275 89 2MB
English Pages 335 Year 2016
Relevance Theory Recent developments, current challenges and future directions
edi t ed by Manuel Padilla Cruz
John Benjamins Publishing Company
Relevance Theory
Pragmatics & Beyond New Series (P&bns) issn 0922-842X Pragmatics & Beyond New Series is a continuation of Pragmatics & Beyond and its Companion Series. The New Series offers a selection of high quality work covering the full richness of Pragmatics as an interdisciplinary field, within language sciences. For an overview of all books published in this series, please see http://benjamins.com/catalog/pbns
Editor
Associate Editor
Anita Fetzer
Andreas H. Jucker
University of Augsburg
University of Zurich
Founding Editors Jacob L. Mey
Herman Parret
Jef Verschueren
Robyn Carston
Sachiko Ide
Deborah Schiffrin
Thorstein Fretheim
Kuniyoshi Kataoka
Paul Osamu Takahara
John C. Heritage
Miriam A. Locher
University of Southern Denmark
Belgian National Science Foundation, Universities of Louvain and Antwerp
Belgian National Science Foundation, University of Antwerp
Editorial Board University College London University of Trondheim University of California at Los Angeles
Susan C. Herring
Indiana University
Masako K. Hiraga
St. Paul’s (Rikkyo) University
Japan Women’s University Aichi University
Universität Basel
Georgetown University Kobe City University of Foreign Studies
Sandra A. Thompson
Sophia S.A. Marmaridou
University of California at Santa Barbara
Srikant Sarangi
Teun A. van Dijk
University of Athens Aalborg University
Marina Sbisà
University of Trieste
Universitat Pompeu Fabra, Barcelona
Yunxia Zhu
The University of Queensland
Volume 268 Relevance Theory. Recent developments, current challenges and future directions Edited by Manuel Padilla Cruz
Relevance Theory Recent developments, current challenges and future directions
Edited by
Manuel Padilla Cruz University of Seville
John Benjamins Publishing Company Amsterdam / Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
doi 10.1075/pbns.268 Cataloging-in-Publication Data available from Library of Congress: lccn 2016029912 (print) / 2016043519 (e-book) isbn 978 90 272 5673 7 (Hb) isbn 978 90 272 6648 4 (e-book)
© 2016 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Company · https://benjamins.com
Table of contents
Introduction: Three decades of relevance theory Manuel Padilla Cruz
1
Part I. Issues on procedural meaning and procedural analyses The speaker’s derivational intention Thorstein Fretheim
33
Cracking the chestnut: How intonation interacts with procedural meaning in Colloquial Singapore English lah Junwen Lee and Chonghyuck Kim
59
Reference assignment in pronominal argument languages: A relevance-theoretic perspective Helga Schröder
81
Conceptual and procedural information for verb tense disambiguation: The English Simple Past Cristina Grisot, Bruno Cartoni and Jacques Moeschler
103
Part II. Discourse issues Relevance theory and contextual sources-centred analysis of irony: Current research and compatibility Francisco Yus
147
Distinguishing rhetorical from ironical questions: A relevance-theoretic account Thierry Raeber
173
vi
Relevance Theory
Part III. Interpretive processes Relevance theory, epistemic vigilance and pragmatic competence Elly Ifantidou
193
Evidentials, genre and epistemic vigilance Christoph Unger
239
Part IV. Rhetorical and perlocutionary effects of communication Rhetoric and cognition: Pragmatic constraints on argument processing Steve Oswald
261
Perlocutionary effects and relevance theory Agnieszka Piskorska
287
Conclusion: Some directions for future research in relevance-theoretic pragmatics Manuel Padilla Cruz
307
Contributors Index
321 325
Introduction
Three decades of relevance theory Manuel Padilla Cruz University of Seville
Almost three decades have elapsed since the publication of Dan Sperber and Deirdre Wilson’s (1986) seminal and most influential book Relevance. Communication and Cognition, and exactly two since the publication of its second edition (Sperber and Wilson 1995). In them, the authors presented and revised a series of claims and principles about human verbal communication and the foundational postulates of what is now known as ‘relevance theory’. Over all these years, their two books and numerous subsequent publications have given rise to a strand of research in pragmatics with a psychological basis and cognitive orientation. Often alluded to as ‘relevance-theoretic pragmatics’, this strand seeks to unravel how the mind processes utterances – and, more widely, discourse – the contribution of diverse linguistic elements (e.g., discourse markers, particles, adverbials, intonation, etc.) to comprehension, why the mind arrives at a particular interpretation and the effects that may follow from understanding utterances in one way or another. Relevance theory has certainly awoken the interest of many pragmatists and linguists in general by posing many intriguing problems and thought-provoking questions. Relevance theorists’ continuous challenging of often-taken-for-granted assumptions, claims, generalisations, and even whole models, has also brought fresher air to those disciplines. Indeed, they have analysed in depth a wide variety of linguistic and communicative phenomena from a different perspective and with a new theoretical apparatus, which has shed much light onto underexplored or overlooked issues. This book celebrates these happiest anniversaries and, most importantly, the fact that relevance theory still continues to appeal to researchers, who find in it a very valuable model for understanding the intricacies of linguistic communication. This is a collection of papers, which very sincerely acknowledges the extensive work carried out not only by the authors of the theory themselves, but also by a large number of researchers who have elaborated on some of its postulates and distinctions, empirically tested some of its predictions or applied the theory to diverse domains or neighbouring fields, thus expanding its scope. Consequently, doi 10.1075/pbns.268.01cru © 2016 John Benjamins Publishing Company
2
Manuel Padilla Cruz
a tribute of deepest gratitude is paid to all of them for their brave efforts to answer questions related to an incredibly complex human activity. Ten papers are collected here. Five of them elaborate on issues that have traditionally concerned practitioners in relevance theory: the intentional nature of communication, how speakers guide hearers to recover intended meaning, how specific types of utterances or linguistic elements are interpreted, or the consequences of communication. Nonetheless, these papers present recent developments in these areas and show how relevance theorists have faced various challenges therein. Other two papers apply the relevance-theoretic apparatus to account for features of some linguistic varieties or languages, thus helping to understand how they are structured and function. In so doing, these two papers also show interesting applications of relevance theory to other linguistic disciplines. The remaining three papers also present some of the new directions that the theory is following, as they consider epistemic vigilance mechanisms (Mascaro and Sperber 2009; Sperber et al. 2010). More precisely, those papers also address the interaction of some linguistic elements with such mechanisms, explain their role in rhetoric and argumentation, or consider their importance in second language teaching and acquisition. Thus, these papers also suggest interesting collaboration between relevance theory and fields like argumentation studies or interlanguage pragmatics. Finally, the book concludes with a chapter that suggests further possible future directions for research in relevance-theoretic pragmatics. These works offer new insights and seek to fuel what can be considered an authentic revolution in pragmatics, particularly in its cognitive branch, which had scarcely been convulsed since the various publications on meaning and communication by philosopher of language Herbert Paul Grice. Although most of the chapters summarise essential postulates and explain the key notions they rely on, in what follows Sperber and Wilson’s (1986/1995) views on communication, underlying assumptions, fundamental claims and basic concepts are presented. Since many readers will surely be familiarised with them, the following section is simply intended as a reminder. For those novel to the theory, this discussion supplies the necessary background that enables them to conveniently understand the works gathered here, as well as what relevance theory has meant. Next comes a review of research done in various areas from a relevance-theoretic perspective, which helps situate the chapters of this book in the wider panorama of relevance- theoretic pragmatics. Finally, this introductory chapter is closed with a description of the chapters.
Three decades of relevance theory
1. The relevance-theoretic revolution In a field like linguistics, where most theories, models and frameworks have been constantly revised, refined, questioned and eventually abandoned in favour of others, referring to the contribution made by a particular one as a ‘revolution’ might sound presumptuous and even biased. Undeniably, proponents of new theories, models and frameworks have always been encouraged by a perennial and daring interest in unraveling how linguistic systems are organised and how they work at the service of communication, as well as in overcoming limitations in prior approaches. In a relatively young discipline like pragmatics, scholars have incessantly attempted to grasp the enormous complexity of human verbal communication, where linguistic systems are put to use to convey information and achieve goals that may crucially impact on human relationships, and the outputs of such systems have to be processed by an until fairly recently practically unknown mechanism like the human mind. Relevance theory originated as an exciting endeavour to find convincing answers to a series of only apparently simple and obvious questions for which many pragmatists could have thought there were final answers: – – – –
How do humans understand each other? How do we arrive at a particular interpretation of what others say? What makes us end up with a specific interpretation of an utterance? Why do we select or reject an interpretation?
In their search for answers to these questions, Sperber and Wilson challenged many tacitly accepted assumptions about how languages are used, how the human mind works and what might happen in it while processing input. Thus, they seriously questioned the code model of communication, which was deeply entrenched in the western linguistic tradition, and showed its many drawbacks, gaps and inconsistencies. Some progress had been made when the role of inference in communication was taken into consideration. Indeed, the also influential work by Grice (1957, 1975) – wherein relevance theory is rooted and to which it is greatly indebted – meant a huge leap. However, Sperber and Wilson (1986/1995) went well beyond the so-called Gricean pragmatics by proposing a model that draws from valuable insights from disciplines such as philosophy of mind, developmental psychology or cognitive anthropology, to name but some. Thus, relevance theory incorporated some of their notions and views to pragmatics in order to offer a profoundly cognitive, psychological perspective that has implications for the nature and role of semantics and pragmatics.
3
4
Manuel Padilla Cruz
1.1
Intention, manifestness and cognitive environments
Like Grice (1975), Sperber and Wilson (1986/1995) conceive of communication as an intentional activity, in which the speaker is prompted by two intentions: (i) her informative intention, which is her desire to make manifest to the audience a particular set of assumptions, and (ii) her communicative intention, which is her desire that the audience recognise that she actually has a particular informative intention. Recognising these two intentions requires theory of mind abilities – i.e., being able to attribute mental states to other individuals – and, therefore, metarepresenting the intentions and beliefs of others – i.e., creating representations of other public or private representations (Apperly 2012). These abilities make it possible for individuals to understand whether the communicator believes what she says is true – first-order mental states – or what the communicator knows the audience know, what the communicator intends the audience to believe, or what the communicator believes the audience believe – second-order mental states (Leekam 1991; Happé 1994; Sullivan et al. 1995; Sullivan et al. 2003; Wilson 2013). These layers of metarepresentation can be depicted as follows (Sperber 1994: 195): Speaker intends (attribution of communicative intention – 4) me to know (communicative intention – 3) that she intends (attribution of informative intention – 2) me to believe (informative intention – 1) that p
The notion of manifestness is a remarkable innovation. A certain fact or assumption is manifest to an individual at a certain time if that individual is capable of creating a mental representation of it and accepting that representation as true or probably true (Sperber and Wilson 1995: 39). The set of facts or assumptions of which an individual actually has a mental representation at a certain moment, or of which he may potentially form a mental representation, make up his cognitive environment. If two (or more) individuals are capable of forming similar, though not completely identical, representations of the same physical or psychological objects, those individuals share a mutual cognitive environment. Through the notion of manifestness Sperber and Wilson (1995: 18) overcome the drawbacks of notions like common knowledge (Lewis 1969) or mutual knowledge (Schiffer 1972), which implied that, for communication to succeed, individuals should mutually share a certain amount of knowledge. It is virtually impossible to distinguish the amount (and type) of knowledge that two (or more) individuals merely share from the knowledge that is truly mutual. Perceptual systems and cognitive mechanisms act like filters and greatly determine what individuals perceive and how they represent it. What is more, checking that two (or
Three decades of relevance theory
more) individuals actually share some knowledge would require time-consuming and almost never-ending mental operations. However, the innovativeness of relevance theory does not end here. Though sharing the Gricean view of communication as an intentional activity, Sperber and Wilson also react against some of Grice’s (1957, 1975) own ideas by arguing that, instead of the Cooperative Principle, two other general principles govern human cognition and communication. 1.2
A theory based on two general principles
Although based on Gricean pragmatics, relevance theory builds on it by rejecting some of its foundational postulates. Grice (1957, 1975) claimed that the Cooperative Principle and a series of maxims – those of quantity, quality, relevance and manner – govern communication. Communicators may abide by, covertly violate or blatantly flout those maxims in order to convey implicit contents and achieve certain effects. Also, Grice (1975) acknowledged the existence of other maxims of a social or aesthetic nature, such as that of politeness, which led to some adherents to his ideas to elaborate on those maxims and suggest different models to account for (im)politeness (e.g., Lakoff 1973; Leech 1983). Others, in turn, modified the number or raised the status of the original maxims to that of principles (e.g., Horn 1996; Levinson 2000), thus giving rise to neo-Gricean approaches. However, the Cooperative Principle and its maxims seemed to be based on intuitions about communication and observations of a series of regularities. Relevance theory, in contrast, is a post-Gricean model; it questions the existence of the Cooperative Principle and its maxims, since their origin is unclear, they do not seem to have universal validity and their operation seems to have different effects depending on circumstances (Wilson and Sperber 1991a: 381, 1991b: 586). If anything, the Gricean Principle and maxims would have the status of (cultural) norms, understood as “[…] internalised, unconscious patterns that the individual follows without even noticing that he is complying with an unwritten model” (Escandell Vidal 2004: 349). Relying on a constant tendency that has propelled the evolution of the human species in general and the human mind in particular and greatly contributed to its efficiency – maximisation of gain in exchange of effort invested – Sperber and Wilson reject the Cooperative Principle and its maxims and propose two very general principles that govern communication and cognition.1 Indeed, by ‘principle’ they understand a formalization of how a particular system works; in other 1. In their 1986 work Sperber and Wilson only proposed a single principle.
5
6
Manuel Padilla Cruz
words, a “[…] causal, mechanical explanation” (Escandell Vidal 2004: 349). These two principles are the cognitive and communicative principles of relevance. The former states that human cognition is geared towards the maximization of relevance and is argued to reflect how the human mind functions. The latter, in turn, claims that every act of intentional communication communicates a presumption of its own optimal relevance. This second principle operates in all cases of intentional communication and is responsible for the selection of an interpretation out of the many possible ones that an utterance may have: Communicators and audience need no more know the principle of relevance to communicate that they need to know the principles of genetics to reproduce. Communicators do not ‘follow’ the principle of relevance; and they could not violate it even if they wanted to. The principle of relevance applies without exception […] It is not the general principle, but the fact that a particular presumption of relevance has been communicated by and about a particular act of communication, that the audience uses in inferential comprehension. (Sperber and Wilson 1995: 162)
These two new principles are based on the notion of ‘relevance’, which is the true cornerstone of the theory. Sperber and Wilson also characterise this notion in more precise terms than Grice (1957, 1975) actually did. 1.3
An underlying key notion
Grice (1975) included the maxim of relation and worded it as “be relevant”, but unfortunately he did not define what he took ‘relevance’ to be, nor did he clearly explain its role in communication. In contrast, Sperber and Wilson (1986/1995) portray relevance as a property of the stimuli individuals produce and, therefore, of a very special sub-set thereof: utterances. They define it on the basis of two factors: a. The positive cognitive effects that information yields, or the improvements to our mental representation of the world or set of beliefs about it. Those improvements may be strengthening of previous beliefs, contradiction and eventual rejection of those beliefs or the formation of new beliefs from the interaction of previous ones with new information. b. The processing or cognitive effort required by an item of information. This depends on the complexity of the linguistic form of the utterance that conveys that item of information (i.e., its syntactic structure, lexical items, etc.) or the effort of memory needed to retrieve or select a suitable context for processing it.
Three decades of relevance theory
These two factors are also essential for understanding what the presumption of optimal relevance involves: (i) utterances will normally be relevant enough – i.e., they will result in enough cognitive effects – for the hearer to decide to invest the cognitive effort necessary to process them, and (ii) the formulation of utterances will normally be the one the speaker thinks, given her abilities and preferences, will result in a satisfactory amount of cognitive effects (Sperber and Wilson 1995: 270). The first part of this presumption means that hearers normally expect some cognitive reward which they cannot obtain otherwise, i.e., positive cognitive effects or “[…] a worthwhile difference to [their] representation of the world […]” (Wilson and Sperber 2002: 251). In turn, the second part means that the speaker will normally be interested in producing utterances that are easily comprehensible and provide the hearer with enough evidences for the intended cognitive effects or additional cognitive effects rendering utterances optimally relevant (Wilson and Sperber 2002: 256–257). Nevertheless, the speaker’s performance will depend on her own cognitive skills and capabilities, which may be conditioned by absentmindedness, tiredness, boredom, etc., and her goals, among which is complying with norms dictating, for instance, register, amount or type of information to dispense, formality, etc. (Mazzarella 2013: 33–35). In addition to the two principles of relevance, Sperber and Wilson (1995: 166) initially proposed a criterion of consistency with the principle of relevance, according to which an interpretation is consistent with the principle of relevance if, and only if, the speaker expects it to be optimally relevant. This means that, if the hearer finds an interpretative hypothesis optimally relevant, he should not think that the speaker intended a particular utterance to be optimally relevant under another interpretation. The hearer will think that an interpretative hypothesis was intended upon checking that it yields enough cognitive effects that offset a reasonable amount of cognitive effort. By clearly defining relevance and arguing that the constant search for optimal relevance governs human cognition and communication, Sperber and Wilson go well beyond Gricean pragmatics. However, as pointed out, relevance theory also reacts against the well-established code model of communication. Indeed, Sperber and Wilson conceive of communication as a human activity requiring a great amount inference, which has significant implications for understanding the role of semantics and pragmatics. 1.4
A new conception of communication
The code model metaphorically depicted communication as a process in which the code enables the speaker to package her thoughts (i.e., encoding) in a parcel (i.e., utterance) and send it over to the hearer (i.e., articulating and speaking). The
7
8
Manuel Padilla Cruz
hearer’s task is to open that parcel and match the encoded input to corresponding semantic representations (i.e., decoding) in order to decipher, so to say, what the speaker means. However, the code alone does not suffice to arrive at speaker meaning: not everything that the speaker means is encoded and there is no guarantee that the hearer interprets what she means correctly (Wilson and Sperber 1991b: 584–585; Sperber and Wilson 1995: 27). Hearers have to segment sounds and delimit words; parse and disambiguate constituents; narrow or broaden concepts; assign reference to some expressions; recover elided linguistic material; determine the attitude the speaker projects towards the proposition expressed, her degree of certainty about it or if she is performing a certain action, and supply any assumption that is necessary for arriving at implicit contents. All these tasks rely on inference, so Sperber and Wilson (1987: 698, 1995: 10) opt for describing communication as an ostensive-inferential activity. It is ‘ostensive’ because the speaker shows something to the hearer – an utterance, which is indirect evidence for her communicative intention – and uses it to attract and direct the hearer’s attention to something: her informative intention. Communication is ‘inferential’ because the hearer has to work out the speaker’s meaning and her underlying intention when drawing the hearer’s attention.2 Grice (1957, 1975) emphasised the reliance of communication on inference, but he envisaged its role as limited to the determination of the spatio-temporal coordinates of an utterance, assignment of reference to some expressions and recovery of the implicit contents. Accordingly, hearers arrive at ‘what is said’ through decoding and very little inference, while arriving at ‘what is implied’ greatly depend on inference. This unveils a view of the role of semantics and pragmatics in communication in which the former contributes to the recovery of the proposition expressed, while the latter was crucial for arriving at implicit contents (Wilson and Sperber 1993: 3). On the contrary, by proving that inference intervenes in all the tasks listed above, Sperber and Wilson (1986/1995) present a more intricate picture of comprehension where pragmatic processes are indispensable for the recovery of explicit content. Showing that communication involves much more than an encoding- decoding process also leads Sperber and Wilson to contend that individuals may use utterances in order not to simply pack and send over their own thoughts. By means of utterances communicators may reproduce words or phrases, represent diverse states of affairs and even allude to the thoughts of other people. 2. As a cognitive-pragmatic model, relevance theory focuses on comprehension and aims to account for how hearers arrive at speaker’s meaning through a series of simultaneous inferential tasks. This should not mean that Sperber and Wilson (1986/1995) are not aware of the crucial role that inference plays in production.
1.5
Three decades of relevance theory
Utterances as metarepresentations
Sperber and Wilson (1986/1995) believe that individuals can employ any natural or artificial phenomenon in order to represent another phenomenon it somehow resembles in certain respects. By means of utterances, which are public representations – i.e., perceptible, audible – individuals represent their own private, mental representations. Moreover, utterances may represent existing or desirable states of affairs or the thoughts or representations that an individual thinks another individual, or group of individuals, entertains. This is the basis for Sperber and Wilson’s (1995: 228–232) distinction between descriptive and interpretive dimensions of language usage. Since utterances are representations of representations, they have metarepresentational uses (Noh 2000). One of these is metalinguistic, when utterances publicly reproduce other public representations like words, phrases or utterances, or their logical or conceptual content. Clear examples are direct speech and quotations. Another metarepresentational use is interpretive, when utterances publicly resemble other public or private representations, like other utterances or thoughts. A typical example is indirect speech. Furthermore, if utterances metarepresent the thoughts or utterances that (an) other individual(s) entertain(s) or say(s), they are used attributively or function as attributive metarepresentations, as the source of those thoughts or utterances can be identified more or less easily (Wilson 1999: 148). In contrast, if utterances only represent a word, phrase or sentence produced by an unidentifiable source, they work as non-attributive metarepresentations. Examples of non-attributive metarepresentations are (Wilson and Sperber 1988; Wilson 1999): a. Negative and disjunctive sentences, which metarepresent possible information or thoughts. b. Interrogative and exclamative sentences. These metarepresent desirable information or thoughts. Interrogatives are requests for information if the speaker metarepresents an answer that the hearer can give, while they are offers of information if the speaker metarepresents an answer that she can give to the hearer. c. Imperative sentences. If the speaker metarepresents a state of affairs as desirable from her own viewpoint, they work as requests; if she metarepresents a state of affairs as desirable from the hearer’s viewpoint, they work as suggestions. When metarepresenting another individual’s thoughts or words, the speaker may also express an attitude to them, so the utterance becomes echoic. The attitudes that the speaker may express are numerous, as she may “[…] indicate that she
9
10
Manuel Padilla Cruz
agrees or disagrees with the original, is puzzled, angry, amused, intrigued, sceptical, etc.; or any combination of these” (Wilson 1999: 147). However, three attitudes seem to be essential for understanding some types of utterances: endorsing, questioning and dissociative/rejecting. This last one is characteristic of irony, which Sperber and Wilson (1986/1995) analyse as a case of echoic attributive metarepresentation. An utterance is interpreted as ironic if the hearer identifies (i) that the speaker echoes her own or somebody else’s thoughts or words, (ii) the source of those thoughts or words and (iii) that the speaker dissociates from or rejects those thoughts or words. This analysis of irony contrasts with both the classical and Gricean definitions, as well as with other more recent treatments in pragmatics and psychology. Understanding utterances, then, involves much more than identifying words or arranging these in constituents. Hearers have to adjust concepts, assign reference, recover material that is not overtly present and, very importantly, determine the attitude the speaker projects towards what she says or whether what the speaker says is her own thoughts or somebody else’s thoughts. As pointed out, all these tasks rely on inference, so Sperber and Wilson depict comprehension as a process in which the constant search for relevance causes the human make various inferences at the same time and at an incredibly fast pace. 1.6
A new picture of comprehension
Decoding only yields a set of conceptual representations or logical form, which is not fully propositional and needs enriching through inference. Inferential enrichment amounts to performing (some of) the tasks listed above and results in the explicature of an utterance. This is the explicit content communicated by an utterance. This content can be related to any information that the hearer thinks that the speaker intends or expects him to access – i.e., implicated premises – in order to arrive at some implicated content – i.e., implicated conclusions. Both implicated premises and conclusions amount to the utterance implicit content and are often alluded to by the umbrella term ‘implicatures’. These are strong if the hearer has enough evidence to think that the speaker expects him to access or draw them, and weak if the hearer lacks enough evidence, so they may be contents that the speaker derives at his own responsibility (Sperber and Wilson 1995: 199–200). The comprehension or pragmatic module does not perform all these inferential tasks sequentially, but holistically.3 It processes linguistic input and generates interpretative hypotheses by simultaneously decoding, considering possible 3. Like other modules, it has a very specific domain of action, is mandatory and works in a fast and frugal manner (Fodor 1983).
Three decades of relevance theory
disambiguations, conceptual adjustments and reference-assignments; looking for elided constituents, constructing possible speech-act or propositional-attitude descriptions, searching for implicated premises and anticipating possible implicated conclusions. As a result, utterance comprehension can be described as a process of mutual adjustment of both explicit and implicit content (Carston 2002; Wilson and Sperber 2002, 2004). Whether its outcomes is one or another depends on considerations about which option might result in a satisfactory number of cognitive effects in exchange of a reasonable amount of cognitive effort. This means that the pragmatic module does not search for the most relevant outcome, which would involve constructing all the possible interpretative hypotheses and assessing them. From this picture of comprehension follows the relevance-theoretic comprehension heuristics, which captures how the mind works in comprehension. Accordingly, hearers follow the path of least effort when adjusting explicit and implicit content of utterances, thus formulating interpretative hypotheses, and stop when their expectations of relevance are satisfied (Wilson 1999: 136; Wilson and Sperber 2002, 2004: 612). It is reasonable for hearers to do so because they will normally expect speakers, depending on their abilities and preferences, to formulate utterances in an easy and straightforward manner (Wilson and Sperber 2002: 259). Since relevance decreases as cognitive effort increases, hearers will very likely regard a particular interpretative hypothesis as intended if they can easily construct it. Moreover, it is reasonable for hearers to stop when an interpretative hypothesis satisfies their expectations of relevance because there should only be one optimally relevant interpretation. Two (or more) optimally relevant interpretations would detract from optimal relevance, as hearers will have to invest the additional effort to assess them.4 1.7
A new conception of the role of the speaker
The very nature of inference does not guarantee that the outcome of mutual adjustment is the expected or intended one. The audience may misunderstand the speaker if they fail at any of the tasks involved (Yus Ramos 1999a, 1999b). However, communicators may guide the audience to the intended interpretations by means of their linguistic or expressive choices, thus ensuring that the audience understand them correctly. Style is therefore seen by Sperber and Wilson (1995: 219) as consequence of the communicator’s willingness to be optimally relevant and assist the audience in comprehension. Speakers may take advantage 4. See Allot (2002) for a discussion on rationality and the relevance-theoretic comprehension procedure.
11
12
Manuel Padilla Cruz
of linguistic repertoires in order to generate specific cognitive effects that hearers could not obtain otherwise. Formulations of a specific message that turn out indirect, lengthy, repetitive, and probably costlier in terms of processing effort, may evidence the speaker’s desire to communicate (a wide array of) weak implicatures that contribute to optimal relevance. Among those implicatures, hearers may deduce information about how the speaker feels about a particular state of affairs, the reasons why the speaker phrases her message in a particular manner, how she treats the audience, their social relationship, etc. Weak implicatures like those would be of a behavioural nature (Jary 2013) and their derivation is essential for understanding phenomena like (im)politeness or literary communication (Pilkington 2000). 1.8
The conceptual-procedural distinction
Communication may now be metaphorically described as a ball where the speaker initially takes the lead in her conversational turn and then passes it to the hearer, who will take it in his turn and pass it back to the speaker again. As this exchange of turns takes place, meaning progressively emerges and is confirmed or negotiated. Although speakers and hearers collaborate in the joint endeavor of co-constructing meaning, the former can also assist the latter in their interpretative tasks. Speakers can direct hearers’ attention to specific sets of assumptions, or bring to the fore those they expect hearers to exploit, by means of the conceptual items in utterances. Thus, speakers help hearers select the mental context wherein to interpret what is said and, if necessary, figure out the premises necessary to arrive at implicit contents. On the other hand, speakers can also indicate how hearers should assign reference to some expressions, relate diverse items of information or construct adequate descriptions capturing their attitude to the proposition expressed, their degree of (un)certainty about it or what they intend to achieve with their words. This is possible because a variety of linguistic elements encode procedural meaning: instructions that the comprehension module follows when computing information, which somehow impose constraints on inferences. The distinction between conceptual and procedural expressions has been another major contribution of relevance theory and has led a number of scholars to analyse a series of expressions accordingly (Wilson 2016; Carston 2016). Nouns, verbs and adjectives encode conceptual content, even if that content is amenable to subsequent adjustment resulting in what is known as ad hoc concepts: one-off, occasion-specific concepts (Wilson and Carston 2007; Carston 2012). In contrast, discourse markers encode procedures that indicate the relationships between
Three decades of relevance theory
specific propositions and, therefore, steer the comprehension module toward one direction or another. The notion of procedural meaning enables practitioners in relevance-theoretic pragmatics to overcome drawbacks of the Gricean distinction between conversational and conventional implicatures, the latter of which were thought to result from the linguistically encoded material. Some procedural elements like proper names and personal pronouns encode some conceptual content, even if schematic (Wilson and Sperber 1993).5 That conceptual content is integrated in the lower-level explicatures of an utterance together with the concepts encoded by other words. In contrast, the conceptual content of other expressions like attitudinal adverbials (e.g., ‘happily’, ‘unfortunately’), illocutionary adverbials (e.g., ‘frankly’, ‘seriously’), evidential adverbials (e.g., ‘obviously’, ‘evidently’), hearsay adverbials (e.g., ‘allegedly’, ‘reportedly’), and some parenthetical expressions (e.g., ‘they say’, ‘I hear’) becomes part of higher-level explicatures. While attitudinal and illocutionary adverbials indicate to hearers the sort of attitudinal description under which they must embed the proposition expressed, evidential and hearsay adverbials and parenthetical elements indicate whether the speaker has or lacks adequate evidence about the information dispensed (Wilson and Sperber 1993; Wilson 1999). 2. Research within relevance-theoretic pragmatics In a paper published on the occasion of the first decade of relevance theory – the title of this chapter is clearly inspired by it – Yus Ramos (1998) extensively reviewed the contributions made by relevance theorists thus far. The publication of the theory was followed by some special issues in journals (e.g. Smith and Wilson 1992; Wilson and Smith 1993; Mateo Martínez and Yus Ramos 1998). An edited collection presented diverse applications and implications to understand irony, metaphor, metonymy, hearsay particles, some adverbials or scalar implicatures (Carston and Uchida 1998), while another presented in-depth analyses of the conceptual-procedural distinction, intonation, focus phenomena, semantically underdetermined linguistic forms or the role of pragmatic inference (Rouchota and Jucker 1998). The theory was also applied to literary communication (e.g., Pilkington 1991, 1992), media discourse (e.g., Yus Ramos 1995, 1997, 1998), translation (e.g., Gutt 1989, 1991) or humour (e.g., Jodłowiec 1991; Curcó 1995, 1996, 1997; Yus Ramos 1997, 1998). And a notable bulk of research centred on grammar and discourse, addressing issues such as: 5. Note, however, that some authors consider pronouns to be purely procedural items (e.g., Scott 2016).
