance Theory and the interpretation of literary works. UCL Working Papers in Linguistics 23: 47-68. —. 2013. Irony comprehension: A developmental perspective. Journal of Pragmatics 59: 40-56. Wilson, Deirdre and Dan Sperber. 1992. On verbal irony. Lingua 87: 5376. —. 2002. Relevance Theory. UCL Working Papers in Linguistics 14: 249287. —. 2012. Explaining irony. In Meaning and relevance. Cambridge: Cambridge University Press, 123-145. Yus Ramos, Francisco. 1997-98. Irony: Context accessibility and processing effort. Pragmalingüística 5-6: 391-410.

Code and Inference in the Expression of Irony

195

—. 1999. Misunderstandings and explicit/ implicit communication. Pragmatics 9, no. 4: 487-517. —. 2000. On reaching the intended ironic interpretation. International Journal of Communication 10, nos. 1 and 2: 27-78. —. 2000-01. Literal/non-literal and the processing of verbal irony. Pragmalingüística 8-9: 349-74. —. 2009. Saturación contextual en la comprensión de la ironía. In Dime cómo ironizas y te diré quién eres: Una aproximación pragmática a la ironía, edited by L. Ruiz Gurillo and X. Padilla García. Frankfurt: Peter Lang, 309-331. —. 2012. Relevance Theory and contextual sources-centred analysis of irony. Current research and incompatibility. Paper delivered at EPICS V, University of Pablo Olavide, Seville, March.

PART IV: HUMOROUS DISCOURSE  

CHAPTER TEN DEGREES OF ‘PUNNINESS’? A RELEVANCE-THEORETIC ACCOUNT OF PUNS AND PUN-LIKE UTTERANCES AGNIESZKA SOLSKA, UNIVERSITY OF SILESIA, POLAND

1. Introduction It is generally accepted that one of the key tasks a comprehender has to perform while processing utterances is disambiguation: in order to recover the intended speaker meaning hearers have to identify the referents of indexicals, they have to select the contextually pertinent senses of lexically ambiguous words and to construct the contextually appropriate meanings of numerous expressions by broadening and/or narrowing their lexically encoded meanings. Nevertheless, the fact that some words linguistically encode (and almost all words have the potential to communicate) more than one concept is often exploited for communicative purposes: on occasion speakers intentionally produce utterances whose relevance depends on conveying two meanings simultaneously. The paradigm case of such utterances is puns, especially zeugmatic puns, such as (1), which is phrased in such a way as to force the interpreter to establish and accept two senses of the pivotal word banks. However, not all utterances exploiting the ambiguity inherent in natural language are puns. Consider, for instance, utterance (2), which seems to be metalinguistic rather than ambiguous in nature, and is not particularly likely to be described as a pun despite the fact that the noun crane clearly refers to two totally different kinds of things. (1)

Bank robbers and dam busters blow up banks. (Lascarides et al. 1999, 44)

Degrees of ‘Punniness’?

(2)

199

There is more than one kind of crane, as the birdwatcher pointed out to the construction worker.

Example (2) is an utterance of a type which Wilson and Carston (2007, 238) describe as “pun-like cases involving an element of equivocation or word play”. In their paper “A unitary approach to lexical pragmatics: relevance, inference and ad hoc concepts”, they describe two kinds of such utterances. They illustrate the first kind with the example presented here as (3), in which, they argue, “the word ‘bank’ is intuitively interpreted as referring both to the set of river banks and the set of financial institutions” (Wilson and Carston 2007, 238). (3)

Not all banks are river banks.

The second kind can be observed in what they call “pun-like comparisons”, which they illustrate with examples listed here as (4) – (8), containing adjectives which “have to be simultaneously understood in both physical and psychological senses” (Wilson and Carston 2007, 238): (4) (5) (6) (7) (8)

His mind was as cold as the ice forming on the windscreen. His eyes were as cold as polar ice. Jane is as hard as nails. Sue is as tough as old leather. Jimmy is as sharp as a knife.

The existence of utterances which can be described as being pun-like might indicate that puns are not a clearly defined category of utterances, that in fact there are degrees of ‘punniness’. This seems surprising considering that punning results from ambiguity, which is not a gradable notion. In this paper I would like to take both kinds of utterances discussed by Wilson and Carston (2007) under closer scrutiny and attempt to determine what aspects of meaning make them merely ‘pun-like’ and thus different from actual bona fide puns. Working within the framework of Sperber and Wilson’s (1995) Relevance Theory, in particular its account of lexical meaning (Sperber and Wilson 1995, ch. 2; Carston 2002, ch. 5), I will attempt to specify the conditions that have to be met for the so-called punning effect to arise and to establish to what extent this effect is recreated in pun-like utterances such as the ones above. I will show that the configuration of conceptual meaning conveyed by the pivotal expressions is different in puns vs. pun-like utterances and will argue that

200

Chapter Ten

unlike in puns, in which the ambiguity of the key expression is retained, in the utterances in question it is in fact resolved, albeit in two different ways: (i) in equivocating utterances, such as (2) and (3), two distinct ‘input concepts’ made available by the key element get incorporated into a new hybrid ad hoc concept, and (ii) in ‘pun-like’ comparisons such as (4) – (8), though two distinct ‘input concepts’ are entertained during the interpretation process only one of them fully contributes to the relevance of the utterance.

2. Lexical meaning and disambiguation in Relevance Theory In the cognitively motivated, inferential pragmatic model of utterance interpretation provided by Relevance Theory, words used in utterances are treated by language users as ostensive stimuli, i.e. as deliberately produced verbal signals serving as evidence of the addresser’s intention to communicate a certain meaning. In attempting to establish that meaning, the addressee starts off by identifying mentally represented concepts which are encoded by the words and expressions he has encountered. In doing so he gains access to the threefold information stored in the lexical, logical and encyclopaedic entries of each concept, which specify respectively: (i) the phonetic structure and grammatical properties of the expression encoding the concept, (ii) the defining properties of the concept and (iii) its denotation and/or extension. Although the linguistically encoded meaning usually falls short of what the speaker intends to communicate, the hearer typically has no difficulty filling the gap between the two, thanks to the interpretive abilities characteristic of the human mind. Guided by the presumption that a rational communicator will produce an optimally relevant utterance,1 the comprehender will consider the encoded meaning of the expression in the light of whatever assumptions he has been able to establish based on the linguistic and/or non-linguistic material accompanying the word that is being processed, and he will either select one concept out of many concepts the word can encode or he will treat the encoded concept as a springboard for constructing an occasion-specific concept which the producer had in mind. These so-called ad hoc concepts2 are derived 1

As spelled out in the Communicative Principle of Relevance, “Every ostensive stimulus conveys a presumption of its own optimal relevance” (Sperber and Wilson 1995, 260). 2 For a detailed discussion of ad hoc concepts see Carston (2002, 320-375).

Degrees of ‘Punniness’?

201

inferentially “in response to specific expectations of relevance raised in specific contexts” (Carston 2002, 322), and are constructed by either narrowing or broadening the lexically encoded concepts3. Effectively, the two processes of (i) meaning selection and (ii) meaning construction result in disambiguation of the potentially ambiguous linguistic material. This is what happens in the utterances in (9), which are unambiguous despite containing exactly the same key words bank and cold, which exhibit “the element of equivocation” in the utterances under analysis. Thus, when interpreting utterance (9a), the addressee will select the contextually appropriate meaning of the homonymous verb bank and, when interpreting utterance (9b) or (9c), he will modify the lexically given concept BANK and construct an ad hoc concept BANK*, which in the case of interpreting (9b) will involve narrowing the range of objects which could be called bank to a specific building, and in the case of interpreting (9d), broadening the range of objects to which this noun can apply. Arguably, in utterance (9a), once the linguistically encoded concept BANK has been indentified, it will also get adjusted, i.e. fine-tuned to pick out its most textually appropriate referent. (9)

a. b. c.

After going swimming in the river, I went to a bank and cashed a cheque. Turn left at the village bank. (Pointing to a cash machine): Let me get some money from the bank.

Disambiguation of ambiguous words and the modification of the concepts they encode proceeds according to a certain interpretational heuristic, dubbed the relevance-theoretic comprehension procedure, which underlies all tasks involved in deriving speaker intended meanings. According to this heuristic, the addressee takes a path of least effort in constructing interpretations, formulates hypotheses about the possible content of words and utterances in order of accessibility, tests these hypotheses in the light of available contextual assumptions and stops as soon as the expectations of optimal relevance have been satisfied (Wilson and Sperber 2004, 613). A graphic representation of the processes involved in deriving the meaning of the word bank in utterance (9a) has been shown in figures 1a and 1b. More than one figure has been proposed because the ostensive 3

For the discussion of concept modification in metaphors see Carston (2002, 320375).

Chapter Ten

202

signal, i.e. the homonymous key expression bank, makes available multiple concepts, which conceivably might be accessed in different order depending on which meaning is more accessible for a particular comprehender. For the sake of simplicity, only two such concepts have been indicated: figure 1a captures the situation when the first concept to be accessed happens to be the one intended by the speaker. Figure 1b shows what happens if the interpreter started off by indentifying the ‘wrong’ concept. In the two figures the following aspects of interpreting lexically ambiguous lexical items have been indicated: (i)

(ii)

(iii)

(iv)

(v) (vi)

The ostensive signal: the word which triggers the comprehension process and which can take either the spoken (e.g. /bæƾk/ or the written form, e.g. ). The “input” concepts: the lexically encoded concepts, which the ostensive signal makes available for the interpretation process and which are accessed in the order indicated by the subscript numbers. The three entries of each concept: these are presented to show that the two concepts encoded by the ambiguous words share the same lexical entry, yet have different logical addresses and different encyclopaedic entries providing access to different sets of encyclopaedic data. The aspects of conceptual meaning which the two concepts have in common are indicated by the symbols of equality (=) placed between corresponding entries of the concepts, the shaded sections and the heavy frames placed around the concepts’ lexical entries. The aspects of their meaning which are different are indicated by the inequality symbols (). The “output” concepts: the speaker intended concepts, which are conveyed by the use of the lexical item being processed. They are marked with an asterisk to show that have been pragmatically fine-tuned. The process of accessing as well as inferentially deriving concepts is indicated by the down-pointing arrows (Ë, È or Ì). Failure to consider a concept or the rejection of a concept which has been considered yet found contextually inappropriate has been marked with the x symbol (¯).

Though grossly simplifying what is involved in deriving lexical meanings, figures of this kind will later be used to compare the processes of deriving meanings of the key expressions in puns and pun-like utterances.

Degrees of ‘Punniness’? OSTENSIVE SIGNAL:

/bæƾk/ or ¯

Ë

INPUT CONCEPTS:

203

CONCEPT1

CONCEPT2

e.g. BANK1 [=institution]

e.g. BANK2 [=riverside]

LEXICAL ENTRY1 LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

= 

LEXICAL ENTRY2



ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY2

È CONCEPT1*

e.g. BANK1* OUTPUT CONCEPT:

LEXICAL ENTRY1 LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

Figure 1a. Derivation of the concept conveyed by the use of noun bank in utterance (9a), with the speaker intended concept accessed first. OSTENSIVE SIGNAL:

/ bæƾk / or Ë

INPUT CONCEPTS:

Ì

CONCEPT1

CONCEPT2

e.g. BANK1 [=riverside]

e.g. BANK2 [=institution]

LEXICAL ENTRY1 LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

¯

= 

LEXICAL ENTRY2



ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY2

È CONCEPT2*

e.g. BANK2* OUTPUT CONCEPT:

LEXICAL ENTRY2 LOGICAL ENTRY2 ENCYCLOPAEDIC ENTRY2

Figure 1b. Derivation of the concept conveyed by the use of noun bank in utterance (9a), with the ‘wrong’ concept accessed first.

Chapter Ten

204

As shown in figures (1a) and (1b), whether the comprehender considers both concepts made available by the ostensive signal or one, the outcome is the same: in the end only one concept is accepted as the one intended by the speaker. In the case of puns and pun-like utterances the outcome of the interpretation process is different. The interpreter does not choose one concept to the exclusion of others. Instead, he is forced to derive two meanings of the same ostensive signal and to accept them both. In what follows I am going to demonstrate that what sets puns off from pun-like utterances is the fate of the two “input” concepts determined by the nature of the contextual assumptions against which the meaning of the key expression is processed.

3. The oscillating effect and the conditions on ‘punniness’ The hallmark feature of puns is the so-called oscillating effect: a pun has been successfully interpreted if its interpreter ends up swinging back and forth between two conflicting interpretations of one linguistic form without being able to abandon either. This inability of the interpreter to derive a single meaning of one linguistic form can be observed not only in puns with two viable meanings, such as (1) but also in those with only one such meaning, such as (10). It is unlikely that the producer of this advertising slogan intended to convey the idea that the flowers called roses grow on the addressee. Rather, his intension was to instil in the addressee the belief that he would find himself liking the chocolates called Roses more and more. Nonetheless, instead of vanishing, the inadmissible meaning lingers in the interpreter’s mind as a pun-creating counterpoise to the valid meaning. (10)

Roses grow on you. (advertising slogan for Cadbury chocolates)

Arguably, in order for the oscillating effect to arise the following three conditions must be met. They can be regarded as the necessary and sufficient conditions for an utterance to be perceived as a pun. (i)

The utterance must contain a connector, i.e. a potentially ambiguous linguistic form, which in the structure of the utterance can appear once or more than once, on each occasion conveying a different meaning. In this paper the discussion is limited to the former type of puns, sometimes described as paradigmatic puns, since they best resemble the pun-like utterances under analysis in which the pivotal element likewise makes only one appearance.

Degrees of ‘Punniness’?

205

(ii) The context against which the utterance is processed must contain at least two disjunctive elements,4 i.e. linguistic and/or extralinguistic material which makes available two sets of assumptions forcing the comprehender to derive and accept two (or more) of the connector’s meanings5. (iii) The two meanings of the connector which the interpreter is forced to derive must be sufficiently distinct from one another. It is the third condition which helps us address the issue of varying degrees of ‘punniness’, since it captures the fact that there seems to be a cline of cases among the utterances exploiting ambiguity, with some utterances more and some utterances less likely to be perceived as puns. If the first two conditions are met, the oscillating effect is almost certain to arise if the connector is homonymous, i.e. if it encodes two clearly diverse concepts, as is the case with utterance (1). Other obvious candidates for distinct meanings are the two concepts made available by homophonous and paronymous connectors as in (11) and (12) respectively, though only one of the two is explicitly expressed in the utterance. A comprehender can identify the other one, the so-called ‘target concept’, only if his cognitive environment contains the pertinent information allowing him to do so. (11) (12)

Everybody kneads it. (kneads/needs; advertisement for Pillsbury flour) I used to be a doctor, then I lost patients. (patients/patience)

On the other hand, the oscillating effect does not occur if the denotations of the two concepts are subsets of the denotation of some more general concept, as in utterance (13), which cannot be taken to mean that Peter admired a bird called crane (concept CRANE*) while his wife admired the painting representing such a bird (concept CRANE**), since both concepts, i.e. CRANE* and CRANE** are merely subsets of the same basic concept CRANE. (13)

4

Peter and his wife admired the crane.

Conceivably, more than two meanings can be involved. In his analysis of humorous puns, Attardo (1994), who introduced the terms ‘connector’ and ‘disjunctor’, recognized only one disjunctive element, causing passage from one meaning to another. Solska (2012c) argued why two such disjunctive elements must be recognized. 5

206

Chapter Ten

Now in between these two extremes there is a gray area of utterances which meet the first two criteria yet do not completely fulfill the third one. These would include utterances whose connectors make available the linguistically given and metaphorically and/or hyperbolically extended meanings of unambiguous words. Different comprehenders may differ with respect to whether they see such two meanings as distinct enough. In what follows I am going to demonstrate how these conditions are met in punning utterances such as utterance (1) and attempt to determine what prevents them from being completely fulfilled in pun-like utterances.

4. Juxtaposition of concepts in puns In utterance (1) the fragment serving as the connector is obviously the homonymous noun banks. The interpreter’s decision as to which particular concept is intended by its use is based on the information stored in the encyclopaedic entries of the two concepts made available by the noun phrases bank robbers and dam busters. These two serve as the two disjunctive elements, which make available two mutually exclusive sets of contextual assumptions. It is these sets of assumptions that lead the interpreter to derive and accept two competing meanings of the pivotal expression, i.e. two distinct contextually appropriate concepts: BANK1, pertaining to financial institutions, and BANK2, pertaining to riversides. Unable to let go of either concept, the interpreter ends up oscillating between two interpretations. The particular conceptual content of the pivotal word in the homonymic pun in (1) has been shown in figure 2. Unlike in figures (1a) or (1b), in figure 2 neither of the concepts made available by the ostensive signal is rejected. Instead, the ambiguity of the key expression is retained and the resulting interpreter’s oscillation between them is indicated by the double-ended arrow. For simplicity’s sake only one figure has been presented though the order in which the two concepts encoded by the key word bank are accessed may be different for different comprehenders, with the more accessible concept derived first. Since in utterance (1) it is the disjunctor bank robbers that appears first, it is feasible that the addressee will start off by deriving the ‘bank-as-institution’ meaning of the ostensive signal bank. He will, however, be unable to halt the interpretation process on deriving this meaning. After all, at this point his cognitive environment will still include the set of contextual assumptions made available by the second disjunctor dam busters. These assumptions will give him reason to believe that he has not yet obtained the full range of cognitive effects the connector noun can yield. That is why he will

Degrees of ‘Punniness’?

207

reapply the procedure and derive the other, ‘bank-as-riverside’ meaning of the ostensive signal bank. OSTENSIVE SIGNAL:

/ bæƾk / or Ì

Ë

INPUT CONCEPTS:

CONCEPT1

CONCEPT2

e.g. BANK1 [=institution]

e.g. BANK2 [=riverside]

LEXICAL ENTRY1 LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

OUTPUT CONCEPTS:

= 

LEXICAL ENTRY2



ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY2

È

È

CONCEPT1

CONCEPT2

e.g. BANK1* [=institution]

Ù

e.g. BANK2* [=riverside]

LEXICAL ENTRY1

= 

LEXICAL ENTRY2



ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

LOGICAL ENTRY2

Figure 2. The derivation of concepts conveyed by the use of the key term in puns. What is the relevance of such an outcome? A detailed discussion of this issue can be found in Solska (2012a). In short, punning utterances achieve relevance by juxtaposing concepts made available by a single linguistic form. Bringing two meanings together in one form forces the interpreter of a pun to seek a commonality of meaning between two disparate domains. In utterance (2), the suggested link between bank robbers and dam busters might be that they both engage in what is essentially a destructive activity. If the two meanings are perceived as incompatible, the result may be a humorous effect. As observed by many scholars (Raskin 1985; Pollio 1996; Attardo and Raskin 1991), humor is a product of two unreconcilable meanings present in a text. Utterance (1) is not particularly humorous but, in general, puns, whose connectors make available two meanings which often clash with each other, are by their very nature geared for evoking humor.

208

Chapter Ten

5. Incorporations of concepts in ‘equivocating’ utterances To what extent do equivocating utterances such as (2) and (3) and the punlike comparisons (4) – (8) meet the conditions for ‘punniness’ outlined above? Utterances (2) and (3) clearly fulfill the first of the three conditions. In (2) the connector is the noun crane, in (3) the noun bank. As for the disjunctive elements, they are explicitly present only in utterance (2), where the nouns birdwatcher and construction worker give access to the information stored in the encyclopaedic entries of the two concepts: BIRDWATCHER and CONSTRUCTION WORKER. The former allows the hearer to derive the ‘crane-as-bird’ meaning of the connector, the latter the ‘crane-as-machine’ meaning. In utterance (3), only one disjunctor is explicitly present: the noun phrase river banks. The other disjunctor can however be easily identified though it is merely implied by the quantifying expression not all. This expression signals to the interpreter that the predicate that follows the key noun banks will apply only to a subset of entities included in the denotation of that noun. However not all does more than divide the set of all things which can be called banks into two sets, it also introduces an element of vagueness which does not appear in bona fide puns. Thus instead of deriving two clearly delineated concepts, the interpreter is made to derive only one: in contrast to the ‘bank-as-riverside’ meaning, the ‘bank-as-everything-except riverside’ meaning remains vague. As it happens, utterance (2) also contains a component which introduces an element of vagueness: the expression more than one kind of, which contains a vague quantifier more than one. This expression alerts the hearer to the fact that the denotation of the noun phrase that appears in its scope will not include a single set, but multiple sets of things. The actual number of such sets will remain vague until the two disjunctive elements will (tentatively) limit it to just two: the ‘crane-as-bird’ and the ‘crane-as-device’. Note that neither not more than one nor not all introduce ambiguity. If no disjunctive elements appear in the utterance, as is the case in utterances (14) and (15), the pivotal noun will remain univocal. As noted by Channell (1994, 111), “much of the apparent effect of a particular quantifier actually derives from the choices in the surrounding linguistic context”. (14)

In Asia we can encounter more than one kind of crane: in addition to the common crane this vast continent is home to the sandhill crane, the demoiselle crane, and black-necked crane, to name just a few.

Degrees of ‘Punniness’?

(15)

209

In my home town not all banks open at 9.

Would the oscillating effect arise if instead of a vague quantifier a ‘precise’ one was used? Utterance (16)6, in which the noun crane appears within the scope of the precise numeral two, does seem to evoke the punning effect, though arguably not a particularly strong one. Apparently the oscillating effect is more pronounced if the two meanings of the connector have to be inferred instead of merely accessed, an idea which requires further investigation, which is nevertheless beyond the scope of this paper. (16)

Peter came back from the trip with a fantastic picture of two cranes: a magnificent red-crowned crane hovering over a huge construction crane in the field.

What then is the outcome of placing a potentially ambiguous word in the scope of a vague quantifying expression? In their discussion of utterance (3), Wilson and Carston (2007, 238) point out that it seems metarepresentational7 in character and that the word bank is intuitively felt to refer to “things called bank”. The same could be said of the word crane in utterance (2). It might be argued however that utterances such as (2) and (3) are not really cases of metalinguistic use since in typical cases of such use, shown in (17) – (18), reference is made only to the key word’s linguistic properties or to its orthographic form, while the logical and encyclopaedic properties of the concept the word encodes are irrelevant. However, this is not so in utterances (2) and (3), in which all aspects of conceptual meaning play a role. In utterance (2) the speaker is not talking about the word bank but about things called bank. Similarly, in utterance (2), reference is made not to the word crane but to the things called crane, which include the set of wading birds of the family Gruidae, the set of machines used for lifting and moving heavy weights in suspension, and possibly other unspecified sets. (17) (18)

6

Bank is a countable noun. Bank has four letters.

I would like to thank the anonymous reviewer of the previous version of this paper for providing this example. 7 For the discussion of the notion of metarepresentation see Wilson (2000) and Noh (2000).

210

Chapter Ten

I would like to argue that if disjunctive elements are present, placing the key word within the scope of a vague quantifying expression leads the interpreter to create a new hybrid occasion-specific concept which includes the denotations of both. Since two concepts encoded by the key word continue to coexist, complementing each other and forming a new hybrid occasion-specific concept which includes the denotations of both, to describe the outcome of the meaning derivation process in utterances such as (2) and (3) one could use the term incorporation of concepts. Utterances such as these seem to achieve relevance by alerting the addressee to the homonymy of a lexical item, i.e. to the fact that it encodes more than one concept. The content of the input concepts and the resulting output concept BANK* can be represented in a graphic form, such as figure 3 below. Two sets of single-headed arrows indicate first the concepts which are made available by the ostensive signal, i.e. by either the spoken or the written form of the key expression, then later the hybrid concept emerging from processing the two input concepts in the context in which they occur. The shaded sections in the figure show which aspects of the conceptual information the input concepts have in common. The equation mark (=) between the corresponding entries of these two input concepts indicates their identity and the inequality symbol () indicates the lack of identity between the corresponding concept entries. The incorporation of concepts is indicated by the addition mark (+). As can be seen, it is the input concepts’ logical and encyclopaedic entries, which were different in the input concepts, that get combined in the output concept:

Degrees of ‘Punniness’? OSTENSIVE SIGNAL:

211

/ bæƾk/ or Ë

Ì CONCEPT2

CONCEPT1

e.g. BANK2* [=other things called bank]

e.g. BANK [=riverside] INPUT CONCEPTS:

LEXICAL ENTRY1 LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

= 

LEXICAL ENTRY2



ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY2

Ì

Ë CONCEPT3

OUTPUT CONCEPT:

e.g. BANK** [=things called bank] LEXICAL ENTRY LOGICAL ENTRY1 + LOGICAL ENTRY2 ENCYCLOPAEDIC ENTRY1 + ENCYCLOPAEDIC ENTRY2

Figure 3. The derivation of the hybrid concept conveyed by the key term in the equivocating utterance (3). As can be seen, the conceptual content of the pivotal expression in incorporations is quite different from the conceptual content of the pivotal expression in puns. The feeling of equivocation is caused by the fact that indeed two concepts are entertained during the process of deriving the speaker intended meaning. However, there is no oscillation between meanings because eventually only one concept is formed and the utterance achieves relevance by highlighting the homonymy of the key word. As for the impression that the key words in (2) and (3) have been used metarepresentationally, it can be easily explained if we consider the conceptual information which is accessed through the intentionally ambiguous expression in such utterances: as can be seen in figure 1, the output concept’s lexical entry is the same as the lexical entry of either of the input concepts: to highlight this fact heavy frames have been placed around the lexical entries.

212

Chapter Ten

5. Processing pun-like comparisons Let us now turn to the other type of pun-like utterances. In contrast to ‘regular’ as... as... comparisons, whose function is to assert that the entity referred to by the subject noun phrase possesses a certain feature which is also possessed by the entity to which the first entity is being compared, the raison d’être of pun-like comparisons is to imply that the entity identified by the subject noun phrase possesses some quality to a particularly high degree. For instance, utterance (4) conveys the idea that some man’s mind was totally unaffected by emotions, i.e. very cold or, to be more exact, very COLD*. Similarly, utterance (6) conveys the idea that Jane is callous and unsympathetic, i.e. very hard or, to be more exact, very HARD*. This effect is achieved by likening the entity to be described (some man’s eyes, a woman called Jane) to some other entity (ice, nails) known to prototypically possess the quality identified by the pivotal adjective. Utterances (4) – (8) meet the first two of the three criteria for ‘punniness’. They all contain a connector, which in this case is an adjective referring to some quality. They also contain two disjunctive elements: two explicitly expressed noun phrases referring to two entities being compared with respect to that quality. Characteristically, these two entities belong to disparate classes of things: the first disjunctor is a NP referring to a person or some aspect of a person’s character (his mind, his eyes seen as the expression of that person’s inner qualities), the second disjunctor is a NP referring to some physical object. As a result, the connector is expected to oscillate between its psychological and the physical senses. However, the oscillation seems to be much weaker than in regular puns, and for some comprehenders it may not appear at all, which is why the utterances in question have been described as merely pun-like. I believe that what prevents such utterances from exhibiting the fullyfledged oscillating effect is a combination of several factors.

5.1 Inadequate disparity between concepts The most obvious reason why Wilson and Carston’s examples fail to behave like regular puns is they do not or only marginally meet the third condition on ‘punniness’, presented in section 3 of this paper. This is to be expected considering that the psychological senses of the connector adjectives such as cold or hard can be seen as metaphorical modifications of the lexically encoded senses of the key words. As such, they either include or overlap with the senses they have been derived from. For example, though in utterance (4), the adjective cold can apply to the noun

Degrees of ‘Punniness’?

213

phrase ice only in its strictly literal sense of ‘being characterised by low temperature’ (i.e. lacking physical warmth) and to the noun phrase his mind in its conventionally metaphorical sense of ‘imperturbable’ (i.e. unaffected by emotions, exhibiting lack of emotional warmth), the two concepts COLD and COLD* share a fragment of their logical and encyclopaedic entries. Similarly, though the adjective hard in (6) can apply to the noun phrase nails only in its strictly literal sense of ‘firm and stiff’ (and thus in certain circumstances likely to cause physical damage/capable of withstanding impact) and to the proper noun Jane in its conventionally metaphorical sense of ‘showing no kindness or sympathy towards other people’ (and thus likely to cause them emotional pain), the two concepts at play, namely HARD and HARD* have a portion of their logical and encyclopaedic meaning in common. Not being sufficiently distinct, the two concepts cannot be juxtaposed, hence the lack of an oscillating effect, or only a marginal one. A similar situation can be observed in as…as… comparisons (19) – (22), in which other types of concept modifications can be identified: (19) (20) (21) (22)

His face was as round as a full moon. [approximation1/approximation2] His action was as quick as lightning [encoded sense/hyperbolic extension] My car is as old as the hills. [narrowing1/narrowing2] Her mind is as sharp as her tongue. [metaphorical extension 1/metaphorical extension 2]

In utterance (19), in which a different degree of roundness can be ascribed to the full moon and to a person’s face, we find two different approximations of an encoded concept ROUND, namely ROUND* and ROUND**, both involving a departure from a perfectly circular shape. An inferentially constructed concept QUICK*, which is a hyperbolic extension of the concept QUICK, can be found in utterance (20), which (implausibly) proclaims identity between the speed with which a person’s action is performed and the speed of lightning. Different narrowings of the same concept OLD can be observed in (21), in which a different time span, reflected in such concepts as OLD* and OLD**, is applicable to the two things being compared, i.e. the car and the hills. Finally, two different metaphorical extensions of the same concept can be observed in example (22), in which the adjective sharp does not convey its encoded meaning of SHARP (i.e. ‘capable of dissecting physical objects thanks to having a thin edge’) but appears in a comparison whose function is to liken one abstract

214

Chapter Ten

entity, namely a person’s mind, which can be SHARP* (i.e. ‘mentally acute and penetrating, thus capable of mentally dissecting reality’) to another abstract entity, namely a person’s tongue (itself a metaphor for a manner of speaking), which can be SHARP** (i.e. ‘unmerciful’ or ‘capable of hurting people’s emotions’). In all these cases the two concepts involved, though not identical, are insufficiently distinct for the oscillating effect to appear.

5.2 Fusion of the two input concepts or univocality? What then is the fate of the two concepts made available in the process of interpreting pun-like comparisons? One possibility is that they merge, producing a fusion of the pivot’s encoded ‘physical’ meaning and its ‘psychological’ metaphorical extension. If this is the case, then the interpretation process would result in the formation of an ad hoc output concept whose lexical entry would be the same as the lexical entries of either of the input concepts, but whose logical and encyclopaedic entries would include a mixture of some but not all aspects of the corresponding logical and encyclopaedic entries of the input concepts. The explanation for the weak oscillation would then be that the two concepts the interpreter is made to entertain are not only insufficiently distinct but appear only in the early stages of the interpretation process, by the end of which only one concept is in fact derived. A graphic representation of such a hypothetical fusion of concepts is presented in figure 4a. As in incorporations, the output concept’s lexical entry is the same as the lexical entry of either of the input concepts, which is indicated by the shaded sections in the figure as well as the ‘equal to’ mark (=) placed between the corresponding lexical entries. The ‘approximately equal’ symbols (§) placed between the logical and encyclopaedic entries of the input concepts indicate their relative similarity. The fact is also indicated by the shaded sections, which cover a fragment of these entries in the first concept. The slashes (/) in the output concepts’ logical and encyclopaedic entries indicate the ‘fused’ aspects of conceptual meaning, i.e. the fact that the output concepts include a mixture of only some information contained in the logical and encyclopaedic entries of the input concepts.

Degrees of ‘Punniness’?

