Humorous Garden-Paths: A Pragmatic-Cognitive Study 1443801402, 9781443801409

Citation preview

Humorous Garden-Paths

Humorous Garden-Paths: A Pragmatic-Cognitive Study

By

Marta Dynel

Humorous Garden-Paths: A Pragmatic-Cognitive Study, by Marta Dynel This book first published 2009 Cambridge Scholars Publishing 12 Back Chapman Street, Newcastle upon Tyne, NE6 2XX, UK This manuscript was reviewed and approved by Dr. Andy Nercessian at CSP.

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2009 by Marta Dynel All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-0140-2, ISBN (13): 978-1-4438-0140-9

TABLE OF CONTENTS

Acknowledgments ...................................................................................... ix Introduction ................................................................................................. 1 Chapter One Methodology and Introduction to Garden-Paths ......................................... 4 1. Methodology and exemplification........................................................... 4 1.1.Humorousness and funniness........................................................... 7 2. Jokes vs. conversational humour ............................................................. 8 2.1. (Canned) jokes ................................................................................ 9 2.2. Conversational humour: any verbal humour but jokes ................. 11 3. Witticisms and one-liners ...................................................................... 13 4. Humorous keying .................................................................................. 16 5. On-line processing of language ............................................................. 17 6. Garden-path sentences vs. garden-path humour .................................... 18 7. What the GP mechanism is not.............................................................. 24 7.1. The red-light mechanism .............................................................. 25 7.2. The crossroads mechanism ........................................................... 29 8. The structure of GP jokes, witticisms and one-liners ............................ 32 Chapter Two Theoretical Approaches to Humour: Incongruity and Incongruity-Resolution ....................................................................... 41 1. Families of humour theories .................................................................. 41 2. Philosophical incongruity theories ........................................................ 42 3. Incongruity in contemporary studies ..................................................... 45 4. Bisociation............................................................................................. 48 5. Incongruity model or incongruity-resolution model.............................. 51 6. The I-R model: Queries and problems................................................... 54 7. Novelty and surprise.............................................................................. 61 8. Defeated expectations............................................................................ 64

vi

Table of Contents

Chapter Three Humour Proposals ..................................................................................... 69 1. Shultz’s and Suls’s two-stage models: The incongruity-resolution model ......................................................................................................... 69 2. A catastrophe theory model................................................................... 73 3. The SSTH and the GTVH ..................................................................... 76 3.1. The SSTH ..................................................................................... 76 3.1.1. Script ................................................................................. 77 3.1.2. Script opposition/overlap................................................... 78 3.2. The GTVH .................................................................................... 81 3.2.1. Script opposition and logical mechanism .......................... 84 3.3. The SSTH and the GTVH vs. I-R theories ................................... 86 3.4. Script enforcement: The trigger and on-line processing ............... 88 4. Bisociation as schema conflict............................................................... 93 5. Isotopy-disjunction model ..................................................................... 96 6. The frame-structuring model ............................................................... 101 7. Graded salience and Marked Informativeness Requirement ............... 108 8. Forced reinterpretation ........................................................................ 113 9. Conclusion and comments on GPs ...................................................... 116 Chapter Four Ambiguity-Based Types of GPs .............................................................. 117 1. (Non-)verbal and (non-)linguistic humour .......................................... 117 1.1. Criteria for the linguistic vs. non-linguistic distinction............... 120 2. Ambiguity............................................................................................ 123 2.1. Ambiguity vis-à-vis other phenomena ........................................ 124 3. Ambiguity types .................................................................................. 125 4. Syntactic ambiguity and GPs............................................................... 129 5. Lexical ambiguity and punning GPs.................................................... 131 5.1. Puns............................................................................................. 131 5.2. The basic mechanism of garden-path puns ................................. 132 5.3. Polysemy and homonymy........................................................... 133 5.4. Literalisation of phrasemes: phrasemic polysemies.................... 139 6. Pragmatic ambiguity and GPs ............................................................. 143 6.1. GPs based on purely inferential ambiguity ................................. 145 6.2. GPs based on linguistically-motivated inferential ambiguity ..... 146 6.3. GPs based on presuppositional ambiguity .................................. 150 7. Final comments ................................................................................... 154

Humorous Garden-Paths

vii

Chapter Five Grice’s Philosophy and Garden-Path Humour ........................................ 155 1. Humour and the CP ............................................................................. 156 2. The Gricean model .............................................................................. 157 2.1. Maxim nonfulfillment and inferences......................................... 157 2.1.1. Maxim violations............................................................. 161 2.1.2. Maxim flouts ................................................................... 162 2.2. Rationality and intentionality at the heart of cooperation ........... 164 2.2.1. Intentionality, i.e. the speaker’s meaning ........................ 165 2.2.2. Rationality, reasoning and mutual observance of the CP .. 166 2.3. The purpose of communication................................................... 168 2.4. Can the CP be violated? .............................................................. 170 3. The Gricean model and humor ............................................................ 173 3.1. Other discussions of the Gricean model and humour interface... 174 3.2. Non-bona-fide mode? ................................................................. 178 3.3. Humour-CP? ............................................................................... 184 3.4. Maxim violation? ........................................................................ 185 3.5. CP violation?: Rationality and cooperative goals ....................... 187 4. Concluding remarks: the CP and maxim flouts in humour.................. 192 5. Generalised conversational implicatures in GPs.................................. 196 6. Meaning............................................................................................... 197 6.1. What is said................................................................................. 198 6.2. Implicature types and their features ............................................ 200 6.2.1. Conventional implicature ................................................ 201 6.2.2. Conversational implicatures ............................................ 202 6.2.3. Features of implicatures................................................... 206 7. Ambiguity............................................................................................ 208 7.1. Disambiguation before determining what is said........................ 209 7.2. Manner maxim flouts – the level of PCIs ................................... 211 8. Conventional meanings vs. GCIs exploited in GPs ............................. 213 9. Conclusion: Defeasible GCIs conducive to GPs ................................. 217 Chapter Six Default and Salient Meanings in GPs...................................................... 222 1. Neo-Gricean and post-Gricean studies ................................................ 222 1.1. Default reasoning and meanings ................................................. 224 2. Bach’s implicitures .............................................................................. 227 3. Levinson’s Presumptive Meanings...................................................... 230 3.1. Heuristics and the resultant implicatures .................................... 237 3.2. The I-heuristic............................................................................. 240 3.3. Criticism of the model ................................................................ 244

viii

Table of Contents

3.4. GCIs and ambiguity resolution ................................................... 246 3.5. GPs and presumptive meanings’ cancellation............................. 247 4. Giora’s salient meanings ..................................................................... 251 4.1. Competitive models of lexical processing .................................. 252 4.2. Salience....................................................................................... 257 4.3. Salience, priming and predictive prior co-text ............................ 259 4.4. Equibiased ambiguities ............................................................... 264 4.5. GPs and salient meanings’ cancellation...................................... 267 5. Conclusion........................................................................................... 271 Final Comments....................................................................................... 273 Appendix ................................................................................................. 275 References ............................................................................................... 280 Index........................................................................................................ 307

ACKNOWLEDGMENTS

I am immensely grateful to Rachel Giora, Bruce Fraser and, primarily, to Giovannantonio Forabosco and Neal Norrick for their comments and warm encouragement. No standard acknowledgment can express my gratitude to Ken Turner, and again, Neal Norrick and Giovannantonio Forabosco. Without their invaluable help, my academic development and research on humour would not have been possible in Poland. Thanks are also due to my family, friends and students who, at times unwittingly, contributed some of the examples to my data.

INTRODUCTION

The book aims to investigate the workings of the garden-path (GP) mechanism underlying verbal humour, whose scope, for heuristic purposes, is restricted to witticisms and one-liners, referred to as gardenpaths (GPs). In essence, these are short humorous texts couched in covert ambiguity (of various types) emergent only at the final stage of on-line processing, when the initially overt and obvious (default/salient) interpretation needs to be cancelled and superseded by an alternative meaning, so far covert. Don’t drink and drive. You might hit a bump and spill your drink. It’s not that I always see a glass half empty. I always see a glass half full of poison. He told me I was in shape. It was nice with the exception that round is a shape as well. She has her looks from her father. He’s a plastic surgeon.

The primary analytic perspective adopted here conforms to the widely acknowledged incongruity-resolution framework (Suls 1972, 1983), which gives insight into on-line interpretation of humorous texts. However, all their forms cannot be fully explained by one principle. A precise description of humour processes entails an anti-essentialist approach, i.e. a detailed multi-faceted analysis of distinct language phenomena within clearly delineated borders. The goal is thus to shed light on the GP mechanism manifesting itself in particular humorous forms and boasting a variety of realisations. The multifarious methodological apparatus employed to account for the humorous verbalisations in focus is derived from semantics, cognitivism and pragmatics. The discussions will aim to verify whether and how the humorous forms can be described with recourse to postulates originating from the field of humour research and from general studies on language production and comprehension. The first chapter serves as an introduction to the work, presenting a number of methodological assumptions and familiarising the reader with fundamental concepts. Most importantly, the GP phenomenon is introduced in view of earlier theories rooted in literature on humour and grammar. GP humour is conceptualised vis-à-vis two other mechanisms

2

Introduction

underpinning short humorous texts. Finally, various forms of humour (jokes vs. forms of conversational humour) are discussed and the scope of research is narrowed down to GPs, understood as one-liners and witticisms contingent on the mechanism in question. The second and third chapters provide a background to humour studies and survey the prevailing humour theories which tackle the issue of online processing of humorous texts, verifying their aptness in the description of the chosen range of humorous phenomena. At the outset, the major theories (supposedly germane to all forms of humour) are revised. The primary focus is the incongruity-resolution approach (Suls 1972, 1983), the prevailing model in contemporary humour literature, which is also given support here, albeit with a few reservations. Those are made after several vexing issues pertinent to the framework have been raised. In chapter 3, the discussion is averted to a number of proposals widely acknowledged in linguistics, which describe the mechanisms of on-line interpretation of humorous texts, i.e. jokes, together with one-liners. The models are revisited and scrutinised for their applicability to all the three joke categories distinguished in chapter 1. It emerges that a few authors present the interpretation process as if all humorous texts centred on factors inherent to the GP mechanism. Further, in chapter 4, major subtypes of GPs are differentiated, in the light of semantic ambiguity types on which they capitalise, i.e. pragmatic and linguistic (primarily lexical, rarely together with syntactic), the latter coinciding with puns. Also, several subrealisations of GPs within both of the major ambiguity types are distinguished, with a view to gravitating towards a most adequate description of GP phenomena. To meet this goal, the notion of ambiguity, its categorisations and semantic manifestations are first expounded on. Chapters 5 and 6 offer two discrete explanatory approaches to the GP mechanism, undertaking to explain the two fundamental features of GPs in the light of broader pragmatic and cognitive proposals concerning the phenomena inherent to human communication. One of the characteristics is the defaultness or salience of the first interpretation, which, in the case of GPs, prevents an alternative meaning from arising until the final stage of the verbalisation’s processing. The other pivotal feature is the cancellability or defeasibility of this default meaning, given the co-text (verbal context) developing contrary to the meaning inferred. Chapter 5 analyses GPs within the Gricean framework (e.g. Grice 1975/1989b, 1978/1989b, 1981/1989b), verifying generalised conversational implicatures as regards their explanatory potential for GPs. These implicatures are made by default across contexts and are subject to

Humorous Garden-Paths: A Pragmatic-Cognitive Study

3

cancellation, which appears to suit the description of the first inference in GP processing. Nevertheless, the Gricean philosophy of language is complex and problematic, rendering the discussion challenging. Before this implicature type is addressed, the thorny issue of the relation between the Gricean model and humour must be elaborated. The objective is primarily to provide argumentation against the well-entrenched contention that humor (not only of the GP type) constitutes the non-bona-fide mode of communication with its own humour principle standing vis-à-vis Grice’s (Raskin 1985, Raskin and Attardo 1994) and even violates the Gricean Cooperative Principle and its subordinate maxims (Attardo 1990, 1993, 1994, 1996a, 2006). Therefore, the goal is to testify that the Gricean model is germane to the analysis of humour, with special attention paid to the type resting on the GP mechanism. The last chapter, based on independent pragmatic and cognitive theories, pursues the strand of the previous chapter and aims to account for the initial default but cancellable text interpretation in GPs. The chapter concentrates principally on the notion of generalised conversational implicatures reconceptualised as presumptive meanings (Levinson 2000, cf. 1983, 1998, 1995) and on salient meanings (Giora 1997, 1999, 2003), which are argued to correspond to the initial “obvious” interpretations of ambiguity-based chunks conducive to GPs. Albeit divergent, these approaches together help explain the workings of GPs exploiting lexical (punning) or pragmatic forms of covert ambiguity, to the effect that one meaning is effortlessly activated only to have to be legitimately rejected in favour of an initially unavailable interpretation on the strength of the final part of the verbalisation.

CHAPTER ONE METHODOLOGY AND INTRODUCTION TO GARDEN-PATHS

The present chapter serves as a general introduction to the study on humorous units pivoting on the garden-path mechanism. It opens with a number of observations on the methodology and exemplification used in the monograph. The major part of the chapter is devoted to the conceptualisation of the units of analysis, i.e. witticisms and one-liners, and the humorous mechanism in focus, standing vis-à-vis other mechanisms which can be found in the two humorous forms, as well as jokes.

1. Methodology and exemplification In literature on humour1, two main methodological strands can be distinguished, viz. universalist and descriptive theories (Ritchie 2004). The former embraces general proposals, e.g. from a philosophical standpoint, meant to account for a wide range of humour phenomena, usually failing to produce adequate descriptions of real-life instances. On the other hand, the alternative approach entails conducting detailed analyses of particular humour phenomena. The present work represents the latter perspective, focusing on garden-paths (GPs), i.e. witticisms and oneliners capitalising on the garden-path (GP) mechanism. As Ritchie (2004) rightly observes, a descriptive approach is advantageous in the sense that it holds for the collected exemplification and yields falsifiable hypotheses, as long as the researcher garners extensive data with examples which can be grouped into categories, in order to ascertain that the realisations are not circumstantial.

1

Contrary to the general tendency in topical literature to use the form “humor”, British English spelling of the term is preferred here, given that the language of whole book conforms to this standard.

Methodology and Introduction to Garden-Paths

5

Both the top-down (theory-driven) and the bottom-up (data-driven) procedures of analysis are employed. The two must be combined to provide a detailed picture of one particular humour phenomenon, together with its subtypes. The bottom-up procedure guarantees adequate description of language phenomena, while theoretical grounding and hypotheses (to be proved or disproved) are indispensable as a pre-existing skeleton for the study. This approach facilitates the application of varied theoretical frameworks from the fields of cognitivism and pragmatics, within not only the realm of humour but also general studies on discourse production and comprehension. Objectivity in the choice of data collection is difficult, and even impossible, to achieve in the study of humour. It should be acknowledged that the majority of contemporary language researchers in humour either conduct meticulous discourse analyses or corroborate their theses by providing exemplification chosen from sources specified. This method is obviously not exclusive to humour research but recurs in analyses of discourse in general. Subjectivity and introspection are then the problems many linguists face, irrespective of their field of interest. Humour researchers endeavour to circumvent the problem of subjectivisation by dint of the mutual guarantee technique, according to which an analyst is entitled to regard a given utterance as an instance of humour if it has already been perceived as such by others (OlbrechtsTyteca 1974). Secondly, empirical evidence legitimising a verbalisation as a humorous item is its occurrence in a collection of jokes or other humorous texts. However, the major flaw of such a criterion is again the subjective perspective of the authors of humour collections. In either case, linguists finally resort to introspection, which may not be a perfectly reliable criterion but apparently the only one available (Attardo 1994). The book is rich in examples. Discussions of revisited humour proposals are exemplified with instances provided by the original authors, while postulates propounded here are illustrated with GP witticisms and one-liners (and a few longer jokes in this chapter) gathered a priori, primarily by the present author. Nevertheless, for the sake of mitigating complete subjectivisation, some were also submitted by English Philology students, who found examples of GP humour (mostly on the Internet), having been briefly familiarised with this phenomenon. The fact that they were capable of this testifies that language users, guided by intuition, can determine GP examples, even if not being able to conceptualise them theoretically. The instances derive from real-life oral conversations and electronic exchanges, film dialogues, as well as Internet collections and published

6

Chapter One

volumes of jokes and quotations (see the end of the reference section). Such diversification of sources may raise serious doubts, which must instantly be dispersed. It is here maintained that regardless of whether its provenance is a real-life conversation (private or media), a fictional dialogue, a scripted interaction in a TV programme or a collection of freefloating humorous texts, each humorous verbalisation is invariably created by one speaker, a public persona, an ordinary language user or a scriptwriter, who attributes it to a fictional character. Once a humorous item it produced first, it will be repeated by language users. It must also be remembered that the success of a film script is determined by whether it reflects the language of the contemporary society, i.e. the target audience. As Wardhaugh rightly observes, realistic fictional representations of conversations can still offer valid insights into naturally-occurring interactions since they must be based on what people think happens if they are to be effective. (Wardhaugh 1992: 319)

Therefore, it can be extrapolated that contemporary scriptwriters engage in a meta-play with the target audience within an external communicative network, making use of the fictitious characters as their mouthpieces. Furthermore, the caesuras between fictional, media and reallife discourses are blurred, for individual speakers tend to incorporate various humorous chunks from the media discourse into their idiolects. It is also worth noting that numerous examples recurred in various resources. Finally, what is most significant here is the presentation of the mechanism of humorous forms in focus, rather than socio-pragmatic factors of their occurrence. Inasmuch as GP witticisms and one-liners are easily isolated from discourse, no contextual factors (if present at all, i.e. if the original source is not a collection of humorous texts) have to be elucidated, as is the case of canned jokes. In other words, the present work concentrates on humorous verbalisations which are independent of contextual factors and can be enjoyed by interpreters across situations. This is because these humorous units encapsulate all the information indispensable for understanding (Long and Graesser 1988). What is interesting, the context in which an utterance might have been produced and in which it may be used again can easily be inferred on the basis of the utterance’s meaning (Kecskes 2008). In linguistics, context normally comprises the whole gamut of factors (linguistic, epistemic, physical and social), which affect the interpretation of utterances (and nonverbal communication). From the traditional

Methodology and Introduction to Garden-Paths

7

external perspective, which is assumed here, context modifies and/or specifies word meanings. This is in contrast to the dynamic model of meaning with its broad conception of context as combining both prior and present experience (of an individual or a social group) with the present context of the outside world (Kecskes 2008). If the word “context” is used in the analyses of GP-related phenomena, it pertains to the text within a GP utterance, preceding or following its ambiguous segment. Given its verbal nature, such context is here also dubbed co-text (cf. Halliday and Hasan 1976), which encompasses the text neighbouring the GP pivot, i.e. a covertly ambiguous chunk with one default/salient interpretation. If a broader, not only language-based, context is meant, the epithet “situational” is added, or the term “situational factors” is used instead. For the sake of terminological simplification, the terms “utterance”, “verbalisation”, or “humorous text” are used interchangeably in reference to GPs regardless of whether they are encountered in spoken or written forms. Also, the term “speaker” will be equated with the term “producer” in reference to the author of a humorous statement (not necessarily the original one), while the terms “hearer”, “recipient”, “addressee”, “interpreter” or “audience” will refer to the person or people analysing a humorous utterance.

1.1. Humorousness and funniness Naturally, the perception and assessment of humour is idiosyncratic. Humour appreciation, i.e. the perception of its funniness, is an individual matter. Humour (like beauty) is something that exists only in our minds and not in the real world. Humour is not a characteristic of certain events (such as cartoons, jokes, clowning behaviour, etc.) although certain stimulus events are more likely than others to produce the perception of humour. (McGhee 1979: 6)

This individualisation of humour perception is particularly problematic from the perspective of humour researchers who feel obliged to assert that the examples chosen need not be considered funny and laughable by readers (cf. Raskin 1985; Attardo 1994, 2001). Thus, the reader of any scholarly monograph or paper on humour need not read it with a view to being genuinely amused. A similar precautionary measure is taken here. It must be stated at the outset that the focus of interest is material carrying potential for humorousness, but not necessarily funniness (Carrell 1997). Humorousness is a binary category representing a stimulus’s theoretical

8

Chapter One

capacity to induce a humorous response, while funniness is a gradable category indicating the degrees of appreciation of a humorous text, differently perceived by individuals (Carrell 1997). Differentiation is thus made between humour comprehension and appreciation, the latter being affected by motivational, emotional and situational factors (Suls 1972, 1983). The aim will be to discuss the workings and the comprehension process of the humorous GP mechanism, not to account for its appreciation and genuinely amusing value. No attempt will be made to address the question why humour is funny in an individual’s perception, or why and how humorous responses come into being, which is the focus of psychological studies on humour (see e.g. Ruch 1998). As Raskin (Aymone 2007) rightly observes, the aim of any linguistic2 analysis is to describe humour, rather than explain its funniness. It must also be highlighted that the interpretation process described is the ideal model realised by the ideal speaker and the ideal interpreter. Hence, the underlying assumption is that each humorous verbalisation is produced and interpreted according to one pattern. No claim is made that this will always be the case in all actual situations in which a particular GP is produced. However, it cannot be denied that, thanks to their skilful structuring and underlying cognitive operations, the humorous texts discussed here should normally be comprehended in the same default manner.3

2. Jokes vs. conversational humour The aim of this section is to give an overview of verbal humour’s pragmatic types so that the taxonomic position and nature of the two forms in focus can be later discussed. There have been numerous proposals of humour classifications within philosophical and linguistic literature (see Raskin 1985, Attardo 1994). In contemporary linguistic analyses, the scope of humour is frequently

2

Throughout this book, the epithet “linguistic” preceding words such as “theory”, “research”, proposal”, “analysis”, “literature”, etc. means “related to the study of language, linguistics”, while preceding “meaning”, “code”, “expression”, “phenomena”, it means “conveyed by means of language”. By contrast, collocated with humour-related terms “humour”, “ambiguity” or “GP”, the epithet carries a narrower sense and means “associated with the very surface structure of an utterance” (cf. chapter 4). 3 This generalisation may not obtain, for example, in the case of people with brain damage.

Methodology and Introduction to Garden-Paths

9

reduced to the “canned joke”. However, it must be emphasised this is only one form of humour conveyed by means of language. Jokes are often divided into conversational or canned types (e.g. Fry 1963; Attardo 1994, 2001), the main criterion being that while the latter are repeated verbatim, the former are improvised and anchored in the situational context of a particular conversation conducted by minimally two interlocutors (Brzozowska 2000). Attardo (1994), following Mulkay (1988), perceives conversational jokes as spontaneous, for they are improvised on the spur of the moment, and as situational, given that they are indigenous to a particular situational context and emerge as virtually incomprehensible to an outsider. Attardo (1994, 2001) also perceives a number of dissimilarities between the two types. Whereas canned jokes are normally repeated, recounted by a narrator and decontextualised, conversational jokes are original, non-narrative and contextually linked. Nonetheless, as Attardo (1994) also observes, this dichotomy is not a rigid one. For example, once constructed, a short conversational joke may develop into a text repeated verbatim, rather than be spontaneously produced. Additionally, as noted by Zajdman (1991), canned jokes may be adopted for the purposes of the joke-telling situation. Notwithstanding these observations, it could be argued that the former is still a conversational form, while the latter is still a canned joke. Also, the term “conversational joking” has broader applicability (see the next section), which is why it is here protested that, for the sake of clarity, the vague category “conversational joke/joking” should be avoided and substituted with second-order types, all of which represent conversational humour (cf. Coates 2007). On the other hand, in accord with the general tendency (see Attardo 1994, cf. Raskin 1985, Attardo and Chabanne 1992, Hockett 1972/1977, Ritchie 2002), the term “joke” is used in the sense of a “canned joke”, a fully-fledged humorous text. Thanks to this, the epithet “canned” is elided as redundant. In addition, the one-liner, normally conceived as a one-line canned joke, is distinguished as a discrete type of humorous text.

2.1. (Canned) jokes Even if the concept of a joke is by no means unfamiliar to lay language users, who intuitively grasp its meaning, it does pose definitional problems. Most frequently, a joke is defined in terms of its constituent parts.

10

Chapter One

Attardo and Chabanne (1992) understand a joke as a text type, a narrative which is recounted by the speaker and which is composed of the following elements: 1. Introduction – the information aimed to provide contextual factors (status of the characters, time, place), which sets “the background against which and in reason of which the punch line4 appears incongruous” (Attardo and Chabanne 1992: 109) 2. Dialogue – usually two, three or four characters are quoted 3. Punchline – the last sentence closes the narration, rendering the earlier part of the joke sensible and triggering the hearer’s laughter. It should be observed that the definition is not flawless, since dialogues and narratives are non-obligatory elements and only either element may suffice to form a joke, while the punchline can be reduced to a phrase or a single word. Hence, it appears feasible to champion the broadest possible structure of a joke and a necessity for dividing jokes into subtypes. Such an approach, much older than the one cited above, is credited to Hockett (1972/1977), according to whom, a joke comprises a build-up and a punch. This definition is often repeated in literature, albeit with different terminology. For instance, Sherzer defines a joke as “a discourse unit consisting of two parts, the set up and the punch line” (Sherzer 1985: 216). Hockett also (1972/1977) proposes an organised categorisation of jokes, juxtaposing them with sentences. He thus distinguishes between finite jokes commencing with narration and ending in a punchline and infinite ones, which are devoid of either or even both the elements. The former are further divided into: simple ones, i.e. one narrative part followed by one punchline, complex ones with many punchlines, and joke-within-joke structures. Nota bene, it could be hypothesised that there can only be one genuine punchline introducing the main incongruity (see chapter 2), while the preceding ones only contribute to the ultimate humorous effect. Among infinite jokes, Hockett (1972/1977) enumerates shaggy-dog stories and humorous dialogues, i.e. riddles. To summarise, it is the safest to posit that, in terms of structuring, a canonical joke usually consists of a variously formulated set-up (a story and/or a dialogue) and a punchline. The set-up can be differently structured, thereby yielding various joke forms. The most crucial is the distinction between the main body of a joke and its punchline. There is an easy procedure for locating the boundary between build-up and punch. Starting at the end, one finds the shortest terminal sequence, the

4

Unless quoted from literature, the term is written here as one word.

Methodology and Introduction to Garden-Paths

11

replacement of which by suitably chosen other words will transform the joke into a nonjoke. (Hockett 1977: 259)

Ritchie (2004) criticises this definition, pointing out that elements of the punchline can actually be altered, deleted or added with no loss of the joke. It is, however, undeniable that the punchline is an indispensable trigger of each joke, whose mechanisms are a few (see section 7). All in all, one might conclude that a joke is a humorous short text repeated (almost) verbatim whose content is usually irrelevant to, and dissociated from, the conversation and is geared towards amusement. Nonetheless, it cannot be ruled out that jokes can carry relevant propositional content (see Oring 2003). A few different forms can also be categorised as special types of jokes. One of the most prominent of such forms is a humour-oriented riddle realised by one speaker, who both asks the hearer an “unanswerable” question and provides the answer to it, which can be conceived of as a punchline. Another type of joke is a shaggy-dog story, i.e. a long joke whose quasi-punchline is formed by a nonsensical ending. The most significant type of joke in the present analysis is the one-liner, which will be discussed below. Because of its importance in the present work, this category is treated independently from that of a joke. The term “joke” used throughout the book refers to a fully-fledged joke (longer than one line/sentence) comprising a set-up and a punchline (but this may no always be thus if the term appears in quotations cited verbatim from works by other authors).

2.2. Conversational humour: any verbal humour but jokes Many contemporary humour linguists narrow down the scope of humour under investigation to jokes, due to their heuristic advantages over other humour forms (cf. Attardo and Raskin 1991, Attardo 1994). Nevertheless, there are a few authors (e.g. Long and Graesser 1988; Norrick 1993, 2003; Kotthoff 1996, 1999, 2006, 2007; Boxer and CortésConde 1997; Brzozowska 2000; Veale et al. 2006) who do fill the lacuna in the realm of humour research, examining forms used in interactions. Usually, articles focus on particular humour phenomena in isolation, distinguished as pragmatic categories or formal realisations, which need not be placed on any hierarchical tier in humour taxonomy, e.g. stock conversational witticisms, i.e. retorts (punning and non-punning) and quips (Norrick 1984, 1986); teasing, i.e. benevolent jibes carrying no genuinely aggressive potential (cf. e.g. Norrick 1993, Drew 1987, Hay 2000, Lampert and Ervin-Tripp 2006, Dynel 2008); banter, i.e. longer

12

Chapter One

teasing exchanges (cf. Norrick 1993, Dynel 2008); joint fantasising, i.e. fantasy/imaginary scenarios created by both interlocutors (Kotthoff 2007); irony (e.g. Attardo 2000, 2001, Partington 2007); or adversarial humour achieved via trumping i.e. humorous verbalisations subverting the communicative import of the preceding turns (Veale et al. 2006, Brône forth.). The few authors who assume a bird’s eye view of the field of humour beyond the joke are hardly unanimous regarding the terminology and the conceptualisation of the phenomena discussed. For example, Boxer and Cortés-Conde (1997), introduce the term conversational joking, equating it with situational humour, which may be deemed as causing epistemological convolution, since the denotation of the term “humour” embraces (is superordinate to) “joking”, as will be explained below. More importantly, not all conversational joking must be immanently associated with particular contextual factors, the lack of which deprives the text of its humorous force. Additionally, Boxer and CortésConde (1997) propose a tripartite function-based division of conversational joking into teasing, understood as poking fun at the interlocutor, joking about absent others and self-denigration. Obviously, conversational joking so understood does not exhaust the list of potential realisations of conversational humour. Long and Graesser (1988) explore conversational humour which they dub “wit” manifesting itself in: irony, satire, sarcasm, overstatement and understatement, self-deprecation, teasing, replies to rhetorical questions, clever replies to serious statements, double entendres (puns), transformations of frozen expressions and puns. It is self-evident that these eleven categories operate on various classificatory criteria, thus representing rhetorical figures or pragmatic types, which could easily overlap. Norrick (1987, 1993, 2003) also holds a broad view of humour realisations trespassing jokes, distinguishing conversational joking, a blanket term for a multitude of “verbal playing” types. From this vantagepoint, conversational joking encompasses primarily wordplay, punning, teasing, sarcasm, mockery, anecdotes and joint narratives (Norrick 1993). Similarly to Long and Graesser’s (1988) list, this one is not entirely satisfactory either, for the types are not distinguished according to one criterion (e.g. formal structuring or pragmatic function). Also, they are by no means mutually exclusive, e.g. teasing may entail punning, while mockery may involve sarcasm. For the sake of terminological clarity, the blanket category advocated here is conversational humour (cf. Coates 2007). While Norrick (2003) uses this term in reference to all forms of humour in conversations, including jokes, for heuristic purposes, it is here assumed that conversational

Methodology and Introduction to Garden-Paths

13

humour is any type of humour except for jokes. The term “humour” appears to be more advantageous than “joking/jokes”, inasmuch as it covers units such as neologisms (e.g. “pupkus”, i.e. a mark of a dog’s nose on a window pane) or witticisms, which are humorous but cannot be labelled as joking, owing to their “non-joke” form. Units of conversational humour range from single lexemes to whole utterances and even exchanges interwoven into otherwise non-humorous discourse. In essence, conversational humour is here treated as an umbrella term for various intentionally formed verbal chunks, both linguistic (i.e. hinged on wordplay) and non-linguistic (see chapter 4), uttered for the sake of amusing the recipient, either directly contributing to the semantic content of the ongoing conversation or diverting the flow of the conversation into the humorous mode, in which the speaker need not support the propositional content of the jocular chunk (see section 4). Even if some may argue that conversational humour is necessarily a locally structured social phenomenon (Kotthoff 2007), it can be either spontaneously formed or standardised (Norrick 1993). There may exist (exchanges of) spontaneous humorous verbalisations inextricably connected with a given situation, which are unlikely to be remembered and reused. However, a humorous unit may come into being in the course of a particular interaction, whether private or media, only to gain a permanent position within the idiolects of conversationalists, who repeat it verbatim in appropriate conversational situations. Obviously, such humorous chunks are unlikely to be lexicalised, thus always retaining their exceptionality and the quality of novelty (but for humorous conversational routines, e.g. “Look what the cat has dragged in” said upon someone entering). It is noteworthy that if the hearer knows a certain humorous item, he/she will normally not consider it as funny as it will have been upon first encounter, even if it is still humorous. Nevertheless, it cannot be denied that repeatedly heard jokes may retain their funniness (see section 7, chapter 2).

3. Witticisms and one-liners The present conceptualisation of witticisms (wisecracks, quips5 or epigrams6) is rooted in Norrick’s (1984, 1993, 1994, 2003) notion of stock 5

According to Norrick (1984, 1986), a quip is a subtype of witticism produced in response to non-verbal behaviour of the other interactionist. 6 Esar (1952) differentiates between epigrams and wisecracks, both clever remarks, depending on whether they refer to a general group of people/things or a particular person or thing, respectively.

14

Chapter One

conversational witticisms, i.e. witty and humorous units of text entwined in conversational exchanges, not necessarily of humorous nature. They are context-bound but mode spontaneous in contrast to jokes, which constitute integral parts per se, dissociated from the whole discourse (Long and Graesser 1988). However, witticisms can easily be isolated from conversations and still be perfectly comprehensible. Additionally, they may either be formed on the spur of the moment or, once verbalised, gain a permanent status in an individual speaker’s idiolect or in the language of a social group (Norrick 1993, 1994, 2003), particularly if they are used by respected members of a peer group or appear in popular media discourse targeted at broad audiences. In other words, a witticism comes into being as a novel verbalisation produced by a fellow conversationalist, a fictional character or a media persona. Having heard it, speakers may quote it verbatim in relevant situational contexts, either acknowledging its origin or pretending to be producing it spontaneously (except for allusive quotations, e.g. “Shaken but not stirred” known from the Bond series). The allure of witticisms resides in the fact that speakers may boast their wit and reap rewards for the creation of witty chunks, having obtained them from media discourse or exchanges with other interlocutors (Norrick 1993). Witticisms can be categorised according to semantic phenomena or rhetorical figures on which they operate (e.g. metaphor or irony). Additionally, they may assume various communicative forms such as definitions or comments (cf. Chiaro 1992), thanks to which they serve communicative purposes. My motto is: I’m a light eater. I start eating as soon as it’s light. [a GP definition] When you meet Mr Right, you don’t know his first name is Always. [a GP comment]

Witticisms can also be categorised regarding pragmatically oriented humour types they represent, such as teasing (jocular mocking, whose aggressive import need not be treated seriously), putdown humour (a genuinely offensive comment) or self-deprecating humour (a jocular comment directed at oneself) (cf. Dynel 2008). Yes, darling, I know you have an open mind. I can feel the draught from where I’m sitting. [a GP in the form of teasing between spouses]

Methodology and Introduction to Garden-Paths

15

Could you pass me the recipe for your ratatouille? So I can tear it up. [a GP in the form of putdown humour between two female antagonists] How I managed? Well, there are remarkable people in this world, but I’m not one of them. [a GP in the form of self-deprecating humour]

The one-liner, the shortest form of a joke, is regarded as a slippery category (Chiaro 1992), lingering on the border between prototypical jokes and conversational humour. One-liners, although predominantly considered to be canned jokes such as those produced by stand-up comedians (Norrick 1986), are here claimed to be very similar to witticisms, because they may occur in a conversation, being contextually prompted and not always interfering with its development, as opposed to jokes, which are, more often than not, disruptive. In essence, one-liners may coincide in function (and hence in form) with witticisms, or may act as short jokes for a moment diverting the flow of the interaction to the amusement of the hearer but conveying little or no informational content. This is because they often display absurdity. A woman came to ask the doctor if a woman should have children after 35. I said 35 children is enough for any woman. [a GP one-liner] I’m addicted to chocolate. I snort cocoa. [a GP one-liner, first part of the verbalisation outside the humorous frame, i.e. the speaker does love chocolate] It is not a book to be tossed lightly aside. It should be thrown with great force. [a GP witticism]

The division between the two types of GPs, i.e. witticisms and oneliners, appears not to be a clear one, as the two categories often merge in their pragmatic function. Both carry humorous potential and may be relevantly applied in conversations as responses to the preceding turns or comments on the situation and the interactant’s non-verbal behaviour. What may be the distinguishing parameter is whether the speaker supports the import of the humorous utterance or whether the latter is produced within a humorous frame (see section 4). In the case of a witticism, the speaker not only amuses the hearer but also conveys pertinent propositional meaning, while a one-liner serves humorous purposes, not necessarily communicating any propositional force. Nonetheless, even in the latter situation, some meanings may be conveyed outside the humorous

16

Chapter One

frame, e.g. that the speaker wants to relieve tension or mitigate the force of a threatening act.