13
14
Manuel Padilla Cruz
– discourse markers (e.g., Blakemore 1987, 1988, 1989; Jucker 1993; Moeschler 1993; Rouchota 1995); – different types of adverbials and particles (e.g., Blass 1989, 1990; Ifantidou 1992, 1993; Itani 1994; Imai 1998); – mood (e.g., Wilson and Sperber 1988, 1993; Clark 1993; Rouchota 1994) and modality (e.g., Berbeira Gardón 1993, 1998; Groefsema 1995); – tense and aspect (e.g., Moeschler 1993; Wilson and Sperber 1993), or – reformulations (Blakemore 1992, 1993, 1994). Research from a relevance-theoretic perspective has continued with renewed enthusiasm and a vivid impetus. The prominence given to the role of inference in the determination of the explicit content of utterances led Carston (2002) to collect and profoundly revise a series of her most illuminating works on the pragmatics of explicit communication. Her ideas were also celebrated years later with the publication of another volume that showed their explanatory potential (Soria and Romero 2010). Quite similarly, the evolution of the theory subsequently encouraged other scholars to revisit issues like literary communication (Pilkington 2000; Unger 2006), evidentiality (Ifantidou 2001), particles (Iten 2002), humour (Yus Ramos 2003, 2008, 2013, 2016; Solska 2012), discourse markers (Blakemore 2002; Hall 2007; Moeschler 2016), figurative speech (Vega Moreno 2007), intonation or paralanguage (Wilson and Wharton 2006; Wharton 2009). The resulting adjustments of the theory have also been presented in a series of works by Wilson and Sperber (2002, 2004, 2012). Later innovations, applications and suggestions for further research have also been included in some manuals, which make the theory accessible to students and novel researchers (e.g., Blakemore 1992; Yus Ramos 1997; Clark 2013). For the sake of exemplification, suffice it to mention those in areas like: – misunderstanding (e.g., Yus Ramos 1999a, 1999b; Jodłowiec 2008) and pragmatic failure (e.g., Padilla Cruz 2013a, 2014); – phatic communion (e.g., Žegarac 1998; Žegarac and Clark 1999a, 1999b; Padilla Cruz 2005a, 2007a, 2007b); – (im)politeness (e.g., Escandell Vidal 1996, 1998; Jary 1998a, 1998b, 2013); – historical linguistics (e.g., Ruiz Moneva 1997; Padilla Cruz 2003, 2005b; Clark 2016); – language acquisition, pragmatic competence and interlanguage pragmatic development (e.g., Padilla Cruz 2013b; Ifantidou 2014); – computer-mediated communication (e.g., Yus Ramos 2001, 2010, 2011); – expressive meaning (e.g., Moeschler 2009; Blakemore 2011; Piskorska 2012a; Wharton 2016);
Three decades of relevance theory
– communication disorders and clinical pragmatics (e.g., Papp 2006; Leinonen and Ryder 2008; Wearing 2010), or – the relationship between epistemic vigilance and understanding (e.g., Mascaro and Sperber 2009; Sperber et al. 2010; Padilla Cruz 2012; Mazzarella 2013, 2015). The latest directions relevance theory has taken can also be seen in other more recent edited collections and special journal issues. These gather works which delve into the conceptual-procedural distinction, the nature and role of ad hoc concepts, lexical pragmatics, the role of context and metarepresentation, the role of valence, phatic communion, focal stress, the interactions of modality and evidentiality with epistemic vigilance, humour, (im)politeness, miscommunication, interjections and response cries, various speech acts, emotions, implicit communication, translation or style (Mioduszewska 2004; Korzeniowska and Grzegorzewska 2005; Wałaszewska, Kisielewska-Krysiuk and Piskorska 2010; Escandell Vidal, Leonetti and Ahern 2011; Piskorska 2012b; Wałaszewska and Piskorska 2012; Wałaszewska 2015; Sasamoto and Wilson 2016). Finally, one more volume will soon present new applications to computer-mediated discourse, psychotherapeutic discourse, literary discourse, humorous discourse, lexical pragmatics or morphology (Wałaszewska and Piskorska, in press). The list of topics and references given here is obviously far from exhaustive, but it seeks to give an idea of the impressive amount of work done or in progress, to which the chapters gathered in this book purport to add up. Readers will certainly gain access to more specific references in databases, bibliographic repositories or catalogues, as well as on the “Relevance Theory Online Bibliographic Service.” Created some years ago and monthly updated by Yus Ramos (http://personal.ua.es/francisco.yus/rt2.html), this service facilitates access to an immense number of works through an index of authors in alphabetical order or a list of thematic sections. 3. This book This collection is divided in four parts on the basis of common thematic threads: procedural meaning, discourse issues, interpretive processes, and the rhetorical and perlocutionary effects of communication. Although, as shown in the preceding section, the range of interests of relevance theorists is quite wide, the parts in which this volume is divided address topics that have traditionally intrigued them and areas where significant contributions and progresses are being made. Moreover, although the chapters in each part delve into specific issues or phenomena,
15
16
Manuel Padilla Cruz
they have different orientations and implications for fields as diverse as linguistic description, morphology, machine translation, rhetoric and argumentation, or interlanguage pragmatic development. 3.1
Issues on procedural meaning and procedural analyses
The first part groups four papers that rely on the notion of procedural meaning. The first two papers look into the interaction of procedural elements with intonation, thus broadening the scope of procedural analyses, while the third paper presents what can be considered a traditional relevance-theoretic procedural analysis of some linguistic elements in an African language, thus contributing to its description. The last paper explores the procedural meaning of verbal tenses and shows the usefulness and applicability of procedural analyses for a field like machine translation. The first chapter is “The speaker’s derivational intention”, by Thorstein Fretheim. Based on the notion of derivational intention, which alludes to the route the audience should follow in order to reach the content of the speaker’s informative intention, this chapter shows the usefulness of that notion and argues that it cannot be separated from that of informative intention. Fretheim claims that the speaker’s derivational intention is constrained by encoded conceptual semantics and encoded procedural semantics alike. However, a given piece of procedural information may be at odds with the speaker’s derivational intention in a given context, and hence with her informative intention. Accordingly, he contends that the procedural meaning encoded by one expression may override and set aside the procedural meaning encoded by a co-occurring expression, if there is an intuitively felt conflict between the constraints on interpretation that the two expressions encode. In such cases, the more powerful procedural constraint is consonant with the speaker’s derivational intention and the less powerful procedural constraint is not. By means of two sections on the pragmatic functions of certain intonational phenomena in Norwegian, Fretheim illustrates how native speakers differently ranked co-occurring procedural constraints on pragmatic interpretation. This caused their audience to adhere to the procedural information conveyed by the expression with the higher rank and to disregard the information coming from the more modestly ranked encoder of procedural meaning. The first section reports on results from a listening comprehension experiment. It showed that the more highly ranked procedural constraint prevailed. The second section discusses the extent to which Optimality Theory may be a suitable framework for handling conflicts between co-occurring procedural constraints.
Three decades of relevance theory
The second chapter is “Cracking the chestnut: How intonation interacts with procedural meaning in Colloquial Singapore English lah”, by Junwen Lee and Chonghyuck Kim. Also assuming that procedural meaning may interact with intonation, it looks into the different pragmatic functions of this particle, which are argued to result from the interaction between its unitary semantic meaning and the effect of pitch. This chapter questions previous analyses of this particle as a marker of solidarity, warmth or informality, attenuation or emphasis, assertion or accommodation. Due to the variety of pitch contours with which this particle can be pronounced, it had generally been regarded as either a set of homonymic variants, or as a unitary particle with a monolithic meaning despite tonal differences. For Lee and Kim, ‘lah’ describes the preceding proposition as being of high strength. The falling tone characteristic of declaratives leads the hearer to interpret that proposition as referring to an actual situation, while the rising tone characteristic of interrogatives leads him to interpret it as alluding to a desirable thought. The various pragmatic functions ascribable to ‘lah’ then arise as a consequence of processing. Additionally, this chapter accounts for a particularly troublesome phenomenon: the ability of ‘lah’ to pragmatically strengthen declaratives but to weaken imperatives. Through the chapter, its authors show that relevance theory provides a useful framework for analysing the effect of intonation on the processing of this discourse particle, which opens up new ways of characterising other particles in Singapore English. In “Reference assignment in pronominal argument languages: A relevance- theoretic perspective”, Helga Schröder adheres to the relevance-theoretic view that reference assignment is part of the process of explicature construction and that pronominal expressions procedurally constrain it (Wilson and Sperber 1993; Sperber and Wilson 1995). By instructing hearers to pick out a reference, pronominals contribute to the computational side of comprehension. Based on data from Toposa, an Eastern Nilotic language spoken in South Sudan, and Kiswahili, spoken in Kenya, Shröeder demonstrates that two procedures are involved in reference assignment in African pronominal argument languages: an incorporated pronoun in the verb helps identify the referent, while that pronoun in the verb helps an attributive expression achieve referential status. Quite often, African languages do not mark their nominal expressions for definiteness or indefiniteness, so they enter the conversation as undetermined. Disambiguation of such expressions requires pragmatic enrichment. Those languages seem to follow a ‘double strategy’ in terms of reference assignment. Pronominals can be marked in the verb or occur independently. Independent pronouns do not encode procedural information guiding the selection of the correct referent, whereas incorporated pronouns do. Processing independent pronouns, Schröeder
17
18
Manuel Padilla Cruz
concludes, requires extra effort but yields cognitive effects related to focus identification and contrastive focus. The last paper in this part has an empirical orientation and combines corpus work and linguistic experiments. In “Conceptual and procedural information for verb tense disambiguation: The English simple past”, Cristina Grisot, Bruno Cartoni and Jacques Moeschler seek to elucidate which features should be included in a model explaining and predicting cross-linguistic variation in the translation of tenses. Thus, the authors also try to explain how a source language verb tenses may be disambiguated in order to choose from among different translation possibilities and which features could contribute to improving the output of Statistical Machine Translation systems. Corpus analysis reveals a lack of correspondence between English and French tenses. One of the most frequent divergences is that of the English simple past. Its semantic and pragmatic domains may be rendered in French through three tenses: passé composé, passé simple and imparfait. The authors assume that the conceptual and procedural contents of the English simple past can be used as disambiguation criteria in order to search for a French equivalent. They test this hypothesis through annotation experiments, whose results partially confirm their hypothesis. 3.2
Discourse issues
The two chapters in the second part of this book are connected by the fact that they address discourse. While the first chapter deals with the clues that hearers can rely on in order to perceive discourse as ironical, the second one analyses how specific manifestations of discourse, namely, rhetorical and ironical questions, may be distinguished. Obviously, these two chapters are further connected by the fact that they address irony from a relevance-theoretic perspective. Francisco Yus Ramos had previously argued that irony comprehension involves the simultaneous or sequential activation of one or several contextual sources. A successful activation of these sources, especially the saturation of information that they may give rise to, leads to the so-called “criterion of optimal accessibility to irony.” Also, the simultaneous or sequential activation of these sources generates a number of prototypical cases in the comprehension of irony. This view of irony, the author thinks, is fully compatible with the extant relevance-theoretic approach based on the idea that irony is a case of echoic metarepresentational use of language, in which the speaker expresses a dissociative attitude towards certain words or thoughts attributable to some other (group of) individual(s).
Three decades of relevance theory
In “Relevance theory and contextual sources-centred analysis of irony: Current research and compatibility”, Francisco Yus Ramos has two aims. On the one hand, he checks whether those prototypical cases of irony are still valid and do cover all the possible ironic situations, together with the introduction of a more fine-grained notion of narrowed cognitive environment. On the other hand, he assesses to what extent his approach fits the latest relevance-theoretic research on irony such as the one involving different types of metarepresentation that are activated in successful irony comprehension. Next, “Distinguishing rhetorical from ironical questions: A relevance- theoretic account”, by Thierry Raeber, aims to offer criteria for differentiating rhetorical from ironical questions. Some of their features have been conflated on the grounds of the interpretative effects ironical questions seem to yield. Indeed, ironical questions have often been regarded as subtypes of rhetorical questions (Bonhomme 2005), or vice versa, rhetorical questions have been considered as subtypes of ironical questions (Gibbs 2000) owing to the obvious answer they call for, their persuasive power and the cognitive effects they result in. However, Raeber suggests that rhetorical and ironical questions are to be distinguished because the former implicitly assert a proposition bearing relevance – in most cases the answer itself or its consequences – while such implicit proposition, though manifest to the hearer, is not relevant at all by itself in ironical questions. The proposition rhetorical questions implicitly assert is recoverable thanks to a biased choice of answer. In contrast, ironical questions exhibit their in-context inaccuracy, thus giving rise to specific effects resulting from the contextual absurdity of pragmatic expectations, which are to be mobilized in order to motivate the question. Accordingly, Raeber proposes that failure at satisfying the hearer’s expectations of relevance triggers the attitudinal, non-propositional effects associated with ironical questions. 3.3
Interpretive processes
The third part of the book also includes two chapters, but these are connected by the fact that they consider interpretation. While the first one centres on the interpretive processes by learners of a second language, the second one analyses how different linguistic elements are put to work in order to guide interpretation. The first chapter shows that relevance theory has significant implications for, and very helpful applications to, the field of interlanguage pragmatics. The second chapter, in turn, offers interesting insights into evidentials. Furthermore, both chapters are connected by the fact that, although in different domains, they take into account the role of epistemic vigilance mechanisms. These mechanisms
19
20 Manuel Padilla Cruz
have recently started to attract the interest of relevance theorists and are currently receiving due attention from them, as these works show. Assuming that little attention has been paid to the cognitive underpinnings of L2 learners’ pragmatic competence, Elly Ifantidou addresses the role of vigilance mechanisms in interlanguage pragmatic development in “Relevance theory, epistemic vigilance and pragmatic competence.” The author evaluates the effect of explicit pragmatic instruction in an EFL academic context and of language proficiency on learners’ metapragmatic awareness. This is an empirical study which considers different conditions of treatment, as determined by type of pragmatic input and level of language proficiency. It compares data from three groups of undergraduate students of English Language and Literature, which are used as control and experimental groups. Through newspaper editorials, Ifantidou examines how learners may exercise epistemic vigilance towards the source of information and thus avoid being accidentally or intentionally misinformed. In the metapragmatic awareness task she administered, learners relied on (i) relevance-theoretic mechanisms for comprehension of content and search for relevance, and (ii) epistemic vigilance for acceptance of both the content and source of communicated information. Since understanding is a precondition, even though not sufficient, for believing (Sperber et al. 2010: 368; Wilson 2010), learners’ pragmatic competence may benefit from sophisticated mindreading if they exercise epistemic vigilance towards the informant’s (i.e., newspaper journalist’s) epistemic states (e.g., acceptance, doubt, rejection) and intentions (e.g., to inform, to mislead). Therefore, the development of L2 learners’ pragmatic competence is contingent on the development of their epistemic vigilance mechanisms. The second paper analyses the interaction of epistemic vigilance with genre and evidentials – i.e., grammatical elements indicating the type of evidence the speaker has for making a statement. Authored by Christoph Unger, “Evidentials, genre and epistemic vigilance” shows that existing accounts have mainly explored the contribution of these elements to comprehension of speaker’s meaning. However, communicators may exploit evidentials in order to indicate that they are competent in distinguishing their information sources (Wilson 2011). Hence, evidentials help communicators portray themselves as trustworthy individuals and enable the audience to evaluate the evidential status of the communicated information. Drawing from Wilson (2011), Unger argues that evidentials may be used as genre indicators. Indeed, Aikhenvald (2004) had shown that reported evidentials very often appear in traditional narratives. This suggests that reported evidentials and traditional narrative genres may be conventionally related. Accordingly, the author contends that, as opposed to general markers of metarepresentational use
Three decades of relevance theory
and other indicators of information source, true reported evidentials raise the activation of a whole array of cognitive mechanisms specialised in checking support and coherence of the communicated content with existing beliefs. Furthermore, the relevance of traditional narratives resides in the validity of cultural values or norms communicated through exemplification. For this reason, the processing of such narratives strongly engages the argumentation module, whose activation is raised by reported evidentials. 3.4
Rhetorical and perlocutionary effects of communication
Before closing this book with additional directions for future research, the two chapters included in its fourth part analyse different effects of communication from a relevance-theoretic angle. The first chapter considers how arguments may succeed in rhetorical terms, thus introducing the machinery of relevance theory to the field of argumentation. In contrast, the second chapter looks into perlocutionary effects, an area that is increasingly drawing the attention of relevance theorists and where significant developments have been made. Moreover, both chapters also take into account the role of epistemic vigilance mechanisms, which links them to the contents of the previous part. “Rhetoric and cognition: Pragmatic constraints on argument processing” adopts a cognitive-pragmatic perspective on rhetorical effectiveness in order to hypothesise that the information-selection mechanisms in interpretation positively influence the outcome of subsequent argumentative evaluation. Moreover, author Steve Oswald moves for the inclusion of a cognitive-pragmatic component in a theory of argumentation, which has typically refrained from adopting cognitive standpoints (van Eemeren and Grootendorst 2004: 74). Indeed, individuals need to understand arguments before evaluating them, which makes comprehension, to some extent at least, responsible for the selection of premises that will constitute the input to the evaluative stage of argumentation. Accordingly, arguers who want to convince their audience can be said to seek to constrain their interpretative processes so that they only presumably select ‘rhetorically-friendly’ contextual premises. Rhetorical effectiveness is defined as the propensity for an argument to ensure that its conclusion appears to be optimally relevant with respect to its premises. In other words, successful arguments are those whose conclusions (i) are derived with little effort and (ii) generate significant cognitive effects given a series of premises. This definition encompasses both cases of sound argumentation, where potential counter-arguments are weighed and dismissed as epistemically weaker than the argument under consideration, and cases of fallacious argumentation,
21
22
Manuel Padilla Cruz
whose success rests on the audience’s inability to mobilise counter-evidence and discard the argument as fallacious (Maillat and Oswald 2009, 2011; Oswald 2010, 2011). Fallacious moves are therefore seen as attempts to manipulate those two conditions by making ‘rhetorically-friendly’ assumptions epistemically strong and accessible, and ‘rhetorically-unfriendly’ assumptions weak and less accessible. Perlocutionary effects should not be merely understood as consequences of the comprehension process, but as phenomena influencing the way communicated stimuli are actually processed. In “Perlocutionary effects and relevance theory”, Agnieszka Piskorska narrows down the traditional Austinian definition of perlocution and confines it to intentionally evoked mental states, which may have a cognitive or affective nature. Although the notion of ‘perlocution’ was initially alien to relevance theory, there has recently been a growing interest in the speech-act theoretic concept (e.g., Jary 2010; see also Witczak-Plisiecka 2013 for a review). Sharing this interest, Piskorska briefly overviews relevance-theoretic insights into speech acts, such as its reductionism of speech act types or its account of illocutionary force in terms of procedural meaning. Then, she presents arguments supporting her claim. The main one is that on many occasions effects like amusement, warning/ threatening or offending are the reasons why individuals interact. If relevance is assessed solely in terms of positive cognitive effects, something important is missing. Indeed, extant relevance-theoretic analyses of politeness phenomena (Escandell Vidal 1998, 2004) and metaphor (Pilkington 2010) question the view that meaning equals the sum of explicatures and implicatures. Emotion and cognition must be integrated in pragmatic analyses of meaning and its consequences, as argued in (neuro)psychology (e.g., Damasio 1994). Therefore, Piskorska makes two suggestions about the relationship between comprehension and affective states, which are consistent with the massive modularity model of the mind (Sperber 2005). Nevertheless, experiments on the role of perlocutions are needed, although their results could be limited due to the difficulties at modelling actual emotions in experimental settings.
References Aikhenvald, Alexandra. 2004. Evidentiality. Oxford: Oxford University Press. Allot, Nicholas E. 2002. “Relevance and Rationality.” UCL Working Papers in Linguistics 14: 69–82. Apperly, Ian. 2012. Mindreaders. The Cognitive Basis of Theory of Mind. New York: Psychology Press. Berbeira Gardón, José L. 1993. “Posibilidad epistémica, posibilidad radical y pertinencia.” Pragmalingüística 1: 53–78.
Three decades of relevance theory
Berbeira Gardón, José L. 1998. “Relevance and Modality.” Revista Alicantina de Estudios Ingleses 11: 3–22. doi: 10.14198/raei.1998.11.02 Blakemore, Diane. 1987. Semantic Constraints on Relevance. Oxford: Basil Blackwell. Blakemore, Diane. 1988. “’So’ as a Constraint on Relevance.” In Mental Representations. The Interface between Language and Reality, ed. by Ruth M. Kempson, 183–195. Cambridge: Cambridge University Press. Blakemore, Diane. 1989. “Denial and Contrast: A Relevance Theoretic Analysis of But.” Linguistics and Philosophy 12: 15–37. doi: 10.1007/BF00627397 Blakemore, Diane. 1992. Understanding Utterances. Oxford: Blackwell. Blakemore, Diane. 1993. “The Relevance of Reformulations.” Language and Literature 2: 101–120. Blakemore, Diane. 1994. “Relevance, Poetic Effects and Social Goals: A Reply to Culpeper.” Language and Literature 3: 49–59. Blakemore, Diane. 2002. Relevance and Linguistic Meaning. The Semantics and Pragmatics of Discourse Markers. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511486456
Blakemore, Diane. 2011. “On the Descriptive Ineffability of Expressive Meaning.” Journal of Pragmatics 43 (14): 3537–3550. doi: 10.1016/j.pragma.2011.08.003 Blass, Regina. 1989. “Pragmatic Effects of Co-ordination: The Case of ‘and’ in Sissala.” UCL Working Papers in Linguistics 1: 32–51. Blass, Regina. 1990. Relevance Relations in Discourse: A Study with Special Reference to Sissala. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511586293 Bonhomme, Marc. 2005. “Flou et polyvalence de la question rhétorique: L’exemple des fables de La Fontaine.” In Les états de la question, ed. by C. Rossari et al., 191–209. Québec: Nota Bene. Carston, Robyn. 2002. Thoughts and Utterances. The Pragmatics of Explicit Communication. Oxford: Blackwell. doi: 10.1002/9780470754603 Carston, Robyn. 2012. “Word Meaning and Concept Expressed.” The Linguistic Review 29 (4): 607–623. doi: 10.1515/tlr-2012-0022 Carston, Robyn. 2016. “The Heterogeneity of Procedural Meaning.” Lingua 175–176: 154–166. doi: 10.1016/j.lingua.2015.12.010
Carston, Robyn, and Seiji Uchida (eds). 1998. Relevance Theory. Applications and Implications. Amsterdam: John Benjamins. doi: 10.1075/pbns.37 Clark, Billy. 1993. “Let and let’s: Procedural Encoding and Explicature.” Lingua 90: 173–200. doi: 10.1016/0024-3841(93)90066-6
Clark, Billy. 2013. Relevance Theory. Cambridge: Cambridge University Press. doi: 10.1017/CBO9781139034104
Clark, Billy. 2016. “Relevance Theory and Language Change.” Lingua 175–176: 139–153. doi: 10.1016/j.lingua.2015.12.007
Curcó, Carmen. 1995. “Some Observations on the Pragmatics of Humorous Interpretations: A Relevance Theoretic Approach.” UCL Working Papers in Linguistics 7: 27–47. Curcó, Carmen. 1996. “The Implicit Expression of Attitudes, Mutual Manifestness, and Verbal Humour.” UCL Working Papers in Linguistics 8: 89–99. Curcó, Carmen. 1997. The Pragmatics of Humorous Interpretations: A Relevance-Theoretic Approach. PhD diss., University of London. Damasio, Antonio R. 1994. Descartes’ Error. New York: Putnam.
23
24
Manuel Padilla Cruz
Eemeren, Frans van, and Rob Grootendorst. 2004. A Systematic Theory of Argumentation. The Pragma-dialectical Approach. Cambridge: Cambridge University Press. Escandell Vidal, M. Victoria. 1996. “Towards a Cognitive Approach to Politeness.” Language Sciences 18: 629–650. doi: 10.1016/S0388-0001(96)00039-3 Escandell Vidal, M. Victoria. 1998. “Politeness: A Relevant Issue for Relevance Theory.” Revista Alicantina de Estudios Ingleses 11: 45–57. doi: 10.14198/raei.1998.11.05 Escandell Vidal, M. Victoria. 2004. “Norms and Principles. Putting Social and Cognitive Pragmatics Together.” In Current Trends in the Pragmatics of Spanish, ed. by Rosina Márquez- Reiter and M. Elena Placencia, 347-371. Amsterdam: John Benjamins. doi: 10.1075/pbns.123.27esc
Escandell Vidal, M. Victoria, Manuel Leonetti, and Aoife Ahern (eds). 2011. Procedural Meaning. Problems and Perspectives. Bingley: Emerald/Brill. doi: 10.1163/9780857240941 Fodor, Jerry. 1983. The Modularity of Mind. Cambridge, MA: MIT Press. Gibbs, Raymond W. 2000. “Irony in Talk among Friends.” Metaphor and Symbol 15: 5–27. doi: 10.1080/10926488.2000.9678862
Grice, Herbert P. 1957. “Meaning.” Philosophical Review 66: 377–388. doi: 10.2307/2182440 Grice, Herbert P. 1975. “Logic and Conversation.” In Syntax and Semantics vol. 3: Speech Acts, ed. by Peter Cole and Jerry L. Morgan, 41–59. New York: Academic Press. Groefsema, Marjolein. 1995. “Can, May, Must and Should: A Relevance Theoretic Account.” Journal of Linguistics 31: 53–79. doi: 10.1017/S0022226700000566 Gutt, Ernst-August. 1989. “Translation and Relevance.” UCL Working Papers in Linguistics 1: 75–94. Gutt, Ernst-August. 1991. Translation and Relevance. Cognition and Context. Oxford: Basil Blackwell Hall, Alison. 2007. “Do Discourse Connectives Encode Concepts or Procedures?” Lingua 117 (1): 149–174. doi: 10.1016/j.lingua.2005.10.003 Happé, Francesca G. E. 1994. “An Advanced Test of Theory of Mind: Understanding of Story Characters’ Thoughts and Feelings by Able Autistic, Mentally Handicapped, and Normal Children and Adults.” Journal of Autism and Developmental Disorders 24 (2): 129–154. doi: 10.1007/BF02172093
Horn, Larry. 1996. “Presupposition and Implicature.” In The Handbook of Contemporary Semantic Theory, ed. by Shalom Lappin, 299–320. Oxford: Blackwell. Ifantidou, Elly. 1992. “Sentential Adverbs and Relevance.” UCL Working Papers in Linguistics 4: 193–214. Ifantidou, Elly. 1993. “Parentheticals and Relevance.” UCL Working Papers in Linguistics 5: 193–210. Ifantidou, Elly. 2001. Evidentials and Relevance. Amsterdam: John Benjamins. doi: 10.1075/pbns.86
Ifantidou, Elly. 2014. Pragmatic Competence and Relevance. Amsterdam: John Benjamins. doi: 10.1075/pbns.245
Imai, Kunihito. 1998. “Intonation and Relevance.” In Relevance Theory. Applications and Implications, ed. by Robyn Carston and Seiji Uchida, 69–86. Amsterdam: John Benjamins. doi: 10.1075/pbns.37.06ima
Itani, Reiko. 1994. “A Relevance-based Analysis of Hearsay Particles: Japanese Utterance-final tte.” UCL Working Papers in Linguistics 6: 379–400. Iten, Corinne. 2002. Linguistic Meaning, Truth Conditions and Relevance. The Case of Concessives. Basingstoke: Palgrave Macmillan.
Three decades of relevance theory
Jary, Mark. 1998a. “Is Relevance Theory Asocial?” Revista Alicantina de Estudios Ingleses 11: 157–169. doi: 10.14198/raei.1998.11.12 Jary, Mark. 1998b. “Relevance Theory and the Communication of Politeness.” Journal of Pragmatics 30: 1–19. doi: 10.1016/S0378-2166(98)80005-2 Jary, Mark. 2010. Assertion. Basingstoke: Palgrave Macmillan. doi: 10.1057/9780230274617 Jary, Mark. 2013. “Two Types of Implicatures: Material and Behavioural.” Mind & Language 28 (5): 638–660. doi: 10.1111/mila.12037 Jodłowiec, Maria. 1991. The Role of Relevance in the Interpretation of Verbal Jokes: A Pragmatic Analysis. PhD diss., Jagiellonian Uiversity. Jodłowiec, Maria. 2008. “What’s in the Punchline?” In Relevant Worlds: Current Perspectives on Language, Translation and Relevance Theory, ed. by Ewa Wałaszewska, Marta Kisielewska- Krysiuk, Aniela Korzeniowska, and Małgorzata Grzegorzewska, 67–86. Newcastle upon Tyne: Cambridge Scholars Publishing. Jucker, Andreas H. 1993. “The Discourse Marker Well: A Relevance-Theoretical Account.” Journal of Pragmatics 19: 435–452. doi: 10.1016/0378-2166(93)90004-9 Korzeniowska, Aniela, and Małgorzata Grzegorzewska (eds). 2005. Relevance Studies in Poland, Vol. 2. Warsaw: University of Warsaw. Lakoff, Robin T. 1973. “The Logic of Politeness; or, Minding Your P’s and q’s.” In Papers from the Ninth Regional Meeting, ed. by Claudia W. Corum, Thomas C. Smith-Stark, and Ann Weiser, 292–305. Chicago: Chicago Linguistic Society. Leech, Geoffrey. 1983. Principles of Pragmatics. London: Longman. Leekam, Susan R. 1991. “Jokes and Lies: Children’s Understanding of Intentional Falsehood.” In Natural Theories of Mind. Evolution, Development and Simulation of Everyday Mindreading, ed. by Andrew Whiten, 159–174. Oxford: Basil Blackwell. Leinonen, Eeva, and Nuala Ryder. 2008. “Relevance Theory and Language Disorders.” In The Handbook of Clinical Linguistics, ed. by Martin J. Ball, Michael Perkins, Nicole Müller, and Sara Howard. Oxford: Blackwell. Levinson, Stephen C. 2000. Presumptive Meanings. The Theory of Generalized Conversational Implicature. Cambridge, MA: MIT Press. Lewis, David. 1969. Convention. Cambridge, MA.: Harvard University Press. Maillat, Didier, and Steve Oswald. 2009. “Defining Manipulative Discourse: The Pragmatics of Cognitive Illusions.” International Review of Pragmatics 1 (2): 348–370. doi: 10.1163/187730909X12535267111651
Maillat, Didier, and Steve Oswald. 2011. “Constraining Context: A Pragmatic Account of Cognitive Manipulation.” In Critical Discourse Studies in Context and Cognition, ed. by C. Hart, 65–80. Amsterdam: John Benjamins. doi: 10.1075/dapsac.43.04mai Mascaro, Olivier, and Dan Sperber. 2009. “The Moral, Epistemic, and Mindreading Components of Children’s Vigilance towards Deception.” Cognition 112 (3): 367–380. doi: 10.1016/j.cognition.2009.05.012
Mateo Martínez, José, and Francisco Yus Ramos. 1998. Special Issue on Relevance Theory. Revista Alicantina de Estudios Ingleses 11. Mazzarella, Diana. 2013. “‘Optimal Relevance’ as a Pragmatic Criterion: The Role of Epistemic Vigilance.” UCL Working Papers in Linguistics 25: 20–45. Mazzarella, Diana. 2015. “Pragmatics and Epistemic Vigilance: The Deployment of Sophisticated Interpretative Strategies.” Croatian Journal of Philosophy 15 (44): 183–199. Mioduszewska, Ewa (ed.). 2004. Relevance Studies in Poland, Vol. 1. Warszawa: University of Warsaw.