OSTENSIVE SIGNAL:

INPUT CONCEPTS:

215

/kԥ‫ݜ‬ld/ or Ë

Ì

CONCEPT1

CONCEPT2

e.g. COLD* [=emotionally cold]

(Ù)

e.g. COLD [=physically cold]

LEXICAL ENTRY1

= §

LEXICAL ENTRY2

§

ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

Ì

LOGICAL ENTRY2

Ë CONCEPT3

OUTPUT CONCEPT:

e.g. COLD** LEXICAL ENTRY LOGICAL ENTRY1 / LOGICAL ENTRY2 ENCYCLOPAEDIC ENTRY1 / ENCYCLOPAEDIC ENTRY2

Figure 4a. A hypothetical derivation of the hybrid ad hoc concept representing the fusion of the encoded and metaphoric sense of the key adjective in as...as.... comparisons. Underlying the idea of the fusion of concepts is the assumption that the two disjunctors make an identical contribution to the relevance of the utterance. This, however, does not seem to be the case. In utterances (4) – (8), some quality is predicated only of whatever the subject noun phrase functioning as the first or disjunctor refers to. The other disjunctor serves merely as an example of the kind of entity that prototypically exhibits that quality. For instance, utterances (4) and (6) do not predicate anything of ice or nails. Instead they predicate some quality only of some person’s mind or of a woman called Jane. The contribution of the second disjunctor to the interpretation process is to signal that the referent of the first disjunctor exhibits a notably large amount of that quality, e.g. is emotionally very cold (i.e. COLD**). Thus it is more likely that instead of two concepts being conflated into one, only one of them will remain as the concept intended by the producer. The fleshing out of the intended meaning of the connector adjective may proceed in the following way. On encountering the connector the addressee interprets it in the light of the information provided by the subject NP disjunctor. This allows him to derive the metaphorical meaning, which is already a modified (broadened and/or narrowed) version of the encoded meaning. Then, on encountering the second disjunctor he accesses the encoded meaning, which either intersects with

216

Chapter Ten

or is a subset of the metaphorical meaning. Of the two only the first one (e.g. the metaphorically extended concept COLD* or HARD*) is fully relevant. The other one, i.e. the basic concept (e.g. COLD and HARD) gets activated but is found to contain the same core information as the fully relevant concept. Therefore there is no need for the interpreter to process it any further. Its role boils down to strengthening the assumptions about the content of the already derived metaphorical meaning. As was explained above, the weak oscillation which may be experienced by some interpreters is the result of the interpreter having accessed two dissimilar concepts in the early stages of the interpretation process. However, the utterance is essentially unambiguous. A graphic representation of such a derivation process has been shown in figure 4b. As in figure 4a, the ‘approximately equal’ symbols (§) reflect the fact that the two input concepts are not entirely distinct. The possible weak oscillation is marked by the double-ended arrow in parentheses ((Ù)) and the x symbol (¯) indicates that the second concept, once it has affected the interpretation of the connector, is abandoned and plays no further role in the derivation of connector’s meaning. The figure also shows that the single ‘output’ concept is a further modification of the first of the input ‘concepts’, carried out under the influence of the second ‘input’ concept has.

Degrees of ‘Punniness’? OSTENSIVE SIGNAL:

INPUT CONCEPTS:

217

/kԥ‫ݜ‬ld/ or Ë

Ì

CONCEPT1

CONCEPT2

e.g. COLD* [=emotionally cold]

(Ù)

e.g. COLD [=physically cold]

LEXICAL ENTRY1

= §

LEXICAL ENTRY2

LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

§

ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY2

È

¯

CONCEPT1

OUTPUT CONCEPTS:

e.g. COLD** [=emotionally very cold] LEXICAL ENTRY1 LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

Figure 4b. The derivation of the single concept conveyed by the use of the key adjective in as...as.... comparisons.

5.3 Degrees of conventionalisation and the depth of processing Interestingly, in addition to merely pun-like as… as…. comparisons, there are comparisons which do trigger a full-blown oscillating effect. This is what we can observe in the vaguely humorous utterances which Dynel (2010, 118-119) calls ‘punning similes’ and which she illustrates with examples listed as (23) and (24) below. The existence of such utterances indicates that one more factor is at work in precluding the strong oscillating effect from arising in utterances such as (4) – (8). (23) (24)

I’m as wound as a Timex. I’m up and down like a toilet at a mixed party.

Utterances (23) and (24) do not seem to be essentially different from utterances (4) – (8)8. In both the two entities being compared represent disparate classes of things, with the subject NP disjunctor belonging to the

8

Strictly speaking, utterance (24) is not an as… as… comparison. However, despite a slightly different structure it contains a distinguishable conjuctor (up and down) and two disjunctive elements (I and a toilet at a mixed party).

218

Chapter Ten

category of human beings and the second disjunctor referring to inanimate objects. In both the denotations of the two concepts made available by the connector overlap. Utterance (23) conveys the idea that its speaker is very tense, or to be more precise, emotionally tense or nervous. This sense of the key word wound seems to be a metaphorical extension of the lexical meaning of wound. The other sense of that word, namely ‘being made operational by turning a knob and making the inner spring tense’ seems to be a narrowing of the connector’s encoded sense. Utterance (24) conveys the idea that the speaker keeps changing his position by standing up and sitting down, much as a toilet seat constantly changes its position during a party attended by people of both sexes, during which it continually gets raised and lowered. In this utterance the two concepts made available by the phrase up and down seem to be different modifications of the basic concept and thus they too share a portion of their encyclopaedic and logical entries. Though the processes of deriving the meaning of the connector adjective in (23) and (24) is not different from the corresponding process in utterances (4) – (8), only punning similes (23) and (24) effectively juxtapose two disparate domains: a person’s emotional condition and the workings of a mechanical device. The result is the humorous effect, absent in pun-like comparisons. I believe that the difference in the effect utterances (23) – (24) and utterances (4) – (8) have on the comprehender can be attributed to the differing degree of conventionality the two sets of utterances exhibit. In (23) – (24) we are dealing with novel, creative metaphorical uses of the connector items. Utterances (4) – (8) belong to a vast set of conventionalized similes, among whose many members we find as white as snow, as cool as a cucumber, as drunk as a lord, etc. Now, in Relevance Theory conventional meanings, exhibited for instance by metaphors and idioms, are seen as resulting from the repeated use of specific utterances. According to Vega Moreno (2007, 196), “repeated processing of an idiom may recurrently direct a hearer along the same inferential route, which may at some point develop into a fullfledged pragmatic routine.” The result is that the comprehender no longer needs to delve into the encyclopaedic entries of the encoded concepts and may find “his attention and processing resources automatically directed towards the sort of contextual assumptions and implications that have generally led to a successful outcome in processing this idiom” (Vega Moreno 2007, 197). This is what we observe in (4) – (8): the concepts made available by the connector, instead of being occasion specific, are standardised ways of talking about various aspects of the physical world

Degrees of ‘Punniness’?

219

(ice, nails, etc.) as well as various aspects of human character and emotions. As such they are processed in a rather shallow manner9, requiring only a minimal effort on the part of the comprehender. In contrast, when processing novel utterances (23) – (24), the interpreter is forced to reach deeply into the content of the encyclopaedic entries of the connector and two disjunctors as this is the only way to flesh out the intended meaning of the somewhat bizarre utterance. Both disjunctive elements thus make an unequal yet valid contribution to the utterance’s relevance. As a result, the two meanings of the connector, which the disjunctors help to establish, can be perceived as not only relevant but also as distinct enough to be juxtaposed. The derivation process of the connector’s meaning in punning similes has been presented in figure 4c. The element that makes it different from figure 4b is the lack of parentheses around the double-ended arrow (Ù), indicating the strong oscillation between the two input concepts. OSTENSIVE SIGNAL:

INPUT CONCEPTS:

/Wa‫ݜ‬nd/ or Ë

Ì

CONCEPT1

CONCEPT2

e.g. WOUND1* [=emotionally tense]

Ù

e.g. WOUND2** [=physically tense]

LEXICAL ENTRY1

= §

LEXICAL ENTRY2

§

ENCYCLOPAEDIC ENTRY2

LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

È

LOGICAL ENTRY2

¯

CONCEPT1

OUTPUT CONCEPTS:

e.g. WOUND1* [=emotionally very tense] LEXICAL ENTRY1 LOGICAL ENTRY1 ENCYCLOPAEDIC ENTRY1

Figure 4c. The derivation of the single concept conveyed by the use of the key adjective in punning as...as.... comparisons.

9 The notion of the depth of processing during online interpretation of verbal inputs is discussed by Sanford (2002).

220

Chapter Ten

6. Concluding remarks In this paper three different types of utterances exploiting ambiguity have been distinguished and discussed. Puns, specifically zeugmatic puns (Bank robbers and dam busters blow up banks), and two kinds of utterances, which Wilson and Carston (2007) describe as merely pun-like: utterances which seem to be metarepresentational in character (Not all banks are river banks) and as… as… comparison (Jane is as hard as nails). In puns, two meanings of one linguistic form get juxtaposed and the interpreter ends up going back and forth between them, unable to let go of either. This so-called oscillating effect is much weaker in pun-like utterances, even though they too exhibit an element of equivocation. In an attempt to address the issue of whether the quality of being a pun, or ‘punniness’, is indeed a gradable notion, I first considered the conditions that have to be met if an utterance is to perceived as a pun. Then I went on to establish what factors play a role in preventing these conditions from being completely satisfied in the other types of utterances and I examined the specific conceptual content of the key element (or connector) of each utterance type. I hope to have demonstrated that in equivocating ‘metarepresentational’ utterances, multiple meanings conveyed by one linguistic form combine with each other and yield a single new hybrid concept in which each of the input concepts retains its distinctiveness, contributing to the overall relevance of the utterance by highlighting the homonymy of the connector word. The key factor that contributes to this conceptual outcome, which in this paper has been described an incorporation of concepts, is an element of vagueness present in the utterance, which prevents the interpreter from forming two clearly defined concepts. In the case of pun-like comparisons, I argued that though the key adjective makes available two concepts, only the one that applies to the subject NP makes an adequate contribution to the relevance of the utterance. The other one contains the same core encyclopaedic information as the first one and its role in the interpretation process is limited to strengthening the already established assumptions about intended meaning of the connector. Thus the utterance is essentially unambiguous and the impression of oscillation between the ‘psychological’ and ‘physical’ senses of the adjective, which some interpreters may experience, results from the fact that the interpreter did access two not entirely similar concepts in the early stages of the interpretation process. This impression may, however, turn into a full-scale oscillating effect if instead of a conventional comparative phrase (as cold as ice) a creative one is used (as

Degrees of ‘Punniness’?

221

wound as a Timex), requiring the comprehender to construct a novel oneoff concept, whose derivation involves delving deeply into the encyclopaedic entries of the two noun phrases whose whose referents are being compared.

Acknowledgement I would like to thank the anonymous reviewer of the earlier draft of this paper for invaluable comments which helped me realize that my previous ideas were untenable.

References Attardo, Salvatore and Victor Raskin. 1991. Script Theory revis(it)ed: Joke similarity and joke representation model. Humor 4, nos. 3 and 4: 293-347. Carston, Robyn. 2002. Thoughts and utterances: The pragmatics of explicit communication. Oxford: Blackwell. Channell, Joanna. 1994. Vague language. Oxford: Oxford University Press. Dynel, Marta. 2010. How do puns bear relevance. In Relevance studies in Poland: Exploring translation and communication problems. Vol. 3, edited by M. Kisielewska-Krysiuk, A. Piskorska and E. Waáaszewska. Warsaw: Warsaw University Press. Lascarides, Alex, Ann Copestake and Ted Briscoe. 1996. Ambiguity and coherence. Journal of Semantics 13, no. 1: 41-65. Noh, Eun-Ju. 2000. Metarepresentation: A relevance-theory approach. Amsterdam: Benjamins. Pollio, Howard R. 1996. Boundaries in humor and metaphor. In Metaphor: Implications and applications, edited by J. S. Mio and A. N. Katz. Mahwah: New Jersey: Lawrence Erlbaum. Raskin, Victor. 1985. Semantic mechanisms of humor. Dordrecht: Reidel. Sanford, Anthony J. 2002. Context, attention and depth of processing during interpretation. Mind and Language 17, nos. 1 and 2: 188-206. Solska, Agnieszka. 2012a. The relevance of the juxtaposition of meanings in puns. In Exploring language through contrast, edited by W. Skrzypczak, T. Fojt, S. Wacewicz and W. Skrzypczak. Newcastleupon-Tyne: Cambridge Scholars Publishing. —. 2012b. Relevance-theoretic comprehension procedure and processing multiple meanings in paradigmatic puns. In Relevance Theory: More

222

Chapter Ten

than understanding, edited by E. Waáaszewska and A. Piskorska. Newcastle-upon-Tyne: Cambridge Scholars Publishing. —. 2012c. The relevance-based model of context in processing puns. Research in Language 10, no. 4: 387-404. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. Vega Moreno, Rosa E. 2007. Creativity and convention: The pragmatics of everyday figurative speech. Amsterdam: Benjamins. Wilson, Deirdre. 2000. Metarepresentation in linguistic communication. In Metarepresentations: A multidisciplinary perspective, edited by D. Sperber. Oxford: Oxford University Press. —. 2009. Irony and metarepresentation. UCL Working Papers in Linguistics 21: 183-226. Wilson, Deirdre and Robyn Carston. 2007. A unitary approach to lexical pragmatics: relevance, inference and ad hoc concepts. In Pragmatics, edited by N. Burton-Roberts. Basinstoke and New York: Palgrave Macmillan. Wilson, Deirdre and Dan Sperber. 2004. Relevance Theory. In: The handbook of pragmatics, edited by L. R. Horn and G. Ward. Oxford: Blackwell.

CHAPTER ELEVEN “TO CLASSIFY OR NOT TO CLASSIFY?” – A CRITICAL SURVEY AND QUANTITATIVE ANALYSIS OF RELEVANCE-THEORETIC CLASSIFICATIONS OF JOKES MAGDALENA BIEGAJàO, UNIVERSITY OF WARSAW, POLAND

1. Introduction Relevance-theoretic analyses of humorous texts, especially canned jokes, have received considerable attention in recent years (Curcó 1995, 1996a, b, 1997; Dynel 2010, 2012; Groefsema 1998; Jodáowiec 1991a, b, 2008; Larkin 2000; Yus 2003, 2004, 2008, 2012, 2013; Martínez-Sierra 2009). This paper first attempts a critical overview of Jodáowiec’s (1991a), Curcó’s (1995, 1996a, b, 1997) and Yus’s (2008, 2012, 2013) relevancetheoretic taxonomies of jokes. This is then backed up by a quantitative analysis, illustrating how the mechanisms discerned by the above mentioned accounts appear in a randomly selected set of jokes. In other words, the objective of this analysis is to seek empirical support for the theoretical accounts. The results suggest that ambiguity, conveyed on explicit and implicit levels, is the most frequently attested feature of jokes when analysed according to the three classifications. It is therefore shown that there is a single mechanism uniting the categorisations.

2. Defining a joke One of the possible ways to define a joke is in terms of a two- or three-part structure. Hockett ([1972] 1977) claims that the text of a joke is composed of a ‘build-up’ and a ‘punch’. Many authors have endorsed this view, albeit with slight changes in the terms used for the two parts. For example,

224

Chapter Eleven

Sherzer (1985) speaks of the ‘set-up’ and the ‘punch line’, whereas Attardo (1994) distinguishes the ‘the narration/presentation’ and the ‘dialogue’. A tri-partite division of jokes, in turn, has been advocated by Attardo and Chabanne (1992), who define a joke as a text type which comprises a narrative, a dialogue and a punchline. How jokes are described in terms of constituents is germane to the explanation of the way in which humorous effects emerge. A two-part joke structure is central to the joke taxonomies proposed by Yus, Jodáowiec and Curcó, who concentrate on different techniques employed by the humorist to evoke amusement on the hearer’s part.

3. Jodáowiec’s classification of verbal jokes Jodáowiec (1991ab) champions a two-fold classification of verbal jokes, comprising ambiguity jokes and implicit import jokes. Her two types vary in “formal properties and stages of interpretations involved, the presence of certain elements in the text, and ‘felicity conditions’” (Jodáowiec 1991b, 91-92). Joke type 1: Ambiguity jokes Ambiguity-based jokes create humorous effects based on an ambiguous language expression. The hearer has to be able to identify the duality within a linguistic material and to follow the relevance-driven comprehension procedure in switching from one sense to the other. We can note here, anticipating somewhat the discussion below, that Yus likewise espouses a similar view in this respect, claiming that the manipulation of ambiguity resolution is a source of humour. In his model, which was developed some years later than Jodáowiec’s pioneering work, Yus demonstrates that not only the exploitation of disambiguation but also other stages of comprehension can lead to humour. Joke type 2: Implicit import jokes In implicit jokes it is essential ‘what is not explicitly said’. The sole felicity condition which has to be satisfied is that since the setting and the punchline make some assumptions manifest, the addressee has to infer these background assumptions to make the telling of the joke successful. As regards the mechanisms responsible for the creation of humorous effects in Jodáowiec’s ambiguity- and implicit import-based jokes, they are similar: there has to be a clash between the first interpretation extracted from the setting and another interpretation derived from the punchline. To

“To Classify or not to Classify?”

225

be more specific, the stages of comprehension may be less easily identifiable in ambiguity jokes than in implicit import jokes. Jodáowiec’s taxonomy is straightforward and there is a point of agreement with Yus’s (2008) later taxonomy, discussed below, as they adduce evidence that funniness can be created both explicitly and implicitly. However, a major problem with her account is the claim that it is the extraction of a vast array of weakly communicated assumptions that makes a joke potentially funny to the audience. To refute that claim, we can point to poetic uses of language, where the poet intends the reader to arrive at as many weak assumptions as possible (which may differ from audience to audience), yet no humorous effect necessarily obtains. To illustrate this shortcoming from a different angle, let us look at one of the jokes discussed by Jodáowiec under the label ‘ambiguity jokes’ in the 1991b paper: (1)

Henry was trying to sell his battered old car for £65. His friend Tom said he would pay 10% less than the price Henry was asking for the car. But Henry was not very good at figures so he said he would think about Tom’s offer. That evening when he was in his usual bar Henry asked the barmaid: ‘If I offered you £65 less 10% what would you take off’. The barmaid hesitated slightly, then replied: ‘Everything except my ear-rings’. (Jodáowiec 1991b, 242243) Weak assumptions: the price is ridiculously low, Henry is not bright, Tom is a bargain-hunter, etc.

It is legitimate to note that comic effects revolve around the pivot (cf. Hockett [1972] 1977), which in this example is the ambiguous phrasal verb take off with the meaning either to remove a piece of clothing or to deduct as discount. Most of the problem with Jodáowiec’s account lies in the fact that although ambiguity is responsible for the delivery of humorous effects, it is also necessary to extract weak assumptions. A different issue this problem entails is that the weakly communicated assumptions presented under joke (1) are not humorous at all. Hence, the explanation of comic effects in terms of weak implicatures seems illadvised and cannot satisfactorily account for ambiguity-driven jokes. The funniness of both implicit import jokes and ambiguity jokes depends on weak implicatures and thus the two groups share a single mechanism. In Jodáowiec’s comprehension schema concerning implicit import jokes, the addressee is supposed to derive two interpretations, extracted on

226

Chapter Eleven

the basis of two contexts made manifest from the setting and the punchline, respectively. The two interpretations have to be in a clash to yield the intended effects. I claim that Jodáowiec’s account is further undermined by the pattern of comprehension she applies to both types of jokes. She claims, first, that the two interpretations derived from the setting and the punchline are in opposition/clash and, second, that the interpretation from the setting is replaced by the one derived from the punchline. When one studies the jokes presented in her papers, however, it can be noticed that these two interpretations are not in fact in opposition. Although they are two readings derived from the two parts of a text, one is not replaced with the other. The transition/switch from the first reading to the second/humorous one helps the hearer arrive at the humorous resolution and establish overall relevance.

4. Curcó’s classification of humorous texts Curcó (1995, 1996a) puts forward a three-fold classification of humorous texts, with jokes being at the centre of her attention. The bedrock for her taxonomy is the interdependence between perception and manipulation of the incongruous and inferential computations made by the hearer. In her papers, she touches upon issues of explaining the complex nature of jokes by employing the relevance-theoretic toolkit of mutual manifestness and cognitive environment, which will not be discussed in the present paper. Instead, her three mechanisms are presented below. Mechanism 1: Conflicting1 propositional forms Curcó resorts to the concept of two contradictory propositional forms which, with the use of language, are conveyed implicitly. When they are delivered on the explicit level, no humorous effect arises. The first assumption is a strongly implicated premise which clashes with the second assumption made manifest from either the context or the explicit content, as in the following example: (2)

1

There is something tragic about the enormous number of young men there are in England at the present moment who start life with a perfect profile and end up by adopting some useful profession (Oscar Wilde).

The label ‘conflicting’ can be alternatively replaced with ‘contradictory’, ‘clashing’, or ‘discrepant’.

“To Classify or not to Classify?”

227

(a) A useful profession is a cause for congratulation (b) A useful profession is a cause for commiseration (Curcó 1995, 29) Curcó (1995) argues that a strongly manifest assumption as implied by the humorist is supplied first (2a) and it can be speculated that this interpretation is to be recovered from the initial part of a text and is used as a premise. Propositional form (2a) is then extracted from the context of interpretation. To me, a proposition (2a) is naturally communicated at the first stage of a comprehension process, which is then in opposition to the second proposition (2b). Unfortunately, there are many unresolved issues in Curcó’s paper, for instance how a relevant interpretation is achieved on the audience’s part. Curcó’s claim of two propositional forms as a source of humour is on a par with Yus’s and Jodáowiec’s accounts. Mechanism 2: Foreground and background swapping: a shift in relevance search fields Curcó (1995, 31) conceptualizes this mechanism in the following way: “a hearer may be led to expect relevance in a given direction and suddenly discover some other unpredicted way in which utterance achieves it.” There is a dichotomy of implications on the scale of anticipatory hypotheses which raise contextual effects, viz. ‘foreground implication’ (relevant in its own right) and ‘background implication’ (not relevant in its own right, but raising relevant questions). Also, these two assumptions divide one’s cognitive environment into two fields, i.e. foreground and background search fields. Mechanism 3: A clash between the expectations of the way in which upcoming material will achieve relevance and the way in which it actually does This mechanism, added in Curcó (1996), may, but does not necessarily, result from the mixture of the first two mechanisms. In the same vein as the other two, this mechanism is also connected to the recovery of two implicit assumptions. Having looked at all three of Curcó’s mechanisms, we can conclude that she always speaks of two assumptions/discrepant propositional forms which are a part of the implicit import. To better capture my point, let me emphasise that I agree with Dynel (2012), who argues that the three mechanisms proposed by Curcó can be subsumed under a single category in which the hearer extracts the key assumption and the target one.

228

Chapter Eleven

5. Yus’s classifications of jokes Yus (2003, 2004, 2008, 2012, 2013) addresses many humour-specific issues but due to space limitations I cannot discuss every aspect of his approach; special attention will be devoted here to selected claims which are crucial for my critical comments. Yus (2008) advances a four-fold classification of jokes based on the explicature/implicature dichotomy, including jokes in which explicit interpretation is questioned, jokes in which the explicit content clashes with contextual information, jokes in which implicated premises and implicated conclusions are at work, and jokes which target background encyclopaedic assumptions. In addition, Yus (2013) provides the ‘Intersecting Circles Model’ embracing seven different strategies performed by the communicator which can be manipulated in order to create humorous effects on the part of the hearer.

5.1 Yus’s four-fold classification2 Joke type 1: Jokes in which explicit interpretation is questioned The fundamental tenet underlying this type of jokes is that they create humorous effects through the humorist’s manipulation of the interpretive steps taken by the hearer, as predicted within the theory of relevance. These inferential phases include reference assignment, ambiguity resolution and conceptual adjustment3. Jokes from the first group reflect the MGI/SCI pattern of comprehension, which can be compared to Suls’ (1972) two-stage model. The first part of a joke, labelled as the multiplegraded interpretations (MGI) part of a joke, communicates the first highly accessible, but ultimately incorrect interpretation. When the speaker creates a cognitive dissonance, the hearer processes the second part of a joke, referred as the single-covert part (SCI), which makes the second interpretation highly relevant in the context. It can be shown that jokes classified as belonging to the first type can also be found under the fourth category of jokes, where funniness is based on targeting background stereotypical assumptions. The fourth group of jokes is like an umbrella category encompassing an excessive number of 2

Note that three of the types of jokes (the first, third and fourth) were presented earlier in Yus (2003), although he did not there put forth a fully-fledged classification. 3 The choice of a particular inferential path depends on specific requirements of an utterance, for example when a joke or any utterance in general contains an ambiguous term/phrase, the hearer, to achieve a relevant output, has to resolve potential ambiguities.

“To Classify or not to Classify?”

229

jokes, raising the question of what defining criterion must be satisfied in the processing of a joke for it to be assigned to the first group but not at the same time ascribed to the fourth type. The first group reflects the incremental interpretation in which the first reading is invalidated and replaced with the second one. This cannot be said about jokes addressing stereotypes, since they come in different structures and the invalidation of the first relevant interpretation is not so obvious. Moreover, there are sexually/culturally themed jokes which are funny when implicatures are derived. Treating the stereotype-based jokes as a separate category, i.e. a fourth group, adds complication to the otherwise straightforward classification put forward by Yus: the first three groups are distinguished on the basis of criteria associated with processing, whereas the fourth one is related to the theme of a joke. The boundaries between joke types therefore become fuzzy and a joke cannot be unambiguously assigned to one group. Another problematic area concerns the MGI/SCI pattern of comprehension. In every joke, a first interpretation has to be invalidated and replaced with a second one to make the telling successful. As noted by Dynel (2012), not every joke presented in Yus’s contribution deals with the cancellation of the first interpretation. I have also found that there is a group of jokes in which two interpretations are equally important for the derivation of humorous effect: (3)

An English teacher wrote the words, “Woman without her man is nothing” on the blackboard and asked the students to punctuate it so that it made sense. The boys wrote: “Woman, without her man, is nothing.” The girls wrote: “Woman! Without her, man is nothing.” (Weeks 2011, 10-11)

There are two possible ways in which a joke achieves its relevance: first, by presenting boys’ punctuation of a sentence ([woman] [without her man] [is nothing]), and second, when girls’ view is presented ([woman!] [without her] [man is nothing]). The humorist plays with a grammatical arrangement of sentence constituents. It shows that there is no division of a text of a joke into two parts from which two interpretations are arrived at. Along the same lines, Yus (2008) claims that the manipulation of conceptual adjustment can produce comic effects. He presents a joke in which the addressee has to adjust a concept encoded by the noun shark to denote lawyers.

230

(4)

Chapter Eleven

Q: Why won’t sharks attack lawyers? A: They don’t attack animals of their own species. (Yus 2008, 150)

In this joke, there is no explicit interpretation which is invalidated in the course of foregrounding of a humorous one. I suppose that only one relevant interpretation can be derived in which the hearer attributes some encyclopaedic features of sharks to lawyers, for instance sharks are dangerous to humans. It is also doubtful whether jokes presenting social stereotypes which involve metaphorical conceptualisations can depend only on conceptual adjustment. This case is exemplified by a joke (5) below: (5)

There once was a man who owned a sausage factory, and he was showing his arrogant son around his factory. Try as he might to impress his snobbish son, his son would just sneer. They approached the heart of the factory, where the Father thought “This should impress him!” He showed his son the machine and said, “Son, this is the heart of the factory. This machine here we can put in a pig, and out come sausages.” The son, unimpressed, said “Yes, but do you have a machine where you can put in a sausage and out comes a pig?” The father, furious, said, “Yes son, we call it your mother.” (Yus 2008, 150)

The humorous effect is created through ambiguity in which there is a metaphorical sense of the two words sausage and pig. In my opinion, a joke in which there is a play between literal and metaphorical meanings creates ambiguity. In essence, the point I am making here is that although many jokes classified to the first group reflect the MGI/SCI strategy, it is not always the case. It happens frequently that only one interpretation is extracted by the addressee. Joke type 2: Explicit interpretation clashing with contextual assumptions Here a humorous effect is created through the clash between an explicit interpretation and contextual assumptions. Yus (2008) endorses Curcó’s (1995, 1996a, b, 1997) main lines of analysis, who calls these competing assumptions the key assumption and the target one. The former assumption is a strongly implicated premise whereas the latter is a weakly communicated assumption from the context becoming strongly manifest as a result of the contradiction.

“To Classify or not to Classify?”

(6)

231

A woman walked up to a little old man rocking in a chair on his porch. “I couldn’t help noticing how happy you look,” she said. “What’s your secret for a long happy life?” “I smoke three packs of cigarettes a day,” he said. “I also drink a case of whiskey a week, eat fatty foods, and never exercise!” “That’s amazing,” the woman said. “How old are you?” He thought for a moment, and replied, “Twenty-six.” (a) Target assumption: It is impossible to live very long with such bad habits. (b) Key assumption: It is possible to live very long with such bad habits (Yus 2008, 151-152)

Example (6) should suffice here to show that the inferential procedure operating in the second type of jokes corresponds to the MGI/SCI schema of comprehension. Yus (2008) claims that the key assumption, being an explicit one, is derived from the initial part of a joke (the MGI, 6b). A set of assumptions made manifest from the first part includes: The very fact that an old man who drinks whiskey, eats junk food and never exercises enjoys a good life is an exemplar of having bad habits. But, upon the punchline, the target assumption, which was irrelevant and treated as not worth of being foregrounded, is subjected to a “sudden and drastic change” (Curcó 1996b, 61). Then, the assumption derived from the context is: in fact it is impossible to lead a healthy and long life when one has bad habits. In this discussion, I would like to demonstrate that the first type of jokes is a broad category which encompasses not only jokes in which a stage of interpretation is exploited but also jokes which create humorous effects via a clash between two assumptions. Due to this fact, these jokes favour the same pattern of comprehension. Certainly, there is some sort of hierarchy, so that a joke has to be told in a specific way in which the addressee may arrive at two interpretations. There is also a subtler point to cover: Yus claims that the target assumption from joke (6) is inferred from the hearer’s encyclopaedic knowledge. At this point, one has to decide to which group this problem can be assigned: in particular, it seems that either Yus has not fully endorsed Curcó’s claim or (which appears more probable to me) he has purposefully altered Curcó’s claim for the sake of his own research.

232

Chapter Eleven

Joke type 3: Implicated premises and implicated conclusions are at work As regards the third type of jokes, they generate humorous effects by the hearer’s extraction of implicatures which are not prompted by any material included in the initial part of the joke. Yus (2008) emphasises that it should be enjoyable for the hearer to find congruence in the text beyond the explicit level. (7)

A dietician was once addressing a large audience in Chicago. “The material we put into our stomachs is enough to have killed most of us sitting here, years ago. Red meat is awful. Vegetables can be disastrous, and none of us realizes the germs in our drinking water. But there is one thing that is the most dangerous of all and all of us eat it. Can anyone here tell me what lethal product I’m referring to? You, sir, in the first row, please give us your idea.” The man lowered his head and said, “Wedding cake?” (a) Implicated premise: One eats wedding cake if one has just got married. (b) Implicated conclusion: Getting married is very bad for your health. (Yus 2008, 153)

The definition of this group of jokes involves a certain methodological problem, connected to the inferential path favoured by the addressee. It is clear that the hearer has to derive implicit information as intended by the humorist to get this joke. I see this example just as another case of the MGI/SCI pattern of interpretation. Yus intended his classification to be concentrated around the last line of a joke, i.e. the punchline, but this idea does not seem to work. If we analyse the joke above in terms of a two-fold structure, it is evident that the first relevant interpretation includes assumptions regarding lethal products, i.e. arsenic, which are subverted by the punchline, which makes manifest assumptions concerning marriages, wedding cakes and their impact over men’s health. Throughout this section, what I am seeking to demonstrate is that that given any (or almost any) joke which is classified to the first, second and third type, the hearer has to derive two interpretations and hence the MGI/SCI schema can be followed. Joke type 4: Targeting background encyclopaedic assumptions The construct of the fourth type, contingent on the idea of targeting broad encyclopaedic information which can be reinforced, contradicted, eliminated or reminded, is beset by myriad problems.

“To Classify or not to Classify?”

233

First and foremost, it is an undeniable fact that the vast majority of jokes address different stereotypes, be it sexual, religious or cultural. In fact, almost every individual’s characteristic can be thought of in terms of a stereotype, for example: a dumb blonde or a beer-loving German. Moreover, Yus’s (2008) classification was supposed to be based on the explicature/implicature distinction, yet this criterion is no longer functional when he proposes a stereotype-based joke category, with this fourth type of jokes not based on any inferential procedure (this has also been pointed out by Dynel 2012). The second problem deals with the division between stand-up comedies and short jokes. Again, without going into an elaborate discussion about “short” jokes and those humorous texts which are longer (i.e. narrative) and which at the same time may fall outside the scope of his classification, Yus states that short jokes “reinforce it [stereotypical information], contradict it and eliminate it, or simply remind the audience of the collective quality of this [stereotypical] information” (Yus 2008, 153), which is then negated on the following page: “short jokes ... do not exhibit a private/collective duality or attempt to challenge stereotypes” (Yus 2008, 154). In my opinion, short jokes may address stereotypes in a variety of ways, for instance by challenging, reinforcing or eliminating one’s encyclopaedic assumptions, to mention but a few. Another troublesome aspect of this joke/stand-up comedy bifurcation is that Yus (2008) presents only cultural stereotypes as a source of humorous effects in stand-up performances. A query arises as why other stereotypes are not included in his classification.