4. Humorous keying Jokes and a number of manifestations of conversational humour need to be perceived as utterances produced within a humorous frame (Bateson 1953, Gumperz 1982, Goffman 1974, Coates 2007), as a special form of keying (Goffman 1974; Kotthoff 1999, 2007) or footing (Goffman 1981). Hearing humour, the recipient recognises that “this is play” (Bateson 1953) and that the interlocutor does not seriously mean what he/she is uttering. The appreciation of humour entails mode adoption, i.e. plunging into the world defined by the speaker (Mulkay 1988, Attardo 2001). Interlocutors can change frames many times in one speech encounter. Tannen and Wallat (1993) deem frames as interactive structures of interpretation sustained and modified by interacting participants. Regardless of their perspectives and goals, participants must cooperate to negotiate a frame, which is, as a result, their joint creation (Tannen and Wallat 1993). In humour, this entails not necessarily equal verbal contributions produced by both the interlocutors but the hearer’s humorous reaction to the utterance made (exclusively) by the speaker. This should not be mistaken for Goffman’s (1974) frame fabrication, i.e. the speaker’s intentional effort to produce an utterance which will mislead the conversationalist, causing the latter to have a partial or simply false notion of the ongoing interchange. This will, however, be the case of a backhanded compliment, whose genuine import is unavailable to the addressee (e.g. “You’ve got peach skin.” meant as “Your skin is fluffy.”). A question may arise if frame fabrication is actually a case of GP humour. The answer is negative, because in the latter case, even if the interpreter is initially deceived on purpose, he/she ultimately appreciates the alternative interpretation, which is the source of the humorous surprise experienced. Humour often appears within contextualisation cues (Gumperz 1982), e.g. gestures, peculiar prosody, facial expressions, code switching, social stylistics, interjections and laughter, which create humorous keying associated with contextual presuppositions indispensable for conversational inferencing (Kotthoff 2007). Very often, the speaker signals his/her intention by using adequate non-verbal cues (body language, facial expressions, intonation, etc.) before, while or after producing a humorous unit, or by uttering introductory utterances, such as “Have you heard the one?”, whereby he/she enters a humorous frame and signals that the hearer should follow suit (e.g. Mulkay 1988, Carrell 1997,

Methodology and Introduction to Garden-Paths

17

Chiaro 1992, Norrick 1993, Kotthoff 2007, Coates 2007). By contrast, introductory verbal cues do not invariably accompany units of conversational humour, which are produced more spontaneously, not always in the sense of being spontaneously created, though. However, non-verbal cues can accompany conversational humour, emphasising the speaker’s jocular intention. Although the dichotomy between serious and humorous modes appears to be clear in theory, there do exist units of conversational humour which cannot be classified as belonging solely to either. For example, witticisms are only tinted with humour and do convey meanings relevant to the ongoing discourse. This is why such forms of humour cannot be unanimously subsumed under the humorous frame, lying on the border between humorous keying and serious mode of communication. The speaker may hence convey vital information, i.e. intentionally support the message conveyed, and simultaneously amuse the interlocutor. Admittedly, one may venture a claim that almost any form of humour can carry significant information, even jokes (Oring 2003). Humour may convey potentially offensive information which the speaker refrains from communicating in a serious mode and uses humour as a mitigator or as a vehicle for retracting the message under the “only-joking” pretence (e.g. Norrick 1993).

5. On-line processing of language Crucial to the present analysis is the assumption that language production and reception are necessarily linear processes, whether language is used in a spoken or written form. Thus, meanings conveyed and received by language users rely on time dimension. It is widely acknowledged that language creation and interpretation processes take place on-line (e.g. Carston 2002), i.e. incrementally (e.g. Levinson 2000, Kamp 1981, Kamp and Reyle 1993) or linearly (Attardo 1997a). This means that in human language, words are strung out in time, even if it should be a matter of milliseconds. Consequently, the comprehension process occurs on-line, with meanings being constructed on the spur of the moment, immediately upon the recipient’s encounter of the elements evoking them, and not at the end of the utterance perception process (e.g. Hockett 1961, Van Dijk and Kintsch 1983, Just and Carpenter 1980). Also, the linear organisation of the text presupposes temporal ordering of the chunks and the affect of the preceding elements on the consecutive ones. All the processed elements form the preceding verbal context against which the consecutive material is processed, but contextually inadequate

18

Chapter One

meanings can be activated due to their salience, only to be instantly suppressed (see chapter 6). With the constituent members of a linearly organised text being analysed consecutively, the global meaning is constantly revised as more elements are added. Meanings are constructed and revised in the light of incoming information and background knowledge. On-line construction of meaning entails an active process in which the speaker integrates ongoing aspects of perceptual and conceptual information with more abstract information available from long-term memory. (Coulson 2001: 29)

On-line meaning construction is a natural process which precludes delays in interpretation and often relies on cognitive shortcuts in the process of ambiguity resolution. Interpreters resolve ambiguities, not even being aware of alternative meanings. This is because the first interpretation is rapidly chosen by default and appears to be correct, as no contextual factors point to the contrary. All the same, hearers run the risk of misinterpretation. Processing traps are conducive to mistakes triggered by the processes which facilitate regular text interpretation. Such traps are conceived of as GP constructions, which entail compensatory strategies of rejecting prior readings and activating readings so far unobserved.

6. Garden-path sentences vs. garden-path humour The term “garden-path” derives from the idiom “to lead somebody up the garden path” (in American English, “lead somebody down the garden path”) meaning “to cause someone to believe something that is not true; to deceive someone”. This idiom offers a basis for a grammatical problem as well as for the humorous mechanism proposed here. The phenomenon of garden-path sentence operates on the deception of the hearer thanks to syntactic ambiguity. In more scholarly terms, the phenomenon of a garden-path sentence is anchored in syntactic ambiguity resulting from misparsing consequent upon the decisions taken in on-line processing, which is conducive to a grammatical mistake (Pritchett 1992). The linear string of words is consistent with more than one syntactic structure, and hence sentence meaning, but only one is perceived upon the first reading/hearing. The addressee first makes an interpretation that turns out to be wrong, clashing with the next part of the text, which prompts him/her to reanalyse the sentence from the beginning and find the second (correct) interpretation. The classic example of a garden-path sentence which keeps reappearing in topical literature is credited to Bever (1970: 316), i.e. “The horse raced past the barn fell.” First, the interpreter

Methodology and Introduction to Garden-Paths

19

analyses the sentence according to the pattern, where “raced” is the main verb used in past simple, but the final element of the sentence (“fell”), which does not fit the structure, causes him/her to retrace the whole sentence anew to find another interpretation congruent with the final element as well, viz. where “raced” is treated as a passive participle. Pritchett (1992) posits that garden-path sentences are pivoted on local ambiguity, which leads to ungrammaticality but not global ambiguity. Specifically, a garden-path sentence is defined as a grammatical but unprocessable sentence which results from the combination of (a) a local parsing decision which ultimately proves not to be consonant with a global grammatical representation, and (b) the parser’s inability to perform the reanalysis necessary to obtain a grammatical representation. (Pritchett 1992: 7)

The initial ungrammaticality and resistance to processing can, however, be redeemed thanks to higher non-automatic cognitive processes. The parser needs to backtrack and reinterpret the sentence so as to change an incorrectly chosen local hypothesis that has produced the garden-path effect. Appreciating the fact of having wrongly analysed a portion of the sentence, the interpreter must go back and reanalyse, or at least correct this misanalysis (Milne 1982). Several structures giving rise to garden-path sentences can be distinguished in English. a. b. c. d. e.

Main Clause – Relative NP Ambiguity The horse raced past the barn fell. Complement Clause – Relative Clause Ambiguity The doctor told the patient he was having trouble with to leave. Object-Subject Ambiguity After Susan drank the water evaporated. Double Object Ambiguity Todd gave the boy the dog bit a bandage. Lexical Ambiguity The old train the children. (Pritchett 1992: 12)

It is interesting to observe that Pritchett (1992) categorises lexical ambiguity, represented by homonymy (e.g. “train”: “to coach”/“a means of transport”), as subordinate to the syntactic type. However, this approach is not supported here (see chapter 4), given that, most often, lexical ambiguity will not inherently entail categorical contrast, being thus independent from syntactic ambiguity. The two types of ambiguity may mesh, but it is here argued that the lexical one can be accompanied by syntactic ambiguity, not vice versa.

20

Chapter One

The garden-path effect is discussed as a syntactic problem in psycholinguistics and artificial intelligence research (e.g. Frazier 1987, Frazier and Rayner 1982, Fodor et al. 1994). There are also other perspectives, e.g. literary garden-paths with their authorial intent functionality (Fish 1989), as well as more complex GP narratives, such as short stories and novels (Mey 1991). The idea of leading the interpreter up the garden path appears to be mentioned in passing in a number of humour studies (Paulos 1980, Dolitsky 1992, Giora 2003, Coulson et al. 2006). Nevertheless, in none of these works is this statement sufficiently explained. Nor is such humour compared with the concept of a garden-path sentence. In contemporary studies on humour within linguistics, the garden-path phenomenon is also known as one of the logical mechanisms within the General Theory of Verbal Humor (Attardo and Raskin 1991, Attardo 1994, Attardo 2001, Attardo et al. 2002 inter alia, see section 3, chapter 3). It is there defined as a mechanism obtaining when the audience of a joke is led to entertain one interpretation or assumption only to have to backtrack the text and appreciate the second reading (Attardo and Raskin 1991). Postulating this logical mechanism, Attardo and Raskin present two jokes leading the interpreter up the garden path, for instance by “manipulating the acceptable level of the obvious” (Attardo and Raskin 1991: 306). George Bush has a short one. Gorbachev has a longer one. The Pope has it but does not use it, Madonna does not have it. What is it? A last name. (Attardo and Raskin 1991: 305) Should a person stir his coffee with his right hand or his left hand? Neither. He should use a spoon. (Attardo 1991: 306, quoted from Esar 1952: 21)

Unfortunately, the authors equate the mechanism of garden-path with that of false priming, as in the sentence, “The astronomer married a star”, in which the name of the profession primes the inappropriate, i.e. celestial meaning of the term “star”, whereby comprehension is hampered (Attardo and Raskin 1991: 306). Certainly, the two phenomena cannot be equated. Priming (e.g. Leeper 1935, Lackner and Garrett 1972, Meyer and Schvaneveldt 1971, see section 4.3, chapter 6) is a phenomenon potentially affecting the interpretation of incoming stimuli and prompting the interpreter to opt for one interpretation. Indeed, Attardo and Raskin’s (1991) example about the astronomer exploits this phenomenon but cannot be deemed as garden-path humour, which is characterised by an effortless interpretation of the set-up (covertly ambiguous), usually without the support of any prior contextual effects. The first meaning is later cancelled

Methodology and Introduction to Garden-Paths

21

on the strength of the following text, as the two jokes quoted above indeed show. Later, Attardo et al. (2002: 18) dub the garden-path logical mechanism a complex error in reasoning and a meta-form of logical mechanism, which directly manipulates the interpreter’s expectations. However, it is not the authors’ aim to expound on the mechanism in full detail, which is why it is hard to evaluate its importance, owing to the number of other logical mechanisms the authors distinguish (see section 3, chapter 3). Attardo and Raskin are not the first to have employed the term “garden path”. The concept “garden-path joke” should be credited to Hockett (1972/1977), who exemplifies this category with a one-liner, a genre of humour in focus here. If you hit me I’ll wax wroth, and Roth will be all slippery.

This instance, coming from the Marx Brothers’ film “Horse Feathers”, capitalises on homophonous terms “wroth” and “Roth”, as well as conventional reading of “to wax wroth”, i.e. “to be furious”. Hockett makes a number of generalisations with a view to defining this type of joke. The context is so constructed as to force one of the actual possibilities into the hearer’s awareness: the punch then reveals the ambiguity. We shall call punning jokes of this variety GARDEN-PATH JOKES. (Hockett 1972: 160) The garden-path joke is not a joke at all unless the build-up includes a pun; this pun may be called the PIVOT of the build-up (…) Evidence for pivots is the fact that, in retelling a joke, a great of the build-up can be recast in widely varying ways; but if the pivot is altered or omitted, even if the punch is delivered correctly, the joke is either no longer a joke or, if it is still funny, is a different joke. (Hockett 1972: 161-162)

First of all, the claim that garden-path jokes “force” any interpretation into the interpreter’s consciousness appears unfounded. More likely, the hearer willingly follows the path of least effort and makes inferences of his/her own accord, given that there is no contextual, specifically cotextual, information to the contrary. May (1991) also suggests this. The force that leads us gently down the garden path is effective by not being there, as such: it is the hidden force of the context, inasmuch as it presupposes a common user attitude to the world and a common user

22

Chapter One familiarity with the means the users’ language employs to describe and interpret that world. (Mey 1991: 243)

Admittedly, under the term “context”, Hockett (1972/1977) and Mey (1991) mean the set-up of the joke or the body of a narrative, respectively, which the hearer perceives as bearing one unequivocal interpretation until the punchline discloses another meaning. In his paper on a non-humorous garden-path mechanism, Mey (1991) concentrates not so much on the sources of latent ambiguities and their default interpretations as on the context which permits such to occur. By letting the context speak for itself and not allowing ourselves to interfere (positively or negatively) with its message, the resulting absence of “marks” will bring out those characteristics which the users hold as evident or normal for a particular situation. Using the context, we put the default to work; in other words, we use the pragmatic forces that are already there, in the context, to move the reader or hearer into one particular direction. The ambiguity is sustained, without the reader being aware of it. (Mey 1991: 243)

Mey subscribes to the standard pragmatic perspective that “context determines the sense of any utterance and of any of its parts” (Mey 1991: 241), with the revealing context being incomplete until the very end. By contrast, it is here protested that this process is the reverse in short humorous units. Default interpretations are activated first with the context posing no obstacles (see chapter 6), before the final stage of textual processing arrives, where information to the contrary emerges, causing retroactive re-evaluation of the default. As regards the source of ambiguity, it is self-evident that Hockett (1972/1977) reduces the scope of garden-path jokes to those which operate on puns, i.e. surface-level (mostly lexical) ambiguity, which does not, however, exhaust all ambiguous realisations, because there are also punless jokes in which the hearer’s inferences are manipulated in a very similar manner. It is also Yamaguchi (1988) who appreciates that a garden-path joke need not rely on puns and redefines Hockett’s (1972/1977) definition of a GP joke. A joke in which the context is potentially ambiguous in that it has potentially a first and a second reading, the former being replaced by the latter at the end of the joke. (Yamaguchi 1988: 325)

Observing the existence of non-punning garden-path jokes, Yamaguchi (1988) analyses them as violations of Grice’s maxims (whether they do

Methodology and Introduction to Garden-Paths

23

violate maxims will be discussed in chapter 5), i.e. “these violations of the maxims are used as a resource for ambiguity, just as a pun is used in the garden-path pun” (Yamaguchi 1988: 327). Discussing the case of a garden-path novella, Mey also notes that a GP is “a balancing act on the tight-rope of some syntactic and/or semantic ambiguity or implicitness” (Mey 1991: 236). Here, two major types of semantic ambiguity necessarily underlying garden-path humour will be distinguished, i.e. lexical and pragmatic (see chapter 4). This is why the original conceptualisation of the garden-path mechanism in grammar (e.g. Bever 1970, Pritchett 1992, Milne 1982, Frazier 1987, Frazier and Rayner 1982, Fodor et al. 1994) needs to be modified to meet the purpose of the study on humour. Essentially, garden-path sentences as such are hardly humorous. She told me a little white lie will come back to haunt me. The cotton clothing is made of is grown in Mississippi.

While garden-path sentences are dependent on syntactic ambiguity, humorous utterances reliant on the garden-path mechanism are different. Although there do exist instances of exploitation of syntactic ambiguity for humorous purposes (e.g. Groucho Marx’s one-liner or e. e. cummings’s response, both of which entail the syntactic position of prepositional phrases), this phenomenon is an infrequent one. One morning I shot an elephant in my pajamas. How he got in my pajamas, I don’t know. (Groucho Marx) Would you hit a woman with a baby? No I’d hit her with a brick. (e. e. cummings)

The survey of the corpus of GPs collected for the purpose of the present study yields a conclusion that humorous use of the GP mechanism hinges primarily on lexical ambiguity and pragmatic ambiguity, while the former may rarely coincide with syntactic ambiguity, predominantly the case of categorical contrast, e.g. the famous one-liner, “Time flies like an arrow. Fruit flies like bananas”, where “flies” is either a verb or a noun and “like” is a verb or a conjunction. Taking all the above into account, the GP mechanism, which will be discussed in this book from various theoretical standpoints, can now be briefly characterised. It should be highlighted at the outset that the whole GP inference process is rationally devised by the speaker to be followed by the hearer. The initial part of a humorous verbalisation, i.e. the lead-up,

24

Chapter One

presents itself to the interpreter as unambiguous and subject to effortless interpretation, while its ending, i.e. the switch, stands in the relation of incongruity (see chapter 2) with this first interpretation and forces the interpreter to revisit the preceding part of the verbalisation (the lead-up) in order to discover its second interpretation compatible with the ending. The interpreter must backtrack the lead-up, ignoring its initial interpretation, and seek another interpretation congruent with the switch. The sine qua non for each GP is covert (initially unobserved) semantic ambiguity, primarily lexical and pragmatic (see chapter 4), giving rise to two inferences, i.e. the one that is default/salient (see chapter 6) but “wrong”, and the “right” one. All the italicised terms appearing in this paragraph will be clarified in the course of the monograph.

7. What the GP mechanism is not Chapter 3 will aim to vet competitive postulates concerning the mechanisms underlying jokes, substantiating that a few authors, using various methodologies and terminologies, analyse jocular texts as if they all relied on what is here dubbed the GP mechanism. Also, a number of researchers make similar claims in passing. The comic effect arises when an alternative, non-favored and therefore non-expected interpretation is revealed, at the punchline, as the correct one. (Dascal 1985: 95) The humorous effect comes from the listener’s realisation and acceptance that s/he has been led down the garden path (…) In humour, listeners are lured into accepting presuppositions that are later disclosed as unfounded. (Dolitsky 1992: 35) My approach to the internal organization of jokes (…) the punch semantically reverses the sense we would expect from the build-up, and forces an unexpected sense to our attention. (Norrick 2001: 258)

Contrary to what the authors quoted above may suggest, not all jokes entail ambiguity, and hence two competitive interpretations of the set-up, only one of which is initially overt. Nor do all jokes end in a punchline bringing the ultimate suppression of the meaning inferred and causing the enforcement of an alternative interpretation, so far unobserved. In other words, the GP mechanism is not inherent to all humorous forms (witticisms, one-liners and jokes). It is here argued that two different mechanisms can be distinguished. By analogy to the term “garden-path

Methodology and Introduction to Garden-Paths

25

mechanism”, those are dubbed the red-light mechanism and the crossroads mechanism. Although all the three categories of jocular texts (governed by three different mechanisms) can be subsumed under the incongruityresolution model (see sections 5, chapter 2 and section 1, chapter 3), according to which the final element of a text enforces incongruity to be ultimately resolved, each of them features its peculiarities, which will now be explored. It should also be mentioned that, even if rarely, jokes may centre around mechanisms characteristic of two types, e.g. the crossroads mechanism (incomprehensibility of the set-up) can be entwined with the GP mechanism (hidden ambiguity of the set-up). For the sake of clarity, such merger examples are not discussed here. Given the general tendency in humour studies, the discussion below is illustrated not only with witticisms and one-liners but also with jokes, even though those are not in the focus of this book and will not be used further, unless illustrating other authors’ proposals.

7.1. The red-light mechanism A humorous unit may end with a surprising element, which, however, neither proves the set-up ambiguous nor invalidates any earlier inference, being in stark contrast (incongruity) with it, but provides additional surprising import. Such a humorous form draws on surprise formed by an unexpected piece of information, which can be deemed as incongruous with the preceding part of the text. The final portion of the text contributes information as to which the interpreter will have made no inferences. This process can be metaphorically called the red-light mechanism. Accordingly, the hearer follows the interpretation path unobstructed until he/she needs to stop upon encountering the surprising red light diverting the interpretative process to a destination which could not have been envisaged earlier. The red light corresponds to the switch in a one-liner or the punchline of a joke introducing the element of surprise and incongruity. After this momentary pause caused by the red light, the hearer continues the interpretation process by resolving the incongruity, i.e. finding a new route to be followed. Also, the pause often remains unobserved if incongruity resolution does not pose major problems. Each of the humorous witticisms and one-liners in this section is contingent upon some surprising information coming at the stage of the switch, i.e. a punchline reduced to a single word, phrase or clause. Admittedly, the categories distinguished below constitute only a sample of the “surprising-end” realisations found in units of conversational humour. Firstly, those can operate on the surprising and explicit cancellation of

26

Chapter One

(part of) the meaning conveyed by the earlier part of the utterance. Such forms should not be mistaken for GPs. The parameter of cancellation is not a sufficient one, as this is a matter of negation, rather than defeating the hearer’s interpretation, on the strength that an alternative but latent one is also possible. Only two things are certainly infinite – the universe and human stupidity but I’m not sure about the former. Let me help you solve this problem. Just a second…why should I? It’s not mine! Look, I’m ageing. I’m turning 30. Hey, I’m not! I’m only 29.5!

That cancellation alone does not suffice for a verbalisation to be regarded as a GP is also corroborated by quips (coinciding with teases, self-denigrating humour or putdowns, not necessarily dependent on semantic ambiguity) embraced by a humorous frame, within which speakers do not need to support the truth of their verbalisations. If their objectives are not transparent, having produced quips, speakers may retract their words, asserting that they were “only kidding” or otherwise communicate their humorous intentions underlying their words, which interpreters may have taken as genuinely meant. Put a plastic bag on your head. It will solve all your problems… (laughter) I sing James Blunt’s “You’re beautiful” every morning as I start shaving. (said by a woman)

Secondly, the switch may undermine information the interpreter normally takes for granted, although not provoking the negation/suppression of the import carried initially by the first part of the semantically unambiguous verbalisation. I’m not 50 – I’m 18 with 32 years’ experience. I used to be a bartender at the Betty Ford Clinic.

The examples above capitalise on the revelation of information which will not normally be expected or which contradicts common-sense assumptions, to the effect of being paradoxical, e.g. the famous rehabilitation centre for addicts has a drink bar.

Methodology and Introduction to Garden-Paths

27

You’re rich. You’ve got a, mansion a car… not as good as mine, though. The song is supposed to seem virgin-like so pretend.

In either of the two conversational quips above, the speaker closes the utterance with surprising (and biting) information, which appears to act as a humorous aside, not affecting the communicative force carried by the preceding textual material. He collected cardboard boxes and once he was able to buy a loaf of bread, some butter, a second-hand suit and a new Porsche. Some girls are beautiful, others are quite cute, and still others are average or just ugly or even unsightly. And there are, of course, those from the Electric Faculty at Technical University.

The hyperbolic expressions above, albeit developing gradually and preparing the recipient for the forthcoming increase, are still built on the element of surprise coming with the hyperbolic climax, which simply cannot have been envisaged. What is most significant, GPs should not be equated with humorous texts which finish with ambiguous chunks but manifest no ambiguity in the set-up/lead-up. This type of humorous text coincides with what Attardo (1994) and Attardo et al. (1994) dub the non-distinct configuration (see section 5, chapter 3). A very similar type of humorous verbalisation will also be found within the crossroads type, discussed below. War does not determine who is right but who is left. I’m so ugly that even the ceiling won’t fall for me.

The point of similarity such verbalisations share with GPs is that initially the final ambiguous element (“left”: “direction”/“remaining” and “fall (for)”: “feel attracted to”/“fall down on”, preceded by a wrong preposition) is interpreted according to the salient meaning, additionally primed by the preceding part of the text (as in Attardo and Raskin’s example of an astronomer that married a star, which they wrongly classify as garden-path humour). This reading, however, must instantly be rejected due to its incongruity with the preceding text, being thus ousted by the alternative reading of the ambiguous chunk. On the other hand, both the meanings, which are activated consecutively, may be retained as congruous and relevant, yielding a double-retention pun carrying two logical interpretations (Dynel forth.).

28

Chapter One TV is mass /distrækœԥn/. (destruction/distraction). What to expect when you are expecting. (a book for mothers-to-be)

Given the multifariousness of conversational quips and witticisms and the rhetorical figures they may employ, the above types should be treated only as a sample. Their common denominator is the surprising and incongruity-producing chunk coming at the end. This “surprising-end” mechanism can also be found in jokes. The major difference in the latter is that, aware of the jocular frame, the hearer will be somehow prepared for a twist, albeit being oblivious to its nature. A retired gentleman went to the Social Security office to apply for Social Security. The woman behind the counter asked him for his driver’s license to verify his age. He looked in his pockets and realised he had left his wallet at home. He told the woman that he was very sorry but he seemed to have left his wallet at home. “I will have to go home and come back later.” The woman said, “Unbutton your shirt.” So he opens his shirt revealing curly silver hair. She said, “That silver hair on your chest is proof enough for me” and she processed his Social Security application. When he gets home, the man excitedly tells his wife about his experience at the Social Security office. She says, “You should have dropped your pants. You might have got disability too.”

The interpreter follows the story-line of the joke, finding no element inexplicable (even if finding the Social Security worker’s behaviour strange) and adding the incoming information to the so-far formed interpretation of the text. The punchline coincides with the wife’s last sentence, which adds completely new information. The interpreter must draw new inferences from the last sentence, which does not entail cancellation of any information generated on the basis of the set-up. A man is eating a stew at a restaurant. Suddenly he feels something sharp in his mouth. The object turns out to be an earring. The man instantly starts rebuking the waiter, who says, “I’m terribly sorry but you can’t imagine how happy the chef will be to get it back. It’s over three weeks since she lost it.”

In the joke above, the punchline redirects the focus of the narrative and reveals new information about the earlier-mentioned stew, without disclosing any hidden ambiguities or causing major reinterpretations. In fully-fledged jokes, it is quite common for the punchline to refer to the preceding portion of the text and bring out information which the hearer

Methodology and Introduction to Garden-Paths

29

could not have inferred (even via presuppositions) and, therefore, could not have been led to make a wrong inference about, due to the lack of any cues in that direction. Finally, similarly to the phenomenon found in shorter humorous forms, jokes of the red-light type may also end in ambiguity, i.e. an ambiguous element of the punchline (rather than the set-up, which is the case of GP jokes). Moreover, the first interpretation of this ambiguous element may be consistent with the interpretation generated upon the set-up and it is only later that the interpreter appreciates the alternative meaning. An English bishop received the following note from the vicar of a village in his diocese: “Milord, I regret to inform you of my wife’s death. Can you possibly send me a substitute for the weekend?” (Raskin 1985: 106)

The punchline in the form of a request operates on referential ambiguity, thus yielding the twofold meaning that the vicar wants the bishop to find a priest to substitute him during weekend masses, or a woman who will replace his late wife. It is debatable which of the meanings will be inferred first, being more salient, and which is actually the one intended, i.e. the one not to be rejected (for a different view, see Martin 2007: 65). Nota bene, analysing a riddle in which ambiguity occurs within the punchline, viz. “Why did the cookie cry? Because his mother had been a wafer so long.”, Shultz (1974) and Pepicello and Weisberg (1983) have different viewpoints as to which meaning is observed first, i.e. “away for” or “a type of cookie”. Nonetheless, it is undeniable that ambiguities closing jocular texts, however resolved, foster the effect of surprise, ambiguity and incongruity, halting the inferential process and making the hearer stop and ponder on the resolution.

7.2. The crossroads mechanism The third mechanism is similar to the red-light mechanism (and also the GP mechanism, for that matter) in that the final textual element also introduces new information causing surprise. The crucial difference is that in this new category, the set-up includes incomprehensible premises, surpassing absurdity typical of many forms of humour governed by their own “local logic” (Ziv 1984). It could then be argued that such texts entail two incongruities, one causing puzzlement within the set-up, and the other (more important) within the punchline. The interpreter is at a loss to make rational inferences, not to mention default ones, and cannot fully comprehend the set-up without the speaker’s adding more information. In particular, the hearer cannot produce any sensible interpretation of (some

30

Chapter One

part of) the text or the characters’ words or actions, expecting them to be later resolved. This is the situation opposite to that typical of the first stage of GP comprehension process, in which the interpretation process takes place by default, or the red-light mechanism, where on-line processing of the set-up progresses unobstructed. However, the point of resemblance between this interpretative pattern and the GP mechanism is the sudden discovery as to the import of the first part of the text, preceded by incongruity perception. This type of mechanism underlying humorous texts can be metaphorically dubbed the crossroads mechanism, which symbolises the second-order incongruity, i.e. an inexplicable element, the hearer encounters while interpreting the set-up. The interpreter does not know what road to take, not even observing any alternatives. Hence, the hearer relies on the speaker’s instructions, trusting that he/she will reach some destination in the end. Indeed, the hearer eventually is provided with a surprising switch/punchline. It initially produces another incongruity with its surprising import, which is almost instantly resolved, while the preceding incomprehensible chunk is explained. Obviously, in short humorous texts, the two processes occur instantly one after another. Shhh. You will destroy the harmony in nature. You are shouting so much that stags in the rut can’t hear one another.

In the example above, the part of utterance following the “shhh” sound appears irrelevant, and thus incomprehensible, even if the interpreter assumes that it must be somehow related to the act of hushing him/her. A full understanding comes with the second sentence, which conveys the derogatory content of the statement. A similar mechanism can also be realised with a lexically ambiguous chunk coming in the final position of the text, which is parallel to the textual type found in the red-light category, but revolves not only around the surprising end but also around the initial incomprehensibility of the preceding text. There is no gravity. The Earth sucks. A great many people have a soldier’s stomach – everything they eat goes to the front.

Initially, both the witticisms are virtually incomprehensible, based on a premise which either violates a common presumption (gravity exists) or which escapes understanding (what is a soldier’s stomach?). The second part of either text is anchored in a pun, i.e. in polysemy (“suck”: “be a failure”/“draw powerfully” and “front”: “a movement against opposing

Methodology and Introduction to Garden-Paths

31

forces”/“the most forward part”) yielding two meanings, one of which resolves the problem emergent from the preceding part of the verbalisation. A notable subcategory is formed by creative similes and metaphorical comparisons appearing incomprehensible to interpreters, who cannot find the tertium comparationis, i.e. the point(s) of similarity, between the incongruous elements juxtaposed. The punchline elucidates the compared entities’ shared characteristics, admittedly resistant to the hearer’s appreciation in the independent process of inference. Also, the comparison may rely on a punning (ambiguous) lexeme coming at the end (as in the case mentioned above), whose (two) meanings refer to either of the elements compared, as in the second example below. Men resemble women because of beer. When they drink it, they gain weight, talk excessively and with little sense, become emotional and can’t drive. Awards are like haemorrhoids… sooner or later every asshole gets one.

In the case of a longer crossroads joke, an incomprehensible element may arise at any stage of the set-up but always before the punchline. The interpreter is aware of the humorous frame and of the joke signalled by the speaker, and predicts a resolving twist in the punchline, the nature of which he/she cannot envisage, not having heard it before. This unpredictability surrounding what is tacitly expected is the key to the humorous effect. Beyond a shadow of a doubt, it cannot be said that the interpreter is led up the garden path, as he/she does not draw any default inferences. If any are made, whether later subverted or not, they are not effortless, which would be the case of the GP mechanism. An old woman walks into a bank and asks for the manager. He takes her into a small room, and she asks him if she could take out a loan of $500.000. He asks her how she is ever going to pay it back. She replies, “I make bets with people, and win their money. Take this for example: I’ll bet you $25,000 that your balls are square.” “You’re on,” says the bank manager, feeling confident. The next day the elderly woman and her accountant come by the bank to see the bank manager. “Now then,” she says, “to make this fair, I have brought along my accountant. Now pull your trousers down.” As he has done it, she asks, “I’ll just get a wee feel now, to make sure, okay?” The manager agrees. While the woman is touching the manager’s testicles, the accountant is banging his head on the wall. “They’re round!” cries the delighted bank manager and asks, “By the way, what’s wrong with your accountant?” “I bet him $500,000 that I’d hold the bank manager by the balls on Friday morning!”

32

Chapter One

The interpreter is probably flabbergasted by the old lady’s actions and later the accountant’s behaviour but, in all likelihood, cannot find the motivation for it prior to the emergence of the punchline, i.e. the old lady’s last utterance. The final surprising piece of information resolves the earlier doubts or queries the interpreter will have had, failing to have dispelled them on his/her own. The joke below, derived from the “Ally McBeal” TV series can also serve as an example here, with the inexplicable element occurring much further in the joke than in the previous case. Two fleas meet each year for their vacation in Florida. One year, one of the fleas comes into the hotel where they always stay, shivering and complaining of the cold. “What’s the matter?” asks the other flea. “It’s summer in Florida!” “Yeah, well, I came all the way down in the moustache of some biker, and the wind chill and all just got to me,” the first flea responds, with not just a little touch of bitterness. “Ahh. You just don’t know how to do it. This is how you travel to Florida, first, go hang out in an airport bar and have a few drinks. Then, find a nice stewardess who’s working a flight to Florida. Jump onto her leg, make your way up and snuggle into her … well, you know. Then, you can just sleep the whole way there!” “What an idea!” says the first flea. “Next year for sure!” So, the next year the flea arrives in the same state, cold and angry. “What went wrong?” asked the wiser second flea, who had been there a while and had already finished off a few drinks. “I did just what you said: airport, bar, stewardess, and … err … you know. So I go to sleep, and when I wake back up, I find myself back in the damn biker’s moustache!”7

The unfortunate flea’s second arrival produces incongruity and, initially, the interpreter’s lack of understanding, since this is not what is supposed to have happened. This conundrum is solved at the stage of the punchline, viz. the flea’s final explanation, which helps the interpreter infer what must have occurred only to result in the given state of events. The interpreter has to complete the information missing from the account, drawing taboo inferences in order to arrive at an understanding of the poor flea’s fate.