25
26 Manuel Padilla Cruz
Moeschler, Jacques. 1993. “Relevance and Conversation.” Lingua 90: 149–171. doi: 10.1016/0024-3841(93)90065-5
Moeschler, Jacques. 2009. “Pragmatics, Propositional and Non-propositional Effects: Can a Theory of Utterance Interpretation Account for Emotions in Verbal Communication?” Social Science Information 48 (3): 447–464. doi: 10.1177/0539018409106200 Moescheler, Jacques. 2016. “Where Is Procedural Meaning Located? Evidence from Discourse Connectives and Tenses.” Lingua 175–176: 122–138. doi: 10.1016/j.lingua.2015.11.006 Noh, Eun-Ju. 2000. Metarepresentation. A Relevance-Theory Approach. Amsterdam: John Benjamins. doi: 10.1075/pbns.69 Oswald, Steve. 2010. Pragmatics of Uncooperative and Manipulative Communication. PhD diss., University of Neuchâtel. Oswald, Steve. 2011. “From Interpretation to Consent: Arguments, Beliefs and Meaning.” Discourse Studies 13 (6): 806–814. doi: 10.1177/1461445611421360e Padilla Cruz, Manuel. 2003. “A Relevance Theoretic Approach to the Introduction of Scandinavian Pronouns in English.” In Interaction and Cognition in Linguistics, ed. by Carlos Inchaurralde and Celia Florén, 123–134. Hamburg: Peter Lang. Padilla Cruz, Manuel. 2005a. “On the Phatic Interpretation of Utterances: A Complementary Relevance-theoretic Approach.” Revista Alicantina de Estudios Ingleses 18: 227–246. doi: 10.14198/raei.2005.18.11
Padilla Cruz, Manuel. 2005b. “Relevance Theory and Historical Linguistics: Towards a Pragmatic Approach to the Morphological Changes in the Preterite from Old English to Middle English.” Revista Canaria de Estudios Ingleses 51: 183–204. Padilla Cruz, Manuel. 2007a. “Phatic Utterances and the Communication of Social Information.” In Studies in Intercultural, Cognitive and Social Pragmatics, ed. by Pilar Garcés- Conejos Blitvich, Manuel Padilla Cruz, Reyes Gómez Morón, and Lucía Fernández Amaya, 114–131. New Castle upon Tyne: Cambridge Scholars Publishing. Padilla Cruz, Manuel. 2007b. “Metarepresentations and Phatic Utterances: A Pragmatic Proposal about the Generation of Solidarity between Interlocutors.” In Current Trends in Pragmatics, ed. by Piotr Cap and Joanna Nijakowska, 110–128. New Castle upon Tyne: Cambridge Scholars Publishing. Padilla Cruz, Manuel. 2012. “Epistemic Vigilance, Cautious Optimism and Sophisticated Understanding.” Research in Language 10 (4): 365–386. doi: 10.2478/v10015-011-0040-y Padilla Cruz, Manuel. 2013a. “Understanding and Overcoming Pragmatic Failure in Intercultural Communication: From Focus on Speakers to Focus on Hearers.” International Review of Applied Linguistics in Language Teaching 51 (1): 23–54. Padilla Cruz, Manuel. 2013b. “Metapsychological Awareness of Comprehension and Epistemic Vigilance of L2 Communication in Interlanguage Pragmatic Development.” Journal of Pragmatics 59 (A): 117–135. doi: 10.1016/j.pragma.2013.09.005 Padilla Cruz, Manuel. 2014. “Pragmatic Failure, Epistemic Injustice and Epistemic Vigilance.” Language & Communication 39: 34–50. doi: 10.1016/j.langcom.2014.08.002 Papp, Szilvia. 2006. “A Relevance-theoretic Account of the Development and Deficits of Theory of Mind in Normally Developing Children and Individuals with Autism.” Theory & Psychology 16 (2): 141–161. doi: 10.1177/0959354306062532 Pilkington, Adrian. 1991. “Poetic Effects: A Relevance Theory Perspective.” In Literary Pragmatics, ed. by Roger D. Sell, 44–61. London: Routledge. Pilkington, Adrian. 1992. “Poetic Effects.” Lingua 87: 29–51. doi: 10.1016/0024-3841(92)90024-D
Three decades of relevance theory
Pilkington, Adrian. 2000. Poetic Effects. A Relevance Theory Perspective. Amsterdam: John Benjamins. doi: 10.1075/pbns.75 Pilkington, Adrian. 2010. “Metaphor Comprehension: Some Questions for Current Accounts in Relevance Theory.” In Explicit Communication: Robyn Carston’s Pragmatics, ed. by Esther Romero and Belén Soria, 156–171. Basingstoke: Palgrave-Macmillan. Piskorska, Agnieszka. 2012a. “Cognition and Emotions – A Joint Effort at Obtaining Positive Cognitive Effects?” In Relevance Studies in Poland. Vol. 4. Essays on Language and Communication, ed. by Agnieszka Piskorska, 102–111. Warsaw: WUW. Piskorska, Agnieszka (ed.). 2012b. Relevance Studies in Poland. Vol. 4. Essays on Language and Communication. Warsaw: WUW. Rouchota, Villy. 1994. “On Indefinite Descriptions.” Journal of Linguistics 30: 441–475. doi: 10.1017/S0022226700016716
Rouchota, Villy. 1995. “Discourse Connectives: What Do They Link?” UCL Working Papers in Linguistics 7: 199–212. Rouchota, Villy, and Andreas H. Jucker (eds). 1998. Current Issues in Relevance Theory. Amsterdam: John Benjamins. doi: 10.1075/pbns.58 Ruiz Moneva, María A. 1997. “A Relevance-theory Approach to the Scandinavian Influence upon the Development of the English Language.” Revista Alicantina de Estudios Ingleses 10: 183–191. doi: 10.14198/raei.1997.10.13 Sasamoto, Ryoko, and Deirdre Wilson (eds). 2016. “Little Words: Communication and Procedural Meaning.” Lingua 175–176: 1–166. doi: 10.1016/j.lingua.2015.12.009 Schiffer, Stephen. 1972. Meaning. Oxford: Claredon Press. Scott, Kate. 2016. “Pronouns and Procedures: Reference and Beyond.” Lingua 175–176: 69–82. doi: 10.1016/j.lingua.2015.07.005
Smith, Neil V., and Deirdre Wilson. 1992. Special Issue on Relevance Theory. Lingua 87. doi: 10.1016/0024-3841(92)90022-B
Solska, Agnieszka. 2012. “Relevance-theoretic Comprehension Procedure and Processing Multiple Meanings in Paradigmatic Puns.” In Relevance Theory. More than Understanding, ed. by Ewa Wałaszewska and Agnieszka Piskorska, 167-182. New Castle upon Tyne: Cambridge Scholars Publishing. Soria, Belén, and Esther Romero (eds). 2010. Explicit Communication. Robyn Carston’s Pragmatics. Basingstoke: Palgrave Macmillan. Sperber, Dan. 1994. “Understanding Verbal Understanding.” In What is Intelligence? ed. by Jean Khalifa, 179–198. Cambridge: Cambridge University Press. Sperber, D. 2005. “Modularity and Relevance: How Can a Massively Modular Mind Be Flexible and Context-sensitive?” In The Innate Mind: Structure and Content, ed. by Peter Carruthers, Stephen Laurence, and Stephen Stich, 53–68. Oxford: Oxford University Press. doi: 10.1093/acprof:oso/9780195179675.003.0004
Sperber, Dan, and Deirdre Wilson. 1986. Relevance. Communication and Cognition. Oxford: Blackwell. Sperber, Dan, and Deirdre Wilson. 1987. “Précis of Relevance: Communication and Cognition.”and “Authors’ Response.” Behavioural and Brain Sciences 10: 697–710, 736–754. doi: 10.1017/S0140525X00055345
Sperber, Dan, and Deirdre Wilson. 1995. Relevance. Communication and Cognition, 2nd edition. Oxford: Blackwell. Sperber, Dan, Fabrice Clément, Christophe Heintz, Olivier Mascaro, Hugo Mercier, Gloria Origgi, and Deirdre Wilson. 2010. “Epistemic vigilance.” Mind and Language 25 (4): 359–393. doi: 10.1111/j.1468-0017.2010.01394.x
27
28
Manuel Padilla Cruz
Sullivan, Kate, Ellen Winner, and Natalie Hopfield. 1995. “How Children Tell a Lie from a Joke: The Role of Second-order Mental State Attributions.” British Journal of Developmental Psychology 13 (2): 191–204. doi: 10.1111/j.2044-835X.1995.tb00673.x Sullivan, Kate, Ellen Winner, and Helen Tager-Flugsber. 2003. “Can Adolescents with Williams Syndrome Tell the Difference between Lies and Jokes?” Developmental Neuropsychology 23 (1–2): 85–103. doi: 10.1080/87565641.2003.9651888 Unger, Christoph. 2006. Genre, Relevance and Global Coherence. The Pragmatics of Discourse Type. Basingstoke: Palgrave. doi: 10.1057/9780230288201 Vega Moreno, Rosa E. 2007. Creativity and Convention. The Pragmatics of Everyday Figurative Speech. Amsterdam: John Benjamins. doi: 10.1075/pbns.156 Wałaszewska, Ewa. 2015. Relevance-theoretic Lexical Pragmatics. Newcastle upon Tyne: Cambridge Scholars Publishing. Wałaszewska, Ewa, Marta Kisielewska-Krysiuk, and Agnieszka Piskorska (eds). 2010. In the Mind and across Minds. A Relevance-Theoretic Perspective on Communication and Translation. Newcastle upon Tyne: Cambridge Scholars Publishing. Wałaszewska, Ewa, and Agnieszka Piskorska (eds). 2012. Relevance Theory. More than Understanding. Newcastle upon Tyne: Cambridge Scholars Publishing. Wałaszewska, Ewa, and Agnieszka Piskorska (eds). In press. From Discourse to Morphemes. Applications of Relevance Theory. Newcastle upon Tyne: Cambridge Scholars Publishing. Wearing, Catherine. 2010. “Autism, Metaphor and Relevance Theory.” Mind and Language 25 (2): 196–216. doi: 10.1111/j.1468-0017.2009.01386.x Wharton, Tim. 2009. Pragmatics and Non-Verbal Communication. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511635649 Wharton, Tim. 2016. “That Bloody So-and-so Has Retired: Expressives Revisited.” Lingua 175– 176: 20–35. doi: 10.1016/j.lingua.2015.08.004 Wilson, Deirdre. 1999. “Metarepresentation in Linguistic Communication.” UCL Working Papers in Linguistics 11: 127–161. Wilson, Deirdre. 2010. “Understanding and Believing.” Plenary talk delivered at the 4th International Conference Intercultural Pragmatics and Communication, Madrid 15–17 November. Wilson, Deirdre. 2011. “The Conceptual-procedural Distinction: Past, Present and Future.” In Procedural Meaning: Problems and Perspectives, ed. by M. Victoria Escandell Vidal, Manuel Leonetti, and Aoife Ahern, 3–31. Bingley: Emerald. doi: 10.1108/S1472-7870(2011)0000025005
Wilson, Deirdre. 2013. “Irony Comprehension: A Developmental Perspective.” Journal of Pragmatics 59 (A): 40–56. doi: 10.1016/j.pragma.2012.09.016 Wilson, Deirdre. 2016. “Reassessing the Conceptual-Procedural Distinction.” Lingua 175–176: 5–19. doi: 10.1016/j.lingua.2015.12.005 Wilson, Deirdre, and Robyn Carston. 2007. “A Unitary Approach to Lexical Pragmatics: Relevance, Inference and Ad Hoc Concepts.” In Pragmatics, ed. by Noel Burton-Roberts, 230– 259. Basingstoke: Palgrave. doi: 10.1093/oxfordhb/9780199234769.003.0018 Wilson, Deirdre, and Neil V. Smith. 1993. Special issue on relevance theory. Lingua 90. Wilson, Deirdre, and Dan Sperber. 1988. “Mood and the Analysis of Non-declarative Sentences.” In Human Agency: Language, Duty and Value, ed. by Jonathan Dancy, J. M. E. Moravcsik, and C. C. W. Taylor, 77–101. Stanford: Stanford University Press. Wilson, Deirdre, and Dan Sperber. 1991a. “Inference and Implicature.” In Pragmatics: A Reader, ed. by Steven Davis, 377–393. Oxford: Oxford University Press.
Three decades of relevance theory
Wilson, Deirdre, and Dan Sperber. 1991b. “Pragmatics and Modularity.” In Pragmatics: A Reader, ed. by Steven Davis, 583-595. Oxford: Oxford University Press. Wilson, Deirdre, and Dan Sperber. 1993. “Linguistic Form and Relevance.” Lingua 90: 1–26. doi: 10.1016/0024-3841(93)90058-5
Wilson, Deirdre, and Dan Sperber. 2002. “Relevance Theory.” UCL Working Papers in Linguistics 14: 249–287. Wilson, Deirdre, and Dan Sperber. 2004. “Relevance Theory.” In The Handbook of Pragmatics, ed. by Larry Horn and Gregory Ward, 607–632. Oxford: Blackwell. Wilson, Deirdre, and Dan Sperber. 2012. Meaning and Relevance. Cambridge: Cambridge University Press. doi: 10.1017/CBO9781139028370 Wilson, Deirdre, and Tim Wharton. 2006. “Relevance and Prosody.” Journal of Pragmatics 38: 1559–1579. doi: 10.1016/j.pragma.2005.04.012 Witczak-Plisiecka, Iwona. 2013. From Speech Acts to Speech Actions. Lodz: Lodz University Press. Yus Ramos, Francisco. 1995. “La significación social de las máximas de Grice: el caso del cómic alternativo inglés.” Revista Canaria de Estudios Ingleses 30–31: 109–128. Yus Ramos, Francisco. 1997. Cooperación y relevancia. Dos aproximaciones pragmáticas a la interpretación. Alicante: Publicaciones de la Universidad de Alicante. Yus Ramos, Francisco. 1998. “A Decade of Relevance Theory.” Journal of Pragmatics 30: 305–345. doi: 10.1016/S0378-2166(98)00015-0
Yus Ramos, Francisco. 1999a. “Towards a Pragmatic Taxonomy of Misunderstandings.” Revista Canaria de Estudios Ingleses 38: 217–239. Yus Ramos, Francisco. 1999b. “Misunderstandings and Explicit/Implicit Communication.” Pragmatics 9: 487–517. doi: 10.1075/prag.9.4.01yus Yus Ramos, Francisco. 2001. Ciberpragmática. El uso del lenguaje en Internet. Barcelona: Ariel. Yus Ramos, Francisco. 2003. “Humor and the Search for Relevance.” Journal of Pragmatics 35: 1295–1331. doi: 10.1016/S0378-2166(02)00179-0 Yus Ramos, Francisco. 2008. “A Relevance-theoretic Classification of Jokes.” Lodz Papers in Pragmatics 4: 131–157. Yus Ramos, Francisco. 2010. Ciberpragmática 2.0. Nuevos usos del lenguaje en Internet. Barcelona: Ariel. Yus Ramos, Francisco. 2011. Cyberpragmatics. Internet-Mediated Communication in Context. Amsterdam: John Benjamins. doi: 10.1075/pbns.213 Yus Ramos, Francisco. 2013. “An Inference-centered Analysis of Jokes: The Intersecting Circles Model of Humorous Communication.” In Irony and Humor: Highlights and Genres, ed. by Leonor Ruiz Gurillo and Beatriz Alvarado, 59–82. Amsterdam: John Benjamins. doi: 10.1075/pbns.231.05yus
Yus Ramos, Francisco. 2016. Humour and Relevance. Amsterdam: John Benjamins. doi: 10.1075/thr.4
Žegarac, Vladimir. 1998. “What is Phatic Communication?” In Current Issues in Relevance Theory, ed. by Villy Rouchota and Andreas H. Jucker, 327–361. Amsterdam: John Benjamins. doi: 10.1075/pbns.58.14zeg
Žegarac, Vladimir, and Billy Clark. 1999a. “Phatic Interpretations and Phatic Communication.” Journal of Linguistics 35: 321–346. doi: 10.1017/S0022226799007628 Žegarac, Vladimir, and Billy Clark. 1999b. “Phatic Communication and Relevance Theory: A Reply to Ward & Horn.” Journal of Linguistics 35: 565–577. doi: 10.1017/S0022226799007707
29
Part I
Issues on procedural meaning and procedural analyses
The speaker’s derivational intention Thorstein Fretheim
Norwegian University of Science and Technology
Adopting the basic tenets of relevance theory, Powell (2010) introduces the concept of derivational intention as something separate from a speaker’s informative intention. The derivational intention of a speaker is an intention concerning the pragmatically inferred route that the hearer should take in order to recognize the speaker’s informative intention. This paper addresses what can happen when a speaker’s derivational intention is at odds with a particular piece of procedural information encoded by some linguistic expression, with the conceptual semantics of a lexical item, or with the procedural meaning of an intonation pattern employed by the speaker. The procedural meaning of one expression may override that of a co-occurring expression when there is a conflict between them. Keywords: intonation, procedural meaning/information, informative intention, derivational intention, focus
1. Introduction 1.1
The concept of “derivational intention”
Powell (2010) observes that, in addition to Sperber and Wilson’s concept of informative intention (Sperber and Wilson 1986/1995), “a speaker will have intentions concerning the route via which her audience should reach the content of her informative intention” (Powell 2010: 29). This he refers to as the speaker’s derivational intention. Powell does not develop his concept of derivational intention much, but he notes that there may be cases “[…] in which the speaker’s derivational intention does not, in fact, lead to the content of her informative intention” (Powell 2010: 30), and then he adds that, “[…] intuitions may be weighted towards either of the intentions, thus allowing for the possibility of two distinct intuitions concerning what is said by one and the same utterance […]” (Powell 2010: 30). doi 10.1075/pbns.268.02fre © 2016 John Benjamins Publishing Company
34
Thorstein Fretheim
My feeling is that the concept of derivational intention may be an important addition to the relevance theorist’s toolbox, but I fail to see how the derivational intention behind a speaker’s utterance can be in conflict with her informative intention. (A possible exception might be when a speaker erroneously believes that her description is satisfied by the intended referent.) For me, the speaker’s derivational intention is for the hearer to pursue an inferential route that is consonant with those procedural constraints that enable him to work out the speaker’s informative intention. However, I shall argue that a piece of linguistically encoded procedural information in the relevance-theoretic sense (Blakemore 1987, 2002; Wilson and Sperber 1993) may occasionally contradict a speaker’s derivational intention. A hearer has successfully recognized the speaker’s derivational intention if, and only if, the inferential route pursued leads to recognition of the informative intention. On the other hand, the hearer may identify the speaker’s informative intention even if the encoded procedural meaning of one particular linguistic expression points in a different direction, to an inferential path that fails to lead to the result intended by the speaker. It is always a combination of conceptual and procedural information that constrains the hearer’s pragmatic search for the speaker’s derivational intention and informative intention. The hearer’s recognition of the speaker’s derivational intention is based on a fused impression of how conceptual information and procedural information together constrain the relevance of a given utterance. It is my objective to show that the encoded procedural meaning of a linguistic expression is fallible, just as the encoded conceptual meaning of a linguistic expression is fallible because taking it at face value in the pragmatic phase of utterance interpretation will sometimes lead to a wrong interpretation. 1.2 Mismatch between derivational intention and encoded procedural information The following is the opening sentence of the Norwegian novel De Udødelige (‘The immortal ones’) by the author Ketil Bjørnstad: (1) Tanken kom først lenge etter at hun hadde sovnet thought.DEF came first long after that she had slept.INCHO ‘The thought occurred [to NN] only long after she had fallen asleep’
The reader does not know who the feminine pronoun hun (‘she’) refers to, or who the thought belongs to. Tanken (‘the thought’) has a conceptual meaning that restricts the denotatum to something identifiable as a thought, and the suffixed definiteness marker n (tanke-n) encodes the procedural information that
The speaker’s derivational intention
the reader should be able to associate the referring expression with a determinable thought, though the reader cannot access any such thought when reading (1). For pragmatic reasons one would be inclined to infer that the thought does not belong to the woman represented by the personal pronoun hun. This conclusion depends partly on the lexical meaning of the inchoative verb sovne (‘fall asleep’) in (1) and partly on an implicature derived from this opening sentence, an inferential process that requires access to the conceptual meaning of the verb sovne. Although people who remember thoughts that occurred to them in their dreams do exist, most of us do not normally form thoughts in our sleep that we are able to re-activate and remember accurately afterwards. Despite formal linguistic appearances, the thought in (1) is unlikely to be the thought of the female referent of the pronoun hun, a thought that occurred to her while she was asleep. The next sentence, here presented as (2), supports the assumption that the thought does not belong to the referent of the pronoun hun in (1). (2) Da hadde han ligget og vridd seg i minst en time then had he lain and twisted REFL in least one hour ‘Then he had been lying turning from side to side for at least an hour’
The indexical da (‘then’) in (2) must refer to the point in time when the thought mentioned in (1) was formed. (2) tells the reader that the bedfellow of the woman is a man, and it invites the reader to infer that the referent of the masculine pronoun han (‘he’) is the one who formed the as yet unidentifiable thought, and that the thought came to this man a long time after his female companion had fallen asleep. A personal pronoun like hun (‘she’) or han (‘he’) encodes two pieces of procedural information: (i) the reader should seek out a candidate referent of the female or male gender, respectively (some would say that “female” is the conceptual meaning encoded by hun, and that “male” is the conceptual meaning encoded by han), and (ii) this referent is already in the hearer’s focus of attention. The next paragraph of the opening page of the novel reveals that the man and the woman lying side by side in bed are a married couple. In the course of the next three pages, the reader receives a substantial amount of background information about this couple, as well as information that shows how the first word tanken (‘the thought’) in (1) should be linked to a possessive modifier so that a unique proposition for sentence (1) can be derived through backtracking. Pronouns in opening sentences of novels and short stories would appear to be problematic for the Givenness Hierarchy (GH) theory about the cognitive status of discourse referents proposed by Jeanette Gundel and her collaborators (in particular, Gundel, Hedberg and Zacharski 1993). This theory postulates six universal, implicationally related cognitive statuses of referents associated with different types of nominal referring expressions. These statuses are, from the highest (most
35
36
Thorstein Fretheim
constraining) to the lowest (least constraining): In Focus, Activated, Familiar, Uniquely Identifiable, Referential and Type Identifiable. A given cognitive status is entailed by any cognitive status that is higher in the hierarchy, so that whenever a referent is in focus of attention for speaker and hearer, giving it the highest possible cognitive status In Focus, it is by necessity also Activated, Familiar, Uniquely Identifiable, and so on. In certain papers written in the wake of the cutting-edge article from 1993, Gundel calls the encoded information about the relationship between a referring expression and its intended referent’s cognitive status “procedural information.” She writes: “GH statuses are mental states (not linguistic entities); the linguistic forms that encode these statuses provide procedural information, in the sense of relevance theory […] about how to access (a mental representation of) the referent […]” (Gundel 2011: 207). It is true of all tokens of unaccented anaphoric pronouns in discourse and many tokens of accented ones that access to a mental representation of the intended referent depends on information that is “in focus of attention” for speaker and hearer at the time of utterance (Gundel 2011: 208), while resolution of the reference of a definite descriptive noun phrase depends on a combination of the conceptual semantics that the phrase itself encodes plus easily accessible contextual information. In (1)–(2) there exists no referent which is In Focus, nor one that is Activated, so that it would license use of a personal pronoun. The personal pronouns in (1) and (2) and the description tanken (‘the thought’) in (1) fail to satisfy the procedural constraints encoded by a pronoun or a definiteness marker. The reader has to read on in order to get a chance to resolve the reference of these referring expressions. There is no antecedent phrase that helps to determine the referent of the two pronouns hun (‘she’) and han (‘he’), so they are not anaphoric. This is also no example of cataphora in the traditional syntactic or discourse-structural sense, because descriptive information that will cause the reader to link the personal pronouns to unique individuals within the author’s fictional universe is presented in scattered places several paragraphs later. It would be wrong to conclude that the GH approach to the procedural semantics of pronouns, demonstratives and definiteness markers is in need of revision simply because the pronouns in (1)–(2) and the definite description in (1) do not offer the reader sufficient information to identify the entities that these expressions refer to. I would like to propose that, for pragmatic reasons, a reader of the sequence of (1)–(2) will understand that the derivational intention of the author is at odds with the procedural semantics of the pronouns and the marker of definiteness, and that there is an asymmetry between the author’s own knowledge of what the referring expressions represent and the knowledge that he has decided to share with the reader in the novel’s opening paragraph. The author’s derivational intention in (1) and (2) includes the assumption that readers suspend
The speaker’s derivational intention
their expectation that the pronouns refer to a female gender entity and a male gender entity of which they are able to form a unique mental representation at the point in the text where they occur. Similarly, the author’s derivational intention includes the assumption that readers should suspend their expectation that the definite description tanken (‘the thought’) in (1) refers to a uniquely identifiable thought. However, the pronouns and the definiteness marker encode the same procedural information here as elsewhere. There is a conflict between the procedural information that the personal pronoun and the definite description in (1) offer the reader and the author’s derivational intention that readers defer their reference resolution task until a bit later in the reading process. By using referring expressions in the opening lines of the novel that encode a higher cognitive status for their referents than what the reader’s knowledge permits, the author has utilized a well-known literary technique which helps to resolve the apparent conflict between the author’s choice of referring expressions and the author’s derivational intention. We can account for and justify the referring expressions in (1)–(2) by invoking a specific pragmatic dictum familiar from relevance theory (Sperber and Wilson 1986/1995), namely that the linguistic semantics of a verbal stimulus underdetermines the speaker’s informative intention, hence, I would add, the speaker’s derivational intention. The linguistic semantics encoded by a sentence includes strictly procedural information as well as concepts contributing to the proposition expressed. It is an accepted truth within relevance theory that encoded conceptual meaning should be processed with caution in the pragmatic phase of utterance comprehension and that it is frequently in need of context-sensitive adjustment (e.g. Wilson and Carston 2007; Carston 2012), but encoded procedural information should be treated with caution as well. Successful processing of the referring expressions in (1)–(2) demands the reader’s awareness that reference resolution will be possible only later, when the author has given the reader more contextual information about the main characters in the novel. An anonymous reviewer remarked that data of the type presented in (1)–(2) is not typical of everyday communication. I agree. One could even argue that the missing antecedents of the tokens of the personal pronouns hun and han in (1)–(2) justify the judgment that a rule of grammar is violated. Due to an impoverished linguistic input, the explicature of (1) cannot be determined until we read on and access supplementary information. Readers of the opening sentence (1) who possess basic knowledge of certain genre conventions are not likely to be frustrated because (1) fails to supply information sufficient to determine the author’s informative intention. They will typically suspend their expectation of a relevant linguistic input, because they acknowledge that here the procedural meanings encoded by a personal pronoun and by a definite description are of no
37
38
Thorstein Fretheim
help in the pragmatic processes of reference resolution. They will therefore infer that the author intends them to be patient and watch out for pertinent information in forthcoming sentences, information that enables them to fill any truth- conditional gaps that have so far hampered comprehension of the opening lines of the novel. 1.3 Mismatch between derivational intention and encoded conceptual information A piece of encoded conceptual information may fail to direct the hearer to the speaker’s derivational intention. In such cases it is likely to be ignored, because what it instructs the hearer to do in the pragmatic processing phase is felt to be counterproductive, leading to an output that would make the utterance irrelevant. One very good example of this is the following English text appearing at the back of all passenger seats in aircraft cabins all over the world: (3) FASTEN SEAT BELT WHILE SEATED
No rational mind can possibly misunderstand this advice to the passengers. They are encouraged not to unclasp their seat belts while they are seated, but this is not what the lexically inchoative verb fasten appears to instruct the passengers to do. A literal reading of (3) is irrelevant no matter whether it is while seated or fasten seat belt that is identified as the focus of information. A linguistic form that captures the communicator’s informative intention through a relevant indication of the communicator’s derivational intention is seen in the alternative instruction in (4), which I have never observed on board any aircraft: (4) KEEP SEAT BELT FASTENED WHILE SEATED
Above all seat rows there are seat belt icons that light up whenever passengers should fasten their seat belt. This is a sign to keep it fastened as long as the icon remains lit. The temporal adjunct while seated in (3) shows that the written imperative Fasten seat belt while seated has a different function than the seat belt icon. The adjunct reveals that the derivational intention is for a passenger to interpret (3) as if the text had been (4). While the conceptual semantics of the verb fasten in (3) encodes a change-of-state from not fastened to fastened, all available contextual information causes passengers to realize that this aspect of the lexical meaning of fasten should not be entered into their mental representation of the written instruction in (3).