5.2 Yus’s ‘Intersecting Circles Model’ Yus (2013) attempts a broad picture of joke comprehension, singling out interpretive procedures which account for humorous effects. The main difference between the previous taxonomy and current one is that in the earlier version Yus concentrated on the explicit/implicit dichotomy and stereotypes whereas in the ‘Intersecting Circles Model’ he also pays heed to interferential phases, but argues that they work in parallel with cultural information and make-sense frames. For the sake of his taxonomy, Yus coins the term ‘make-sense frame’ which comprises three types of information, i.e. word-associated schemas (encyclopaedic features), sequence-associated scripts (actions) and situation-associated frames (prototypical situation). Every mechanism is the result of a combination of a make-sense frame, a cultural frame and a path of comprehension manipulated by the humorist. Sometimes make-sense frames and cultural frames overlap.

234

Chapter Eleven

Type 1: Make-sense frame + cultural frame + utterance interpretation This joke category creates humorous effects via the hearer’s activation of a make-sense frame and a cultural frame together with the communicator’s manipulation of one of the inferential phases. Let us consider the example below to draw attention to the way in which humorous effects are created: (8)

A young couple got married and left on their honeymoon. When they got back, the bride immediately called her mother. Her mother asked, “How was the honeymoon?” “Oh, mama,” she replied, “the honeymoon was wonderful! So romantic...” Suddenly she burst out crying. “But, mama, as soon as we returned Sam started using the most horrible language... things I’d never heard before! I mean, all these awful 4-letter words! You’ve got to come get me and take me home. ... Please mama!” “Sarah, Sarah,” her mother said, “calm down! Tell me, what could be so awful? What 4-letter words? You must tell me what has you so upset... Tell your mother these horrible 4-letter words!” Still sobbing, the bride said, “Oh, mama... words like DUST, WASH, IRON, COOK...!” (Yus 2013)

On hearing joke (8), the process of interpretation starts with the identification of a make-sense frame for honeymoon which, as argued by Yus, also is tied up with cultural information. In most societies, a honeymoon is perceived as an enjoyable time to celebrate one’s marriage. Then, there is a different make-sense frame activated for sex roles and wicked husbands. This frame, in turn, makes the audience find referents for the expression 4-letter words, which is understood as ‘swearwords’. Then, this state of affairs is invalidated by the punchline which communicates a cultural stereotype of a woman as a housewife. Type 2: Make-sense frame + cultural frame The second type capitalises on a play between make-sense frames and cultural ones. It is maintained that the recovery of the two types of frames is connected to the disconfirmation of a make-sense frame and the identification of a cultural frame. The joke (9) falls within the scope of this group of jokes: (9)

Teacher: “Everything you do is wrong. How can you expect to get a job when you leave school?” Pupil: “Well, sir! I’m going to be a TV weatherman.” (Yus 2013)

“To Classify or not to Classify?”

235

In the joke under analysis (9), the set-up part of a joke forces the hearer to activate the make-sense frame for education and school so that the phrase do wrong is understood as something negative with reference to finding a job. However, the punchline leads to the recovery of cultural information concerning weathermen who are not very good at weather forecasts. Type 3: Make-sense frame + utterance interpretation For the sake of humorous effects, the speaker manipulates an inferential step taken by the hearer in which the first interpretation is invalidated and replaced with the second interpretation extracted from the punchline. Together with the exploitation of the first accessible interpretation, the hearer’s construction of a make-sense frame from the setting should also be subverted when the subsequent potion of the text is communicated. According to RT, the communicator possesses a mind-reading ability to predict possible interpretations represented in the hearer’s mind. This is successfully used in joke comprehension, since the humorist should be able to predict all interpretations so that the humorous resolution is not at stake. The example below illustrates the case under consideration: (10)

A customer enters a sports shop and asks the man in charge: “Excuse me, but do you have the balls to play tennis?” “Yes, of course!” replies the man. The customer says: “Then, tomorrow at 10.”

On hearing joke (10), the audience identifies the make sense frame for sports shop and hence extracts the first highly salient and accessible interpretation that the customer wants to buy balls to play tennis. When the discourse proceeds, another interpretation is formulated in which the phrase to have the balls is understood idiomatically and which at the same time invalidates the initially activated make-sense frame. In this particular case, Yus (2013) does not explain which inferential procedure is manipulated by the humorist. He only shows how a makesense frame is created when processing the setting and thus which set of assumptions is extracted and then how the hearer formulates a different interpretation from the punchline. In other words, I argue that a sudden change of assumptions, i.e. the transition from the setting to the punchline, is satisfied for the humorous resolution in every joke so that the hearer finds the relevance in the first portion of the text and then tries to arrive at the second interpretation which suits the whole text.

236

Chapter Eleven

Type 4: Make-sense frame The fourth group of jokes is based on the activation of make-sense frames so that the humorist attempts at incongruity, which emerges in the course of interpretation. Type 5: Cultural frame + utterance interpretation Humorous effects in this group are created via the derivation of cultural frames and the manipulation of one of the inferential phases, i.e. ambiguity resolution, reference assignment or conceptual adjustment. Type 6: Cultural frame As it happens frequently, many jokes depend on the strengthening or manipulation of cultural information concerning humans roles, jobs or professions. As a result, this group is comparable to the fourth type of jokes in Yus’s (2008) previous research. Type 7: Utterance interpretation Other jokes achieve humorous effects when a particular inferential procedure is manipulated. Since this joke category is the same as the first type of jokes in Yus’s (2008) four-fold classification, it poses a similar set of problems which are not reiterated here.

6. Quantitative analysis Having completed this overview of relevance-theoretic classifications of jokes, we will now move on to presenting the results of a small quantitative study performed in this light.

6.1 Methodology The sample study consists in 54 canned jokes drawn at random from a 4 collection of jokes . The motivation behind the choice of these particular jokes was the author’s firm conviction that some, if not all, should bring about humorous effects upon the audience. The very fact that the jokes are culled from a joke collection may preserve the data from being criticised for the author’s subjective assessment of humorousness of the jokes, or lack of it. The prime objective of this small-scale study was to detect which pragmatic mechanism/type of joke is most prevalent. In these 4

The set of jokes chosen for my analysis can be accessed via my page on academia.edu: http://uw.academia.edu/MagdalenaBiegaj%C5%82o

“To Classify or not to Classify?”

237

collections, a vast majority of jokes are classified by topic to predetermined categories, be it blackmail, city life, dating, etc., whereas other jokes can be assigned to different types. The jokes were chosen randomly in order to avoid a bias toward one inferential process as predicted in the relevance-based theory. The results from this quantitative analysis and the above critical comments upon the three classifications of jokes may prove useful to those who attempt to advance their own relevance-theoretic classification. The results presented here may come under criticism, of course, for the relatively small number of jokes chosen for this study, with their credibility perhaps being challenged on this basis. However, be that as it may, I claim that more examples would not negate or reverse the outcome.

6.2 Analysis I will first present the general findings from the quantitative analysis, then discuss a problem encountered while distinguishing the pragmatic mechanism at work. Lastly, I will focus on analyzing the data with respect to the three accounts of jokes presented in sections 6.3-6.5. More generally, the results demonstrate that jokes based on ambiguity constitute the most frequent type, which is supported by the fact that in all the classifications presented the manipulation of ambiguity resolution is seen as one of the mechanisms. According to Yus, ambiguity resolution is exploited when an explicit interpretation is questioned by the speaker. In Jodáowiec’s parlance, ambiguity originates on the explicit level of syntax, lexis and phonology. In Curcó’s terms, ambiguity/incongruity arises on an implicit level as a result of two clashing propositional forms. My data also indicates that ambiguity is the most frequently attested mechanism. Of the 54 jokes surveyed, in 98% of cases it was possible to single out an inferential procedure responsible for the perceived funniness. In my opinion, the remaining 2% of jokes are ones containing three analogous events (cf. Chiaro 1992). That is to say, the very presence of three punchlines renders the selection of pragmatic mechanisms difficult: (8)

The three amateur archaeologists were exploring a temple deep in the jungles of the Amazon when they were captured by a tribe of hostile natives. “You have profaned the ancient temple of our fertility goddess,” exhorted the chief. “For that, you must surrender your manhood.” Approaching the first man, he asked, “What is your job in your native land?”

238

Chapter Eleven

Puzzled the man said, “I- I’m a butcher, sir.” The chief ordered, “Then we shall chop it off!” Approaching the second man, he said, “What is your job in your native land?” Drawing himself up proudly, he said, “I’m a fire fighter.” “Then we shall burn yours off,” he said, and went to the third man. “And what is your job in your native land?” Having seen what was ordered for his friends, the man said confidently, “I’m a lumberjack Your Lordship.” “Very well,” he said. “For you, we shall jack it off!” (Rovin 1987, 29) Here there are three symmetrical events in which the three characters perform a similar action, with the third conveying the most important punchline. Since the punchline is not only in the final position (as there are three of them), whereas Yus, Jodáowiec and Curcó all favour a bi-partite structure of a joke text, it is hard to classify the pragmatic mechanism involved here. To study the data with regard to Jodáowiec’s classification was the easiest task, owing to the fact that she propounds the most straightforward model. I had to set aside the problem of ambiguity jokes for whose successful delivery it is also necessary to retrieve weak assumptions, because otherwise I would be forced to state that all ambiguity jokes may also be classified as implicit import jokes but not vice versa. As for ambiguity jokes, I set the boundaries that only phonological, lexical and syntactic ambiguity fall under this group. As a result, I found that humorous effects result from ambiguity conveyed explicitly in 79% of cases. It is a separate issue and it may open a further area of investigation where one draws a boundary line, using the conceptual tools, between ambiguity communicated on an explicit level and an implicit one5. As for implicit import jokes, they were also frequent, constituting 21% of all cases. I am well aware that some readers may question the way I investigated the data with reference to Jodáowiec’s taxonomy, but I believe that was a necessary move. The most demanding and troublesome case was to verify Curcó’s classification of humorous texts with respect to the data. Given the fact that her taxonomy is based on different ways incongruity is achieved by the speaker and how two forms clash, I found that the first mechanism was the most prevalent (58%), followed by the second and third mechanisms 5

By implicit ambiguity, the author means the cases of ambiguity which are not associated with syntactic or lexical ambiguities, but which arise in the process of recovering implicatures on the part of the hearer.

“To Classify or not to Classify?”

239

(each 21%). I am convinced that the data obtained from Curcó’s taxonomy is the least satisfactory because the three types of mechanisms can be put under one label “Two discrepant forms”. In the light of this argument, the quantitative analysis from this section is itself replete with shortcomings which echo those of Curcó’s categorization. As regards Yus’s classification (2008) of jokes, the fourth type of jokes has not been included in my analysis, since, as argued above, many jokes create humorous effects through presenting a cultural/social/sexual stereotype. With reference to this, the targeting of broad encyclopaedic information has been found in 66% of the jokes. The data also corroborates my claim concerning the first type of jokes, that it constitutes a vast category in which the abovementioned model of comprehension (MGI/SCI) is generally followed, although, as I tried to show, that is not always the case. Empirically, approximately 77% of the jokes follow this schema of comprehension. In the same vein, I argued in section 3 that jokes which belong to the second or third type may, and actually usually do, follow the MGI/SCI pattern, so one and the same joke can be assigned to two groups. The second type of jokes was much less frequently encountered, at 18%. The third type was extremely rare, found in only 5% of cases. Yus’s (2013) Intersecting Circles Model bears resemblance to his earlier research (Yus 2008), however, he introduces the two new concepts of a make-sense frame and a cultural frame in explaining the emergence of humorous material. The main problem exhibited in the previous work (2008) and which is repeated in this model is the presence of overlapping joke categories. In explaining a joke, Yus (2013, 31) does not consider the manipulation of an interpretive stage even though he claims that it is responsible for a comic effect, and hence it may be stated that this joke can also fall within joke type 4. Another problem, connected to the first shortcoming, is that Yus changes his perspective on the manipulation of interferential procedures, first asserting that ambiguity resolution, reference assignment or conceptual adjustment have to be exploited, then later applying a basic explanation of the occurrence of humorous effects as simply when incongruity emerges at the punchline stage, so that the audience has to derive two interpretations. Let me now present the results from the quantitative analysis and discuss them with regard to my data. There are several interesting claims to be made. First, the notion of a make-sense frame is crucial for the explanation of how incongruity emerges on the part of the hearer and in fact it enables us to provide a broader picture of the cognitive mechanisms employed by the audience to

240

Chapter Eleven

understand a joke and accordingly to entertain the overall relevance of a humorous text. Second, the above-mentioned problematic joke in which three simultaneous punchlines are present was here not excluded from the quantitative analysis, since the Intersecting Circles Model provides appropriate tools to explain such comic effects. In other words, I argue that it should be assigned to the fourth category of jokes because it requires the activation of three different make-sense frames for butchers, fire fighters and lumberjacks. Third, my results show that there are two categories which were frequent: the third group (based on make-sense frames and utterance interpretation) and the seventh group (based on utterance interpretation) and this fact indicates that the manipulation of an inferential procedure is again the leading way of achieving humorous effects.

7. Conclusions The present paper sought to achieve two purposes. First, I critically reviewed the three relevance-theoretic classifications of jokes, viz. Jodáowiec’s (1991a, b), Curcó’s (1995, 1996a) and Yus’s (2008, 2013), pointing out that each of these classifications suffers from rigidity, mostly because categories meant to be distinguished on the basis of a single mechanism also in fact involve other mechanisms at work at the same time. Second, I looked at a small set of empirical data with regard to the three classifications. The sample study serves as an introduction for further research. Probably taking a larger set of data would enable a researcher to come up with their own classification, which would evade the above mentioned issues. For the sake of this paper, I was not willing to take that risk in order to present my classification based on 54 jokes. In my opinion, the most promising results have been obtained from the analysis of Yus’s (2008, 2013) categorisations, since the other two are much weaker with regard to their complexity. On a critical note, Jodáowiec’s (1991a) taxonomy is the earliest attempt at putting jokes into groups using the RT framework, and as such, it is the most basic one with merely two types being singled out. As a result, there is not much work to be done to scrutinise the data using her classification, and what is even worse, the fact that in both types some kind of ambiguity needs to emerge does not contribute new ideas to the development of humour research, besides the one that has been taken by Yus (2008), who developed the classification. As regard Curcó’s (1995, 1996a) taxonomy, the study of the data using her categorisation also posed two knotty problems: first, all

“To Classify or not to Classify?”

241

humorous mechanisms boil down to one type, which resides in extracting two clashing assumptions, and second, these mechanisms originate from implicatures. For the reasons which I explicitly stated above, in spite of the critique I advanced regarding the three joke classifications, I have found that they may be plausibly applied to the quantitative analysis of jokes, albeit with some doubts remaining as to Curcó’s pragmatic mechanisms. On the whole, the critical comments concerning the validity of the three classifications presented in sections 3-5 are indirectly confirmed by the results from my small-scale study. The findings from the quantitative analysis give rise to an interesting query: is it really an indispensable feature of all humorous texts, or only jokes, to be subscribed to a general model of comprehension in which incongruity is conveyed explicitly and implicitly? Let us now close by considering the question posed in the title of this paper: “To classify or not to classify?” Given the many methodological shortcomings inherent in the existing relevance-theoretic classifications of jokes, rather than attempting such exhaustive classificatory frameworks in the first place, perhaps it should be sufficient to observe that Relevance Theory provides many theoretical tools capable of satisfactorily accounting for the myriad pragmatic processes that are at work in jokes, even though these tools may not provide an adequate basis for an exhaustive classification.

References Attardo, Salvatore. 1994. Linguistic theories of humor. Berlin, New York: Mouton de Gruyter. Attardo, Salvatore and Jean-Charles Chabanne. 1992. Jokes as a text type. Humor 5, nos. 1-2: 165-176. Chiaro, Delia. 1992. The language of jokes: Analysing verbal play. London and New York: Routledge. Curcó, Carmen. 1995. Some observations on the pragmatics of humorous interpretations. A relevance theoretic approach. UCL Working Papers in Linguistics 7: 27- 47. —. 1996a. The implicit expression of attitudes, mutual manifestness, and verbal humour. UCL Working Papers in Linguistics 8: 89-99. —. 1996b. Relevance Theory and humorous interpretations. In Automatic interpretation and generation of verbal humor, edited by J. Hulstijn and A. Nijholt. Enschede: University of Twente.

242

Chapter Eleven

—. 1997. The pragmatics of humorous interpretations. A relevance theoretic approach. Ph.D. diss., UCL Department of Phonetics and Linguistics, London. Dynel, Marta. 2010. How do puns bear relevance. In Relevance Studies in Poland, edited by M. Kisielewska-Krysiuk, A. Piskorska and E. Waáaszewska. Warszawa: Wydawnictwa Uniwersytetu Warszawskiego. —. 2012. What RT cannot do, IR can – On the incremental interpretation of jokes in (non)relevance-theoretic terms. In Relevance: More than understanding, edited by E. Waáaszewska and A. Piskorska. Newcastle: Cambridge Scholars Publishing. Groefsema, Marjolein. 1998. Processing for Relevance. Revista Alicantina de Estudios Ingleses 11: 95-116. Hockett, Charles. [1972] 1977. Jokes. In Studies in linguistics in honor of George L. Trager, edited by E. Smith. New York: Mouton. Reprinted in The view from language: Selected essays 1948-1974. Athens: University of Georgia Press. Jodáowiec, Maria. 1991a. The role of relevance in the interpretation of verbal jokes: A pragmatic analysis. Ph.D. diss., Jagiellonian University. —. 1991b. What makes jokes tick. UCL Working Papers in Linguistics 3: 241-253. —. 2008. What‘s in the punchline? In Relevant worlds: Current perspectives on language, translation and Relevance Theory, edited by E. Waáaszewska, M. Kisielewska-Krysiuk, A. Korzeniowska and M. Grzegorzewska. Newcastle: Cambridge Scholars Publishing. Larkin Galiñanes, Cristina. 2000. Relevance Theory, humour, and the narrative structure of humorous novels. Revista Alicantina de Estudios Ingleses 13: 95-106. Martínez-Sierra, Juan José. 2009. The relevance of humour in audio description. Intralinea 11. Rovin, Jeff. 1987. 1001 Great jokes: From the delightfully droll to the truly tasteless. New York: New American Library. Sherzer, Joel. 1985. Puns and jokes. In Handbook of discourse analysis, edited by T. van Dijk. London: Academic. Sperber, Dan and Deirdre Wilson. 1986 [1995]. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. Suls, Jerry. 1972. A two-stage model for the appreciation of jokes and cartoons: An information-processing analysis. In The psychology of humor, edited by J. Goldstein and P. McGhee. New York: Academic Press.

“To Classify or not to Classify?”

243

Weeks, W. 2011. The Great Big Book of Hilarious Jokes [e-book] ( 27 December 2012). Wilson, Deirdre. 2002. Relevance Theory: A tutorial. In Proceedings of the Third Tokyo Conference on Psycholinguistics, edited by Y. Otsu. Tokyo: Hituzi Shobo. Wilson, Deirdre and Dan Sperber. 2004. Relevance Theory. UCL Working Papers in Linguistics 14: 249-287. Yus, Francisco. 2003. Humour and the search for relevance. Journal of Pragmatics 35, no. 9: 1295-1331. —. 2004. Pragmatics of humorous strategies in El club de la comedia. In Current trends in the pragmatics of Spanish, edited by R. MárquezReiter and M. E. Placencia. Amsterdam: John Benjamins. —. 2008. A relevance-theoretic classification of jokes. Lodz Papers in Pragmatics 4, no. 1: 131-157. —. 2012. Relevance, humour and translation. In Relevance Theory: More than understanding, edited by E. Waáaszewska and A. Piskorska. Newcastle upon Tyne: Cambridge Scholars Publishing. —. 2013. An inference-centered analysis of jokes: The Intersecting Circles Model of humorous communication. In Irony and humor: From pragmatics to discourse, edited by L. Ruiz Gurillo and B. Alvarado. Amsterdam and Philadelphia: John Benjamins.

PART V: MORPHOLOGICAL ISSUES AND LEXICAL PRAGMATICS

CHAPTER TWELVE BROADENING THE SCOPE OF LEXICAL PRAGMATICS: THE CREATION OF NEOLOGISMS IN TOPOSA MARTIN C. SCHRÖDER, SIL INTERNATIONAL AND VU UNIVERSITY AMSTERDAM, THE NETHERLANDS

1. Introduction 1.1 Overview A lot has been written from a relevance-theoretic perspective about literal use versus loose talk, especially approximation, metaphor, and hyperbole (e.g. Carston 1997; Wilson 2003/2004, 2006/2007; Sperber and Wilson 2002; Wilson and Carston 2007), and how comprehension of these requires lexical adjustment via the construction of ad hoc concepts. Far less has been written about the category of ‘neologisms and word coinages’ which are located somewhere between metaphor and catachresis1 (Wilson 2006/2007, 1.7)2 and might “provide further data for a theory of lexical pragmatics and shed some light on the nature of the mental mechanisms involved” (Wilson 2004, 346, repeated in Wilson and Carston 2007).

1

Catachresis is a figure of speech in which a familiar word is used in an unconventional way. 2 I am aware that this work is not a published article but a set of – very well written – lecture notes graciously provided to me and my colleagues by the author herself. (These lessons were also accessible on the UCL website for a while.) I have decided to make reference to them since the only other references to neologisms I could find in relevance-theoretic publications are in Wilson 2004 and Wilson and Carston 2007, and these lectures are far more specific.

The Creation of Neologisms in Toposa

247

I therefore want to examine how neologisms are formed and communicated, then supplement what we know about these processes in English with data from an ethnic group that has undergone massive changes in their language and culture over the last twenty-five years: the Toposa of South Sudan. In the early eighties, when my wife and I lived among them for several years, the Toposa were still an isolated tribe, but then civil war came and led to the internal displacement of many ethnic groups and massive migration into Toposa areas, the experience of modern warfare technology, coupled with a flow of Western aid and development, as well as a Bible translation project in combination with language development and a literacy project.3 All these changes brought with them a wide array of cultural contacts which introduced many new ideas that suddenly had to be expressed in Toposa.

1.2 Some introductory observations Before we begin to look at data, it might be helpful to consider a few preliminary issues. 1.2.1 The scope of the term neologism According to Wilson (2006/2007, 1.7), the creation of neologisms covers a range of situations “in which words are invented, blended or transferred from one syntactic category to another.” Looking at Wilson’s data, it seems that she thought mainly of innovation within one and the same language and cultural sphere. However, if we broaden the scope somewhat and ask in which contexts lexical innovation typically takes place, then the answer for communities like the Toposa must be that the creation of neologisms in the vast majority of cases is driven by influences “from outside”, i.e. a situation requires that a concept that is known and lexicalized in one language/culture needs to be expressed in another language/culture because it is not known and it has become necessary or desirable to be able to use that concept. In other words, cultural contact 3

My wife Helga and I lived with the Toposa, analyzed their language and helped them to develop a suitable orthography for use in schools and adult literacy, produced literacy materials and trained teachers in the use of these new materials. First we worked among the Toposa of Riwoto (1981-1984), later among displaced populations in other parts of Eastern and Western Equatoria (1986-1988), and after that with refugees in Kenya (1988-1991). Later I worked with the Toposa Bible translation project in Kitale/Kenya (1999-2003) and Entebbe/Uganda (2003-2006) as translation consultant.

248

Chapter Twelve

and rapid technological innovation brought from the outside are the main influences here. Considering the nature of neologisms from a slightly different angle, it appears that they come in two types: first, there are one-off word creations that remain just that: a flash in the pan. Such words may serve one purpose, and one purpose only. It may be that the new word communicates something particularly well in one situation, in that it is particularly suited in that context, or it adds a special effect, such as being funny or poetic. Secondly, there are neologisms which are created in some situation – for whatever reason – but they somehow stick and gain currency. Wilson also refers to both types, although most of her examples are of the short-lived type (ibid., 8), but I would like to go beyond this general observation and note that the more playful type of neologism which is a one-off occurrence is clearly more what we find in situations involving just one language, whereas the type that lasts is found in both monocultural and multicultural contexts, usually driven by a real need to bridge a lexical gap – usually accompanied by a conceptual gap.4 I think that it makes good sense to view all forms of lexical innovation, those that are short-lived and those that last, as well as those that are more creative-expressive and those that are dictated by some tangible need, as belonging to the same general phenomenon.5 However, the type of neologisms I want to concentrate on in this investigation is the one reflecting genuine innovation, which also tends to be the type that lasts. 1.2.2 The type of data used This paper draws from natural language data that reflect developments over the last three decades. I also have a large number of examples from the realm of Bible translation, where the Toposa translation team has 4

Of course, it is a fundamental tenet of RT that in any language only a fraction of the conceptual repertoire is lexicalized and that “there are a great many stable and effable mental concepts that do not map onto words” (Sperber and Wilson 1998, 189, italics in original). However, I am dealing with those instances here where some new idea exists that has not been lexicalized yet, but where there is a real need to do so as a way of referring to an invention or an innovative idea introduced by cultural exchange. This may lead to either using a placeholder term like whatnot, whatsit, gizmo, doodah, doohickey or whatchamacallit in a particular instance, or to the creation of a new lexicalized form. 5 As I have demonstrated elsewhere, there are close parallels between creating a neologism in communication within one and the same language and introducing a new term referring to a so far unknown idea in the translation process from one language to another (Schröder 2009).

The Creation of Neologisms in Toposa

249

conscientiously tried to craft new lexical entries in their effort to express biblical ideas which were foreign to their audience, but have decided for methodological reasons not to present these data here. We shall however briefly address some questions raised by these two different sets of data at the end of this study (see section 4).

2. Types of neologisms in Toposa For the reasons described in the introduction, Toposa has absorbed a large number of neologisms over the last few decades, and we shall now take a look at the different types of processes how the Toposa have created new words in order to cope with new ideas.

2.1 Derivation A common method of expressing new concepts in Toposa is derivation. This is probably not limited to Toposa but true of most languages, because it is the main function of derivation to encode new concepts. Where I expect languages to differ is in terms of which types of derivation a particular language may prefer to use in the creation of neologisms. When a new idea is expressed via derivation, a speaker takes known elements of words (morphemes) and combines them in novel ways to form a new word. In Toposa I have so far found the following derivational processes in conjunction with neologisms: a new combination of known roots and derivational affixes which works across word classes (generally from verb to noun), the change of a noun from one grammatical category to another (of which there are two subtypes: change in gender and change in noun class), and compounding. 2.1.1 New combination of known roots and derivational affixes Toposa can produce neologisms by using a known morpheme as the root and combining it with derivational affixes in novel ways. This way of forming new derivations is particularly common with instrumental nouns and agentive nouns, both of which employ highly productive derivational processes in the language,6 for example:7

6

Instrumental nouns are formed by using a gender-number prefix and the instrumental suffix -ete֔ , agentive nouns are formed by using a gender-number prefix, the derivational prefix ka- and one of the agentive suffixes -ani֔ ~ -aka֔ for –

Chapter Twelve

250 (1a) nya-ki-nok8 (1b)

(2a) (2b) (2c)

known: light sth (transitive verb, root *nok) nya-ki-nok-ete literally: something that lights/shines F/SG-DER-light-INS new senses: mirror, flashlight, headlight/headlamp nya-mojiri known: turn sth (transitive verb, root *mojiri ~ mojir) nya-mojir-ete literally: instrument for turning F/SG-turn-INS new senses: steering wheel, rudder nye-ka-mojir-ani literally: one who turns M/SG-DER-turn-AG:SG new senses: driver, captain, pilot

What happens in these derivations at the lexical-pragmatic level? First of all, there are several stages to consider – and these will be the same for all processes of how neologisms are created: We need to consider the processes that happen in the mind of the speaker who crafts a new word, as well as what happens in the mind of a hearer who hears a new word for the first time, and what happens in the minds of hearers once a new concept has been established and more people in a community have the new word in their vocabulary, i.e. it has entered the public lexicon. At this point we need to take a detour and cover some ground that I have found virtually nothing about in the literature. All extant studies in RT-based lexical pragmatics focus on concepts encoded by full words, not so much on the meaning encoded by morphemes such as roots and affixes and how best to differentiate between these types of information provided by these elements.9 However, it is probably correct to say that a root ATR nouns and -oni֔ ~ -oko֔ for +ATR ones. (For an overview of gender number prefixes in Toposa see Table 1 below.) 7 The spelling system for Toposa used here is the one which the Toposa community has chosen for writing their language. The main differences from IPA are the use of the grapheme for [݄] and < a࡮ e࡮ i࡮ o࡮ u࡮ > to represent the voiceless vowels [ ঀ eࡢ iࡢ oࡢ uࡢ ]. 8 This is the citation form for Toposa verbs. There are two morphological verb classes: one class takes the prefixes nya-ki-, the other class drops the ki- as shown in (2a). 9 One possible exception might be the unpublished work of Chun on morphopragmatic aspects of Korean (2007). In the wider field of pragmatics, research in morphopragmatics has been pioneered by Wolfgang Dressler, Lavinia Merlini Barbaresi, and Jörg Meibauer. However, these are mainly studies on German and Italian diminutives, augmentatives and intensifiers, i.e. elements which reflect speaker attitudes, whereas so far I have found no research on the fine-tuning that needs to take place in the understanding of word-forming elements

The Creation of Neologisms in Toposa

251

morpheme gives access to certain encyclopaedic assumptions which are common to all occurrences of that morpheme, although the morpheme by itself does not contain all the conceptual content that is encoded by the full lexical entry. Under this perspective, let us consider example (2c) in greater detail, i.e. how the various building elements contribute to the meaning of this lexical entry: nye-ka-mojir-ani M/SG-DER-turn-AGT:M/SG root morpheme *mojiri ‘turn sth’ conceptual meaning: something to do with ‘turning s.th.’

‘turner’

affixational pattern nye-ka-[xyz]-ani procedural meaning: read this as an agentive noun number: singular, gender: masculine procedural meaning: conceptual meaning: take this as the main indicator of doer, someone who does “xyz” the intended conceptual content male person combined meaning of root plus affixation (the new concept): one who turns [the wheel in a car] = DRIVER narrowing (depending on context): “one who turns s.th. in a car” => driver “one who turns s.th. in a ship” => captain “one who turns s.th. in a plane => pilot

There is already a transitive verb nyamojiri࡮ ‘turn’ in the language, which builds on the root *mojiriˎ. This root morpheme, wherever it occurs, points to meaning properties which all have to do with the basic idea ‘turn’ but differ somewhat, depending on context. At the same time the affixational pattern points simultaneously to a particular word class (for example a verb or a noun) and additionally to a particular inflected form. For example, in the verb form e-mojir-i the person agreement prefix e- and the aspect suffix -i signal not only ‘verb’ but also ‘third person singular in the present tense’. Root and affixes combine to encode ‘he turns’.10 Or, in the word (2c) nye-ka-mojir-ani࡮, the gender-number prefix nye-, ‘masculine/singular’, together with the derivational prefix ka- and the agentive suffix -ani࡮ as a whole pattern point the hearer to the like root morphemes and affixes in different contexts, as is naturally required in highly inflectional languages like Toposa. 10 Tense in Toposa is indicated by the tonal pattern over the entire word (tone has not been marked for the purposes of this investigation).