8. The structure of GP jokes, witticisms and one-liners The GP mechanism is considerably different from the red-light and the crossroads mechanisms. The switch/punchline evokes the concealed sense of the preceding text, bringing to light its ambiguity, and cancels the first 7

Bubel and Spitz (2006) discuss this joke as well, albeit from a different theoretical perspective.

Methodology and Introduction to Garden-Paths

33

effortlessly made (default/salient) interpretation. The prerequisite for GP humour (any humorous GP text) is thus twofold. The first part of the text must entail covert ambiguity, with only one meaning being effortlessly accessible, and the second part of the text must invalidate the earlier inference and prompt the interpreter to backtrack and reprocess the initial part of the text to appreciate an alternative, hitherto unobserved meaning, compatible with the import of the final textual chunk. As already signalled, the present analysis of the GP mechanism in humour is restricted to its occurrence in one-liners and witticisms, collectively referred to as garden-paths (GPs). The mechanism is by no means exclusive to such short forms of conversational humour. However, the discussion of these jocular verbalisations facilitates the most adequate description of the mechanism per se, without having to account for many accompanying elements and factors found in longer jokes. As already mentioned, these long forms are multifarious in structuring and may also include various embellishments, indispensable for successful joke-telling but immaterial to the operation of the mechanism as such. Mey (1991) shows that ambiguity in a longer GP text, a novella, needs to be sustained longer. While the reader is very often exposed to it thanks to innuendos, the ambiguity is normally not appreciated and remains latent until the very end. Several complications found in jokes are listed below. Nonetheless, the pivotal elements of a GP is the two-fold interpretation of the leadup/set-up, initially perceived as monosemic, and the switch/punchline which is incongruous with the interpretation and causes backtracking in search for a more compatible interpretation. In a GP, the developing text leads us from the starting line to the (unexpected) “denouement” of the punch line, in which one of the possible interpretations of the narrative (viz., the one that has been hidden from our sight) is suddenly revealed as the real one. (Mey 1991: 237)

In jokes, the garden-path process may be instigated by an element within the narrative, e.g. a backgrounded piece of information, a detail omission in the narrative or a misformulation, which recedes to the background and is interpreted by the hearer so that it should fit the context of the joke (cf. Yamaguchi 1988). A lawyer who works in Texas gets a call about an emergency which requires him to immediately fly out of the state for a short period of time. He doesn’t even have enough time to pack, so he calls home to tell his wife about the trip. The maid answers the call, but is quite hesitant about putting his wife on the phone. After quite a bit of interrogation, she admits that the

34

Chapter One wife is upstairs in bed with the mailman! The lawyer is furious, and wants to rush right home, but of course there is this emergency he must take care of. So instead, he tells the maid to go get the gun from the desk drawer and kill both his wife and the mailman. The maid protests, but the lawyer explains that under Texas law it is legal to kill your adulterous spouse and his/her lover. Using his silver tongue, he finally convinces the maid to do it, offering a hefty bonus as a reward. She puts down the phone, and soon the lawyer hears the sound of two gunshots, a scream, a dozen or so loud thumps, and finally, two splashes. The maid comes back to the phone. The lawyer asks, “Did you kill them?” “Yes,” she replies. The lawyer questions her again, “What did you do with the bodies?” “I threw them in the pool,” she responds. There is a brief pause from the lawyer. He asks her, “Did you say the pool?” “Yes! I threw them in the pool!” she says. “Uh, is this 555-8234?”

In the joke above, the GP mechanism relies on the main protagonist’s action, i.e. his having dialled a wrong number, of which he is unaware. So is the hearer, being misled as to the interpretation of the whole sequence of events. In other words, there is no single verbalisation that carries the ambiguity, while the interpretation of the chunk “calls home” is taken for granted to mean a successful attempt. The whole story has a second interpretation revealed at the very end of the joke, i.e. that the character intended to call his own home but reached a different number. The pragmatic ambiguity of this joke resides in the default presumption that if the protagonist calls home, he does manage to do that, rather than dial a wrong number, as he really does. This piece of information is concealed from the listener. What is significant, as the instance presented above also shows, jokes may entail a character, together with the interpreter, being led up the garden path, causing what could be called double layering. This double layering may perhaps cushion the impact of the GP mechanism and its deception, to which the interpreter is not directly exposed as in the case of conversational GPs. In jokes, the imaginary character is the deceived or misunderstanding party, while the hearer is distanced from the joke scene. This observation coincides with Ritchie’s (2006) view of the hearer’s or story character’s viewpoints. The interpretation and the subsequent reinterpretation may not take place directly within the audience’s viewpoint but within a nested viewpoint available to the audience. In the joke presented above, the GP effect is the result of the character’s own action, while in the example below one character is deceived by another’s behaviour.

Methodology and Introduction to Garden-Paths

35

A guy at a bar is sitting and looking at his drink. He stays like that for half of an hour. Then, this big trouble-making truck driver steps next to him, takes the drink from the guy, and just drinks it all down. The poor man starts crying. The truck driver says, “Come on man, I was just joking. Here, I’ll buy you another drink. I just can’t stand to see a man cry.” The weeping man responds, “This day is the worst of my life. First, I fall asleep, and I go late to my office. My boss, outrageous, fires me. When I leave the building, to my car, I find out it was stolen. The police say that they can do nothing. I get a cab to return home, and when I leave it, I remember I left my wallet and credit cards there. The cab driver just drives away. I go home, and when I get there, I find my wife in bed with the gardener. I leave home, and come to this bar. And just when I was thinking about putting an end to my life, you show up and drink my poison.”

In this joke, the final sentence of the long utterance produced by the unfortunate man sheds new light on his earlier non-verbal response, i.e. his reaction to the truck driver’s drinking from his glass, which the latter (just as the joke addressees) will have interpreted as a childish or at least highly emotional reaction to having a drink stolen. As the story progresses, the hearer additionally motivates the man’s reaction with his misery, while the real reason emerges at the very end of the joke. Nota bene, even if ostensibly similar to the crossroads joke on fleas, this one is significantly different, because here the interpreter does arrive at a default, albeit “wrong” (as it emerges at the stage of the punchline) explanation for the man’s reaction, while in the flea joke, the reason for the insect’s state is not available until the punchline. Either of the GP jokes above relies on a pivotal (ambiguous) element which is not a verbalisation produced by a joke character. However, very often it is a character’s utterance, whether in direct speech or reported, not an element of the narrative (e.g. a description of a character’s action) that is conducive to the GP mechanism. A woman walks into the bar with a duck on a leash. The bartender says, “Where’d you get the pig?” The woman says, “This is not a pig. It’s a duck.” The bartender says, “I was talking to the duck.”

In the joke above, the bartender offends the customer by deliberately producing a statement aimed to deceive the latter. The hearer is equally deceived but does not face the putdown as the joke character does. On the other hand, a garden-path verbalisation may be produced, as if unintentionally, by one of the joke characters.

36

Chapter One Ms. Collins sat nervously before the psychiatrist and said, “I caught my son and the little girl next door examining each other with their pants down.” “That’s not so unusual, said the doctor. “Children like to compare things. I wouldn’t get upset.” “I’m already upset,” said the woman, “and so is my son’s wife.” (Yamaguchi 1988: 325)

This joke hinges on the patient’s verbalisation which, allegedly unbeknownst to her, carries two meanings, only one of which is initially perceived by the in-joke interlocutor and the audience. Within the joke’s world, this is the case of unintended misunderstanding, the doctor being unintentionally led up the garden path. Sometimes none of the joke characters is deceived just as the interpreter is. For example, the annulment of the first interpretation comes as a result of the second persona’s intended misinterpretation of the preceding utterance produced by the interlocutor. Again, the hearer is somehow dissociated from the comprehension process, following the interpretative path taken by the character. At the interactional level between joke characters, such interchanges may be taken as non-punning or punning retorts (Norrick 1984, 1986, 1993), or a form of trumping (Veale et al. 2006, Brône et al. 2007, Brône in press), which entail finding a second sense in the first interlocutor’s utterance. “Is the doctor at home?” the patient asked in his bronchial whisper. “No,” the doctor’s young and pretty wife whispered in reply. “Come right in.” (Raskin 1985: 32) “Waiter what is the fly doing in my soup?” “I believe this is the backstroke.”

In either of the two jokes above, the characters wittily respond to the opening questions grounded in pragmatic ambiguity not perceived by the producers thereof or by the joke hearer. As the responders bring out the hidden meanings, the interpreter feels misled and experiences the humorous GP mechanism. Moreover, jokes can be structurally complex. The set-up accumulates tension before the GP mechanism is enforced, with the punchline ultimately provoking the reinterpretation of only a small portion of the preceding text. An international feminists’ meeting. One of the delegates postulates, “Sisters! No more female exploitation! We won’t cook or do the laundry for men.” The motion is unanimously accepted. At the next meeting, some delegates share their experiences. An English woman starts, “I returned

Methodology and Introduction to Garden-Paths

37

home and said, ‘John, from now on, I won’t cook or do the laundry.’ The first day I saw nothing, the second day I still saw nothing, and the third day I saw John boiling an egg.” The whole audience applauds. A French delegate starts speaking. “I returned home and I said, ‘Pierre, from now on, I won’t cook or do the laundry.’ The first day I saw nothing, the second day I still saw nothing, and the third day I saw Pierre washing his underwear.” A standing ovation follows. A Polish woman starts speaking “I returned home and I said, ‘Janek, from now on, I won’t cook or do the laundry.’ The first day I saw nothing, the second day I still saw nothing, and the third day I still saw nothing, but the fourth day I could see with my left eye.”

The bulk of the text is not rooted in the GP mechanism, for it entails no ambiguity, which is introduced only with the appearance of the Polish woman. It is at the stage of the last clause that the interpreter realises that her utterance needs to be retraced for the second interpretation capitalising on the polysemous word “saw”. Additionally, the first interpretation of the polysemous item is strengthened by the priming effect (see section 4.3, chapter 6). Accordingly, the earlier part of the joke operates on a repeated pattern, i.e. women’s similar accounts contingent on the meaning of the word “saw” in the sense of “observing”, which is ultimately replaced by the second sense “being capable of perceiving”. Witticisms and one-liners serve as adequate material for the present analysis thanks to their brevity, which facilitates a more meticulous discussion of the workings of the GP mechanism per se. Despite their formal succinctness, GPs manifest structuring similar to many fullyfledged jokes, being comprised of a minuscule version of the set-up and the punchline. However, this terminology is inadequate, since the punchline is reduced to a short sentence, a phrase or even a single word. Some authors (e.g. Coulson 2001; Giora 1991, 2003; Norrick 1986; Ritchie 2004) suggest that the term “punchline” can also be used in the analysis of one-liners, referring only to parts thereof, which is here believed to be confusing, owing to the composition and the meaning of the blend “punchline”. Attardo (1997a), on the other hand, posits that oneliners should be equated with punchlines. Such a view subverts the common conceptualisation of a punchline as the final part of a joke. Even if one-liners are punchlines, they must have a second-order punchline element. GP one-liners (and witticisms) have a condensed structure embracing the two composite elements typical of jokes (cf. Norrick 2001). Given the forms of humour analysed, i.e. GP witticisms and one-liners, different terminology is employed in reference to the two fundamental chunks of the text. The first part of a GP yielding covert ambiguity is dubbed the lead-up, while the final portion of the text which enforces the

38

Chapter One

second sense and brings to light the ambiguity of the preceding chunk is dubbed the switch. It is not always the whole lead-up but only its ambiguous chunk that generates the necessary covert ambiguity. By the same token, it may be only the pivotal part of the switch, not the whole of it, that is directly responsible for the change of the lead-up’s meaning. Those less important elements may be, for instance, those which guarantee grammaticality, e.g. subjects, disjuncts, conjuncts or adjuncts. The central element of the lead-up is a lexically or pragmatically ambiguous chunk whose default or salient meaning is the only one perceived upon the first reading/hearing of the utterance. In the case of some pragmatically ambiguous GPs and lexically ambiguous GPs based on phrasemes, often the whole lead-up coincides with the ambiguous chunk. The remaining part of the GP is dubbed co-text, preceding or following the ambiguous chunk, the former type being non-obligatory. The following co-text may then either belong to the lead-up (found only in some GPs) or coincide with the switch (the necessary element). Below a few GPs couched in different ambiguity types are presented and partitioned into chunks. I’m a one-man woman | One at a time. Ļ Ļ Lead-up switch I’m | a one-man woman. | One at a time. Ļ Ļ Ļ preceding ambiguous following co-text chunk co-text (the speaker is faithful to one man ĺ the speaker has had relationships with many men, one after another) As you know, I don’t repeat gossip so | listen carefully. Ļ Ļ Lead-up switch As you know, I don’t | repeat gossip | so listen carefully. Ļ Ļ Ļ preceding ambiguous following co-text chunk co-text (the speaker never gossips ĺ the speaker will not repeat the news after he/she says it once)

Methodology and Introduction to Garden-Paths

39

Sex on television can’t hurt you | unless you fall off. Ļ Ļ lead-up switch Sex on television | can’t hurt you unless you fall off. Ļ Ļ ambiguous following co-text chunk (watching sex on TV is not harmful to one’s psyche ĺ having sex on the TV box is not harmful as long as one does not fall off)

In case of fire do not use elevators, | use water. Ļ Ļ lead-up switch In case of fire do not use elevators, | use water. Ļ Ļ ambiguous chunk following co-text (if there is fire, one should not use elevators but stairs ĺ if there is fire, one should use not elevators but water to extinguish it) Giving up smoking is easy. | I’ve done it a thousand times. Ļ Ļ lead-up switch Giving up smoking is easy. | I’ve done it a thousand times. Ļ Ļ ambiguous chunk following co-text (the speaker believes that breaking the smoking habit once and for all is not difficult ĺ the speaker believes that the act of getting rid of the smoking habit is not difficult, as he/she has done it many times)

As in the examples above, throughout the work, the result of the GP inference process will be presented (if necessary) according to this pattern: the first salient/default inference made on the basis of the lead-up ĺ the second understanding of the lead-up in the light of the import of the switch

40

Chapter One

The divisions between the lead-up and the switch, or between the cotext, preceding or following the ambiguous chunk, will not be signalled so that the reader can appreciate the workings and humorousness of the examples. Additional explanatory material is sometimes provided in square brackets. As explained at the beginning of the chapter, broader situational or cotextual factors (if present at all in the source material) are dismissed as irrelevant to the discussion of the GP phenomenon. Accordingly, to appreciate the nature and workings of GPs, one must analyse them in isolation just like fully-fledged jokes, which are to be equally understood (but not necessarily found funny) by the target group, i.e. competent language users. GPs are anchored in salient/default meanings and the underlying cognitive and pragmatic processes, which will normally be in force. This is why GPs ought to be perceived in the same way by linguistically competent interpreters, succeeding as humorous phenomena across various situational contexts. Finally, it should be added that GPs can be viewed as the speaker’s creation process, or as the potential hearer’s comprehension process, which coincide with the objective description of how the humorous verbalisation progresses from the lead-up to the switch, within the ideal model constructed here. The whole book will be devoted to the exploration of the GP humorous phenomenon introduced above. By necessity, the work will boast a discursive character, with the discussions being often diverted from the main issue in the direction of humour studies on the whole. Additionally, a number of semantic, pragmatic and cognitive proposals and theories concerning human communication, not necessarily humorous, will be revisited and applied to the analysis of GPs. The next chapter’s aim is to position the GP mechanism within the classical framework of humour analysis, viz. the incongruity approach, in particular, the incongruityresolution model prevailing in humour studies within linguistics.

CHAPTER TWO THEORETICAL APPROACHES TO HUMOUR: INCONGRUITY AND INCONGRUITYRESOLUTION

The aim of this chapter is to present a theoretical background to the linguistic studies on humour. Primarily, the concepts of incongruity and the model of incongruity-resolution will be revisited, the latter being prevalent in humour studies within linguistics. The incongruity-resolution model is also considered germane to the present discussion of GPs, albeit with a number of reservations.

1. Families of humour theories There’s no ingeniousness in the observation that the issue of humour has held scholarly interest since ancient times. At first, it was the early philosophers that discussed humour, but perspectives multiplied with the advent of new realms of study, i.e. psychology and sociology. Finally, linguists started contributing to humour research, allowing insight into its semantics, pragmatics and cognitive operations. Three directions of linguistic analyses can be found, viz. object-related, which focuses on the topic or form of humour; subject-related, which is devoted to the analysis of human reaction; and functional, which describes how humour affects communication (Buttler 1966). Contemporary research is multifarious in terms of both methodologies and topics of studies. Independent theories and postulates abound within each approach and across disciplines. The most widely acknowledged humour theories can be grouped into three main categories (cf. Monro 1951, Wilson 1979, Attardo 1994, Dziemidok 1993), i.e. superiority theories (conflict theories); relief and release theories; and incongruity theories, together with contrast theories. Apart from those, Keith-Spiegel (1972) lists biological, instinct and evolution theories, ambivalence theories and configurational theories. The theoretical approaches fall into three families and can be tabulated as follows:

Chapter Two

42 Cognitive Incongruity Contrast Deviation Configuration

Social Hostility Aggression Superiority Triumph Derision Disparagement

Psychoanalytical Release Sublimation Liberation Economy

The Three Families of Theories (cf. Attardo 1994: 47)

It should be highlighted that the categorisation above is not heuristically flawless, for both the families and their members can overlap.1 Given the abundance of theories and particular proposals within each of the approaches above, it would be impossible to recapitulate, not to mention discuss in detail, all of them. Nor is there a reason for doing so, because many are beyond the scope of the present linguistic research. This short introduction is only meant to testify to the wide spectrum of approaches to humour, an immensely diversified phenomenon. What is most significant here, incongruity theories, together with contrast, deviation and configuration theories, determine the formal object of humour and provide a basis for the majority of contemporary linguistic and psychological analyses of humour. The next subchapter succinctly recapitulates the premises of incongruity theories, as well as configurational theories, contrast theories and norm deviation theories. This philosophical heritage offers foundations for the cognitive two-stage incongruityresolution model (Suls 1972, 1983), which is considered to capture the majority of contemporary linguistic analyses of verbal humour.

2. Philosophical incongruity theories Among others, Grimes (1955), Keith-Spiegel (1972) and Morreall (1987) enumerate a host of philosophers, ranging from ancient thinkers to 20th-century scholars, who utilised variously conceptualised incongruity as the kernel of their theories of humour. One could observe, however, that some of the early philosophical ideas, such as Aristotle’s “deformity not productive of pain”, appear to have been misinterpreted as forerunning the cognitive incongruity theory, since the original intentions underpinning

1

I would like to thank Giovannantonio Forabosco for pointing this out to me.

Theoretical Approaches to Humour

43

philosophical notions will have been significantly different from incongruity (Forabosco 1992). As many (e.g. Raskin 1985, Keith-Spiegel 1972, Shultz 1976, Rothbart 1976, Suls 1983, Wilson 1979, Martin 2007) report, one of the earliest mentions of incongruity goes back to Beattie’s (1776) writings on the origin of laughter. Laughter arises from the view of two or more inconsistent, unsuitable or incongruous parts or circumstances, considered as united in one complex object or assemblage, or as acquiring a sort of mutual relation from the peculiar manner in which the mind takes notice of them. (Beattie 1776: 348)

Beattie (1776) acknowledged the juxtaposition of two disparate elements emergent from their false similarity or illogical reason-result connection. What is significant, Beattie observed that laughter may be blocked by three conditions. The first two occur when there is no element of surprise or if a different strong emotion is produced, such as moral disapproval, distaste, fear and the like. The third condition pertains to savoir vivre rules, which require that laughter be restrained. Apart from the third stipulation, the first two are still recognised in discussions on incongruity. The foundations of the incongruity approach are also credited to Kant and Schopenhauer. Kant (1790/1951) proposed a theory of laughter hinged on unfulfilled expectations. In this view, the humour process, which he called “play of thought” starts with a certain expectation, further reduced to nothingness, which results in a pleasant surprise, giving rise to delight through the change of sensations. In everything that is to excite a lively convulsive laugh there must be something absurd (in which the understanding, therefore, can find no satisfaction). Laughter is an affectation arising from the sudden transformation of the strained expectation into nothing. (Kant 1790/1951: 177)

Schopenhauer (1819) is widely known to have introduced the term “incongruity” in reference to the contradiction which produces humour. The incongruity occurs between the notion of an object and a real object which at first sight seems to correspond to the notion but which actually possesses only some features ascribed to it by the notion while some other of its qualities remain mutually exclusive with the features included in the notion which we want to apply to the object. (Schopenhauer 1819: 76)

44

Chapter Two

Similar to incongruity, albeit less widely acknowledged in linguistics, are contrast, configurational and deviation theories. These three can also be classified as cognitive theories focusing on the object of humour which manifests some form of incongruity. One of the first theoreticians to have used the term “contrast” in reference to humour is Jean Paul. Among other philosophers, perceived as the pioneers of the model, there is Hutcheson (1750), who propounded his theory of idea association. The rudimentary mechanism of humour is a clash of contradictory ideas, such as magnitude and perfection with minuteness, or human wisdom with animal instinct. Another theory of contrast was advanced by Spencer (1860), who wrote about shifts from “great to small things”. The theory rests on the assumption that a clash occurs between the knowledge humans absorb via senses and abstract knowledge which they resort to while assessing real objects. Laughter is the outcome of the perception of a mismatch between an abstract idea and real objects attributed to it, which leads to the accumulation of mental energy, whose surplus manifests itself in laughter. Lipps (1898) proposed yet another contrast theory, according to which humour stems from the substitution of one value in focus of human attention for another one, until it becomes of minor importance. The new value manages to intrigue the mind, because it purports not to be the one expected. Keith-Spiegel (1972) differentiates between incongruity and configurational theories (e.g. Maier 1932, Schiller 1938, Bateson 1953), which manifest psychological orientation and are usually treated as subordinate to incongruity theories. While both are grounded in the cognitive mechanism of humour appreciation and the disparity between elements, they differently conceptualise the point of humour emergence. Incongruity theories highlight the disconnectedness, while configurational theories focus on the surprising congruity between elements. A few ideas underpinning these perspectives can be traced back to Quintilian’s and Hegel’s writings on the intelligibility of a situation that initially seemed to be unintelligible (Keith-Spiegel 1972). Maier (1932) employs the approach of Gestalt-oriented theories, positing that humour occurs when a certain configuration of elements with particular meanings suddenly turns into another. In the process of understanding, therefore, the first train of thought is stopped and supplanted by another one. Specifically, a humorous experience invariably originates from the feeling of surprise caused by objectively experienced elements, which are harmonised, explained and unified, while ridiculousness resides in the fact that harmony and logic apply only to its own elements. On the other hand, Schiller (1938) attributes laughter to amusement and pleasure people

Theoretical Approaches to Humour

45

experience as the benefit of reasoning, such as solving riddles but not more profound problems. At first, the recipient is perplexed by a bizarre situation but, able to reason, resolves the issue, rendering the situation sensible. This also explains why the sense of humour correlates with mental capabilities and intelligence. Although the theory of deviation is classified as one vying with incongruity, it could be compared with the latter or even categorised as its subtype. The theory of deviation, first constructed by Gross (see Dziemidok 1993), pivots on the assumption that humour is produced by deviation from an established norm. Aubouin (1948) also tackled the theory but presented the definition of deviation as relevant particularly to humans bearing characteristics deviant from what was traditionally conceived as the norm, which rendered such humans liable to ridicule but could not cause their suffering. Also, Aubouin (1948) extended the deviation theory to actions and objects different from those expected. The theory was further developed to fit deviation from conventional, and thus normal logical rules, conceptual standards, customs and social or linguistic conventions (Dziemidok 1993). Humour was also deemed as the result of transgression of an accepted linguistic routine. Broadly speaking, deviation from a norm can be perceived as a subtype of incongruity, because it is related to a disparity between a certain norm and a particular phenomenon. The philosophical theories are, allegedly, meta-theses applicable to all humour manifestations, whether verbal or non-verbal. Therefore, they offer very general tenets which are nowadays considered methodologically imperfect and insufficient for analyses. However, these theories will have offered inspiration for incongruity theories. Paulos (1980) rightly observes that his predecessors meant different oppositions as the manifestation of incongruity: expectation versus surprise, the mechanical versus spiritual, superiority versus competence, balance versus exaggeration and propriety versus vulgarity. Indubitably, many more oppositions may be found to describe particular instantiations of humour. Regardless of this, incongruity is used as a terminological unifier, which appears to have superseded such terminology as contrast, ambiguity, discrepancy, dissonance, etc. (Forabosco 1992).

3. Incongruity in contemporary studies The notion of incongruity as a correlate of humour is known to have been initially advocated in philosophical and psychological literature. It was much later that it aroused interest also within linguistic studies on

46

Chapter Two

humour, although being differently conceptualised. Whereas psychologists’ aim is to discuss the subject’s perception of incongruity and their processing of a humorous stimulus compared with the cognitive model of reference (see e.g. McGhee 1979, Forabosco 1992, Martin 2007), linguists focus on the structural features of the stimulus in the form of text. In psychological studies, the perception of incongruity is variable, for it depends on the perception of a given subject (e.g. a child or an adult) or even of one subject at different times.2 By contrast, in linguistic discussions, incongruity is considered stable and inherent to a particular example as perceived by “an ideal recipient”, which is also the view assumed here. The generalisation emergent from the bulk of linguistic literature is that humour arises from incongruity, as already observed by many (e.g. Keith-Spiegel 1972; Ruch 1992; Forabosco 1992, 2008; Staley and Derks 1995; Ritchie 2004; Partington 2006).3 In other words, the underlying mechanism of humour, whether verbal or non-verbal, is captured by the concept of incongruity. Nevertheless, it has earned an array of definitions and characterisations, by no means mutually exclusive. For example, incongruity is considered to be the arrangement of the constituent elements of an event is incompatible with the normal or expected pattern (McGhee 1979: 6-7) [something] unexpected, out of context, inappropriate, unreasonable, illogical, exaggerated and so forth (McGhee 1979: 10) the simultaneous presence of two or more habitually incompatible elements, where an element can be either an object or an event (Shultz 1972: 457) the simultaneous occurrence of normally incompatible elements (Staley and Derks 1995: 97) a mismatch, disharmony or contrast between ideas or elements in the broadest possible sense (Attardo 1994:48)

Here, support is also given to the premise that incongruity is the common denominator for all forms of humor-inducing stimuli. A 2

I am grateful to Giovannantonio Forabosco for this observation. A divergent opinion is held by Latta (1999), who criticises the incongruity approach and dismisses it as entirely implausible. His argumentation is, altogether, unconvincing, if not unfounded (see Oring 1999).

3

Theoretical Approaches to Humour

47

humorous stimulus derives from the (surprising) juxtaposition of two unrelated (i.e. incongruous) elements. Undoubtedly, presenting humour merely as a phenomenon contingent on some form of incongruity is just a generalisation, which, even if apt, does not suffice as an explanation of humour workings. Each of the numerous humour manifestations (e.g. pictures, drawings, canned jokes, lyrics, satirical stories, etc.) has its own subordinate incongruity-based mechanism. Therefore, there simply must exist various unequivocal conceptualisations of incongruity mechanisms if they are propounded in reference to different humour phenomena. Due to this diversification, Ritchie (2004) rejects the notion entirely, emphasising that there is no unification even in analyses of jokes. Indeed, authors use incongruity in reference to two incongruous meanings consequent upon ambiguity within the set-up (e.g. Oring 1992, 2003) or the incongruity between the set-up and the punchline (e.g. Suls 1972, 1983; Shultz 1972, 1974, 1976). It should be noted that the formulation of incongruity is even more problematic when applied to other manifestations of humour. However, this does not mean that the generalisation ought to be rejected. Following Wyer and Collins (1992), Martin rightly espouses a belief that incongruity is a constellation of cognitive elements which appears to characterize all forms of humor such as amusing slips of the tongue or the proverbial person slipping on the banana peel, the laughtereliciting peek-a-boo games and rough-and-tumble play of children, and even the humor of chimpanzees and gorillas. (Martin 2007: 6)

Incongruity should be regarded as inherent to all humour forms, whether intentional or not, whether verbal or non-verbal. For example, a picture showing a big dog and a cat hugging is incongruous, as it violates the common beliefs that the two species are enemies and that animals are incapable of hugging. On the other hand, in humorous neologisms (e.g. “alcoholiday”), incongruity will reside in the juxtaposition of the two composite meanings (“alcohol” and “holiday”). As regards puns (acting as jokes, witticisms, slogans or headlines), incongruity may be argued to operate between the two meanings of the ambiguous chunk and between the two interpretations of the whole verbalisation. Such examples could be multiplied, but it is not the aim here. Suffice it to say that in jokes, incongruity emerges at the junction of the set-up and the punchline (Suls 1972, 1983; Shultz 1972, 1974, 1976). Within this model, a particular conceptualisation of incongruity is proposed for GP one-liners and witticisms, in which the switch is incongruous with the default/salient interpretation of the lead-up.

48

Chapter Two

It is also worth noting that the phenomenon of incongruity is not exclusive to humour but is relevant also to non-humorous phenomena (e.g. detective stories or annoying incompatibilities). The arousal consequent upon the perception of incongruity need not induce a humorous response but may produce emotions other than enjoyment such as moral disapproval, fear, shock, puzzlement or anxiety (cf. e.g. Berlyne 1960, 1972; Rothbart 1976; Morreall 1989). Incongruity is a necessary but not sufficient condition for humour. A question must be raised as to what renders incongruity humorous. In his answer to this query, Suls (1972) highlights that humorous incongruity entails unexpectedness, illogicality and ultimate resolution. Suls (1972) adds that incongruities are funny if they are resolved and simply if one knows that a particular stimulus is a joke. The recognition of this meta-level, i.e. the humorous intention underlying a text, is exactly the pattern on which almost all studies on humour are based. Any analysis of humour, uses material already classified as heuristically adequate, chosen rationally with a view to proving or disproving various theses. Other authors endorse different opinions on a sine qua non for humorous incongruity, e.g. a facilitating (pleasant, safe) context (Rothbart 1976), or a playful frame of mind (Apter 1982). The issue of what renders incongruity not only humorous, i.e. potentially amusing, but also actually funny, i.e. genuinely amusing to the recipient, is obviously also open to discussion and, primarily, subject to psychological and psycholinguistic research. For example, Staley and Derks (1995) discuss several social and idiosyncratic emotional parameters determining whether an incongruous stimulus will elicit a humorous response.

4. Bisociation Besides incongruity, a separate, albeit in several respects similar, humour theory embracing philosophical and psychological perspectives was advanced by Koestler (1964). The underlying claim is that humorous forms, together with poetry and science, entail creativity and the discovery of hidden similarities between juxtaposed elements. The existence of humour is a testament to human creativity and the ability to operate on two different planes of thought simultaneously. The creative act by connecting previously unconnected dimensions of experience enables him (man) to attain a higher level of mental evolution. It is an act of liberation, the defeat of habit by originality. (Koestler 1964: 96)

Theoretical Approaches to Humour

49

A considerable part of Koestler’s (1964) proposal falls within the framework of relief theories, accounting for the provenance of laughter, which is not discussed in detail here, given the linguistic approach assumed. In essence, laugher acts as a safety valve for the tension accumulated by the simultaneous perception of one idea in two incompatible frames of reference. Such a stipulation is reminiscent of the Freudian explanation of laughter as momentary ascendancy of the subconscious over the conscious, and as a consequence of the accumulation of energy, which finds its release after the mind falters. The sudden bisociation of an idea or event with two habitually incompatible matrices will produce a comic effect, provided that the narrative, the semantic pipeline, carries the right kind of emotional tension. When the pipe is punctured, and our expectations are fooled, the now redundant tension gushes out in laughter, or is spilled in the gentler form of the sou-rire. (Koestler 1964: 51)

From the present linguistic vantagepoint, more vital than the notion of release is Koestler’s coinage bisociation introduced in reference to a nonresolvable dichotomy, a juxtaposition of two incongruous frames of reference, coupled with the discovery of various similarities or analogies implicit in them. All creative phenomena, most importantly humour, entail two habitually incompatible frames which appear to be somehow coherent (congruous). In other words, bisociation can be understood as mediation between two realms of meaning, between which there exists a conceptual conflict which is ultimately resolved (Johnson 1976). As Koestler himself puts it, bisociation is the perceiving of a situation or idea L, in two self-consistent but habitually incompatible frames of reference, M1 and M2. The event L, in which the two intersect, is made to vibrate on two different wavelengths, as it were. (Koestler 1964: 35)

Diagrammatic representation of bisociation (Koestler 1964: 35)

50

Chapter Two

As a result of bisociation, human cognitive attention alternates between two incompatible matrices, i.e. frames. A frame of reference constitutes the matrix of thought defined as “any ability, habit or skill, any pattern of fixed behaviour governed by a code or fixed rules” (Koestler 1964: 38). Bisociation may indeed be regarded as a general theory of humour, while a more detailed analysis must be conducted so that mechanisms underlying specific humour phenomena can be explained. Despite the elapsing decades, the idea of incompatible matrices (differently termed) appears to be a relatively plausible description of admittedly all humour phenomena. The notion of bisociated matrices is mirrored by the postulate of conflicting but compatible schemas/schemata (Bartlett 1932), frames (Bateson 1972) or scripts (Minsky 1975, 1979; Rumelhart and Ortony 1977; Schank and Abelson 1977). As Norrick (1986: 227) perceptively notes, “many writers both before and since Koestler (1964) have advanced bisociation theories of humour in various guises and in various forms.” Indeed, many authors revisit the theory before embarking on their analyses on humour manifestations (e.g. Suls 1972, Norrick 1986, Partington 2006, Paulos 1980, Coulson 2001). Bisociation appears to lie at the heart of Apter’s (1982) cognitive synergy, i.e. two compatible incompatible/contradictory interpretations of one object simultaneously activated in the mind. The notion of bisociation also underlies Oring’s (1992, 2003) concept of appropriate incongruity, viz. “the perception of an appropriate interrelationship of elements from domains that are generally regarded as incongruous” (Oring 1992: 2). Oring, however, takes as his departure point not Koestler’s bisociation (which he regards as an incongruity model) but Monro’s “element of appropriateness in the inappropriate” (1951: 255). In this view, incongruity will hold unresolved, but common points between the incongruous elements will be found. Oring (1992, 2003) regards appropriate incongruity as a state, not accounting for the stages in humour processing. Nevertheless, in the light of the examples (mainly riddles) the author provides, it is transparent that the process takes place on-line, with incongruity preceding the phase of (resolved) appropriate incongruity. As Oring (1992: 2) himself acknowledges, “a riddle question proposes an incongruity that the riddle answer must in some way appropriately resolve.” Incongruity, which the interpreter must render appropriate, may also emerge at the junction of the question and the answer. In essence, it is here argued that the notions of bisociation or appropriate incongruity refer to the final stage of a humorous texts’ processing when the incongruity has been perceived and somehow motivated, i.e. resolved (see section 5).