The speaker’s derivational intention
The linguistic form in (3) may have been chosen at the expense of the more accurate form in (4) because of its relative brevity. (5), appearing inside Air France cabins, is the French version corresponding to (3), or rather to (4): (5) GARDEZ VOTRE CEINTURE ATTACHÉE
In (5) there is no mismatch between encoded linguistic meaning and communicated meaning. Gardez … attaché corresponds to keep … fastened, a stative rather than an inchoative predicate, and the linguistic form of (5) also makes it superfluous to add a temporal adjunct, like pendant que vous êtes assis (‘while you are seated’) (p.c. Astrid Nome). 2. Procedural indicators differ in strength 2.1
The procedural meaning of intonation as a cue to identification of the constituent targeted by a focus particle
An additive focus particle like English also gives the hearer the procedural information that, when there is a pitch accent on the particle itself, the targeted focus is a preceding phrase, not a following one. (6) illustrates this. A pitch accent assigned to also causes the hearer to identify the focus constituent as the preceding subject NP Florence, not the following VP: (6) John says that Mary is someone you can trust, and Florence has ALSO said some very nice things about her
The focal accentual handling of also indicated by the caps in (6) is acceptable. Uttering (6) implicates that calling someone trustworthy is to say something nice about that person. Both John and Florence have said something nice about Mary. That the phrase said some very nice things about her expresses discourse-given information that is true of someone other than Florence is signaled by the lack of a primary pitch accent on all word forms subsequent to the particle also. The prosody of the second conjunct in (7), however, indicates that Florence has said something nice about Mary in addition to her sharing John’s view that Mary is a trustworthy person. Due to also and the primary pitch accent on things, (7) implicates that saying you trust someone does not qualify as saying something nice about that person. (7) John says that Mary is someone you can trust, and Florence has also said some very nice THINGS about her
39
40 Thorstein Fretheim
There are two focal phrase-tones (see the pitch contours in Section 2.2) in the Norwegian utterance represented by the labeled bracketing notation in (8) below. This tone is instantiated as a mandatory pitch peak in the final syllable of a Focal Phrase (FP), the pragmatically most important constituent in Norwegian intonational phonology and an immediate constituent of the topmost prosodic node, the Intonation Unit (IU) (e.g. Fretheim 1978, 1981, 1992). The peak at the end of an FP gives special prominence to accented phrases within that FP, prior to the peak. In (8) the location of the right edge of the first of the two FPs determines whether også modifies the earlier subject or the later predicate nominal: (8) [[´han er ´OGså]FP [´GOLFspiller]FP ]IU he is also golfplayer ‘He is also a golfer’/‘He too is a golfer’
With reference to similar data, Fretheim (1978) argued that the speaker’s intonation indicates that the focus constituent targeted by the particle også (‘also’) is located in the same FP as the particle. In (8) the subject pronoun han and the particle are in the same FP, and the utterance-medial FP boundary serves as a barrier that blocks association of også with the predicate in the next FP. By varying the position of the boundary between two FPs, a Norwegian speaker can indicate which sentence constituent is the information focus targeted by the focus particle. Since også and the predicate golfspiller belong to the same FP in (9), the hearer is here instructed to identify the predicate as the focus constituent. Golfspiller was presented as an activated concept in (8), despite its prosodic prominence, but it stands for a discourse-new, rhematic concept in (9), where the subject is a contrastive topic: (9) [[´HAN er]FP [´også ´GOLFspiller]FP ]IU he is also golfplayer ‘He is even a golfer’/‘He is a golfer too’
If the pronominal subject han is unaccented but the FP boundary is located between også and golfspiller as in (8), then the interpretation will be same as for (9), in spite of the speaker’s intonational phrasing. The syntactic constituent targeted by the focus particle requires a word-accent but does not need a focal phrase-tone appearing at the right edge of an FP. It is sometimes necessary for a hearer to disregard a particular piece of procedural information because common sense, context-dependent inference and/or a co-occurring but contrary procedural device shows that adhering to it in the process of pragmatic interpretation leads to an impasse. The procedural information encoded by the Norwegian focus particle også and the FP structures in (8) and
The speaker’s derivational intention
(9) directly mirrors the speaker’s derivational intention. However, the lexically encoded procedural meaning of an additive focus particle may fail to guide the hearer to the speaker’s derivational intention because the procedural information derived from the global FP structure points to a different derivational intention. Let us turn to another Norwegian additive focus particle, og (‘too’), for an illustration of this kind of conflict between co-occurring linguistic indicators. Like the English focus particle too, its Norwegian counterpart og occupies a position after the focus constituent that it modifies, and not necessarily adjacent to it. There are, however, certain differences between the impact of Norwegian og and the impact of English too on pragmatic interpretation, which I attribute to the procedural meanings of different types of intonational phrasing in spoken Norwegian, compared to the fact that no comparable prosodic phenomenon exists in spoken English. The encoded information that the focus phrase precedes the focus particle og points to the subject han in (10) as the focus, but the prosodic inclusion of og and the accented predicate golfspiller within the same FP gives the hearer the procedural information that the particle targets the following predicate: (10) [[´HAN er]FP [´og ´GOLFspiller]FP ]IU he is too golfplayer ‘He is even a golfer’
The procedural meaning of the FP structure in (10) overrides the procedural meaning associated with the lexical item og, i.e. the instruction that og connects with an already produced utterance chunk. By respecting the procedural information inherent in the FP structure in (10) but ignoring the procedural information that the particle og selects an earlier phrase as the focus of information, the hearer is able to infer that the speaker’s derivational intention is for the hearer to associate the focus particle with the later predicate, just as in the utterance of (9) where the focus particle is også, an item that has more in common with English also than with English too. Christoph Unger (p.c.) has suggested that, when two linguistic expressions encode mutually incompatible procedural meanings and one of them is lexical and the other prosodic, the effect of the latter is presumably more likely to be overridden by the former than the other way around. I suspect that this may frequently be true but (10) illustrates that it is also possible to use an intonation pattern to set aside what would otherwise be a pragmatic effect of procedural information derived from a function word.
41
42
Thorstein Fretheim
2.2 Parenthetical and non-parenthetical uses of verbs of propositional attitude A number of verbs or adjectival phrases in English and other languages acquire a truth-conditional function in some contexts and a non-truth-conditional parenthetical function in other contexts. I can use an utterance of the sentence in (11) to inform you of my fear that a miscalculation may have been made, or I can use it to inform you that there has been a miscalculation and that I deplore this: (11) I’m afraid it’s a miscalculation
When the predicate afraid is truth-conditional, an utterance of (11) communicates the speaker’s attitude of uneasiness directed at a discourse-given state of affairs. When the predicate afraid is not truth-conditional, a ground-floor explicature is expressed in the embedded complement clause, while the main clause expresses the speaker’s attitude of regret toward this explicated thought. A prerequisite to what I am going to say about the relationship between procedural meaning and the speaker’s derivational intention in the present section is a brief introduction to certain procedural means of indicating that a given verb (or adjective) of propositional attitude should be understood either as a “parenthetical” modifier of the truth-conditional content of the utterance or as a truth- conditional predicate integrated in the proposition expressed. It is a well-known fact that the parenthetical syntactic position of I’m afraid in a sentence structure like (12) encodes the procedural information that this right-detached chunk does not contribute to truth-conditional content: (12) It’s a miscalculation, I’m afraid
It is also possible to exploit intonational contrasts in order to convey the procedural information that I’m afraid in the matrix clause of (11) should be interpreted either as an integral part of the proposition expressed or as a non-truth-conditional comment on the proposition of the embedded complement. A strong pitch accent on the adjective afraid, as indicated in (13), directs the hearer to a truth- conditional interpretation of that word, while a pitch accent on miscalculation and no stress on afraid, as shown in (14), conveys the same procedural information about the speaker’s derivational intention as the syntactic form in (12) with its appended parenthetical clause. (13) I’m a´FRAID it’s a miscalculation. (But I hope it’s not) (14) I’m afraid it’s a miscalcu´LAtion. (I’m sorry about that)
The speaker’s derivational intention
A paper by van Dommelen and Fretheim (2013) reports on the outcome of two listening comprehension tests, one of which featured the Norwegian verb skjønne (either ‘understand’ or parenthetical ‘see’ of ‘you see’) with a second person subject pronoun and with an embedded complement clause as its direct object. The syntactic form of the stimuli was constant but the intonation structure superposed on the sentence was varied systematically. 25 informants agreed to participate in the test. They were seated individually in a phonetic studio and listened to fourteen distinct, prosodically manipulated versions of the same sentence. Each stimulus was presented twice, and the order of presentation was randomized. The listeners’ task was to determine the meaning of each stimulus, which for us amounted to determining whether the cognitive verb in the main clause was used as a predicate whose embedded complement was presupposed to be true, or as a non-truth-conditional verb that restricts the truth-conditional content to the complement. Figures 1 and 2 illustrate two minimally different pitch contours, accompanied by waveforms, which were superposed on the sentence used in the test. They are identical up until the final word form klage (‘complaint’), whose second syllable ends on a high pitch representing a high terminal boundary tone (H%) in Figure 1, while the corresponding high-pitched point is followed by a fall to a pitch level representing a low terminal boundary tone (L%) in Figure 2. There is exactly one focal phrase-tone, and therefore no more than a single FP, in Figure 1 as well as in Figure 2. The phrase-tone is manifested phonetically as a pitch peak in the utterance-final syllable of the FP, here the unaccented second syllable of the noun klage. It may sound counter-intuitive that a tone at the supra-lexical level of prosodic structure manifests itself in an unaccented syllable, but one should bear in mind that the pitch peak, the phonetic sign of a focal
Pitch (Hz)
250 200 150 100 50
0
Time (s)
1.8
[[du skjønner at jeg ´sendte inn en ´KLAge]FP H%]IU you see that I sent in a complaint ‘You see, I filed a complaint’ or ‘You do understand that I filed a complaint?’
Figure 1. Broad-focus intonation and High terminal boundary tone
43
44 Thorstein Fretheim
Pitch (Hz)
250 200 150 100 50
0
Time (s)
1.8
[[du skjønner at jeg ´sendte inn en ´KLAge]FP L%]IU you see that I sent in a complaint ‘You see, I filed a complaint’
Figure 2. Broad-focus intonation and Low terminal boundary tone
phrase-tone in East Norwegian intonation, highlights a syntactic phrase in the FP terminated by the raised peak just in case that phrase carries a word-tone (or word-accent), which happens to be a Low tone for the so-called Accent 1 and a High tone for the lexically opposite Accent 2 in East Norwegian speech. The respective H% and L% boundary tones do not belong to the FP constituent but are properties of the IU that comprises the whole utterance (see Fretheim and van Dommelen 2012, for an analysis of the East Norwegian boundary tones). The fundamental frequency (f0) endpoint in Figure 1 is at 200 Hz (a male voice), which is higher than any preceding high points in the utterance. This high terminal point represents two distinct phonological values, both the focal phrase-tone that terminates the FP and the high terminal boundary tone H% that terminates the IU (Fretheim and van Dommelen 2012). In Figure 2, the utterance-final, unstressed syllable includes first a rise to the peak representing the focal phrase-tone and then a fall from that peak to a lower f0 level indicative of L%. So much is happening within that short, unstressed syllable that the f0 level of the focal phrasetone in Figure 2 cannot possibly reach as high a level of pitch as in Figure 1. The space is cramped, especially when the speaker’s choice of boundary tone is L%. The people who participated in the listening test were Norwegians whose intonation is of the East Norwegian type with focal phrase-tones realized as the high endpoint of a rising pitch movement no matter whether that f0 maximum point is in an accented or an unaccented syllable, as opposed to West and North Norwegian intonation where the phonetic locus of focal phrase-tones always coincides with an accented syllable. The participants were asked to determine whether the different spoken versions of the sentence Du skjønner at jeg sendte inn en klage meant either ‘Can you confirm that you have understood that I filed
The speaker’s derivational intention
a complaint?’ (skjønner as a semi-factive predicate that contributes to the proposition expressed), or else ‘I filed a complaint, you see’ (skjønner as a parenthetical verb which indicates that the explicature expressed in the complement clause is offered as an explanation for, and possibly justification of, something brought up in the preceding discourse). The upper field of the computer screen that the listeners were facing in the studio gave them the cue Du skjønner det? (‘You understand that?’) and the lower field the cue Jeg klaget, skjønner du (‘I complained, you see’). They were instructed to mouse-click either in the upper or in the lower field when they had made their decision, an operation that automatically triggered presentation of the next stimulus. Let us now take a look at how the minimal difference between Figure 1 and Figure 2 affected the outcome of our test. The listeners almost univocally (98% of the responses) found that the utterance represented in Figure 2, with the L% boundary tone, was used to assert that the speaker filed a complaint. Thus, the main clause verb was non-truth-conditional and the propositional content was restricted to the embedded clause. When the boundary tone was high (H%), as in Figure 1, there was no longer an overwhelming majority who opted for the assertion reading (70%). Even though the observed difference in the reactions to Figure 1 and Figure 2 was statistically significant, it is natural to ask why the H% vs. L% distinction, the only difference between the contours of Figure 1 and Figure 2, did not cause many more people to associate the H% in Figure 1 with the act of asking if the hearer had understood what had happened. What intonational feature is responsible for the fact that a comfortable majority of responses still supported the parenthetical interpretation of the main clause verb and the belief that the complement proposition was asserted, in spite of the high-pitched endpoint (H%)? The prosodic handling of the main clause verb turned out to be a key factor. It was a more powerful procedural indicator than the terminal boundary tone. Skjønner is an unaccented verb form both in Figure 1 and Figure 2. Two of the stimuli included in our test showed global intonation patterns similar to these, except that skjønner was made more prominent because it carried a word-tone. This led to an increase in the number of responses in favor of the request for confirmation interpretation, both with the H% and the L% at the end, but not sufficiently many to create a significant difference between the stimuli with an unstressed main clause verb and the stimuli whose main clause verb had a word-tone. However, when a word-tone on skjønner combined with an utterance-medial focal phrase-tone that gave extra prominence to this verb, the distribution of responses was completely different. Regardless of the terminal boundary tone, this prosodic highlighting of the main clause verb proved to be a very powerful indicator of
45
46 Thorstein Fretheim
the message that the speaker requests the hearer to confirm a proposition whose semantic material comprises the main clause as well as the embedded clause. There was a minuscule difference between the listeners’ reactions to the minimally different alternatives indicated in (15) and (16) where the main clause verb is highlighted by a focal phrase-tone at the end of the first FP but where the terminal boundary tone is H% in (15) and L% in (16). The request for confirmation interpretation was selected 96% of the time when the terminal boundary tone was H% and 92% of the time when it was L%. Boosting the main clause verb by means of the focal phrase-tone of the first FP in (15) and (16) is an efficient way to change the assertion in Figure 1 and Figure 2 into a request for confirmation: (15) [[du ´SKJØNNer at jeg]FP [´sendte inn en ´KLAge]FP H%]IU (16) [[du ´SKJØNNer at jeg]FP [´sendte inn en ´KLAge]FP L%]IU ‘You understand that I filed a complaint, don’t you?’
Pitch (Hz)
Figure 3 below shows the f0 contour of the stimulus of (15). Here the effect of the first FP-terminal pitch peak is that it boosts the main clause verb, not because there is a communicated contrast between this lexical verb and some other predicate but because this phrase-tone indicates that the affirmative polarity of the proposition expressed is the focus of information, the only new information in the utterance. There is another FP-terminal pitch peak at the end of the utterance, but the information expressed in that second FP covering the embedded complement clause is discourse-given. There is a steeply rising pitch movement in Figure 3 from the low-pitched word-tone (Accent 1) of the present tense form skjønner to the pitch peak that terminates the first FP. This is radically different from the level pitch of the first five unaccented syllables of the utterances represented in Figure 1 and Figure 2.
250 200 150 100 50
0
Time (s)
1.8
[[du ´SKJØNNer at jeg]FP [´sendte inn en ´KLAge]FP H%]IU ‘You understand that I filed a complaint, don’t you?’
Figure 3. Double-FP intonation with polarity focus and High terminal boundary tone
The speaker’s derivational intention
The f0 contour superposed on the sentence-final noun klage in Figure 3 is identical to the corresponding contour at the end of Figure 1, but while the instruction to associate H% with a question had an impact on the test subjects’ judgments in Figure 1, the importance of H% as a procedural indicator of a (biased) question was found to be negligible when the main clause verb was prosodically highlighted as shown in Figure 3. The focal phrase-tone that restricts the information focus to the affirmative polarity of the complex sentence is a procedural device that overrides the potential effect of the difference in procedural meaning between the H% of Figure 3 and the opposite tone L% seen as the endpoint of the pitch contour in Figure 2. When the intonation was the double-FP structure visualized in Figure 3, a mere 8% of the responses failed to support the request for confirmation interpretation even when the boundary tone on the utterance-final syllable was L%, and only 4% failed to select that interpretation when the boundary tone was H%, the contour shown in Figure 3. The procedural meaning of H% is the same in Figure 1 and Figure 3. A high boundary tone instructs the hearer to test for relevance the hypothesis that the speaker is asking for information rather than offering the hearer information, but a variety of language-specific linguistic facts and contextual factors determines whether this procedural meaning will be fruitfully exploited by the hearer in his pragmatic processing of the stimulus, or whether it should rather be ignored because there exists evidence to the contrary in the form of procedural meaning that nullifies its potential effect. In the latter case the hearer’s attention will be turned away from the speaker’s choice of terminal boundary tone. This is in keeping with Sperber and Wilson’s view (1986/1995) that hearers only pay attention to phenomena that appear to be relevant to them. While the procedural information associated with the phonological opposition between the two boundary tones did affect the responses of a fair number of our test subjects when the main clause verb was unaccented as in Figure 1 and Figure 2, the same boundary tone distinction had virtually no impact on the listeners’ judgments when the main clause verb was made prominent by a focal phrase-tone, as in the pair of (15)/ Figure 3 and (16). The rising f0 movement to the first f0 maximum in Figure 3 is a perceptually conspicuous procedural indicator of the truth-conditional status of the main clause verb, and the combination of polarity focus on the verb, the global intonational phrasing, and the H% boundary tone jointly provides the procedural information that the speech act is a request for confirmation. When the cognitive verb in the main clause is unaccented, the low boundary tone L% in Figure 2 is a better procedural pointer to the overall favorite interpretation ‘I filed a complaint, you see’ than the high boundary tone H% in Figure 1. The L% in Figure 2 discriminates better between the two competing interpretations than the H% of Figure 1, because L% interacts relevantly with the lack of pitch
47
48 Thorstein Fretheim
accent on the main clause verb. A parenthetical verb should not be accented, or stressed, in any language. This constraint applies both to a parenthetical verb that appears in a right-detached clause, as in English (12) above, and to a parenthetical that occupies the in situ position of the finite verb in a main clause. When the verb form skjønner in our sentence is unaccented, the utterance directs the hearer to a parenthetical interpretation of the main clause verb and a ground-floor explicature confined to the complement string. A terminal L% supports this interpretation, because the low boundary tone sits well with an explicature interpretation of the complement proposition. A terminal H%, however, is a marked choice if the function of the complement is to explicate the thought that the speaker filed a complaint. An H% in conjunction with an unstressed main clause verb detracts from relevance, because there is a risk, however small, that it directs the hearer to an unintended question interpretation of the stimulus. An L% in a similar phonological environment arguably increases the probability that a processing of the main clause verb as a parenthetical and the embedded proposition as a groundfloor explicature is consonant with the speaker’s derivational intention. The procedural information derived from the intonation in Figure 3 is due to the procedural meaning of a specific global intonation structure in Norwegian, a pattern that involves more than the main clause verb furnished with an elevated phrase-tone. It so happens that there must also be a word-accent somewhere in the intonation structure subsequent to the focally accented verb form. This condition is met by the intonation represented in Figure 3. If there is nothing after the focally accented verb, or just a string of unaccented word forms, the procedural meaning of the intonation structure changes radically, in ways not relevant to the topic of this paper. When structural input conditions are satisfied, a phrase-tone on a finite verb indicates polarity focus, which implies that the verb is included in the propositional content. The procedural information derived from the unaccented main clause verb in Figures 1 and 2 is due to a lexical property of the Norwegian verbs that display the same duality of function as skjønne (‘understand’ ‘see’) with a clausal complement: parenthetical vs. non-parenthetical. Provided the subject of this verb is 1st person or 2nd person, an unaccented form skjønner indicates that the verb is a parenthetical, and yet a number of interacting linguistic conditions must be met in order for a hearer to recognize the use of the verb as parenthetical without fail. And even under the linguistically most favorable circumstances there is a risk that some contextual premise causes the hearer to choose a different interpretation. The hypothesis that should be tested for relevance before any contrary hypothesis is that the unaccented verb skjønner is a parenthetical. If it does not meet the hearer’s expectation of optimal relevance (Sperber and Wilson 1986/1995), it will
The speaker’s derivational intention
be abandoned and replaced by an interpretation of skjønner as a verb that contributes to the propositional content of the utterance. The upshot of my discussion of the data examined in the present section is that a particular procedural meaning encoded by a linguistic device may direct the hearer to the speaker’s derivational intention and informative intention in one context but may fail to do so in a different context. A rather trivial problem with this conclusion is that, behind data invented for the purpose of testing some hypothesis, including stimuli prepared for a laboratory experiment, there is no human communicator with an intention to inform. We can only stipulate that a rational, competent language user, an ideal one, produces the stimulus on the basis of an intention that is determinable to the hearer, given available linguistic and extra-linguistic input. With regard to the study described in this section (van Dommelen and Fretheim 2013) it must be legitimate to conclude that an interpretation that is almost unanimously accepted by a body of participants in a listening comprehension test is in actual fact in agreement with the intention of a competent East Norwegian speaker, both the informative intention and the derivational intention that a hearer has to grasp in order to identify the informative intention. Since 98% of the responses to the utterance represented in Figure 2 supported the interpretation ‘I filed a complaint, you see’, with a parenthetical present-tense form skjønner (‘understands’) in the main clause and an explicated proposition in the complement clause, this is indeed what we can reasonably assume that a speaker in a real conversational event would have meant by an utterance with the intonational form of Figure 2. In Figure 1 the terminal boundary tone was high (H%). This prosodic gesture does not interact optimally with the unaccented verb form skjønner; specifically, it does not support the hearer’s perception of the embedded clause as the linguistic locus of an explicature. And yet, recall that 70% of the responses expressed the judgment that the speaker explicated the complement proposition ‘Speaker filed a complaint’. This suggests that, for the majority of listeners to the utterance represented in Figure 1, the potential procedural effect of H% was overridden by the information expressed by the lack of accent on skjønner, a sign that this main clause verb has the function of a parenthetical. In Figure 3, where the main clause verb was highlighted by means of a focal phrase-tone that indicates its role as the linguistic mark of polarity focus, the difference in procedural meaning between H% (requesting information?) and L% (offering information?) was truly insignificant, as those utterances were identified as requests for confirmation of the proposition of the complex sentence by nearly all listeners, regardless of whether the terminal tone was H% or L%. Although the terminal boundary tone was their most recent impression of a pitch-based
49
50
Thorstein Fretheim
gesture in the stimulus presented, the difference between H% and L% was largely disregarded by the listeners. A still unpublished follow-up paper (Fretheim and van Dommelen 2013) explores the relative strengths of intonational parameters in complex Norwegian sentences consisting of a main clause second person subject pronoun, the cognitive verb vite (‘know’) in the present tense, and an embedded complement clause as direct object. In contrast to the experiment described above, the new stimuli differed not only intonationally but also syntactically, as we varied the word order in the complement clause. Three stimulus sentences were used: (17) with a subordinate clause order in the embedded complement, i.e. the negation marker ikke before the finite copula verb er, (18) with a main clause order in the embedded complement, i.e. the finite verb in the second position (V2), and (19) where the predicate forbudt (‘forbidden’, ‘illegal’) in the embedded complement incorporates negation and so conceals the difference in procedural meaning between the complement clause word orders of (17) compared to (18). (17) Du vet at det ikke er lov å parkere her you know that it not is law to park here ‘You know that it isn’t legal to park here, don’t you?’ (18) Du vet at det er ikke lov å parkere her you know that it is not law to park here ‘You know, it isn’t legal to park here’ (19) Du vet at det er forbudt å parkere her you know that it is forbidden to park here ‘You know that it’s forbidden to park here’
If the main clause verb vet (‘know’) is used as a parenthetical, the meaning of (17), (18) and (19) is ‘It’s forbidden to park here, you know’, a reminder to the hearer. If vet is a non-parenthetical lexical verb form, the meaning is ‘You are aware that it’s forbidden to park here, aren’t you?’, a request for confirmation of the assumption that the hearer is aware of the fact that the proposition expressed in the complement clause is true. Our expectation was that the constituent order difference between (17) and (18) would be a fairly powerful procedural indicator, as it is natural to think that a main clause order in the complement of (18) implies that the truth-conditional part of the utterance is restricted to what is expressed in the complement, while the contrasting subordinate clause order in (17) would presumably direct the hearer to the other interpretation. In (19) the choice between the reminder and the request for confirmation interpretation is determined exclusively by the intonation structure. The intonation offers the hearer the only context-creating procedural cue in (19), but we foresaw that there might be a
The speaker’s derivational intention
conflict in (17) and (18) between what the syntactic string indicates and what the intonation indicates. It turned out that the prosody had an overriding effect when a double-FP intonation similar to what was shown in Figure 3 was employed in the production of (17) and (18). A focal phrase-tone on the main clause verb in combination with a high terminal boundary tone resulted in 84% responses in favor of the request for confirmation interpretation of (18). The main clause verb was perceived as non-parenthetical in spite of the V2 word order in the embedded clause, the order that is mandatory in matrix clauses. In Section 2.1 it was shown that the lexical properties which distinguish the two focus particles også (‘also’) and og (‘too’) were not only contradicted by the intonational phrasing shown in (10) but the intonation actually set the procedural effect of the lexical choice aside. Likewise, the listeners’ comprehension of sentence (18) produced with the intonational phrasing and high boundary tone indicated in (20) revealed a very strong tendency for listeners to ignore the procedural information expressed by the main-clause word order er ikke (‘is not’) in the embedded clause. (20)
[[du ´VET at det]FP [´er ikke ´lov å par´KERe her]FP H%]IU Preferred interpretation: ‘You know that it’s not legal to park here, don’t you?’ Dispreferred interpretation: ‘You know, it’s not legal to park here’
The procedural constraint of the prosodic sort was seen to be perceptually more important than the competing word-order constraint. Thus, in the global intonation pattern indicated in (20), the speaker’s derivational intention was reflected most reliably by the focal phrase-tone that gives tonal prominence to the verb form vet (‘know’). 3. Optimal and non-optimal stimuli Accounts of how to deal with conflicts between opposite and often countervailing cues in the production as well as the comprehension of linguistic expressions are rife in pragmatic literature. Gricean pragmatics is a typical example. A speaker intent on observing conversational maxims and a hearer engaged in utterance comprehension may, according to Gricean reasoning, be forced to sacrifice the First Maxim of Quantity, Q1 (“Make your contribution as informative as is required”) in order to be true to the Maxim of Quality (“Try to make your contribution one that is true”). Violation of the former maxim may be judged by a
51
52
Thorstein Fretheim
speaker in a given communicative setting to be of a less severe sort than violation of the latter, and occasionally it turns out to be impossible to observe both. Levinson (2000) proposes two optimization principles. His Q-Principle corresponds to Grice’s First Maxim of Quantity, Q1, and his I-Principle corresponds to Grice’s Second Maxim of Quantity, Q2 (“Do not make your contribution more informative than is required”). In relevance theory, potential conflicts between minimization of processing cost and maximization of cognitive effects for the hearer caused Sperber and Wilson (1986/1995) to develop the theoretical concept of optimal relevance characterized by a sound balance between efforts expended and effects achieved. Optimality Theory (OT) argues that linguistic forms arise from the interaction between conflicting constraints. It sees grammars as systems that provide mappings from inputs to outputs and defines one output, among a set of candidates, as optimal if it only violates universal, hierarchically ranked constraints whose violation is less severe than the violations incurred by any of its competitors. OT is no longer a single theory but a group of theories that share some fundamental tenets. One branch is the “bidirectional” OT designed specifically for the development of pragmatic research based on optimality-theoretic principles (e.g. Blutner and Zeevat 2004; Blutner 2007). The neo-Gricean Q- and I- Principles are central here. While the I-Principle helps the addressee select the optimal interpretation of a linguistic input among a set of candidate interpretations, the Q-Principle “[…] compares different possible syntactic expressions that the speaker could have used to communicate the same meaning” and “[…] acts as a blocking mechanism which blocks all the outputs which can be grasped more economically by an alternative linguistic output” (Blutner 2007: 74–75). I said in Section 2.2 that the procedural meaning of a focal phrase-tone that highlights a finite verb in a spoken utterance of a Norwegian sentence includes the information that the verb is a truth-conditional constituent of the input to the hearer’s pragmatic processing, and that the lack of pitch accent on a cognitive verb form like skjønner (mental ‘see’) encodes the procedural information that the verb is a parenthetical. The encoded procedural semantics of the prosodic contrast between a focally accented and an unaccented constituent has consequences for the hearer’s search for an interpretation consonant with the speaker’s derivational intention and informative intention. The very modest effect of the choice between the boundary tone H% in (15) compared to the boundary tone L% in (16) in Section 2.2 might give one the impression that the procedural constraints expressed by the two opposite boundary tones, an assertion indicated by L% and a question indicated by H%, have a more humble rank in spoken East Norwegian than the accentual handling of the cognitive verb in the main clause. This observation of a possible difference in rank between constraints is reminiscent of the
The speaker’s derivational intention
language-specific ranking of conflicting universal constraints in OT, which is supposed to give rise to variation in generated output across languages. However, the results of the listening test described in the as yet unpublished follow-up paper (Fretheim and van Dommelen 2013) suggest that it would be premature to rank the boundary tone constraint lower than the constraint encoded by the difference between an unaccented and a focally accented main clause verb exclusively on the basis of what the earlier listening test (van Dommelen and Fretheim 2013) seemed to tell us. It was the co-occurrence of H% and a focal phrase-tone highlighting the main clause verb that swayed the opinion to a comfortable majority vote for the request for confirmation interpretation of (20). This happened in spite of the fact that the main-clause (V2) word order in the complement clause in (18)/(20) suggests that the ground-floor explicature of the utterance is not meant to include the main clause subject and predicate. The procedural meaning of the V2 order in the complement clause was overridden by the joint effect of polarity focus on the verb form vet (‘know’) and the boundary tone H% in (20). Slightly less than the majority of responses supported the request for confirmation interpretation of (18) when the main clause verb was focally accented but the boundary tone was L%. If a speaker employs the Norwegian sentence Du skjønner at jeg sendte inn en klage (‘You understand that I filed a complaint’) with the intention to tell the hearer that a complaint was filed, the optimal intonational output combines an unaccented verb skjønner and the boundary tone L%. If the speaker’s intention is to use that sentence to ask for confirmation of the assumption that the hearer understands that a complaint had been filed, the optimal intonational output combines a focal phrase-tone on skjønner and the boundary tone H%. Non-optimal combinations of prosodic features may affect the relevance of a stimulus negatively, but not all non-optimal linguistic outputs, viewed as candidates for the expression of a given meaning, are faulty linguistic forms. Unlike the situation in OT, there is no unique winner output that blocks all other outputs. Some candidate output forms do block access to a given interpretation but more than one output can do the same job as the optimal candidate. The hearer may obtain a correct mental image of what the speaker meant even if the stimulus is not optimal, though the task of arriving at the right result may involve less mental effort if an optimal stimulus is employed. Naturally, a stimulus with two or more procedural indicators that interact positively by reducing processing cost for the hearer is more relevant than a stimulus with two or more conflicting procedural indicators; it reflects the speaker’s derivational intention in a more successful way. The less powerful of two indicators that encode contrary procedural meanings may be disregarded by the hearer at a subliminal level because it appears not to contribute to relevance, but we have no guarantee that the hearer will disregard it. If the hearer notices it,
53
54
Thorstein Fretheim
increased processing cost may be one type of infelicitous result, and the worstcase scenario is that the hearer even misunderstands the speaker’s informative intention. 4. Conclusion A speaker’s informative intention is captured by the addressee insofar as the latter succeeds in forming a mental representation of the speaker’s communicated thought based on the speaker’s ostensive-inferential behavior, verbal or non- verbal (Sperber and Wilson 1986/1995). The concept of derivational intention (Powell 2010) relates to the dynamic phase of inferential processing of an utterance in a context, and on my account of the relationship between the speaker’s derivational intention and her informative intention there is no discrepancy between the two notions. Linguistic devices that encode a procedural meaning, whether single linguistic items, grammatical constructions or intonation patterns, place constraints on the hearer’s choice of an inferential route that enables him to arrive at the content of the speaker’s informative intention, but as two co-occurring procedural markers may direct the hearer to distinct and mutually incompatible inferential routes, hence to distinct outputs of the pragmatic phase of utterance interpretation, the hearer has no guarantee that the speaker’s derivational intention is identified correctly just because inferences have been drawn in accordance with a specific constraint on interpretation encoded by a single procedural marker. Procedural meaning as a clue to the speaker’s derivational intention is fallible, and when two distinct expressions that encode a procedural meaning direct the hearer to two incompatible inferential routes, the instruction prompted by one of them has to be sacrificed. Conceptual meaning works in tandem with procedural meaning for the benefit of the audience processing a stimulus. The former constrains the hearer’s search for the right inferential path as much as the latter. One piece of conceptual information may strengthen the pragmatic effect of a given encoder of procedural meaning, while a different piece of conceptual information may weaken the potential effect of the same encoder of procedural meaning. I have elsewhere defended the view that descriptive meaning in a definite noun phrase used as a referential term does not contribute a concept to the truth-conditional content of the utterance but serves to constrain the addressee’s search for a relevant interpretation (Fretheim 2011, 2012). Although the description contained in the referential term encodes concepts, those concepts are not supplied to the proposition to be derived at the addressee’s end. Section 1.3 of the present paper showed how
The speaker’s derivational intention
one aspect of the conceptual meaning of the English verb fasten makes it to the addressee’s mental representation of the imperative appearing at the back of passenger seats in aircraft cabins, while the encoded concept of change-of-state from unfastened to fastened pertaining to the same verb fasten had better be ignored by the passengers, because its co-occurrence with the adjunct while seated makes the inchoative reading of this verb pragmatically unacceptable. It is physically impossible to fasten one’s seat belt if one is not already seated. In Section 2.1, I argued that it is possible for a prosodic marker of procedural meaning to override procedural information derived from a particular lexical item. This is true of the interaction of the procedural meaning associated with the additive focus particle og (‘too’) and the procedural meaning associated with a Norwegian intonation structure that groups the particle and the following predicate nominal together in a single prosodic constituent. What counts more than the lexically induced instruction to locate the focus constituent to the left of the focus particle is the procedural meaning of the intonational phrasing: the focus particle og and the focus constituent that it targets are located in the same FP. The impact of the intonational phrasing is that the hearer ignores the encoded procedural information that og, like English too, targets a phrase earlier in the utterance. Section 1.2 showed that a novelist’s derivational intention may be for readers to suspend their expectation that those personal pronouns in the very first paragraph of a novel (cf. (1)–(2)) behave in accordance with the procedural instruction that the pronouns encode. Antecedent-free personal pronouns do not give a reader access to the sort of information that is necessary to form a unique mental representation of the referent. A masculine or a feminine 3rd person singular pronoun in the very first sentence of a novel constrains the reader’s search for the referent considerably, but it is only when the information about gender and number can be matched against different pieces of descriptive information in the following pages of the novel, that the reader will be able to resolve the reference of these pronouns. When they appear in the initial paragraph, the writer’s derivational intention is at odds with the procedural constraints that the pronouns encode but the author’s derivational intention and the procedural meanings of the pronouns are united when the same 3rd person masculine and feminine pronouns reappear a few pages later. Two or more co-occurring procedural markers may be mutually supportive, obviously, so that they jointly constrain the relevance of the stimulus in an optimal way. For instance, the joint impact of a focal phrase-tone on the verb form skjønner (‘understand’, ‘see’) and the boundary tone H% in the data discussed in Section 2.2 constrains interpretation more than an intonation pattern terminated with H% but without a strongly accented main clause verb. This paper has shown,
55
56
Thorstein Fretheim
however, that co-occurring procedural markers may also convey conflicting procedural information. Indeed, a given stimulus might contain two equipotent procedural markers that express contrary instructions to the hearer, with the result that the stimulus is actually rendered ill-formed, a situation that might confuse the hearer so much that no relevant interpretation is obtained. I have concentrated on a different kind of outcome of the co-occurrence of two procedural markers that convey contrary instructions to the hearer, focusing on cases where the pragmatic effect of one encoder of procedural meaning may be nil if the stimulus includes another procedural marker that turns the hearer’s attention to a different inferential route. Two co-occurring procedural markers may not be equally highly ranked for native speakers and hearers, so that one may win out and the other one may be ignored in the pragmatic phase of interpretation. The less powerful procedural marker may be relevant enough in an utterance where its potential effect on the hearer is not challenged by contrary contextual information or the co-presence of a procedural marker that guides the hearer to a different inferential route, but when two procedural markers that encode conflicting information co-occur in an utterance, one of them may simply not be relevant enough to be given any notice by the addressee. A possible result of the kind of situation where two co-occurring but mutually incompatible procedural markers are unequally strong is that the overall relevance of the stimulus is somewhat reduced due to disturbing ‘noise’ caused by the presence of the weaker marker, resulting in gratuitous processing on the part of the hearer. It must be legitimate to presume that, in such situations the more powerful, outranking procedural marker is the one that directs the hearer to the speaker’s derivational intention. The relatively weaker procedural marker typically fails to do so. A reviewer of this paper was left with the question of why speakers would ever allow two conflicting procedural markers to co-occur in an utterance. Why would a speaker use a procedural marker if it is to be ignored? My answer is that speakers have no conscious control of the multifarious linguistic devices that presumptively contribute to the relevance of a given linguistic stimulus, some of which encode conceptual meaning, others procedural meaning. Use of a lexical item that encodes a concept may be less than optimal because the encoded concept does not quite capture what the speaker intended that lexical item to convey. In spite of this, the complex interaction of a variety of mentally activated verbal and contextual cues may direct the hearer to an interpretation that successfully recovers the speaker’s informative intention. In the same vein, I conclude that a speaker has no conscious awareness of the fact that two co-occurring procedural markers may turn out to be in conflict. A linguistic researcher may be able to demonstrate that such situations exist but native speakers do not consciously avoid them in
The speaker’s derivational intention
their production of linguistic stimuli. As always in ostensive-inferential communication, hearers will only pay attention to those aspects of a linguistic stimulus that immediately strike them as contributing to relevance. I have argued in favor of a theoretical distinction between a speaker’s use of linguistic devices that encode a piece of procedural information on the one hand and the speaker’s intention that the hearer sticks to a specific inferential path consonant with the speaker’s informative intention on the other hand. Following Powell (2010) I have referred to the latter as the speaker’s derivational intention. It remains to be seen to what extent this theoretical distinction proves to be useful cross-linguistically in the exploration of pragmatic implications of a wider array of linguistic expressions and constructions.