252

Chapter Twelve

understanding of ‘agentive noun’, and in this particular instance ‘agentive noun in masculine singular’.11 Thus, the affixational pattern provides a certain (but limited) amount of procedural meaning, for example ‘read this as an agentive noun in masculine singular’. It even contains a limited amount of conceptual meaning such as ‘doer of [xyz]’, where the content of ‘xyz’ is determined by the root, which always is the major contributor of conceptual content. So when the root morpheme mojiri࡮ which encodes the meaning in its broadest form is combined with this agentive affix pattern (which at the same time encodes masculine/singular), the resulting full lexical form encodes the concept SOMEONE [MALE] WHO TURNS (or TURNER) – which then still needs to be narrowed, depending on context. One could further say that even in the root there resides a certain procedural meaning, which we could loosely describe as an instruction to take this element as the main contributor of conceptual content in the overall word construction. As has become increasingly clear in the last few years (Wilson 2011, 12-14, 17-18), the border between conceptual and procedural is often not clear-cut. As I believe I have just demonstrated for Toposa, this is certainly true on the word level in highly inflectional languages, where I would like to suggest that roots generally encode conceptual content and affixes procedural content, but that even here the borders are fuzzy in that every root can be seen to contain a minimal procedural instruction, and every affix (or affix pattern, as the case may be) can be said to contain a limited amount of conceptual content.12 However it is only in combination that root and affixational pattern constitute a lexical form which encodes a fuller concept (that also includes inflectional information like gender and number), a pointer to meaning that goes way beyond what the root morpheme signals by itself. Nevertheless, even this full lexical form only points to the meaning the speaker intends to encode. In order for correct understanding to take place, the hearer needs to infer the intended meaning which in that context makes the best sense (i.e. appears optimally relevant), for example in the context of a car, the hypothesis that the man behind the steering wheel is the one who turns something appears optimally relevant in that instance, so the hearer may assume that DRIVER is indeed the intended meaning and stops processing. Note that this understanding was the first new idea that nyekamojirani࡮ 11

The corresponding feminine singular form is nya- (as in examples (1b) and (2b)). For a full description of Toposa nominal prefixes see section 2.1.2.2. 12 In order for this content to be accessed correctly, enrichment also may need to take place, but this area requires further research.

The Creation of Neologisms in Toposa

253

referred to, which was later extended to also refer to those who steer ships and aircraft. Once these options had become available to a larger number of speakers, of course, the processing of the neologism nyekamojirani࡮ requires narrowing in the same way as other concepts, i.e. via an ad hoc concept SOMEONE WHO STEERS* in order to reach the understanding that is correct and intended in a particular context, i.e. DRIVER in the context of a car, CAPTAIN in the context of a ship, or PILOT in the context of an aircraft. There are other aspects to the process of decoding and inferring in order to understand a neologism, such as what happens in the minds of a hearer when his hypothesizing does not immediately reach satisfying conclusions, because the intended new concept is more difficult to infer, which causes the hearer to set up an ad hoc concept of a type which at first has only a very limited amount of conceptual content under its encyclopaedic entry some of which may be stored with a tag that says “awaiting verification”, a type of concept that I propose to call ‘tentative concept’. 2.1.2 Change of grammatical category Toposa speakers also create new lexical forms by changing the grammatical category of a word. This corresponds to what Wilson refers to as “the transfer of a word from one syntactic category to another”, although in her English examples the change is always from noun to verb (2006/2007, 7.9),13 whereas in Toposa changes of grammatical category occur in two forms: as change of gender within the same noun class, and as change of noun class. 2.1.2.1 Change of gender Traditional Toposa has some noun pairs that occur with different gender prefixes14 and have related meanings, for example: (3a) (3b)

13

nye-kitoe M/SG-tree/wood nya-kitoe F/SG-tree/wood

tree wood

Wilson’s examples 13a-c. Toposa marks three genders in the type of noun we are discussing in this section: masculine, feminine, and diminutive. For a summary of all gender prefixes in all classes of nouns see table 1 below. 14

Chapter Twelve

254

(4a) (4b)

nye-moru M/SG-rock nya-moru F/SG-rock

mountain stone, rock

The semantic relationship expressed in such lexical pairs is not one of natural gender oppositions, like masculine-feminine,15 but one of partwhole (or possibly of dead-alive). A similar derivational process can be put to work in the creation of neologisms: one changes the gender of a noun from masculine to feminine (and probably also the other way around)16 as in the following example: (5a) (5b)

nye-putiri֔ M/SG-warthog nya-putiri֔ F/SG-warthog

warthog armored vehicle (tank)

What happens in such creations from a lexical pragmatics perspective? Similar to the first type of derivation we discussed, this type also recombines a known root morpheme and existing nominal affixes in a new way, except that this type of derivation employs a systematic change in gender that is utilized elsewhere to signal a semantic relationship between the two words involved. However, the semantic mapping between the old and the new use of the root meaning does not follow the same part-whole pattern as in traditional word pairs – such deviation may simply be the prerogative of those who craft neologisms, but it could also be a variation of another lexical process that is very common and shall be discussed as meaning extension below in section 2.2 As we shall see, in meaning extension a subset of encyclopaedic assumptions is activated which focuses on the common meaning properties normally activated by that root. For example, 15 Toposa marks natural gender fairly consistently: nye-kile ‘man’, nya-beru ‘woman’. It also has a third “diminutive” gender, which is often used to mark small items like nyi-koku ‘baby’, nyi-kale ‘kid goat’. However, as in other languages that mark masculine and feminine, all those Toposa words that cannot be associated with a natural gender carry grammatical gender. Finally, even in rare cases natural gender is not always applied consistently, for example nyi-tooni࡮ ‘person’, which is diminutive but is usually applied to men (c.f. also German ‘das Mädchen’). 16 All my examples are from masculine to feminine, but it seems too early to formulate a rule.

The Creation of Neologisms in Toposa

255

from a Toposa perspective, tanks are thought to share certain meaning properties with warthogs, such as their ability to run someone over, to move through mud, and their darkish brown color. Thus, the meaning of the new concept ARMORED VEHICLE is not immediately obvious, the new lexical entry provides merely a starting point for the hearer from where he needs to build the new concept. 2.1.2.2 Change of noun class Changing the noun class to produce neologisms is closely related to the most productive process in the language for creating names and terms for animals and plants based on known morphemes, or to create nicknames for people. Consider the following examples: (6a) (6b) (7a) (7b) (7c)

nya-upwale֔ F/SG:C1-shield lo-upwale֔ M/SG:C2-shield nya-kopile֔ F/SG:C1-umbilicus lo-kopile֔ M/SG:C2-umbilicus na-kopile֔ F/SG:C2-umbilicus

shield cobra (lit.: “one with shield”) umbilicus boy with umbilical hernia (nickname) girl with umbilical hernia (nickname)

Before we describe these examples, it may be helpful to briefly outline the three Toposa noun classes which can be charted as follows:

masc dim fem

class 1 Sg pl nyeƾinyinyaƾa-

class 2 sg loØ na-

Pl ta-loØ ta-na-

class 3 sg Ø

pl ta-

Table 1. Toposa noun classes Toposa nouns belonging to class 1 distinguish between three genders in the singular and two in the plural, whereas class 2 nouns only distinguish between masculine and feminine and usually employ the same prefix ta- to form plurals. There is a third class without any class prefixes which comprises only a few kinship terms and some other rare nouns.

Chapter Twelve

256

As the data (6b) and (7b-c) demonstrate, class 2 nouns are used in naming plants or animals, or to refer to people (where the noun is more of an ad hoc creation and is then understood to be a nickname). Many of these class 2 nouns referring to plants and animals exist even where there is no corresponding class 1 noun, in those cases the root may be known from another word class (most frequently adjectives). It is not surprising then, that such a productive naming process is also employed in the creation of neologisms, which of course is also about ‘naming’ new ideas. Consider the following example: (8a) (8b)

nya-laar F/SG:C1-fence na-laar F/SG:C2-fence

fence angle iron (lit.: “fence material’)

We found even one example where the process has gone the other way around from class 2 to class 1: (9a) (9b)

na-giligili֔ gecko F/SG:C2-ideo nya-giligili֔ F/SG:C1-ideo

teaspoon

The word for gecko, which belongs to noun class 2, existed long before teaspoons were introduced, which employ the same root, but combined it with a class 1 prefix. The root in each case describes the metallic noise caused by geckos and teaspoons.17 2.1.3 Compounding The naming process using the prefixes lo- and na- that we just described can also co-occur with compounding, i.e. joining two root elements together, for example: (10)

lo-bityanga֔ -kurwoni֔ M/SG-crazy-black

drongo (lit.: crazy black [one])18

17 Note that example (9b) is ideophonic, a special case that will be dealt with under descriptives in 2.3.2. 18 This songbird (dicrurus adsimilis) has glossy black plumage and characteristically jumps around in erratic movements, so the Toposa compound is a very fitting designation.

The Creation of Neologisms in Toposa

257

In the same way, Toposa speakers can craft new words, compounding two known roots to create a new word. Consider the following examples: na-yoko-tyaƾa֔ game warden F/SG-look-animals (lit.: [one who] looks [after] wild animals) (12) na-dapa-kiti֔ cell phone, mobile phone (lit.: hold.to:ear [thing]) F/SG-hold-ear

(11)

This type of construction is very similar to, but slightly different from the process “change of noun class” because usually there is no corresponding compound noun with a prefix of noun class 1. What is the same is that the prefixes lo- and na- indicate that some naming process occurs here, and the two joined root morphemes provide clues in which direction the combined meaning is to be sought: in example (11), out of all those who ‘look after wild animals’ the encoded concept refers to the type employed by the government to do so, i.e. a game warden, and in example (12), out of all the things one could ‘hold to [one’s] ear’, reference is intended to that device we hold to our ears for mobile communication, i.e. a mobile phone.

2.2 Extending the meaning of a known term Another widespread and productive approach to lexicalizing a new idea in Toposa is to take an existing word in the language and to extend its meaning. For example: (13a) nyamoru (13b) nyatarukoto֔ (13c) nyepoolo֔ (13d) nyakime֔ (13e) nyakoƾu֔

19

originally: stone now also: battery originally: vulture now also: airplane originally: colon now also: inner tube (bicycle/car) originally: fire now also: electricity19 originally: eye now also: headlight20

Note that this competes with the loan nyekaraba. Note that headlight can also be nyakinokete֔ /nyakunokete֔ (see example (1b) above).

20

258

(13f) ƾakonyen (13g) nyemali֔ (13h) ƾikwoi (13i) nyecokiti֔

Chapter Twelve

originally: eyes now also: glasses21 originally: arrow now also: bullet originally: quills (porcupine) now also: bullet originally: seed (inside fruit) now also: bullet

This type of innovation is not restricted to nouns, it occurs with verbs as well: (14a) kituduƾe

(14b) todetak

(14c) nya-kicuny-akin (14d) nya-ki-pen-ar (14e) nya-nyara-re (14f) nya-mojiri֔

originally: put out, extinguish (fire) now also: turn off light now also: turn off engine originally: flick (away) now also: turn on light and: pull trigger originally: light fire now also: weld originally: sharpen now also: grinding with disc originally: call now also: call on the phone originally: turn now also: drive (car etc.) and: steer (ship)

Let us take a moment here and ask what happens in this type of process from a lexical pragmatics perspective. When the meaning of a known term is extended to comprise a new concept, i.e. to add a sense as it were, there are at least some properties of meaning that are carried over, as for example in the case of (13b) above. A fixed-winged aircraft is not a living thing, but its flight pattern is gliding like that of vultures, and whoever first extended the meaning of vulture to airplane intended this analogy to be inferred. Similarly with (13c), an inner tube is not part of the digestive tract of mammals, but it looks like an intestine in terms of size and shape. 21 There is another word for glasses which is a loan from Arabic: ƾamadarae (from Sud. Arabic: nadaara).

The Creation of Neologisms in Toposa

259

Considering that the Toposa are pastoralists who frequently slaughter goats, sheep and cattle, this analogy transfer is an obvious choice. Thus when a speaker extends the meaning of a known term to a new idea in this way for the first time (usually when a member of this category is also visible), he expects his hearer to single out and activate, among all possible encyclopaedic entries associated with the known lexical entry, particularly those properties of meaning which resemble the known item in some respect. This approach to encoding new concepts relies on the capability of hearers to draw these parallels that seem most relevant in that particular instance from a Toposa cultural perspective and to construct the intended meaning based on the intended context. The human mind is very well equipped to do so because these are the same kinds of pragmatic processes we find in the creation and understanding of metaphor (Vega Moreno 2007, 87-120). Finally, compare examples (13g-i) which show that one unknown idea like BULLET can give rise to several lexical entries that may then compete with one another, an observation that we shall pick up in more detail in section 3.3 below.22

2.3 Descriptives 2.3.1 Descriptive phrases Another way to introduce a new idea is via using a descriptive phrase, usually in the form of a noun phrase, where an additional word is used to make a certain aspect of the concept explicit. This descriptive element can either refer to the outer appearance (such as color or size), to the function of the new concept, or to how it is used: (15)

(16a)

nya-putiri֔ ka ƾi-rikwo F/SG-warthog with M/PL-chains nya-giligili֔

tank (lit.: armored vehicle with chains)

teaspoon

F/SG-spoon (16b) (17a)

nya-giligili֔ na-a-polo-ni֔ F/SG-spoon F/SG-DER-big-ADJ nye-midani֔ ka nya-teere

tablespoon airport (from Arab. midan teyara)

M/SG-field of F/SG-plane 22

For other competing forms see the footnotes on examples (12e) and (12f) above.

260

Chapter Twelve

(17b)

nye-midani֔ ka nya-teere lo-dooci M/SG-field of F/SG-plane M/SG-small

(17c)

nye-midani֔ ka nye-kura M/SG-field of M/SG-football tire, wheel (of car) nya-keju ka nya-gariri֔ 23 F/SG-foot of F/SG-car (lit.: foot of car) nya-ki-myete֔ ka nya-tukutuku֔ 2-stroke oil F/SG-DER-oil of F/SG-motorcycle (lit: oil of motorcycle) nye-makana lo-dony-ete֔ sewing machine M/SG-machine M/SG:ADJ-sewing-INS nya-buku֔ naka nya-ceke֔ cheque book F/SG-book that:of F/SG-cheque

(18) (19) (20) (21)

airstrip (lit.: small airport/runway) football field

These data show that descriptive constructions in Toposa can take either the shape of a possessive phrase (as in 17a-b, 19 and 21), of a phrase consisting of a noun and an adjective (as in 16b and 20), or of a coordinative noun phrase (as in 15). Within such descriptive constructions, one element may in itself be a neologism, either in the form of a meaning extension – as in nyakimyet and nyakeju in (18) and (19) – or in the form of a loan as in nyemakana and nyabuku࡮ in (20) and (21). What happens in such descriptive constructions at the lexical pragmatic level? When a speaker uses a descriptive phrase to lexicalize a novel idea, the descriptive element points more explicitly to a certain conceptual content than the non-descriptive element in the phrase. In combination, both elements provide a starting point for the hearer, similar to what we found under derivation and meaning extension, except that via the descriptive the speaker provides stronger guidance towards the intended meaning. For example, the “foot of a car” provides two concepts, that of foot and that of car, which, if taken literally by the hearer, will not provide satisfactory cognitive effects because cars don’t have feet. However, the foot in humans and animals is the part where the body rests on the ground or which is used for motion, and this is exactly the association that is intended to be made.24 Often the creation of a descriptive phrase is motivated by the need to differentiate a newly introduced idea further, as example (15) shows when 23 The word nyagariri֔ itself is a loan from Kiswahili gari, which in turn is a loan from English ‘car’. 24 Probably all the more in that the Toposa word for foot includes the whole leg. (The precise extent that is intended to be referred to needs to be narrowed down in each context.)

The Creation of Neologisms in Toposa

261

compared to (5a-b): at some stage of the civil war it became necessary to distinguish between armored vehicles and tanks, hence the added descriptive “with chain”, as that is the main feature that distinguishes tanks from armored vehicles which run on rubber tires. Thus, descriptive additions are particularly helpful in situations where the modified concept itself is a neologism or a loan. Finally, I would like to point out that creating descriptive phrases is similar to compounding, except that here the descriptive elements are at phrase level rather than at word level. 2.3.2 Ideophones A special case of descriptive phrase is where the description of the new idea is ideophonic, i.e. it mimics the sound, a common occurrence in African languages. Consider the following examples:

(22)

nya-coowo F/SG-sound nya-giligili֔

(23)

nya-tukutuku֔

(24)

nya-meele

(25)

na-peeowae

(26)

nya-waawa

(27)

aleekubar

(21)

water tap, shower (ideophone: sound of running water coowo) 1. teaspoon 2. helicopter (ideophone gili-gili࡮, ‘metallic sound’) motorcycle (ideophone tuku-tuku࡮ ‘sputtering’) bicycle (ideophone melele ‘sound of chain’) airplane, aircraft (ideophone peeoo ‘sound of steady movement’) plastic bag (ideophone wa-wa-wa ‘sound of rustling’) mosque (ideophone from Sud. Arab.: Allahu Akbar)

Note how all the examples except one follow standard noun patterning. (27) is different because the ideophonic sound itself is a loan from Arabic. Let us examine what happens in this type of word creation. For example, in (21) the concept WATER TAP is encoded in a new word that is marked as a noun and has as its root the ideophone coowo which in Toposa mimics the sound of pouring water. Thus again, the form of this lexical entry provides a starting point, while the ideophonic root encodes all fixtures in the kitchen and in the bathroom that provide water accompanied by this particular sound, either water taps or shower heads, as the intended context may be. In a similar, but slightly more complex

262

Chapter Twelve

process, the concepts TEASPOON and HELICOPTER in (22) are both encoded as a lexical entry that is marked as a noun belonging to noun class 1 and has as its root an ideophone that mimics the sound of gecko calls. Since these are perceived of as metallic, this ideophone is used to describe both teaspoons (when several of them are put together) and helicopters (i.e. the chopping noise of their rotor blades). Thus, for correct understanding to occur in a particular instance, narrowing has to take place even after the two new concepts have been learnt.

2.4 New expressions (collocations) As we have just seen with descriptive phrases, what constitutes a concept and has been lexicalized as just one word in one language may be expressed as a phrase in another. It should therefore come as no surprise that some new concepts in Toposa have become encoded in the form of new collocations,25 for example: taram nyekwaam [nika ati] hit wind to someone (29) kicwaar nyekwaam send wind (30) nyakigir ƾakiro nauresik write matters short (31) nyariori֔ nyakwaan stretch body (32a) kicak nyakimyet step oil (32b) kicak nyeporomila step brake (33a) kigiraki֔ nyaceke֔ write cheque (33b) tataca֔ ka nyaceke֔ pay with cheque (34a) tolemu֔ ƾaropiyae kalo benki take money from bank (34b) kuwaaki֔ ƾaropiyae lo benki put money in bank (28)

25

call [someone] with mobile phone (lit.: hit the wind) send SMS/text message (lit.: send with the wind) write a short message26 exercise (lit.: stretch the body) accelerate, step on it [lit.: step on the oil] brake, slam on the brake (lit.: step on the brake) write a check pay by check withdraw cash deposit cash

I have not yet determined whether some of these expressions should be classified as idioms, more data are required here. 26 The verb write, now well established, once started its own life as a neologism: its original primary meaning is ‘carve, mark’.

The Creation of Neologisms in Toposa

263

As these data show, new expressions do not operate at phrase level, but combine a verb and a noun (which itself may consist of a noun phrase, as (30) shows). Such expressions either consist of two well-known lexical entries, or the new expression itself can incorporate a neologism. This new element can be either a known lexical form that has taken on an extended meaning (nyakimyet, tolemuˎ, kuwaaki࡮), or a loan (nyeporomila, nyaceke࡮). The difference between new expressions and descriptives is that here we do not have a process where one concept singles out some properties of meaning of the other concept in the construction, but in new expressions both elements contribute equally and only in combination yield the intended meaning which often cannot be guessed very easily. Like all other neologisms, the combined elements used in the explicature do not contain the intended meaning fully, but only provide some evidence towards it, the construction of the full meaning requires inferential work and a learning process.

2.5 Blends Wilson lists the blending of words as a basic lexical process for forming new concepts (2006/2007: 1.7), i.e. two existing words are fused together to encode a new concept that merges properties of meaning of both original words. While this is a well-established pattern in English, we have not found one example of this type of process in Toposa. Apart from the well-known example SMOG, two neologisms that I came across on the Internet recently demonstrate that blending of concepts to achieve a conceptual hybrid is a common mental process in English: (35a) smishing (35b) bankster

phishing via SMS (text messaging on smartphones) gangster specialized in robbing banks

The fact that no such process can be documented for Toposa (so far), should lead us to ask why certain processes of forming neologisms may be less utilized than others, or not at all. I will raise this question – together with some others – in my discussion in section 3.4. In my analysis, blending is a process similar to compounding, except that blends involve fusion, i.e. one or both combining building blocks lose an element. I therefore suggest to categorize blends as a special case of compounding – which exists in Toposa and is probably a universal process.

Chapter Twelve

264

2.6 Loans Another highly productive means of introducing a new idea into a language is to transfer it by borrowing the lexical entry that encodes it from the source language/culture, i.e. via a loan. This is an option not listed in Wilson 2006/2007, but I would like to advocate that loans be added as a subcategory of neologisms in view of how frequently borrowing occurs and what an important role it plays in the encoding of new concepts.) 2.6.1 Nativization Depending on the differences between source language and receptor language, such lexical transfers may be either straightforward transliterations, or they may involve some adaptation at the phonological and or graphematic level, and even at the grammatical level. Consider the following examples: (36) (37) (38) (39) (40) (41) (42) (43) (44)

nye-batariya nye-karaba nya-kompyuuta nya-caja nya-satipikete֔ nya-pikca nya-likaptar nye-poromila nye-mijiƾi֔

(45) (46) (47)

nya-mece֔ lo-plastik lo-ka-ƾapepera

27

flashlight [from Sud. Arab. batariya] electricity [from Sud. Arab. karaba]27 computer, laptop charger certificate picture helicopter brake (in car) [from Arab. faramil] fuel (petrol, diesel, kerosene) [from Arab. benzin] fighter (plane) [from Engl. MiG]28 made of plastic29 made of paper

In northern Sudanese Arabic this is phonetically [kax ҕraba], but in SouthSudanese Arabic from which it is borrowed, the fricative is dropped. Usually the pronunciation of loans in Eastern Equatoria is closer to Juba Arabic than to the Sudanese Colloquial Arabic spoken in the North. 28 MiG in turn stems from the Russian Mikoyan-i-Gurevich Design Bureau for fighter aircraft. 29 Adjectives in Toposa are prefixed by lo-/lu- for masculine singular/plural and na-/na- for feminine singular/plural. Additional prefixes and suffixes may occur.

The Creation of Neologisms in Toposa

265

Note how in Toposa the vast majority of all borrowed concepts are nativized at the grammatical level, i.e. nouns receive a noun-gender prefix belonging to noun class 1, as in (36)-(45), and adjectives take adjectival affixation, as in (46) and (47). Additionally, when the sounds are foreign, the pronunciation and the spelling of the loan word may be adapted to the Toposa phoneme system as deemed necessary. Where sounds of the source language do not exist, Toposa replaces them with the phonetically closest option, like the in certificate becomes

.30 Where syllable structure is different, Toposa is more accommodating. For example, the consonant clusters m-py31 in (38), k-c in (41), and p-t in (48) violate the strict CV syllable structure in Toposa (even with the syllable boundary between the two consonants). On the other hand, in some words speakers prefer to drop a sound in order to conform to syllable structure, as in (40) where the r is dropped to avoid the sequence r-t, or to add a word-final vowel, as in (43).32 Only very few loans in Toposa are borrowed without grammatical nativization. The only ones found so far are:33 (48) (49) (50)

bapra ~ bafra antinopo࡮ senke [seƾke]

cassava (from Sud. Arab.: bafra) bomber, cargo plane (from Russian: Antonov) motorcycle (from Chinese brand name Senke)

As with other class 3 nouns, the plural for these forms is formed by prefixing ta-. From a lexical-pragmatic perspective, loans presuppose at least partial knowledge of another culture and language and do not provide by themselves any structural or semantic clue as to how they are intended to be understood. Thus they are harder to interpret and need to be learned in pretty much the same way as completely new vocabulary in language acquisition, i.e. they either require harder inferential work, or some explanation that helps to achieve correct understanding.34 30

Toposa does not have the consonants [f, v, h, and ‫]ݕ‬. Toposa allows palatalization and labialization of all its consonants which are written as sequences of consonants, for example py or pw. 32 Here a has been added, although words ending in a voiceless vowel are quite common and are often used in cases where the loan ends in a closed syllable, as in (40). 33 There are some others, but these were introduced by Italian missionaries, not Toposa native speakers. Example (54a) below demonstrates this. 34 When it comes to introducing new ideas in translation, the use of loans puts monolingual individuals at a disadvantage for obvious reasons. In terms of 31

Chapter Twelve

266

2.6.2 Under- and overdifferentiation While most loans come to acquire the same common encyclopaedic information they have in the source language, some loans undergo semantic change, which may result in either underdifferentiation or overdifferentiation. The following examples show how not all distinctions that the source language makes are transferred to Toposa, resulting in loans that encode more than one concept in the source language: (50) (51) (52)

nyewotele֔ nyasidi nyapidiyo

1. hotel 2. restaurant 1. CD 2. DVD 1. video, movie 2. video player, VCR player 3. TV, television set

Consider example (52). The loan word nyapidiyo has experienced a meaning extension from video to TV, because when the first TV sets came to Southern Sudan, there were no TV channels and all movies came recorded on video tapes, thus a separate expression for TV was not relevant at the time. Now that reception via satellites has been established, the former loan is no longer precise enough, it may be only a matter of time until its meaning gets adjusted. For some speakers the concept TELEVISION has already been encoded differently: (53)

nyatelepisono࡮ [alt.: nyatelebisono࡮]

TV, television set

It is to be expected that the next concept to be encoded separately is likely to be the DVD PLAYER. While this type of underdifferentiation in Toposa loans is quite common, the opposite effect, overdifferentiation, is rare: (54a) (54b)

piino nyewayini֔

wine (from Ital.: vino) wine (from Engl.: wine)

While piino refers to wine as introduced decades ago by the Roman Catholic church for the celebration of mass, nyewayini࡮ is the type of wine that is now sold in shops, a fairly recent development, and currently limited to those Toposa speakers that have been exposed to modern common translation terminology, one could also say that the introduction of a loan is far more foreignizing than any of the other approaches we have looked at so far, all of which are more on the domesticating side of the scale.

The Creation of Neologisms in Toposa

267

consumer goods. Even though the same substance is referred to, the distinction between sacral and profane contexts have led to separate encodings. From a lexical-pragmatic perspective, it seems best to explain these competing terms in terms of greater accessibility and relevance in their respective contexts.

3. Further observations and analysis 3.1 Types of processes in Toposa To summarize, the processes in the creation of neologisms which we encountered in our Toposa data are as follows: x derivation o new combination of known roots and derivational affixes o change of grammatical category ƒ change of gender ƒ change of noun class o compounding x extending the meaning of a known term x use of descriptive words or phrases o describing appearance/function/use o describing sound via an ideophone x new collocations/expressions x borrowing/loans When we compare our data from Toposa with another language like English, it becomes apparent that not all languages employ exactly all the same methods of creating neologisms. For example, English does not employ change in gender or change in noun class, while Toposa does not know the blending of concepts.35

3.2 Explicature and inferential processing In order to understand these lexical processes better, it makes good sense to ask how they differ or overlap. In our analysis, let us begin with what seems common to the processes we have observed. Please refer to Table 2 for a summary of these processes. 35

As data from Yoruba show (Stark 2000), other languages again differ from both English and Toposa.

word class => noun => subtype root meaning (can be quite similar but also extended)

conceptual/procedural meaning encoded

new combination of known words (N+N or N+ADJ)

new combination of 2 known words at clause level (v+o)

unknown word(s)

Expressions

Loans

root morpheme

Table 2. Types of processes in the creation of Toposa neologisms other languages: no clues

Toposa: known affix pattern

syntetic pattern known meaning of elements

phrase pattern known meaning of elements (one explicitly states certain properties of meaning)

root meaning a mere starting point

2 known roots to form new stem known word pattern meaning properties belonging to 2 root morphemes word class known affix pattern

affix pattern (mostly procedural meaning) root morpheme(s) (mostly conceptual meaning)

linguistic elements used

Descriptives (phrases/ideophones)

Meaning extension

change of grammatical category (gender, noun class) compounding

Derivation new combination of known roots + derivational affixes

type of process

entirely new concept to be inferred and learned without any indication as to intended meaning whatsoever (except context ± explanations)

new concept to be inferred and learned

new concept to be inferred and learned

significant meaning extension to be inferred/learned

analogous properties of meaning to be identified/accessed

2 relevant sets of encyclopedic entries to be identified/accessed new resulting concept to be inferred and learned

analogous properties of meaning to be identified/accessed based on context new word meaning (concept) to be inferred

inferential processing required

268 Chapter Twelve

The Creation of Neologisms in Toposa

269

First of all, at the lexical level, there are pragmatic processes at work which are similar to those that occur at utterance level, in that all neologisms are encoded as linguistic forms which can be decoded… to a degree. The much larger amount of processing work lies in attempting to infer correctly what the encoded concept seems to represent until one has achieved a full conceptual content. (This again cross-links to the issue of how exactly new concepts are understood for the first time, how full conceptual content is learned over time and how understanding spreads from one individual to another.) One important observation I want to make here is that there is a qualitative difference between creative processes like derivation, meaning extension and forming descriptives on one hand and using loans on the other. Loans are different in that the whole concept is expected to be carried over, it needs to be grasped “in one chunk” because there are no known elements in the lexical entry (except the affixational patterns in those loans that are nativized a little) – whereas all other creative processes utilize some properties of meaning as a starting point for the hearer to access the intended new meaning, i.e. they lead along a gentler path from the known to the new.

3.3 Innovation as process What the data as presented in this study do not show is that neologisms do not simply begin to exist. By virtue of their nature, data are static, the reality behind them is much more fluid: there is a multitude of complex processes in operation from the first time someone coins a new word until that term has gained more widespread acceptance in a language community. This can happen either gradually or almost explosively, but there is always a process involved how new words are created and new ideas are transmitted by them.36 Something else that follows from this observation is that it depends on the nature of the new idea, and the context in which it is transmitted, how readily this new concept may be properly represented in the minds of a community of speakers and hearers. Understanding, which I suggest to define as mentally representing someone else’s ideas and intentions correctly, is a lot easier if a new lexical entry refers to something concrete, 36

Important in this context is Sperber’s extensive work in cognitive anthropology (1996 and a host of articles before and after this landmark book), attempting to explain culture in terms of an ‘epidemiology’ of mental representations and examining under which circumstances there occurs a relative stabilization of mental representations and productions in human populations.

Chapter Twelve

270

i.e. something that is accessible to the five senses during communication (vision, touch, sound, taste, smell). It also helps when the function or purpose of this new item is immediately obvious and does not need to be guessed. If, however, the new lexical entry refers to some abstract concept, and maybe one that is even quite foreign to the culture and which is furthermore introduced and used without any explanation, correct understanding involves a more complex process of learning, i.e. the hearer sets up a new conceptual address and begins with an incomplete set of encyclopaedic assumptions, sometimes mere guesses, while he tests various hypotheses as to what the intended conceptual content might be. The same happens in the minds of a whole community of speakers and hearers, until most of them have reached an adequate understanding of the new concept and the lexical entry has become part of the “public lexicon”.

3.4 Misconceptions A corollary of this learning process is that not all word creations are necessarily correctly understood right away, i.e. attempts to discover their intended meanings may be based on false assumptions and therefore yield the wrong cognitive effects, at least initially. For example, the Toposa have observed condensation trails of aircraft flying at high altitude for a long time, however it never occurred to them to relate this phenomenon to airplanes (which they know from the Kapoeta airfield and flying at low altitude without condensation trails), hence their word for what they observed high up in the sky was a tentative concept with the basic meaning “steadily moving object”. Only later, when educated Toposa confirmed that these clouds were caused by aircraft, the term for UNKNOWN OBJECT MOVING STEADILY ACROSS THE SKY also began to mean AIRCRAFT: (55)

napeeowae

1. unknown object moving steadily across the sky 2. airplane, aircraft

There are other concepts which at present still fall under the category of misconceptions that have not been correctly identified yet – even though this is probably only a matter of time:

The Creation of Neologisms in Toposa

(56)

nya-kanyer-iti֔ na elosi F/SG-star-SG which goes

271

“star that walks around” (in reality: satellite)37

Obviously, when it comes to learning new lexical entries via tentative concepts, we may expect to encounter similar phenomena as those we observe in language acquisition, both that of children and second language learning, be it at school or later in life.