Theoretical Approaches to Humour

51

Koestler’s model as such can be infallibly applied to many forms of humour where the two incompatible elements are (nearly) simultaneously juxtaposed, as in similes, funny neologisms or pictures. However, not all humorous stimuli, GPs being a case in point, revolve around almost instantaneous presentation of two meanings. In most jokes, including those based on the GP mechanism, the second frame is imposed after the first one has already been appreciated. In GPs, a schema established by the early part of the text is overturned by subsequent information. All the same, even is such cases, Koestler’s oscillation between two frames will be applicable to the final stage of humor appreciation, i.e. after the incongruity between the lead-up and the switch and the two incompatible meanings of the lead-up have been appreciated. The meanings within both of the incongruous/incompatible juxtapositions can be bisociated. She has her looks from her father. He’s a plastic surgeon.

Comprehending this GP, the interpreter first draws an inference that the female resembles her father, which is incongruous with the surprising and seemingly irrelevant information that he’s a plastic surgeon. Eventually, the bisociation of these two inferences takes place as well as the bisociation of the two competitive meanings of the ambiguous lead-up, i.e. the woman inherited the looks/the woman had her looks changed by her father. In linguistic literature, the concept of bisociation, albeit widely known, seems to have been almost entirely ousted by that of “incongruity”, or specifically “incongruity-resolution” (except for Norrick 1986 and Coulson 2001, see chapter 3), which describes the two stages of the humour interpretation process. Propounding the incongruity-resolution model, Suls (1972) appreciates Koestler’s (1964) bisociation but interprets the bisociative humour appreciation process (contrary to the original formulation) as one sudden shift, “an abrupt transfer of the train of thought from one matrix to another governed by a different logic or rule” (Suls 1972: 95), not the juxtaposition of two matrices and continuous oscillation between them. Suls (1972) observes that there must be a problem-solving process underlying the integration of the discrepant parts, which Koestler (1964) leaves unaccounted for.

5. Incongruity model or incongruity-resolution model In linguistic studies, it is commonly assumed that incongruity alone does not suffice as an explanation for humorous phenomena but that it

52

Chapter Two

must be resolved. Broadly speaking, the incongruity-resolution (I-R) model, credited primarily to Suls (1972, 1983), holds that incongruity is first observed and later resolved, i.e. made congruous. The I-R model is the widely accepted departure point for many contemporary linguistic analyses of humour. A host of authors discussing humour mechanisms at least allude to the I-R approach, even if their discussions divert from the original formulation. Although Suls (1983) modestly reports that two other authors independently advocated the incongruity-resolution model (Jones 1970, Shultz 19724), it is Suls (1972) that is widely credited for the two-stage framework of the reception of jokes and captioned cartoons. This is because it focuses on the cognitive mechanism of comprehension and is only accompanied by discussions on additional psychological factors, such detours from appreciating humour, which are of secondary importance to linguists. By contrast, Shultz (e.g. 1972, 1974, 1976) proposes to combine the incongruity-resolution framework with Berlyne’s (1969) arousal jag, thereby contending that the perception of incongruity correlates positively with the interpreter’s level of arousal, while the resolution restores it back to the baseline. Furthermore, Shultz’s primary focus of interest is an analysis of developments in children’s humour perception. Finally, as will be discussed, Shultz’s model of joke comprehension appears to concern primarily jokes and riddles couched in ambiguity (see section 1, chapter 3), contrary to Suls’s, which obtains for all jokes. To recapitulate, from a linguistic perspective, Suls’s (1972) model is more insightful, as it primarily offers a skeleton for a theory of joke comprehension, rather than appreciation, which is the focus of linguistic studies, while psychological factors are marginalised. The proposal concentrates on the process of comprehension but does not aspire to account for the emotional or social aspects which may help explain why humour is enjoyable (Martin 2007), which is a merit from the viewpoint taken here. Suls’s (1972) model pertains to both verbal and pictorial humour but particularly to jokes of all types. It should be appreciated that no claims are made as to the phenomenon of humour on the whole, and thus non-verbal humour (other than captioned cartoons) or conversational humour phenomena, such as teasing or neologisms. Despite the prevalence of the view that incongruity as such is insufficient for humour and that it must be resolved so that humour can come into being (Suls 1972), proponents of incongruity alone can also be 4

Shultz’s research on the incongruity-resolution model dates back to 1970, when he completed his unpublished PhD thesis.

Theoretical Approaches to Humour

53

found. In the light of findings yielded by the famous experiment during which subjects were asked to lift a series of boxes, the last of which weighed considerably more or considerably less than the previous ones, Nerhardt (1970) postulates that it is only incongruity, understood as divergence from expectation, that is essential to provoke a humorous response, and hence that resolution is not obligatory. Similarly, studies on children’s appreciation of humour corroborate that incongruity itself suffices to experience a humorous response, given that young subjects are incapable of conducting resolution processes (e.g. Rothbart 1976, Pien and Rothbart 1976, see Martin 2007). Suls (1983: 47), however, maintains that most humor, particularly verbal, has an incongruity-resolution structure and criticises the unresolved incongruity approach, following from the premise that incongruous stimuli may produce laughter but not necessarily humour, which is the case of Nerhardt’s findings. Admittedly, Suls (1983) presupposes that there is no inherent correlation between humour and laugher. Indeed, laughter may be the outcome of nervousness, while humour need not unfailingly cause laughter but, for example, only a smile (see Rothbart 1976, Raskin 1985, Attardo 1994 inter alia). Nevertheless, further experiments exploiting Nerhardt’s weight judgment paradigm (Deckers 1993, Deckers et al. 1997) indicate that subjects’ smiling and laughter must be associated with amusement rather than embarrassment and thereby testify to the incongruity model without resolution. Later, Suls (1983) also relaxes the resolution stipulation when stating, “most humor, particularly of verbal form, has an incongruity-resolution structure” (Suls 1983: 47), while “some humorous experiences are the result of mere incongruity” (Suls 1983: 48). Humorous stimuli which do not exhibit the incongruity-resolution structure are, in his view, for instance, quips or shaggy-dog stories, which he dubs meta-humour, because “they provide enjoyment by playing off the fact that they pretend to be humor but do not possess all of the requisite features” (Suls 1983: 53). Suls’s statement seems to suggest that jokes are the only legitimate form of verbal humour and thereby minimises the importance of other forms which can be genuinely humorous as well. What Suls (1983) may mean is that there are forms which do not seem in any way logical. All the same, Suls (1983) acknowledges that resolution in such cases comes from the interpreter’s understanding that the speaker has humour-oriented intentions. It may be argued that incongruity is always resolved, even in the case of non-verbal and childish humour (e.g. a man wearing a clown’s nose or someone making strange faces). Even if no clear logical rule resolving the incongruity, as proposed by the two-stage I-R model, can be found, there

54

Chapter Two

must be something that renders the incongruous stimulus cognitively feasible. As Forabosco (2008) rightly posits, the element of sense is always present in the background, as the perceiver always exerts mental control over the stimulus. Therefore, it may be extrapolated that even seemingly irresolvable incongruities are always somehow resolved (e.g. via the realisation that a man does not normally wear a red shiny ball on the nose or that the face is unnaturally changed). The above hypotheses notwithstanding, it can be safely affirmed that the I-R model, as originally propounded by Suls (1972), holds for virtually all jokes as well as one-liners and witticisms. The pattern of joke interpretation will be presented in detail in the first section of the next chapter. In a nutshell, according to Suls (1972), the hearer/reader listens to/reads the first part of the joke and organises it in a narrative schema, which helps him/her predict the following part of the joke, which culminates in a surprising punchline. The latter part of the humorous narration is incongruous with the preceding part. As a result, the hearer starts searching for a cognitive rule, which paves the way for the combination of the punchline with the preceding text. The ultimate arrival at a satisfactory cognitive rule (i.e. the solution) determines the experience of amusement, without which the text is nothing but astounding. In essence, the classic processing account of humour comprehension involves the recognition of incongruity and its subsequent resolution via the adoption of another set of assumptions (Suls 1972). It is also crucial that the resolution is prompted by a trigger in the text but takes place in the mind of the recipient, who cognitively processes the verbal stimulus presented. Referring to the I-R model, authors do not precisely quote Suls’s formulation of the joke interpretation process but succinctly summarise the approach. This generalisation is that a stimulus or a situation is humorous if the perceiver observes some incongruity, which may additionally violate his/her expectations and/or surprise him/her. Thereupon, the interpreter is able to resolve the incongruity on condition that the stimulus can be deemed as sensible if viewed in an appropriate way.

6. The I-R model: Queries and problems Even if there is a general agreement among linguists about the existence of the two-stage incongruity-resolution process of perceiving and understanding humour, there is little agreement about the exact conceptualisation of incongruity or its resolution (cf. Forabosco 1992, Ritchie 2004, Partington 2006). Contemporary discussions of the I-R

Theoretical Approaches to Humour

55

model are incompatible and even contradictory. One bone of contention is whether two incongruous meanings are obtained and appreciated simultaneously or consecutively. In addition, researchers do not concur regarding the time course of meaning suppression. Nor do they agree on whether the resolution is complete or partial. The first debatable issue is when each of the alternative meanings becomes available. As regards the issue of meaning activation, crucial is Ritchie’s (2004) dichotomy between static and dynamic incongruities. This dichotomy is pertinent to a stimulus’s form of presentation. Static incongruity represents a status quo in the configuration or instantaneous presentation of two elements, e.g. in cartoons, being subject to easy bisociation (Koestler 1964). Even if the twofold interpretation emerges instantly in the perception of the interpreter, one meaning will always be observed earlier, even if only by a fraction of a second, due to its objective salience, defaultness or givenness consequent upon the form of presentation. In essence, the human mind forms individual mental structures and is partial to this interpretation which it can access more easily (see chapter 6). On the other hand, dynamic incongruity entails temporal sequencing of ideas both in the presentation and in the interpretation processes, which appears to underlie most texts. This type of incongruity is inherent to the GP mechanism discussed here. A related query is whether or not the two interpretations are at some point simultaneously available and what eventually happens to them in the resolution phase. On the one hand, advocates of the concurrent activation view (cf. Vaid et al. 2003)5 aver that the two competitive meanings must co-exist in the interpreter’s mind in the incongruity phase so that the interpreter can appreciate the dissimilarity between them, and in the resolution phase so that they can be compared and a cognitive rule governing the resolution of incongruity can be found. For example, Paulos (1980) concedes that a prerequisite for humour is that two aspects of an entity must be perceived together and compared. Nerlich and Clarke (2001) also support the idea that meanings ought to be kept in mind simultaneously, relying on the research substantiating that understanding polysemous words entails the activation of inappropriate meanings. One can find in such postulates the reverberation of Koestler’s bisociation (1964). By contrast, according to proponents of the selective activation view (Vaid et al. 2003), the second meaning has the effect of erasing the 5

The authors attribute this approach to Attardo and Raskin (1991) and Attardo (1997), which is far-fetched, since Attardo and Raskin’s model is not focused on the mechanisms of meaning activation/suppression and may also be conceived of as independent from the I-R model (see the next chapter).

56

Chapter Two

first one (e.g. Giora 1991, 2003). Such proposals are proven findings of neuropsychological laboratory analyses. For instance, Coulson and Kutas (2001) substantiate that the arrival at the second interpretation is facilitated if the first incompatible meaning is suppressed. Yet different findings are reported by Vaid et el. (2003), partly testifying to both the approaches. In the initial part of the joke, only one, i.e. the initially favoured, meaning is activated, two meanings are activated in the incongruity detection phase, whereas in the resolution phase only the second meaning is primed (Vaid et al. 2003). It should be observed that the divergences in the theses championed and the research data presented might, at least partly, result from the types of humour stimuli, in particular, the joke types that are discussed. Theoretical discussions on puns testify that humorous lines operate on diversified processes of meaning activation and retention/suppression (Yus 2003, Van Mulken et al. 2005, Dynel forth.). Therefore, no definite answer can be provided as to whether incongruous meanings are appreciated simultaneously, and whether this occurs in the incongruity phase or the resolution phase. Although the theoretical approaches may seem contradictory, they are not mutually exclusive, if one appreciates that they describe different humour phenomena or perceive incongruity as either the conflict of the set-up with the punchline or the conflict between the two senses of an ambiguous chunk occurring at various stages of the text. For example, Giora (1991, 2003) focuses on one-liners which are mostly based on the annulment of the earlier interpretation of an ambiguous chunk, coinciding with GPs. On the other hand, in the ambiguous pictures or double-retention puns (e.g. “Contraceptives should be used at every conceivable time.”), where both meanings are relevant, incongruous meanings are activated almost instantly and neither is to be rejected as implausible, which is why they are sustained as pertinent. What is most crucial here, the stage of incongruity in GPs enforced by the switch is instantly followed by resolution, because no (or very short) text follows the incongruous element. The stage of resolution entails rejecting the first interpretation of the lead-up and finding another hidden meaning within it, congruous with the final portion of the text. It must be highlighted, however, that after this interpretation process has been completed, the interpreter may keep revisiting the two meanings of the ambiguous chunk and their respective incongruity or congruity with the switch (cf. bisociation). To summarise, it appears that complete incongruity resolution holds only for a moment, while the phase of humour acknowledgement following comprehension as such will entail the oscillation between two

Theoretical Approaches to Humour

57

incongruous elements. Having gone through the resolution phase, the interpreter may revisit the whole joke development process and juxtapose the two meanings, appreciating the humorous effect. It could then be extrapolated that the final stage of humour appreciation resides in bisociation between the two interpretations (Koestler 1964) or incongruity’s appropriateness (Oring 1993, 2003). This conforms also to Forabosco’s (1992) attention-shift situation, which relies on the subject’s repeated perceptual passing from congruence to incongruity and vice versa. Also, Forabosco (2008) dubs this stage of humour processing second level processing, a meta-level on which the interpreter appreciates the message and the value of a humorous stimulus. In the light of the above, the view supported here is that incongruity must not be removed entirely at the resolution stage (Suls 1983; Ruch and Hehl 1998; Forabosco 1992, 2008; Attardo and Raskin 1991). A complete removal of incongruity would disallow the appreciation of two competitive meanings, and hence the interpretation of the whole humorous text. Indeed, Suls (1983) admits that resolution does not completely reconcile incongruous elements. According to Forabosco (1992), the incongruity which lingers after the resolution phase is the essence of humour experience, not a side effect. If incongruity were resolved and a stimulus were perceived as perfectly sensible and congruous, the processing would be called to a halt and no humour appreciation would ensue (Forabosco 2008). As a result, a stimulus would cease to be humorous. As Forabosco (1992) rightly contends, the resolution of incongruity does not mean eliminating it but arriving at an incongruity that makes sense, a congruent incongruity. Oxymoronic as this may seem, the postulate appears to be most plausible. Incongruity resolution should not be regarded as “incongruity dissolution” (Attardo and Raskin 1991). In addition, it should be emphasised that the resolution stage is variously instigated in humour stimuli. In the majority of jokes, particularly of the GP type, the interpreter is forced to seek a resolution, since the text suddenly becomes incongruous and does not make sense. By contrast, there are cases of mostly static incongruity, e.g. pictures (a pig dressed up as a human baby), or puns (A sign on the lawn at a drug re-hab centre said “Keep off the Grass!”), in which no such prompt is provided. The resolution which does take place coincides with the interpreter’s understanding of two incongruous meanings, together with points of similarity and dissimilarity between them. The view that resolution may be equated with the appreciation of congruous and incongruous aspects of the juxtaposed elements is compatible with McGhee’s (1979) observation that for two- or three-year-olds resolution may centre around the awareness of

58

Chapter Two

what an appropriate element (which in the experiment was the right name) is. On the other hand, Giora (1991) does concede that the process of ambiguity perception is dual and repetitive but makes a reservation that texts and visual vignettes which maintain incongruity but do not explicitly resolve it cannot be labelled as funny but merely witty. Admittedly, in Giora’s view, wit carries less laughable potential than humour. Nonetheless, it must be remembered that there is no clearly distinguished caesura separating the denotations of the terms “humour” and “wit”, the latter being usually thought of as a clever, sometimes even biting, form of humour (cf. Gruner 1978, Martin 2007). Besides, similarly to other linguistic analyses, the present one is focused on incongruity appreciation and resolution as the general process of comprehension conducted by the ideal recipient, rather than on the recipient’s enjoyment of humour, which is an idiosyncratic matter. In essence, both incongruity and its resolution are crucial to humour appreciation. Morreall (1989) holds a view that the prerogative for enjoying incongruity is the human pleasure of breaking norms and rules, while the pleasure of solving problems and comprehension coincide with the stage of incongruity resolution. Forabosco (1992) agrees to this, reconciling the proposals of resolved incongruities and incongruities which remain, asserting that both are complementary sources of humour. The former correspond to the pleasure of reasoning, while the latter to breaching schemata or rules. Moreover, Forabosco extrapolates that the degree of fulfilment of these two conditions increases the humorous force. Yet another problem concerns incongruities not only lingering intact after the resolution phase but also emerging at this stage. Rothbart (1976) and Rothbart and Pien (1977)6 differentiate between possible and impossible incongruities, and complete and incomplete resolutions thereof. Impossible incongruities (incongruent with one’s knowledge of the world) cannot be fully resolved, and thus rendered meaningful, for the situation still remains impossible. Rothbart and Pien (1977) and McGhee et al. (1990) maintain that the punchline may actually bring only partial or no resolution, or may even introduce new incongruities or absurdities. To exemplify these phenomena, Rothbart (1976) and Rothbart and Pien (1977) analyse the riddle (“Why did the elephant sit on the marshmallow? Because he did not want to fall in the hot chocolate.”) which retains some 6

The authors’ emergent postulate is that the degree of humour perceived is commensurate with the number of incongruity resolutions, the number of unresolved incongruities, the degree of each incongruity, the difficulty in resolving the incongruity and the degree of its resolution. The intensification of any of the first three produces a rise in the humorous force.

Theoretical Approaches to Humour

59

incongruity after the first incongruity is ostensibly resolved, “challenging any knowledge of elephants and chocolate we may possess” (Rothbart 1976: 41). It emerges, therefore, that incongruities may be a few and that they may occur on various planes, which is why the resolution of one of them need not entail the resolution of another. The I-R model, however, is concentrated on the main incongruity, not on the subordinate ones. Rightly, Attardo et al. (2002) differentiate between focal and background incongruities, the former being in the focus of a joke. Background incongruities may pertain to the underlying absurdity of the joke, which can never be really resolved. As Attardo and Raskin (1991: 307) observe, “a joke must provide a logical or pseudological justification of the absurdity or irreality it postulates.” The resolution may pivot on absurdity and involve a willing suspension of disbelief (Attardo 1994), which corresponds to Aubouin’s7 (1948) justification/acceptance or Ziv’s (1984) local logic, which allow for the masking of absurdity. As conceptualised by Ziv (1984), local logic brings explanation to the incongruity within the reality of the joke, which the hearer is willing to appreciate (see Attardo and Raskin 1991; Attardo 1994, Attardo 1997a). A variety of other problems may also emerge in the course of the analysis of humour examples. Those do not undermine the truthfulness of the I-R model but highlight additional factors, which only testify to the complexity and multifariousness of humour phenomena. For example, some hearers may even search for further explanation, thinking that they have miscomprehended a joke, given that its resolution does not make sense or is very shallow. The resolution may sometimes seem to be farfetched or demand little elaboration although one may seek it, investing a lot of cognitive effort in pursuit of one. Three men happen to catch a goldfish, which promises to fulfil each fisherman’s three wishes. The first man wants to have a mansion, a posh car and a beautiful wife. The second one requests that he should be rendered clever and that he should become a famous businessman and a mayor of his town. The third one finds it difficult to make up his mind, but he says, “I’d like to have my left arm always hanging down, my head shaking and my right hand hitting against my forehead.” The goldfish is extremely surprised to hear this and asks the third fisherman if this is what he really wants. Since the man confirms it, she makes all the wishes come true. They all meet next year. The first man is happy with his wife and 7

However, Aubouin fails to appreciate that incongruity and the errors of judgement and language are most legitimate in the faulty logic of humor (Attardo 1994).

60

Chapter Two possessions. The second one is satisfied with his life, being a distinguished persona. The third man, with his left arm dangling, head shaking and right hand hitting against his forehead, says, “You know what, I may have made a mistake.”8

Analysing the joke above, the interpreter will, in all likelihood, be flabbergasted by the absurdity of the third fisherman’s choice, expecting that the incongruous decision will be somehow resolved by the punchline, which does not take place. Instead, the character admits to have chosen wrongly. The joke hinges on resolution of the incongruity rendered via the character’s appreciation of his absurd choice. This example testifies to Ruch’s (1992) perception of nonsense humour, which does manifest resolution of incongruity, with the cognitive rule escaping a clear description. Moreover, there are cases, even if rare, when the first available interpretation does not produce incongruity and the punchline is perfectly consistent with the body of the text, while the interpreter, aware of the fact that a joke is intended, will search for the incongruous interpretation, as in the case of “How was your stay at a nudist camp?” “The first three days were the hardest.” (Forabosco 1992: 62, quoted from McGhee 1979: 6). The punchline in this example is ambiguous, with the incongruous meaning emerging later than the congruous one (Forabosco 1992). In addition, Ritchie (2006) rightly observes that the perspective(s) from which the reinterpretations occur vary, since jokes may operate on misunderstanding experienced by the audience or by a character within a story. Obviously, for one-liners and witticisms, only the former is the case. To conclude, the well-entrenched incongruity-resolution model is very broad, and thus applicable to a vast area of humour, which can be perceived both as a merit and as a flaw. As Ritchie (2004) asserts, one instance of humour may even be analysed in a number of ways. In some cases, the resolution stage may consist in the appreciation of the similarities and disparities between incongruous elements, as in non-verbal humour, such as a squirrel burping her ABCs (a scene from “Over the Hedge”), or witty neologisms and phrasemes, e.g. “a monsterpiece”, “Helen of Troy wannabe” or “like a parasite knocking at your door looking for a host”. Most importantly, the incongruity-resolution model, is adequate in the analysis verbal humour in the form of canned jokes and 8

I owe this joke to my brother, Krzysztof, who rendered it not only verbally but also by means of non-verbal expression, impersonating the third fisherman in his preposterous posture. Admittedly, the body language rendition accounted for a major part of the humorous effect.

Theoretical Approaches to Humour

61

one-liners or witticisms. In essence, the final part of the text is somehow incongruous with the preceding part, necessitating ultimate resolution. A distinct realisation of the I-R model proposed in this work describes GP humor, the distinctive feature being covert ambiguity of the lead-up. Notwithstanding numerous differences, the common denominator for various forms of humour is, indubitably, resolved, albeit not dissolved, incongruity. This phenomenon is sometimes combined with novelty causing surprise and the violation of expectations (e.g. McGhee 1971, Suls 1972, Nerhardt 1976, Staley and Derks 1995). The two accompanying factors will now be revisited, also as regards their relevance to GP humour.

7. Novelty and surprise The contention that surprise is a prerequisite for the appreciation of humour is rooted in philosophical tradition (see Keith-Spiegel 1972). Hobbes (1651), a supporter of the superiority approach, is known to have heralded the analysis of humour in terms of subjective experiences, thanks to which he is regarded as a forerunner of psychological and cognitive theories of humour (see e.g. Keith-Spiegel 1972, Morreall 1987). The notion of superiority aside, his theory emphasises that observing the comical rests on suddenness and unexpectedness, which give rise to unexpected satisfaction and delight manifesting themselves in laughter. Kant (1790/1951) also stresses the significance of surprise for the occurrence of humour appreciation. Nowadays, it is widely acknowledged in philosophical, psychological and linguistic literature of the subject that the experience of humour (not only in the form of jokes) is correlated with the experience of surprise (Raskin 1985), novelty, and hence the fact that incongruity is introduced abruptly (Katz 1993). The postulate that humorous incongruity emerges unexpectedly, involving novelty and the element of surprise, lays the foundation for not only jokes’ formal mechanisms (the focus of attention here) but also psychological premises for humour’s existence. From the latter vantagepoint, humour offers delight and escape from boredom and monotony thanks to its freshness, novelty and playfulness (Monro 1951). In more recent literature, Giora (2003) argues that surprise and novelty seem to be of crucial significance for humans’ mental functioning. Her proposal of the optimal innovation hypothesis stems from the acknowledgement that people derive pleasure from finding the familiar in the novel (Freud 1905/1960, Berlyne 1971). The author subverts the folk theory that the juxtaposition of salience (cf. chapter 6) and aesthetic

62

Chapter Two

novelty results in contradiction, adding that the two are complementary and that innovation relies on allowing for unprecedented, unthinkable insights into salient concepts. Surprise is the unexpected experience of novelty but ought to be attenuated by finding familiar in the novel. Giora (2003) also hypothesises why innovations are in demand. Accordingly, the human mind craves for novelty, bored with linguistic routines of lexicalised and conventionalised expressions. It could thus be extrapolated that novelty of expression will manifest itself, for example, in the use of verbal humour, which functions as a diversion from the main topic of a conversation, or from conventionalised patterns of expression. Forabosco (1992) advances a subjective model of joke interpretation, according to which incongruity is the consequence of a departure from the attribution model arising in the process of joke reception. During the act of joke-telling, a subjective model of a specific concept or event forms in the mind of the recipient. Forabosco (1992, 2008) proposes that a stimulus is incongruous when it differs/diverts from the cognitive model of reference. The underlying premise is that an individual garners knowledge in his/her cognitive history, which amounts to models (i.e. unbounded schemata) formed thanks to prior experience and learning. Any new stimulus is evaluated against a relevant cognitive model. Should the former differ from the latter, (surprising) incongruity arises. Forabosco (1992) also adds that the perception of incongruity is largely idiosyncratic and even variable for one individual, depending on particular circumstances. To be perceived as incongruous, a stimulus needs to pass the threshold of conformity with a given cognitive model. This is indeed a plausible general conceptualisation of incongruity from the perspective of cognitive psychological studies, largely coinciding with a cognitive surprise. The incongruity-resolution model rests on the thesis that a joke must entail surprise coming at the stage of the punchline. Suls (1972) even contends that surprise is correlated with incongruity, the need for its resolution and, ultimately, humour appreciation. It is suggested that degree of incongruity is directly related to the amount of surprise experienced, and the amount of surprise that the punch line creates should produce a corresponding need to solve the problem. (...) The prediction here is that the more surprising the punch line, the more one should want to overcome the surprise. When the problem is solved, the recipient should experience greater appreciation. (Suls 1972: 91)

Although the postulate of the significance of a surprising punchline is most plausible, a question remains whether, and if so, how the respective degrees can be measured. In addition, it is certainly not only the sole factor

Theoretical Approaches to Humour

63

of surprise in jokes that accounts for their humorous force and funniness. Indeed, Suls (1972) contends that humour appreciation relies on four factors: the degree of incongruity of the punchline, the degree of complexity of the problem to be solved, the time needed for the resolution, and salience. Additionally, crucial are also, for instance, the topic of a joke, and the hearer’s idiosyncratic preferences. It should be acknowledged that the occurrence of a punchline as such is hardly surprising in jokes, for recipients familiar with this humour genre expect it to emerge at the end of the text. Even if a jocular intention on the part of the speaker is presupposed, what is still surprising is the quality of the punchline, i.e. its import which can hardly be envisaged. On the other hand, in the case of GP witticisms and one-liners interwoven unexpectedly into non-humorous discourse, the recipient will be surprised not only by the import but also by the very occurrence of the humour-inducing switch. The latter introduces incongruity and causes the interpreter to search for its resolution, which in GPs invariably entails reanalysing the preceding text. Irrespective of the importance of novelty in the perception of humour, it cannot be denied that it is also familiar stimuli that can be considered humorous. Accordingly, there do exist incongruities which make recipients laugh even though the latter encounter them more than once (Rothbart and Pien 1977). Admittedly, almost everybody has a favourite joke, a sketch performed by a comedian or a comedy to which they return whenever they want to raise their spirits or to amuse themselves. Suls (1972) also attempts to explain why some jokes may still be funny regardless of the recipient’s repeated exposure to them. One of the explanations he provides is simply the recipient’s having forgotten a joke. This, however, accounts for very few cases and does not explain the humorous response to a stimulus which the recipient deems as familiar. An alternative and much more plausible explanation is that some jokes (and also other forms of humour) have several levels of interpretation. This is certainly the case of many joke or sketch renditions, which are entertaining to the audience thanks to, e.g. the speaker’s use of body language, facial expressions and non-verbal aspects of speech (such as voice modulation). The third type of explanation is the familiarity effect coupled with the mere exposure effect (Zajonc 1968), and hence the positive associations the interpreter has with a certain humour stimulus, for the sense of safety it guarantees. Forabosco (2008) also lists the familiarity effect as a potential explanation of humorous stimuli manifesting a decrease in novelty and surprise. Yet another hypothesis he forms is suspension of knowledge, the effect of human ability to control what otherwise would be a spontaneous reaction.

64

Chapter Two

8. Defeated expectations In psychological studies on humour, the notions of incongruity and surprise are very often associated with that of defeated expectations. The arrangement of the constituent elements of an event is perceived as incompatible with the normal or expected pattern.(…) When the arrangement of the constituent elements of an event is incompatible with the normal or expected pattern, the event is perceived as incongruous. (McGhee 1979: 6-7) (…) the greater the divergence of a stimulus from expectation in one or many dimensions,9 the funnier the stimulus. (Nerhardt 1976: 59) What amuses us is some object of perception or thought that clashes with what we would have expected in a particular set of circumstances. (Morreall 1987: 6) In every form of humour, the audience is led in a certain direction and encouraged to form certain expectations, and these are then turned upside down. (Ziv 1984: 89)

Forabosco (1992) posits that individuals develop implicit expectations as to the new incoming stimulus, which entails assigning some subjective probability value to it. (…) there is for a subject, a given model, a threshold beyond which deviation of the stimulus from the model generates that particular experience of cognitive rupture, a sensation of the alien (or strangeness), of mental tension, etc., which corresponds to the perception of incongruity. (Forabosco 1992: 55)

Of most crucial importance here is the tenet that jokes rely on the violation of expectations, and even the equation between such violation and incongruity. Incongruity of the joke’s ending refers to how much the punch line violates the recipient’s expectations. (Suls 1972: 92)

9

Nerhardt (1976) puts forward a thirteen-point theory of humour, according to which a person will classify a stimulus in the context of other stimuli depending on a number of relevant dimensions.

Theoretical Approaches to Humour

65

Incongruity is usually defined as a conflict between what is expected and what actually occurs in a joke. (Shultz 1976: 12) The concept of incongruity refers to the notion that something unexpected happens in a joke which serves to arouse, surprise or mystify the listener. (Shultz and Robillard 1980: 60) Deliberately constructed to violate the listener’s expectations, jokes highlight the various sorts of expectancies we make as a matter of course. (Coulson 2001: 49)

Suls’s (1972) incongruity-resolution model, which appears to have laid the foundations for all the followers, hinges on the stipulation that the interpreter makes predictions and enjoys expectations analysing a text, while incongruity in jokes emerges due to the violation of expectations. (…) the perceiver finds his expectations about the text disconfirmed by the ending of the joke… In other words, the recipient encounters an incongruity – the punch line. (Suls 1972: 83)

Unfortunately, at no point does Suls define expectations, which yields the conclusion that the term bears common-knowledge denotation. Suls upholds a view that readers will generate schemas upon the words they read and form hypotheses as to what will follow. Advancing the notion of violated expectations, Suls (1972) capitalises on the premise that the interpreter cannot entertain a number of predictions to find one of them correct but usually follows only one train of thought. While the postulate of the one-track mind appears to be most plausible and pertinent to the analysis of GPs, the factor of making predictions as regards the forthcoming text is not entirely feasible. A question arises if interpreters do develop any expectations about the forthcoming text. It may be wiser to posit that joke recipients make on-line inferences as the text progresses. Very often (but not always), given textual complexity, it is hardly possible for interpreters to make any predictions about the punchline of a long joke, as even the direction into which the text will develop may be entirely unpredictable. The crux of the matter is that this unpredictability is a sine qua non for the success of a joke. It appears safer to assume that each (prototypical) joke finishes with a punchline coinciding with a surprising element which its recipients would not have envisaged, while the factor of expectations’ violation can only be assessed post factum, i.e. after the surprising chunk has emerged. Upon the punchline presentation, interpreters experience surprise thanks to the incongruous element, which they assess, with the benefit of hindsight,

66

Chapter Two

as one they cannot possibly have expected. The punchline may either entirely subvert the inferences drawn so far, as in the case of GP texts, or only add a new incongruous piece of information (the red-light type, the cross-roads type). As already argued in the section on surprise above, this does not mean that the recipient does not anticipate any sudden twist coming with the punchline. Expectations about structure constitute some sort of meta-level prediction, whereby the ritual of joke-telling may arouse an anticipation that the text will accord with known joke patterns. (Ritchie 2003: 104)

Nevertheless, this observation is inapplicable to the humour forms in the focus of the present analysis, i.e. witticisms and one-liners which appear spontaneously in discourse. Therefore, no joke-related prediction can be present until the switch appears. GPs, primarily their lead-ups, are initially analysed outside the humour frame. The main query is whether the tenet of expectations being defeated by the text ending is relevant to GPs. It may indeed be so in GPs contingent on distorted clichés, sayings or proverbs (see chapter 4 for a detailed analysis). In these cases, the interpreter can presuppose how a given verbalisation will develop according to convention, but his/her expectations are thwarted, as the GP utterance manifests a different ending. The expectation may entail either one particular conventional chunk of text operating across contexts (the first example below) or a verbalisation which is variously formed and contextually adjusted but carries one type of predicted proposition (the second example below). If God had intended us to fly, he would have made it easier to get to the airport. [cf. If God had intended us to fly, {expectation} he would have given us wings.] If you have missed it the first time, here’s a golden opportunity – you can miss it again. [cf. If you have missed it the first time, here’s a golden opportunity – {expectation} you can see it now/you can watch the programme tomorrow, etc.]

Yet another group of GPs in which the speaker may indeed make a prediction (eventually proved wrong by the switch) embraces utterances whose final slots can be completed by default, for example, by generality narrowing (example one) or reference to a stereotype (the second example).