Acknowledgements I am grateful to Manuel Padilla Cruz for his excellent editorial work that surely improved the quality of this paper, and to four anonymous referees.
References Blakemore, Diane. 1987. Semantic Constraints on Relevance. Oxford: Blackwell. Blakemore, Diane. 2002. Relevance and Linguistic Meaning: The Semantics and Pragmatics of Discourse Markers. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511486456 Blutner, Reinhard. 2007. “Optimality Theoretic Pragmatics and the Explicature/Implicature Distinction.” In Pragmatics, ed. by Noel Burton-Roberts, 67–89. Basingstoke: Palgrave. Blutner, Reinhard, and Henk Zeevat (eds). 2004. Optimality Theory and Pragmatics. Basingstoke: Palgrave. doi: 10.1057/9780230501409 Carston, Robyn. 2012. “Word Meaning and Concept Expressed.” The Linguistic Review 29: 607–623. doi: 10.1515/tlr-2012-0022 Dommelen, Wim A. van, and Thorstein Fretheim. 2013. “Intonational Cues to the Identification of Propositional Form and Speech-act Type in Norwegian.” In Nordic Prosody: Proceedings of the XIth Conference, ed. by Eva Liina Asu and Pärtel Lippus, 401–410. Frankfurt am Main: Peter Lang. Fretheim, Thorstein. 1978. “When Syntax Fails to Determine Semantic Scope.” In Nordic Prosody, ed. by Eva Gårding, Gösta Bruce, and Robert Bannert, 5–14. Lund: Lund University Press. Fretheim, Thorstein. 1981. “Intonational Phrasing in Norwegian.” Nordic Journal of Linguistics 4: 111–138. doi: 10.1017/S0332586500000718 Fretheim, Thorstein. 1992. “Themehood, Rhemehood and Norwegian Focus Structure.” Folia Linguistica XXVII: 111–150.
57
58
Thorstein Fretheim
Fretheim, Thorstein. 2011. “Description as Indication: The Use of Conceptual Meaning for a Procedural Purpose.” In Procedural Meaning: Problems and Perspectives, ed. by Victoria Escandell-Vidal, Manuel Leonetti, and Aoife Ahern, 131–156. Bingley, UK: Emerald Group. Fretheim, Thorstein. 2012. “Relevance Theory and Direct Reference Philosophy: A Suitable Match?” In Relevance Theory: More than Understanding, ed. by Ewa Wałaszewska and Agnieszka Piskorska, 331–352. Newcastle upon Tyne: Cambridge Scholars Publishing. Fretheim Thorstein, and Wim A. van Dommelen. 2012. “A Pragmatic Perspective on the Phonological Values of Utterance-final Boundary Tones in East Norwegian.” The Linguistic Review 29: 663–677. Fretheim, Thorstein, and Wim A. van Dommelen. 2013. “Intonation as a Guide to a Parenthetical vs. a Non-parenthetical Interpretation of a Norwegian Cognitive Verb.” Paper presented at the 19th International Congress of Linguists, Geneva, July 21–27. Gundel, Jeanette K. 2011. “Child Language, Theory of Mind, and the Role of Procedural Markers in Identifying Referents of Nominal Expressions.” In Procedural Meaning: Problems and Perspectives, ed. by Victoria Escandell-Vidal, Manuel Leonetti, and Aoife Ahern, 205–231. Bingley, UK: Emerald Group. Gundel, Jeanette K., Nancy Hedberg, and Ron Zacharski. 1993. “Cognitive Status and the Form of Referring Expressions in Discourse.” Language 69: 274–307. doi: 10.2307/416535 Levinson, Stephen C. 2000. Presumptive Meanings. The Theory of Generalized Conversational Implicature. Cambridge, MA: MIT Press. Powell, George. 2010. Language, Thought and Reference. Basingstoke: Palgrave. doi: 10.1057/9780230274914
Sperber, Dan, and Deirdre Wilson. 1986/1995. Relevance. Communication and Cognition. Oxford: Blackwell. (2nd edition with a Postface) Wilson, Deirdre, and Robyn Carston. 2007. “A Unitary Approach to Lexical Pragmatics: Relevance, Inference and Ad Hoc Concepts.” In Pragmatics, ed. by Noel Burton-Roberts, 230– 259. Basingstoke: Palgrave. doi: 10.1093/oxfordhb/9780199234769.003.0018 Wilson, Deirdre, and Dan Sperber. 1993. “Linguistic Form and Relevance.” Lingua 90: 1–25. doi: 10.1016/0024-3841(93)90058-5
Cracking the chestnut How intonation interacts with procedural meaning in Colloquial Singapore English lah Junwen Lee and Chonghyuck Kim
Brown University / Chonbuk National University
This chapter argues against previous analyses of the Colloquial Singapore English (CSE) particle lah. While homonymic approaches (e.g. Wong 2004) conflate pragmatic function and semantic meaning, unitary approaches (e.g. Gupta 2006) ignore the systematic differences in function that correlate with tonal differences. Using a relevance-theoretic framework, this chapter proposes that lah describes the preceding proposition as being of high epistemic strength, and the hearer interprets this description as being either of an actual situation (signalled by the falling tone of a declarative), or of a thought that is desirable to the hearer (signalled by the rising tone of a interrogative). One phenomenon that this analysis explains is how lah can strengthen declaratives but weaken imperatives. Keywords: Colloquial Singapore English, intonation, procedural meaning, lah, discourse particles
1. Introduction A wide range of pragmatic functions have been attributed to the Colloquial Singapore English (CSE) particle lah, which has been described as “one of the hardest and most intriguing chestnuts in the field of cross-cultural pragmatics” (Besemeres and Wierzbicka 2003: 3). For instance, lah can be used to lend emphasis to utterances, e.g.: (1) A: Do you want to go? I’m not going lah.1 (I emphasise to you that I’m not going.)
(Kwan-Terry 1992: 69)
1. Discourse participant labels have been amended in this chapter – the speaker of utterance marked by the CSE particle is labelled as A and is taken to be female, while the hearer is labelled as B and is taken to be male, unless otherwise determined by the dialogue itself. doi 10.1075/pbns.268.03lee © 2016 John Benjamins Publishing Company
60 Junwen Lee and Chonghyuck Kim
However, lah can also serve to soften the force of utterances, or mark as salient the close social relationship between speaker and hearer: (2) A: No use trying to hide our roots lah. (As one Singaporean to another, there is no use in trying to hide our roots.) (Ler 2006: 155)
Since these functions, among others, have been observed to correspond with specific pitches with which lah is pronounced, some researchers (Bell and Ser 1983; Kwan-Terry 1978; Wong 2004) have concluded that there are different variants of the particle, each with its own pragmatic function. However, we propose instead in this chapter that lah has a unitary procedural meaning that indicates that the lah-marked proposition can be accepted by the hearer as true with a high degree of confidence. The different pragmatic functions that other researchers argue are inherent to the particle, e.g. emphasising solidarity or softening the impoliteness of an utterance, result from the interaction between this core meaning and the modality of the utterance, which is signalled through the tone of the particle. 2. Discourse particles in CSE As an immigrant nation, the predominance of Chinese settlers and traders in the early twentieth century has resulted in a majority Mandarin-dialect-speaking population in Singapore (primarily Hokkien and Teochew), with Malay and Tamil speakers respectively forming the next largest proportions (Bao 2001: 281). These languages provided the loan words and even entire grammatical structures and constructions that were incorporated through successive generations of Singaporean speakers into a variety of English – CSE – that is this chapter’s object of study. The sheer number of discourse particles in CSE, mostly loans from Cantonese and Hokkien (Ler 2005:35), have been analysed in many previous studies (e.g. Bell and Ser 1983; Besemeres and Wierzbicka 2003; Gupta 1992; Kim and Wee 2009; Kwan-Terry 1978; Ler 2006; Loke and Low 1988; Pakir 1992; Platt and Ho 1989; Richards and Tay 1977; Wee 2002; Wong 2004). Though syntactically optional, these particles are essential elements in conversation, performing key pragmatic functions such as indicating the speaker’s stance or emotional tone (Wee 2004). Some of these observed functions are listed below (paraphrases in brackets) (Wee 2004: 118–124):
Intonation and procedural meaning in CSE lah
a. meh – indicates scepticism (3) A: You got girlfriend meh? (Are you sure you have a girlfriend? I don’t believe it.)
b. wat – indicates information as obvious and contradictory (4) A: I dam stupid lah! I shouldn’t have stopped [my dance classes]. (I’m so stupid! I shouldn’t have stopped [my dance classes].) B: You can start now wat! (It should be obvious to you that you can start classes again. I don’t believe you can’t do that.)
c. hor – asserts and elicits support for a proposition (5) A: Remind me to tell Jim about the group meeting hor. (I assert that I want you to remind me to tell Jim about the group meeting, and wish you will agree to it.) B: Sure. No problem.
While the particles listed above have fairly specific functions, the function of the particle lah appears to be more nebulous, and multiple functions have been attributed to it over the years. However, as Ler (2005) points out, previous characterisations of lah focus on different aspects and functions of the particle, which may not overlap or may even be contradictory. The following section will elaborate on the characterisations of lah made by previous researchers. 3. Previous characterisations of lah 3.1
Lah as a marker to indicate social relationships
Richards and Tay observe that lah can be used in various types of speech acts, “provided that there is a positive rapport between speakers and an element of solidarity” (1977: 145), and propose that lah is a code marker that identifies “rapport, solidarity, familiarity and informality” (1977: 146) between the speaker and hearer. Elaborating on lah’s function as a social marker, both Bell and Ser (1983) and Kwan-Terry (1978) argue that lah can indicate social distance as well as intimacy, with two forms that differ in phonological stress. However, while Kwan-Terry believes that it is the stressed lah that may be used to convey “a certain softening of tone or attitude” (1978: 23), Bell and Ser believe that it is the unstressed or contracted lah that “‘softens’ the meaning of the utterance, [and] introduces a degree of ‘warmth’ into the interaction” (1983: 14).
61
62
Junwen Lee and Chonghyuck Kim
Through this ‘softening’ effect, speakers are able to use lah to change a direct command into a persuading utterance, e.g.: (6) a. A: (to B) Come with us. (Direct command) b. A: (to B) Come with us lah. (Persuading tone)
(Ler 2005: 297)
In contrast, the other form of lah that indicates social distance can be used to convey impatience or even hostility, e.g.: (7) a. A: (to B) Eat your food. (Direct command) b. A: (to B) Eat your food lah. (Impatient tone)
3.2
(Ler 2005: 291)
Lah as an assertive marker
Another observed function of lah is to emphasise the speaker’s utterance. In two papers, Gupta proposes that CSE discourse particles indicate the speaker’s commitment to the “validity of the propositional content” (2006: 246) of the marked utterance, and arranges them on a single scale of assertiveness ranging from contradiction (which is maximally assertive) to tentativeness (which is minimally assertive) (2006: 250). On this scale, lah “covers the full range within the assertive continuum” (Gupta 1992: 42), allowing the speaker’s utterance to be used to establish a proposition, negate a proposition by the hearer, or to persuade or direct the hearer. Similarly, Kwan-Terry argues that the stressed form of lah may also be used to “express emphasis, though the emphatic meaning is usually modified in meaning according to the context” (1978: 22). For example, A uses lah in (8) to emphasise the truth of her utterance, indicating that it should be obvious to B that she should wear an evening dress for the evening event. (8) B: What should I wear this evening? A: Evening dress lah.
3.3
(Kwan-Terry 1978: 23)
Lah as a group with different lexical variants
To encompass the various observed functions of lah, Wong (2004) proposes that lah is used when the speaker wishes to change the mind of the hearer, and characterises the underlying core meaning of lah as a request for the hearer to agree with the speaker’s lah-marked assertion. Consistent with this characterisation, the use of lah seems odd when the hearer is already in agreement with the speaker:
Intonation and procedural meaning in CSE lah
(9) (A and B are in a museum) a. A: (pointing at a painting) This is nice! B: Yes, I think so too!2 b. A: (to B)? This is nice lah! B: # Yes, I think so too!
Wong (2004) analyses lah as having three variants3 in his study, characterised by their lexical tone and differing in their pragmatic effect. La3 (pronounced in a low level tone) is the most forceful as the speaker aims to impose her own view on the hearer; this imposition can thus account for how the particle has been observed to indicate impatience and social distance (Kwan-Terry 1978; Bell and Ser 1983; Ler 2005, 2006). La2 (pronounced in a mid-rising tone) is used when the speaker aims to persuade the hearer “without excessive pressure”, and accounts for lah’s effect of softening direct imperatives to persuading utterances. Finally, la4 (pronounced in a high falling tone) conveys a sense that the speaker’s view should have been obvious to the hearer, similar to Kwan-Terry’s (1978) ‘emphatic’ lah. However, although Wong is able to provide an underlying meaning of lah that can account for almost all observed uses of the particle, this meaning seems to focus on lah’s epistemic function of asserting propositions and neglect its social function of marking solidarity and rapport, reducing it to the persuasive aspect of la2. Moreover, it is also unable to account for situations when the hearer does not actually disagree with the lah-marked utterance, e.g.: (10) A: Maybe he goes to the park when he’s free? B: He’s got kidney problem and carries a bag around. A: So he cannot go anywhere lah.
(Wong 2004: 769)
In (10), A comes to the conclusion the person they are referring to cannot go anywhere because of a kidney problem, and states this explicitly to seeks confirmation from B (Wong 2004: 770). Wong (2004) describes lah in (10) as encoding the following meaning:
He cannot go anywhere la4 = I think something now (= I know he cannot go anywhere) I think you don’t think like this I think you could have thought like this before I think it will be good if you think like this
2. Examples that do not come from cited sources are obtained from personal conversations that the author has recorded and edited for clarity. 3. Wong (2004) leaves out one variant of lah in his study as it is rarely used relative to the other three variants. However, this variant will be covered in a later section of this chapter.
63
64 Junwen Lee and Chonghyuck Kim
I think if I say something, you can think like this I say it now (“he cannot go anywhere”)
However, this characterisation of lah is problematic as it supposes that A believes that B does not think that “A knows that [the person in question] cannot go anywhere” and seeks to change B’s mind regarding this, whereas the pragmatic function of the utterance is in fact to “seek confirmation” about whether the person spoken of can move about freely with a kidney problem. Since according to Wong (2004), all three lahs indicate the speech act of assertion (with varying nuances), it is unclear how his account can explain how lah is used to seek confirmation here. 3.4
Lah as a unitary marker
In contrast to Wong (2004), Besemeres and Wierzbicka (2003) propose that lah is a unitary marker indicating the speaker’s expectation that the hearer is able to understand her rather than a request for him to agree with her asserted proposition, and derive the different pragmatic functions of lah from this unitary semantic meaning. For example, when the hearer does not seem to be able to understand the speaker, the speaker’s use of lah to highlight her expectation that the hearer should be able to understand her can be interpreted by the hearer as a plea for understanding, accounting for how lah-marked utterances can seem to be ‘pleading’; alternatively, the speaker may also convey her disapproval of the fact that the hearer does not seem to be able to understand her even though she should, causing lah to sound ‘impatient’. Ler (2005, 2006) also characterises lah as a unitary marker, but proposes instead that it is used to highlight a shared contextual assumption that the hearer is expected to access and accommodate in order to reach an optimally relevant interpretation of the lah-marked utterance. However, her characterisation of lah can be seen as an elaboration of Besemeres and Wierzbicka’s (2003) proposal, as the hearer is presumably able to arrive at the speaker’s intended interpretation of her lah-marked utterance in the first place only by accessing and accommodating an implicated premise that is not explicitly communicated. However, even though these two characterisations can account for all acceptable uses of lah, they seem to be too general, and are unable to explain the unacceptability of the previously raised (9b): (9) b. A: (to B)? This is nice lah! B: # Yes, I think so too!
If we interpret A’s use of lah in (9b) as indicating to B that he should be able to understand why she likes the painting (maybe because it is very well-painted, or
Intonation and procedural meaning in CSE lah
because she likes the subject of the painting) or as indicating that B should access and accommodate the previous reasons as shared assumptions, there is no reason why either A’s use of lah to mark her praise of the painting, or her wish for B to agree with her, should be unacceptable. Unitary characterisations of lah also ignore the relationship between the particle’s intonation and pragmatic function, attributing the different pragmatic functions of lah solely to the interaction between its core meaning and context. As such, they are unable to explain why certain pragmatic functions seem to be strongly linked to the intonation of the particle even if the context is identical. For example, even in the same context, A can only convey that it should be obvious that B should wear an evening dress by pronouncing lah with a high falling tone and not a low level tone: (11) B: What should I wear this evening? a. A: Evening dress la4. (Kwan-Terry 1978: 23) (It should be obvious that you should wear an evening dress.) b. A: ?? Evening dress la3. (It should be obvious that you should wear an evening dress.) c. A: Evening dress la3. (I am asserting the proposition that you should wear an evening dress.)
3.5
A problematic observation – lah as an enumerative marker
Richards and Tay (1977) note that lah in Hokkien can be used to mark enumeration, e.g.: (12) kè-nî gê sî-chueh tak-gê si thâi koe lah, thâi ah lah, bóe ti-kha lah, bóe bah lah. (At New Year time, everyone kills chicken lah, kills duck lah, buys pigs’ trotters lah, buys pork lah.) (Richards and Tay 1977: 151)
Although they did not explicitly recognise CSE lah as having this function as well, my informal respondents as well as my own intuitions as a native speaker of CSE agree that CSE lah can similarly be used to mark items on a list.4 However, this function of lah is curiously absent in characterisations of the particle, and cannot
4. These judgements were obtained through convenience sampling from five university undergraduates in their twenties who are bilingual in Mandarin and Standard Singapore English, and who use CSE in their social interactions. They do not have any training in linguistics or pragmatics. All agreed that when pronounced with a high level tone, lah can be used to mark items on a list in the CSE utterance “At New Year time, everyone kill chicken lah, kill duck lah, buy pigs’ trotters lah, buy pork lah”, i.e. the CSE equivalent of (12).
65
66 Junwen Lee and Chonghyuck Kim
be accounted for by either Ler (2005, 2006), Besemeres and Wierzbicka (2003) or Wong (2004) – it is unclear what shared assumption the speaker is highlighting to the hearer, or why the speaker has to indicate that the hearer can understand that people perform such actions during the New Year; the speaker is also not attempting to convince the hearer about anything or impose his view on the hearer. A complete characterisation of lah must thus be able to account for this function, together with the other functions listed throughout this section. 4. A relevance-theoretic characterisation of lah 4.1
Lah as an indicator of high strength
One of the pragmatic functions of lah that is commonly documented is persuasion (Besemeres and Wierzbicka 2003; Kwan-Terry 1978; Ler 2006; Loke and Low 1988; Wong 2004): (13) B: Are you going to the party tomorrow? a. A: I cannot make it. Sorry. (I may have no good reason for missing the party.) b. A: I cannot make it lah. Sorry. (I am missing the party because of a very good reason.)
In (13b), A seems to be using lah to persuade B that she cannot make it to the party. More specifically, the difference between (13a) and (13b) seems to be that (13a) leaves open the possibility that A has no justifiable reason for missing the party, while (13b) suggests that A has a good reason for doing so. In relevance-theoretic terms, when a speaker uses lah to mark persuasive utterances, she is trying to convince the hearer of the reliability of her utterances, i.e. to increase the hearer’s assessment of the epistemic strength of the proposition(s) asserted. As such, we propose that the procedural meaning encoded by lah is an indication to the hearer that the lah-marked proposition has high epistemic strength. To illustrate, (13b) can be analysed as the inferential relation (14): (14) a. Premise 1: If A has something on at the time of the party, A cannot go for the party. b. Premise 2: A has something on at the time of the party. c. Conclusion: A cannot go for the party.
Using lah, A indicates to B that the epistemic strength of (14c) is high; if B accepts A’s assertion, B would have to assume that both (14a) and (14b) are also of high strength, as the epistemic strength of a deductive implication depends on the
Intonation and procedural meaning in CSE lah
strength of the premises from which it is derived (Sperber & Wilson 1995: 110). Although B may have to just trust A and accommodate this assumption, A can also explicitly justify the high strength of (14b), e.g. by following (13b) with “I have a test to study for tomorrow.”5 However, this explanation is still not enough to explain why the previously raised (9b) (reproduced below as (15a)) is awkward while (15b) is acceptable: (15) (A and B are in a museum) a. A: (pointing at a painting)? This is nice lah! B: # Yes, I think so too! b. A: (trying to convince B) This is nice lah! I can’t see why you don’t like it – I think it’s absolutely perfect!
The contrast between (15a) and (15b) can be accounted for when we recall the concept of optimal relevance – the cognitive benefits to be derived must not require more cognitive effort than they are worth. (15b) is therefore acceptable as the additional cognitive effort required by B to process lah is rewarded by a potential increase in the epistemic strength of the asserted proposition, since A may succeed in convincing B that the painting is a good one. However, if B is already prepared to take A’s assertion as certain, the additional cognitive effort in processing lah will not be justified by the benefits obtained (since the proposition “This painting is nice” would be of high strength anyway) and so using lah would violate the presumption of optimal relevance in this case. This restriction therefore explains why lah is most often used when the speaker believes the hearer to disagree with her utterance or hold the opposite view, as it is an indication for the hearer to increase the low epistemic strength that he would otherwise assign to her utterance. From the hearer’s perspective, whether he should accept the lah-marked proposition as true may be motivated by how well it agrees with either his own beliefs or subsequent utterances by the speaker, but another important factor would be the speaker’s relationship with the hearer. For instance, consider (16): (16) A: Quite nice lah. (I’m your friend; consider my opinion.) p: [It is] quite nice.
(Ler 2006: 156)
In (16), A uses lah to indicate to B the high epistemic strength of the proposition p. In order to decide whether or not to revise p, B may consider factors such as A’s general reliability, but may also take into account his relationship with A – in this
5. Consistent with this, speakers do commonly follow a lah-marked negation with the reason for the negation.
67
68 Junwen Lee and Chonghyuck Kim
case, Ler (2006) suggests that the fact that A is B’s friend is an important factor that determines whether B accepts A’s opinion. Although the speaker’s authority might be another motivating factor for the hearer to revise the strength of the lah-marked proposition, lah’s emphasis of social closeness seems to preclude the interpretation that the speaker is using her authority to compel the hearer to accept her assertion (e.g. as a teacher to a student, parent to child, etc.) (Richards and Tay 1977). For example, A is unable to use lah in (17a) when she tries to use her authority as B’s boss to convince B to accept her proposition; in contrast, lah in (17b) is acceptable when A explicitly states that she is not arguing from a position of authority. (17) a. A: ( to B) ? As your boss, I’m telling you that doing this would be good for you lah. b. A: (to B) As your friend, I’m telling you that doing this would be good for you lah.