3.5 Competing terms When new ideas migrate from one culture to another, one concept is not necessarily tied to one lexical entry only. In the process of naming these new ideas, different terms may arise and begin to compete, as the following data will show: (57a) (57b) (58a) (58b) (59a) (59b) (60a) (60b) (61a) (61b) (62a) (62b) (62c)

ƾa-madarae ƾa-konyen bafra lo-cine-bu nye-karaba nya-kime֔ nya-teere nye-tarukot nya-kitabe֔ nya-buki֔ nya-likaptar nya-kani֔ -dege na-giligili֔

(63a) (63b) (63c)

nya-kaloboci֔ nya-malaga na-giligili֔

glasses (from Arab.: nadaara) glasses (lit.: eyes) cassava (from Arab.: bafra) cassava (lit.: faeces of hyena)38 electricity (from Sud. Arab.: karaba] electricity airplane (from Sud. Arab.: tiyaara) airplane (originally ‘vulture’) book (from Arab.: kitaab ‘book’) book (from English: ‘book’) helicopter (transliterated from English) helicopter (lit.: hand-aircraft) helicopter (from ideophone gili-gili࡮ ‘metallic noise’) spoon spoon (from Arab.: malaga) spoon (from ideophone gili-gili࡮ ‘metallic noise’)

37 There are several satellites that move across the skies above Toposaland, including the brightly visible International Space Station, which is also still in the category of moving stars for the vast majority of speakers. 38 This is not necessarily a nickname. Due to the high calcium content of crushed bones, the faeces of hyenas are often whitish and look surprisingly like the piles of cassava sold in the market. (Many goods in the local market in Kapoeta are sold in piles rather than by weight.)

272

Chapter Twelve

(64a) nya-tukutuku֔ (64b) senke (65a) nyemali֔

motorcycle (idephone tuku-tuku࡮ ‘sputtering’ motorcycle (from Chinese brand name Senke) originally: arrow now also: bullet (65b) ƾikwoi (ƾukwoi) orginally: quills (porcupine) now also: bullet (65c) nyecokiti֔ orginally: seed (inside fruit) now also: bullet

The first four example sets (57)-(61) demonstrate synonymous pairs where one lexical entry was borrowed from Arabic, but a local word was coined as well. Since there is no written record of these developments, we can only make some educated guesses. What we know is that with the intensification of the civil war and now after the separation of the Sudan into two sovereign states, loans from Arabic have been perceived more and more as belonging to the language of the enemy. While this has not affected most older loans from Arabic, we observe today a strong shift away from Arabic towards crafting Toposa terms or borrowing from English (with the occasional ‘intrusion’ from Kiswahili). Nevertheless, as Toposa shows, competing entries can continue to co-exist for decades. Examples (61a-b) show that even two loans can compete. Typically, the competition is between loans first borrowed from Arabic and later reborrowed from English, but this isn’t necessarily always so: in the case of BOOK the concept might have been introduced when the British colonial administration allowed missionaries to establish an education system which was taken over by the Arabic-speaking Northern Sudanese government during the time of the first civil war (1955-1972). Examples (62a-c) show that Arabic had already fallen into disrepute when the war brought the first helicopters to Toposaland, and furthermore how more than one locally crafted term can fill a lexical gap. It remains to be seen which term will ultimately ‘win’, i.e. become the dominantly preferred lexical entry. Of the three words for bullet in (65a-c), the one derived from the word for arrow is the oldest. It already existed before the war, while the other two based on quill and seed are more recent additions. How can we best explain the creation of competing lexical entries? There may be several factors responsible. The first factor is clearly sociolinguistic and has to do with the personal willingness and tolerance for continuing to use a lexical entry borrowed from a language and culture that is seen not only as foreign but likely also as hostile, hence the shift in Toposa away from loans based on Arabic and the increased shift to loans from English und Kiswahili, or to crafting a more indigenous term.

The Creation of Neologisms in Toposa

273

Other factors involved lend themselves more readily to lexicalpragmatic analysis. One option is that speakers simply follow different paths of encoding a new idea without having contact with one another. Normally a speaker will try to craft a term that is easily accessible and relevant to him and his hearers, i.e. it provides sufficient cognitive benefits in a particular instance. In one case this can be the analogy of quills and bullets (possibly including the cartridge), on another occasion this is the similarity between the bullet (without the cartridge) and a seed. Even though this is hard to measure and prove, it seems reasonable to assume that each innovator assumed that the meaning of each of these forms was relatively easy to access, i.e. it offered more relevance in each situation to his hearers. In the long run it is to be expected that the competing form that ‘wins’ will be the one which offers the greatest overall relevance to the largest number of speakers in the community.

3.6 Why are some processes more frequent than others? We noted above that Toposa does not seem to be producing conceptual blends. This led us to ask why certain types of creative processes exist in one language, but not in another. One answer must be that some processes are only possible in a particular language due to its structural properties, like the change of noun class in Toposa – other languages do not have that grammatical option. Furthermore, it seems that language communities develop certain preferences. For example, while blending thrives in English, it is not very productive in Polish, and non-existent in Toposa. However, there seems to be no reason why, for example, Toposa speakers should not begin to blend morphemes together in the future. At this point it is a technique they do not have in their repertoire and they have not yet learned from other languages like English. Also, the introduction of blending is not an urgent matter for the Toposa, because it is nothing more than a special case of joining morphemes together, and compounding already exists in Toposa and is widely used.

4. An important practical application This whole field of how neologisms are created is not only a fascinating study in itself, it also has an important practical application, that is in the realm of translation. For years I have served as consultant for a Bible translation project where a team of Toposa translators have had to struggle with the challenge of transferring new ideas from ancient source texts in

274

Chapter Twelve

Hebrew and Greek, reflecting various source cultures, into the rapidly changing modern-day context of the Toposa language and its culture. Time and space do not allow me to present data from that Bible translation project and lexical innovations in the context of the churches operating in Toposaland. However, I would like to at least briefly point out that it is very instructive to take real-life data, like the examples I have presented in this paper, and to compare them with the approaches which mother-tongue translators employ in their translation. Such a comparative study might help greatly in determining which techniques of bridging lexical gaps are available to the translators, which approaches work well (and why), and which are harder to process and are less likely to succeed. In other words: examining natural language first clearly provides valuable insight into how translators can most profitably craft new terms in situations where there are lexical and even conceptual gaps, i.e. employ approaches that are more likely to communicate successfully.39

5. Conclusion The examination of our data has shown that when native speakers form neologisms in Toposa, they normally choose one of the strategies listed in detail in section 3.1. That taxonomy also shows us that there are more strategies of how to encode and introduce new concepts than listed in Wilson 2006/2007 and Schröder 2009.40 In terms of broader categories, we found that Toposa speakers employ derivational processes, meaning extension of existing lexical inventory, formation of new phrases, either using a descriptive element or in the form of a new collocation, as well as loans as their preferred approaches to encoding new ideas. In our lexicalpragmatic analysis we saw that with the exception of loans, all the formation processes in some way point to the intended meaning via their affixational patterning and some meaning properties of their roots – usually an analogy that needs to be worked out by the hearer. In this regard neologisms are similar to utterances (most of which are also unique and new expressions of some speaker intent): All neologisms have a linguistic form (the lexical entry) which in some way encodes the 39 Due to what we have learnt the hard way I would certainly recommend such a comparative approach to all translators who need to communicate new ideas between different cultures. 40 These are the only works within the RT framework that I know which list different types of neologisms. The only other article that mentions the creation of neologisms and word coinages as lexical process is Wilson 2004, 346 (and its 2003 sibling).

The Creation of Neologisms in Toposa

275

intended meaning, providing clues to conceptual content, mainly through their root, and often procedural information through their affixational morphology, but a large part of the intended meaning usually needs to be inferentially worked out by the hearer. We further looked at how lexical innovation is always a dynamic process that requires learning over time, i.e. the gradual forming of ideas from tentative concepts to adequate conceptual content. Sometimes this journey leads via the detour of misconceptions, sometimes there may be parallel paths of competing terms over longer periods of time. The data from Toposa we considered further demonstrated that not all languages employ exactly all the same methods in crafting neologisms. While this is partly due to structural constraints, languages also develop certain preferences which approaches speakers opt to choose. Furthermore, when we bring the insights of Relevance Theory to bear on this question, the most likely explanation seems to be that what is most important in the choice of approach is the perception of the communicator as to which form of encoding may be more easily accessible and relevant to his particular audience in a particular communication situation. Finally, I have pointed out that a clearer understanding of how neologisms are formed in a language can help translators to improve naturalness and comprehension of their translations when it comes to transferring unknown ideas from one language to another.

References Carston, Robyn. 1997. Enrichment and loosening: complementary processes in deriving the proposition expressed? Linguistische Berichte 8 (Special Issue on Pragmatics): 103-27. Chun, Chong-Hoon. 2007. A study of Korean conjunctive verbal suffixes: Towards a theory of morphopragmatics. PhD Thesis. University of New South Wales. Schröder, Martin C. 2009. The transfer of unknown ideas as neologisms: A fresh look at an old translation problem from the vantage point of lexical pragmatics. Paper presented at the 2009 Conference on Bible Translation, Oct 16-20, in Dallas, TX. SIL-International, [CD-ROM]. Sperber, Dan. 1996. Explaining culture: A naturalistic approach. Oxford: Blackwell. Sperber, Dan and Deirdre Wilson. 1998. The mapping between the mental and the public lexicon. In Language and thought, edited by P. Carruthers and J. Boucher. Cambridge: Cambridge University Press, 184-200.

276

Chapter Twelve

—. 2002. Truthfulness and relevance. Mind 111: 583-632. Stark, John E. 2000. The role of oral literature in the dissemination of neologisms. PhD thesis, University of Ilorin, Nigeria. Wilson, Deirdre. 2003. Relevance and lexical pragmatics. Italian Journal of Linguistics/Rivista di Linguistica 15, no. 2: 273-291. —. 2004a. Relevance and lexical pragmatics. UCL Working Papers in Linguistics 16: 343-360. —. 2004b. Relevance and word meaning: The past, present and future of lexical pragmatics. Modern Foreign Languages 27: 1-13. —. 2006/2007. Issues in pragmatics: Lexical pragmatics. [Course notes (PLIN 3001), 2006-07]. —. 2011. The conceptual-procedural distinction: Past, present and future. In Procedural meaning: Problems and perspectives, edited by V. Escandell-Vidal, M. Leonetti and A. Ahern. Bingley, UK: Emerald, 331. Wilson, Deirdre and Robyn Carston. 2007. A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In Pragmatics, edited by N. Burton-Roberts. Houndmills, Basingstoke: Palgrave Macmillan, 230-259.

CHAPTER THIRTEEN CONTRASTIVE REDUPLICATION AND RELEVANCE EWA WAàASZEWSKA, UNIVERSITY OF WARSAW, POLAND

1. Introduction The term ‘contrastive reduplication’, or ‘lexical cloning’ has only recently gained currency in linguistics, mainly due to the growing popularity of the phenomenon it refers to. Contrastive reduplication is a type of reduplication involving a repeated occurrence of the same word with contrastive stress on the first item of the doublet. It is quite common in English (at least in colloquial American English), but can be encountered in other languages as well. The paper attempts to describe this phenomenon, showing how it differs from other types of reduplication and other related phenomena such as compounding, contrastive stress and the rhetorical effect of repetition (epizeuxis). To do this, reduplication is first reviewed, with the focus on various types and their classification. Next, contrastive reduplication is discussed in view of previous research and attested examples. The discussion leads to the conclusion about the pragmatic nature of the phenomenon and about inadequacy of previous analyses based on prototypes; examples show that interpretations of utterances including cases of contrastive reduplication are not restricted at all to prototypical interpretations. It turns out that contrastive reduplication works in a way similar to hedges as another linguistic device whose function is to indicate the need for concept adjustment, and specifically, for the narrowing of a lexicalized concept. Additionally, the phenomenon is redescribed within Relevance Theory as involving procedural meaning.

278

Chapter Thirteen

2. An overview of reduplication According to Sapir (1921, 76), “[n]othing is more natural than the prevalence of reduplication,” a morphological process “in which part or all of a word is repeated” (Stockwell and Minkova 2001, 18). It is hard not to agree with this observation since reduplication is a common process in both inflection and derivation though, undoubtedly, it is more frequent in the area of word-formation (Bybee 1985, 97). Reduplication is regarded as even a more common type of word-formation among the languages of the world than different forms of affixation (Braun and Plag 2003, 91). On the other hand, the popularity of reduplication depends on the language family; as observed by Booij (2005, 43), “languages of the Austronesian family make wide use of reduplication patterns, whereas this does not apply to most Indo-European languages of Europe.” In Germanic languages, reduplication is considered to be an unusual word-formation process (Booij 2005, 59); unsurprisingly then, in English, it is treated as rather marginal, though often amusing, way of expanding vocabulary (Stockwell and Minkova 2001, 18). Cross-linguistically, productive types of reduplication include, among others (a) reduplication of verbal bases with nominal effect (e.g. ‘instrument for Ving’, ‘act of Ving’, ‘someone who Vs/something that Vs’), (b) intensifying reduplication such as reduplication of adjectives with the resulting meaning ‘very A’ and reduplication of members of other word-classes, such as adverbs, nouns or determiners, also with intensifying effect, (c) resultative reduplication where verbal bases are reduplicated with the purpose of creating adjectives with the meaning ‘result of Ving’ (Braun and Plag 2003, 91-92). Reduplication, especially full reduplication (where the doubled part, ‘reduplicant’, to use standard terminology, fully reproduces the ‘base’) is often associated with iconicity in language: the doubled form is, in a sense, indicative of doubled or intensified meaning (Booij 2005, 59). In Sapir’s words: “The process is generally employed, with self-evident symbolism, to indicate such concepts as distribution, plurality, repetition, customary activity, increase of size, added intensity, continuance” (1921, 76, italics mine). Partial reduplication is typically employed to create both inflectional and derivational meanings which involve little iconicity, for example marking third person singular possessive in Tarok (Benue-Congo) (Inkelas and Zoll 2005, 14). Although reduplication is not regarded as one of the typical wordformation processes in English, there are copious examples of reduplicatives (words or pairs of words formed by means of this process) in the language. Moreover, people are not ignorant of their abundant

Contrastive Reduplication and Relevance Theory

279

presence, which can be seen in the vast number of names that are used to refer to reduplicatives: ‘echo words’, ‘sing-song words’, ‘wishy-washy words’ or ‘rhyming words’, to mention but a few (Mattiello 2013, 141). This high, though underestimated, occurrence of reduplicatives in English was already commented on, at the beginning of the 20th century, by Sapir (1921, 76), who aptly pointed out that “[s]uch locutions ... are far more common, especially in the speech of women and children, than our linguistic textbooks would lead one to suppose.” It is worth noting that it is the sixteenth century that marks the beginning of the growing popularity of reduplicatives. Currently, reduplication is usually employed to produce original meanings, many of which are just nonce formations. On the basis of some recent English examples, Mattiello (2013, 143) shows that reduplication is typical of song lyrics, political and advertising slogans, newspaper headlines, brand names, slang terms and character names in children’s cartoons or television series (e.g. Tinky-Winky in Teletubbies). In English, reduplication is associated with informal varieties of language; it is mostly used in disparaging contexts (e.g. art-shmart), humorous or playful constructions (e.g. Humpty Dumpty) and baby talk (e.g. wee-wee). The use of reduplicatives by adults, especially of those expressions that are related to nursery language, may signal emotional closeness and reduction of social distance. Mattiello (2013, 142) observes that what might be expressed by the use of reduplicatives is “a pragmatic meaning of non-seriousness.” Apart from the user’s ‘non-serious attitude’, reduplicatives can be characterized by a certain amount of semantic indeterminacy and vagueness since they often encode “vague concepts such as indecision, confusion, carelessness, disorder, foolishness” (Merlini Barbaresi 2008, 235), contempt (Sapir 1921, 76) and smallness (e.g. itsybitsy, bitsy-witsy, itty-bitty, teeny-weeny) (Mattiello 2013, 164). Another property that is standardly ascribed to reduplicatives is iconicity. In examples such as flip-flop and zig-zag, the vowel alternation can be viewed as representing movement in opposite directions. Pairs of rhyming words such as willy-nilly and shilly-shally will be interpreted as conveying indecision. Reduplicatives based on the exact repetition of the base such as bye-bye seem to be indicative of reiterated action (repeated waves of the hand) (Mattiello 2013, 164-165). At this point, it is worth noting that reduplication can be approached from two perspectives: according to the broad view, reduplication comprises both syntactic repetitions such as long long (‘very long’) and syllabic repetitions e.g. kiwi; according to the narrow view, reduplication is limited to repetition with nonsense bases, e.g. hubba bubba (Mattiello 2013, 144). This shows that the phenomenon of reduplication in English is

280

Chapter Thirteen

by no means homogeneous and that a classification of reduplicatives is a challenging task. An interesting recent classification of reduplicatives has been offered by Mattiello (2013), who suggests two general macro-classes based on the traditional division into full (total) and partial reduplication, supplied with a number of sub-classes. The first macro-class contains socalled ‘copy’ or ‘exact’ reduplicatives, in which the reduplicant is the exact copy of the base, e.g. ‘baby-talk’ reduplications such as choo-choo, wee-wee. In the second macro-class, there are reduplicatives involving repetition of only a part of the base. This macro-class can be further divided into (1) ‘ablaut’ reduplicatives or combinations (e.g. flim-flam, zigzag, sing-song, pitterpatter, riff-raff, mish-mash), which involve repetition with an alternation of the internal vowel; (2) ‘rhyming reduplicatives’ or ‘rhyme combinations’ (e.g. super-duper, willy-nilly, pall-mall, okey-dokey, hanky-panky), in which the reduplicant is a meaningless element linked to the base by rhyme; (3) ‘rhyming compounds’ (e.g. chug-drug ‘the sound of an explosion’, funny bunny ‘peculiar’), which are characterized by rhyme but, unlike rhyming reduplicatives, which are based on two meaningful elements; (4) ‘onomatopoeic reduplicatives’ or ‘ideophones’1 (e.g. bow-wow, trit-trot, tick-tack), which involve repeated monosyllables imitating natural and other sounds. Three of the classes listed by Mattiello clearly overlap with those suggested by Ghomeshi et al. (2004, 309) – baby-talk reduplications, rhyme combinations and ablaut combinations. Apart from these, Ghomeshi et al. identify the following types of reduplication: (1) ‘multiple partial reduplications’ (e.g. hap-hap-happy), typically found in song lyrics; (2) ‘deprecative reduplication’ or the Yiddish-inspired ‘shm-reduplication’ (e.g. table-shmable, baby-schmaby, linguist-shminguist), in which the shm-copy has deprecative or belittling effect; (3) ‘intensive reduplication’ (e.g. You are sick sick sick!, Let’s get out there and win win win!, Prices just keep going up up up), which requires a sequence of at least three items; (4) ‘contrastive reduplication’ (e.g. I’ll make the tuna salad, and you make the SALAD-salad.), in which the reduplicant bears contrastive stress.2

1

The existence of this class is questionable since most of its members fall into other classes. 2 In my analysis, I will follow Ghomeshi et al.’s convention and use capital letters to mark the contrastively stressed element.

Contrastive Reduplication and Relevance Theory

281

3. Contrastive reduplication The phenomenon of contrastive reduplication has also been referred to as ‘lexical cloning’ (Horn 2006) or ‘identical constituent compounding’ (Hohenhaus 2004, 297). These terms show that cases of contrastive reduplication can be treated as compounds in the case of which both elements are the same, or to put it differently, one element is a clone of the other. Such a treatment, however, seems to go against the common understanding of compounds according to which they are items combining dissimilar constituents (Hohenhaus 2004, 299). Contrastive reduplication is a fairly recent phenomenon typical of colloquial English, considered today to be quite frequent especially in American English.3 It may target different grammatical categories (nouns, verbs, adjectives, proper names, pronouns) as well as lexicalised expressions. The effect contrastive reduplication has on the target element is “to focus [its] denotation ... on a more sharply delimited, more specialized, range” (Ghomeshi et al. 2004, 308); specifically, as Ghomeshi et al. initially observe, to focus its denotation on the prototypical instance of the reduplicated lexical expression. Let me illustrate the phenomenon with the examples in (1): (1)

3

a. I’ll make the tuna salad, and you make the SALAD-salad. b. That’s not AUCKLAND-Auckland, is it? c. Are you LEAVING-leaving? (Ghomeshi et al. 2004, 308, 312) d. Somewhere in the dark park, laced by low fog. Grant, Timmy and Lex walk. Lex: I’m tired. Timmy: You’re too big to be carried. Lex: But I’m TIRED-tired. [...] Grant picks her up. Grant: Oof! You’re heavy. (Hohenhaus 2004, 330, from Jurassic Park, first draft by Michael Crichton, re-write by Maria Scotch Marmo 3/14/92, at: http://blake.prohosting.com/horrorsu/scripts/JurassicPark_lstDr afl.txt) e. Geoff: So you didn’t bring any food?!? Mike: Not FOOD-food ... Geoff: What’s ‘not FOOD-food’, then? Mike: Got bubble gum ...

There is a collection of over 200 examples in the online Corpus of English contrastive focus reduplications available from http://home.cc.umanitoba.ca/~krussll/redup-corpus.html.

282

Chapter Thirteen

Geoff: How you expect to survive on gum for three days, you muppet?!? (example from Hohenhaus 2004, 301, from The Hole, directed by Nick Hamm, Universal Pictures, 2001) In these examples, contrastive reduplication “signals that one meaning of the word is being contrasted with other possible meanings” (Ghomeshi et al. 2004, 317). In example (1a), the expression SALAD-salad is used to denote a specific type of salad (prototypical plain green salad) as opposed to other salads (e.g. chicken salad or shrimp salad), whereas in example (1b), the sequence AUCKLAND-Auckland is used to denote the city in New Zealand, not just any city with this name. In (1c), the reduplicated verb is taken to mean ‘really leaving’ or ‘leaving for good’ (as opposed to ‘stepping out for a moment’), which seems to be the default, prototypical interpretation of leave. In (1d), the reduplicated form is interpreted as ‘really tired’ or ‘tired to the point of exhaustion’. Finally, in (1e), the expression FOOD-food is used to denote prototypical food (something that contains essential body nutrients) as opposed to chewing gum, which, though not at all nutritive, is also classified as food. It is worth noting that functional words such as auxiliaries and determiners do not undergo the process of contrastive reduplication, which is illustrated by the ungrammaticality of examples (2a) and (2b): (2)

a. *Are you sick, or ARE-are you sick? b. *I didn’t just read the book, I read THE-the book! (Ghomeshi et al. 2004, 313)

Contrastive reduplication can also be found in a number of other languages, for example Russian, Spanish, French, Italian, to mention just a few. Unfortunately, there is a lot of terminological confusion around this phenomenon. Most of the properties that Ghomeshi et al. (2004) attribute to contrastive reduplication in English have already been mentioned by Wierzbicka (2003, 255) in her discussion of what she calls ‘syntactic reduplication’, commonly used for expressive purposes in Italian: both processes operate on words rather than units smaller than words and both of them can affect different lexical categories (e.g. nouns, adjectives, etc). Wierzbicka insightfully insists on distinguishing cases of syntactic reduplication such as adagio adagio (‘slowly slowly’), in which, formally, there is no comma between the elements, from cases of repetition such as adagio, adagio (‘slowly, slowly’), where a comma signals the presence of a pause. On the other hand, Levinson (2000, 150) in his account of English

Contrastive Reduplication and Relevance Theory

283

examples of (contrastive) reduplication and repetition, apparently does not make any distinction between the two.4 It might be possible to distinguish cases of reduplication such as prettypretty from those involving repetition such as pretty, pretty by means of criteria originally suggested by Thun (1963), based on phonetic/prosodic, morphological and semantic information. Consequently, there is a prosodic difference between prétty-pretty (reduplication) and prétty-prétty (repetition); there is a morphological difference since, unlike repetitions, reduplicatives may undergo pluralization (pretty-pretties); and, finally, there is a semantic difference since the repetition does not drastically affect the meaning of the adjective, whereas the reduplicative has a derogatory meaning. However, as Forza (2011, 18) critically observes, the morphological and semantic differences between the repetition pretty, pretty and the reduplicative pretty-pretty appear to be by-products of the change of grammatical category: pretty, pretty is a sequence of two adjectives, whereas pretty-pretty is a noun. As for the meaning of syntactic reduplication, Wierzbicka (2003, 263264) proposes that by means of this construction the speaker insists on the validity or accuracy of his utterance. In describing someone’s eyes as neri neri (‘black black’) the speaker communicates that their colour was not close to black but exactly black; the reduplication implies that no exaggeration is involved. In more general terms, she suggests that “[b]y repeating a word (‘XX’) the speaker draws attention to that word, and insists on its strict correspondence with what is meant (‘I mean X, not something a little different from X’)” (Wierzbicka 2003, 264). Furthermore, as Wierzbicka’s (2003, 266) argument shows, syntactic reduplication adds an emotional dimension to the utterance – expressions such as bella bella (‘beautiful beautiful’) contain an emotional component as opposed to expressions with an intensifier such as molto bella (‘very beautiful’). On the basis of examples such as (1a-e), Hohenhaus (2004, 301) suggests the following meaning generalization: the doubled construction as a whole picks out ‘a prototype reading’ (for nouns ‘an XX is a proper/prototypical X’) or one of ‘extreme degree’ of what the duplicated item denotes (for adjectives, adverbs and verbs, ‘XX is really/properly/ extremely X’). He also notices that such constructions are typically used with linguistic (co-textual) elements which guide towards this reading. Specifically, such constructions are used in interrogatives or accompanied 4

Interestingly, the example Let it cool till it’s thick thick given by Sapir (1921, 76) in his discussion of reduplication seems to be an example of contrastive reduplication.

284

Chapter Thirteen

by negation: either the doubled construction or the single element is negated in overt contrast to the other: ‘not XX/X but rather X/XX’. Similarly to Hohenhaus (2004), Ghomeshi et al. (2004, 317) notice that in a number of cases, the doubled construction is explicitly contrasted with a non-reduplicated version of the same phrase in the nearby discourse. They also show that there is some similarity between contrastive reduplication and contrastive focus. In contrastive reduplication, the reduplicant bears a focus accent, which makes it similar to contrastively focused modifiers, for example not a BOOK-book vs. not a RED book. The difference between contrastive focus and contrastive reduplication boils down to the type of contrast signalled: contrastive focus on a word signals that it is being used in contrast with other words belonging to the same category; contrastive reduplication “signals that one meaning of the word is being contrasted with other possible meanings” (Ghomeshi et al. 2004, 317). This is illustrated in (3) and (4): (3)

(4)

a. I didn’t give the book to JOHN. [contrastive focus] Contrast set: {John, Bill, Dave, Sue, . . . } b. I didn’t give the book to JOHN-John. [contrastive reduplication] Contrast set: {John1, John2, . . . } a. It wasn’t a GOAT. [contrastive focus] Contrast set: {goats, horses, pigs, sheep, . . . } b. It wasn’t a GOAT-goat. [contrastive reduplication] Contrast set: {prototypical goats, non-prototypical or nonliteral (euphemistic, figurative) goats} (from Ghomeshi et al. 2004, 317)

3.1 Types of contrastive reduplication In his discussion of meaning of contrastive reduplication, or to use his term, lexical cloning, Horn (mentioned by Ghomeshi et al. 2004, 314) identifies four types: prototype meaning, literal meaning, intensified meaning and ‘value-added’ meaning. In the case of prototype meaning, contrastive reduplication signals that what is intended is the default or prototypical sense of the lexical item, as illustrated in (1a) above, for example. With reference to literal meaning, contrastive reduplication signals a literal, rather than loose, interpretation of a given expression. In (5):

Contrastive Reduplication and Relevance Theory

(5)

285

[Dialogue between a married couple, recently separated and now living apart.] A: Maybe you’d like to come in and have some coffee? B: Yeah, I’d like that. A: Just COFFEE-coffee, no double meanings. (Ghomeshi et al. 2004, 315)

the reduplicated expression indicates that what is intended by A is a literal, not euphemistic, interpretation of coffee (A’s utterance is an invitation to have coffee, not to have sex). The third type of lexical cloning is referred to as intensified meaning. It typically involves adjectives (adverbs or verbs) whose meaning gets intensified. In the example below, NERVOUS-nervous is taken in the sense ‘really/extremely/properly nervous’. (6)

A: Are you nervous? B: Yeah, but, you know, not NERVOUS-nervous. (Ghomeshi et al. 2004, 315)

The last type, the so-called ‘value-added’ meaning is illustrated in (7): (7)

A: I hear you guys are, um, living together now. B1: Well, we’re not LIVING TOGETHER-living together. B2: Well, we’re only LIVING TOGETHER-living together.

where the meaning of living together is enriched with context-dependent information. In B1, the reduplicated expression is understood as ‘living together as lovers’, whereas in B2, the meaning conveyed is that of ‘living together as roommates’ (Ghomeshi et al. 2004, 315; Huang 2009, 138). The last example shows that the choice of interpretation for a reduplicated expression is largely determined by context, which can be further illustrated by the use of the same form (HOME-home in (8) and DRINKdrink in (9)) to mean differently in different contexts: (8)

a. “... things have gone so bad that I decide to go home – HOME home. Mom and Dad home ...” (example from Hohenhaus 2004, 300, after Nick Hornby, High Fidelity, London: Penguin, 1995, p, 105) b. Lorelai: ... Mia is selling the inn. And that hit me hard ... I’m gonna be without a home. Luke: What do you mean? This is your home.

286

Chapter Thirteen

Lorelai: No, I mean a HOME home. A memory home. The inn is where Rory took her first step. It’s where I took my first step. It’s more of a home to me more than my parents’ house ever was.” (Hohenhaus 2004, 327, from Gilmore Girls, season 2, episode 8) (9)

a. …if you must have a ‘DRINK-drink’ go with the hard liquor. Why is hard liquor better than beer? b. ‘Do you want a bottle of wine?’ Mac asks. ‘I think I’ll have a DRINK-drink,’ I say, and when the waiter comes, I order a martini. c. and drink prices are never over the top (around 3 euros a shot and 8 euros a DRINK-drink). d. Are you looking for alcohol? Or just a DRINK-drink? (Whitton 2006, 19-21)

As can be seen from the speakers’ self-repairs, in (8a), HOME-home is used to mean ‘Mom and Dad home’, whereas in (8b) it conveys the narrowed sense of ‘memory home’. In (9a), DRINK-drink refers to both hard liquor and softer alcoholic drinks such as beer; in (9b) it denotes hard liquor only to the exclusion of softer alcoholic drinks such as wine; in (9c) DRINK-drink refers to a mixed drink as opposed to a shot; finally, in (9d) DRINK-drink refers to a non-alcoholic drink. Finally, homonyms can also be subjected to contrastive reduplication, with the result of disambiguation, as shown in (10). (10)

a. Son: I want a bat for my birthday. Dad: Aren’t you afraid it will bite you? Son: No, a BAT-bat. For little league. (= a baseball bat) b. Son: I want a bat for my birthday. Dad: A baseball bat? Son: No, a BAT-bat. With wings. (= a flying mammal) (adapted from Song and Lee 2011, 454)

Whitton argues that examples such as (10) would be rather unexpected under the prototype/intensification-based analyses suggested by Ghomeshi et al. (2004) and Hohenhaus (2004). The question she poses is how one meaning could be more prototypical if the two meanings of a word are essentially unrelated.

Contrastive Reduplication and Relevance Theory

287

3.2 Context-based approaches to contrastive reduplication The above discussion shows that contrastive reduplication is heavily context-dependent and that the vast majority of examples cannot be interpreted properly out of context. Furthermore, cases of contrastive reduplication are quite frequently misunderstood by addressees, which is why speakers often initiate (or are asked to launch) self-repairs by means of paraphrasing or clarifying the intended meaning. All this strongly suggests that analysing contrastive reduplication only in terms of prototype and intensification is largely an oversimplification (see Huang 2009). Whitton (2006), for example, rejects the prototype-based analysis of the meanings of contrastive reduplications, and instead suggests an analysis of their meanings in terms of context-based scales, containing possible meaning alternatives (contrast sets). In other words, it is context that allows for the construction of a contrast set for each contrastive reduplication. Subsequently, the contrast sets for examples (9a-d) are {(alcoholic drink, ~nonalcoholic drink)}, {(martini, ~wine)}, {(mixed drink in a glass, ~shot)} and {(nonalcoholic drink, ~alcoholic drink)}, respectively.5 An interesting context-sensitive approach to the meaning of contrastive reduplication has been presented by Song and Lee (2011), who put forward the so-called category-based account of contrastive reduplications involving a prototype category, dynamic prototypes and contrastive focus effects. According to Song and Lee (2011, 446) contrastive reduplications may refer to each of the three levels of a category: the prototype of a category, illustrated in (11), its subcategories, illustrated in (12), and the category itself, illustrated in (13). (11)

a. Mary: Would you like a wine? Jason: No, a DRINK-drink. I’ll have some vodka. b. Mary: Would you like a coke? Jason: No, a DRINK-drink. I’ll have some water.