Theoretical Approaches to Humour

67

My favourite machine in the gym is the vending machine. [expectation: any workout machine in the gym, e.g. a rowing machine, a bike, etc.] Having drunk once, he resolved never to move his eyeballs again. [expectation: in such a situation, one decides to refrain from drinking]

Similarly, explaining the notion of thwarted expectations in jokes, Ritchie (2004) avers that in the case of some, but not all, forcedreinterpretation jokes (whose mechanism coincides, to a large extent, with the GP mechanism, see section 8, chapter 3), interpreters may develop expectations as to the wording of the textual ending, while the punchline deviates from this prediction, disappointing the hearer’s expectations (Ritchie 2002, 2004). Some jokes seem to lead the audience to expect a particular textual ending, but the punchline deviates from this prediction (…) Here we refer to the narrow case where there is a very specific prediction of some wording, not just the wider case where (as in all FR jokes) the audience’s interpretation is confounded. (Ritchie 2004: 180)

However, with the exception of the GP categories presented above, it is difficult to speak of the hearer’s expectations as to the following part of the text until they are violated, i.e. until the humorous verbalisation ends with meanings the interpreter cannot possibly have envisaged. In either of the examples below, expectations, if at all present, would be varied. It is thus questionable whether any particular expectations are built as regards the incoming material. Interpreters do not appear to predict what will happen but infer how a given part of the already produced should be understood. Even if they do make predictions, those would be aplenty, while their violation by the forthcoming part of the utterance entails the subversion of the inference drawn. It is thus not expectations but inferences engendered on the basis of the textual material that are of utmost importance in the GP interpretation process. Additionally, no expectations are nursed as to the alternative meaning of ambiguity, which does surface, unexpectedly. Oysters enhance your sexual performance but they don’t work for me. Maybe I put them on too soon. [inferences from the lead-up: the speaker’s organism is not affected by the substances found in oysters he/she eats expectations built?: “Do you know any other boosters?”, “I’m allergic to them”, etc.]

68

Chapter Two Gambling? I once won a lot. I bought my first Reubens with my poker winnings. Two Reuben sandwiches. [inferences from the lead-up: the speaker has won money by playing poker and bought a painting by Reubens expectations built?: “the painting is now hanging in the speaker’s house”, “the painting has been stolen”, “the speaker has been playing poker regularly” etc.]

In a similar vein, Ritchie (1999, 2004) observes that in some jokes, the notion of expectation is hardly adequate, for it may be reduced to the belief that the “obvious” (here salient/default) meaning generated on the basis of the set-up is the correct one. If so, the mechanism resides not in expectations but in on-line inferences, which are ultimately cancelled. Therefore, the postulate of violated expectations should be reconceptualised as the violation of inferences if it is to cover all GPs. The stipulation is that inferences drawn cannot be in tune with the import of the switch for a GP to be successful, i.e. surprising and, consequently, humorous. It is only post factum, as the ending, i.e. the switch, is performed, that the interpreter can judge that the text does not end as he/she would have expected and would never have conceived of such an ending. At this stage, the interpreter perceives the incongruity and starts the incongruity-resolution process, i.e. backtracks the text, seeking an alternative interpretation.

CHAPTER THREE HUMOUR PROPOSALS

Over the past three decades several authors have independently advanced proposals focused on joke comprehension processes. The most widely appreciated models presenting on-line interpretation will be revisited (chronologically, according to the year of publication)1 in the light of their applicability in the analysis of jokes on the whole, i.e. the three types introduced in chapter 1. Special attention will be paid to those based on the GP mechanism, which is in the focus of the present work. It will also be argued that a few frameworks present the on-line interpretation process as if all jokes were of the GP type.

1. Shultz’s and Suls’s two-stage models: The incongruity-resolution model As already reported in the previous chapter, the proposition of the incongruity-resolution model can be traced back to writings by Shultz and Suls. Shultz (1976) acknowledges that within the incongruity-resolution framework, “humor appreciation is conceptualised as a biphasic sequence involving first the discovery of incongruity followed by a resolution of the incongruity” (Shultz 1976: 12) and that in jokes “incongruity consists in the relation between the last line, or punchline, and the part that precedes the last line” (Shultz 1976: 13). Two structural elements of a joke are regarded as two parts of incongruity. Shultz (1972, 1976) proposes that the punchline creates incongruity with the set-up by providing information incompatible with the latter. In addition, Shultz (1974) applies the twostage model in the discussion on riddles. In a riddle, a question is followed by an incongruous answer, normally provided by the riddle teller, while 1 Relevance-theoretic accounts of jokes (e.g. Yus 2003; Curcó 1995, 1996) are not revisited here, given that they do not focus on the incremental development of jocular texts per se but on the layers of meanings with recourse to relevancetheoretic terminology. Also, it is the neo-Gricean, rather than post-Gricean, i.e. relevance-theoretic, approach that is supported here, see chapters 5 and 6.

70

Chapter Three

the hearer’s task is to discover why the answer is not nonsensical in terms of the question. In his succinct analyses of on-line interpretation, Shultz (1972, 1974, 1976) appears to be preoccupied almost exclusively with jokes and riddles anchored in ambiguity (e.g. the famous W.C. Fields’s response to “Do you believe in clubs for young men?” “Only when kindness fails”), stipulating that the punchline causes the comprehender to backtrack the set-up to find an ambiguous item, which can be interpreted in a different way, allowing the punchline to be compatible with the preceding part of the text. The ambiguity, therefore, facilitates the resolution of incongruity. It is not clear if the author allows for jokes devoid of any ambiguity. Shultz (1976: 13) explicitly states that half the jokes he analysed are resolved thanks to linguistic ambiguity, i.e. lexical ambiguity or semantic ambiguity (the latter term is not defined; for the denotation supported here, see chapter 4), admitting that other forms of ambiguity are also common, i.e. phonological and syntactic, both surface-structure and deep-structure. On the other hand, Shultz (1976: 14) adds that many jokes have resolutions which depend on “general, non-linguistic knowledge”, illustrating this observation with a joke earlier discussed by Suls (1972). Admittedly, no hidden ambiguity is present in this instance, which Shultz does not explicitly concede. Also, Shultz (1974, 1976) discusses riddles which entail ambiguity in the punchline, i.e. in the answer, not in the question, as in “Why did the cookie cry? Because his mother had been a wafer so long.”, where the incongruity is contingent on understanding the word “wafer” as a type of cookie and is resolved when the homophonic meaning “away for” is perceived. On this basis, it cannot be concluded that what Shultz proposes is restricted to texts with covert ambiguity of the set-up, as Ritchie (1999) suggests. It cannot be denied, however, that Shultz (1974, 1976) does not present jokes without any ambiguity. Even if Shultz put forward the two-stage incongruity-resolution model in his unpublished PhD thesis in 1970, hence earlier than Suls (1972), he is not the one to be credited for it. This is not only because the work may not be widely known among linguists but also because in his articles Shultz (1972, 1974, 1976) does not elaborate on the on-line interpretation process, while the succinct descriptions which are provided pertain not to all humorous forms but only to those couched in ambiguity, whether in the set-up/riddle’s question or in the punchline/riddle’s answer. It is hardly surprising that this linguistic discussion should be so scant and insufficient, inasmuch as Shultz (1972, 1974, 1976) is engaged not in a linguistic analysis but in research on children’s perception of humour.

Humour Proposals

71

The incongruity-resolution model for humour in linguistics and psychology is widely credited to Suls (1972, 1983), as already discussed in chapter 2. Suls’s focus of interest is the humour of jokes and cartoons “constructed to lead the recipient astray and produce surprise” (Suls 1972: 84). As regards cartoons, Suls writes about the incongruity emergent at the junction of the image and the accompanying text, i.e. the interpreter has “his expectations about the picture disconfirmed by the caption” (Suls 1972: 82). It emerges that the model in its original form would be inapplicable to cartoons without captions. Given the present scope of interest, more important are, obviously, jokes. According to Suls (1972), after a moment of surprise caused by the incongruity of the punchline, the second stage of interpretation is initiated, i.e. resolution, during which the interpreter tries to find a cognitive rule to render the punchline congruent with the preceding text with recourse to heuristics, the basic one being a means-end analysis. Upon finding a cognitive rule to render the punchline congruous, the interpreter arrives at an understanding of the joke. In other words, the recognition of incongruity is followed by its resolution via the logical adoption of another set of assumptions. The I-R processing model is framed as a cognitive problem-solving task and can be summarised as follows (cf. Ritchie 2004): 1. make predictions in on-line interpretation and keep going if there is nothing conflicting with them 2. if the input is in conflict with your predictions in the final chunk of the text, try to find a cognitive rule 3. if the rule is found, this is a case of humour. It should be observed that the model does not demand that any ambiguity be present in the set-up or in the punchline. Nor does it immanently entail backtracking and reinterpretation of the preceding text for the sake of finding an alternative meaning. It will, in all likelihood, cover all jokes, irrespective of their formal structuring and the variability of the underlying mechanisms. It will prove relevant to one-line jokes (on the assumption that a punchline can be formed even by a single word) and to long jokes. It will easily accommodate all the three joke types, without accounting for their distinctive characteristics (incomprehensibility of some element within the set-up, which might be understood as secondorder incongruity; or covert ambiguity of the lead-up). The only condition that the model imposes is the existence of a punchline introducing some form of incongruity, which must be somehow made congruous. The notion of the cognitive rule causing resolution appears to be an indeterminate one. Suls defines the cognitive rule only as “a logical proposition, a definition, or a fact of experience” (1972: 82) and adds that

72

Chapter Three

each individual may even generate a different explanation of one joke, which highlights an idiosyncratic aspect of the comprehension process. Even speakers’ idiosyncrasies aside, there may be as many rules as there are jokes and, as a result, a new rule needs to be (intuitively) generated for each joke. This indeterminacy is then understandable, for any taxonomy of such rules will always be subject to expansion. Suls (1972: 83) provides two examples to illustrate his approach. Fat Ethel sat down at the lunch counter and ordered a whole fruit cake. “Shall I cut it into four or eight pieces?” asked the waitress, “Four,” said Ether, “I’m on a diet.” One prostitute said to another, “Can you lend me ten dollars until I get back on my back?”

In the first joke, the cognitive rule helping resolve the incongruity brought about by the last utterance is the realisation that the female protagonist ignores the rule of quantity and produces the illogical statement. In the second joke (involving a horizontal pun, where the punning element is repeated twice in the surface structure), the hearer is surprised at the final word in the idiom “get back on my feet”, which violates the expectations with the “back” ending. The incongruity, however, emerges as sensible, given the nature of the woman’s profession. As the examples above illustrate, the theory does not demand any covert ambiguity in the set-up of the humorous text, although it is not explicitly excluded either. This means that although Suls (1972) provides no example subsumable under what is here called the GP mechanism, his original framework would capture it without accounting for the differences between this and the remaining two joke types. This generality of the model is its undeniable merit. Ritchie (1999) advocates the existence of two sub-models of verbal humour within the incongruity-resolution model, drawing a distinction between the two-stage model (Suls 1972) and his surprise disambiguation model (Ritchie 1999), also called the forced reinterpretation model (Ritchie 2002, 2004, 2006), coinciding what is here dubbed the GP mechanism. Ritchie’s dichotomisation is not lent support here. The major reservation against it is that his model is subordinate to Suls’s. In other words, thanks to its generality, Suls’s two-stage model, equated with the IR model, is superordinate and is relevant to jokes of the GP type as well as those manifesting the crossroads and the red-light mechanisms. All the joke types converge in the point that the punchline is initially incongruous with the set-up of a joke but is rendered congruous in the process of

Humour Proposals

73

resolution, as Suls (1972, 1983) asserts in his I-R model. Suls’s (1972, 1983) framework is broad enough to contain jokes (one-liners included) but obviously fails to appreciate the distinctive features of GP jokes and, as a consequence, to distinguish them from other joke forms. The inherent feature of the GP joke is that although the set-up, or lead-up, has two different interpretations, only one is initially available to the audience. However, since the punchline/switch conflicts with the interpretation generated, the hitherto hidden meaning is evoked to be proven compatible with it. Therefore, the GP mechanism demands that any manifestation of this particular logical rule be found so that the incongruity formed by the punchline can be resolved. In conclusion, the incongruity-resolution model adjusted specifically to GP jokes entails a provision that the punchline (here, the switch) forces the audience to backtrack the set-up (here, the lead-up) to perceive its different, so far covert, interpretation congruent with the text’s ending. Admittedly, the model will also have numerous subrealisations in jokes, as exemplified in chapter 1. However, as regards GPs discussed here, complexities will be fewer.

2. A catastrophe theory model Propounding the cusp or catastrophe theory model of jokes and humor, Paulos (1980) compares a mathematical axiomatic system allowing two interpretations to humorous incongruity within an ambiguous story or statement. The notion of ambiguity is then modelled on the cusp catastrophe. Humour revolves around the reversal of interpretation resulting in the sudden perception of a situation, statement or person in a way different to, and incongruous with, the earlier one. At the heart of this proposal lies Rene Thom’s (1975) catastrophe theory concerned with discontinuities (jumps, switches, reversals) and focused on situations where the controlling parameter’s changes provoke qualitative changes in the system. Paulos’s account abounds in discussions on mathematical paradigms and axioms, which are omitted in this overview, owing to the linguistic perspective supported here. Also, elided is the hypothesis concerning the release theory of laughter, with which Paulos (1980) couples his version of the incongruity model. The author stresses the fact that the model does not hold for “jokes that are all punch line” (Paulos 1980: 91), e.g. uncaptioned cartoons, caricature or exaggerated gestures. By this vague formulation, anchored in a most peculiar use of the term “punchline”, Paulos purports to restrict the model

74

Chapter Three

to verbal humour, while his examples are predominantly its subtypes, i.e. jokes and riddles, which he considers to follow one pattern. A joke can thus be considered a kind of structured ambiguity, the punch line precipitating the catastrophe of switching interpretations. It adds sufficient information to make it suddenly clear that the second (usually hidden) interpretation is the intended one. (Paulos 1980: 85)

Paulos (1980) posits that each joke is based on ambiguity, understood as more than one possible meaning. He also adds than only one of them is apparent, or if both should be thus, only one actually surfaces in a context. The epithet “apparent” and why only one meaning is overt are left unaccounted for. Here, such phenomena will be explained by salience and defaultness of meanings in chapters 5 and 6. As the humorous story unfolds, its elements suit both meanings, while the interpreter is aware of only one of them and follows one path of interpretation until encountering an abrupt interpretation switch, which triggers the second interpretation. Paulos (1980) also advances the phenomenon of hysteresis, which corresponds to the fact that reinforcing the first meaning is difficult (or impossible) when the second one has been revealed. It may be so indeed, but the interpreter can oscillate between the two interpretations after the incongruity has been resolved. Paulos (1980) allows for rapid alternations between two interpretations only in ambiguous pictures.

Graphic representation of the catastrophe model (Paulos 1980: passim)

The first meaning is represented as the layer ascending over the region of ambiguity. The switch from the first to the second meaning brought about by the punchline is presented as dropping from the upper to the lower layer of the graph. The gap between the two surfaces represents the

Humour Proposals

75

fact that only one interpretation is possible at a time. Also, the width of the gap between the upper and the lower levels corresponds to the strength of the drop and to the laughter-provoking value, on which the author does not expound. One may, however, observe that is difficult to estimate what causes the gap to be wider, i.e. how greater incongruity between two meanings can be achieved. Paulos (1980) also suggests that the laughterprovoking value is somehow correlated with building expectations, which may again be deemed as an elusive conceptualisation (cf. chapter 2). What Paulos appears to suggest is that the interpreter should be positive that the interpretation chosen is the right one, which is why he/she is unprepared for a different one to come. Additionally, Paulos (1980) raises the issue of the anxiety-provoking content matter of humour, such as sex and authority, as the parameter contributing to the gap’s width and thus laughter-provoking potential. Nevertheless, these assumptions cannot be either proved or disproved and manifest little scholarly value, especially that Paulos (1980: 92) vaguely writes, “More laughter, of course, does not necessarily mean funnier.” As exemplification of his theory, Paulos provides the widely known W.C. Fields jocular response (albeit in a slightly changed form) based on the homonymy of the word “club”, and a “story of a young man who registered his requirements at a computer dating service. He wanted someone who enjoyed water sports, liked company, was comfortable in formal attire and was very short. The computer sent him a penguin” (Paulos 1980: 25-26). Indeed, both the examples perfectly illustrate the cusp catastrophe, one manifesting linguistic ambiguity and the other pragmatic ambiguity. However, formulating his unfounded thesis of “the funniness of words containing the sound k” (Paulos 1980: 92), Paulos provides an example which does not conform do the catastrophe theory although the author assumes it to manifest “the catastrophe-producing punch line” (Paulos 1980: 92). A Greek regularly eats breakfast in a Chinese restaurant, where he always orders two fried eggs. The Chinese waiter always serves him politely, saying, “Two flied eggs, sir.” Finally, after years of this, the Greek gets fed up and explodes, “You idiot, learn to speak English. Two FRIED eggs, not two FLIED eggs. Understand? Two FRIED eggs, two FRIED eggs!” The next morning the Chinese waiter serves him his eggs, saying, very politely, “Two FRIED eggs…you Gleek plick.” (Paulos 1980: 92)

What Paulos fails to appreciate is that the set-up of the joke is devoid of ambiguity, viz. two competitive meanings. Therefore, the punchline, even if surprising, does not prompt the text’s backtracking bringing to

76

Chapter Three

light the second meaning, so far unobserved. It is transparent that, irrespective of the counterexamples Paulos (1980) himself provides, he perceives all jokes as subsumable under the category in the focus of the present work, i.e. the GP type. As emphasised on numerous occasions in the present work, covert ambiguity followed by the switch from one interpretation to another is not typical of all jokes. Paulos’s (1980) mathematically oriented interpretation of jokes, even if interesting, is insufficient from a linguistic vantage point, for it fails to account for the red-light and crossroads mechanisms. On the other hand, it is insufficient as a model for garden-path humour, because it fails to explain its subtypes according to any criteria and, more importantly, to motivate the transparency of the first interpretation.

3. The SSTH and the GTVH Given the high status of Raskin’s and Attardo’s output in contemporary linguistic humour research, no work must fail to discuss it. Thus, the present subchapter aims to provide an overview of Raskin’s (1985) Semantic Script Theory of Humour (SSTH), and its offshoot, the General Theory of Verbal Humour (GTVH) (Attardo and Raskin 1991, Raskin and Attardo 1994; Attardo 1994, 1996a,b, 1997a, 1998, 2001, 2002; Attardo et al. 2002). However, the two models appear to be significantly different from the incongruity-resolution framework supported here and say very little about on-line joke interpretation processes, with the exception of a few postulates introduced as if in passing.

3.1. The SSTH Due to its reliance on the concept of a script, Raskin’s (1985, 1987) proposal grew to be known as the Semantic Script Theory of Humour (SSTH). Contrary to this name and the title of the book, i.e. “Semantic Mechanisms of Humor”, Raskin (1985) focuses, not on humour on the whole, but primarily on full-fledged jokes and one-liners, even if he also mentions the applicability of his theory to spontaneous joking (puns) or self-deprecating humour (Raskin 1985: 142-144). As originally constructed, the model has a purely semantic focus, accounting for the semantic workings of jokes in the humor-competent interpreter’s perception, and usurps no right to explain any cognitive mechanisms, as Raskin (1985) explicates. Nevertheless, it is now believed to share some premises with a cognitive linguistic framework (Brône et al.

Humour Proposals

77

2004, 2006; Attardo 1997a, 2006). Two essential components determining humour competence are lexicon, i.e. the knowledge a speaker has about words and their meanings, and combinatorial rules, which enable the speaker to interpret the whole utterance by summarising the componential meanings and switching into the non-bona-fide mode of communication (see section 3, chapter 5). Raskin’s (1985) theory is hinged on the notion of a joke as a text, at least partly, compatible with two opposing semantic scripts. A text can be characterized as a single-joke-carrying text if both of the conditions are satisfied: The text is compatible, fully or in part, with two different scripts The two scripts with which the text is compatible are opposite. (Raskin 1985: 99)

3.1.1. Script The pivot of both the SSTH and the GTVH is the script (see Attardo 2001 for an extensive discussion). Developing his definition of the “script”, Raskin (1985) appreciates the existence of neighbouring terms, e.g. “schema” or “frame”, and acknowledges the provenance of the focal term and its denotation in psychological, sociological and artificial intelligence literature (e.g. Bartlett 1932, Bateson 1972, Schank and Abelson 1977, Minsky 1975). However, Raskin (1985) chooses to formulate the script as a chunk of semantic information evoked by a word, the semantic information that can be common to the whole community or that can constitute idiosyncratic information. A script consists of slot-filler pairs and is usually, but not necessarily, linked to a “lexematic handle”, which differentiates it from a psychological or experiential object (Attardo 2002). Raskin (1985) describes a script as a multi-dimensional web of words linked by semantic meanings, e.g. cause-effect relations and “a graph with lexical nodes and semantic links between the nodes” (Raskin 1985: 81). Raskin posits that all scripts of a language constitute one graph, while each word’s lexical entry is its domain and simultaneously the central node associated with adjacent and farther nodes. A significant provision is that a lexematic handle in the form of an ambiguous word is the centre of more than one domain (Raskin 1985). What is also of crucial importance from the present perspective, Raskin (1985) proposes that competitive scripts evoked by one word manifest different degrees of markedness, and thus probability.

78

Chapter Three Unless some clues to the contrary are present in the context, the combinatorial rules will always push the unmarked script for the word forward. If it turns out to be impossible (…), the combinatorial rules change the marking of the scripts for the word temporarily, for the purposes of the processed discourse and declare another script unmarked. (Raskin 1985: 89)

Minimal markedness is here viewed as the salience/defaultness of senses. However, it is here contended that only one sense is initially evoked, while the alternative one is activated later, as necessitated by the following part of a humorous verbalisation. The conceptualisation of a script changes in the consecutive accounts of GTVH, which may be regarded as giving rise to methodological convolution (Ritchie 2004). Nonetheless, these different approaches are not mutually exclusive. In the first publication on the GTVH, the script is defined as “a chunk of structured semantic information”, which “can be understood for the purposes of this article as an interpretation of the text of a joke” (Attardo and Raskin 1991: 308). Later, Attardo defines script differently. It is a cognitive structure internalized by the speaker which provides the speaker with information on how a given entity is structured, what are its parts and components, or how an activity is done, a relationship is organized, and so on, to cover all possible relations between entities (including their constituents). (Attardo 2001: 2-3)

Attardo (2001) further elaborates on the notion, providing its tripartite division into lexical scripts, which are abstract and reside in the lexicon; sentential scripts, which are more concrete, composed of activated scripts in context; and inferential scripts, activated by context without any lexematic handle. Finally, it is crucial to observe that in his in-depth analysis of the script theory in texts, Attardo (2001) adds that the first stage of a text’s processing entails activating all scripts prompted by the text and finding combinatorial rules to combine the various scripts. It is only later that the fundamental opposing and overlapping scripts are found. 3.1.2. Script opposition/overlap The overlap of scripts concerns the text being partly or fully compatible with the semantic interpretation of the whole text. Partial overlap means that some elements of the text are incompatible with one or both activated scripts. The concept of opposition should not be equated

Humour Proposals

79

with the literal understanding of the term, as the scripts are opposite “in a special sense” (Raskin 1987: 16). Raskin distinguishes four types of oppositions: negation (e.g. senators are gentlemen vs. senators are not gentlemen), antonyms (e.g. wise vs. foolish), antonyms with paraphrase (e.g. sexual activity vs. impotence) and local antonyms which are “opposite only within a particular discourse and solely for the purposes of this discourse” (1985: 108). It thus emerges that the denotation of opposition can be modified to fit the purpose of the analysis of a given joke. In addition, Raskin (1985) raises the issue of varied distance between the two opposing scripts, which may be “very closely related to each other” (Raskin 1985: 113) or which may be accidentally juxtaposed, which is the case of puns. Supposedly, by this close relation, Raskin means that scripts are related topically. A question may be pondered whether the term “opposition” is the most fortunate and whether a different one, namely “incongruity”, would not be more adequate. The notion of “script opposition” could then be substituted with “script incongruity”. Raskin (1985) is, however, intent on avoiding the latter term, for he construes his model as being independent from the incongruity approach, which will be expounded on below. This terminological shift does not change the fact that some form of incongruity (though not always between the set-up and the punchline, as devised by Suls and Shultz) and script opposition are actually equivocal. The oppositions of scripts are hierarchical and their three main levels can be distinguished (Attardo and Raskin 1991). Raskin (1985: 111) posits that, on the highest plane, all jokes analysed represent the superordinate script, i.e. real vs. unreal situation, which is divided into three subtypes, i.e. actual vs. non-actual/non-existing situation; normal, expected state of affairs vs. abnormal/unexpected state of affairs; possible plausible situation vs. fully or partly impossible or much less plausible situation. Admittedly, the real vs. unreal opposition, together with its three subrealisations, pertain to the ontology of, i.e. the reality within, the joke, not necessarily coinciding with the reality of the external world. The division into three subtypes is not flawless, given the fuzzy boundaries between the types. Raskin (1985) himself observes that some jokes are subsumable under more than one type. Therefore, a question arises if this tripartite division is a necessary heuristic tool. Obviously, these oppositions need to be evaluated intuitively as there exists no theoretical definition of actual/non-actual, normal/abnormal, etc. This can also be a point of critique of the entire proposal (Ritchie 2004). Raskin (1985) distinguishes another dimension of script opposition at the level of a number of binary categories essential to human life, e.g.

80

Chapter Three

good vs. bad, death vs. life, money vs. no money, sex vs. no sex, all of which can be reduced to the main three dichotomies. As Attardo and Raskin (1991) observe the dichotomies are not equal in generality and it may even be extrapolated that the good vs. bad opposition is a secondlevel type. Also, each of the second-order oppositions may again exhibit a wide array of manifestations, admittedly as many as there are jokes. In addition, numerous script oppositions can be found for one joke. Raskin (1985) allows for simple and compound jokes, the latter of which immanently entail multiple script oppositions as well as elementary and non-elementary scripts. It can easily be inferred that the list of all oppositions is inexhaustible, just as jokes are. As Attardo, “a hardwired list of SOs has always been a ‘necessary evil’ in the SSTH” (Attardo 1997a: 404), while Raskin’s (1985) intention was to give them only an expository character, which is why SOs have to be formed anew for each humorous text (Attardo 1997a). This substantiates that SOs are, and must be, largely under-defined, while their determination for each joke appears to rest on the interpreter’s creativity. Furthermore, although Raskin (1985) does not explicate it, scripts can be grouped as meta-scripts characterising joke types. The last three chapters of Raskin’s book focus on three types of jokes pivoting on respective scripts (or simply topics), i.e. sexuality, ethnicity and politics. Needless to say, within each of the types there are further categorisations while each of the jokes manifests different and distinctive scripts. In these analyses, Raskin no longer concentrates on script oppositions and shifts from one script to another but presents scripts as topics, whereby he changes the orientation of the book. The semantic script analysis is regarded as a milestone in linguistic research on humour and has attracted many supporters. Nevertheless, upon closer inspection, it appears to be based on intangible heuristics. Even though the notion of a script is conceptualised to a minute detail, its application to the analysis of jokes is elusive, as already signalled above. Providing exemplification for script opposition, Raskin states, “The scripts are denoted somewhat loosely, exclusively for the purpose of positive identification” (Raskin 1985: 107). To appreciate the model and its problems, one may focus on Raskin’s primary example. “Is the doctor at home?” the patient asked in his bronchial whisper. “No,” the doctor’s young and pretty wife whispered in reply. “Come right in.” (Raskin 1985: 100)

Raskin analyses it as operating on the “doctor” and “lover” scripts, which are evoked consecutively and whose overlap is “a visit at home”,

Humour Proposals

81

observing that the real/unreal situations epitomise the actual/non-actual type of script opposition and the obscene/non-obscene dichotomy. A query may be put forward, however, whether it would be more feasible to call the opposed scripts “patient” and “lover”, which would also appear plausible, or even more adequate, for such a distinction focuses on the altered conceptualisation of the key character as the joke progresses. Surprisingly enough, discussing Raskin’s example, Attardo (2001: 20) perceives the “patient” and “lover” scripts as overlapping and opposing, even though, as usual, he repeats also Raskin’s well-entrenched “doctor” vs. “lover” opposition (2001: 22). This testifies to the insufficient rigidity of the model as an interpretative tool. In later discussions on the phenomenon of local antonymy, Attardo (1997, 2001) also highlights the significance of accessibility or salience (e.g. De Palma and Weiner 1992) associated with informativeness (Giora 1991) within scripts. Accordingly, the underlying assumption, which tallies with the cognitive linguistic approach, is that scripts manifest default, salient2 and unmarked sets of elements (e.g. for the “doctor” script it is “curing patients”). For the sake of local antonymy, at least one of the two scripts must have a contextually forced and non-default slot filler pair. Additionally, Attardo et al. (2002) present this proposal with recourse to set and graph theories. There are also issues related to the method of script introduction. The “unreal” type of script can manifest different types of relationships to the real script, e.g. being externally parallel to it, or internally parallel when brought about by one of the joke’s characters (Raskin 1985). Also, one of the scripts may not be evoked directly but by implication (Raskin 1985). Further discussion on the nature of scripts’ activation is conducted in the section on the on-line interpretation process.

3.2. The GTVH Widely criticised and revised, the SSTH gave rise to the General Theory of Verbal Humour (GTVH) (Attardo and Raskin 1991; Raskin and Attardo 1994; Attardo 1994, 1996a,b, 1997a, 1998, 2001, 2002; Attardo et al. 2002), which centres on six hierarchically organised Knowledge Resources (KRs). As Attardo and Raskin (1991) and Attardo (1994) note, the SSTH collapses all the six parameters onto the script opposition/overlap 2

The two terms, i.e. salience and defaultness, are of most crucial significance in the context of GPs. However, their understanding is entirely different in the present approach.

82

Chapter Three

mechanism, while the GTVH is more detailed than its ascendant and encompasses tenets derived from various scholarly perspectives, not only semantics but also textual linguistics and pragmatics. Although the GTVH, given its name, appears to have been devised to account for all forms of humorous texts, it concentrates on canned jokes. In their first article on the GTVH, Attardo and Raskin (1991) emphasise that the theory accounts for verbal humour3 but they focus on jokes as its most representative subset. Attardo and Raskin (1991) rebut one of Morreall’s, i.e. the article’s pre-publication reviewer’s, critical observation that their theory will not cover all verbal humour such as rhymes, alliteration, funny sayings, etc., asserting that such “jokes” (not the most fortunate term, cf. chapter 1) will be manageable by the GTVH (Attardo and Raskin 1991: 333-334). Unfortunately, the authors fail to account for this claim, apart from highlighting the fact that some of the forms mentioned are grounded in the meta-linguistic use of language and that the phenomenon of script opposition/overlap can indeed be found in such forms. Attardo (1994, 1996b, 1998, 2001) avers that the GTVH is also applicable to other texts, notwithstanding their varied formal requirements, particularly more complex ones, such as anecdotes, short stories or even novels, whose humorous force originates from repeated application of many micro-elements. Moreover, Attardo (2007) highlights that the scope of the GTVH has been broadened even further and, contrary to its name, it is also applied to non-verbal humour, e.g. cartoons. However, a claim can be ventured that the model is not the most useful in the analysis of certain forms and pragmatic categories of conversational humour, such as witticisms or teasing. Even if some criteria can indeed be employed in analyses of conversational humour, primarily the script opposition (cf. Norrick 2003, Archakis and Tsakona 2005), many knowledge resources will be left empty, proving the approach inadequate. According to the GTVH, each humorous text is characterised by six Knowledge Resources, which form a hierarchy. Those in higher positions are more abstract and determine those more concrete, placed at lower levels. The hierarchical template presented below serves to measure similarity between jokes (Attardo and Raskin 1991, Raskin and Attardo 1994, Attardo 1994), which renders GTVH an adequate metric for humour translation (Attardo 1994, 2002). 3

Contrary to Attardo’s (1994, 1996a) understanding of verbal jokes as being vis-àvis referential ones (see chapter 4), verbal humour covers all forms of humour conveyed by means of language/texts.

Humour Proposals

83

Script Opposition (SO): the source of incongruity Ļ Logical Mechanism (LM): the resolution of incongruity, if any Ļ Situation (SI): the setting of a joke Ļ Target (TA): the butt of the joke Ļ Narrative Strategy (NS): the kind of communication the joke is Ļ Language (LA): the signifier of the semiotic.