In this way, the observation that lah indicates social intimacy or solidarity (Bell and Ser 1983; Kwan-Terry 1978; Kwan-Terry 1992; Loke and Low 1988; Richards and Tay 1977) is accounted for, as lah highlights the social closeness between speaker and hearer in particular when the hearer considers whether to accept the lah-marked proposition as being of high strength. Lah’s de-emphasising of speaker authority will be discussed again in the following section. Based on the previous examples, one might relate the meaning of lah to Sperber et al.’s (2010) conception of epistemic vigilance as a mental module that is activated by the hearer in determining the truth of the speaker’s assertions, and conclude that lah indicates the speaker’s wish for the hearer to lower his epistemic vigilance and accept the lah-marked assertions as true. It is also unsurprising that lah highlights social intimacy or solidarity between speaker and hearer, since the hearer is more likely to lower his epistemic vigilance and assume that the speaker is trustworthy if they have a close social relationship. However, while we agree that lah does indeed have this function when it marks assertions, we argue that it does not directly indicate that the speaker wishes the hearer to lower his epistemic vigilance, as the particle is interpreted in different ways when marking other higher-level explicatures that are expressed by the speaker’s utterance, such as in the examples involving imperatives in the following section. 4.2
Lah and imperatives
Ler (2005) explains how lah-marked imperatives can appear to be persuading or pleading as the speaker attempts to convey to the hearer the shared assumption
Intonation and procedural meaning in CSE lah
behind her utterance. If the hearer is unable to recognise the shared assumption, the speaker’s insistence that he do so may also convey impatience or annoyance. This contrast is shown in (18): (18) a. A: (to B) Come with us lah. (More polite) b. A: (to B) Eat your food lah. (More impatient)
(Ler 2005: 297) (Ler 2005: 292)
In (18a), the speaker uses lah to indicate to the hearer that it will be good for him to go along with them; in (18b), B does not appear to recognise that he should finish his food faster as they are late, despite the speaker’s use of lah, and this results in the sense of impatience in the imperative. However, it is unclear how the hearer can focus on the precise shared assumption that the speaker wishes to be highlighted each time, given how disparate the assumptions seem to be for each situation, and we believe that the characterisation of lah in the previous section – that lah asserts the high epistemic strength of the preceding proposition – can account better for its function in imperatives. According to relevance theory, the uttered imperative (18a) makes manifest the fact that it is potential and desirable to either the speaker or hearer that the hearer makes real the situation described in the imperative (that the hearer follows the speaker), and this proposition serves as a premise for the hearer to decide whether to obey the speaker or not. The speaker may increase the likelihood for compliance by increasing the force of the uttered imperative, which seems to vary with an emphasis on the speaker’s authority to compel or threaten the hearer: (19) a. A: Put your hands up. b. A: I am the police. Put your hands up. c. A: Put your hands up or I’ll shoot.
Compared to (19a), (19b) has greater imperative force as A indicates to the hearer that she is the police, and thus has sufficient authority to compel the hearer to obey her imperative; (19c) has still greater imperative force as A motivates the hearer to obey the imperative by using a physical threat. In all these cases, A’s desire for the hearer to do what she wants him to is stated but not emphasised. Conversely, when a speaker emphasises her desire for the hearer to make the described situation actual, it implies that the speaker either has no authority to compel the hearer or chooses not to use her authority; in fact, there seems to be an inverse relationship between the extent of the speaker’s desire that is conveyed and the force of the imperative. Compared to (20a), (20c) and (20d) are markedly lower in imperative force, functioning in fact as pleas:
69
70 Junwen Lee and Chonghyuck Kim
(20) a. A: Put your hands up. (Command) b. A: I want you to put your hands up. (Command/Request) c. A: I really want you to put your hands up. (Request/Plea) d. A: I really want so badly for you to put your hands up. (Plea)
We propose that when lah is attached to imperatives, it indicates that the strength of the higher-level explicature expressed by the imperative, i.e. that it is potential and desirable to either the speaker or hearer that the hearer follows the speaker, is high. However, since the situation’s potentiality is generally presupposed in an imperative, lah emphasises solely the speaker’s desire for the situation described to be made actual. By highlighting the speaker’s desire rather than her authority over the hearer, the implication is that the speaker cannot compel the hearer to obey the imperative, but can only hope to persuade him to fulfil her desire. Since the speaker’s authority is no longer a motivating factor to the hearer, the hearer will most likely fulfil her desire if her happiness is important to him, e.g. in situations where the speaker and hearer have a close social relationship. In support of this, we find that speakers will often highlight the close social relationship between the hearers and themselves in order to better convince the former to accede to their pleas, e.g. “Come with us (because I am your friend)” or “Help me (because I am your friend).” Therefore, rather than attributing the differences in how lah changes the force of imperatives to the assumptions that the hearer is supposed to accept (Ler 2005), we argue that they can be more accurately accounted for by differences in the social relationship between speaker and hearer and the nature of the speaker’s desire for the situation described to be made actual. For example, lah emphasises in (18a) (reproduced as (21a)) the speaker’s desire for the hearer to go along with them: (21) a. A: (to B) Come with us lah. (More polite) b. A: (to B) Eat your food lah. (More impatient)
(Ler 2005: 297) (Ler 2005: 292)
If we assume that A has no authority over B, e.g. if they are friends, B would recognise that A has no means to compel him into fulfilling her desire, and the particle would weaken the force of the imperative. The pragmatic effect of (21a) would be to indicate to B that A wishes to persuade B to go with her on the basis of their friendship, since that would be the strongest motivation for B to fulfil A’s desire. In (21b), lah emphasises the speaker’s desire for the hearer to eat faster. However, Ler (2005) describes both A and B as having to go off soon for an appointment, so B knows that he needs to eat his food quickly in order not to be late. In this situation, B is not just motivated to fulfil A’s desire based on their friendship, but also for his own self-interest, and so interprets the lah-marked imperative as
Intonation and procedural meaning in CSE lah
a command or request rather than a plea. Like Ler (2005), we analyse the impatience that is conveyed in (21b) as an emotional state that is specific to the speaker in that situation, which is separate from the core meaning or even pragmatic function of lah. However, lah’s emphasis of A’s desire for B to eat faster does result in an increased sense of urgency that serves to complement A’s tone of impatience. The function of lah to persuade the hearer to carry out the lah-marked imperative by emphasising the close relationship between the speaker and the hearer is also consistent with the observation in the previous section that lah de- emphasises the speaker’s authority when indicating the lah-marked utterance’s high strength. In both cases, the speaker indicates to the hearer that her authority is to be irrelevant in the exchange, which would in turn highlight the speaker’s solidarity or rapport with the hearer in convincing the hearer to either believe the speaker’s utterance or to obey her imperative. 5. The effect of intonation contour on lah As previously stated, Wong (2004) distinguishes three pitch contours that lah is consistently pronounced in: a low level pitch, a mid-rising pitch and a high falling pitch, also identified by Kwan-Terry (1978) and Loke and Low (1988). However, Wong also notes a fourth high level variant that is of relatively low frequency, which will be covered in Section 5.1.4. Following his terminology, we propose in this chapter that the different pitches that lah can be pronounced in are not lexical tones that distinguish between different variants of the particle, but are intonational differences that are independent of the particle. This approach thus attempts to reconcile maintaining a unitary meaning of lah with acknowledging the correspondence between its different pragmatic functions and the pitch with which the particle is pronounced. Clark and Lindsey (1990) suggest that aspects of intonational meaning are constant across languages, and Clark (1991) proposes that this universality can be accounted for if we posit an iconic aspect to intonational meaning – a rising tone can signal affective non-resolution as it is universally associated with tension in the speaker, while a falling tone can signal affective resolution as it is associated with a relaxed emotional state. The affective state of the speaker can in turn indicate the grammatical mood of her utterance, as illustrated in (22): (22) (to B, pointing to the shelf) a. A: The box goes here? (Rising tone) b. A: The box goes here. (Falling tone)
71
72
Junwen Lee and Chonghyuck Kim
Relating intonation to (procedural) meaning, Clark (1991) proposes that a rising tone encodes the procedural constraint that the speaker is not entertaining the thought of which the proposition expressed is an interpretation. As such, (22a) with a rising tone will tend to be interpreted as a question, in which A indicates that she has not accepted the thought that the box should be placed on the shelf, and would require the hearer to either confirm or disprove the truth of this proposition. Conversely, a falling tone encodes the procedural constraint that the speaker is entertaining the thought of which the proposition expressed is an interpretation (Clark 1991: 195), and thus (22b) with a falling tone will tend to be interpreted as a declarative or imperative, in which A is certain that the proposition “The box should be placed on the shelf” is true, and is asserting it either as a fact or as a command or request to B. 5.1
Low and falling pitch contours
The affective resolution that Clark (1991) proposes a low or falling tone signals correctly accounts for the stronger propositional nature of the low level and high falling pitch variants of lah (Loke and Low 1988: 156; Wong 2004: 773). For these two variants, lah’s core meaning indicates to the hearer the high epistemic strength of the lah-marked utterance, and the low and falling tones emphasise the speaker’s belief in or commitment to this assertion. 5.1.1 Low level lah To illustrate how the effect of a low level intonation interacts with the core meaning of lah, consider (23): (23) A: (to B) No need to count lah. I’m sure that the number is right. (Kwan-Terry 1992: 64) p: [There is] no need to count.
In (23), lah indicates to B that the lah-marked proposition, p, is of high epistemic strength, and the low level intonation indicates that A has taken this thought to be true. As such, B will interpret the utterance as A’s attempt to convince B about the truth of p. However, the force of such persuasion depends on contextual factors – as explained previously in the previous section, whether lah is interpreted as “contribut[ing] an attitude of gentle persuasion to the utterance” (Kwan-Terry 1992: 64) or a stronger ‘imposition’ of the speaker’s views onto the hearer (Wong 2004: 764) will depend on the emotional tone that the speaker conveys in her utterance, as well as the relationship between speaker and hearer.
Intonation and procedural meaning in CSE lah
However, recall (10) (reproduced below as (24)): (24) A: Maybe he goes to the park when he’s free? B: He’s got kidney problem and carries a bag around. A: So he cannot go anywhere lah.
(Wong 2004: 769)
This is problematic for characterisations of lah that analyse the particle as asserting a proposition for the hearer’s acceptance, as A is trying to confirm her hypothesis with B rather than convincing B about it. In such cases, we propose that lah in (25) is interpreted in the same way as in (26): (25) A: So he cannot go anywhere lah. p: He cannot go anywhere.
(Wong 2004: 769)
(26) (A and B are discussing B’s recent holiday.) B: Ya we went to a lot of places and ate a lot of good food. A: So you had a good time lah. B: Ya.
In (26), the proposition that A marks with lah is not a new one but rather a summary or conclusion of what B had said earlier. Since the epistemic strength of a conclusion follows the strength of the premises from which it is derived (Sperber and Wilson 1995: 110), A’s conclusion (if correct) would be of high strength given the high strength of B’s earlier propositions (e.g. that he had fun going to different places and eating good food). A’s statement in (26) can thus be seen as a description of the conclusion that she has derived in her own cognitive environment, with lah indicating the high strength that A assigns to it – however, as the inferential route that A takes may be wrong, she would require B to either confirm or disprove this conclusion. Similarly, A concludes in (24) that, based on B’s words, the man with a kidney problem would not be able to go anywhere. In order to confirm her hypothesis, she presents it to B, using lah to indicate the high strength that she has assigned to it. The use of lah in such cases is thus accounted for: rather than changing the epistemic strength of a proposition in the hearer’s cognitive environment, lah indicates the high strength of a proposition from the speaker’s own cognitive environment, which the hearer then either confirms or disproves. 5.1.2 High falling lah Although both low level and high falling variants indicate the speaker’s assertion of the high epistemic strength of the lah-marked proposition, the high falling variant would convey greater pragmatic emphasis as its initial high pitch indicates greater stress than the low pitch of the former. The hearer may then interpret this
73
74
Junwen Lee and Chonghyuck Kim
greater pragmatic emphasis of the proposition’s high strength as indicating obviousness,6 accounting for Wong’s (2004) characterisation of la4 as a specific indicator of obviousness. Both Kwan-Terry’s (1978) characterisation of the contracted form lâ as conveying greater authority as well as irritation and Loke and Low’s (1988) observation of high lah of long duration as expressing irritation can also be explained, as the greater stress indicated by the high pitch contour may reflect these emotional stances from the speaker even though they are not inherent in the particle’s meaning. 5.1.3 Mid-rising lah As stated earlier, Clark (1991: 195) proposes that a rising intonation signals affective non-resolution, or that the speaker is not entertaining the thought of which the proposition expressed is an interpretation. Although a rising intonation tends to signal questions, rather than interpreting utterances marked by mid-rising lah as true questions, we propose that they are interpreted in a way similar to rhetorical questions, e.g. (27b): (27) a. A: (to B) I cannot do it lah. I really cannot. (Wong 2004: 771) p: A cannot do what B wants. b. A: (to B) I cannot do it, okay? I really cannot. i. Unitary meaning of lah: p is high-strength. ii. Meaning of mid-rising lah: The thought that [p is high-strength] is desirable to the hearer.
In contrast with a true request for information, in which the speaker requires information from the hearer to confirm or disprove the proposition that would be relevant to herself, the answer of a rhetorical question is already known to both speaker and hearer; in effect, the hearer is led to recognise that what the speaker is indirectly asserting is obviously true given the ‘obvious’ answer of the rhetorical question. Similarly, in (27a), A indicates that the assertion of p being of high epistemic strength is relevant to B; however, rather than asserting p as being of high strength directly, A leaves it to B to arrive at this conclusion himself. This accounts for Wong’s observation that mid-rising lah is “less effectual […] as a persuasive tool” (2004: 772) than the low level variant. While he sees this difference in pragmatic force as the straightforward result of a difference in meaning between mid-rising and low level lah, we propose that it is due to a difference in grammatical mood arising from the different intonations of the utterances: 6. In fact, the intonational contour of high falling lah seems to match Sag and Liberman’s (1975) ‘surprise-redundancy contour’ in Standard American English in both intonational contour as well as its pragmatic function of indicating redundancy and obviousness.
Intonation and procedural meaning in CSE lah
(28) (A is trying to stop B from buying a new car.) a. A: (to B) You are mad! You can’t afford a new car! p: B is mad in wanting to buy a new car. b. A: (to B) Are you mad? You can’t afford a new car!
Comparing between (28a) and (28b), we see the same difference observed by Wong (2004): while A is directly asserting in (28a) that B is mad, her rhetorical question in (28b) appears to be less forceful as she is leading B to recognise for himself the obvious answer to the question, i.e. that he is mad. However, mid- rising lah does not just assert a proposition but the proposition’s high epistemic strength, and thus its pragmatic function is to lead the hearer to recognise the lahmarked proposition’s high strength by himself and reject the opposite viewpoint, i.e. to convince or persuade the hearer of the truth of the proposition. 5.1.4 High level lah Wong (2004) also notes that lah can be pronounced in a high level pitch, but did not include it in his study due to its relatively low frequency of use (2004: 762). However, any unified characterisation of lah must be able to accommodate its use, which, based on informal respondents as well as my own intuitions as a native speaker, superficially appears to be one of enumeration. Richards and Tay (1977) documents an example in Hokkien that nevertheless maintains its acceptability when translated into CSE: (29) a. kè-nî gê sî-chueh tak-gê si thâi koe lah, thâi ah lah, bóe ti-kha lah, bóe bah lah. b. At New Year time, everyone kills chicken lah, kills duck lah, buys pigs’ trotters lah, buys pork lah… all these things.7
Although one might be led to account for this function of lah by including some sort of enumerative feature in its semantic meaning, this is in fact unnecessary as the utterance can be interpreted as a list even without lah. The factor contributing to the interpretation of (29) and (30) as lists seems to be the high level intonation of the noun phrases (“chicken”, “duck”, etc.). Goh characterises the level tone in CSE as being associated with “short tone units and indicating incomplete information” (2000: 40), and gives the following as an example:
7. The phrase “all these things” were added to the CSE translation as it is unacceptable for high level lah to end an utterance. However, the meaning of the utterance (an enumeration of some of the ways people prepare for Chinese New Year) remains unchanged.
75
76
Junwen Lee and Chonghyuck Kim
(30) //→ so i would classiFY YOU// → as the TRUE// → SMART// → consuMER// (Stresses capitalised, ‘//’ indicates a tone group and ‘→’ indicates a level tone within the tone group.) (Goh 2000: 40)
Similarly, Halliday characterises a mid-level intonation in British Standard English as expressing “some form of dependence or incompleteness” (1970: 24). As such, we propose that, in relevance-theoretic terms, the high level pitch indicates to the hearer that the utterance is not intended to supply the desired relevance by itself, but must be interpreted in parallel with subsequent utterances. Since the effect of the high level pitch conveys a sense of incompleteness, or a sense that the utterance is an example, we would expect that this sense would be emphasised by lah. How lah functions in (29) (reproduced below as (31)) can thus be analysed as follows: (31) At New Year time, everyone kills chicken lah, kills duck lah, buys pigs’ trotters lah, buys pork lah… all these things. (During Chinese New Year, everyone performs tasks like killing chickens (as an individual, non-exhaustive example), killing ducks (as another individual, non-exhaustive example), buying pigs’ trotters (as another individual, non- exhaustive example) and buying pork (as another individual, non-exhaustive example).) p: p1: Everyone kills chicken [during Chinese New Year]; p2: Everyone kills duck [during Chinese New Year]; p3: Everyone buys pigs’ trotters [during Chinese New Year]; p4: Everyone buys pork [during Chinese New Year] a. Unitary meaning of lah: p is high-strength. b. Meaning of high level lah: [p is incomplete] is high-strength.
In (31), killing chickens, killing ducks, buying pigs’ trotters, and buying pork are interpreted as four individual, non-exhaustive exemplars of tasks that people perform during Chinese New Year, which together may suggest other related tasks that people may perform in preparation for Chinese New Year. Lah in a high level tone can thus be characterised as emphasising to the hearer the incompleteness of the lah-marked proposition, prompting him to access other propositions of which the lah-marked proposition is an exemplar. (32) provides further support for this analysis of high level lah: (32) (B is asking A about the secretary training course she had attended recently.) B: What was the training like? a. A: We learnt how to answer calls lah, take notes lah, all these things.
Intonation and procedural meaning in CSE lah
b. A: # We just learnt three things – how to answer calls lah, take notes lah, (and) prepare for meetings. c. A: We just learnt three things – how to answer calls (High level pitch), take notes (High level pitch), and prepare for meetings.
Even though A responded to B’s question with a list in (32a), she is not telling B that the training only involved answering calls and taking notes, but that the two are non-exhaustive examples of what the training involved. In contrast, we see that while a high level intonation is still acceptable when used in an exhaustive list in (32c), high level lah is unable to enumerate an exhaustive list in (32b). 6. Conclusion Addressing the issue of lah’s multiple pragmatic functions, this chapter proposes that its pragmatic functions can be analysed as the results of the interaction between its intonation and its core procedural meaning. Instead of positing four lexical variants of lah, we propose that the particle encodes the procedural instruction to indicate to the hearer that the lah-marked proposition is of high epistemic strength. The ‘impositional’ nature when lah is pronounced with a low level intonational contour is attributed to the authority associated with the low level pitch contour, while the sharp fall in pitch of lah pronounced with a high falling contour may reflect greater pragmatic emphasis and correspondingly a sense that the high strength of a proposition is so great it becomes obvious. The relatively weaker force of lah in a mid-rising contour can in turn be explained by its similarity to rhetorical questions. The proposed framework can also accommodate high level lah, which functions as an enumerative marker; this use of lah has been ignored in almost all previous literature that had attempted to describe the particle. This proposal is similar to Padilla Cruz’s (2009) suggestion that interjections interact with intonation in contributing to the higher-level explicatures expressed by an utterance. However, while it is unclear if interjections do encode procedural meaning, it seems reasonable to assume that lah and other discourse particles do. Therefore, while Padilla Cruz (2009) proposes that interjections may serve to reinforce the higher-level explicatures that the hearer has already constructed based on the intonation of the utterance and other suprasegmental features, we claim here that it is intonation that modifies the hearer’s interpretation of the lahmarked utterance based on the core meaning of lah. Developing this proposal further would hopefully contribute to the growing body of research on how intonation in general contributes to hearers’ construction of higher-level explicatures
77
78
Junwen Lee and Chonghyuck Kim
(e.g. Clark 1991; Clark and Lindsey 1990; Escandell Vidal 1998; Fretheim 1998; Padilla Cruz 2009; Wilson and Wharton 2006). We believe that this approach is especially applicable to the study of CSE particles as many can be pronounced with different pitch contours. For example, the CSE particle hor has been described as a topic marker or a marker to elicit agreement when it is pronounced with a rising pitch, and as an emphatic marker when pronounced with a falling tone (Kim and Wee 2009). Assuming that the effect of intonation on utterances that we propose here is consistent within CSE, it would be interesting to learn to what extent we can apply them to explain the different effects of hor. How our proposal applies to particles that can only be pronounced in one way is also another potential area of research. Wong (2004) notes that many CSE loan words from Chinese languages retain lexical tone from their language of origin, and so it is possible that the hearer will interpret the intonation of particles that cannot be pronounced with multiple tones as a ‘default’ tone that does not carry intonational meaning. Alternatively, another possibility is that the semantic meaning of a particle may restrict the intonation contours it can take. This could thus prove to be another fruitful direction for further research.
References Bao, Zhiming. 2001. “The Origins of Empty Categories in Singapore English.” Journal of Pidgin and Creole Languages 16: 275–319. doi: 10.1075/jpcl.16.2.03zhi Bell, Roger T., and Larry Peng Quee Ser. 1983. “‘To-day la?’ ‘Tomorrow lah!’; the LA Particle in Singapore English.” RELC Journal 14: 1–18. doi: 10.1177/003368828301400201 Besemeres, Mary, and Anna Wierzbicka. 2003. “Pragmatics and Cognition: The Meaning of the Particle ‘lah’ in Singapore English.” Pragmatics and Cognition 11: 1–36. doi: 10.1075/pc.11.1.03bes
Clark, Billy. 1991. Relevance Theory and the Semantics of Non-declarative Sentences. PhD diss., University College London. Clark, Billy, and Geoff Lindsey. 1990. “Intonation, Grammar and Utterance Interpretation: Evidence from English Exclamatory-Inversions.” UCL Working Papers in Linguistics 2: 32–51. Escandell Vidal, María Victoria. 1998. “Intonation and Procedural Encoding: The Case of Spanish Interrogatives.” In Current Issues in Relevance Theory., ed. by Villy Rouchota and Andreas H. Jucker, 169–204. Amsterdam: John Benjamins. doi: 10.1075/pbns.58.09esc Fretheim, Thorstein. 1998. “Intonation and the Procedural Encoding of Attributed Thoughts: The Case of Norwegian Negative Interrogatives.” In Current Issues in Relevance Theory, ed. by Villy Rouchota and Andreas H. Jucker, 205–236. Amsterdam: John Benjamins. doi: 10.1075/pbns.58.10fre
Intonation and procedural meaning in CSE lah
Goh, Christine. 2000. “A Discourse Approach to the Description of Intonation in Singapore English.” In The English Language in Singapore: Research on Pronunciation, ed. by A. Brown, D. Deterding, and E. L. Low, 35–45. Singapore: Singapore Association for Applied Linguistics. Gupta, Anthea Fraser. 1992. “The Pragmatic Particles of Singapore Colloquial English.” Journal of Pragmatics 18: 31–57. doi: 10.1016/0378-2166(92)90106-L Gupta, Anthea Fraser. 2006. “Epistemic Modalities and the Discourse Particles of Singapore.” In Approaches to Discourse Particles, ed. by K. Fischer, 243–263. Amsterdam: Elsevier. Halliday, M. A. K. 1970. A Course in Spoken English: Intonation. Oxford: Oxford University Press. Kim, Chonghyuck, and Lionel Wee. 2009. “Resolving the Paradox of Singapore English hor.” English World-Wide 30: 241–261. doi: 10.1075/eww.30.3.01kim Kwan-Terry, Anna. 1978. “The Meaning and the Source of the la and the what Particles in Singapore English.” RELC Journal 9: 22–36. doi: 10.1177/003368827800900202 Kwan-Terry, Anna. 1992. “Towards a Dictionary of Singapore English – Issues Relating to Making Entries for Particles in Singapore English.” In Words in a Cultural Context, ed. by A. Pakir, 62–72. Singapore: Unipress. Ler, Vivien. 2005. An In-depth Study of Discourse Particles in Singapore English. PhD diss., National University of Singapore. Ler, Vivien. 2006. “A Relevance-theoretic Approach to Discourse Particles in Singapore English.” In Approaches to Discourse Particles, ed. by K. Fischer, 149–166. Amsterdam; Oxford: Elsevier. Loke, Kit-Ken, and Mei-Yin Johna Low. 1988. “A Proposed Descriptive Framework for the Pragmatic Meanings of the Particle la in Colloquial Singaporean English.” In Asian-Pacific Papers, ed. by B. McCarthy, 150–161. Applied Linguistics Association of Australia. Padilla Cruz, Manuel. 2009. “Might Interjections Encode Concepts? More Questions than Answers.” Lodz Papers in Pragmatics 5: 241–270. Pakir, Anne. 1992. “Dictionary Entries for Discourse Particles.” In Words in a Cultural Context, ed. by A. Pakir, 143–152. Singapore: Unipress. Platt, John T., and Mian Lian Ho. 1989. “Discourse Particles in Singaporean English: Substratum Influences and Universals.” World Englishes 8: 215–221. doi: 10.1111/j.1467-971X.1989.tb00656.x
Richards, Jack C., and Mary W. J. Tay. 1977. “The la Particle in Singapore English.” In The English Language in Singapore, ed. by W. Crewe, 141–156. Singapore: Eastern Universities Press. Sag, Ivan, and Mark Liberman. 1975. “The Intonational Disambiguation of Indirect Speech acts.” Paper presented at the Papers from the Eleventh Regional Meeting. Sperber, Dan, Fabrice Clément, Christophe Heintz, Olivier Mascaro, Hugo Mercier, Gloria Origgi, and Dierdre Wilson. 2010. “Epistemic Vigilance.” Mind and Language 25: 359–393. doi: 10.1111/j.1468-0017.2010.01394.x
Sperber, Dan, and Dierdre Wilson. 1995. Relevance. Communication and Cognition. Oxford: Blackwell. Wee, Lionel. 2002. “Lor in Colloquial Singapore English.” Journal of Pragmatics 34: 711–725. doi: 10.1016/S0378-2166(01)00057-1
79
80 Junwen Lee and Chonghyuck Kim
Wee, Lionel. 2004. “Reduplication and Discourse Particles.” In Singapore English: A Grammatical Description, ed. by L. Lim, 105–126. Amsterdam: John Benjamins. doi: 10.1075/veaw.g33.07wee
Wilson, Deirdre, and Tim Wharton. 2006. “Relevance and Prosody.” Journal of Pragmatics 38: 1559–1579. doi: 10.1016/j.pragma.2005.04.012 Wong, Jock. 2004. “The Particles of Singapore English: A Semantic and Cultural Interpretation.” Journal of Pragmatics 36: 739–793. doi: 10.1016/S0378-2166(03)00070-5
Reference assignment in pronominal argument languages A relevance-theoretic perspective Helga Schröder
University of Nairobi
Using data from Toposa and Kiswahili, this article demonstrates that the reference assignment in pronominal languages does not only constrain the process of identifying the intended referent, but the pronominal marking on the verb also assists with disambiguating whether a nominal expression has attributive or referential value. A nominal expression enters the discourse as underspecified regarding for referential/attributive meaning and therefore needs to be enriched through saturation. The pronominal marking on the verb triggers the process of enrichment. If the overt pronoun occurs in addition to the default incorporated pronoun on the verb, it carries the procedural marker of identificational or contrastive focus, similar to the contrastive stress in English. Keywords: pronominal argument languages, conceptual/procedural meaning, referential nominalisation, attributive nominalisation, reference assignment
1. Introduction Reference assignment is understood in relevance theory (RT) as an online process of decoding and enriching an explicature (Sperber and Wilson 1986/1995). It is a procedure that contributes to the computational side in utterance interpretation and it is truth-conditional, because it enriches the propositional content of an explicature. The procedural status of reference assignment has been well documented (Sperber and Wilson 1986/1995; Wilson 1992; Blakemore 2002; Carston 2002; Cram and Hedley 2005; Hedley 2005, 2007; Wilson 2011). In this paper I want to show that reference assignment in pronominal argument languages requires a different analysis than languages where the pronoun is
doi 10.1075/pbns.268.04sch © 2016 John Benjamins Publishing Company
82
Helga Schröder
overt and not integrated (like English).1 Pronominal argument languages – a term coined by Jelinek (1984) – have an incorporated pronoun encoded as a default construction on virtually every verb. Speaking in relevance-theoretic terms, these languages encode a pronominal reference in every utterance via the verb. These references are agreement affixes on the verb. In this paper I shall refer to them as incorporated pronouns.2 Many of these pronominal argument languages are found in Africa.3 Also some more widely known languages like Greek, Spanish, Italian and Arabic have incorporated pronouns. Pronominal argument languages posit a challenge to the relevance-theoretical framework. Firstly, the procedural interpretations of pronouns for those that are incorporated and for those that are free standing differ. In this paper, I will claim that the incorporated pronouns that are found in pronominal argument languages do not only have the function of identifying the referent, as freestanding pronouns do. In addition, they carry the function of enriching nominal expressions for the value of definiteness and indefiniteness. Secondly, these languages employ a double strategy, which allows for the simultaneous occurrence of an overt pronoun and an incorporated one. I want to show that in instances where the pronoun and the incorporated pronoun both occur, the unbound pronoun does not only trigger reference assignment which is already taken care of by the incorporated pronoun, but it carries also a focus feature. Thus, the focus applied to that free-standing pronoun is a procedural computational instruction that has a double function: it helps to identify the referent, but it is also a stylistic feature that expresses either contrastive focus or focus of identification, similar to what Sperber and Wilson (1995: 202–217) and Wilson and Wharton (2006: 12–13)4 have described as contrastive stress.5 I shall
1. I wish to thank two anonymous reviewers for their valuable comments and suggestions. 2. I call these pronouns ‘incorporated’, because they are not free standing, instead, they are incorporated into the verbal complex as affixes. I also use for free standing pronouns the term unbound pronouns. Both free standing and unbound pronouns are contrasted with the term incorporated pronouns, where the latter refers to the affixes in the verb and the former to the overt pronouns. 3. Integrated pronominal affixes are known to occur in Nilotic languages spoken in East Africa, and for Bantu languages spoken in Sub-Saharan Africa and parts of West Africa. 4. Page numbering follows the Internet version, not the printed one. 5. Contrastive stress is a prosodic device that shows the focus in an utterance through placing a stress on that constituent that the speaker wants the hearer to draw special attention to and secondly it indicates contrast that means the hearer implicitly makes a choice between several items in his mind. For example a speaker can say. Peter went to the bank vs. Peter went to the bank. In the first utterance the stress is placed on Peter so the speaker intends to say that Peter
Reference assignment in pronominal argument languages
be using data from Toposa, an Eastern Nilotic language spoken in South Sudan, and from Kiswahili, a Bantu language widely used across East Africa. This paper has five sections. The first section will discuss the conceptual procedural distinction in relevance theory. The second section will address the status of the incorporated pronoun in pronominal languages from a relevance-theoretic point of view. The third section will deal with the disambiguation of nominal expressions. The fourth section will consider the occurrence of the overt pronoun and its communicative effect. Finally, the fifth section will draw some theoretical conclusions. 2. Conceptual and procedural distinction in relevance theory According to RT, there is a difference between procedural and conceptual meaning. Conceptual meaning typically encodes lexical information, content words that feed into the conceptual representation system of utterance comprehension. So conceptual meaning represents the ‘language of thought’, a term coined and discussed in Fodor (1975, 1998). The procedural meaning on the other hand is semantically ‘empty’, it provides instruction to the hearer how to process the conceptual information provided by the stimuli in the online process of comprehension. So procedural information supports the inferential process. It helps the hearer to constrain the constructing of cognitive effects in search for a relevant single interpretation of the stimulus. The distinction between the conceptual and procedural meaning has been well discussed and established with various syntactic categories. Blakemore (1987) was the first to point out that connectors such as ‘but’, ‘because’, ‘so’, ‘also’ and ‘after all’ guide the inferential process by giving instructions to how to obtain cognitive effects in the search for relevance, see the following example for ‘because’: (1) Mary could not buy the expensive dress because she did not have enough money.