In (11a), the word wine introduces the context of alcoholic drinks and hence the expression DRINK-drink will be interpreted as referring to the prototype of the alcoholic drink subcategory, whose most salient example is vodka, whereas in (11b) the word coke invokes the context of nonalcoholic drinks, in which the reduplication will refer to the prototype of the non-alcoholic drink subcategory, whose best member is undoubtedly 5

For a criticism of Whitton’s approach, see Song and Lee (2011, 445-446).

288

Chapter Thirteen

water (Song and Lee 2011, 446-447). Based on the observation that the prototype of the drink category varies according to the context of use of DRINK-drink, Song and Lee (2011, 450) suggest so-called ‘dynamic prototypes’ as the key to the successful interpretation of contrastive reduplications whose denotations vary in context. The role of the contrastive focus accent then boils down to making the hearer attend to the “immediately relevant alternatives” available in the context (e.g. the previous discourse) and “select one of them in terms of prototypicality” (Song and Lee 2011, 451). (12)

a. Jason: I want a drink. Mary: Here, have some coke. Jason: No, I want a DRINK-drink. With alcohol. b. Jason: I want a drink. Mary: Here, have some beer. Jason: No, I want a DRINK-drink. With no alcohol. (Song and Lee 2011, 451)

Assuming that there are two general subcategories of the drink category based on alcoholicity, the reduplication DRINK-drink will refer either to the alcoholic subcategory (12a) or to the non-alcoholic subcategory (12b). “[T]he contrastive focus serves to disambiguate the ... reduplications referring to the subcategories of a category” (Song and Lee 2011, 452). The expression DRINK-drink in (12a) will be interpreted as denoting an alcoholic drink (be it vodka, wine, beer, etc.) to the exclusion of nonalcoholic drinks such as water, coke, tea, etc., whereas in (12b) the interpretation of DRINK-drink will pick out a non-alcoholic drink to the exclusion of those containing alcohol. (13)

[Two attractive models are auditioning for a magazine at a photo shoot. During the audition, they are asked to hold a variety of props while posing for photographs. Jason holds sunglasses and a bottle of sunscreen while Mary holds a fan and a plastic replica of a glass of lemonade.] Jason: It’s really hot in here. I’d love a drink right now. Mary: (handing him the lemonade prop) Here! Jason: No. A DRINK-drink, silly! (from Song and Lee 2011, 453)

Contrastive Reduplication and Relevance Theory

289

In (13), DRINK-drink does not refer to the (context-dependent) prototype of the drink category (as in (11)) or to its subcategory (as in (12)), but to the drink category per se. Hence, the reduplication will be interpreted as having the meaning of ‘a genuine (real) drink’ as opposed to that of ‘a fake drink’ or ‘a drink prop’.

4. Contrastive reduplication and loose uses of language Given the crucial role played by context in the production and comprehension of reduplicated forms, contrastive reduplication seems to be essentially a pragmatic phenomenon. Ghomeshi et al. (2004, 311) rightly notice that “[t]he use of a word or phrase often leaves open some vagueness, lack of precision, or ambiguity” and that contrastive reduplication is used as a way to clarify such situations by restricting the denotation of a lexical item, even though they certainly miss the point by suggesting that the restriction is to the prototype only. It is worth noting that Ghomeshi et al. (2004, 316) admit that their approach to contrastive reduplication bears similarity to Lasersohn’s (1999) account of loose uses of language (what he calls ‘pragmatic slack’) and his analysis of so-called ‘slack regulators’. Lasersohn insightfully observes that utterances people produce are hardly literally true, though they are close enough to the truth for communicative purposes to be fulfilled. The amount of pragmatic slack (looseness) in one’s utterance depends both on the context of the utterance and on the presence (or absence) of a slack regulator, that is an expression which regulates the degree of slack, by reducing its amount. One of Lasersohn’s slack regulators is the word exactly as used in (14a): (14)

a. Mary arrived at exactly three o’clock. b. Mary arrived at three o’clock.

Even though, from a semantic perspective, the expressions exactly three o’clock and three o’clock have the same denotation: the set of Mary’s arrivals at the hour specified in the expression, the sentences in (14) will not be interpreted in the same way. The difference between (14a) and (14b) boils down to the amount of pragmatic slack each of them allows, or in other words, to the way exactly interacts with the ‘pragmatic halo’ of the expression three o’clock. Lasersohn defines a pragmatic halo as a set of items which (more or less closely) resemble the denotation of an expression. Due to the presence of exactly, there is less looseness in (14a) than in its counterpart in (14b), which shows that exactly has tightened the pragmatic halo of three o’clock (so that it more closely matches the

290

Chapter Thirteen

denotation of three o’clock) by removing from it Mary’s arrivals too remote from the literally specified time, for example, those between 2:50 and 3:10, which are perfectly acceptable as members of the pragmatic halo of three o’clock in (14b).

4.1 Contrastive reduplication, concept adjustment and hedges Lasersohn’s analysis of pragmatic slack and its regulators, on the one hand, and Ghomeshi et al.’s (2004, 316) claim that contrastive reduplication serves to narrow down the range of appropriate referents of a lexical item, on the other, are both reminiscent of the relevance-theoretic discussion of ad hoc concept formation, especially that of narrowing, and the role of linguistic devices such as hedges in the process of concept adjustment. In the relevance-theoretic framework, hedges can be successfully analyzed as expressions which signal the process of lexical concept adjustment, or to put it differently, they somewhat guide the hearer towards the speaker’s meaning. The group is by no means homogeneous: there are expressions that signal the broadening of a lexically encoded concept (e.g. sort of, regular, real) or its narrowing (e.g. typical), but there are also expressions (e.g. exactly, very) which, similarly to Lasersohn’s slack regulators, lower the extent of looseness in approximations (Waáaszewska forthcoming 2014; for the relevancetheoretic discussion of sort of and typical, see also Itani 1995 and Miskovic-Lukovic 2009). The existence of hedges in a language is only to be expected: according to Sperber and Wilson (1995, 289, n. 21; after Blakemore e.g. 1987), “a language might develop certain structures whose sole function was to guide the interpretation process by stipulating certain properties of context and contextual effects.” The idea that hedges and possibly other devices have developed to guide hearers in the interpretation of speakers’ meaning is a natural outcome of two basic assumptions of Relevance Theory.6 First, successful communication is not guaranteed since, on Sperber and Wilson’s (1995, 45) approach, “communication is governed by a less-than-perfect heuristic ... [and – E.W.] ... failures in communication are to be expected.” Second, an utterance serves as only a piece of evidence about the speaker’s meaning and it is by inferential work (based on the utterance and its context) that the hearer can arrive at the 6

Obviously, hedges are not claimed to be an indispensible element in an act of communicating. In the interpretation process, the hearer is naturally guided towards the speaker’s meaning by expectations of relevance raised by the very act of communicating.

Contrastive Reduplication and Relevance Theory

291

meaning conveyed by the speaker. This is linked to the view, endorsed by relevance theorists (e.g. Carston 2002; Wilson 2003; Vega Moreno 2007; Wilson and Sperber 2012), that there does not have to be (and rarely is) a relation of sameness between the concept communicated by the speaker’s use of a certain word and the concept encoded by that word. A lexically encoded concept is inferentially adjusted by the hearer in the process of utterance interpretation to yield an occasion-specific sense, referred to as an ad hoc concept. There are two types of pragmatic processes involved in ‘ad hoc’ concept formation: lexical narrowing and broadening. In narrowing, a word is used “to convey a more specific sense than the encoded one, with a more restricted denotation ...” (Wilson and Carston 2007, 232). For example, in (15) (15)

I will never marry Tom. He is a bachelor.

the word bachelor may be interpreted not as conveying the lexically encoded concept BACHELOR denoting unmarried men, but as communicating the ad hoc concept BACHELOR* whose denotation is restricted only to a subset of the bachelor category – those unmarried men for whom sexual conquest is more important than a stable relationship with one partner. The process of concept narrowing can be signalled by means of hedges such as typical. According to Itani (1995, 92-94), typical guides the hearer to construct an ad hoc concept which will pick out only stereotypical (or prototypical) members of the original category. Hence, typical in (16) signals that the ad hoc concept BACHELOR* denotes only those bachelors who display stereotypical properties such as machismo and promiscuity. (16)

I will never marry Tom. He is a typical bachelor.

In broadening, a word is used “to convey a more general sense than the encoded one, with a consequent expansion of the linguistically-specified denotation” (Wilson and Carston 2007, 232). It seems that the process of broadening may lead to the dropping of the definitional (logical) properties of a concept or at least to the change of their nature – they can become fully context-dependent non-definitional properties (Itani 1995, 90). ‘Being unmarried’ is no longer a definitional property of the concept communicated by the use of the word bachelor in (17), as indicated by the fact that the statement does not involve contradiction. (17)

Tom has just got married but he is still a bachelor.

292

Chapter Thirteen

The ad hoc concept BACHELOR** in (17) will denote those married men who exhibit properties attributable to a stereotypical bachelor, e.g. being an inveterate womanizer. Just as there are hedges indicating the process of concept narrowing, there are also hedges that signal the process of broadening. Undoubtedly, example (18) will be interpreted along the same lines as (17). (18)

Tom has just got married but he is still sort of a bachelor.

Sort of, then, functions as a signal of the process of ad hoc concept formation in which the lexically encoded concept it premodifies undergoes pragmatic broadening. Generally speaking, what hedges such as sort of and typically have in common is the way they affect the relevance of the utterance in which they are used. It seems that they are used by the speaker to reduce the mental effort the hearer needs to spend on reaching the intended interpretation, and as such may encode procedural meaning (e.g. sort of as argued by Miskovic-Lukovic 2009). At this point, it should be noted that in Relevance Theory there is a distinction between two types of encoded meaning as by words: conceptual (concepts) and procedural (procedures). The first to address this distinction was Diane Blakemore (1987), who analyzed discourse connectives such as but, so or after all as encoding procedures, in contrast to ‘content’ words such as nouns or verbs, standardly assumed to encode concepts. This generally means that, unlike concepts, procedures do not contribute to the truth conditions for an utterance, but rather place certain constraints on the process of pragmatic inferencing necessary to arrive at the intended interpretation (Carston 2002, 57).7 In other words, “[p]rocedural meanings ... are encoded instructions that specify computational operations to be performed during interpretation and, more precisely, to access a particular context for interpretation” (Escandell-Vidal et al. 2011, xix). A classic example of a word encoding a procedure is but, whose role is to instruct the hearer that the part of a statement introduced by this conjunction will counter an implication triggered by the first of the

7 It is worth noting that the initial claim that the distinction between conceptual and procedural meaning should be seen as coinciding with the distinction between truth-conditional and non-truth-conditional meaning has long been abandoned. According to Wilson and Sperber (1993), both concepts and procedures may, or may not, be truth-conditional since each of them may, or may not, contribute to the proposition expressed (for a thorough discussion of the conceptual-procedural distinction, see e.g. Bordería 2008; Wilson 2011).

Contrastive Reduplication and Relevance Theory

293

conjoined elements (Blakemore 2002, 90-91), e.g. It’s raining (implication: ‘the grass is wet’), but the grass is not wet.

4.2 Contrastive reduplication and procedural meaning To the best of my knowledge, the only relevance-theoretic account of reduplication has been suggested by Dansieh (2011), who has analysed a number of translations from English into West African languages, particularly Dagaare. Dansieh (2011) shows that there are two types of uses of reduplication in the target language: those that have become grammaticalized so that their meanings are not modified in context, and those that are context-sensitive. The latter are regarded as conveying some general procedural information that can direct the addressee toward a range of possible interpretations that need to be further developed by means of inferential processing. My claim is that, similarly to those non-grammaticalized contextsensitive uses of reduplication discussed by Dansieh, the contrastive reduplication construction in English should also be analyzed in terms of procedural meaning. Specifically, it is the element that bears contrastive stress (reduplicant) that encodes a procedure which directs the addressee’s attention to the lexical element that has been duplicated and facilitates the choice of interpretation of that lexical element by narrowing down the range of potential readings to the most relevant ones. This proposal seems to agree with Sperber and Wilson’s (1995, 212) observation that contrastive stress is a natural device used for highlighting a particular aspect of an utterance and also with Wilson and Wharton’s (2006, 1571) claim that the notion of procedural encoding applies to properly linguistic prosodic elements such as lexical stress, lexical tone and fully grammaticalized aspects of sentence stress and intonation. Undoubtedly, contrastive reduplication introduces an additional dimension into utterance interpretation in that a reduplication such as LINGUIST-linguist in (19) (19)

... the linguist who does not display at least a token interest in the Chomskyan endeavor is not considered a ‘linguist linguist’ in the back of the minds of a great many in the field ... (John McWhorter, The Power of Babel, New York: Times Books, p. 282)

cannot be paraphrased as ‘a real linguist’. This shows that the use of this process gives rise to some kind of novelty for sarcastic, humorous or other

294

Chapter Thirteen

effects (Huang 2009, 141) which may result from the fact that, formally, contrastive reduplications sound like imitations of baby-talk reduplications. On the other hand, bearing in mind that reduplication is one of the most natural word-formation processes, speakers may resort to contrastive reduplication to cope with the discrepancy between the need to communicate certain meanings and a temporary or permanent lack of the right word or expression. This may bring to mind the Communicative Principle of Relevance, according to which the choice of the most relevant stimulus depends on the communicator’s abilities and preferences – in this case, apparently the communicator’s lexical limitations affect his/her choice of a stimulus. Finally, the procedural account of contrastive reduplication presented here shows that, as rightly observed by Wierzbicka (2003), it is a distinct phenomenon from repetition, or epizeuxis, one of the classical stylistic devices, illustrated in (20). (20)

a. There were houses, houses everywhere. b. There’s a fox, a fox in the garden. c. My childhood days are gone, gone.

Sperber and Wilson (1995, 219) show that the effects of repetition on utterance interpretation are by no means constant and have to be worked out in different ways for different examples. Accordingly, (20a) conveys that there were a great number of houses, (20b) suggests that the speaker was excited/surprised about the presence of a fox in the garden, (20c) conveys that the speaker was moved by the fact that her childhood days had disappeared. Pilkington (2000), who, alongside his discussion of metaphor has developed a relevance-theoretic account of poetic effects with reference to sound patterning and repetition, argues that cases of repetition redirect the hearer’s attention to the lexical concepts already activated, in this way inviting further derivation of weak implicatures. For example, in the case of (20b), apart from deriving a strongly communicated implicature like ‘The chickens are in danger’, the hearer may feel invited by the speaker’s repetition to explore the concept FOX more deeply, and consequently arrive at weaker implicatures such as ‘Seeing a fox in this area is a rarity’ clearly related to the speaker’s mental or emotional state (disbelief, puzzlement, excitement, etc.) (Sperber and Wilson 1995, 219-220; Pilkington 2000, 125-126). This is clearly different from contrastive reduplication, because in the latter phenomenon, which is found mostly in conversation rather than literature, the stressed reduplicant does not invite the hearer to expend extra linguistic processing effort on

Contrastive Reduplication and Relevance Theory

295

further exploration of the lexically encoded concept, but, in a sense, it facilitates the hearer’s task by signalling the need for narrowing the concept encoded by a given word.

5. Conclusion Contrastive reduplication, also known as lexical cloning or identical constituent compounding, is a fairly common and highly contagious wordformation process observable mostly in colloquial spoken American English but also in other dialects of English and other languages as well. Contrastive reduplication is an interesting phenomenon, sufficiently distinct from related or similar phenomena to deserve separate consideration. Formally, it involves combining two identical words of which the first receives contrastive stress, and as such it is substantially different from other types of reduplication as well as from compounding, which it superficially resembles. The presence of contrastive stress makes contrastive reduplication structurally similar to contrastive focus but indicates different contrast, not between words of the same grammatical category that could have been used, but between possible meanings of one word. This contrast serves as a signal that a certain specific meaning has to be derived from the use of a doubled word, literal, prototypical, intensified, or some other, contextually determined. Thus, from a functional angle, contrastive reduplication works in a way similar to hedges, which, in relevance-theoretic terms, involve a procedure indicating the need for concept adjustment. In the case of contrastive reduplication, the adjustment takes the form of narrowing of the concept encoded by the repeated word. The presented description of contrastive reduplication shows that it is not sufficient to focus on the formal properties of the phenomenon, and that it is necessary to pay attention to its use and to the context-dependence of its meaning. The attempted analysis demonstrates that a pragmatic theory such as Relevance Theory is capable of accounting for morphological phenomena. The relevancetheoretic approach also clarifies the difference between contrastive reduplication and repetition (epizeuxis), which redirects the hearer’s attention to the lexical concepts already activated, inviting further derivation of weak implicatures.

296

Chapter Thirteen

References Blakemore, Diane. 1987. Semantic constraints on relevance. Oxford: Blackwell. —. 2002. Relevance and linguistic meaning: The semantics and pragmatics of discourse markesrs. Cambridge: Cambridge University Press. Booij, Geert. 2005. The grammar of words: An introduction to linguistic morphology. Oxford: Oxford University Press. Bordería, Salvador Pons. 2008. Do discourse markers exist? On the treatment of discourse markers in Relevance Theory. Journal of Pragmatics 40: 1411-1434. Braun, Maria and Ingo Plag. 2003. How transparent is creole morphology? A study of Early Sranan word-formation. In Yearbook of Morphology 2002, edited by G. Booij and J. van Marle. Dordrecht: Kluwer Academic Publishers, 81-104. Bybee, Joan L. 1985. Morphology: A study of the relation between meaning and form. Amsterdam and Philadelphia: John Benjamins. Carston, Robyn. 2002. Thoughts and utterances. Oxford: Blackwell. Dansieh, Solomon Ali. 2011. The pragmatics of reduplication: Implications for translating. Journal of Pragmatics 43: 164-174. Escandell-Vidal, Victoria, Manuel Leonetti and Aoife Ahern. 2011. Introduction: Procedural meaning. In Procedural meaning: Problems and perspectives, edited by V. Escandell-Vidal, M. Leonetti and A. Ahern. Bingley, UK: Emerald, xvii-xlv. Forza, Francesca. 2011. Doubling as a sign of morphology: A typological perspective. Journal of Universal Language 12, no. 2: 7-44. Ghomeshi, Jila, Ray Jackendoff, Nicole Rosen and Kevin Russell. 2004. Contrastive focus reduplication in English (the SALAD-salad paper). Natural Language and Linguistic Theory 22: 307-357. Hohenhaus, Peter. 2004. Identical constituent compounding – a corpusbased study. Folia Linguistica 38: 297-331. Horn, Laurence. 2006. Speaker and hearer in neo-Gricean pragmatics. Waiguoyi 164: 2-26. Huang, Yan. 2009. Neo-Gricean pragmatics and the lexicon. International Review of Pragmatics 1: 118-153. Inkelas, Sharon and Cheryl Zoll. 2005. Reduplication: Doubling in morphology. Cambridge: Cambridge University Press. Itani, Reiko. 1995. A relevance-based analysis of Lakoffian hedges: Sort of, a typical and technically. UCL Working Papers in Linguistics 7: 87105.

Contrastive Reduplication and Relevance Theory

297

Lasersohn, Peter. 1999. Pragmatic halos. Language 75: 522-551. Levinson, Stephen C. 2000. Presumptive meanings: The theory of generalized conversational implicature. Cambridge, MA: MIT Press. Mattiello, Elisa. 2013. Extra-grammatical morphology in English. Berlin and Boston: Walter de Gruyter. Merlini Barbaresi, Lavinia. 2008. Extra-morphological morphology: English reduplicatives. In Words in action: Diachronic and synchronic approaches to English discourse. Studies in honour of Ermanno Barisone, edited by J. Douthwaite and D. Pezzini. Genova: ECIG. Miskovic-Lukovic, Mirjana. 2009. Is there a chance that I might kinda sort of take you out to dinner?: The role of the pragmatic particles kind of and sort of in utterance interpretation. Journal of Pragmatics 41: 602-625. Pilkington, Adrian. 2000. Poetic effects: A Relevance Theory perspective. Amsterdam and Philadelphia: John Benjamins. Sapir, Edward. 1921. Language: An introduction to the study of speech. New York: Harcourt Brace & Company. Song, Myounghyoun and Chungmin Lee. 2011. CF-reduplication in English: Dynamic prototypes & contrastive focus effects. Proceedings of SALT 21: 444-462. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and cognition. 2nd ed. Oxford: Blackwell. Stockwell, Robert and Donka Minkova. 2001. English words: History and structure. Cambridge: Cambridge University Press. Thun, Nils. 1963. Reduplicative words in English. A study of formations of the types ‘tick-tock’, ‘hurly-burly’, and ‘shilly-shally’. Lund: Carl Bloms. Vega Moreno, Rosa Elena. 2007. Creativity and convention: The pragmatics of everyday figurative speech. Amsterdam and Philadelphia: Benjamins. Waáaszewska, Ewa. Forthcoming 2014. How to signal lexical concept adjustment: A relevance-theoretic perspective on hedging. International Review of Pragmatics 6, no. 2. Whitton, Laura. 2006. The semantics of contrastive focus reduplication in English: Does the construction mark prototype-prototype? Unpublished manuscript, Stanford University, CA. Wierzbicka, Anna. 2003. Cross-cultural pragmatics: The semantics of human interaction. 2nd ed. Berlin and New York: Mouton de Gruyter. Wilson, Deirdre. 2003. Relevance and lexical pragmatics. Rivista di Linguistica 15, no. 2: 273-291.

298

Chapter Thirteen

—. 2011. The conceptual-procedural distinction: Past, present and future. In Procedural meaning: Problems and perspectives, edited by V. Escandell-Vidal, M. Leonetti and A. Ahern. Bingley, UK: Emerald, 331. Wilson, Deirdre and Robyn Carston. 2007. A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In Pragmatics, edited by N. Burton-Roberts. Houndmills, Basingstoke: Palgrave Macmillan, 230-259. Wilson, Deirdre and Dan Sperber. 1993. Linguistic form and relevance. Lingua 90: 1-25. —. 2012. Meaning and relevance. Cambridge: Cambridge University Press. Wilson, Deirdre and Tim Wharton. 2006. Relevance and prosody. Journal of Pragmatics 38: 1559-1579.

CHAPTER FOURTEEN ON THE ORIGIN AND MEANING OF SECONDARY INTERJECTIONS: A RELEVANCE-THEORETIC PROPOSAL MANUEL PADILLA CRUZ, UNIVERSITY OF SEVILLE, SPAIN

1. Introduction Interjections are linguistic elements that individuals use to express mental states, attitudes or reactions to perceived stimuli. Quite often, they are seen as instinctive, involuntary or uncontrolled verbalisations, i.e. quasireflexes (Nicoloff 1990, 214), owing to their almost automatic and/or unconscious production. However, in many cases interjections are not produced instinctively (ĝwiatkowska 2006), but fully intentionally in ostensive-inferential communication. With them, speakers try to make manifest in a more precise or less precise way a certain informative intention amounting to assumptions about feelings, emotions or attitudes which they experience because of or project to some state(s) of affairs. In these cases, the production of interjections involves a conscious evaluation of the spatio-temporal setting of a conversation or specific elements thereof, and the selection of an item from among a wider or narrower set of possible candidates on the basis of its suitability for what speakers want to express. The peculiar and anomalous formal features of interjections have confused grammarians throughout history and led some to regard them as peripheral linguistic elements (Quirk et al. 1985). Their lack of constant meaning and context-boundedness has also induced others to doubt their status as words. This might justify why interjections have received so little attention over the years. In fact, it was not until the publication of special issues of Journal of Pragmatics and Langages in 1992 and 2006, respectively, and a series of studies from different frameworks that they

300

Chapter Fourteen

have progressively attracted more attention. One of those frameworks has been Relevance Theory (Sperber and Wilson 1986, 1995; Wilson and Sperber 2002, 2004), where researchers have examined what interjections communicate and how they contribute to communication (Wharton 2003, 2009; Waáaszewska 2004; Blakemore 2010, 2011; Padilla Cruz 2009a, 2009b, 2010). This paper does not focus on the whole class of interjections, but just on a sub-type of them: ‘secondary’ interjections. These are linguistic elements which, borrowed from other word-classes, such as nouns, adjectives, verbs and adverbs, are frequently used to communicate feelings, emotions or attitudes. The aim of this paper is twofold. Firstly, it seeks to propose an account of the origin of secondary interjections grounded in Relevance Theory (Sperber and Wilson 1986, 1995; Wilson and Sperber 2002, 2004). More specifically, it purports to explain why lexical elements belonging to distinct word-classes can be transferred to the class of interjections. Secondly, it tries to suggest an explanation of the wide expressive potential of secondary interjections. This paper starts by summarising some of the general characteristics of interjections and explaining what primary and secondary interjections are.  ext, its third section reviews recent research on interjections. Then, it introduces the relevance-theoretic approach to concepts and lexical pragmatic processes, on which the proposal about the origin of secondary interjections and their expressive potential made in this paper rests squarely. This is presented in its fifth section, which tries to elucidate if the members of the word-classes from which secondary interjections undergo the same lexical pragmatic processes and why the resulting interjections may have differing expressive meanings. Finally, the concluding section summarises its main claims and suggests some directions for future research.

2. General characteristics and types of interjections Interjections are normally considered paralinguistic elements because of their phonological and morphological anomalies, their relative syntactic independence, and their occurrence in discourse accompanying other linguistic chunks. This consideration has resulted in a historical lack of agreement as to whether they are one of the traditional word-classes or parts of speech1. As opposed to linguists who do not include interjections 1 The word-class classification was firstly made by Dionysius Thrax in the 2nd century B.C. for Classical Greek. Based on a series of morphological, syntactic and

On the Origin and Meaning of Secondary Interjections

301

among their word-classes (Huddleston 1988), for those who do so interjections are an ‘open’ (Buridant 2006) or ‘closed’ class (Quirk et al. 1972; Quirk and Greenbaum 1973; Quirk et al. 1985; Greenbaum 2000), depending on the possibility of incorporating new items2. Their class consists of language- and culture-specific items which express emotions and other modalities, whose most remarkable phonological, morphological, semantic, syntactic traits are the following (Quirk et al. 1972, 1985; Wierzbicka 1991, 1992; Ameka 1992, 2006; Aijmer 2004)3: 1.

2. 3.

4.

They do not tend to be homophonous or homonymous with other lexical items and their peculiar phonological layout places them outside the regular linguistic system4. They usually are monomorphemic and invariable, as they do not receive inflectional or derivational affixes5. They do not have denotative, but indexical meaning, i.e. they behave as pointing devices that signal elements in the external reality. They normally appear as stand-alone utterances or independent tone units, so they are loosely attached to the rest of the constituents of a sentence.

semantic properties, he distinguished eight word-classes: nouns, verbs, participles, articles, pronouns, prepositions, adverbs and conjunctions. This was later modified by Apollonius Dyscolus and Priscian, in order to adapt it to Latin. Latin having no articles, these grammarians replaced that class by interjections in order to preserve the initial number of classes. For a review, see Buridant (2006). 2 Quirk et al. (1985, 67) consider interjections a marginal and anomalous wordclass, the other classes being closed (prepositions, pronouns, determiners, conjunctions, modal verbs and primary verbs), open (nouns, adjectives, full verbs and adverbs) and numerals. 3 See López Bobo (2002) for a comparison between interjections and conjunctions, sentential adverbs, imperatives or vocatives. 4 Although many interjections are monosyllabic, others are non-syllabic (‘tsktsk!’, ‘shh!’), contain unusual sounds (e.g. the dento-alveolar click in ‘tut-tut’) or sound combinations in a language (e.g. the bilabial vibrant in ‘brrr!’), show iteration (e.g. the Spanish interjection ‘ayyyyyyyy!’) or reduplication (e.g. the Spanish interjections ‘¡ayayay!’ or ‘¡huyhuyhuy!’) (Gehweiler 2008, 73). 5 However, when interjections are converted into nouns or verbs, they can receive inflectional morphology (Greenbaum 2000, 183). For instance, ‘pooh-pooh’ and ‘wow’, which can be verbs, or ‘boo’ and ‘tut-tut’, which can be nouns or verbs, can take third person singular and preterit or plural morphemes.

302

Chapter Fourteen

This broad formal characterisation has led authors to gather under the label of interjections words that express emotions (1), words and expressions used to carry out some conversational routines (2), expletives, swear words and imprecations (3), attention-getting signals (4), some particles and response words (5), words directed at animals (6), and onomatopoeias (7) (Wierzbicka 1991, 1992; Ameka 2006; Gehweiler 2008)6: (1) (2) (3) (4) (5) (6) (7)

Yuk! Ugh! Phew! Wow! Oh! ¡Huy! ¡Ah! Hello! Thank you! Good bye! Ok! ¡Hola! ¡Gracias! ¡Adiós! ¡Vale! Shit! Bastard! Hell! Jesus! God! Christ! ¡Mierda! ¡Hostia! ¡Coño! ¡Jesús! ¡Dios! ¡Señor! Hey! Psst! Eh! Look! ¡Ojo! ¡Mira! ¡Oye! Yes!  o! ¡Sí! ¡ o! Whoa! ¡Arre! Hehe! ¡Jeje!

Those monomorphemic interjections which are not homophonous or homonymous with other words, constitute independent, non-elliptical utterances and do not co-occur with other word classes are normally alluded to as ‘primary’ interjections (Wierzbicka 1991, 1992; Ameka 1992, 2006). Goffman (1981) called them ‘response cries’. In contrast, those words which behave as interjections, even if not sharing all their prototypical characteristics, are transferred from other word-classes, have an independent semantic value and are also used as non-elliptical utterances to express a mental state are ‘secondary’ interjections. Within this broader category a further distinction can be made between ‘simple’ secondary interjections – i.e. only one item – exemplified by calls of alert or attention (8) or some swear words and imprecations (9), and ‘interjectional phrases’ (10) – i.e. with phrasal structure (Hill 1992; Wilkins 1992): (8) (9) (10)

Help! Fire! Careful! Damn! Hell! Heavens! Fuck! Shit! Bloody hell! Dear me! My Goodness! My God!

Swear words and imprecations are also known as ‘expletives’. They are further sub-divided into ‘taboo expletives’, i.e. with homonyms connected 6

Examples are given in English and Spanish.

On the Origin and Meaning of Secondary Interjections

303

to taboo areas of religion, sex or excretions; ‘moderated expletives’, which somehow camouflage their taboo origin (e.g. ‘gee!’), and ‘euphemistic taboo expletives’, i.e. with homonyms that are not their base forms (e.g. ‘goodness!’) (Biber et al. 1999, 1094; Gehweiler 2008, 73-74). If compared to primary ones, secondary interjections are more creative and open to newcomers from the word-classes of nouns, adjectives, verbs and adverbs (Quirk et al. 1985, 74; Buridant 2006, 7). This is the type of interjections that this paper centres on. It excludes from discussion both moderated and euphemistic taboo expletives, as an account of their origin and meaning would have to address historical considerations and would hence far exceed the scope of this paper. More specifically, this paper tries to explain which lexical process(es) might make it possible for some words initially belonging to other categories to enter that of secondary interjections.

3. Contributions on interjections In spite of their historical neglect, the word-class of interjections has recently awoken the interest of researchers from different linguistic disciplines, who have undertaken rather illuminating descriptive and contrastive studies on, for example, the usage and peculiarities of interjections in certain languages (Eastman 1992; Aijmer 2004), their values and functions in specific interactional contexts (O’Connell and Kowal 2005; O’Connell, Kowal, and Ageneau 2005; Shenhav 2008) or judgements about individuals’ verbal fluency depending on their presence in speech (Hegde and Hartman 1979). Researchers have also carried out developmental studies on their acquisition by children and their progressive incorporation in speech (Meng and Schrabback 1999; Montes 1999). Relevance theorists have not been oblivious to interjections either. Relying on the procedural and conceptual distinction (Blakemore 1992; Wilson and Sperber 1993), Wharton (2003, 2009) re-analysed them in procedural terms. In his view, interjections would not encode conceptual content or contribute to the truth-conditional content of utterances, but encode a procedural meaning that would assist hearers in the construction of higher-level explicatures concerning the emotion, feeling or attitude that the speaker expresses. This re-analysis of interjections has been the basis for Blakemore’s (2010, 2011) work on the ineffability of their expressive meaning, according to which they would work very much like gestures and tones of voice, giving rise to cognitive effects that are weakly communicated.