The rising number of converging KRs correlates positively with the similarity of the jokes compared. Jokes are predicted to be more similar in direct proportion to the number of parameters they have in common, and conversely to differ more if the values of many parameters are different. (Attardo 1994: 228)

Additionally, in the light of the linear variation hypothesis verified by Ruch et al. (1993), the degree of perceived difference between jokes increases linearly with the hierarchical position of the KR in which two jokes are divergent. Namely, two jokes different at the level of Language are considered to be relatively similar while those governed by different Script Oppositions are regarded as maximally divergent. In essence, the degree of perceived difference between jokes increases linearly in accordance with the hierarchical organization of Knowledge Resources, the only reservation being the switch between Logical Mechanism and Situation. Admittedly, such a phenomenon may result from the fact that a change in Situation is conspicuous and more easily perceivable than Logical Mechanism, which works beyond the realisation of laymen. The lowest KR is the Language, which contains all indispensable information for the verbalisation, i.e. the surface structure of a text, including “all the choices at the phonetic, phonologic, morphophonemic, morphologic, lexic, syntactic, semantic and pragmatic levels of language structure that the speaker is still free to make” as well as “a few specifically humorous elements and relations” (Attardo and Raskin 1991: 298). In essence, this KR governs the precise phrasing and placement of the punchline. LA is subject to crosslinguistic translation and interlingual translation, i.e. transformation (paraphrase with recourse to synonyms and different syntactic constructions) on condition that the meaning of a joke, i.e. its semantic content, is preserved. The exception is the verbal joke (here dubbed linguistic), which heavily depends on the exact wording, and

84

Chapter Three

thus linguistic ambiguity. Discussing LA in the case of verbal (here linguistic) jokes, Attardo proposes, “the exact wording of the punch line of verbal jokes is extremely important because it is necessary for the linguistic element to be ambiguous and to connect the two opposed senses of the text” (Attardo 2001: 23). By this, the author appears to suggest that ambiguity must occur in the punchline, which is obviously not necessarily the case, and transparently not the case of GPs, where ambiguity obligatorily appears in the text before the punchline. Another KR is the Narrative Strategy, which determines the organisation of a humorous narrative. In other words, this KR aims to capture each form of humour as a type of narrative (Attardo 2001). As Attardo and Chabanne (1992) suggest, all jokes are narrative in nature and embrace, e.g. a narrative, a dialogue or a riddle. On the other hand, not each manifestation of humour is a textual narrative but, for instance a cartoon. The third KR is the Target, which accounts for the butt of the joke, whose categories are many, e.g. a stereotypical group or an ideological target (e.g. marriage). It must be borne in mind that there do exist jokes which are non-aggressive and have an empty value for this parameter, as Attardo (1994, 2001) also observes. Yet another KR is the Situation offering a description of objects, participants, instruments and activities relevant to a joke, even if not overtly mentioned. Attardo (1994) admits that the dearth of research devoted to this KR may be due to the fact that it is a largely intuitive phenomenon corresponding to various scripts activated by a text. Therefore, it could be theorised that the richness of this parameter will correlate with the length and complexity of a humorous text. It should also be noted that the number of potential realisations of each component varies, affecting the possibility of taxonomisation thereof. While some categorisations (even if only general) can be made as regards the Script Opposition/Overlap (Raskin 1985) and the Logical Mechanism (Attardo 1988), the realisations of the Situation KR and the Language KR are unbound, which is why no exhaustive classification can be provided. Finally, some Knowledge Resources tend to be immaterial, since jokes may be devoid of, for example, the Target or the Situation. 3.2.1. Script opposition and logical mechanism The two most essential KRs are the Script Opposition and the Logical Mechanism. The former is inherited from Raskin’s Semantic Script Theory. It should be noted that although the aspect of overlap appears to

Humour Proposals

85

have been backgrounded in the GTVH, it is still present, inasmuch as the developed proposal retains the premises of the SSTH (Attardo and Raskin 1991: 308, 326; Attardo 1994: 226). This is why the acronym SO may need to be bifurcated into the SOp and the SOv (e.g. Hempelmann 2004: 390). However, Attardo (e.g. 1997a) tends to use the acronym “SO” solely in reference to the script opposition. In various writings, Attardo champions the equations between Script Opposition and incongruity, and between Logical Mechanism and resolution (Attardo 1994, 1997a, 2001; Attardo et al. 2002). Attardo (1997a) also compares the set-up with the stage where the script overlap arises. In essence, Attardo states that incongruity and script opposition, as well as resolution and logical mechanism are “different conceptualisations of the same phenomena” (1997a: 403). Attardo (1997a) and Attardo et al. (2002) postulate that the logical mechanism is the resolution of script opposition and it is comparable to incongruity resolution, viz. “the LM is the resolution itself, not just the enabling mechanism thereof” (Attardo 1997a: 409). Indeed, in Suls’s framework, incongruity emerges at the junction of the set-up and the punchline, while in the GTVH, the second script is also enforced at the punchline stage. Similarly, this definition of the LM does not appear to contradict Suls’s (1973) idea of incongruity resolution, which hinges on finding a cognitive rule for each joke. It emerges that the I-R model and the GTVH offer parallel descriptions of the same linguistic phenomena, resorting to different methodological apparatuses (see section 3.3). Attardo and Raskin (1991) enumerate logical mechanisms such as juxtaposition, garden-path mechanism (already discussed in chapter 1), figure-ground reversal, faulty reasoning (which is later deemed as too broad a category), chiasmus and false analogy. Later, the list is expanded, covering the mechanisms presented in the table (Attardo et al. 2002: 18). role-reversal vacuous reversal garden-path almost situations inferring consequences coincidence proportion exaggeration meta-humour

role exchanges juxtaposition figure-ground reversal analogy reasoning from false premise parallelism ignoring the obvious field restriction vicious circle

potency mappings chiasmus faulty reasoning self-undermining missing link implicit parallelism false analogy Cratylism referential ambiguity

86

Chapter Three

It should be appreciated that the mechanisms do not form taxonomy of distinct operations. Attardo et al. (2002) argue that several groups of mechanisms can be found. The first one embraces reversals (vacuous reversal, chiasmus, figure-ground reversals, role reversals, exchanges and potency mappings), which are placed at different levels of abstraction. Secondly, juxtaposition and sequence, together with parallelisms, viz. proportion and (implicit and explicit) parallelisms, capitalise on similar mechanisms related to direct spatial relationships. The third group is comprised of reasoning LM, encompassing its faulty and correct types, metahumour, garden-path, as well as self-reflexive and undermining LMs. More importantly, these LMs appear not to be distinguished according to one methodological criterion or even a clearly defined perspective. LMs represent intellectual processes to be realised by the joke addressee, the processes within the narrative or joke structure, or the intellectual processes performed by joke characters, which is corroborated by the examples the authors provide in their two articles (Attardo and Raskin 1991, Attardo et al. 2002), apparently failing to observe the diversification of the layers of analysis. In addition, jokes may be described by more than one mechanism. For instance, to illustrate the phenomenon of juxtaposition, Attardo and Raskin (1991: 206) provide the following example “‘Who supports Gorbachev?’ ‘Oh, nobody. He is still able to walk on his own.’” It is transparent, however, that this riddle is also a clear manifestation of the GP mechanism couched in lexical ambiguity. The authors’ meticulous analysis of mechanisms may be interesting but fails to capture the main cognitive patterns of joke comprehension supported here. Even if the authors do distinguish the garden-path LM, which is here believed to constitute one of the primary patterns of joke interpretation processes, they put it on the same footing as other (minor) LMs.

3.3. The SSTH and the GTVH vs. I-R theories It is often assumed in linguistic literature that the Semantic Script Theory of Humour (Raskin 1985, 1987), and the General Theory of Verbal Humour (Attardo and Raskin 1991, Raskin and Attardo 1994; Attardo 1994, 1997a, 2001) subscribe to the incongruity-resolution framework. Attardo (1997a) concedes that the General Theory of Verbal Humour was originally construed as a model independent from the incongruityresolution approach but was later classed as such. Attardo (1994, 1997a, 2001) thus draws a number of parallelisms between the two models, primarily the SO being incongruity and the LM being incongruity

Humour Proposals

87

resolution, whereas the other co-author perceives the fact as a serious misinterpretation recurrent in humour literature. Raskin (Raskin, 1985, 1987; Aymone 2007) regards his model as neutral to other humour theories and simultaneously explicitly divorces it from the I-R framework. Raskin openly states, Attardo and I disagree on whether my script-based theory of humour (SSTH, see Raskin, 1985) or our joint general theory of verbal humour (GTVH, see Attardo and Raskin 1991) should be listed as incongruitybased theories just because they do involve a script opposition in the joke text. (Aymone 2007: 222)

In addition, Raskin maintains that proposals propounded by Attardo and himself have an advantage over the incongruity approach and it is the advocates of the latter that should subscribe to their theory, not vice versa (Aymone 2007: 222). This is certainly a bold statement. It would probably be safer to conclude that the two competitive models, based on dissimilar psychological/linguistic premises, are different and independent from each other, notwithstanding the few points of similarity (not equations but correspondences) which can be found. Raskin (Aymone 2007) recognises the aptness of Kant’s conceptualisation of incongruity but remains impartial, or, as he admits, “indifferent” to the psychological perspective on the phenomenon and, admittedly, also to the cognitive linguistic approach, which is very often drawn on in many studies on humour (cf. Suls 1972, 1983). Interestingly enough, Attardo (2007) interprets Raskin’s (Aymone 2007) indifference towards the incongruity approach as his acceptance of the model. It is evident that Attardo (2007) strives to reconcile Raskin’s and his own disparate views, while Raskin explicitly (Aymone 2007) dissociates himself from any cognitive or psychological theories and their “methodological burden”. It is ironic that he should believe so, since the SSTH (together with the GTVH) is considered to be a cornerstone in cognitively oriented research on humour within the I-R model (e.g. Brône and Feyaerts 2004, 2006; Attardo 2006). For example, Brône and Feyaerts write that the SSTH and the GTVH are cognitive linguistic in the sense that they explore the interface between language and cognition in highly creative language use. Witness to this cognitive rather than structural-typological approach is the fact that concepts such as scripts and frames, incongruity and cognitive resolution, which are grounded in cognitive psychological models of humor, are the central pillars of these theories. (Brône and Feyaerts 2004: 362)

88

Chapter Three

However, as already discussed, Raskin (1985) will have conceived of scripts as semantic and not cognitive (psychological or linguistic) phenomena, albeit with the reservation that they had initially been proposed in psychological studies. Secondly, the KRs of language, situation, target and narrative strategy bear little relation to cognitive analyses. Thirdly, the remaining two KRs, i.e. script opposition and logical mechanism, even if concentrated on cognitive phenomena, are not unanimously regarded as synonymous to incongruity and its resolution, respectively. Nor do they capitalise on cognitive research findings. In essence, the SSTH and the GTVH are significantly divergent from the incongruity-resolution framework. Although the GTVH happens to be viewed as including the aspect of incongruity and its resolution in the form of script opposition and logical mechanism, it can hardly be deemed as the I-R model, owing to the abundance of other semantic elements (other KRs), which the latter does not account for, being focused on different aspects of humour-related phenomena. Nevertheless, it must be emphasised that Attardo’s reconceptualisation of the SO and the LM is not misguided. This, on the other hand, provokes a query as to the necessity for the introduction of the SO and the LM in the SSTH and the GTVH, if the same phenomena had earlier been known in humour literature as incongruity and its resolution. As Attardo’s (2006) critical comments levelled at other researchers prove, he is averse to the multiplication of terms denoting the same phenomena.

3.4. Script enforcement: The trigger and on-line processing As mentioned in the brief introduction to the section on the SSTH and the GTVH, neither of the two models expounds in depth on the on-line interpretation process. The authors present a number of postulates but do not appear to treat them as being of pivotal importance to their major theories, which is also why they are often neglected by other authors revisiting Raskin and Attardo’s work. Nevertheless, the postulates on online interpretation have great relevance to the present approach and must necessarily be revisited. Raskin’s (1985) SSTH is primarily focused on the presentation of the joke as a text compatible with two overlapping, albeit opposite, scripts. Apart from constructing the Main Hypothesis, as the foregoing conceptualisation is dubbed, Raskin (1985) discusses the textual organisation of jokes, which may be regarded as an analysis of the on-line interpretation process. Accordingly, Raskin (1985) introduces the concept of the semantic script-switch trigger, a textual element found in many (but

Humour Proposals

89

not all) jokes which triggers the switch from one script to another and whose role is to render the second script “more plausible and less nonactual, abnormal or impossible” (Raskin 1985: 115). Also, the trigger may either be semantically ambiguous, and hence couched in lexical ambiguity, figurative ambiguity, situational ambiguity or quasiambiguity (see chapter 4), or may enforce a second interpretation as the contradiction trigger, with its subtype, the dichotomising trigger. Concentrating on the classification of the divergent examples to which Raskin (1985: 115-116) refers, one may infer that the trigger (in bold, in the examples below) is not flawlessly framed, its status being vague, and offers little insight into the on-line analysis of jokes. “Does the accused have anything else to offer in his defence?” “No, your Honor, I’ve already given everything I own to my lawyer and a couple of jurors, who are friends of his.” (Raskin 1985: 45), “lexical ambiguity trigger” (Raskin 1985: 115) Stalemate. The wife you are tired of. (Raskin 1985: 46), “a quasiambiguous trigger” (Raskin 1985: 116) “He used such nautical terms.” “Yes, sailors always talk dirty.” (Raskin 1985: 29), “a quasiambiguous trigger” (Raskin 1985: 116) Should a person stir his coffee with his right or his left hand? Neither. He should use a spoon. (Raskin 1985: 46), “syntactic ambiguity trigger, with introducing the agent or the instrument” (Raskin 1985: 116) A doctor tells a woman, “Your husband must have absolute rest. Here is a sleeping tablet.” “When do I give it to him?” she asks. “You don’t,” explains the doctor. “You take it yourself” (Raskin 1985: 45-46), “a contradiction trigger” (Raskin 1985: 116) Why does a donkey eat thistles? Because he’s an ass. (Raskin 1985: 26), “figurative ambiguity trigger” (Raskin 1985: 115) “My wife used to play the violin a lot but after we had kids she has had not much time for that.” “Children are a comfort, aren’t they?” (Raskin 1985: 55), “a contradiction trigger or situational ambiguity” (Raskin 1985: 116) By the time the wise man is old enough to marry, the fool has enough children to support him. (Raskin 1985: 29), “a contradiction dichotomising trigger” (Raskin 1985: 117)

90

Chapter Three

The first problematic issue is the relation between the trigger and ambiguity. It is noteworthy that even if the contradiction trigger is not ambiguous, the joke can still operate on ambiguity, even if non-lexical. For example, in the doctor joke above, one may detect the hidden ambiguity underlying the doctor’s advice, which the patient (just as the joke hearer) fails to appreciate initially. By the same token, in Raskin’s primary example about the patient and the doctor’s wife, the contradictory trigger (“no, come in”) enforces the initially latent ambiguity in the preceding part of the joke. As a matter of fact, Raskin claims that ambiguity is present in the majority of jokes. In many if not most jokes, however, ambiguity is deliberate and the intention of the speaker includes two interpretations which he wants the hearer to perceive (…) the hearer knows ‘the rules of the game’ and is not only ready to perceive the second interpretation along with the first obvious one but actually is willing to look for it. (Raskin 1985: 115)

Clearly, what the author means is that ambiguity underlies most jokes, realised within the humorous frame, which the hearer is also willing to assume. However, transparently not all jokes do entail ambiguity, while ambiguity boasts varied formal manifestations (see chapter 4) and realisations as regards its position in the text and the activation of competitive meanings. This, in turn, determines the joke interpretation process. Secondly and more importantly, as the examples quoted above show, the trigger can belong to either the set-up (ambiguity trigger) or the punchline (ambiguity trigger, contradiction trigger), or even both (the dichotomising trigger). A question arises whether in cases when the trigger does not belong to the punchline it should not be regarded as the element facilitating the switch, rather than causing the switch, which appears to be the effect caused by the punchline, i.e. the final phrase/clause/sentence of a humorous text. Irrespective of the multifarious examples provided (e.g. those quoted above), Raskin makes generalisations which are at variance with some of the examples. Retroactively, due to the ambiguity, the second script changes the interpretation of the first part of the text of the joke (…) The usual effect of the trigger is exactly this: by introducing the second script it casts a shadow on the first script and the part of the text which introduced it, and imposes a different interpretation on it, which is different from the most obvious one. (Raskin 1985: 114)

Humour Proposals

91

The contradiction, while operating differently from ambiguity, creates exactly the same effect, namely, a second interpretation retroactively imposed on the whole text preceding the trigger as well as on the text following it, if any. (Raskin 1985: 116)

In these two statements, Raskin (1985) transparently fails to acknowledge all the examples where the trigger occurs in the set-up, in which case it is not the element that retroactively enforces the second script but, constituting a basis for ambiguity, only facilitates the occurrence of the two scripts in the light of the punchline. The latter is element which does cause the reinterpretation, while ambiguity within the preceding text only facilitates this effect. While discussing semantically based trigger types, Raskin (1985) omits the position-related dichotomisation, which is why also the trigger’s function appears to be vague. This is regrettable, since a more adequate discussion on the trigger’s position and function might elucidate the on-line joke interpretation process, i.e. the way in which and the stage at which the two scripts are brought together. Most significantly, as the two quotations above prove, in Raskin’s view, all jocular texts represent the structure typical of garden-path humour, i.e. there are two text interpretations, one of which is instantly perceivable while the other emerges only thanks to backtracking after the vaguely defined trigger, thus preferably the punchline, requires this. Therefore, the text must inherently be covertly ambiguous in the set-up, which does not have to involve an ambiguous trigger but pragmatic ambiguity. Raskin (1985) also appears to uphold a view that ambiguity need not be lexical or syntactic, i.e. need not reside in the surface structure (see chapter 4). Raskin’s prime example about the doctor’s adulterous wife is certainly a GP joke. Therefore, the analysis of the joke in the light of its script opposition and script overlap may bear some elements reminiscent of the approach supported to hold for GPs. Accordingly, at the stage of the script-switch trigger in the form of doctor’s wife response, the script analysis of the text obtained so far will be recognized as the patient’s interpretation, and the theory will start the search for an alternative analysis. (…)This switch will reconfirm the command, already arrived at in the analysis at that time, to the combinatorial rules to start looking for a competing script analysis of the entire text or part thereof in view of the Main Hypothesis and a subsequent discussion of script overlap. Another clue (…) will be to look for a special kind of a competing script, namely a potentially opposed script. (Raskin 1985: 124 - 125)

92

Chapter Three

However, the opposition of the scripts is not always brought about by any hidden ambiguity of the set-up, with the obvious script being eventually ousted by another. Such a description is certainly inapplicable to many humorous forms, including many of the instances Raskin (1985) provides, e.g. the last three jokes quoted earlier (Raskin 1985: 45-46, 26, 55) or the following three. A rogue who was being led out to execution on a Monday remarked: “Well, this week’s beginning nicely” (Raskin 1985: 25), “ordinary week (life) vs. death” (Raskin 1985: 107) “Davey, do you know that your assistant just left in a Jaguar?” “So what? Left in Jaguar, returns in February” (Raskin 1985: 47), “Jaguar vs. calendar” (Raskin 1985: 108) There’s only one trouble with watching TV in glorious color. You turn it off and the world’s in black and white. (Raskin 1985: 106); “literal color vs. figurative color” (Raskin 1985: 108)

It emerges that these jokes have two different interpretations, or opposite scripts, the second of which is introduced at the stage of the punchline, or the trigger. The two scripts, however, are not mutually exclusive as such, i.e. the second one does not cancel the first. Nor do they capitalise on any form of hidden ambiguity and retroactive enforcement of the second meaning. While the first two jokes manifest the red-light structure, the third one pivots on the crossroads mechanism. In the light of Raskin’s (1985) explicit formulations cited earlier, but irrespective of numerous counterexamples he provides, a number of authors also observe that the SSTH presents jokes as operating on two scripts, only one of which is initially available to the interpreter, while the other emerges only due to backtracking in view of the switch (Ritchie 2004, Brône et al. 2006). Ritchie (2004) rightly argues that Raskin’s SSTH is on a par with his forced-reinterpretation model (see section 8), which is compatible with the GP mechanism discussed here, holding only for one class of jokes, in which the final portion of a joke brings into focus the hidden interpretation of the set-up. By contrast, Brône et al. (2006) perceive Raskin’s postulates as pertinent to all joke types, as if all represented the GP category. Jokes, on this view [i.e. the SSTH], are partly or fully compatible with two different, (con)textually opposed scripts, only one of which is saliently activated in the first part of the text. The punch line of the joke turns out to be incompatible with the first script interpretation (incongruity), but there

Humour Proposals

93

is a lexical cue in the text (script switch trigger) that enables the switch or shift from the first interpretation to the second, backgrounded script (resolution). (Brône et al. 2006: 207)

Brône et al. (2006) do not fully appreciate Raskin’s conceptualisation of the trigger and its positioning. As observed here, the switch may belong not to the set-up but to the punchline, and it may not coincide with the ambiguity of the text. Secondly, the authors see no problem in subsuming all jokes under this interpretative model, whereas it appears relevant only to GP jokes. In a similar vein, in the first GTVH article, a postulate may be found suggesting that all jokes entail covert ambiguity/script opposition uncovered by the punchline. The main claim of SSTH is that the text of a joke is always fully or in part compatible with two distinct scripts and that the two scripts are opposed to each other in a special way. In other words, the text of a joke is deliberately ambiguous, at least up to a point, if not to the very end. The punchline triggers the switch from the one script to the other by making the hearer backtrack and realize that a different interpretation was possible from the very beginning. (Attardo and Raskin 1991: 308)

This is a considerable improvement upon Raskin’s original formulation, insofar as it avoids the ambivalent concept of the trigger, and delegates its work, i.e. the shift of scripts, to the punchline. However, this account of joke processing, a perfect description of the GP mechanism, does not apply to the two remaining types of jokes, i.e. the red-light and the crossroads mechanisms, where no ambiguity in the set-up or the lead-up is present. Finally, it should also be highlighted again that the garden-path mechanism (Attardo and Raskin 1991) is listed as only one out of the 27 mechanisms found in jokes (Attardo et al. 2002), which is perhaps how the authors indirectly concede that it is not the mechanism characteristic of all jokes, omitting to recognise the correspondence between this LM and their generalisation on the incremental interpretation of humorous texts. Placing the GP mechanism besides 26 others, many of which are infrequent, makes a false impression that its scope is relatively narrow.

4. Bisociation as schema conflict Even though Norrick’s (1986, 1987) discussion of humour is not preoccupied primarily with on-line processes, elements of such an approach can also be found in it. Norrick (1986, 1987) focuses not only on

94

Chapter Three

jokes and one-liners but also on other categories of conversational humour, i.e. retorts (responses to preceding chunks) and quips (comments on non-verbal behaviour), all of which can be either poetic (here, linguistic) or non-poetic (here, non-linguistic), depending on whether they rely on puns or not, respectively (see chapter 4). Norrick (1986) assumes as his departure point Koestler’s (1964) bisociation of frames of reference, reinterpreting it as schema conflict (Norrick 1986). Norrick (1986: 230) posits that such a reinterpretation offers “sounder footing as a model of human cognitive processes.” Indeed, propounding the notion of a schema (different from Raskin’s script), the author draws on well-established cognitive literature, e.g. works by Minsky (1979), Rumelhart and Ortony (1977) and Schank and Abelson (1977). Essentially, Norrick (1986) proposes that, in accordance with the frame theory, human knowledge is organised in schemas, understood as arrays of relations between variables denoting agents, object, instruments, etc. and acknowledges that scripts coincide with schemas for ongoing activities, with temporal and spatial relations being particularly significant. Once a schema/script is activated with its variables and relations, conversationalists are guided in their expectations, evaluations of relevance, appropriateness and the like. In humour, two schemas are activated, either because the speaker pretends to misinterpret the schema in force and thus imposes another one, or because the speaker deliberately produces a verbalisation which activates two conflicting schemas. A schema conflict occurs on one level, while conflict resolution takes place on a higher level (Norrick 1986). The phenomena of lower-level conflict and higher-level fit (Norrick 1986) may be regarded as parallel to incongruity and its resolution, respectively. In Norrick’s view, humour derives from the bisociation of schema conflict, since the lower-level conflict does not cease to exist notwithstanding the higher-level congruency. Norrick (1986, 1987) presupposes in his discussions the oscillation between two frames of reference. This means that after the stage of resolution, the interpreter may reprocess the humorous verbalisation, assessing the conflict/incongruity and the fit/congruity (cf. chapter 2). Norrick (1986) verifies this general pattern in detailed analyses of examples (falling into different humour types), explaining its wide applicability but the necessary diversification. Perhaps owing to this heterogeneity, Norrick (1986) does not employ any terms to represent textual chunks in on-line processing of retorts, quips and one-liners parallel to the set-up vs. punchline distinction but focuses on the discussion of particular examples.

Humour Proposals

95

As regards quips, the speaker produces a chunk which itself embraces two conflicting schemata, while in retorts, the speaker feigns misinterpretation of the schema in force and skews it, thereby dashing the first speaker’s intentions and expectations, and imposes a new schema, standing in conflict with the previous one (Norrick 1986). One-liners rely on evoking a schema and skewing it within two sentences, clauses or even words. Finally, longer canned jokes entail the schema conflict introduced at the stage of the punchline, which surprises the audience but is resolved at a higher level. In the case of poetic humorous verbalisations entailing lexically ambiguous chunks, Norrick (1986, 1987) appears to understand the term “higher-level fit” as a coincidence of two interpretations in one lexical form. Therefore, the fit is not contingent on the actual compatibility of the schemas juxtaposed but on the acknowledgement of the two different interpretations, and thus two schemas, as being motivated by one verbal chunk. Schema conflict is the divergence between two meanings. What does your watch say? – Ticktock, ticktock (Norrick 1986: 230) Bring that up again and we’ll vote on it. (Norrick 1986: 234) Take my wife – please. (Norrick 1986: 237) A man accompanies an ailing wife to the doctor’s office. After examining her, the doctor calls in the man from the waiting-room and says, “Mr. Smith, I have some good and some bad news for you. The bad news is that your wife has VD; the good news is that she didn’t get it from you.” (Norrick 1986: 240)

Norrick (1986) discusses the examples in the following way. In the first example above (which can be understood as an interactional pun, in which the second speaker exploits ambiguity within the first speaker’s utterance), the first speaker’s schema and expectations about the answer are not met. A schema conflict is formed on the basis of the lexeme “say”. The responder skews the schema involving the time of the day, introducing the schema of sound production. Similarly, in the quip, the schema conflict between a parliamentary situation entailing voting and an embarrassing situation (someone regurgitating) capitalises on the use of the lexically ambiguous term “bring up”, whose meanings are activated on-line. Similarly, in the example of the one-liner above, the listener must appreciate the first schema only to switch to another thanks to the reinterpretation of the lexeme “take” from “consider” to “accept”. The consecutive stage of resolution occurs at the pragmatic level and entails switching from the act of exemplification to a plea. Finally, in the case of the long joke above, there are two schema conflicts and resolutions. One

96

Chapter Three

involves the subversion of the inference as to the goodness of the news, resolved with the acknowledgement that goodness and badness of news are relative; while the other concerns the two different points of view, the doctor’s medical news and the husband’s perception of it, resolved thanks to the recognition that both are valid. Regarding non-poetic verbal humour, the higher-level schema fit appears to coincide with understanding and finding sense in a strange and sometimes seemingly irrelevant utterance. This process entails acknowledging presuppositions and implicatures, whereby the lower-level conflict is resolved. How old are you? – Old enough to know better (Norrick 1986: 234) Beauty before intelligence. (Norrick 1986: 235) I not only don’t believe in ghosts, I’m not afraid of them either. (Norrick 1986: 238)

The retort to the question (a request for objective information) introduces a new schema (subjective judgment on maturity) upon the preceding utterance, while the quip operates on two schemas (of polite behaviour and of rules which do not apply), which achieve a higher-level fit owing to the presuppositions concerning gender, age and looks. Those prompt implicatures of ridicule of the addressee and etiquette norms. Finally, the one-liner exploits a logical conflict between the schemas imposed by the first and the second part of the statement. As the sample analysis shows, Norrick’s (1986) approach to humour is not (and cannot be) homogenous and boils down to a detailed discussion of categories of conversational humour, as seen fit to corroborate the tenet of schema conflict and a higher-level fit. This is then an apt generalisation on verbal humour’s workings. Admittedly, any example of humour could be analysed according to this broad conceptualisation, whether manifesting the GP mechanism or not. Clearly, some of Norrick’s examples do entail covert ambiguity in the first portion of the text, which needs to be backtracked and reanalysed. This also shows that GP mechanism can be found not only in jokes and witticisms but also in interactional puns, in which the second speaker plays on the ambiguity of the preceding turn.

5. Isotopy-disjunction model Attardo and co-researchers (Attardo 1994, Attardo et al. 1994) discuss the linear organisation of jokes (including one-liners) within what they call the isotopy disjunction model (IDM), based on the theories propounded by

Humour Proposals

97

Gremais (1966/1983) and Morin (1966). The isotopy-disjunction model functions practically independently from the GTVH and is preoccupied not with knowledge resources but with the linear organisation of a joke. It is thus most pertinent here, given the focus of the present study. Attardo (1994) takes as his starting point the assumption that at the heart of each joke as a narrative text, there is some type of ambiguity and that the process of disambiguation takes place linearly, i.e. along a chronological axis. In Attardo’s (1994: 93) view, ambiguity is a concept surpassing lexical ambiguity and may refer to any type of linguistic underspecification which yields two coexisting semantically incompatible senses for the sake of a humorous effect. Attardo (1994) borrows the term “isotopy” from Gremais (1966/1983), who used it in reference to different readings of ambiguous expressions, also in non-humorous texts, and postulates that jokes revolve around a jump between isotopies, which is correlated with the experience of novelty or surprise. The interpreter establishes one isotopy, i.e. sense, until he/she encounters an obstructing element, i.e. the disjunctor, which causes a passage to another antagonistic sense, hence shifting the interpretation to a competitive isotopy. In addition, a distinction is drawn between referential (here non-linguistic) and verbal (here linguistic) jokes (see chapter 4). The latter, in contrast to the former, centre on a connector, realised by a textual segment bearing two distinct interpretations, i.e. ambiguity. The disjunctor is a minimal necessary element of the punchline, which could be here equated with the switch. The disjunctor is an element causing a passage from one sense to another (Attardo 1994: 96) but first triggers incongruity (Attardo 1994: 97), admittedly with the preceding texts. It could then be gathered that the disjunctor introduces incongruity and thereby causes the interpreter to make the first step towards its resolution, invoking the need for reinterpretation. Attardo (2001) also analyses this process with recourse to scripts, rather than isotopes, and incongruity, rather than oppositeness, between them. (…) while the setup part of the text establishes a given script, the occurrence of a disjunctor (punch line) forces the reader to switch to a second script (…) In essence, the very concept of incongruity tells us that the second script is non-congruous with the first one, and hence that the punchline, which brings the second, non-congruous script about, cannot be congruous with the script set up by/in the narrative. (Attardo 2001: 83)

The disjunctor is seen as corresponding to Raskin’s (1985) scriptswitch trigger (Attardo et al. 1994, Attardo1994), which is nothing but a misguided conclusion. The latter can appear either in the set-up or in the

98

Chapter Three

punchline, as discussed earlier in this chapter, while the disjunctor is transparently the pivot of the punchline. Attardo (1994: 203) also observes that the trigger “is not exactly equivalent to it [the disjunctor] since the script-switch trigger is not opposed to a ‘connector’ element.” It would be safer to refrain from drawing any parallels between the two concepts. The disjunctor is the core of the punchline or even the whole “punchline”, which is why the term is most useful in the analysis of one-liners. If present, the connector joins two isotopies and provides a playful justification for a redirection of interpretation caused by the disjunctor. In other words, in the case of verbal jokes, the disjunctor forces the recipient to seek a different interpretation, i.e. a different isotopy, of the preceding text and causes the reinterpretation of the ambiguous element, i.e. the connector. The disjunctor plays off of this ambiguous element [the connector] by causing the passage from one possible actualisation of the connector to another, which had been previously discarded by the process of selection. (Attardo et al. 1994: 28)

However, it is here argued that the alternative meaning may never have surfaced at all. Hence, no process of conscious disambiguation/selection will have taken place, which is the key to the GP mechanism. As the fourstage disambiguation process described by Attardo et al. (1994: 24) shows, the analysis proceeds smoothly until it is halted upon the perception of the disjunctor, which forces the interpreter to backtrack the text and discover the second meaning of the connector. The disjunctor, occurring both in verbal and in referential jokes, must occupy the final position of the text but may optionally be followed additional textual material. This will be so, for example, if there are a few punchlines, or more often if the material is semantically empty or explains the punchline (Attardo et al. 1994). The connector may optionally coincide with the disjunctor. Attardo (1994) observes that if the connector and the disjunctor merge (the non-distinct configuration), no actual backtracking will occur, but the interpretation up to that point will be suspended and the alternative meaning of the disjunctor/connector will determine the second interpretation of the text. Nota bene, Bucaria (2004) observes that sometimes the disjunctor can also precede the connector. This pertains to cases when the necessary means to resolve the ambiguity is provided before the ambiguous element appears. Indeed, some examples of this phenomenon can be found (e.g. “Blind [disjunctor] woman gets new kidney from dad she hasn’t seen [connector] in years.”). However, Bucaria’s (2004) examples come from the corpus of headlines and can

Humour Proposals

99

hardly be thought of as jokes, which normally spring a surprise on the recipient. Headlines manifest punning wordplay, sometimes resembling deliberately produced stylistic mistakes. In addition, it should be added that the connector and the disjunctor may coincide in form but appear consecutively in the surface structure, which is the case of horizontal puns (e.g. “In result of marriage, a woman gets a ring [connector] on her finger, while a man gets two rings [disjunctor/connector?] below his eyes.”). Only verbal (here linguistic) jokes include connectors, which, as discussed by Attardo (1994) and Attardo et al. (1994), operate on surfacelevel ambiguity of the lexical or syntactic type. One instance of a verbal joke (Attardo et al. 1994: 28-29) coincides with a GP capitalising on lexical ambiguity. In the famous answer credited to W.C. Fields, the first selected interpretation of the item “clubs”4 in the sense of “social groups” is sustained until the disjunctor “fails” emerges, in result of which the interpreter backtracks the text (rather than discard it as ill-formed) and finds the second sense of the connector, i.e. “sticks”. On the other hand, the referential joke illustrating the model, i.e. “Taxi crashes in Glasgow. 15 injured.” (Attardo 1994: 97, quoted from Olbrechts-Tyteca 1974: 89), does not manifest any ambiguity within the text preceding the disjunctor. It revolves around the final element which introduces a new, unexpected isotopy, whereby it adds new information to the interpretation of the unambiguous text (which Attardo might find ambiguous). The interpreter must conduct additional inferential processes to infer the humorous import of the one-liner, i.e. that taxis in Glasgow are crowded, because the people there are stingy. It is certainly not that the first isotopy (there has been an accident involving a taxi in Glasgow) contradicts the novel sense enforced by the disjunctor. This instance testifies that the IDM is not restricted to jocular texts centred on the GP mechanism, i.e. those which exhibit ambiguity in the set-up (lead-up), with the second meaning introduced by the disjunctor. Additionally, this example also questions Attardo’s claim of ambiguity as underlying all jocular texts. To recapitulate, the IDM holds for both referential (here non-linguistic) and verbal (here linguistic) jokes, the latter of which immanently rely on connectors, segments of the text which is subject to two different readings.5 Attardo et al. (1994) postulate that in either case, the interpreter 4 The authors perceive the lexeme as polysemous, whereas it appears to be a case of homonymy, since the two meanings are not related. 5 Attardo et al. (1994) write also about alliterative verbal jokes (which embrace rhyme, among others). However, this category does not entail ambiguity. Hence, albeit verbal (here linguistic), an alliterative joke would not be deemed as a

100

Chapter Three

processes a joke linearly and disambiguates it, i.e. combines the meanings of consecutive words, to form the first sense of the text (S1) until he/she reaches an anomalous (here incongruous) element, i.e. the disjunctor, which forces an unexpected and sudden passage to the second meaning of the text (S2). The disjunctor triggers the process of backtracking and reinterpretation or disambiguation, optionally facilitated by the justification offered by the connector, i.e. “a playful motivation for the presence of the second sense” (Attardo 1994: 97). The first available sense is regarded as untenable while the second sense as feasible, in consequence of which the former is abandoned or maintained parallel to the latter. There are, however, cases of the non-distinct configuration when no backtracking is necessary (Attardo 1994, Attardo et al. 1994), i.e. when the ambiguous element is both the connector and the disjunctor. The model might serve as a general framework of joke interpretation obtaining for both GP jokes and jokes in which there is no hidden ambiguity within the set-up, even if Attardo insists that all jokes do entail some ambiguity. In other words, the model appears applicable also to jokes which do not reside in any ambiguity per se but in the enforcement of a new and surprising sense at the stage of the punchline, specifically the disjunctor, found in the crossroads and the red-light types of jokes. Nevertheless, it the case of the former the construction of the first sense is not completed, i.e. there is an element within the text which escapes understanding. Moreover, even though the authors posit, “The anomalous element [the disjunctor] in a joke creates a passage from S1 to a different sense, opposed to the first one, which we will call S2” (Attardo et al. 1994: 28), the two senses need not be interpreted as mutually exclusive, even if incongruous or opposite (in Raskin’s sense), and the second need not annul the first, such annulment being inherent only to GPs. The text’s backtracking, which, in the authors’ view, always takes place (with the exception of the non-distinct configuration texts) does not mean that a competitive, so far covert, sense of the set-up must be found. A question arises whether backtracking is necessary if no connector or no other covert ambiguity of the set-up is present. Discussing the textual function of punchlines and disjunctors as disrupting elements, Attardo (2001) mitigates the force of the backtracking claim, by showing an alternative path, allowing for the fact that no reanalysis need actually take place. From this basic structural fact, comes the disruptive nature of punch lines: by forcing the hearer/reader to backtrack and reinterpret the text, or by linguistically based GP, unless a covertly ambiguous element were simultaneously present.