In this utterance ‘because’ guides the hearer to find the reason why Mary could not buy the dress. The reason is that she cannot pay the bill for the expensive dress because she does not have the money. The conjunction ‘because’ connecting the first and second sentence contributes to the propositional content of the utterance through providing a reason for the fact that ‘Mary could not buy the dress’.
and nobody else went to the bank. In the second utterance the intention of the speaker is to express that Peter went to the bank and that he did not go to a shop for example.
83
84
Helga Schröder
That connectors are showing procedural effects was further discussed by Blass (1990). She showed that the interpretive use marker re ‘and’ the particle ma in Sissala contribute to the procedural inferential process. Iten (1997) showed the procedural effects of ‘because’ and ‘although’ and Iten (2002) elaborates on the procedures of ‘even if ’ and ‘even’. Hall (2004) also discussed ‘but’. She argued that ‘but’ is not only indicating the constraint of contradicting and eliminating previous assumptions; instead, she proposes that ‘but’ instructs the hearer to suspend his conclusion. The pragmatic analysis of pronouns was another area where the distinction between conceptual and procedural words was demonstrated. Wilson and Sperber (1993) propose that pronouns constrain the inferential comprehension process. The pronouns help the hearer to identify the right referent among a set of possible candidates, as example (1) demonstrates. In this utterance the procedural information of ‘she’ in the second sentence guides the hearer to identify the referent. So the hearer assigns ‘she’ to Mary. Another researcher who has worked on the procedural meaning of pronouns is Hedley (2005, 2007). Powell (1999) shows that the referential-attributive distinction in definite expressions can be explained in the relevance-theoretic framework. His argument is that definite expressions are neutral: semantically they can either be referential or attributive. A distinctive interpretation of attributive versus referential meaning however relies on “the link between the descriptive concept and the world” (Powell 1999: 125), i.e., it relies on contextual input to the inferential process. The discussion of the conceptual-procedural distinction was extended to linguistic disciplines like prosody. Wilson and Wharton (2006) showed how contrastive stress constrains the interpretation of utterances. Unger (2006, 2011) examined the procedural effects that are signaled through tense, aspect, and mood and how they relate to understanding foreground and background information. He considers tense, aspect, and mood markers as redundant procedural effects. 3. Pronominal languages in syntactic theory There are languages that incorporate the subject and the object of every sentence in the verb, irrespective whether the subject and the object are overt or not. Jelinek (1984) labeled these languages pronominal argument languages. The generative tradition refers to them as pro-drop languages, or deals with them in terms of the Null-Subject Parameter (Chomsky 1981, 1995; Jaeggli and Safir 1989). Van Valin and La Polla (1997) classified them as head-marking and free word order languages. Linguists have been aware of these languages and the challenges they pose to the traditional word-order typologies (Schröder 2008, among others).
Reference assignment in pronominal argument languages
However, the implications of the fact that these languages incorporate subject and object pronoun has never been analysed from a pragmatic angle. Although Jelinek (1984) argues that pronominal argument languages are only those that have both arguments (subject and object) integrated in the verb, I have shown that the same procedures apply to languages that incorporate only the subject as their main argument (Schröder 2012). Therefore I called them either subject-incorporated languages, or split-pronominal argument languages. The following section shows some typical grammatical characteristics of subject-incorporated languages and their pragmatic interpretation using the framework of RT. 3.1
Grammatical realisation of pronominal argument languages
Let me illustrate what it looks like when the subject pronoun in a pronominal argument language is integrated in the verb, using an example from Toposa, a VSO language from South Sudan:6 (2) a. E-muj-i ɲa-beru ɲa-tapa. 3SG-eat-IMP F/SG-woman F/SG-porridge7 ‘A/the woman is eating porridge.’ b. E-muj-i ɲa-kiriŋ daŋa. 3SG-eat-IMP F/SG-meat also ‘She is also eating meat.’
Sentence (2a) has an overt subject ɲaberu ‘woman’, which sentence (2b) does not repeat. However, both sentences have an incorporated pronoun e- on the verb,8 6. I studied Toposa and collected large amounts of data from native speakers between 19821988, while I lived with my husband in Southern Sudan, working under the auspices of SIL International and the Ministry of Education, and also later when we worked with various Toposa speakers in neighbouring countries. Other data presented here have been taken from texts we collected together which my husband compiled and published (M. Schröder 2010). 7. The abbreviations used in glosses are as follows: 1/2/3P = first/second/third person, 3PL = third person plural, 3SG = third person singular, ALL = allative, CL1/3/5/6/9 = noun class 1/3/5/6/9, D/SG = diminutive singular, DER = derivational prefix, F/LOC = feminine locative, F/PL = feminine plural, F/SG = feminine singular, IMP = imperfective, INF = infinitive, M/LOC = masculine locative, M/SG = masculine singular, PL = plural, PRF = perfective, PRG = progressive, PST = past, REL = relative, SEQ = sequential, SG = singular. 8. The incorporated pronoun set in Toposa is different according to verb classes. In TO-class verbs they are all in the order of 1st/2nd/3rd person SG and 1st/2nd/3rd person PL: a-, i-, e-, e, i-, e-. In KI-class verbs, they are e-, i-, i-, i-, i-, i-. This is the set that is used for non-past.
85
86 Helga Schröder
which refers to the same subject that was introduced in (2a). Here one could ask why a language like Toposa duplicates the reference to the subject in its verbs even where it has an overt subject, does this redundancy serve any function? To begin with, the referent ɲaberu ‘the woman’ in sentence (2a) activates the conceptual address ɲaberu in the mind of the hearer. The e- in (2b) encodes the procedural information that constrains the construction of the explicature leading the hearer to correctly identify the referent ‘woman’ of (2a). However, the question can be raised what is the pragmatic function of the pronominal marker on the verb in sentence (2a), where the subject ɲaberu (woman) is overtly stated and the pronominal reference is encoded on the verb. If we consider processing effort, it does not seem very economical to have the subject and its referent in the same sentence. Therefore it seems that the pronominal e- on the verb must have another pragmatic function besides guiding the listener to access the referent ɲaberu in his search for optimal relevance. This raises the following question: What is the pragmatic function of the pronominal marker on the verb in sentence (2a), where the subject ɲaberu ‘woman’ is overtly stated and the pronominal reference is encoded on the verb? Before we attempt an answer, let us consider another example from Kiswahili, which uses a similar pronominal marker on the verb:9 (3) a. M-toto a-na-kula cha-kula. CL1-child 3SG-PRG-eat CL9-food ‘A/the child is eating food.’ b A-na-kunywa ma-ziwa. 3SG-PRG-drink CL3-milk ‘He is drinking milk.’
Again, (3a) has an overt subject mtoto ‘child, which sentence (3b) does not repeat. However, both sentences have an incorporated pronoun a- on the verb, which refers to the same subject. Pragmatically, in (3a) mtoto ‘child’ is activated in the mind of the hearer and can therefore readily be accessed in (3b). Again the question arises what the pragmatic function is of overtly stating ‘child’ and identifying it also via an incorporated pronoun in the same utterance, enriching the explicature in the following way: ‘A/the child, it is eating food.’ The reactivation of the referent on the verb ‘eating’ seems to work contrary to the principle of least effort. According to RT it can be hypothesized that the use of pronouns in pronominal argument languages must yield other cognitive effects besides providing the
9. I collected the Kiswahili data as part of my job teaching linguistics at the University of Nairobi (we have lived in Kenya since 1988). Where I have used connected discourse examples, the sources have been indicated.
Reference assignment in pronominal argument languages
procedural instruction to identify the referent of the utterance. These additional cognitive effects will be discussed in the next section. I have pointed out a grammatical solution to this problem of double marking (Schröder 2008). I stated there that according to the Principle of Economy in Chomsky’s (1995) Minimalist Program, the referent has been identified in sentences like (2a) and (3a) because the subject ɲaberu ‘woman’ in example (2a) and mtoto ‘child’ in example (3a) went through case-feature checking10 after which it can be dropped as overt pronoun and only occurs as affix on the verb. I also claimed at that point that these pronominal argument languages are not really pro-drop languages, as Chomsky (1981) labels them. The reason for that is that they do not drop the pronoun because the pronoun remains incorporated in the verb. With this analysis however I approached the problem merely from a syntactic point of view and did not include its pragmatic aspect, i.e., I did not raise the question which cognitive effects emerge when the referent has to be decoded twice in the same utterance, once as overt NP and again as incorporated pronoun. In the next section I will therefore try to tackle this problem from a pragmatic vantage point. 3.2
Reference assignment in pronominal argument languages
Before I discuss the reference assignment of pronominals that are indicated on the verb, I have to address the problem of definite and indefinite nominal expressions. 3.2.1 Definiteness and indefiniteness in pronominal argument languages It is well established in English that the difference between indefinite and definite expressions is mostly marked by the use of the indefinite article a versus the definite article the. Additionally, in discourse, someone or something gets usually introduced using an indefinite noun phrase. Consider the following: (4) A man came to the door yesterday.
10. Chomsky (1995) claimed that all subjects and objects are subjacent to case feature checking which is realised through the specifier-head relationship of the agreement subject and agreement object phrase before spell-out into the phonetic (PF) and logical form (LF). Schröder (2008: 109–121) pointed out that in languages that have incorporated subject and object pronouns, these pronominal markers on the verb cannot be case-checked as they do not behave like overt constituents. She further argued that a principle of reference has to be integrated into the computational process of Chomsky’s Minimalist Program to take care of the incorporated pronouns syntactically.
87
88
Helga Schröder
Lyons (1977: 189) points out that “Once any information at all has been supplied about an indefinite referent, it can then be treated by the participants as an individual that is known to them both and identifiable within the universe-of- discourse by means of a definite referring expression.” Under the cognitive-pragmatic approach of RT, the referent becomes part of the context that the addressee will have available in short-term memory from that point on and can therefore access it easily. So the common strategy is to refer to the referent now available in the mental lexicon with a pronoun such as he, she, it in the subsequent sentence where reference assignment is applied as enrichment of the explicature. As a conversation unfolds, it is also possible to use the definite article the in its referential function for the known referent and say: (5) The man wanted to sell me a mixer.
In utterance (4) and (5) the distinction between indefinite and definite expressions is achieved through the use of the indefinite versus the definite article, where the indefinite article is used attributively and the definite article referentially.11 In pragmatic terms, the use of the indefinite article ensures that a referent/entity is introduced to the hearer and is placed in his short-term memory for immediate access. Whenever the definite article is used, the entity is accessed as known to the speaker and the hearer. However, the definite article can also be used attributively in a more indefinite sense, and not in its referential function. In the case of generic nouns like ‘the sun’ and ‘the moon’ the definite article is used attributively, because it does not specify a ‘particular sun’ or a ‘particular moon’. An enormous amount of literature has been written about the referential vs. the attributive use of the article the, proving that it can also be used attributively, besides its basic referential function.12 Many languages, particularly in Africa, do not make use of definite and indefinite articles in their nominal expressions, as the following data from Kiswahili and Toposa will show. In Kiswahili, all nominal expressions enter the discourse with their typical Bantu class prefixes: (6) a. m-toto CL1-child ‘a/the child’ 11. The semantics of the attributive/referential use of articles are as follows: in the case of attributive, the indefinite article is expressing an whoever or whatever meaning of an entity and in its referential sense the definite article picks out a particular entity. 12. This discussion was started by Donellan (1966). For a recent relevance-theoretical account, see Powell (1999).
Reference assignment in pronominal argument languages
b. nyumba CL9:house:SG ‘a/the house’ c. mw-alimu CL1-teacher ‘a/the teacher’ d. cha-kula CL5-food ‘(the) food’ e. ma-ziwa CL3-milk ‘(the) milk’
These nouns have neither indefinite nor definite meaning; in grammatical terms, they are unmarked in regard to definiteness/indefiniteness, pragmatically they are underspecified. Toposa nominal expressions, too, are neutral regarding definiteness, but they are marked for gender and number,13 for example: (7) a. ɲe-kile M/SG-man ‘a/the man’ b. ɲa-beru F/SG-woman ‘a/the woman’ c. ɲi-koku D/SG-child ‘a/the child’
The prefix ɲe- indicates masculine singular, the prefix ɲa- feminine singular, and ɲi- diminutive singular. There are equivalent plural prefixes for masculine- diminutive14 and feminine. 3.3
Pragmatic interpretation of reference assignment
The following general observations can be made from the above data. Both in Toposa and Kiswahili the meaning of a nominal expression in form of a noun is underspecified, it either has attributive or referential meaning. Because a noun is encoded underspecified with regard to definiteness the meaning of the noun 13. Toposa distinguishes between feminine, masculine and diminutive. 14. In the plural, the distinction between masculine and diminutive is neutralized.
89
90 Helga Schröder
in a sentence has to be pragmatically enriched, i.e. the value for definiteness and indefiniteness has to be set to make it part of a fully fledged explicature. Now let us go back to our example (2a), repeated here as (8): (8) E-muj-i ɲa-beru ɲa-tapa. 3SG-eat-IMP F/SG-woman F/SG-porridge ‘A/the woman is eating porridge.’
In the above utterance the nominal expression ɲaberu ‘woman’ is underspecified. It can either be an attributive or referential expression. I want to show that in pronominal languages that do not distinguish between definite and indefinite nominal descriptions grammatically, the referential pronominal device on the verb helps to guide the hearer to overcome the ambiguity between attributive and referential constructions. The hearer is guided through the morpheme e- ‘she’ in (8) to identify the referent or entity in the same clause. After this identification, pragmatically one can say that the referent ‘woman’ is now specified and activated in short-term memory for further access. One can say that pronominal languages identify the referent in the same utterance. What the speaker basically says (i.e. the enriched explicature) is: (9) The woman, she is eating porridge.15
So the Toposa prefix e- identifies the person that the speaker wants to talk about, and changes the meaning of the noun, in this case ‘woman’, from a generic expression ‘a woman’ to the definite, referential one. Thus, there is no need to identify the referent with an overt pronoun, as English would do in the following sentence; the pronominal on the verb is sufficient. How the story might continue is shown by the next sentence: (10) E-muj-i ɲa-kiriŋ daŋa. 3SG-eat-IMP F/SG-meat also ‘She is also eating meat.’
As the referent is already placed in the in the short-term memory of the hearer, and is also identified via the personal pronoun prefix e-, there is no need to spell out a pronoun overtly in (10). So the use of the pronominal referent on the verb in (9) enriches the utterance and helps the hearer to pick out the referent, and by doing so it also works as a procedural device that specifies the nominal expression
15. This is not a marked construction, I have only attempted to mirror the Toposa grammar using English word order.
Reference assignment in pronominal argument languages
as referential description and stores it in the short-term memory as recently given information. Kiswahili follows a similar strategy; consider example (3a) from above, repeated here: (11) M-toto a-na-kula cha-kula. CL1-child 3SG-PRG-eat CL9-food ‘A/the child is eating food.’
The meaning of the word mtoto ‘child’ is underspecified in regard to definiteness, and after it is identified through the referential pronominal on the verb a- ‘it’, it is then activated in the mental lexicon of the hearer as definite expression for further reference and accessibility. In the following sentence the overt pronoun can be dropped and the referential a picks out the referent for the interpretation of reference assignment: (12) A-na-kunywa ma-ziwa. 3P-PRG-drink CL6-milk ‘He/she/it is drinking milk.’
These data suggest that reference assignment in Kiswahili and Toposa is carried out in the following way: after the occurrence of a nominal indefinite/definite expression the integrated verbal pronoun is used in two ways: first it picks out the referent of the utterance for reference assignment and secondly it assigns the value of +definiteness to the nominal expression. In its second function the pronominal marker on the verb behaves as an indexical element. 4. Pragmatic disambiguation of nominal descriptions in discourse As established in the previous section, Toposa and Kiswahili do not distinguish nominal expressions as either attributive or referential through a grammatical device, because nominal expressions are not marked for definiteness or indefiniteness through an article. So in narrative discourse or in speech the nominal descriptions of explicatures in Toposa and Kiswahili are underspecified regarding their attributive or referential value. The following example shows a stretch of discourse where nominal expressions are used attributively and referentially and are enriched through saturation and contextual effects:
91
92
Helga Schröder
(13) S1 Bee koloŋo nuwani a-yai ɲi-cye tooni, It.is.said time long.ago 3SG-be another man a-twar-iti ŋa-kinei keŋe. 3SG-herd-PRF F/PL-goats his S2 To-ryam-u ɲi-koku ni-kareŋani puu, SEQ-found-ALL D/SG-child D/SG-red ideo i-tumo ka nidepei ɲa-tapara, na a-beei Lobeleɲe. 3SG-unaware at near F/SG-pond which 3SG-call Lobeleny S3 Na e-yokokino ɲi-tooni ɲi-koku, ta-tama, … When 3SG-see D/SG-person D/SG-child SEQ-think S4 Ta-naŋakini ɲi-koku ɲi-tooni ɲediye, i-syeme iŋesi, … SEQ-notice D/SG-child D/SG-person nearby 3SG-observe it ‘(S1) It is said that long time ago someone herded his goats. (S2) He found a very red child (= a water spirit), it was unaware near the pond, which was called Lobeleny. (S3) When the man saw the child, he thought, … (S4) The child met unexpectedly (= noticed) with the man nearby [who] was observing it.’16
In sentence (S1) the nominal (ɲi)tooni ‘man’ has the value indefinite through the indefinite marker ɲicye ‘another/some’ and so it has an attributive reading. The same referent ɲicye tooni is identified on the verb by a- in ayai ‘he was’, and also in the second clause of (S1) through a- in a-twarit ‘he was herding’. The identification of the referent ɲitooni ‘person’ through the pronominal expression on the verbs also indexes the nominal expression ɲitooni ‘man’ as referential. In (S2) there is another nominal expression ɲikoku ‘child’ whose meaning is underspecified for definiteness and indefiniteness in the sentence ‘he found a/the very red child’. This noun ɲikoku ‘child’ is also identified on the verb itumo, ‘it was unaware’ through the marker i- ‘3SG’. In this way, the noun is specified through the verb and receives the value +definiteness. From sentence (S3) on ɲitooni ‘man’ and ɲikoku ‘child’ are used referentially as they are known and accessible from the encyclopaedic entries in the mind of the hearer as referential. As the discourse above shows, the same nominal expression can be used referentially and attributively, if it is not specifically marked for indefiniteness through ɲicye ‘another/ some’ in (S1). While the hearer is searching for relevance and contextual effects he will be able to differentiate between definite and indefinite readings of nouns with the procedures of saturation for attributive nouns and reference assignment for referential nouns. If we assume that the process of saturation applies to attributive nouns, we can conclude the following: the enrichment of the underspecified nouns where 16. Toposa texts are taken from ‘Toposa traditional texts’ compiled by M. Schröder (2010).
Reference assignment in pronominal argument languages
they are introduced in discourse is guided by the rule that – if not otherwise indicated – the default reading for nominals is attributive. That means that a pragmatic value of ‘whichever’ as in ɲikoku ‘whichever child’ (S2) is added.17 Such a rule also implies that a nominal that enters the interpretation with the default reading ‘whichever’ is assigned the value +definite after going through reference assignment on the verb. After reference assignment the nominal is used with its referential meaning. Note that the sequential marker to-18 in toryamu ‘found’ also represents the third person singular which in this case is realised by a zero morpheme. We know this because the paradigm of sequential markers still shows the incorporated pronoun a- in the first person of the sequential in both singular and plural: (14) a. A-to-ryam-u. 1SG-SEQ-find-ALL ‘I found.’ b. A-to-ryam-u-tu. 1SG-SEQ-find-ALL-PL ‘We found.’
In non-sequential constructions, the subject is expressed by the incorporated pronoun a- ‘third person’ in the word a-twarit ‘he was herding’ in (13/S1) above. It is not frequent that a nominal expression enters the interpretation process with an attributive reading as is the case with ‘another/some person’ in (13/S1), where the nominal was marked by an indefinite marker. Normally, the nominal expression enters the interpretation in a form that remains underspecified for its meaning, as the following example demonstrates: (15) Bee a-yai koloŋo ɲi-tooni nyika Ɲa-kuju, It.is.said 3SG-was long.ago D/SG-man that.of F/SG-God a-beei Nakopor,… 3SG-call Nakopori,… ‘It is said that long ago there was a man of God (= with supernatural powers), he was called Nakopor. …’
Ɲitooni ‘man’ here is not marked as being attributive or referential; the precise meaning needs to be inferred via enrichment. Again, the nominal expression is made definite through the pronominal identification on the verb a-yai ‘he was’. In 17. This fact became clear after a discussion with Kamila Debowska-Kozlowska. 18. Toposa employs a sequential marker to- that indicates the succession of subsequent events. The sequential marker was later analysed as procedural clause chaining marker (Schröder 2013).
93
94 Helga Schröder
the second sentence a-beei ‘he was called’ the pronominal a- ‘3SG’ identifies the referent. Consider another example: (16) S1 Bee koloŋ nuwan, to-lot Kwee ɲakirap ŋa-kee-moogwa, It.is.said time long.ago, SEQ-go jackal to.collect F/PL-his-food to-ryamu ɲa-ate ka ɲi-tooni, i-pwokit ka lo-kupwor. SEQ-found F/SG-cow of D/SG-man 3SG-stuck at M/LOC-pond ‘It is said that very long ago, Jackal went to search for his food, he found a cow of someone, stuck [in the mud] at the pond.’ S2 Ki-ri-u nai iŋesi, ki-lota, SEQ-pull-ALL then it SEQ-washed ta-anyu ɲa-ate ɲatemar i-waca. SEQ-see F/SG-cow that 3SG-pregnant ‘So he pulled it out [of the mud], he washed [it], he saw that the cow was pregnant.’
In (S1) there are two nominal expressions which serve as referents: Kwee ‘jackal’ and ɲaate ‘cow’. As the above story is an animal fable, ‘jackal’ enters the discourse as a referential expression. ‘Jackal’ is not marked grammatically for indefiniteness or definiteness, but is pragmatically disambiguated through the contextual knowledge of the speaker and hearer, because both have jackal in their encyclopaedic entries listed as well-known trickster. In this case, the nominal expression ‘jackal’ functions like a name. Ɲaate ‘cow’ is an attributive expression, which can refer to any of the cows that ‘someone’ has. Here it is introduced to the hearer and is put into short-term memory for further access. The indefinite entity ɲaate ‘cow’ is first identified via the incorporated pronoun i-, in i-pwokit ‘it was stuck’. Since Toposa does not have grammatical marking for definiteness, the nominal ɲaate ‘cow’ in (S1) needs to be inferentially disambiguated through saturation.19 In fairly similar fashion, Kiswahili nouns are also not marked for definiteness and indefiniteness, so they also have to be enriched for attributiveness or referentiality. Consider the following text from an animal fable: (17) S1 Hapo zamani pa-litokea sungura mjanja sana. There long.ago 3SG-was hare cunning very S2 Siku moja a-litaka, a-wapambanishe tembo na kiboko. day one 3SG-want 3SG-make.collide elephant and hippopotamus
19. The question that arises is how bridging procedures work in languages that lack markers for definiteness.
Reference assignment in pronominal argument languages
S3 Basi, a-kaenda hadi kwa kiboko na kumdhihakiya okay 3SG-SEQ.go up to hippo and INF.tease.him kuwa yeye a-na nguvu zaidi yake that he 3SG-have strength more.than his S4 Kiboko a-kazidi kumbishia na baada ya ubishi mkubwa Hippo 3SG-increased to.argue and after of argument big wa-ka-kubaliana, wa-vutane kwa kamba. 3PL-SEQ-agreed.each.other 3PL-pull.each.other with rope S5 Kiboko a-ki-wa majini na sungura nchi kavu. Hippo 3SG-SEQ-was water.in and hare land dry ‘(S1) Long time ago there was a very cunning hare. (S2) One day he wanted to engage an elephant and a hippopotamus in a contest. (S3) So he went to the hippopotamus’ and ridiculed him that he was stronger than him. (S4) The hippopotamus continued arguing and after a heated argument they agreed with each other to compete through a tug-of-war using a rope. (S5) The hippopotamus would be in water and the hare on dry land.’20
In sentence (S1) sungura ‘hare’ is introduced into the discourse as ambiguous. As the hearer does not have ‘hare’ in short-term memory yet, this nominal expression here needs to be read as attributive, i.e. it refers to any hare in the world.21 In the next sentence ‘hare’ is identified on the verb a-litaka ‘he wanted’ by the pronominal prefix a- ‘he’, and on the verb a-wapambanishe ‘he engaged’. The nominal expressions tembo ‘elephant’ and kiboko ‘hippopotamus’ have either a definite or indefinite reading. If tembo and kiboko are understood as names used in the genre of animal fables, then they are known entities and therefore referential. If they are not referred to as names, they enter the discourse with their attributive reading, i.e. they can refer to any elephant or hippopotamus, applying the same inferential procedure as in Toposa that, if not otherwise indicated, the default reading for nominal expressions is attributive. Through the saturation rule the pragmatic value ‘whichever’ is added. The references to hippopotamus in (S3) and (S4) are referential because at this point they are known, so is sungura in sentence (S5). As we compare nominal expressions in Toposa and Kiswahili discourse, I suggest that in both languages nominal expressions encode conceptual meaning together with an implicit procedural instruction. The conceptual meaning represents the content of the concept, while the procedural instruction directs the hearer to look for the attributive versus the referential value. 20. This text was made available to me by Basilio Mungania from his Kiswahili narrative text collection. 21. A contrary view could also maintain that this refers to a very specific hare, i.e. the wellknow trickster who always wins.
95
96 Helga Schröder
This analysis follows some new RT insights regarding the distinction between conceptual and procedural meaning. Wilson (2011) says that a nominal expression can also be regarded as so-called pro-concept. For her, the distinction between a pro-concept and a full concept is expressed as follows: a full concept is semantically complete, and enters the computational process as such and figures directly as a constituent of thought. In contrast, a schematic concept or ‘pro- concept’ is semantically incomplete and has to be further fleshed out in order to act as a constituent of thought (Wilson 2011: 15).22 She further refers to Sperber’s thoughts on conceptual and procedural meaning:23 a. Assume that all lexical items encode procedures (whether or not they also encode conceptual content, as most of them do). b. When a conceptual content is encoded, so is an instruction to inferentially construct an ad hoc concept using the encoded conceptual content as starting point. c. Other instructions of the type familiar from Diane Blakemore’s work may be encoded by any word, whether or not it encodes conceptual content. Wilson further states that “On this approach, most words encode some procedural content. Some would also encode conceptual content, whereas others (e.g. however) would not” (2011: 17). This approach suggests that words are pro-concepts by default; some then develop into conceptual meaning through the construction of ad hoc concepts, while others, like conjunctions (e.g. although) lead to a purely procedural instruction. Using these insights regarding concepts and pro-concepts, I suggest that Toposa and Kiswahili nominal expressions encode both conceptual and procedural content. The conceptual content represents the meaning of the noun, but in terms of definiteness and indefiniteness nominal expressions are always incomplete. The procedural instruction that is encoded implicitly is to search for the attributive or referential reading of an expression. In this way, every nominal expression needs to be narrowed to assign attributive versus referential meaning. In this sense a nominal always has conceptual content, while it can take a different procedural orientation, either it will be interpreted as attributive through saturation, or it will be understood as referential through enrichment after reference assignment.
22. For a full discussion on the notion of pro-concepts versus full concepts see Wilson (2011: 15–19). 23. Sperber’s thoughts can be found in Sperber Relevance e-mail Archives, 3.12.2007 as an answer to Ducrot’s proposal (1972, 1980) that all words should be regarded as having at least some procedural content.
Reference assignment in pronominal argument languages
5. Identificational and contrastive focus in reference assignment Although pronominal argument languages typically have their pronouns incorporated in the verb, these languages occasionally draw from a set of overt pronouns that are also part of their language inventory.24 Consider the following Toposa example: (18) S1 Ɲa-ce paarani ki-lala Ɲe-ŋatuɲu na-moni, F/SG-another day SEQ-walk M/SG-lion F/LOC-bush ɲa-ki-rap ŋa-kee-moogwa, F/SG-DER-find F/PL-his-food S2 ku-rum iŋesi ɲa-koli. SEQ-caught him F/SG-trap S3 Ki-petepet-aki iŋesi, … SEQ-kicked.hard he ‘(S1) One day Lion walked through the bush to search for his food, (S2) a trap caught him. (S3) He kicked very hard, …’
Note how the overt third person pronoun iŋesi occurs in sentence (S3), which refers back to the referent Ɲeŋatuɲu ‘Lion’ in (S1). As the personal pronoun usually occurs on the verb, the occurrence of the overt pronoun iŋesi is unexpected. The occurrence of iŋesi ‘he’ for the reference assignment of Ɲeŋatuɲu ‘Lion’ is not required because the reference assignment is always accomplished on the verb by the sequential prefix ki- where the person is presented as zero morpheme – see the discussion surrounding examples (13) and (14a, b).25 As the pronominal is always part of the verb, the use of the overt pronoun must therefore exhibit a special procedural meaning that goes beyond the ‘normal’ pronominal meaning, i.e. the identification of the referent. According to the economy-based comprehension procedure of RT, the extra effort that is invested in decoding the overt pronoun should yield some additional positive cognitive effects. In this case, the pronoun iŋesi has the special effect of highlighting ‘Lion’ and thus brings ‘Lion’ to the attention of the hearer, i.e. here we have focus of
24. The non-incorporated pronouns in Toposa, also referred to as overt pronouns (or free-standing pronouns), are ayoŋo ‘1P’, iyoŋo ‘2P’, iŋesi ‘3P’ in the singular and isuwa ‘1P/exclusive’, iŋwoni ‘1P inclusive’, eesi ‘2P’ and ikesi ‘3P’ in the plural. 25. According to the economy principle of Chomsky (1995) that basically states that languages are driven by economy and try to avoid redundant features, this overt pronoun is grammatically a redundant feature.