304

Chapter Fourteen

Wharton’s (2003, 2009) re-analysis has been questioned by Waáaszewska (2004) and Padilla Cruz (2009a, 2009b, 2010), who doubt that interjections only contribute to higher-level explicatures in those cases in which there is no lower-level explicature. Padilla Cruz (2009a) reinterprets their procedural meaning as instructions that would make the hearer look for the stimulus that causes the feeling the speaker expresses or the person, object or event to which the speaker projects it. As regards the nonconceptuality of interjections, Padilla Cruz (2009b, 2010) has given some reasons to reconsider their conceptual content and suggested that they might encode very general concepts amenable to pragmatic adjustments. Regarding secondary interjections, a fruitful strand of research has looked into their sociolinguistic distribution and peculiarities in some languages and varieties (Cestero Mancera and Moreno Fernández 2008; Mayol 2008a, 2008b; Tsibulsky 2008; Ljung 2009), their meaning in situational, discourse and social contexts (Kochelman 2003) or their usage as discourse markers (Brinton 1996; Norrick 2009). Nevertheless, the origin of secondary interjections has received scarce attention. Apart from historical explanations of some language-specific items in terms of phonological evolution (van der Hoek 2007), extant accounts agree that secondary interjections arise after a gradual process of ‘grammaticalisation’ by means of which the lexical entries of the items giving rise to them acquire a new grammatical and morphological status. This process progressively causes the semantic content of those items to become more abstract and general, thus enabling them to communicate notions or messages about emotions, feelings and attitudes (Traugott 1989, 1995; Hopper 1991; Hopper and Traugott 1993; Oppermann-Marsaux 2008). This process of grammaticalisation is linked to another, ‘subjectification’ (Traugott 1992, 1995), as a result of which words that initially encoded concepts and contributed to truth-conditional content acquire new functions that assist in the recovery of the speaker’s attitude. The resulting secondary interjections have a diverse communicative potential, can appear in different discourse contexts and become rather fixed expressions, since their constituents cannot be altered or replaced by others (López Bobo 2002; Buridant 2006). An example is Gehweiler’s (2008) account of the origin of ‘gee!’ from ‘Jesus’. Relying on Traugott and Dasher’s (2002) ‘Invited Inferencing Theory of Semantic Change’, she shows that ‘Jesus’ firstly gave rise to a homonymous secondary interjection which foregrounded a pragmatic meaning, while “… the original semantic component, the reference to the religious person, was backgrounded and eventually lost” (Gehweiler 2008, 83). Initiated by speakers and motivated by the occurrence of the original

On the Origin and Meaning of Secondary Interjections

305

proper name in non-religious contexts as an invocation, this process deprived that name of coded meaning and resulted in the item having utterance-token meanings, pragmatically polysemous meanings and new semantically polysemous meanings (cf. Traugott and Dasher 2002, 35). Nevertheless, since the proper name and the expletive were homophonous, speakers might have established some connection between the latter and the religious person referred to by the originating proper name (Gehweiler 2008, 74). However, this account does not clarify what triggers grammaticalisation and subjectification processes or what makes it possible for some lexical items to shift grammatical category and enter that of interjections. The answer to these problems could be sought in lexical processes undergone by some items under certain circumstances. If so, which specific lexical processes could enact the transfer of nouns, verbs, adjectives and adverbs to the word-class of interjections? And, more specifically, would items belonging to diverse classes undergo the same process(es), or would different processes affect them? And a still further problem: when lexical items from different word-classes are transferred to that of interjections, do they stabilise in the latter with one meaning or may they be used to express different emotions, feelings and attitudes? If the latter option is true, how may this be possible? What follows is an attempt to suggest possible answers on the basis of some relevance-theoretic postulates on concepts and lexical pragmatic processes, which the next section summarises.

4. Relevance Theory and concepts 4.1 Words and concepts Like Fodor (1998), Relevance Theory claims that mental concepts may have words as their natural language counterpart. It portrays the mental concepts encoded by words as (i) atomic rather than decompositional7, (ii) lacking definitions in terms of necessary and sufficient component features, and (iii) not structured around prototypes. Atomic concepts consist of an address in memory, which gives access to varied information stored in three types of entries:

7

See Hall (2011) for a detailed explanation of concept atomism.

Chapter Fourteen

306

a)

The ‘logical entry’, or the small, finite and relatively stable set of defining properties of the concept. It specifies its logical relations with other concepts (Sperber and Wilson 1995, 92). b) The ‘encyclopaedic entry’, or information about the extension and/or denotation of the concept, as well as diverse assumptions and personal experience organised in schemas, scenarios, scripts, etc. Its content varies across speakers and times because it can receive new information at any time, whose storage depends on factors like recentness, saliency, easiness or difficulty of accessibility of its items (Sperber and Wilson 1995, 93). c) The ‘lexical entry’, which includes phonetic and grammatical properties of the word corresponding to the concept (Carston 2002, 321-322)8.

Many atomic concepts have these three entries, but others lack one of them. For instance, ‘and’ encodes a concept that has no extension and so lacks an encyclopaedic entry, whereas proper names have an empty logical entry because they do not have logical properties (Sperber and Wilson 1995, 92; Carston 2002, 322). Moreover, correspondences between natural language words and mental concepts are not always accurate (Sperber and Wilson 1997; Carston 2002). Some concepts do not map onto just one single word, but onto different words: synonyms. For other, differing concepts correspond to one single word in natural languages: homonyms. Still other concepts may not be effable through natural language words, but through linguistic structures, like phrases or idioms: complex concepts, whose meanings are determined (at least in part) compositionally. Finally, there are words, like personal pronouns, onto which no concept is mapped; they encode procedures and behave as slot fillers. At most, they encode some very general conceptual content which must be pragmatically 8

In his ‘Dynamic Model of Meaning’, Kecskes (2004, 2008) argues that the meaning of a word consists of ‘coresense’, or its relatively stable denotation; ‘culture-specific conceptual properties’, or idiosyncratic information associated with the word by members of a cultural group; ‘word-specific semantic properties’, or the lexical properties of the word, and ‘consense’, or its actual contextual meaning (Kecskes 2008, 393-395). Coresense and culture-specific conceptual properties would correspond to the encyclopaedic entry of a concept, while wordspecific semantic properties would correspond to the information contained in the lexical entry. In Kecskes’s (2004, 2008) model there is no such thing as a logical entry. Besides, the consense would be the result of conceptual adjustment of a word. Owing to the simplicity and explanatory potential of the relevance-theoretic description of concepts, this work will adhere to it.

On the Origin and Meaning of Secondary Interjections

307

determined: a ‘pro-concept’ that needs fleshing out into a full concept (Wilson and Sperber 1993; Sperber and Wilson 1997; Wilson 1997). Sperber and Wilson (1997) make an even more radical claim: words must be seen as not encoding precise, fully determined concepts but proconcepts that require pragmatic adjustment in order to be specified or restricted. Accordingly, pro-concepts are so common that “… all words behave as if they encoded pro-concepts” (Sperber and Wilson 1997, 108; emphasis in the original). Since words encode pro-concepts, they behave as “… pointers to a conceptual space, on the basis of which … an actual concept … is pragmatically inferred” (Carston 2002, 360). Then, the concept associated with a word is merely considered a clue to the actual concept that the speaker intends to communicate when using that word. Accordingly, almost any content word, even if linguistically unambiguous, can communicate distinct, though related, meanings in different contexts (Sperber and Wilson 1997; Carston and Powell 2005). This seems to be the case for some verbs like ‘put’, ‘take’ or ‘make’9 and adjectives like ‘long’ or ‘empty’ This might also be the case for words liable to be transferred to the sub-type of secondary interjections. When using some words, speakers can convey non-lexicalised concepts or concepts not strictly related to the ones customarily associated with them. They may do so on the grounds of the information contained in the logical and encyclopaedic entries of those concepts. Speakers would trust hearers to work out the occasion-specific concepts by tracing the resemblance relations holding between the concepts customarily associated with those words and the new concepts they intend to communicate (Carston 2002). To put it differently, when some words are used, the concepts that they lexically encode would trigger a pragmatic process which, guided by expectations of relevance, yields different concepts. The resulting concepts may be narrower or broader than those initially encoded, so they are ‘ad-hoc concepts’ (Carston 2002). Their construction requires the interaction of information contained in the logical and encyclopaedic entries of lexically encoded concepts with contextual information and is accomplished as a consequence of the constant search for the optimal relevance of utterances. Hall (2011, 2) underscores that ad-hoc concepts are not “… already established elements of the subject’s conceptual repertoire, but new concepts constructed in the comprehension process, … genuinely novel ones”.

9

In the case of verbs like these, individuals might have several distinct concepts, so the one actually retrieved on a specific occasion could also depend, among other factors, on their direct objects (Hall 2011, 2).

Chapter Fourteen

308

4.2 Ad-hoc concept construction Relevance Theory distinguishes two main types of ad-hoc concept construction: ‘narrowing’ and ‘broadening’ (Carston 1997, 2002; Wilson 2004; Sperber and Wilson 2008; Wilson and Carston 2006, 2007).  arrowing occurs when the linguistically specified denotation of a word is restricted and the word conveys “… a more specific sense than the encoded one” (Wilson 2004, 344). It operates by limiting the denotation of the lexically encoded concept to just a subpart of it (Carston and Powell 2005, 283). As a result, the word preserves its literal meaning because none of the logical properties of its lexicalised concept has been altered or dropped (Hall 2011, 2; Waáaszewska 2011, 317), but one or some components of its encyclopaedic entry is highlighted and elevated to “… a logical (or content-constitutive) status” in the resulting ad-hoc concept (Carston 2002, 339). On the other hand, broadening takes place when the linguistically encoded denotation of a word is expanded and the word conveys “… a more general sense than the encoded one …” (Wilson and Carston 2007, 234). While narrowing preserves literalness, broadening does not, as one or more of the logical properties of the concept can be dropped (Hall 2011, 4; Waáaszewska 2011, 318). The resulting ad-hoc concept may go well beyond the boundaries of the lexically encoded concept. Depending on the extent to which the ad-hoc concept exceeds those boundaries, four varieties of broadening may be distinguished, which form a continuum: a)

(11) (12)

‘Approximation’, in which “… a word with a relatively strict sense is extended to a penumbra of items … that strictly speaking fall outside its linguistically specified denotation” (Sperber and Wilson 2008, 91). Examples of approximation are loose uses of round numbers (11) or geometric figures and shapes (12): Peter earns €2000 a month. Spain is a bull skin.

b) ‘Category extension’, in which a word with a particular sense and denotation is extended “… to a range of items that clearly fall outside its linguistically specified denotation” (Sperber and Wilson 2008, 91). Through this type of broadening, one or some defining or characteristic properties of a concept are applied to a set of items whose features somehow resemble that/those of the concept in question. Typical examples are the use of brand names or proper names for items of different brands:

On the Origin and Meaning of Secondary Interjections

(13) (14)

Have you got a Kleenex? He Xeroxed all the documents for me. c)

(15)

309

‘Hyperbole’, “… a more substantial broadening of the lexically encoded concept than approximation” (Waáaszewska 2011, 318) because the resulting concept can denote items of a diverse class by virtue of their having some similar or adjacent properties (Vega Moreno 2007, 48). The mall was full of people. [There were crowds of people, but it was not literally full]

d) ‘Metaphor’, which involves a broadening on the basis of relatively peripheral or at least contingent properties (Vega Moreno 2007, 48; Waáaszewska 2011, 318): (16) John is a lion/chameleon/shark/bull/George Clooney. Secondary interjections originate in word-classes whose items have conceptual content. The reason why those items can enter the class of interjections and, therefore, be used with a different expressive potential might reside in their concepts undergoing one of these lexical pragmatic processes, namely, broadening. But would the items of each word-class undergo the same type of broadening? And more importantly, what would enact the broadening of the encoded concept? The following section addresses these issues. It seeks to find answers by comparing the transfer of lexical items into the class of interjections to some lexical phenomena characteristic of children’s language.

5. Secondary interjections: A relevance-theoretic account Secondary interjections originate in lexical items belonging to wordclasses like nouns (17), verbs (18), adjectives (19) or adverbs (20): (17) (18) (19) (20)

Hell! Heavens! God! ¡Hostia! ¡Coño! Fuck! Come on! ¡Joder! ¡Anda! Good! Great! ¡Bueno! ¡Estupendo! Well! ¡Bien!

Items initially belonging to those word-classes encode concepts, even if very general conceptual schemas or pro-concepts requiring subsequent

310

Chapter Fourteen

adjustments. For those items to be transferred to the class of interjections, the (pro-)concepts they initially encode would undergo a process of broadening, as a consequence of which they would convey an even more general, diffused or vaguer meaning and have their linguistic denotation (significantly) expanded. From encoding concepts denoting individuals, objects, actions, qualities, etc., nouns, verbs, adjectives and adverbs would progressively have their denotation so loosened that they would refer to emotions, feelings or attitudes, either because such individuals, objects, actions, qualities, etc. cause those emotions, feelings or attitudes, or because such emotions, feelings or attitudes can be projected towards them. The encyclopaedic entries of the concepts initially encoded by the items giving rise to secondary interjections could contain information about stereotypical properties, even if very marginal, of the individuals, objects, actions or events to which those concepts are applied (cf. Gehweiler 2008). As a result of previous personal experience, those entries could also store information concerning emotions, feelings or attitudes associated with those concepts (cf. Kecskes 2004, 2008), either because their denotatum causes or has caused them or because individuals have projected them towards their denotatum. Obviously, the range of emotions, feelings and attitudes associated with those initial concepts would vary across individuals. For instance, the encyclopaedic entry of SHIT might connect the concept with emotions or feelings like disgust, repulsion, etc.; that of GOOD with feelings like wellness, pleasure, delight, approval, etc., and that of FUCK with feelings of disappointment, anger, frustration, wrath, etc. The association of those concepts with such feelings may also enable the words onto which they map to be used metaphorically or hyperbolically (more on this below). When broadened, those initial concepts could shift towards more marginal, but somehow related, properties or emotions associated with them or to emotions, feelings or attitudes that somehow resemble their properties in some respects, thus progressively becoming fuzzier and more inclusive. Consequently, the items giving rise to secondary interjections would instead become pointers to conceptual spaces related to the realm of emotions, feelings or attitudes. In other words, the meaning of those items would shift from individuals, objects, actions, characteristics or manners of actions, to more abstract meanings more or less loosely linked to their properties or to feelings, emotions or attitudes which they can cause or can be projected to them. When used as secondary interjections, those lexical items would only be a clue to an emotional concept that the hearer would have to construct on the fly by taking into account factors such as

On the Origin and Meaning of Secondary Interjections

311

paralanguage. That concept could be in some cases a highly context-bound occasion-specific concept, whereas in others that concept could be more stable, inasmuch as it would point to the same type of emotions or feelings. Some interjections might encode very general, broad, overarching, fuzzy concepts like EMOTION, FEELING or ATTITUDE. Other interjections might encode less general concepts like HAPPINESS or SADNESS corresponding to major types of emotions (Damasio 1994; Goleman 1995), which would be sub-types of those general fuzzy concepts. Still other interjections might encode less broad concepts, like EUPHORIA, JOY, MELANCHOLY, DEPRESSION, DISAPPOINTMENT, etc., which would in turn be sub-types of the major types of emotions. These concepts would need subsequent adjustments on the basis of factors such as intensity or degree of the emotion experienced or expressed (Padilla Cruz 2009b, 2010). When the conceptual content of items giving rise to secondary interjections is broadened, it might shift towards those kinds of (pro)concepts. The concepts encoded by some of those items may perhaps be broadened in the direction of very general and broad concepts related to emotions, while those encoded by other items are broadened in the direction of more specific concepts related to particular emotions. In any case, hearers would have to flesh those concepts out on the basis of paralanguage in order to understand what speakers intend to express (Wilson and Wharton 2006; Wharton 2009). In some cases, the result of such fleshing out might be short-lived ad-hoc concepts created to capture unique types of emotions or very specific nuances of emotions. Those adhoc concepts may not stabilise in the conceptual repertoires of users of a language; they might be one-off, sporadic occurrences. In contrast, if the same kind of ad-hoc concepts are formed repeatedly and often enough, such occasion-specific concepts could subsequently stabilise and become proper concepts, liable to be shared by other communicators (Hall 2011; Solska 2012). Whether individuals will form ad-hoc concepts from the lexical items giving rise to secondary interjections depends on factors such as encountering the same emotions or their same nuances, and availability of already stable concepts denoting them for retrieval from memory (cf. Hall 2011, 5-7). The passing of items from some word-classes to that of interjections would involve an extension of their denotation. In the case of adjectives and adverbs, that extension might not be very dramatic, as the concepts these items encode could refer to emotions or feelings that would be somehow related or bear some resemblance with the qualities, properties or manners they standardly refer to. In contrast, in the case of nouns and

312

Chapter Fourteen

verbs that extension might be more dramatic. But what might trigger such a denotational extension? And under which circumstances might it happen? To some extent, the origin of secondary interjections could be compared to children’s overextensions of words, a phenomenon that has been accounted for in relevance-theoretic terms by Waáaszewska (2011).

5.1 Overextension and secondary interjections Children have been shown to extend or overextend the meaning of words to refer to other entities or actions on the basis of some perceptual similarities (Clark 1973, 1993; Anglin 1977; Thomson and Chapman 1977). Concerning secondary interjections, it would be hard to think of some perceptual similarity between the concepts initially encoded by the items giving rise to them and the concepts those items are subsequently used to communicate. However, the creation of secondary interjections would resemble children’s overextensions in that the latter are frequent in situations in which children need to use a word that they do not know or cannot retrieve at a certain moment (Huttenlocher 1974; Thomson and Chapman 1977; Fremgen and Fay 1980; Gottfried 1997). When this happens, children replace the unknown or irretrievable word with another. Some secondary interjections might also originate in this way. Even if the words giving rise to them do not bear any kind of perceptual similarity to the feeling or emotion speakers intend to express, in some cases speakers could find in their encyclopaedic entries some information or property related, even if remotely, to the feeling or emotion that they intend to express, and, therefore, use those words as interjections. In other cases, on the contrary, speakers, lacking precise terms, would resort to those words as if they were place-holders or slot-fillers, the slot to be filled being the particular emotion that they express. As Waáaszewska (2011, 320) explains, the use of a word to activate a conceptual representation not often associated with it sometimes stems from the speaker’s abilities. The Communicative Principle of Relevance establishes that any act of ostensive communication communicates a presumption of its own optimal relevance. This presumption implies that the processing of a particular ostensive stimulus will be worth the hearer’s effort and that the speaker will select, out of possible candidate stimuli, the one that she thinks will make manifest her informative intention in the easiest and most straightforward way depending on her cognitive abilities and stylistic preferences (Sperber and Wilson 1986, 1995; Wilson and Sperber 2002, 2004). When children overextend some words, they do so

On the Origin and Meaning of Secondary Interjections

313

because of their limited expressive abilities, although they aim for optimal relevance. The usage of nouns, verbs, adjectives and adverbs to express emotions, feelings and attitudes might somehow detract from optimal relevance, if there were other words that could convey the speaker’s informative intention more efficiently and economically (Wilson and Sperber 2004, 612). However, speakers might resort to those items because at some specific moments they cannot retrieve or think of other more precise words to allude to their emotions or they find in their encyclopaedic entries some property or assumption that somehow connects with the emotion, feeling or attitude that they want to express. This would enable speakers to use those items as vehicles to express feelings, emotions or attitudes, above all when emotions, feelings and attitudes are ineffable, vague or difficult to pin down in words. Accordingly, a broadening similar to children’s overextension could be thought to take place when adjectives and adverbs are used as secondary interjections. Think of adjectives like ‘good’, ‘cute’, ‘cool’, ‘awful’ or ‘awesome’, which have interjectional counterparts. Their encyclopaedic entry could contain assumptions about positivity, pleasure, delight, etc., which could be close or related to the feeling, emotion or attitude the speaker wishes to express when using them as interjections. Upon finding no better words to express that feeling, the speaker would resort to such words and expect the hearer to manipulate the conceptual content that they convey. On the other hand, Waáaszewska (2011) also explains that other overextensions can be viewed as the result of children’s preferences. On many occasions, children use wrong words to refer to some concepts – they mislabel objects – even though they know the words conventionally used to name them. Such overextensions reflect early “… metaphorical abilities in young children and indicate children’s linguistic flexibility” (Waáaszewska 2011, 320). To do so, they rely on some similarity, shared property or common encyclopaedic information. This might be what happens when some nouns and verbs are transferred to the class of secondary interjections: they could be transferred as a consequence of metaphoric usage. Many nouns can be metaphorically or hyperbolically used in attributive constructions to describe or assign characteristics on the basis of some contingent or emergent properties stored in their encyclopaedic entry. Likewise, many verbs can also be used metaphorically or hyperbolically to refer to another action which they somehow resemble or evoke. Consider the following examples:

314

Chapter Fourteen

(21)

a. This is hell / shit / heaven. b. Spanish: Esto es un infierno / una mierda / el paraíso. a. John fucked everything up. b. Spanish: Juan (lo) ha jodido todo. a. They fucked me at the exam. b. Spanish: Me han follado en el examen. Esto es la hostia. / ¡Eres la hostia, tío!

(22) (23) (24)

In sentences like (21) or (24) the speaker would not be literally predicating that something is ‘hell/shit/heaven’ or that somebody is ‘the holy host’; rather, some stereotypical features or assumptions connected those nouns – e.g. negativity, pain, suffering, etc. in the case of ‘hell’; positivity, pleasure, delight, joy, etc. in the case of ‘heavens’; superiority, supremacy, etc. in the case of ‘host’ – are attributed to other entities. Those features or assumptions could also be linked to emotions, feelings or attitudes. Along the same lines, a verb like ‘fuck’ (22a)–(23a) or the Spanish ‘joder’ (22b) and ‘follar’ (23b) do not mean what they standardly denote, but would make manifest assumptions related to ruining, destruction, failure, desolation, etc., which can also be associated to some negative emotions or feelings. In these metaphorical or hyperbolical constructions, those nouns and verbs get their conceptual content broadened to denote some property, state or action that shares some features with, or very roughly or ideally resembles, their initial literal referents. Strictly speaking, such property, state or action cannot be literally referred to by means of those lexical items. Since nouns and verbs can have those metaphorical or hyperbolical usages and get their encoded concepts broadened, such usages might also enable those items to be used as secondary interjections, in which case their conceptual content would probably be broadened towards more peripheral properties. Accordingly, metaphorical or hyperbolical usages might make the transfer of nouns and verbs to the class of secondary interjections possible on the basis of common core or peripheral properties. Probably, the broadening of the concepts encoded by nouns and verbs is more radical than that undergone by the concepts encoded by adjectives and adverbs. Such broadening could be represented by means of the following figure, in which each concentric circle represents a slightly different property associated to the initial denotation of a concept and types of lines their centrality or marginality:

On the Origin and Meaning of Secondary Interjections

315

SHIT FUCK

Figure 1. Conceptual broadening of lexical items giving rise to secondary interjections Regardless of whether secondary interjections arise as a consequence of speakers’ expressive limitations or preferences, their transfer from other word-classes would be due to the same lexical pragmatic process: a broadening of their initial lexically encoded concept that yields a fuzzier concept related to emotions, feelings or attitudes. For hearers to interpret secondary interjections, they would have to adjust that concept and create an ad-hoc one that captures the expressive content that they think their producers intend to express. However, when becoming secondary interjections, nouns and verbs, on the one hand, and adjectives and adverbs, on the other hand, seem to show different degrees of conceptual broadening and that broadening seems to operate slightly differently. This might reflect that the items belonging to each group undergo different types of overextensions.

5.2 Secondary interjections and types of overextension Within children’s overextensions researchers have also distinguished two sub-types (Waáaszewska 2011, 321-322): a)

‘Over-inclusion’, also referred to as ‘categorical overextension’ or ‘classic overextension’ (Clark 1973, 1993; Rescorla 1980), in which the overextended word “… is applied to instances of other categories within the same or adjacent conceptual domain” (Waáaszewska 2011, 321). This process is evident when children use terms to refer to other entities which are co-hyponyms of the same superordinate term. This would be the case of ‘doggie’, a

316

Chapter Fourteen

hyponym of ‘animal’, when used to allude to dogs, cats, horses, etc. not only because of perceptual similarity but, more importantly, because of conceptual contiguity – all are animals. b) ‘Analogical extension’, also alluded to as ‘analogical overextension’ (Rescorla 1980; Clark 1993), as a result of which children overextend terms on the basis of perceptual similarity to other terms belonging to different, maybe unrelated, conceptual domains. This is what happens when they use ‘cookie’ to refer to ‘moon’ upon perceiving their roundedness. Adjectives and adverbs could enter the category of secondary interjections as a consequence of over-inclusion. Their lexically encoded concept would contain encyclopaedic information about properties, characteristics or features that resemble those of, or can also be applied to, certain feelings, emotions or attitudes. As in other cases of over-inclusion, the information contained in the logical entries of those adjectives and adverbs giving rise to secondary interjections would provide a link between the initial concept encoded by those lexical items and the ‘new’ non-lexicalised concept that they would subsequently communicate when used as interjections. Thus, the adjective ‘good’ can over-include some feelings, emotions or attitudes on the basis of their belonging to conceptual domains like ‘positivity’, ‘pleasure’, ‘delight’, ‘satisfaction’, etc. Although this adjective normally expresses that the entity it modifies has a particular feature and can thus be described, it may be used as an interjection on the grounds of some property or information which can be extended to a certain feeling, emotion or attitude. In this sense, the adjective and the resulting secondary interjection would behave as hyponyms of broad concepts like POSITIVITY, PLEASURE, DELIGHT, SATISFACTION, etc., and speakers could use that adjective as a secondary interjection because they find in its denotation some feature that can be applied to emotions. In contrast, nouns and verbs would enter the sub-class of secondary interjections as a result of analogical extension. Although the denotation of the concepts initially encoded by those lexical items and the feeling, emotion or attitude that they can subsequently be used to express may certainly not intersect at all, the encyclopaedic entries of those original concepts may perhaps contain or make manifest assumptions about some contingent properties or qualities, even if very marginal or peripheral, which can be applied or related to those feelings, emotions or attitudes. As shown above, nouns and verbs can be used metaphorically or hyperbolically to refer not to the entity or action that they standardly denote, but to some related entity or action which they can allude to on the

On the Origin and Meaning of Secondary Interjections

317

grounds of contingent properties associated with them. However, when nouns and verbs are transferred to the class of secondary interjections, the information in their logical entries – e.g. that such words are normally used to refer to objects, individuals or actions – is ignored, since now those words are used to refer to feelings, emotions or attitudes. Thus, concepts like FUCK, SHIT or HELL may be broadened to the ad-hoc concepts FUCK*, SHIT* or HELL*, which denote some emotion, feeling or attitude that is the consequence of, is related to or is projected to what those concepts initially denoted. Under this account, the lexical items giving rise to secondary interjections have their conceptual content broadened and that content is associated with a new emotional or attitudinal one. When their broadening is accomplished, those items could be associated with proper concepts denoting more or less specific types of emotions or feelings that individuals have already experienced, projected or encountered beforehand. Therefore, the resulting interjections could end up having rather stable usages as expressive vehicles for specific feelings or emotions and their interpretation might not require the construction of ad-hoc concepts any longer, as individuals may already have stored and associated mental entities with those interjections in a stable manner. However, one of the problems that make interjections unamenable to conceptual analyses is precisely their usage with differing, sometimes diametrically opposed, meanings. This phenomenon certainly occurs with secondary interjections and, in effect, new secondary interjections are often used to express a wide variety of emotions, feelings and attitudes. If new secondary interjections were associated to specific (sub-)types of emotions, a new problem would arise: why could some of them be used to express differing or opposed emotions or feelings?

5.3 On the various meanings of secondary interjections When nouns and verbs are used as secondary interjections, speakers invite hearers to broaden their encoded concepts and to construct ad-hoc concepts whose denotations might significantly differ from those of their initially encoded concepts because the encyclopaedic entries of those concepts store some information related to emotions or feelings. In the case of adjectives and adverbs, the concepts that these items initially encoded are extended to include something different on the basis of some conceptual proximity or hyponym relation. When some items from these classes become secondary interjections, they could be initially associated with a general (pro-)concept that denotes some (sub-)type of emotion. If

318

Chapter Fourteen

so, those items would start life as secondary interjections specialised for expressing specific (sub-)type(s) of emotions or feelings. However, many new secondary interjections may be used to express plenty of more or less differing emotions or feelings. The answer to why this may be possible could lie in some usages of secondary interjections. Consider now the following Spanish sentences: (25) (26) (27) (28)

Me has hecho un favor muy bonito. Literally: ‘You did me a very beautiful/great favour’ ¡Bonito favor me has hecho! Literally: ‘Beautiful/Great favour you did me!’ Pedro es menudo. Literally: ‘Peter is tiny’. ¡Menudo es Pedro! Literally: ‘Tiny Peter is!’

(25) and (27) can have a literal reading, with ‘bonito’ and ‘menudo’ having their conventional meaning – i.e. ‘beautiful’ and ‘tiny’, respectively. However, (26) and (28) would be ironical (Yus Ramos, personal communication). The speaker would not be asserting that someone has done a great favour to her (26) or that Peter is tiny (28). Rather, in (26) the speaker would be dissociating herself from the belief that the favour that someone did her was positive, while in (28) she would be communicating something negative about Peter (Wilson and Sperber 2002, 2004). In these two sentences, ‘bonito’ and ‘menudo’ must also be broadened, but the resulting ad-hoc concepts do not have positive connotations. Intonation and the placement of those lexical items at the beginning of those sentences may trigger a more radical broadening, as a result of which their meaning is reversed. Although some items may enter the grammatical sub-category of secondary interjections and stabilise with a (pro-)concept resulting from another concept, their interjectional use to express distinct, maybe radically opposed emotions or feelings might be the result of ironical uses. As a consequence, the broad (pro-)concepts that those interjections would encode would have to be further broadened, maybe to include denotations that were not initially connected with them. Intonation and paralanguage could prompt a further or extreme broadening of an already broadened concept. Solska (2012) uses the term ‘meaning reversal’ to refer to the process by means of which a highly salient meaning of polysemous or homonymous words, or a deeply entrenched figurative meaning of other phrasal expressions, must be rejected and abandoned in favour of another

On the Origin and Meaning of Secondary Interjections

319

meaning or a more literal reading that is backgrounded or part of an adhoc concept10. However, those secondary interjections that stabilise in the linguistic system with an expressive value would not initially be ambiguous in terms of meaning. Their meaning reversal would not involve assessing two (or more) possible meanings and opting for one of them which receives less activation, but broadening and changing their meaning into a different one. If this were right, secondary interjections could express various emotions as a result of a two-stage broadening: the concept encoded by the lexical item giving rise to a secondary interjection is firstly broadened to some very fuzzy (pro-)concept and that already broadened (pro-)concept is subsequently broadened as a result of ironical usages. This process could be labelled ‘extreme’ or ‘radical’ broadening. Since this process results in an association of lexical items with new and distinct concepts related to emotions, feelings or attitudes, it could also be described as a ‘deconceptualisation’ of lexical items because they are deprived of their initial denotation or conceptual load, and a subsequent process of ‘reconceptualisation’, for those lexical items gain new denotations. The following figure could illustrate this process:

10

Meaning reversals differ from puns in that the interpretation of puns achieves an ‘oscillating effect’, as two interpretations can be simultaneously considered and the hearer may hesitate between them. However, in meaning reversals one interpretation is supplanted by another after its unviability is realised (Solska 2012).