Humour Proposals

101

forcing him/her to produce a new and incompatible (locally opposite) interpretation of the text, the punch line cannot be integrated in the narrative it disrupts (which is the one that has set up the first script). (Attardo 2001: 83)

In essence, if the postulates of obligatory backtracking and inherent ambiguity of the set-up are rejected, Attardo’s proposal will present the one-liner as a text ending with a disruptive element, the punchline or the disjunctor, which provokes incongruity or local opposition with the isotopy established in the set-up. It should be observed that because the IDM presents a dichotomy between verbal (here called linguistics) and referential jokes, it clearly distinguishes connector-disjunctor jokes, which coincide with GPs operating on lexical ambiguity, while the issue of GPs contingent on pragmatic ambiguity is not explicitly raised. This category is merged with other jokes, i.e. those showing no hidden ambiguity of the set-up.

6. The frame-structuring model Coulson’s (2001) space/frame-structuring model is anchored in the mental space theory (Fauconnier 1994), the blending theory (e.g. Fauconnier and Turner 1994, 2002), as well as cognitive grammar (Langacker 1987). This cognitive semantic model is most adequate for various discourse analyses, since it easily accounts for the creative combination of frames with a view to constructing new ones in response to contextual demands (Coulson et al. 2006). Frame structures represent knowledge about objects, actions and events (Coulson 2001). Most significant are temporal frames, i.e. scripts, representing sequences (Schank and Abelson 1977), together with situation frames, and embracing both descriptive elements and sequences of events (Sanford and Garrod 1981). In addition, frame models are hierarchically organised and have an attribute-value structure, while default values can be assigned for unspecified attributes (Coulson 2001). In works by Coulson (and co-researchers), the premises of the framestructuring theory are used to advocate two different cognitive models for humour interpretation, one entailing the linear shift of meaning, which is in focus here (Coulson 2001, Coulson and Kutas 1998, Coulson et al. 2006), the other concentrating on (in)congruous juxtapositions of meanings (Coulson 1996, 2005a,b). In the latter strand of research on humour (primarily humorous cartoons), Coulson (2005a,b) champions a view that the comic effect emerges if one idea is linked to self-consistent but habitually incompatible frames of reference forming conceptual

102

Chapter Three

blends. In addition, Coulson (1996) analyses a joke about the “Menendez brothers virus” in the light of frame semantics and blending. The analysis does not focus on the on-line interpretation process but on the conceptual integration occurring in blended spaces, which help explain analogical mappings in the joke about a computer virus which, for a humour purpose, is metaphorically compared to Erik and Lyle Menendez, two brothers who killed their parents (conceptualised as the source domain). The text is thus discussed not as a linearly comprehended joke (which would here be classed as the crossroads joke) but as a humorous metaphor. As regards the linear process of meaning construction, according to which meaning formation progresses linearly and in stages, Coulson’s (2001) basic premise is that the global interpretation of an utterance often relies on understanding what space is being built and what background frames are relevant. A crucial provision is that in the absence of explicit content, frames are filled in by default. Apart from conceptual blending, Coulson (2001) discusses the phenomenon of semantic leaps corresponding to the process of frame-shifting, both of which help language users combine input from different domains in order to understand and express new ideas. Observing that the flexibility of the interpretative process is the most evident in jokes, Coulson (2001) corroborates her theses on human interpretative processes on the basis of one-liners. Coulson (2001), Coulson and Kutas (1998) and Coulson et al. (2006) elaborate on the dynamic aspects of incremental meaning construction and propose frameshifting as the mechanism underlying one-line jokes, in which the final lexical element affects the construal of the whole proposition. The construction of meaning, i.e. the message level representation formation, is dependent on the integration of the verbal input, i.e. lexical processing, and background knowledge (Coulson et al. 2006). Although Coulson does not explicate it, this background knowledge will be responsible for the default/salient interpretation of a verbal chunk, which may sometimes have more than one interpretation. On-line construction of meaning is (…) an active process in which the speaker integrates ongoing aspects of perceptual and conceptual information with more abstract information available from long-term memory. (Coulson 2001: 29)

The lexical level and message level representations are interdependent, but lexical items both benefit from and contribute to the creation of the message-level representation. Firstly, words are interpreted in compatibility with the contextually evoked frame. Also, lexical-level

Humour Proposals

103

expectations revolve around frames active in working memory. Such semantic meanings are more easily accommodated into the activated frame than those unrelated to it. On the other hand, words contribute to the message-level representation by activating long-term memory for the proper construal of the current conceptual content. When interpreting any stimuli (utterances, events, objects, action, etc.), humans constantly invoke background knowledge frames to form expectations and make inferences pertinent to the frame construed. To reformulate, words contribute to the cognitive models active in working memory and prompt the construction of new models, which, in Coulson’s terminology, yields frame-shifting. (…) words benefit from context because structure in the message-level representation facilitates the integration of elements and relations consistent with the contextually-evoked frame. (Coulson et al. 2006: 235)

Coulson (2001) as well as Coulson et al. (2006) speak of the creation of “nonce senses”, i.e. meanings peculiar to a particular verbalisation, created in a nonconventional way. Additionally, meaning construction entails filling in various slots, which function as default hypotheses on the entities which can be found in a given scenario, while comprehension is the result of binding slots or attributes in the activated frame. Because slots in a frame can be thought of as hypotheses about the sorts of entities that might be encountered in a given scenario, a frame-based approach can deal nicely with the fact that people develop expectations in the course of language comprehension. (Coulson 2001: 83)

Coulson’s reverberating postulate concerning the formation of expectations may be questionable (cf. section 8, chapter 2) if those are equated with what interpreters consciously expect to emerge in the text, and not what is entertained below the consciousness threshold. Nevertheless, Coulson’s understanding of expectations appears to coincide with background assumptions, i.e. presuppositions, not always representing the exact form of the forthcoming text, which, however, may also be the case (e.g. the barroom frame described below). Thanks to the expectations which emerge within the activated frame, certain slots are filled in by default, while the following words whose semantic contributions concur with the activated frame are easy to process (Coulson 2001, Coulson et al. 2006). That a frame has been built does not mean that it cannot be altered. The creativity and flexibility of the process of language construction manifest themselves in the phenomenon of frame-shifting, and hence the

104

Chapter Three

construction of new frames on the basis of existing representations. Coulson (2001) posits that frame-shifting reflects the process of a semantic and pragmatic reanalysis due to which existing elements in the contextual representation are reorganised into a new frame. Frame-shifting is prompted by slot-filling constraints which do not allow some semantic contribution, whose circumvention entails entering a different frame. Therefore, jokes are “constructed to suggest one frame while evoking elements consistent with another” (Coulson et al. 2006: 232). This coincides with Minsky’s observation, also quoted by Coulson (2001). (…) unexpected frame substitution, in which a scene is first described from one viewpoint and then suddenly – typically by a single word – one is made to view the scene-elements in another, quite different way. (Minsky 1980: 10)

Coulson (2001) and Coulson et al. (2006) use Attardo’s (1994) and Attardo et al’s (1994) terms “disjunctor”, the element which prompts the frame-shift, and “connector”, the linguistic bridge between the two frames. First, the interpreter extracts a frame based on the combination of the text and knowledge retrieved from long-term memory, before the disjunctor emerges as inconsistent with the message-level expectation, and sometimes also lexical-level expectations, and cannot be accommodated by the frame active in the working memory. Upon encountering the disjunctor, which does not belong in the evoked frame, the interpreter is forced to backtrack and reinterpret the preceding text within a different frame. The second frame is enforced and the earlier elements of the message representation are mapped onto it, causing the creation of an entirely new scenario. This processing model is illustrated with a few examples, some of which reappear in Coulson’s publications (Coulson 2001, Coulson et al. 2006). One example, “When I asked the bartender for something cold and full of rum, he offered his wife.” (Coulson 2001: 57, Coulson et al. 2006: 231), rests on the barroom frame of an ordinary customer-bartender interaction. Because customers often ask bartenders for some drink recommendation, and because the qualities mentioned are conventionally attributed to drinks, the slot to be filled in, structurally emerging in the final part of the verbalisation (the disjunctor), is envisaged as a drink. In other words, the listener builds a cognitive model of an interaction between a customer and a bartender, combining linguistic information with background knowledge. However, the default assumption that the recommendation frame will be filled by a rum drink (e.g. daiquiri) needs to be rejected, because the slot is actually not filled by the name of an alcoholic beverage. The second frame

Humour Proposals

105

introduced and the chunk “cold and full or rum” is reinterpreted according to a new scenario, i.e. the bartender’s description of his wife. This instance can be classed as a pragmatically ambiguous GP. Similar is the case of “By the time Mary had had her fourteenth child, she’d finally run out of names to call her husband.” (Coulson 2001: 49), which depends on what would here be called covert lexical ambiguity of the word “names” evoked by the disjunctor “husband”. In Coulson’s interpretation, the one-liner leans on the necessity of reinterpretation of “names” from “first names” to “derogatory terms” and changing the frame of Mary having problems with finding names for her children for the frame of Mary blaming her husband for the fourteen pregnancies. Yet another instance of the same type is “I let my accountant do my taxes because it saves time: last spring it saved me ten years” (Coulson et al. 2006: 234), which centres on the connector “time” bridging the two frames of a busy professional and a wicked businessman, the latter of which is enforced by the disjunctor “years”. All the three examples discussed above represent what is here called the GP mechanism, couched in pragmatic (the first example) or lexical ambiguity (the other two). In either case, the ambiguity is covert, since the interpreter perceives the default interpretation concurring with the activated frame, which emerges as deceitful. The idea of being “led down the garden path” appears in one of Coulson’s articles (Coulson et al. 2006: 234). However, it is not that the model accounts for jokes of this type only. Coulson (2001) posits that frame-shifting can be prompted neither by lexical nor by higher-level ambiguity, by which she may mean what is here referred to as “pragmatic ambiguity”. By contrast, ambiguity of either type is here deemed as a sine qua non for the GP mechanism. Coulson (2001) avers that it is the violation of the default assumption for a given frame that suffices as a basis for frame-shifting. By way of illustration, in the example “Everyone had so much fun diving from the tree into the swimming pool we decided to put in a little water.” (Coulson 2001: 55), the disjunctor “water” violates the default value that the pool contains water. According to different terminology, the joke may be conceived as reliant on surprise and incongruity consequent upon presupposition violation (“people enjoy jumping into water”), thus presuppositional ambiguity, also conducive to a GP (see section 6, chapter 4). Similar is the case of “My husband took the money we were saving to buy a new car and blew it all at the movies” (Coulson et al. 2006), which exploits the default assumption that the savings amount to a considerable sum, by far exceeding the price of a few cinema tickets.

106

Chapter Three

Not all Coulson’s examples represent the GP mechanism, but what is called the red-light mechanism, in which the final textual chunk adds an incongruous element to the cognitive model built, without causing the cancellation of any inference. For example, “I asked the woman at the party if she remembered me from last year and she said she never forgets a dress.” (Coulson et al. 2006), is the epitome of the red-light mechanism, for it closes with a surprising and incongruous piece of information, which needs to be reconciled with the unambiguous preceding text. Even if the disjunctor appears to violate the hearer’s expectations (distorting the phraseme “never forget names”), the set-up does not have two different meanings. From a different perspective, Coulson also mentions the case of frameshifting for pragmatic goals, as in the example “He told her he still wasn’t himself, but she said she’d noticed the improvement.” (Coulson 2001: 61), where the Sympathy frame is substituted by the Insult frame. This could be viewed as an instance of the red-light mechanism. In addition, the case of non-standard meaning of idioms is discussed with the example, subsumable under the GP type, “She told him he looked like a million and she meant every year” (Coulson 2001: 64), where the Compliment frame is exchanged for the Insult frame. This example would here be classed as a GP rooted in phrasemic literalisation, one realisation of lexical ambiguity (see section 5.4, chapter 4). Similarly, Coulson (2001: 65) discusses a number of examples capitalising on “non-standard meaning of phrases”, which also appear to be cases of lexical ambiguity jokes. In essence, according to Coulson (2001), jokes can be explained by frame-shifting prompted by the violation of slot-filling constraints, which causes the first frame to be abandoned in favour of another one. In other words, frame-shifting is a semantic reanalysis of a text, in which the final chunk of the message-level representation enforces a new frame pertinent to the whole verbalisation. Thus, joke appreciation entails adjusting the whole message-level representation to the second frame. In addition, the message-level meanings, and hence frames, may be affected by particular word(s) used in a sentence, i.e. the connector. Therefore, “the choice of a new frame may change how we interpret the meanings of previously encountered words” (Coulson 2001: 69). This addition suggests that the model accounts for linguistic jokes of the GP category, where the ambiguity of the text preceding the disjunctor is initially covert. In the light of the above, it appears clear that the frame-shifting model of joke interpretation perfectly captures the notion of GP humour, which operates on covert ambiguity of the set-up, uncovered by the disjunctor. In fact, the author places special emphasis on examples which are anchored

Humour Proposals

107

in this mechanism, whether or not intentionally. In addition, Coulson (2001) observes that garden-path sentences (i.e. the strictly grammatical linguistic phenomena) represent frame-shifting at the abstract level, given that, in cognitive grammar (Langacker 1987), grammatical phenomena reflect the operation of schematic frames. If the conditions of the presence of backtracking and abandonment of the first meaning were explicitly presented as germane only to some forms, Coulson’s frame-shifting model could be extended to cover other forms of verbal humour, i.e. all jokes or humorous exchanges (e.g. interactional puns, teasing, etc.). However, it might not be easy for the model to embrace the crossroads joke type, where the hearer’s formation of a frame may be obstructed and postponed until the requisite information is provided in the final chunk of the text. Furthermore, the model might not present so neatly longer jokes, where more than one frame or perhaps a much more complex frame could be construed on the basis of the set-up. Attardo (2006) criticises Coulson’s model on the grounds that it is simply a notational variant of his own framework within the GTVH based on scripts, which, according to the most recent account, are built inferentially and incrementally (Attardo 2001). Also, Attardo virulently disparages frame-shifting as a joke interpretative tool, since it does not “match the SSTH’s sophistication in the analysis of humour and is nowhere near the complexity of the GTVH’s approach” (Attardo 2006: 359). It should be observed that Coulson’s proposal actually benefits from its minimalism, and thus its wide applicability to humorous phenomena, without repeating all the concepts and tenets that can be found in Attardo’s complex model, focused on elements different to on-line processing. Moreover, the theory of semantic leaps benefits from the wealth of cognitive research. On the other hand, the major drawback of the model, from the vantagepoint of humour studies, is that humour, and specifically the oneliner discussed, is not the sole manifestation of the frame-shifting process (Coulson 2001, Coulson and Kutas 1998, Coulson et al. 2006), which is why the framework does not highlight the peculiarities of humorous mechanisms. One may pose a question how a distinction can be drawn between jocular texts and other texts revolving around frame-shifting. No clear answer is provided, but Coulson and Kutas (1998) and Coulson et al. (2006) present the results of their psycholinguistic research on the on-line interpretation processes entailing (humorous) frame-shifting, testifying that sentences with humorous endings take longer to read than nonhumorous sentences which also entail-frame shifting. On the whole,

108

Chapter Three

irrespective of these doubts, frame-shifting offers a credible descriptive model of short jokes’ processing.

7. Graded salience and Marked Informativeness Requirement Giora (1991, 2003) assumes that, in semantic terms, jokes are commonly deemed as involving double entendre (understood as punning), relying on two incompatible interpretations, and addresses the problem of the mechanism underlying ambiguity-dependent jokes in the light of the graded salience hypothesis6, the Marked Informativeness Requirement and the suppression hypothesis. This cognitive model has a solid scientific basis, for the author adduces evidence from psycholinguistic research to support her theses (cf. Giora 2003). In essence, Giora develops a twofold condition for a well-formed joke, viz. 1. the fulfilment of the Relevance Requirement of the last interpretation 2. the fulfilment of the Marked Informativeness Requirement and the linear shift, i.e. interpreter’s cancellation of the unmarked (salient) interpretation in favour of a marked one.7 (Giora 1991: 470)

Giora (2003) also takes as her point of departure the switch of meanings referred to as frame-shifting (Coulson and Kutas 1998, Coulson 2001), which she criticises for failing to answer the question which meaning is activated first and which later, or which is retained and which is suppressed. Giora (2003) addresses one of the problems of the frameshifting approach, explaining that the existence of suppression differentiates between jokes and non-humorous texts which also entail frame-shifting. Jokes reside in salience imbalance (Giora 1991, 2003), being initially interpreted in accordance with the most salient meaning until the punchline (in Giora’s examples of one-liners, often reduced to one word) proves to be incompatible or, as could be reformulated, incongruous with it. Hence, the punchline forces the interpreter to revisit and suppress the 6

The notion of salience is expounded on in chapter 6, given that it is here deemed as being of primary importance to the operation of GPs contingent on lexical ambiguity. 7 Giora (1991) acknowledges that some jokes manifest a different violation of the Graded Informativeness Requirement. A joke may, even if very rarely, end with a less informative element.

Humour Proposals

109

salient meaning in favour of a less salient meaning which appears to be the contextually (the context means the punchline) compatible one. In other words, with the salience imbalance between two interpretations, the comprehender first processes the more salient but incompatible meaning only to reject it in favour of a less salient but more compatible one. It is, therefore, the mind’s proneness to salience that helps the joke to occur. Jokes manipulate the very same tendency to opt for the more salient information first (…). Such discourses deliberately refrain from spelling out the intended, less salient meaning, counting on its slower activation. No wonder we are caught off guard, so to speak, ignoring the less salient yet contextually appropriate meaning while accessing the salient meaning first, only to find out that this interpretation requires reinterpretation. (Giora 2003: 6-7) A consideration of the linear ordering of the joke reveals that jokes manipulate our tendency to minimize the number of readings. The asymmetrical distribution of information in the joke concedes to or, rather, utilizes our ‘one-track-mind’ so that when the second reading is disclosed we are totally unprepared, caught off guard, so to speak. Possibly, what enables such a manipulation of ambiguity is our tendency to save mental effort. Given the hierarchal structure of concepts, where possible, ambiguous constituents will first lend themselves to the unmarked/accessible reading. (Giora 1991: 481)

In the light of the citations above, it transpires that Giora perceives jokes as immanently contingent on ambiguities manifesting salient and less salient meanings. What is most relevant to the present discussion, Giora (1991, 2003) answers the question as to how salient meanings affect ambiguity resolutions and what role context plays in the process (see section 4, chapter 6). Apart from the salience hypothesis, Giora’s (1991) approach to jokes hinges on the stipulation that while non-jokes progress gradually, jokes entail an abrupt increase in informativeness, i.e. the violation of the Graded Informativeness Requirement (Giora 1988), which results in the phenomenon of the Marked Informativeness Requirement. Giora (1991) proposes that jokes abruptly evolve from the least to the most informative element. Jokes and point-stories are markedly informative. Their final informative messages are marked in that they are too distant, in terms of the number of similar features, from the messages preceding them. (Giora 1991: 469)

110

Chapter Three

Giora’s (1991) definition of informativeness is the combination of class membership with the classical information theory. With regard to probability, the most informative is the element which is the least probable of all the options (Giora 1991). From the perspective of the prototype theory, the structure of a joke is categorical à la Rosch (1973). A category is defined as a group of members which share certain similarities. Those similarities are peculiar to a group and distinguish it from other groups. Prototypical members of a set are unmarked and the most accessible but the least informative. On the other hand, a class member is informative if it possesses more features than indispensable for category inclusion, features which are distinctive and differentiating. The most informative members are the least prototypical and accessible, i.e. the most marginal and surprising. Regarding discourse analysis, a verbal chunk is informative if it contributes more information than the remainder of the discourse. In the case of a joke, it is thus the last chunk that is the most informative, i.e. “the joke’s last constituent – the punchline is most distanced cognitively, sharing the least amount of common features with the previous constituents” (Giora 1991: 471). Giora also avers that a joke’s “final element must be cognitively different from the prototypes of the ad hoc set created by the text of the joke” (Giora 1991: 474). Encountering the most informative element, “the reader is made to cancel the first unmarked interpretation upon processing the second marked interpretation” (Giora 1991: 470). What is significant, the categorical structure of a joke suggests that humorous surprise does not derive from the disparity between the two interpretations but from “the likelihood of the less accessible/salient interpretation to be relevant to and included within the category proposed initially” (Giora 2003: 168). This is compatible with the surprise element occurring at the stage of the punchline, a prerequisite for humour in jokes. Giora highlights that the final constituent must be introduced rapidly for the sake of the surprise effect. Greater informativeness will hence correspond to greater unpredictability and surprise (Giora 1991, 2003). In essence, the violation of the Graded Informativeness Requirement depends on a sudden jump from one interpretation to another, and an abrupt passage to the marked interpretation. The third element of Giora’s (1991, 2003) theory is the suppression hypothesis, according to which the salient but incongruous meaning is retained only as long as it is conducive to the appropriate interpretation but it discarded if it interferes with understanding. Giora (2003) also hypothesises that the necessary reinterpretation combined with suppression may explicate the processing difficulty in the comprehension of jokes and

Humour Proposals

111

the resultant longer processing time of jocular texts. Contrary to the assumption that the lack of suppression (ambiguity’s cancellation) renders texts witty but not funny (Giora 1991: 476), a claim could be ventured that in humorous ambiguities the first meaning observed need not be suppressed but retained as relevant, together with the second one, which is the case of double-retention puns (e.g. “A woman was created from a man’s rib. She has been ribbed ever since.”). Giora would counter this argument on the grounds that the retention of the two meanings may yield wit but certainly not humour. As already signalled, however, the caesura between wit or humour is debatable, while the former may be perceived as a special category of humour (whether or not funny). Giora (1991, 2003) exemplifies her proposal with just a few short jokes, mainly found in topical literature, e.g. the famous W.C. Fields’s response, Raskin’s canonical example, Coulson’s (2001) one-liners, as well as a few short jokes, e.g. “‘Did you take a bath?’ a man asked his friend who has just returned from a resort. ‘No,’ his friend replied, ‘Only towels. Is there one missing?’” (Giora 1991: 472) or “Statistics indicate that Americans spend $80 million a year on games of chance, mostly weddings.” (Giora 2003: 172). On the basis of Giora’s (1991, 2003) basic postulates and examples, a conclusion can be drawn that jokes invariably pivot on double entendres, i.e. puns. Admittedly, the lexical ambiguity proviso, most crucial for salience, does not pertain to the Marked Informativeness Requirement, which would also hold for jokes devoid of any ambiguity. Moreover, the model captures almost solely the GP mechanism. Only one meaning is initially activated and so the initial interpretation of the text is unambiguous, as governed by the most salient meaning. The perception of its co-textual misfit allows the contextually adjusted meaning to reach a sufficient level of activation. In essence, a joke operates on initially covert ambiguity of the text before the overinformative punchline causes the salient meaning of the ambiguous chunk to be rejected, paving the way for another one. This view coincides with a perfect description of the lexically-based GP mechanism, which is obviously not intrinsic to all jokes. Giora does use the term “garden path”, albeit informally. To lead our “one-track-mind” down the “garden path”, the initial context of a joke is usually unambiguous, compatible with the salient meaning, so that this interpretation is retained up until the punch line, at which point a sudden incongruity forces reinterpretation. (Giora 2003: 168)

On the other hand, to illustrate context-insensitivity of salience, Giora (2003: 169) provides two examples where the ambiguous element

112

Chapter Three

coincides with the “punchline”, i.e. “What do you call Santa’s helpers? Subordinate Clauses.” and “How do you get holy water? Boil the hell out of it.” In both the riddles, which can be interpreted as reliant on the crossroads mechanism, the ambiguous chunk coincides with the final part of the verbalisation. Therefore, the salient meaning is activated at the very end of the interpretation process only to be instantly suppressed as irrelevant, without any switch following. Moreover, thanks to another example, “He’s so modest he pulls down the shade to change his mind.” (Giora 2003: 172), it becomes transparent that jokes capitalising on salience need not immanently entail covert ambiguity of the co-text (setup) and backtracking, even if they do cause surprise and even violate expectations, in accordance with the red-light mechanism. The author indicates, arguably unbeknown to herself, that there are jokes which do not conform to the pattern she describes in the bulk of her work. Furthermore, Giora’s model will not account for all GPs. Discussing salience imbalance, Giora (1991, 2003) focuses exclusively on jokes reliant on ambiguous lexemes (some of which are here associated with stereotypical presuppositions), which she regards as “semantic ambiguity” (Giora 1991: 466). Giora (1991) openly states that her study does not obtain for jokes residing in “pragmatic expectations”, here deemed as pragmatic ambiguity (see chapter 4), not explaining this heuristic decision. The reason, however, will be that the salience-based proposal is simply inapplicable to ambiguities not consequent upon the interpretation of particular ambiguous lexemes. It is here maintained that Giora’s notion of salience is most relevant to the analysis of the first meaning appreciated in the processing of GPs using lexical ambiguity or lexically-based presuppositional ambiguity but not other forms of pragmatic ambiguity, in which case the pragmatic notion of default meaning proves more tenable (see chapter 6). In conclusion, Giora’s model for joke interpretation is most plausible as long as its scope is explicitly narrowed down to GP jokes centring on lexical ambiguity of the set-up, with no theses being propounded for other types of jokes. As such, it cannot be regarded as an adequate model for jokes revolving around the red-light or crossroads mechanisms.

Humour Proposals

113

8. Forced reinterpretation One of Ritchie’s foci of interest within his humour research largely coincides with the GP mechanism analysed here8. Forced reinterpretation: The set-up has two different interpretations, but one is much more obvious to the audience, who does not become aware of the other meaning. The meaning of the punchline conflicts with this obvious interpretation, but is compatible with, and even evokes, the other, hitherto hidden, meaning. The meaning of the punchline can be integrated with the hidden meaning to form a consistent interpretation which differs from the first obvious interpretation. (Ritchie 2004: 56)

Initially, Ritchie dubs his model for joke interpretation sudden disambiguation (SD) (1999) but later refers to it as forced reinterpretation (FR) (2002, 2004, 2006). The change of terminology is motivated by the fact that not all jokes in focus are rooted in “linguistic meanings”, and hence “linguistic ambiguity” (see chapter 4). Some entail alternative interpretations of whole sequences of events (e.g. the famous doctorpatient joke discussed by Raskin). It is here argued that the latter case can also be regarded as based on ambiguity (of pragmatic nature). Ritchie (1999, 2002, 2004) subscribes to the widely supported division of a joke into the set-up (here, the lead-up) and the punchline (here switch). Admittedly, this is not the most fortunate terminology, insofar as Ritchie analyses not only fully-fledged jokes but also one-liners and riddles. He, however, allows for the fact that the punchline, contrary to what the term suggests, may constitute only one word. Within the FR model, jokes exploit two interpretations within the setup, only one of which is available to the audience, thereby causing a misinterpretation, whose revelation comes with the punchline. Obviously, the model excludes from its scope those jokes which do not operate on such a misinterpretation. Ritchie (1999, 2002, 2004)9 distinguishes four elements of the text crucial for the SD/FR account: the first more obvious interpretation of the set-up, the second less obvious interpretation of the set-up, the meaning of the punchline, and the interpretation formed by 8

I feel obliged to explain that I developed interest in GP humour and started working on my book a year before I familiarised myself with Graeme Ritchie’s work. Needless to say, my theoretical approach is considerably different to Ritchie’s, as regards both the humorous material under investigation and the methodology applied. 9 In his 1999 article, Ritchie does not explicitly appreciate the interpretation integrating the hidden meaning and the punchline.

114

Chapter Three

integrating the meaning of the punchline with the hidden meaning of the set-up. Ritchie also highlights two features inherent to either of the two set-up interpretations. The quality of obviousness is immanent in the nature of the first interpretation. Inappropriateness is ascribed to the second meaning (Ritchie 1999) and later (Ritchie 2004) also to the final interpretation resulting from the combination of the punchline and the second (non-obvious) meaning of the set-up. The quality of inappropriateness (oddness, eccentricity, preposterousness or taboo) means that the integrated meaning must immanently be absurd or taboo, depending on whether it violates the norms of everyday logic or whether it breaches the norms of socially acceptable discourse, respectively (Ritchie 2004). The notion of inappropriateness appears to raise some doubts, as humorous examples (especially witticisms) that are neither taboo nor absurd can also be found, which is why the parameter is here found nonobligatory. Ritchie (1999, 2002, 2004) proposes a list of three relationships, which he believes to have so far been embraced, somewhat confusingly, under the notion of incongruity. Unfortunately, the author provides little evidence for this claim. Essentially, how incongruity is understood relies on the form of humour analysed (cf. chapter 2). In the case of jokes, in accordance with Suls’s model, incongruity is regarded as the relation between the set-up and the punchline. It cannot be denied, however, that the relationships distinguished by Ritchie are most useful in the analysis of GP humour, even if not all of them can be found in other humorous forms, in which incongruity suffices for adequate description. Firstly, the interpretation of the punchline and the obvious interpretation of the set-up are in the relation of conflict, which creates the need for the reinterpretation of the obvious meaning. Conflict will often entail discourse incoherence or a failure of presupposition, but it should by no means be deemed as a semantic clash or logical inconsistency (Ritchie 2002). Ritchie’s conflict corresponds to Suls’s (expectation-violating) incongruity coming with the punchline. Ritchie (1999) also distinguishes two subtypes of the conflict relation, i.e. one setting the punchline against the more obvious interpretation of the set-up, and the other against the predicted interpretation of the set-up. Secondly, the two alternative meanings of the set-up are in the relation of contrast. Ritchie (1999) calls this relationship comparison, understood as a contrastive relationship or even a clash. It is transparent that, given the nature of the relation, the term “contrast” might be preferable. Also, contrast exists between the first interpretation and the interpretation formed by integrating the meaning of the punchline and the initially hidden meaning of the set-up. Thirdly, the punchline is discourse

Humour Proposals

115

coherent and is in the relation of compatibility with the less obvious interpretation of the set-up. All this appears to be most feasible. However, Ritchie (1999) admits that it is not always so that all the criteria need to be met for a text to be humorous. Additionally, the contrast and conflict as well as the obviousness of the first interpretation are gradable parameters (Ritchie 2004). Ritchie (2004) distinguishes two classes of FR jokes. One class, dubbed discourse-coherence misunderstanding, is contingent on linguistic ambiguity. The punchline is coherent with the less obvious interpretation and incoherent with the initially most obvious interpretation. In other words, “CONFLICT consists of lack in discourse coherence, and COMPATIBILITY is discourse coherence” (Ritchie 2004: 96). To illustrate this class, Ritchie provides examples of short jokes and riddles, most of which hinge on what is here called lexical ambiguity. However, a few of the instances may indeed violate discourse coherence but manifest pragmatic, rather than lexical, ambiguity, e.g. “Diner: Waiter there is a fly in my soup” “Waiter: Please don’t shout so loudly – everyone will want one.” (Ritchie 2004: 42). On the other hand, a different subclass covers punchline revision jokes, which may be realised with neither strictly linguistic ambiguity nor discourse incoherence (Ritchie 2004). The punchline appears to be logical and coherent with the set-up but implies the less obvious interpretation. In Ritchie’s (2004: 97) own words, “some of the information supplied by the punchline would, if supplied earlier in the text, have made the less OBVIOUS interpretation to be the more OBVIOUS one.” On the basis of the two long canned jokes quoted (Ritchie 2004: 96-97), it can be gathered that in this subtype of the FR joke, the punchline introduces new information that sheds new light on an earlier event, causing its reinterpretation. This is exactly the type of mechanism that can be observed in the few long jokes based on the GP mechanism of the nonpunning type presented in chapter 1. Such jokes centre on complex contextual factors and thus pragmatic ambiguities originating from situational defaults. This mechanism will, however, not obtain in the case of GP witticisms and one-liners, due to their relatively short and simple text format. On the whole, some of Ritchie’s (1999, 2002, 2004) main arguments, i.e. the basic textual structure, together with its elements as well as the relationships between them, would be most adequate to the analysis of GPs. Unfortunately, the FR model leaves an unexplained lacuna, the primary reasons being that the notion of obviousness and the nature of

116

Chapter Three

ambiguity conducive to the mechanism are not explained. These problems will be addressed here.

9. Conclusion and comments on GPs The overview of several models of joke analysis was meant to present the diversity of heuristic apparatuses authors employ to analyse jokes and one-liners. It is also evident that while some of the models hold for (virtually) all jokes, others fail to do thus. Primarily, a few of the models capture only texts grounded in the GP mechanism, as if no other types existed (with the exception of Ritchie who explicitly focuses on one mechanism). The definition of the GP mechanism provided in the first chapter should now be extended. It is here argued that the GP mechanism can be subsumed under the broadly conceptualised incongruity-resolution model. The distinctive feature of a GP is that it necessarily manifests covert ambiguity of the lead-up, which causes misinterpretation emergent at the stage of the switch. Accordingly, the meaning (i.e. inference or interpretation) of an initially unambiguous lead-up is rendered incongruous with the final part of the textual material, i.e. the switch. As a result, the ambiguity of the lead-up emerges and the first salient/default interpretation must be abandoned in favour of another, earlier unobserved, thereby resolving the incongruity. It is also noteworthy that the two senses, salient/default and nonsalient/non-default, of the ambiguous element/chunk emergent consecutively are in the relation of contrast and necessarily mutually exclusive. What is of utmost importance, the switch is responsible for the emergence of ambiguity, the introduction of incongruity and for prompting its resolution, i.e. causing the interpreter to seek a cognitive rule to render the incongruity somehow congruous. To reformulate, the ambiguity is covert until the switch emerges as incongruous with it. It is only then that the recipient realises that the lead-up is ambiguous and affords another interpretation. The emergent query is what governs such a process of inference, i.e. on what grounds interpreters will let themselves be led up the garden path and why they choose “wrong” interpretations. To explain the phenomenon, GPs need to be dichotomised into those capitalising on lexical (and syntactic) and pragmatic ambiguities. The discussions on the two GP types and their subrealisations will be conducted in the next chapter.

CHAPTER FOUR AMBIGUITY-BASED TYPES OF GPS

This chapter aims to account for the most frequent realisations of GPs, depending on the nature of their underpinning ambiguity. Their two main types are distinguished, i.e. GPs anchored in surface-level linguistic ambiguity, mostly lexical (rarely coupled with syntactic), coinciding with puns, and non-linguistic GPs operating on pragmatic ambiguity of the lead-up. Several second-order mechanisms within either of the categories will also be discussed. Beforehand, however, a number of relevant humour-related concepts and dichotomies will be expounded on for the sake of terminological clarity.