97
98 Helga Schröder
identification.26 Highlighting in English is often achieved through word order or prosody (Sperber and Wilson 1995: 202–217; Wilson and Wharton 2006: 12–14). However, in pronominal argument languages, the very fact that overt pronouns occur additionally to the incorporated pronouns in a sentence construction indicates a grammatical feature [+focus] whenever they are used (Schröder 2008: 122–123). In the relevance-theoretical comprehension procedure pronouns typically contribute to the construction of explicatures by guiding and setting constraints on the search for referents, but in Toposa the occurrence of the overt pronoun involves additional processing effort, as the hearer has to figure out what the occurrence of the overt person pronoun means after the pronominal on the verb has already identified the referent. Constrained by the context of the previous utterance (S2), the hearer is guided to the more costly contextual implication that it is specifically Lion who is kicking very hard. The occurrence of the overt pronoun can also carry contrastive focus, see the following example: (19) Ɲa-bokoko iŋesi a-raa ɲe-rwositi. F/SG-tortoise he 3SG-was M/SG-chief ‘It was tortoise who was the chief.’
This utterance represents a complex contrastive focus construction that employs two grammatical features: one is a change in word order, and the other is the inherent [+focus] feature carried by overt pronouns. Normally in VSO languages the subject always follows the verb, but here Ɲabokoko ‘tortoise’ is fronted and occurs at the beginning of the sentence. This change in word order alerts the hearer that he has to pay special attention to Ɲabokoko. Note also how the overt pronoun iŋesi ‘he’ (which refers to Ɲabokoko ‘tortoise’) contributes to the focus construction via its inherent feature [+focus] as it occurs in addition to the expected pronominal a- ‘3SG’ marking on the verb. The grammatical complex construction above requires the hearer to invest extra effort. He is guided beyond the procedural instruction that tells him to select the referent Ɲabokoko ‘tortoise’. This extra effort is stimulated by the combination of the encoded word order change and the inherent [+focus] feature of the overt pronoun. So by enriching the explicature through the extra procedural instruction the hearer understands that the overt pronoun does not only select the referent, but also picks tortoise out of a group of other possible referents, so he 26. Focus of identification, also known as assertive focus, presupposes known information and highlights the presupposed information and brings it to the attention of the hearer. Contrastive focus also known as selective focus, presupposes known information, and brings the known information to the attention of the speaker but implies a choice out of a set of candidates in the sense he and nobody else.
Reference assignment in pronominal argument languages
comprehends it is tortoise and no one else who was chief. The cognitive selection from a list of known referents is a typical procedure for contrastive focus, which is also known as narrow focus in Generative Grammar (Kiss 1995), or exhaustive focus (Watters 1979).27 In English, contrastive focus is marked by contrastive stress (Wilson and Wharton 2006: 12–13; Sperber and Wilson 2011: 212–217), which Wilson and Wharton (2006: 12) also call ‘natural highlighting device’. The decoding of the difference in meaning between focus of identification, as shown in (18), and contrastive focus, as demonstrated in (19), is guided by the change of word order from the expected VSC to the to unexpected and therefore more highly marked SSVC word order, plus the additional focus feature of the overt pronoun. In Kiswahili a phenomenon similar to what was described for Toposa is found, consider the following examples for focus of identification and contrastive focus:28 (20) a. Yeye a-li-kuwa, a-ki-soma darasa la sita, … He 3SG-PST-be 3SG-SEQ-read class of six ‘He was reading (= studying) in standard six, …’ b. Na yeye ndi-ye a-li-ye-leta kombe mwaka 66. and he really-he 3SG-PST-REL-bring cup year 66 ‘He is the one who brought the cup in 1966.’
In (20a) the prefix a- ‘3SG’ on the verb is sufficient to indicate reference assignment, but because the overt pronoun yeye occurs, extra processing effort is required. In this case the protagonist Kombo (who had been introduced earlier), is highlighted through the unexpected occurrence yeye ‘he’, similar to the focal stress pattern in English. In (20b) the contrastive function is evident as yeye ‘he’ is accompanied by the pronoun ndiye ‘really he (= he is the one)’. So the focus in the second example is contrastive. The difference between the highlighted reading and the contrastive reading is achieved through ndiye in (20b). 6. Conclusion Using data from Toposa and Kiswahili, I wanted to show in this paper that pronominal argument languages have their own method of reference assignment. In this type of language, reference assignment first of all has the effect of identifying 27. Narrow and exhaustive focus are other terms for contrastive focus. They are used as synonyms for the cognitive process that a hearer selects a candidate out of a group of known entities. 28. Taken from the text Malipo duniani hesabu ahera ‘Payment on earth calculations in heaven’, collected by Basilio Mungania.
99
100 Helga Schröder
the referent of an utterance. However, such referential identification also helps to disambiguate between attributive and referential meaning of nominal expressions. Regarding the pronominal reference assignment, I demonstrated how nominal expressions enter the discourse as underspecified regarding their referential/attributive meaning. I argued that nominals have conceptual and implicit procedural meaning. The conceptual meaning represents the content of the word that directly contributes to the computational processing of the utterance. The implicit procedural instruction guides the hearer to enrich the attributive value of the noun through saturation. This process applies the rule that, if not otherwise indicated, nouns indicate the meaning ‘whichever’ or ‘whatever’. Next the hearer establishes the referential value for the noun through reference assignment. The incorporated pronoun on the verb instructs the hearer to identify the referent, thus the referential value of the noun is determined. I also argued that pronominal argument languages have special stylistic means for highlighting referents: as the referential pronoun is always present on the verb, the occurrence of the overt pronoun requires extra processing effort and creates positive cognitive effects of highlighting the referent, or it produces the effect of contrastive focus, similar to what is achieved by contrastive stress in English. The results of this paper can only be regarded as pointing to the need for further research. The following questions should be raised: Firstly, do all pronominal languages that have an incorporated referential subject pronouns coincide with underspecified nominal expressions that then are conceptual and procedural in meaning? Secondly, how do languages that are not pronominal argument languages mark their nominal expressions for definiteness – since they do not have the device of incorporated pronouns on the verb? And thirdly: what theoretical conclusions can be drawn regarding the relationship of conceptual and procedural meaning of nominals that are not marked for definiteness in light of Wilson’s (2011: 16–17) discussion of pro-concepts?
References Blakemore, Diane. 1987. Semantic Constraints on Relevance. Oxford: Blackwell. Blakemore, Diane. 2002. Linguistic Meaning and Relevance: The Semantics and Pragmatics of Discourse Markers. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511486456
Blass, Regina. 1990. Relevance Relations in Discourse: A Study with Special Reference to Sissala. Cambridge Studies in Linguistics, 55. Cambridge: Cambridge University Press. Carston, Robyn. 2002. Thoughts and Utterances. The Pragmatics of Explicit Communication. Oxford: Blackwell. doi: 10.1002/9780470754603 Chomsky, Noam. 1981. Lectures on Government and Binding. New York: Mouton Gruyter.
Reference assignment in pronominal argument languages 101
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. 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. Donnellan, Keith S. 1966. “Reference and Definite Description.” The Philosophical Review 75: 281–304. doi: 10.2307/2183143 Ducrot, Oswald. 1972. Dire et ne pas dire: Principes de semantique linguistique. Paris: Hermann. Ducrot, Oswald. 1980. Les mots du discours. Paris: Editions de Minuit. Fodor, Jerry A. 1975. The Language of Thought. New York: Crowell. Fodor, Jerry A. 1998. Concepts: Where Cognitive Science Went Wrong. Oxford: Oxford University Press. Hall, Alison. 2004. “The Meaning of but: A Procedural Reanalysis.” UCL Working Papers in Linguistics 16: 199–236. Hedley, Paul. 2005. “Pronouns, Procedures and Relevance Theory.” Durham and Newcastle Working Papers in Linguistics 11: 41–55. Hedley, Paul. 2007. Anaphora, Relevance and the Conceptual/Procedural Distinction. PhD diss., University of Oxford. Iten, Corinne. 1997. “Because and Although: A Case of Duality?” UCL Working Papers in Linguistics 9: 1–24. Iten, Corinne. 2002. “Even if and even: The Case for an Inferential Scalar Account.” UCL Working Papers in Linguistics 14: 119–157. Jaeggli, Osvaldo, and Ken Safir (eds). 1989. The Null-Subject Parameter. Dordrecht: Kluwer. doi: 10.1007/978-94-009-2540-3
Jelinek, Eloise. 1984. “Empty Categories, Case, and Configurationality.” Natural Language and Linguistic Theory 2 (1): 39–76. Kiss, Katalin É. 1995. Discourse Configurational Languages. Oxford: Oxford University Press. Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press. Powell, George. 1999. “The Referential-Attributive Distinction – A Cognitive Account.” UCL Working Papers in Linguistics 11: 101–125. Schröder, Helga. 2008. Word Order in Toposa: An Aspect of Multiple Feature-checking. SIL International and the University of Texas at Arlington Publications in Linguistics 142. Dallas: SIL International. Schröder, Helga. 2012. “Incorporated Subject Pronouns in Word Order Typology.” In Proceedings of the 6th World Congress of African Linguistics, Cologne 2009, ed. by Matthias Brenzinger and Anne M. Fehn, 201–213. Cologne: Köppe. Schröder, Helga. 2013. “Clause Chaining in Toposa: A Pragmatic Approach.” Lodz Papers in Pragmatics 9 (1): 25–44. doi: 10.1515/lpp-2013-0003 Schröder, Martin. 2010. Toposa Traditional Texts. Nairobi: SIL International. Manuscript. Sperber, Dan, and Deirdre Wilson. 1995. Relevance. Communication and Cognition. 2nd edition. Oxford: Blackwell Publishing. Unger, Christoph. 2006. Genre, Relevance and Global Coherence. Basingstoke, UK: Palgrave. Unger, Christoph. 2011. “Exploring the Borderline between Procedural Encoding and Pragmatic Inference.” In Procedural Meaning: Problems and Perspectives, ed. by M. Victoria Escandell Vidal, Manuel Leonetti, and Aoife Ahern, 103–127. Bingley: Emerald Group Publishing. Van Valin, Robert D., Jr., and Randy LaPolla. 1997. Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. doi: 10.1017/CBO9781139166799
102 Helga Schröder
Watters, John R. 1979. “Focus in Aghem: A Study of its Formal Correlates and Typology. Part 3.” In Aghem Grammatical Structure: With Special Reference to Noun Classes, Tense-aspect and Focus Marking, ed. by Larry M. Hyman, 137–197. Southern California Occasional Papers in Linguistics 7. Los Angeles: University of Southern California. Wilson, Deidre. 1992. “Relevance and Understanding.” UCL Working Papers in Linguistics 4: 165–191. Wilson, Deirdre. 2011. “The Conceptual-Procedural Distinction: Past, Present and Future.” In Procedural Meaning: Problems and Perspectives, ed. by M. Victoria Escandell Vidal, Manuel Leonetti, and Aoife Ahern, 3–31. Bingley: Emerald Group Publishing. doi: 10.1108/S1472-7870(2011)0000025005
Wilson, Deirdre, and Dan Sperber. 1993. “Linguistic Form and Relevance.” Lingua 90: 1–25. Wilson, Deirdre, and Tim Wharton. 2006. “Relevance and Prosody.” Journal of Pragmatics 38: 1559–1579. doi: 10.1016/j.pragma.2005.04.012
Conceptual and procedural information for verb tense disambiguation The English Simple Past Cristina Grisot, Bruno Cartoni and Jacques Moeschler University of Geneva
This chapter discusses the necessary linguistic and pragmatic support for improving statistical machine translation systems with respect to verbal tenses. The English Simple Past can be translated into French through a series of tenses because of its conceptual, procedural and pragmatic meanings. By testing and validating its usages in offline experiments with a linguistic judgment task, a general predictive model for the cross-linguistic variation of a verbal tense is proposed. The implementation of the procedure encoded by the Simple Past for a statistical machine translation system improved its results in terms of coherence and lexical choices. Thus, this chapter shows that such approach to automatic translation of verbal tenses seems promising and worth pursuing. Keywords: verbal tenses, translation corpora, statistical machine translation, conceptual-procedural distinction, subjectivity
1. Introduction Improving the results of Statistical Machine Translation systems (SMT) is a great challenge and researchers have understood that this cannot be done without an interdisciplinary perspective. If machines have great results for realizing many linguistic tasks (such as syntactic parsing, semantic correlations for dictionaries or logic analyses according to a certain formal language), one domain that challenges them is language use in context. Pragmatics is thus a crucial domain to study when one aims at improving “linguistic capacities” of machines. Within the COMTIS and MODERN Projects,1 our goal is to improve the quality of machine- 1. The COMTIS Project (Improving the Coherence of Machine Translation Output by Modeling Intersentential Relations; project n° CRSI22_127510, March 2010-July 201) and the doi 10.1075/pbns.268.05gri © 2016 John Benjamins Publishing Company
104 Cristina Grisot, Bruno Cartoni and Jacques Moeschler
translated texts by modelling intersentential relations, such as those that depend on verb tenses and connectives. Intersentential relations play an important role for the coherence and cohesion of a discourse. Since the late seventies, an important number of studies in various domains such as semantics, pragmatics, discourse analysis and Natural Language Processing (NLP) have analysed the factors that contribute to discourse coherence (Halliday and Hasan 1976; Hobbs 1983; Mann and Thompson 1987) and propose taxonomies of discourse relations (Mann and Thompson 1987; Sanders 1992). Cohesion is a more specific notion related to coherence, which refers specifically to the linguistic devices used to build coherence between sentences. Verbal tenses represent a type of cohesive ties, among other lexical and grammatical devices such as pronouns, anaphora and discourse connectives. This paper seeks to address the problem of verbal tenses, and in particular how to formalize their distinct usages in order to improve their translation by SMT systems.2 The multilingual objectives of the COMTIS and MODERN Projects reveal the crucial need of a common framework that can be used to describe and to analyse verbal tenses in more than one language at a time. Our research has multiple aims. The departure question was a complex one: why humans choose one rather than another tense when translating from a source language (SL) to a target language (TL) and how can this information be used to improve the results of machine translation systems? In order to answer to this question we needed to identify problematic tenses in translation corpora, propose possible features that explain the choices made by human translators, test them in annotation experiments and finally, use the annotated corpora for SMT systems. Thus, our main research question developed in this article is: which features should be included in a model that explains and predicts the cross-linguistic variation of the translation of verbal tenses? We argue that a reliable description of tenses can be done only within an inferential pragmatics framework, specifically relevance theory (RT) (Sperber and Wilson 1986; Wilson and Sperber 1993, 2004, 2012). We adopt the linguistic underdeterminacy principle for verbal tenses as developed in RT (Sperber and Wilson 1986; Smith 1990; Moeschler et al. 1998; Saussure 2003). We assume that the meaning of a verbal tense form3 is MODERN Project (Modeling discourse entities and relations for coherent machine translation; project n° CRSII2_147653, August 2013–August 2016) belong to the Sinergia interdisciplinary program funded by the Swiss National Science Foundation. 2. This paper reflects partly research carried out in COMTIS and MODERN projects. For the overall results and the final theoretical model, see Grisot (2015, 2016, submitted). 3. We differentiate form of meaning and of usage. Specifically, verbal tense form refers to different forms that verbal system of a language has, such as for example the English (EN) past
Conceptual and procedural information 105
underdetermined and must be contextually worked out through contextual enrichment conforming to expectations of optimal relevance. Verbal tenses are thus a referential category: they are characterized as locating temporal reference for eventualities with respect to three coordinates, specifically speech point S, event point E and reference point R (Reichenbach 1947). Assigning temporal reference for eventualities is an inferential and context-depending process providing the propositional form of the utterance. Two important issues when investigating tense and temporal information at the discourse level are temporal sequencing and cause-consequence relations between eventualities. Sperber and Wilson (1986), Blakemore (1988) and Wilson and Sperber (1998) treat temporal sequencing and cause-consequence relations as inferentially determined aspects of what is said, thus part of explicatures. In this paper, we adopt their approach both for temporal reference assignment and for temporal/causal relations as being inferred information through non- demonstrative inferences. Moreover, we explain temporal reference assignment and temporal/causal relations in terms of the conceptual-procedural distinction (Blakemore 1987). We argue that they encode both conceptual and procedural information (Moeschler 2002; Moeschler et al. 2012; Grisot and Moeschler 2014; Grisot 2015). In our view the conceptual content is given by a specific configuration of Reichenbachian coordinates E and S. In Grisot (2015), it was argued that the specific configuration of the temporal coordinates S and E behaves like pro-concepts (Sperber and Wilson 1998: 15; Wilson 2011). Pro-concepts are semantically incomplete, they are conveyed in a given utterance and have to be contextually worked out. The conceptual content of the French (FR) verbal tenses expressing past time Passé Composé (PC), Passé Simple (PS) and Imparfait (IMP), classically being described as having the configuration E tenses >> verbs class (lexical aspect). The general model defended in this paper predicts that procedural information encoded by connectives, mode, grammatical aspect and tense is stronger than conceptual information encoded by tense and lexical aspect.
118 Cristina Grisot, Bruno Cartoni and Jacques Moeschler
of information but also, progressively, based on experimental evidence (see for example, Moeschler 2016; Zufferey 2012, 2014). Similarly, other phenomena have been investigated regarding the conceptual-procedural distinction and their role for discourse processing (see for example the collection of articles on procedural information edited by Escandell-Vidal et al. 2011), such as mood and modality (see for example Barbet 2013; Saussure 2014), verbal tenses (see for example, Moeschler 2016; Grisot 2015; Grisot and Moeschler 2014; Saussure and Morency 2012; Saussure 2010), pronouns and determiners (see for example, Breheny 1999, on definite expressions), to name but a few. Probably the most well known qualitative description of conceptual and procedural types of information is Wilson and Sperber (1993). They (a) propose an exhaustive hierarchy of types of information conveyed by an utterance and (b) attach cognitive foundations to the conceptual-procedural distinction. In this hierarchy conceptual and procedural information is linguistically encoded (thus semantic information). According to them, encoded information contributes either to explicatures or to higher-level explicatures while procedurally encoded information constraints the explicatures (such as pronouns), the higher-level explicatures (such as the semantic differences between declarative sentences and their non-declarative counterparts) or the implicatures (such as discourse connectives). The first attempts to define and characterize conceptual and procedural information included qualitative features such as truth-conditional vs. non truth- conditional (see Wilson and Sperber 1993 for arguments against this correlation), representational vs. computational, accessible to consciousness vs. inaccessible to consciousness, easily graspable concepts vs. resistant to conceptualization, capable of being reflected on vs. not available through conscious thought (Wilson and Sperber 1993; Wilson 2011), non cancellable vs. cancellable and easily translatable vs. translatable with difficulty (Moeschler et al. 2012). Saussure (2011) proposes a methodological criterion to distinguish between what is conceptual and what is procedural. In his words, an expression is procedural when it triggers inferences that cannot be predicted on the basis of a conceptual core to which general pragmatic inferences (loosening and narrowing) are applied. Grisot (submitted) provides a critical discussion of all qualitative features currently existent in the literature and proposes a quantative measure that allows objective and replicable analyses of linguistic expressions encoding these two types of information. This measure making use of Wilson and Sperber’s (1993) cognitive foundation of the conceptual-procedural distinction is the inter-judge agreement rate (see Section 5 for an illustration of its application to verbal tenses).
Conceptual and procedural information 119
As far as verbal tenses are concerned, two main trends are opposed regarding the nature of their encoded content: on the one hand, a tense encodes procedural information and, on the other hand, we argue a tense encodes both procedural and conceptual information. According to the first trend, verbal tenses have only rigid procedural meanings that help the hearer reconstruct the intended representation of eventualities (Nicolle 1998; Saussure 2003, 2011; Amenós Pons 2011). Saussure (2003) proposes algorithms to follow, consisting of the instructions encoded by verb tenses, in order to grasp the intended meaning of a verb tense at the discourse level. Taking the distinction conceptual-procedural as a foundation, Moeschler (1993), Moeschler et al. (1998), and Nicolle (1997, 1998) claim that verbal tenses have a procedural meaning. Nicolle (1998: 4) argues that tense markers impose constraints on the determination of temporal reference and thus they “may be characterized as exponents of procedural encoding, constraining the inferential processing of conceptual representations of situations and events.” Concerning the status of the temporal coordinates, Saussure and Morency (2012) argue that tenses encode instructions on how the eventuality is to be represented by the hearer through the positions of temporal coordinates. They consider thus that temporal location with the help of S, R and E is of a procedural nature. In this paper, we will defend a conceptualist view of tense, which encodes both conceptual and procedural information. With respect to conceptual information, the assumption is that the specific configuration of the temporal coordinates S and E behaves like pro-concepts (Sperber and Wilson 1998: 15; Wilson 2011). Pro-concepts are semantically incomplete, they are conveyed in a given utterance and have to be contextually worked out through an enrichment process similar to lexical-pragmatic processes. Once the enrichment process is completed the propositional form of the utterance is also available. This temporal information is not defeasible, i.e. it cannot be cancelled. The parameters themselves represent conceptual content, while their contextual values are pragmatically determined. For example, the FR PC allows reference both to past time and to future time. In (17), R is in the past and so the PC refers to past time. In (18), R is given by the temporal adverb so it expresses reference to future time. (17) J’ai fini mon livre. ‘I finished my book.’ (18) Demain, j’ai fini mon article. ‘Tomorrow, I will have finished my book.’
In an offline experiment carried out by Grisot (2015), the participants were asked to conjugate the verb (provided at the infinitive) of each experimental item, as
120 Cristina Grisot, Bruno Cartoni and Jacques Moeschler
exemplified in (19). The results of the experiments indicate a very high agreement rate among annotators, that is, a kappa of 0.86. According to Grisot’s (submitted) scale7 for the interpretation of kappa values with respect to the type of information encoded, which is based on the cognitive features of conceptual and procedural information proposed by Wilson and Sperber (1993), this information is conceptual. (19) Le jeune soldat mis en cause a agi contre les orders de ses supérieurs, il (être) aujourd’hui incarcéré et en attente d’être jugé pour meurtre. ‘The suspected young soldier behaved against his superior’s orders, he (to be) today imprisoned and waiting to be judged for murder.’
Starting from the claim that eventuality types have a conceptual meaning (they have logical properties and add to the propositional content of the utterance) and tenses have procedural meaning, Moeschler (2002) argued that the meaning of any lexical item includes two components: conceptual information, which describes the concept accessible via the lexical entry, and procedural information, which indicates how to reach the descriptive content of concepts. Moeschler (2002) thus proposed that lexicon should be viewed from a perspective that combines both procedural and conceptual information. If we consider example (20) and imagine two different contexts, the distance on the time line between E and S, even if S=E for present tenses is contextually adjusted based on world knowledge. In a first context, a husband is upstairs and his wife is downstairs in their house, he calls her and she answers (20). In the second context, the wife has an hour ride from work to home, he calls her to see when she comes back home and she answers (20). The temporal distance between E and S is between immediately and 2–3 minutes in the first context and a few minutes and one hour in the second context. (20) J’arrive! ‘I am coming!’
Regarding the procedural content of verbal tenses, they help the hearer access the right contextual hypotheses conforming to the principle of relevance to get the intended cognitive effects (Wilson and Sperber 1998). Carston (1998) points out that under normal conditions discourse material is presupposed to be relevant and, when information is not explicitly given, it is filled in. The linguistic content of utterances is thus enriched in the interpretive process: in our case, the basic 7. More precisely, the scale is the following: kappa values 0.7 signals conceptual information. The arguments for this scale are given in Grisot (submitted).
Conceptual and procedural information 121
temporal location of the eventuality represented by conceptual information of a verbal tense is enhanced through the application of the procedure encoded by that verbal tense. In example (21), Binnick (2009) giving a similar example to that proposed by Grice (1989)8 argues that the material in brackets is implicit. We consider that (21) is an example of temporal ordering, and thus the procedural feature [±narrativity] of the SP is active. (21) He took off his boots and [then] got into bed.
We define procedural information of verb tenses in terms of the [±narrativity] feature. We argue that this feature is encoded by verbal tenses and that they can be used for multilingual comparison of different usages of verb tenses. The [±narrativity] feature corresponds to the demand to verify if the events are temporally ordered. If it is positive, then a procedure of temporal ordering calculus is set off. Identification of reference time is either linguistically triggered (through verb tense form or temporal adverb, for example) or pragmatically inferred by the hearer/reader. This procedure of temporal ordering calculus is not a default procedure, as Asher and Lascarides (2003) state, but it is triggered by the positive value of narrativity feature, hence [+narrative]. We provide four arguments in favour of the procedural nature of this feature. Firstly, the temporal information operationalized as the [±narrativity] feature constraints the inferential phase of constructing explicatures. It does not contribute but constraints the construction of the propositional content of utterance (Wilson and Sperber 1998; Binnick 2009; Escandell Vidal and Leonetti 2011). Secondly, temporal sequencing is a discourse property: it needs at least two eventualities for the [±narrativity] feature to be active (hence have a positive value). Procedural content gives information about how to manipulate conceptual representations, corresponding to more than two discourse entities. If a tense has a narrative usage, it means that as soon as its reference time is set, it is used to construct the temporal reference of the next event, and thus time advances. Binnick (2009) points out the role of verb tenses for discourse coherence as temporal anaphors (discourse interpretation depends on the identification of their antecedents). In example (22), the SP of the verb take (specifically took) is bound by that of the verb go (specifically went). Time advances in a narrative sequence because the R point of one eventuality is located just after the preceding one. (22) John went home early. He took the subway. 8. Binnick’s example is a typical example for conversational implicatures (in Grice’s terms, 1989) that follow the maxim “Be orderly.” Carston (1998, 2002) and Sperber and Wilson (1986) treat this content as pragmatically determined aspects of what is said, thus an explicature.
122 Cristina Grisot, Bruno Cartoni and Jacques Moeschler
Thirdly, temporal sequencing can hardly be paraphrased (as with synonyms for conceptual representations), but it can be rendered explicit with the help of temporal connectives, such as and, then, afterwards, because. And fourthly, the [±narrativity] feature is information inaccessible to consciousness resulting in low agreement for two annotators (Grisot and Moeschler 2014). This predictive model regarding verbal tenses and their usages is a discourse model. Kamp and Rohrer (1983) also argued for their discourse semantics model that the meaning of a tense could be established only at the discursive level. We did not aimed at proposing a model for isolated tokens of SP. The model we present here is determined by the need to disambiguate usages of the SP and to improve its translation into FR. Consider example (23). Its translation into a TL is ambiguous. Taken as an isolated token it cannot be disambiguated. Consider now example (24), the second sentence introduces another eventuality and the two eventualities are causally related. According to our model, the SP has a narrative usage and it is translated into FR by a PS/PC. In (25) on the other hand, the second sentence introduces an eventuality that takes place simultaneously. More specifically, the R period of the first SP includes the R moment of the second eventuality. According to our model, the SP has a non-narrative usage and it is into FR by an IMP. (23) John slept. (24) John slept. He got rest. ‘Jean a dormi. Il s’est reposé.’ ‘Jean dormit. Il se reposa.’
(PC) (PS)
(25) John slept. He had a dream. ‘Jean dormait. Il fit un rêve.’
Finally, verbal tenses have been described in the literature also in relation to the notion of subjectivity (Banfield 1982/1995; Reboul 1992; Schlenker 2004; Fleischmann 1990, 1991; Tahara 2000; Saussure 2003 and Binnick 2009, to name but a few). For example, Binnick (2009) observed that there is a perspective or point of view from which the events are narrated and the category of tense is sensitive to this focalization. He notes that narration may be non-focalized [–subjective] or it may adopt the perspective of either an internal or an external focalizer [+subjective] (Fleischman 1990: Chapter 7). In Grisot and Cartoni (2012), we made the assumption that the [±subjectivity] feature has a procedural nature and gave arguments similar to those provided above for the [±narrativity] feature. However, a procedural character was validated experimentally only for the [±narrativity] feature (Grisot 2015).
Conceptual and procedural information 123
Grisot (2016) tested experimentally the [±subjectivity] feature, which native speakers and French and, respectively, English judged the French PC, PS and IMP, and respectively, the English SP regards subjective and non-subjective usages. Subjectivity was defined as the speaker’s psychological perspective and perceptions included into the description of a situation. With respect to this definition, sentences can be subjective or not subjective (objective). A sentence is subjective when the description of a situation or a series of situations is centred on the speaker’s psychological perspective. A sentence is not subjective when the speaker merely reports a situation or a series of situations that are related in the world. The annotators were asked to judge each occurrence of the target verbal tense as being subjective or not subjective. For each language, the agreement rate between judges was very low (kappa of 0.22 for English and 0.31 for French). According to the scale suggested by Grisot (submitted) for the interpretation of kappa values with respect to the type of information encoded, the [±subjectivity] feature is pragmatic information recovered through general pragmatic inference. This feature seems to be specific to certain verbal tenses in particular, such as the IMP and PS in FR. This model represents a fine interweaving consisting of semantic and pragmatic features, in which the features keep their independent character. The various possible combinations of the three features characterize the usages of a verbal tense. In the following subsection, we provide a detailed presentation of these FR tenses and their narrative/non-narrative and subjective/non-subjective usages. 4.3
Predictive model for specific tenses
We argue that both conceptual (S, R & E) and procedural contents of the EN SP represent crucial information for usage disambiguation and utterance interpretation, as well as for discourse coherence. We claim that this information represents disambiguation criteria and can be used as semantic and pragmatic traits for tagging parallel corpora. These parallel corpora were be used for machine learning. In order to test empirically the [±narrativity] feature, we performed annotation experiments that confirmed partially our hypotheses. The procedure and the results of our annotation experiments are provided in Section 5. The feature [±subjectivity] is not directly included in the predictive model presented in Section 4.1 because of its pragmatic character. However, it represents relevant information of characterising usages of verbal tenses. Hence, one question that arises at this point of our discussion is how to apply this pragmatic model to specific verbal tenses, such as the FR PS or IMP. The main hypothesis is that a verbal tense may have only some of the possible usages
124 Cristina Grisot, Bruno Cartoni and Jacques Moeschler
subjective narrative non-subjective PS non-narrative non-subjective
Figure 5. Usages of the French PS
predicted by the model. Our assumption is that the variation of usages of verbal tenses can be explained through the combination of conceptual, procedural and pragmatic types of information. For example, the FR PS has narrative non- subjective and narrative subjective usages and non-narrative non-subjective usages as shown in Figure 5. The first type of usage is ordinary narrative usages as in (26) while the second occurs more rarely, as in (27). The third represent temporal simultaneity, as in (28). (26) Max entra dans le bar. Il alla s’asseoir au fond de la salle. ‘Max entered the bar. He sat in the back.’ (27) Aujourd’hui, personne ne lui adressa la parole (Stendhal, Le Rouge et le Noir) ‘Today, nobody talked with him.’ (28) Bianca chanta et Igor l’accompagna au piano. ‘Bianca sung and Igor played the piano.’
The IMP and PS in FR have the same configuration of temporal coordinates (E=R