Chapter Fourteen

320

ADJECTIVES (Pro-)concepta

ADVERBS (Pro-)conceptb

NOUNS (Pro-)conceptc

OVERINCLUSION

IRONICAL USAGE

(Pro-)concepta

VERBS (Pro-)Conceptd

ANALOGICAL EXTENSION

SECONDARY INTERJECTIONS (Pro-)conceptb (Pro-)conceptc (Pro-)conceptd

EXTREME BROADENING

(Pro-)concepta’

SECONDARY INTERJECTIONS’ (Pro-)conceptb’ (Pro-)conceptc’ (Pro-)conceptd’

Figure 2. Extreme or radical broadening of lexical items when becoming secondary interjections

Conclusion Concepts are not stable mental objects, as they need adjustments during comprehension. Sperber and Wilson (1997) suggested that the lexical pragmatic processes of broadening and narrowing could account for semantic change and evolution. This paper has argued that broadening could explain the transfer of nouns, verbs, adverbs and adjectives to the class of secondary interjections. If this were right, then at least broadening may be responsible for some word-class shifts. As a result of it, lexical items that had specific functions and were used in particular ways may be used in quite different ways, with the function of expressing emotions, feelings or attitudes. This paper has also proposed that some lexical items may be used as secondary interjections because their conceptual content undergoes a broadening similar to the overextensions made by children, either because

On the Origin and Meaning of Secondary Interjections

321

speakers lack words to refer to complex, evasive and ineffable things like emotions, feelings and attitudes, or because speakers use those words metaphorically or hyperbolically owing to the properties or assumptions that they would find in the encyclopaedic entries of the concepts that those items initially encoded. Additionally, this paper has differentiated two ways in which various types of lexical items may give rise to secondary interjections: adjectives and adverbs could enter the class of secondary interjections because their conceptual content would undergo overinclusion, whereas nouns and verbs would shift to that class as a consequence of analogical extension. Regarding the distinct expressive potentials that secondary interjections have, this paper has put forward that these interjections could express different, maybe opposed, emotions, feelings or attitudes as a consequence of a further broadening of their conceptual load triggered by ironical usages. Nevertheless, this paper has addressed the sub-category of secondary interjections as a whole and made some hypotheses about their origins in general terms. A more fine-grained and detailed account should consider individual interjections and look into their original meanings in order to unveil how those meanings are actually broadened. Thus, it would be possible to ascertain which kind of broadening they undergo, i.e. whether over-inclusion or categorical extension, if they stabilise in the linguistic system with a specific expressive potential for some time and when they show traces of such expressive potential evolving.

References Aijmer, Karin. 2004. Interjections in a contrastive perspective. In Emotion in dialogic interaction: Advances in the complex, edited by W. Edda. Philadelphia: John Benjamins. Ameka, Felix. 1992. Interjections: The universal yet neglected part of speech. Journal of Pragmatics 18: 101-118. —. 2006. Interjections. In Encyclopaedia of language and linguistics, edited by K. Brown. Amsterdam: Elsevier. Anglin, Jeremy M. 1977. Word, object and conceptual development. New York: Norton. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward Finegan. 1999. Longman grammar of spoken and written English. London: Longman. Blakemore, Diane. 1992. Understanding utterances. Oxford: Blackwell.

322

Chapter Fourteen

—. 2010. Communication and the representation of thought: The use of audience-directed expressions in free indirect representations. Journal of Linguistics 46, no. 3: 575-599. —. 2011. On the descriptive ineffability of expressive meaning. Journal of Pragmatics 43, no. 14: 3537-3550. Brinton, Laurel. 1996. Pragmatic markers in English. Grammaticalization and discourse functions. Berlin: Mouton de Gruyter. Buridant, Claude. 2006. L’interjection: Jeux et enjeux. Langages 161: 3-9. Carston, Robyn. 1997. Enrichment and loosening: Complementary processes in deriving the proposition expressed. Linguistiche Berichte 8: 103-127. —. 2002. Thoughts and utterances. The pragmatics of explicit communication. Oxford: Blackwell. Carston, Robyn, and George Powell. 2005. Relevance Theory: New directions and developments. UCL Working Papers in Linguistics 17: 279-299. Cestero Mancera, Ana M., and Francisco Moreno Fernández. 2008. Usos y funciones de vale y ¡venga! en el habla de Madrid. Boletín de Lingüística 20, no. 29: 65-84. Clark, Eve V. 1973. What’s in a word? On the child’s acquisition of semantics in his first language. In Cognitive development and the acquisition of meaning, edited by T. Moore. New York: Academic Press. —. 1993. The lexicon in acquisition. Cambridge: Cambridge University Press. Damasio, Antonio. 1994. Descartes’ error: Emotion, reason and the human brain. New York: Avon. Eastman, Carol M. 1992. Swahili interjections: Blurring languageuse/gesture-use boundaries. Journal of Pragmatics 18: 273-287. Fodor, Jerry. 1998. Concepts: Where cognitive science went wrong. Oxford: Clarendon Press. Fremgen, Amy and David Fay. 1980. Overextensions in production and comprehension: A methodological clarification. Journal of Child Language 7: 205-211. Gehweiler, Elke. 2008. From proper name to primary interjection. The case of gee! Journal of Historical Pragmatics 9, no. 1: 71-93. Goffman, Erving. 1981. Forms of talk. Oxford: Blackwell. Goleman, Daniel. 1995. Emotional intelligence. New York: Bantam Books.

On the Origin and Meaning of Secondary Interjections

323

Gottfried, Gail M. 1997. Using metaphors as modifiers: Children’s production of metaphoric compounds. Journal of Child Language 24: 567-601. Greenbaum, Sidney. 2000. The Oxford reference grammar. Oxford: Oxford University Press. Hegde, M. N., and David E. Hartman. 1979. Factors affecting judgements of fluency: I. Interjections. Journal of Fluency Disorders 4: 1-11. Hall, Alison. 2011. Ad hoc concepts: Atomic or decompositional. UCL Working Papers in Linguistics 23: 1-10. Hill, Deborah. 1992. Imprecatory interjectional expression: Examples from Australian English. Journal of Pragmatics 18: 209-233. Hopper, Paul J. 1991. On some principles of grammaticalisation. In Approaches to grammaticalisation, edited by E. C. Traugott and B. Heine. Amsterdam: John Benjamins. Hopper, Paul J., and Elizabeth C. Traugott. 1993. Grammaticalisation. Cambridge: Cambridge University Press. Huddleston, Rodney. 1988. English grammar: An outline. Cambridge: Cambridge University Press. Huttenlocher, Janellen. 1974. The origins of language comprehension. In Theories in cognitive psychology, edited by R. Solso. Potomac: Erlbaum. Kecskes, Istvan. 2004. Editorial: Lexical merging, conceptual blending, and cultural crossing. Intercultural Pragmatics 1, no.1: 1-26. —. 2008. Dueling contexts: A dynamic model of meaning. Journal of Pragmatics 40: 385-406. Kochelman, Paul. 2003. The meanings of interjections in Q’eqchi’ Maya: From emotive reaction to social and discursive action. Current Anthropology 44, no. 4: 467-490. Ljung, Magnus. 2009. A study of expletive interjections in spoken British English. Gothenburg Studies in English 96: 199-211. López Bobo, María J. 2002. La interjección. Aspectos gramaticales. Madrid: Arco Libros. Mayol, Laia. 2008a. Catalan “Déu n’hi do” and conventional implicatures in exclamatives. Catalan Journal of Linguistics 7: 135-156. —. 2008b. Catalán “Déu n’hi do” and levels of meaning in exclamatives. Proceedings of the West Coast conference on Formal Linguistics 26: 375-383. Meng, Katharina, and Susanne Schrabback. 1999. Interjections in adultchild discourse: The cases of German HM and NA. Journal of Pragmatics 31: 1263-1287.

324

Chapter Fourteen

Montes, Rosa G. 1999. The development of discourse markers in Spanish: Interjections. Journal of Pragmatics 31: 1289-1319. Nicoloff, Frank. 1990. How so/such/what/how exclamatories mean. Grazer Linguistische Studien 33-34: 207-225. Norrick, Neal R. 2009. Interjections as pragmatic markers. Journal of Pragmatics 41: 866-891. O’Connell, Daniel C., and Sabine Kowal. 2005. Where do interjections come from? A psycholinguistic analysis of Shaw’s Pygmalion. Journal of Psycholinguistic Research 34, no. 5: 497-514. O’Connell, Daniel C., Sabine Kowal, and Carie Ageneau. 2005. Interjections in interviews. Journal of Psycholinguistic Research 34, no. 2: 153-171. Oppermann-Marsaux, Evelyne. 2008. De l’injonction a l’interjection. La naissance des marqueurs discursifs tien(s) et tenez en français medieval (XIIe-XVe siecles). L’Information Grammaticale 118: 11-15. Padilla Cruz, Manuel. 2009a. Towards and alternative relevance-theoretic approach to interjections. International Review of Pragmatics 1, no. 1: 182-206. —. 2009b. Might interjections encode concepts? More questions than answers. àodz Papers in Pragmatics 5, no. 2: 241-270. —. 2010. What do interjections contribute to communication and how are they interpreted? A cognitive pragmatic approach. In Pragmatic perspectives on language and linguistics. Volume I: Speech actions in theory and applied studies, edited by I. Witczak-Plisiecka. New Castle upon Tyne: Cambridge Scholars Publishing. Quirk, Randolph, and Sidney Greenbaum. 1973. A university grammar of English. Harlow: Longman. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1972. A grammar of contemporary English. Harlow: Longman. —. 1985. A comprehensive grammar of the English language. London: Longman. Rescorla, Leslie. 1980. Overextension in early language development. Journal of Child Language 7: 321-335. Shenhav, Shaul R. 2008. Showing and telling in parliamentary discourse: The case of repeated interjections to Rabin’s speeches in the Israeli parliament. Discourse and Society 19: 223-255. Solska, Agnieszka. 2012. Enforced reversals of meaning: A lexical pragmatic perspective. Paper presented at the 2nd Meaning, Context and Cognition Conference, Lodz, 22-24 March. Sperber, Dan, and Deirdre Wilson. 1986. Relevance. Communication and cognition. Oxford: Blackwell.

On the Origin and Meaning of Secondary Interjections

325

—. 1995. Relevance. Communication and cognition. 2nd ed. Oxford: Blackwell. —. 1997. The mapping between the mental and the public lexicon. UCL Working Papers in Linguistics 9: 107-125. —. 2008. A deflationary account of metaphors. In The Cambridge handbook of metaphor and thought, edited by R. Gibbs. Cambridge: Cambridge University Press. ĝwiatkowska, Marcela. 2006. L’interjection: Entre deixis et anaphore. Langages 161: 47-56. Thomson, Jean, and Robin Chapman. 1977. Who is ‘daddy’? revisited: The status of two-year-olds’ overextended words in use and comprehension. Journal of Child Language 4: 359-375. Traugott, Elizabeth C. 1989. On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65: 3155. —. 1992. From propositional to textual and expressive meanings: Some semantic-pragmatic aspects of grammaticalisation. In Perspectives in historical linguistics, edited by W. P. Lehmann and Y. Malkiel. Amsterdam: John Benjamins. —. 1995. Subjectification in grammaticalisation. In Subjectivitiy and subjectivisation, edited by D. Stein and S. Wright. Cambridge: Cambridge University Press. Traugott, Elizabeth C., and Richard B. Dasher. 2002. Regularity in semantic change. Cambridge: Cambridge University Press. Tsibulsky, Danielle. 2008. Estudio sobre las interjecciones bueno, hombre y huy en español. Gaceta Hispánica de Madrid 6: 1-17. Van der Hoek, Michel. 2007. The origin of Gothic hiri. NOWELE 52: 921. Vega Moreno, Rosa E. 2007. Creativity and convention. The pragmatics of everyday figurative speech. Amsterdam: John Benjamins. Waáaszewska, Ewa. 2004. What to do with response cries in Relevance Theory? In Relevance studies in Poland. Volume I, edited by E. Mioduszewska. Warsaw: University of Warsaw. —. 2011. Broadening and narrowing in lexical development: How Relevance Theory can account for children’s overextensions and underextensions. Journal of Pragmatics 43: 314-326. Wharton, Tim. 2003. Interjection, language, and the ‘showing/saying’ continuum. Pragmatics and Cognition 11: 39-91. —. 2009. Pragmatics and non-verbal communication. Cambridge: Cambridge University Press.

326

Chapter Fourteen

Wierzbicka, Anna. 1991. Cross-cultural pragmatics: The semantics of human interaction. Berlin: Mouton de Gruyter. —. 1992. The semantics of interjection. Journal of Pragmatics 18: 159192. Wilkins, David P. 1992. Interjections as deictics. Journal of Pragmatics 18: 119-158. Wilson, Deirdre. 1997. Linguistic structure and inferential communication. Paper presented at the 16th International Congress of Linguistics, Paris, 20-25 July. —. 2004. Relevance and lexical pragmatics. UCL Working Papers in Linguistics 16: 343-360. Wilson, Deirdre, and Robyn Carston. 2006. Metaphor, relevance and the ‘emergent property’ issue. Mind and Language 21: 404-433. —. 2007. A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In Pragmatics, edited by N. Burton-Roberts. Basingstoke: Palgrave. Wilson, Deirdre, and Dan Sperber. 1993. Linguistic form and relevance. Lingua 90: 1-25. —. 2002. Relevance Theory. UCL Working Papers in Linguistics 14: 249287. —. 2004. Relevance Theory. In The handbook of pragmatics, edited by L. Horn and G. Ward. Oxford: Blackwell. Wilson, Deirdre, and Tim Wharton. 2006. Relevance and prosody. Journal of Pragmatics 38: 1559-1579.

CONTRIBUTORS

Signe Rix Berthelin Magdalena Biegajáo Thornstein Fretheim Jadwiga Linde-Usiekniewicz Manuel Padilla Cruz Agnieszka Piskorska Maria Angeles Ruiz Moneva Jolanta Sak-Wernicka Martin Schröder Agnieszka Solska Elwira Szehidewicz Christoph Unger Ewa Waáaszewska Francisco Yus Edyta ħraáka

Norwegian University of Science and Technology, Trondheim, Norway University of Warsaw, Poland Norwegian University of Science and Technology, Trondheim, Norway University of Warsaw, Poland University of Seville, Spain University of Warsaw, Poland University of Zaragoza, Spain John Paul II Catholic University of Lublin, Poland SIL International and VU University Amsterdam, The Netherlands University of Silesia, Poland University of Warsaw, Poland SIL International and Johannes Gutenberg University Mainz, Germany University of Warsaw, Poland University of Alicante, Spain University of Silesia, Katowice, Poland

INDEX

ad hoc concept, 6, 9, 124-127, 159, 168, 200-201, 215, 253, 290-292 adaptation, 134, 142, 147, 264 affix(ation), 10, 252, 268 Ageneau, C., 303 Aijmer, K., 301, 303 Aikhenvald, A., 33, 45, 48, 54, 61 Alba-Juez, L., 121 allegory, 7, 12-13, 152-163, 165168, 170-175 Allott, N., 9 Alton, T., 45 ambiguity, 7-8, 31, 198-200, 205206, 208, 220, 223-225, 230, 237-238, 240, 289 ambiguity resolution, 224, 228, 236237, 239 Ameka, F., 301-302 Amfo, N. A. A., 66 analogical extension, 12, 316, 321 Anglin, J. M., 312 Animal Farm, 8, 158, 175, 184-185, 188-189, 192, 194 Antaki, C., 123 antecedent, 26, 64-65, 67, 71, 73, 75, 77-78, 81 assumption, 4, 8, 24, 34-39, 52, 5759, 61, 66-70, 75, 78, 81, 107, 118, 125, 153, 159, 161, 165, 181, 215, 226-227, 230-231, 313 Attardo, S., 178, 205, 207, 224 attributive use, 46, 54, 59, 60-61, 181 Barnes, J., 48 Bassnett, S., 135 Beck, J., 118, 120 Bennett-Levy, J., 118 Bercelli, F., 123 Berthelin, S. R., 51

Bezuidenhout, A., 17, 20 Biber, D., 303 Blagden, S., 107 Blakemore, D., 2, 17, 20, 26, 66, 117, 120-121, 176, 290, 292293, 300, 303 Blass, R., 59, 137-138, 140-141 blend, 263, 273 blind people’s comprehension, 104 Booij, G., 278 Bordería, S., 292 Boys-Stones, G. R., 154 Braun, M., 278 Breheny, R., 31 bridging implicature, 4, 64, 67, 7578 Brinton, L., 304 broadening, 9, 11-12, 119, 198, 201, 290-292, 308-310, 313-315, 317321 Brugman, C., 23 Buridant, C., 301, 303-304 Butler, H. E., 152-154 Bybee, J. L., 278 Carr, N., 5, 95 Carston, R., 6, 9, 59, 69, 117, 119, 124, 126-128, 154, 159, 168172, 199-201, 209, 212, 220, 246, 291-292, 306-308 Catford, J. C., 135 cautious optimism, 106 CBTT (cognitive-behavioural therapy theory), 120-121, 127 censorship, 133, 138-139, 148 Cestero Mancera, A. M., 304 Chabanne, J.-C., 224 Channell, J., 208 Chapman, R., 312 Chiaro, D., 237

330 Chun, C.-H., 250 Clark, D., 118 Clark, E. V., 312, 315-316 Clark, H., 64, 177 classification of jokes, 14, 228, 243 cognitive environment, 3-4, 6, 1617, 19, 34-40, 121, 124, 182, 186, 190, 205-206, 226-227 Cognitive Principle of Relevance, 4 cognitive-behavioral therapy, 118 collocation, 10, 262, 267, 274 compounding, 10, 249, 256-257, 261, 263, 267-268, 273, 277, 281, 295-296 comprehension module, 3, 166 comprehension strategies, 106 concept adjustment, 11, 169, 277, 290, 295, 297 conceptual adjustment, 228-230, 236, 239, 306 conditions on punniness, 204 connective, 63, 65-67, 69-71, 73, 77, 80, 82-83, 292 connector, 204-209, 212, 215-216, 218-220 context accessibility, 90, 178 contextual sources, 177-180, 183, 185-186, 190-192, 195 contradiction, 31, 184, 188, 190, 192, 230, 291 contrariety, 64 contrastive reduplication, 10, 277, 280-284, 286-287, 289-290, 293295 Cram, D., 26 Crisp, P., 7, 152-153, 156-158, 160, 171-172 criterion of optimal accessibility to irony, 179 cultural frame, 233-234, 239 Curcó, C., 9, 177, 223-224, 226227, 230-231, 237-241 cyberpragmatics, 4, 90-91, 96 Damasio, A., 311 Daneš, F., 35 Danielewiczowa, M., 29

Index Dansieh, S. A., 293 Dasher, R. B., 304-305 Daviþo, O., 169 Demonte, V., 29 derivation, 87, 98, 207, 210-211, 215-217, 219, 221, 229, 236, 249, 254, 260, 267, 269, 278, 294-295 descriptive phrase, 10, 259-262 direct translation, 8, 183 disambiguation, 198, 200-201, 224, 286 discourse, 2-4, 8, 12, 14, 16-17, 20, 23, 33, 35, 39, 41, 54, 62, 64-65, 67, 78, 81-83, 100, 117, 120121, 123-125, 128-131, 135, 137-138, 140, 149, 159-161, 174, 235, 242-243, 284, 288, 292, 296-297, 300, 304, 322-324 discourse analysis, 12, 117, 121, 123-125, 128-130, 242 disjunctive elements, 205-206, 208, 210, 212, 217, 219 Dorais, L.-J., 44-45 Dukate, A., 134-136, 138-140 Dumarsais, C., 170 Dynel, M., 217, 223, 227, 229, 233 Eastman, C. M., 303 echoic use, 4, 59, 177 Elias, U., 139 Encoding Grammar, 3, 17-28, 30, 33 encyclopedic entry, 127, 268 epistemic modal, 48-49 epistemic status, 45-49, 51-52, 55, 58, 60 epizeuxis, 11, 277, 294-295 equivocating utterance, 200, 208, 211 Eriksson, K., 104 Erteschik-Shir,  ., 16-17, 27, 38 Escandell-Vidal, V., 66, 292 Everett, J., 107 evidentiality, 3, 45, 48, 57, 62-63 explicature, 24, 30, 65, 67-69, 7475, 80-81, 119, 125, 141, 182-

Applications of Relevance Theory: From Discourse to Morphemes 183, 186-187, 192, 228, 233, 263, 303-304 external relevance, 159 Fairburn, C., 118 Fanshel, D., 121-122 Fauconnier, G., 157 Fay, D., 312 Fennig, C. D., 44 Ferrara, K. W., 122-123 fiction, 153, 159 Fletcher, A., 154 focus particle, 32 Fodor, J., 305 Fontanier, M., 170 Fortescue, M., 17, 45-47 Forza, F., 283 Franklin, R. W., 169 Fremgen, A., 312 Fretheim, T., 4, 29, 31, 66 Frost, R., 160 Galasinski, D., 138, 140 Gehweiler, E., 301-305, 310 gender, 10, 20-21, 23, 25-26, 48, 249, 251-254, 265, 267-268 Gerrig, R., 177 Ghomeshi, J., 280-282, 284-286, 289-290 Gibbs, R., 7, 152, 156, 158, 160161, 171 Goffman, 302 Goleman, D., 311 Gottfried, G. M., 312 grammaticalisation, 11, 304-305, 323, 325 Greenbaum, S., 301 Grice, H. P., 1-2, 180-181 Groefsema, M., 223 Gruenler, C., 156 Guillén Galve, I., 183 guru effect, 5, 13, 103-108, 111, 113-115 Gutt, E.-A., 8, 138, 182-184 Hall, A., 307-308, 311 Harris, R. A., 156 Hartman, D. E., 303 hearsay evidential, 44, 47, 61

331

hedge, 11, 277, 290-292, 295 Hedley, P., 26 Hegde, M.  ., 303 Heller, Z., 169 Hengeveld, K., 19 Hernández Iglesias, M., 124 Hill, D., 302 Hockett, C., 223, 225 Hohenhaus, P., 281-286 Hopper, P. J., 304 Horn, L., 281, 284 House, J., 31 Huang, Y., 285, 287, 294 Huddleston, R., 301 humorous effect, 207, 218, 225-226, 229-230 Huttenlocher, J., 312 hyperbole, 119, 129, 172, 182, 246 ideophone, 261, 267-268, 271, 280 implicature, 1, 6, 42, 68, 70-71, 7678, 81, 125-127, 141, 165, 182184, 192, 225, 228-229, 232233, 238, 241, 294-295, 297, 323 incongruity, 236-239, 241 indirect translation, 183 information focus, 51 information structure, 3, 13, 16-18, 30-32, 38, 42 Inkelas, S., 278 instead, 64-67, 69-75, 77-78, 80-81 intention, 1, 6, 16, 25, 37, 45, 50, 70, 82, 87, 91-92, 96, 110, 119, 133-134, 137-138, 141, 148-149, 178, 180, 188, 200, 269, 299, 312-313 internal relevance, 170 Internet, 4, 12, 14, 86-87, 90-91, 9498, 100-102, 263 Internet-mediated communication, 14, 91, 98, 102 interpretive resemblance, 97, 164, 176, 184, 192 Intersecting Circles Model, 14, 228, 233, 239-240, 243 Inuit dialect, 44 Iñupiaq, 3, 44-51, 53-55, 59-63

332 ironical usage, 319, 321 Ishikawa, K., 32 Itani, R., 290-291 Iten, C., 69 Jezierski, G., 104 Jodáowiec, M., 9, 223-227, 237-238, 240 joke, 9, 13-14, 221, 223-226, 228243 Jorgensen, J., 176 juxtaposition of concepts, 206 Kaplan, L. D., 44 Kasten, M., 156 Katz, J., 20 Kecskes, I., 306, 310 Kizeweter, M., 133 Kochelman, P., 304 KosiĔska, K., 104 Kowal, S., 303 Kramina, A., 136 Krauss, M., 45 Kurz, G., 153, 154, 155, 161, 170 Labov, W., 121-122 Lakoff, G., 23 Lambrecht, K., 31-33, 38 Lanz, L, 46 Larkin, C., 223 Lascarides, A., 198 Lasersohn, P., 289-290 Leahy, R., 118 Lee, C., 286-288 Lefevere, A., 135 Leonetti, M., 26, 31 Leudar, I., 123 Levinson, S., 28, 282 Lewis, B., 123-124 Lewis, M. P., 44 lexical entry, 202, 211, 214-215, 249, 251, 255, 259, 261, 263264, 269, 271-272, 274, 304, 306 lexical pragmatics, 9, 11, 14, 83, 131, 174, 199, 222, 246, 250, 254, 258, 275-276, 297-298, 326 Linde-Usiekniewicz, J., 3, 16-18, 24-27, 29, 33-34 Ljung, M., 304

Index loan, 10, 257-258, 260-261, 263267, 269, 272, 274 logical entry, 120, 218, 306, 316317 logical form, 87, 179-180 López Bobo, M. J., 301, 304 Mackenzie, J. L., 19 MacLean, E. A., 44-47, 50, 52-54 make-sense frame, 233-236, 239240 manipulation, 6, 104, 132-141, 144, 147-150, 190, 224, 226, 228229, 234, 236-237, 239-240 Marek, B., 107 Martínez-Sierra, J. J., 223 Mateo, J., 192 Mathesius, V., 27 Matsui, T., 64 Matthewson, L., 49 Mattiello, E., 279-280 Mayol, L., 304 Mel’þuk, I., 18 Meng, K., 303 mental effort, 5, 87, 90-91, 93, 9799, 292 Mercier, H., 3, 166 Merlini Barbaresi, L., 250, 279 metaphor, 6-7, 12-13, 118-119, 123128, 130-131, 152-161, 167-174, 177, 181, 201, 214, 218, 221, 246, 259, 294, 323, 325 metarepresentation, 96, 169-170, 177, 193-194, 209, 222 MGI/SCI, 228-232, 239 Miller, G. A., 176 Miltsakaki, E., 65, 67 Minkova, D., 278 Mioduszewska, E., 104, 106-107 Miskovic-Lukovic, M., 290, 292 modality, 3, 45, 48, 63, 145 modulation, 9, 67, 69-70, 79, 81 Montes, R. G., 303 Moreno Fernández, F., 304 Nagai, T., 44-46 naïve optimism, 106, 113

Applications of Relevance Theory: From Discourse to Morphemes narrowing, 9, 11, 13, 119, 198, 201, 213, 218, 251, 253, 262, 277, 290-293, 295, 308, 320, 325 nativization, 265 Needham-Didsbury, I., 124 neologism, 9-10, 246, 247-250, 254256, 263-264, 267-269, 273-276 Nicoloff, F., 299 Nikolaeva, I., 31 Noh, E.-J., 209 Norrick, N., 304 noun class, 10, 249, 253, 255-257, 262, 265, 267, 273 Nye, C., 124 O’Connell, D. C., 303 Oakley, T., 156 Öhlschläger, G., 45, 49 online discourse, 90 Oppermann-Marsaux, E., 304 oral-written, 90, 96 Orwell, G., 8, 158, 175, 183-185, 187-188, 192 oscillating effect, 204-205, 209, 212-214, 217, 220, 319 overdifferentiation, 266 overextension, 11-13, 312-313, 315316, 320, 325 over-inclusion, 12, 316, 321 Padilla Cruz, M., 300, 304 Paivio, 127 Palmer, F., 45, 48 Papafragou, A., 4, 45, 49, 55-56, 58, 61 paradigmatic pun, 204, 221 Pariser, E., 93-94 Pawelczyk, J., 117 Peräkylä, A., 123 Peterson, T., 49 Pilkington, A., 127, 294 Plag, I., 278 Pollio, H. R., 207 positive cognitive effect, 86, 88, 92, 95, 98, 134 Powell, G., 307-308 pragmatics, 4, 12-14, 18, 24, 31, 40, 49, 54-55, 61-64, 82-83, 91-92,

333

99, 104, 116, 129-131, 150, 153154, 156, 174, 181, 194, 221222, 241-243, 250, 276, 296297, 322, 325-326 prefix, 249-253, 255-257, 264-265 press article, 6, 133, 138-139 procedural meaning, 2-4, 11, 17-18, 26, 33, 38, 39-40, 57, 68, 80, 251-252, 268, 277, 292-293, 303-304 pro-concept, 307, 309 psychotherapeutic discourse, 117118, 121 pun, 8, 111, 198-200, 202, 204-208, 211-212, 220-222, 242, 319 pun-like utterance, 8, 199, 202, 204, 206, 212, 220 Puzynina, J., 138 Quilligan, M., 154 Quirk, R., 299, 301, 303 Raskin, V., 207 Reboul, A., 124-125 Recanati, F., 65 reduplication, 10, 277-284, 287-290, 293-297, 301 reference assignment, 228, 236, 239 Reinecke, M., 118 Rescorla, L., 315-316 rheme, 3, 18, 31-40 Rizzi, L., 33 Romek, Z., 139 root, 249-250, 252, 257, 267-268, 274 Rossano, F., 123 Rovin, J., 238 Rubio Fernández, P., 169 Ruiz Moneva, M. A., 182 Sak-Wernicka, J., 108 Salich, H., 132 Sandburg, C., 168 Sanford, A. J., 219 Sapir, E., 278-279, 283 Saramago, J., 158 saturation, 56, 65, 67, 71, 178, 180 Saussure, L. de, 132, 137-138, 140 Sax, D., 22, 32

334 Schiffrin, D., 121-122 Schrabback, S., 303 Schröder, M. C., 248, 274 secondary interjection, 11-12, 300, 302-304, 307, 310-321 semantic change, 30, 266, 320, 325 semantics, 3, 13, 16, 18, 20, 24, 4041, 49, 54-56, 61-63, 70, 75, 82, 129, 296-297, 322, 326 Shenhav, S. R., 303 Sherzer, J., 224 simile, 124-125, 129, 131, 172, 217219 Simons, G. F., 44 Singer,  ., 93 Slobin, D., 23 social relationship, 105, 111, 113 Solska, A., 205, 207, 311, 318-319 Song, M., 286-288 sophisticated understanding, 106, 115 speaker attitude, 45, 49-50, 61, 250 Sperber, D., 2-3, 5-7, 13-14, 16-18, 20, 24-26, 31-32, 35-37, 43, 54, 56-57, 63, 66, 75, 83, 86-89, 92, 101, 103-107, 113-116, 118-120, 125, 130-131, 133, 137, 150, 156, 159, 166, 168, 173-177, 181-182, 184, 194, 199-201, 222, 242-243, 246, 248, 269, 275, 290-294, 297-298, 300, 303, 306-308, 312-313, 318, 320, 324, 326 Stark, J. E., 267 stereotype, 133, 229-230, 233-234, 239 Stockwell, R., 278 Stott, R., 118 subjectification, 11, 304-305, 325 Suls, J., 228 syntax, 3, 13, 16, 18, 28, 31-32, 38, 41-42, 62, 237 Szarkowska, A., 139 Szehidewicz, E., 118, 128 ĝwiatkowska, M., 299 Taillard, M.-O., 137

Index Tannen, D., 123 Thagard, P., 7, 157-158, 160 theme, 3, 18, 31-40, 158, 160, 229 theme-rheme articulation, 3, 31-36, 39-40 Thompson, C., 99 Thomson, J., 312 Thun,  ., 283 Tolmie, S., 156 Toposa, 9-10, 246-257, 259-268, 270, 272-275 translation, 4, 6, 10, 13-14, 50, 53, 67, 72-76, 115, 132-142, 144, 146-150, 172, 175-176, 179, 182-184, 186-189, 192-194, 221, 242-243, 247-248, 265, 273-275, 293 translation of irony, 175, 182-184, 192, 194 Traugott, E. C., 304 trope, 154, 156, 158 Tsibulsky, D., 304 Turkle, S., 86 Turner, M., 157 underdifferentiation, 266 Unger, C., 153-154, 159, 162-163, 165 utterance-type, 3, 17, 28-31, 33-34, 38-40 Vaidhyanathan, S., 94 van der Hoek, M., 304 Vaskó, I., 66 Vega Moreno, R. E., 218, 259, 291, 309 Vehviläinen, S., 123 Viaro, M., 123 virtual space, 86 visual-verbal, 90, 96 Waáaszewska, E., 12, 290, 300, 304, 308-309, 312-313, 315 Watson, P., 105 Wearing, C., 128, 154, 168-172 Wedgwood, D., 22 Weeks, W., 229 Werth, P., 153 Wharton, T., 293, 300, 303-304, 311

Applications of Relevance Theory: From Discourse to Morphemes Whitton, L., 286-287 Wierzbicka, A., 282-283, 294, 301302 Wilkins, D. P., 302 Wilson, D., 2, 6-7, 9, 13-14, 16-18, 20, 24-26, 31-32, 35-37, 43, 54, 56-57, 63, 66, 69, 75, 83, 86-89, 92, 101, 107, 115-120, 125-126, 128, 130-131, 133, 137, 150, 152-153, 156, 158-159, 161, 166-168, 172-177, 181-184, 188, 194, 199-201, 209, 212, 220, 222, 242-243, 246-248, 252-253,

335

263-264, 274-276, 290-294, 297298, 300, 303, 306-308, 311313, 318, 320, 324, 326 word formation, 10, 278, 294-296 Yus, F., 4-5, 8-9, 86, 89-90, 95-100, 177-180, 183, 185, 190, 192, 223-225, 227-240, 318 Žegarac, V., 121 zero anaphor, 4, 64, 68, 70-71, 7577, 81 Zoll, C., 278 Zufferey, S., 52