1. (Non-)verbal and (non-)linguistic humour This section aims to discuss the primary dichotomies within humour, i.e. humour conveyed by language1 vs. humour communicated otherwise; as well as humour dependent on the use of a particular linguistic expression manifesting itself in wordplay (e.g. puns, neological blends, alliteration, etc.) vs. humour communicated by means of language but not determined by a rigid choice of words. Many authors have already discussed these rudimentary humour categorisations, being hardly unanimous in their choice of nomenclature and the premises for the bifurcation. Cicero is reported (see e.g. Attardo 1994) to have first addressed the classificatory issue distinguishing between two categories of humour, or rather the comical, the term prevailing in aesthetics (Dziemidok 1993). In Cicero’s view, humour can be of two kinds, i.e. de dicto (“of what is said”) and in re (“of the thing”), each of which can be further categorised. Humour de dicto is believed to comprise: ambiguity, paronomasia, false etymologies, proverbs, literal interpretation of figurative expressions, 1

Language is here understood as a system of words, sounds and written symbols used by people, and does not embrace forms of non-verbal communication, such as gestures or the so-called non-verbal aspects of speech, e.g. grunting.

118

Chapter Four

metaphors, antiphrasis, and irony. Humour in re includes anecdotes and caricature (Attardo 1994). Such a division is hardly flawless, as the demarcation line running between the two types of humour is ill-defined. For example, anecdotes may actually rely on the funniness of their rendition rather than on the stories related, while the humorousness of ambiguity may not reside in the use of a particular word as such but the twofold inferences drawn on the basis of the verbalisation. By the same token, the humorousness of a metaphor usually centres on the concepts evoked rather than the linguistic expressions used. In essence, but for the anecdote, humour in re would embrace humorous stimuli conveyed by means other than the use of language, the latter being typical of humour de dicto, whether or not contingent on the application of particular words and wordplay, Cicero’s intended criterion notwithstanding (see the next section). These reservations appear to have remained unnoticed and the two categories have formed the foundations for other authors concerned with humour who (mistakenly) refer to Cicero’s dichotomy, differentiating between humour pivoting on wordplay and humour devoid of it. Moreover, they apply different nomenclature, which effects epistemological chaos. Freud (1905) is known to have distinguished between Wortwitz, hinged on words, and Gedankenwitz, based on thought, thus differentiating between verbal and conceptual jokes. The former category relies on a word (or words) which cannot be changed, while the latter is rooted in an idea which, when conveyed, can take a variety of verbal forms. Another distinction made along the same criteria is referential vs. semantic humour (Morin 1966). Hockett (1972/1977) advocates poetic and prosaic jokes, drawing a parallel between literature and jokes and regarding the latter as a subtype of the former. Poetic jokes are those revolving around accidental resemblances between words in sound and meaning and are unlikely to be translatable. Also, Norrick (1986) uses Hockett’s terms in his analysis of verbal humour, i.e. humour produced by means of language. The most prevalent terminology referring to the same dichotomy is verbal vs. referential humour/jokes (Milner 1972, Todorov 1976; Pepicello and Green 1984; Attardo 1994, 1996a, 2001; Attardo et al. 1994 inter alia), which may again provoke misgivings. Both verbal and referential jokes must be conveyed by language. Jokes of the latter type are grounded solely in the meaning of the text and do not make any reference to the surface realisation of its elements, while verbal jokes “in addition to being based on the meaning of the elements of the text, make reference to the phonological realization of the text” (Attardo 1994: 95). It

Ambiguity-Based Types of GPs

119

is strange that the author should favour such terminology, given that he is the co-author of the General Theory of Verbal Humour (Attardo and Raskin 1991). As already discussed in chapter 3, the theory is devoted to the analysis of humour whose vehicle is language but not necessarily punning or and other wordplay. Thus, the theory of verbal humour covers both verbal and referential jokes. The fact that one qualifier “verbal” is used to denote two humour-related concepts, one subordinate to the other, gives rise to terminological confusion. The name of the theory should probably be credited to Raskin, who is a supporter of different terminology. Raskin (1985, 1987) defines verbal humour as a jokecarrying text while non-verbal humour is a humorous situation not created, described or expressed by any text. However, not only jokes but also other humorous forms (e.g. teasing or neologisms) need to be included in the definition. Admittedly to circumvent this problem, Attardo (1994: 96) introduces yet another concept of verbalised humour, i.e. “expressed by means of a linguistic system”, which is synonymous to the term “verbal humour” used in the broad sense. Similarly, Ritchie (2000, 2004) advances the notion of verbally expressed humour, while Suls (1983), Raskin (1985), Alexander (1997) and Norrick (1985, 2004) refer to humour conveyed by dint of language/text as “verbal”. The term “verbal humour” is also used here. The etymology of the term “verbal” goes back to the Latin word “verbum” (a “word”) and so the adjectival form means “connected to words and their use”. The epithet “verbal” is normally regarded as the opposite of “non-verbal”, i.e. “not conveyed by means of words”. The latter term is used, for instance, in social studies, psychology and anthropology in reference to body language, e.g. in collocations “nonverbal communication” or “non-verbal aspects of speech”. It is noteworthy that if the term “verbal” is applied to humour in Attardo’s narrow sense (in reference to wordplay), a question arises as to how one can differentiate between humour conveyed via language and via non-verbal communication. Therefore, humour not produced by language but, for example, by grimaces, pictures or sounds should best be labelled as non-verbal humour, further divided into visual, musical and kinetic subtypes. As Norrick (2004) rightly asserts, non-verbal humour includes slapstick, cartoons and even canned jokes which are heavily reliant on the joke-teller’s rendition such as his/her body language, accent and other paralinguistic features, e.g. the tone of voice or stammering. In addition, even if a non-verbal stimulus (e.g. a picture) is accompanied by a text but humour does not directly result from the latter, it is still a case of non-verbal humour. By contrast, verbal humour, which is the focus here, is understood as humour

120

Chapter Four

conveyed by language in the form of a (short) text, either written or spoken. Consequently, following the footsteps of Raskin (1987)2 and Alexander (1997), a distinction is here drawn between linguistic verbal humour and non-linguistic verbal humour. In contrast to the latter, the former is dependent on wordplay or what Nilsen and Nilsen’s (1978) dub language play. Linguistic humour covers primarily punning (puns), in which similar or identical structural features (homophony or paronymy, homonymy, homography and syntactic ambiguity) are exploited for the confrontation of two different meanings. Secondly, linguistic humour will include other non-punning types of wordplay such as unusual combinations of sounds, e.g. alliteration or rhyming (cf. Nash 1985, Alexander 1997, Nilsen and Nilsen 1978, Crystal 1998, Apte 1985), as well as single words which are deliberately coined, or rather malformed morphologically or phonologically (Buttler 1968). Wordplay is the result of a conscious metalinguistic action on the part of the speaker, who takes advantage of certain linguistic features for the sake of obtaining humorous effects. It should be mentioned that the present understanding of wordplay does not tally with what some authors advocate. By way of illustration, Norrick (1994) perceives wordplay as a category comprising conversational punning, hyperbole and allusion, while Chiaro (1992) defines wordplay as various conceits including puns, spoonerisms, wisecracks and funny stories. Essentially, within the latter approach, wordplay is any humour conveyed by means of words, irrespective of whether it derives from play on the formal structure of the words used. Many researchers tend to equate wordplay with punning (e.g. Delabastita 1997), thereby narrowing down the scope of the former. Punning is, however, the most prominent type of wordplay and tends to be the focus of most linguistic humour analyses. The present work also demonstrates this tendency.

1.1. Criteria for the linguistic vs. non-linguistic distinction The dichotomy between linguistic (or verbal) and non-linguistic (or referential/prosaic) humour is proposed along various criteria. The one supported here is simply that while the former inherently centres on various forms of wordplay (punning, alliteration, creative word2

Having put forward this distinction between “linguistic” and “non-linguistic” humour, Raskin (1987) rejects the necessity for the dichotomy, as his SSTH obtains for both the categories.

Ambiguity-Based Types of GPs

121

formation), the latter does not. In modern studies, the differentiation parameters for the dichotomy are three, i.e. translatability, transformation (paraphrase/substitution) and ambiguity. Their premises are briefly recapitulated here, the intention being to substantiate that none of them is unquestionable. Paraphrase or transformation is a criterion with a very long tradition, first advanced by Cicero. Resistance to paraphrase, i.e. rewording, serves as an empirically verifiable criterion determining whether a humorous text is verbal (here linguistic) or referential (non-linguistic). What, said in whatever words, is nevertheless funny, it is contained in the thing, what loses its saltiness if the words are changed, has all the funniness in the words. (…) because after changing the words they cannot retain the same funniness, should be considered to rely not in the thing but in the words. (translation of LXII, 252 and LXIV, 258 in Attardo 1994: 28)

A reservation must be made that if a linguistically humorous text is a longer one, it can undergo changes in the structure or in particular words, even if the humorous stimulus is linguistic. Crucial are only the word/words yielding wordplay, which the original proposal fails to emphasise. Also, as Attardo et al. (1994) argue, sometimes wordplay can be deleted, while the text retains its humorous potential. Even if Cicero’s bisection of humour types is vague, it seems to have found many supporters, also as regards the criterion. For example, Buttler (1968) equates the parameter of transformation with substitution of a lexeme for a different one. This reformulation may be considered to be an improvement, because it embraces not only canned jokes but also shorter humorous chunks, even single humorous lexemes. The substitution of lexemes for their synonyms in non-linguistic humour should not render a text non-humorous. It is only the content matter of the verbalisation that purports to be essential and, therefore, its surface structure is subject to alteration. Conversely, the pivot of linguistic humour cannot be substituted by a different word or by different words (Buttler 1968). One drawback of this approach is that it may prove unverifiable, as many words simply lack synonyms. Among others, Attardo (1994) perceives paraphrase as equivalent to endolinguistic translation, as opposed to translation per se, which is interlinguistic. Liability to translation is yet another parameter (Hockett 1972/1977). Referential (here non-linguistic) jokes can be translated interlinguistically and intersemiotically, while the translation of verbal (here linguistic) jokes is impossible or its translation is devoid of a humorous edge. (Un)translatability can thus serve as an acid test for this

122

Chapter Four

dichotomisation of humour, as two languages hardly ever manifest congruities in wordplay. The first reservation against this approach is that it is not always linguistic but cultural constraints, such as customs and etiquette, that may inhibit the translation of a jocular chunk, which is then not linguistic but still untranslatable. Such a joke will be linguistically translatable but will not be understood by recipients unfamiliar with a given cultural element. More importantly, compatibilities between languages do exist, even if rarely, facilitating translation of wordplay from one language to another with similar wordplay (Chiaro 1992), hence proving the parameter fallible. In addition, linguistic humour happens to be successfully rendered in the target language, even if with considerably different linguistic means (functional equivalents), retaining source text’s jocularity and so its perlocutionary effect (Chiaro 1992). The original and translated texts’ equivalence is, in essence, the most significant feature of successful translation (e.g. Toury 1980, Neubert and Shreve 1992). Therefore, the term “translatability” ought to be meticulously elucidated to render the parameter more reliable. The last criterion is simply linguistic ambiguity of meaning (Oaks 1994, Attardo et al. 1994). Such ambiguity can emerge at syntactic, lexical or phonological levels (cf. Chiaro 1992, Alexander 1997, Attardo 1994). What is of crucial importance is the acknowledgement that a distinctly different category of joke, and humour in general, is that dependent on pragmatic ambiguity. Plausible as this theory may appear, the criterion of linguistic ambiguity would not account for types of humour centred on language play other than puns. It is, however, sufficient for the present approach, given that it covers the concept of a pun couched in ambiguity and not any other play on words. A doubt persists as to whether the dichotomy between linguistic and non-linguistic humour is significant, since the two categories share certain features. All forms of verbal humour are conveyed by means of language and the language comprehension process determines their being appreciated. In other words, the production and, likewise, the interpretation of verbal humour is somehow enabled by the formal structure. Secondly, humour, even of the linguistic type, can never be asemantic and must have some propositional content besides the specified surface form, which is essential for its humorous potential. Discussing similar issues, Raskin (1987) contends that linguistic humour (coinciding with puns) and non-linguistic humour are both legitimate in topical studies but adds that the distinction is unnecessary. He also notes that surfacestructure ambiguity is not a sine qua non for humour, while the occurrence of such ambiguity does not immanently produce humour. Consequently,

Ambiguity-Based Types of GPs

123

the proponents of the differentiation do not strive to seek the mechanisms of humour, which must be of the non-linguistic nature, but are engrossed in “amorphous and uninspired description of just one or very few words from the text as ambiguous” (Raskin 1987: 13). Indeed, if analysts focus solely on the surface structure phenomena and discuss inexhaustible lists of sub-realisations, they engage themselves in endless discussions on puns, without accounting for the underlying humorous mechanisms. Nevertheless, it is here argued that the division of verbal humour into the linguistic and non-linguistic types can offer a basis for more insightful discussions on GPs. Those hinge on either linguistic or non-linguistic humour grounded in linguistic ambiguity and pragmatic ambiguity, respectively.

2. Ambiguity By nature, human communication processes are geared towards reducing ambiguities, which are resolved instantly, to the effect that language users are often oblivious to them. Mey (1993: 7) even ventures a claim that in “real life – that is, among real language users – there is no such thing as ambiguity.” This may be a far-fetched conclusion, but it cannot be denied that, on many occasions, disambiguation occurs automatically, so the ambiguity is not even consciously observed (cf. e.g. Hirst 1987, Frazier and Rayner 1982). This is why ambiguity in GP humour is not instantly recognised but emerges in on-line processing, with only the obvious, i.e. salient/default, interpretation being appreciated the moment the ambiguous element emerges. In other words, the GP ambiguity is conducive to two competitive meanings acknowledged sequentially in the on-line process of comprehension. Essentially, this automatic but necessarily defeasible process of ambiguity resolution is the foundation for GP humour. Lyons (1981) aptly notes that scholars often labour under an erroneous assumption that the permanent aim of communicators is to avoid ambiguity for the sake of clarity and no equivocation of meaning, deeming ambiguity as malpractice. Most plausible is the postulate that communication is not always oriented towards reducing multiple meanings (Nerlich and Clarke 2001), which is because purposefully formed ambiguity, i.e. purposive ambiguity, produces crucial semantic, pragmatic and conversational benefits (Kittay 1987), among which humorous effects are prominent. Ambiguity can be employed for rhetorical effects (Hirst 1987), such as humour, which entails a switch from one sense to another or the simultaneous appreciation of two senses. The interpreter needs to

124

Chapter Four

perceive the two competitive meanings, as envisaged by the speaker, whether simultaneously or consecutively. On the other hand, there will also be jokes, not to mention other humorous forms, devoid of ambiguity but only carrying, surprising/incongruous information. Semantic ambiguity (i.e. ambiguity of meaning) is central to GP humour and is immanently latent/covert to the hearer until the switch, which corresponds to the final stage of on-line interpretation. In particular, the lead-up, either as a whole or its element, yields (prototypically) two interpretations, only one of which is initially available to the hearer until the meaning of the switch, incongruous with the first interpretation of the lead-up, brings out the ambiguity and necessitates the retrospective foregrounding of a second interpretation. Contrary to the common conceptualisation of ambiguity as immanently reliant on the surface structure, ambiguity need not coincide with linguistic humour, and hence puns consequent upon phenomena such as homonymy or polysemy. Accordingly, lexical ambiguity (based primarily on polysemy and homonymy, sometimes also coupled with syntactic ambiguity) and pragmatic ambiguity will be distinguished here as the sources of GPs. In other words, two major categories of GPs fall within the scope of linguistic and non-linguistic humour, respectively.

2.1. Ambiguity vis-à-vis other phenomena Ambiguity invariably entails the multiplicity of semantic meanings. For the sake of simplification, the term ambiguity will pertain to the prototypical case when two competing interpretations come into play, although more than two are also theoretically possible. Ambiguity is often mistaken for other phenomena lacking in clarity of expression, such as vagueness, relativity, indexicality, nonliterality, indirection and inexplicitness (Bach 1998a). Nonliterality, indirectness and inexplicitness refer to the meant which is not uniquely determined by the meaning of the words used and gives rise to unclear communicative meaning. The three phenomena are confused with ambiguity, admittedly because in folk terminology non-literal, indirect and inexplicit verbalisations are called “ambiguous”. Of all the false friends of ambiguity, vagueness appears to be the most problematic. Miscomprehension of the terms “ambiguity” vs. “vagueness” tends to provoke epistemological convolution in literature. Sometimes vagueness is equated with semantic ambiguity, since in both cases, word denotations are not well defined. Channel (1994) reports that the first rigorous formulation of the definition of vagueness is attributed to Charles

Ambiguity-Based Types of GPs

125

Peirce. In his view, a statement is vague when it is intrinsically uncertain which of the alternative states of things it allows for and which it excludes. According to Pinkal (1995), ambiguity and vagueness can be perceived as two distinct forms of “indefiniteness”. A sentence is semantically indefinite if, in certain situations, neither “true” nor “false” can be assigned to its truth value. Ambiguity and vagueness can, however, be distinguished depending on whether two specifications are formed, or whether the specification manifests itself as a continuum, respectively. In accordance, the major distinction between ambiguity and vagueness is that the former has two or more distinct meanings (Zhang 1998),3 whereas the latter has more than one meaning which cannot be identified, e.g. “a few” or “approximately” (Channel 1994). In essence, a word is ambiguous in a neutral context and has two dictionary entries, whether or not related in origin.

3. Ambiguity types The notion of ambiguity has been approached from various angles. Many studies have been conducted within psycholinguistics and computational linguistics (e.g. Hirst 1987, Franz 1996, Poesio 1996, Gillon 1990). The two main types of ambiguity which can be distinguished on the basis of several categorisations are lexical ambiguity and structural/syntactic ambiguity (e.g. Hirst 1987, Bach 1998a), both of which are encompassed by the term semantic ambiguity (cf. Poesio 1996). Lexical ambiguity is inherent to a lexical item with two meanings dependent primarily on polysemous or homonymous strings/words. Structural ambiguity occurs when a phrase or a sentence has more than one underlying structure and can accommodate more than one structural analysis, in result of which it has more than one meaning (Gillon 1990). Syntactic ambiguity can be subdivided into a number of types, among which notable is that of class ambiguity created by confusion between different parts of speech (Stageberg 1971). Syntactic ambiguity originates also from ambiguous word order, referential ambiguity or prepositional phrase attachment (Hirst 1987, Gibson and Pearlmutter 1994). Oaks (1994) appreciates the fact that class ambiguity, a type of syntactic ambiguity, will also entail lexical ambiguity but not vice versa. On the other hand, MacDonald et al. (1994) observe that the two main ambiguity 3

However, less plausible is Zhang’s postulate that ambiguity does not cover multifold interpretations which are semantically related, i.e. polysemy, which he refers to as vagueness.

126

Chapter Four

types are closely related and that syntactic ambiguity can be regarded as a type of lexical ambiguity, which is also given support here. For the present research, the most significant are obviously the studies on ambiguity within humour literature. A variety of detailed categorisations are by no means compatible and many appear to raise serious doubts as regards consistence in the choice of categorising criteria. For example, Green and Pepicello (1978) and Pepicello and Weisberg (1984) perceive linguistic ambiguity as grammatical ambiguity manifesting itself on various levels, viz. phonological (including lexical), morphological (e.g. based on irregular morphology) and syntactic (e.g. idiom). Other queries aside, it is particularly difficult to understand the authors’ rationale for considering idioms as syntactically ambiguous. Following a similar train of thought, Alexander (1997) presents linguistic ambiguity types as associated with language levels, i.e. the phonological level (homophony, paronymy), the morphological level (derived and compound words), the lexical level (polysemy) and the syntactic level (ambiguous structures). Here, a question arises why the main manifestation of lexical-level ambiguity is only polysemy and whether this category does not overlap with, or even include, other types distinguished. In his succinct discussion on ambiguous triggers in jokes, which can be equated with the punning element, Raskin (1985) mentions regular ambiguity (e.g. “a gentleman”: “a man”/“a well-behaved man”) and figurative ambiguity (e.g. “catch”: “get hold of”/“become infected”). Admittedly, it is hardly possible to differentiate between the two categories in the light of Raskin’s examples. Raskin proposes the third category, i.e. quasi-ambiguity in reference to purely “phonetical and not semantical relations between words” (Raskin 1985: 116), which lays the foundations for knock-knock jokes. This type is usually dubbed “phonetic”. The last category is that of situational ambiguity, e.g. “My wife used to play the violin a lot but after we had kids she has had not much time for that.” “Children are a comfort, aren’t they?” (Raskin 1985: 55), where two interpretations of one situation exist. This is transparently the case of what is here called “pragmatic ambiguity” dependent on the context. Clearly, it is not the trigger “comfort” alone that is ambiguous. The ambiguity is of pragmatic nature and emerges from the twofold inference generated on the basis of the response in the context of the preceding part of the text. On the whole, the sources of ambiguity Raskin (1985) observes are perhaps the ones which generate the greatest number of humorous instances, but their names and conceptualisations are doubtprovoking in the light of literature on ambiguity.

Ambiguity-Based Types of GPs

127

By contrast, analysing verbal jokes (here dubbed linguistic jokes), Attardo et al. (1994: 28) present phonetic ambiguity as superordinate to other realisations of ambiguity. This categorisation does not appear to be entirely plausible, either. First of all, in the authors’ approach, phonetic ambiguity corresponds to phenomena such as homonymy, polysemy, homophony and paronymy, which are all conducive to puns. However, while all the four phenomena may indeed reside in same/similar phonetic forms, they are significantly different, while their humorous potential is unfailingly the effect of the meanings juxtaposed. In addition, homonymy and polysemy operate not only on the level of phonetics but also on the graphological level. It could be argued that the phonetic/phonological type of ambiguity can be incorporated by lexical ambiguity, for the juxtaposed lexemes carry divergent meanings. Nota bene, Bucaria (2004) differentiates between syntactic, lexical and phonological ambiguities. What is noteworthy, Bucaria finds only one example of the third phenomenon in her corpus of headlines, which testifies to its infrequency and questions the necessity for introducing the category in some humour studies in the first place. Because of its infrequence (and even a lack thereof) in GPs, phonetic ambiguity (realised via homophony and paronymy, i.e. near homophony) is not here isolated as a separate category, either. Interestingly enough, Attardo et al. (1994: 33-34) substitute the notion of “phonetic ambiguity” with “lexical ambiguity”, irrespective of the earlier categorisation in the same article. Ritchie (2004) attempts to settle the disparities between the categorisations, of which those presented above are just a small sample, speculating that ambiguity is observable when a portion of a text (a word or phrase) has one representation at one level but more than one, and usually two, at a different one. The levels constitute a hierarchy of phonological, morphological, lexical, syntactic, semantic and reasoning levels (Winograd 1983). Ritchie (2004) hence presumes that disagreements among researchers on the classification of particular examples stem from the fact that only one level is considered. For example, puns tend to be regarded as rooted in the ambiguity of phonetic strings (Attardo 1994), although the sources of twofold interpretations can differ significantly. Nevertheless, even Ritchie’s (2004) approach does not entirely clarify the issue, since, as the author himself observes, e.g. “read” and “red” could be phonetic-lexical ambiguity, or phonetic-orthographical ambiguity or even phonetic-morphological ambiguity (Ritchie 2004: 41). Ritchie’s postulate may rectify some problems of other approaches but is not infallible, either. None of the detailed categorisations of ambiguity according to levels, whether one or two, is supported here. Taking into account the nature of

128

Chapter Four

GPs, it appears the most reasonable to accept the dichotomy between lexical and syntactic ambiguities as the two higher-order types of ambiguity, which accommodate other second-order phenomena analysed by humour researchers. The two types can be subsumed under the umbrella term of linguistic ambiguity, given that it is couched in the surface-structure realisation (Alexander 1997, Green and Pepicello 1978, Pepicello and Weisberg 1984). Lexical ambiguity is postulated to be superordinate to syntactic ambiguity (MacDonald et al. 1994), for the latter always coincides with a change of lexical meaning. The second type of ambiguity proposed here is the pragmatic type. It refers to the cases of linguistically unambiguous verbalisations, which are, however, semantically ambiguous, for they yield two different inferences. Obviously, linguistic ambiguity also gives rise to two inferences, but those are prompted by ambiguity at the surface level. Semantic ambiguity 1. Linguistic ambiguity (puns) 1.1. Syntactic ambiguity 1.2. Lexical ambiguity 1.2.1. Polysemy-based ambiguity Lexemes Pragmatic and semantic phrasemes 1.2.2. Homonymy-based ambiguity 2. Pragmatic ambiguity 2.1. Purely inferential ambiguity 2.2. Inferential but linguistically-based ambiguity 2.3. Presuppositional ambiguity The above summarises the categorisation of deliberate semantic ambiguity, i.e. ambiguity of meaning, advocated as significant for the study of GPs. The distinction between pragmatic and linguistic ambiguities corresponds to that between non-linguistic and linguistic humour, respectively. Linguistic ambiguities coincide with puns. In contrast, pragmatic ambiguities are dependent primarily on the inferential path, even if exploiting also linguistic structuring. Because of their infrequence in GPs, linguistic ambiguity types based on homophony and paronymy are not included. It should be highlighted that the terminology and categories of ambiguity are used for heuristic purposes, while all ambiguous verbalisations rely on pragmatic processes, which are instigated by linguistic verbalisations and yield inferences. The workings of both punning (i.e. linguistic) and non-punning (i.e. pragmatic)

Ambiguity-Based Types of GPs

129

ambiguities in GPs can be explained by pragmatic and cognitive theories on language comprehension (see chapters 5 and 6). The ambiguity types distinguished will now be discussed in detail to present the whole gamut of GP realisations.

4. Syntactic ambiguity and GPs Discussing syntactic ambiguity, Oaks (1994) enumerates a number of enablers emerging from class ambiguity (Stageberg 1971), e.g. articles, verbs or conjunctions, whose application facilitates the occurrence of humour. Although syntactic ambiguity does provide a basis for humour yielding syntactic puns, Attardo et al. (1994) appreciate the fact that syntactic ambiguity occurs in humorous texts far less frequently than lexical ambiguity. Also, Coulson et al. (2006) observe that in one-liners, listeners are led up the garden path, which, however, does not entail syntactic reinterpretation. As already signalled in the introductory section on the GP mechanism in chapter 1, a similar conclusion can be drawn from the analysis of the corpus of GPs collected for the purpose of the present study. This is particularly interesting, inasmuch as the phenomenon of the garden-path sentence is immanently contingent on syntactic ambiguity. The few GP cases of syntactic ambiguity found depend simultaneously on lexical ambiguity. The examples can thus be reclassified as lexical-ambiguity GPs, revolving around homonymy and polysemy, also of multi-constituent chunks, e.g. idioms, sayings or proverbs (see section 5.4). This dress looks better on …fire. [adverbial “on”: subject complement, “worn”/prepositional part of subject complement => polysemy of “on”] (this dress looks better when a woman puts it on ĺ this dress [is so ugly] that it looks better when it is burning, i.e. it should be burnt) Get a new car for your spouse. It’ll be a great trade. [“for”: in “get N for N” introduces the beneficiary/the “currency” of exchange => polysemy of “get N for N”] (one should buy a car to please his/her spouse ĺ one should exchange his/her spouse for a new car) I want to pay my income tax with a smile but they demand cash. [“with a smile”: adverbial of manner/instrument => polysemy of “with”, literalisation of “with a smile”]

130

Chapter Four (the speaker wants to pay taxes, finding pleasure in it ĺ the speaker wants to pay taxes only by smiling, smiles being a currency, but cash is demanded) Don’t count your chickens will do it for you. [“chickens”: object of the matrix clause/subject of the subordinate clause, which is the object within the matrix clause => polysemy based on literalisation, an anti-proverb] (a warning, based on the saying, that the hearer should not to jump to conclusions ĺ the speaker distorts the saying, telling the hearer that his/her chickens won’t do something for him/her) Never hit a man with glasses. Use your fist! [“with glasses”: part of the object/adverbial, instrument => polysemy of “with” ] (one should never hit anyone wearing glasses, as is conventionally maintained ĺ one should never hit anyone with the use of glasses but with a fist) If you want to see a comic strip, you should see me in the shower. [“a comic strip”: object (noun phrase)/object and object (noun and verb), complement => homonymy based on decomposition of the compound “comic strip”] (if the hearer wants to see a series of drawings telling a story ĺ if the hearer wants to see a comic undress, he/she should see the speaker, a comic, taking a shower)

The emergent question is why it should be so that garden-path sentences known in grammar studies do not purport to show much humorous potential. The problem is why syntactic ambiguity alone does not produce humorous effects and why it is so infrequently exploited in jocular texts. Attardo et al. (1994) venture a claim that garden-path sentences do not give rise to humorous texts, since they would be too difficult for the audience to process. Also, it could be added that such sentences would be difficult to produce on the spur of the moment, which is sometimes the case of GPs. In addition, another hypothesis may be that the absence of syntactic ambiguity, unless intertwined with lexical ambiguity, in GPs can be explained by the low degree of contrast between the two competitive semantic meanings (e.g. “raced” as past simple vs. passive participle of the verb), which does not suffice to bring about the incongruity of the lead-up and the switch. Semantic incongruity is, however, one of the indispensable conditions for GP humour. In conclusion, the few cases of GPs capitalising on syntactic ambiguity entail lexical ambiguity, which is why the former ambiguity type can be

Ambiguity-Based Types of GPs

131

subsumed under the latter. Therefore, the terms “linguistic” and “lexical” ambiguity/GPs, albeit not synonymous, can be used interchangeably.

5. Lexical ambiguity and punning GPs Linguistic/lexical GPs making use of linguistic/lexical ambiguity at the lead-up stage can be perceived as a peculiar type of pun. A pun is here defined as a humorous verbalisation which has (prototypically) two interpretations couched in purposeful ambiguity of a word or a string of words, dubbed the punning element manifesting itself in one formal realisation (or two very similar ones4) but conveying two different meanings. In the case of a GP pun, the first and, initially, the only meaning of the punning element (and the neighbouring lead-up material) generated in on-line processing is invalidated by the switch, which brings out the second, so far unobserved, meaning of the ambiguous element to yield one coherent interpretation of the whole utterance. The ambiguity of the punning element originates from various linguistic phenomena such as homonymy, polysemy, and homophony, together with paronymy. Since no GP example was found representing either of the last two phenomena, those are omitted in discussions. The two primary sources of lexical ambiguity in GPs are polysemy and homonymy, which need to be discussed in detail. Also, it is necessary to highlight several subordinate semantic mechanisms typical of covert lexical ambiguity recurring in GPs.

5.1. Puns The phenomenon of punning is hinged on an amusing use of an ambiguous word or phraseme carrying, prototypically, two different meanings, which give rise to semantic ambiguity of the whole verbalisation. Admittedly, one of the thorniest but the most intriguing issues concerning puns is their categorisation, whose result is the abundance of lexico-grammatical classifications (e.g. Brown 1956, Heller 1974, Milner 1972, Oaks 1994, Alexander 1994, cf. Attardo 1994), arguably constituting rather shallow analyses of the surface level. Most commonly, puns are divided into a number of types determined by the linguistic level at which the similarity occurs, i.e. homonymy, polysemy, 4

Mere similarity of sounds yields imperfect puns (Zwicky and Zwicky 1986, Hempelmann 2004). The present analysis concentrates on perfect puns, which intrinsically entail lexical ambiguity.

132

Chapter Four

homophony, homography, morphemic attraction and paronymy, which correspond to the subdivisions of linguistics (Alexander 1997, Attardo 1994). On the whole, Attardo (1994) distinguishes four types of categorisations depending on the central criteria, i.e. the linguistic phenomenon, linguistic structure, phonemic distance, and the eclectic type combining all the criteria. Attardo (1994)5 rightly observes that none of the widely recognised classifications is flawless, insofar as none exhaustively covers all the potential sources of ambiguity. This heuristic failure is due to the heterogeneity of puns. The present approach concentrates only on one punning category, i.e. GPs, making no claims about other types manifesting different cognitive processes (cf. Dynel forth.).

5.2. The basic mechanism of garden-path puns The GP type of pun pivots on the lead-up couched in the ambiguity of one verbal element with two alternative and mutually exclusive meanings, which are introduced sequentially in the text. It also relies on the abrupt invalidation of the first most salient and accessible interpretation of an ambiguous element, and hence of the lead-up, in favour of a less salient one emergent in on-line processing. In other words, the interpreter initially activates only the salient meaning of the punning element, supported by the neighbouring textual material present in the lead-up. However, the next/final chunk of the verbalisation, i.e. the switch, is incongruous with the earlier interpretation of the punning element within the lead-up, so far deemed as monosemic. The lead-up is immediately backtracked and reanalysed as being ambiguous, i.e. as having another meaning. As a result, the second less salient meaning is acknowledged, provoking abrupt rejection of the first most salient and accessible interpretation to yield one coherent interpretation of the whole utterance. The GP pun appears to be perfectly captured within the IDM (Attardo 1994, Attardo et al. 1994) as the distinct connector configuration (cf. section 5, chapter 3). GP puns inherently contain the connector, i.e. the ambiguous element, and the disjunctor causing a passage from one interpretation to another. A reservation needs to be made that only one (the more salient) reading of the ambiguous chunk will initially emerge, which is why backtracking necessitates the activation of the hitherto latent meaning.

5

All the same, the author appears to be partial towards structural taxonomy due to its higher-level generality.

Ambiguity-Based Types of GPs

133

It is also worth noting that the structuring of puns will depend, even if not heavily, on the nature of occurrence of the punning element in the surface structure. The ambiguous element can appear in the verbalisation either once or twice, producing paradigmatic (vertical) or syntagmatic (horizontal) puns, respectively (e.g. Milner 1972, Sherzer 1978, Hausmann 1974, Ritchie 2004). In the former, the punning element occurs only in one portion of the text, while the latter shows the relation of contiguity, with the punning element appearing in the text twice. Specifically, a syntagmatic (horizontal) pun contains two sound strings which are phonetically similar or identical but have different meanings. By contrast, a paradigmatic pun is couched in one surface form which has two meanings, while the unspoken element (whether identical or similar) must be recognisable, given that it is contextually linked or groups separate units (Ritchie 2004). GP puns will manifest themselves in both the types, but the paradigmatic type will statistically be in majority. I’m on a seafood diet. I see food and I eat it. [horizontal/syntagmatic] (the speaker eats only shrimps, cockles, mussels, etc. ĺ the speaker eats any food he/she sees) I don’t suffer from insanity. I enjoy every minute of it. [vertical/paradigmatic] (the speaker is sane ĺ the speaker is not sane but enjoys the state of insanity)

In all linguistic/lexical GPs, the covertly ambiguous punning element introduced in the lead-up will have the second interpretation thrust upon it by the switch, whether or not it includes the ambiguous element in its surface structure. The interpreter conducts an on-line analysis of an utterance which initially seems to be unambiguous until the final chunk of the utterance, viz. the switch, causes him/her to reject the meaning already acknowledged, and, simultaneously, to uncover the ambiguity of the punning element and so its alternative interpretation.

5.3. Polysemy and homonymy Although there are also other pun-inducing phenomena (homophony, paronymy and homography), the two most frequent sources of GP puns are homonymy and polysemy, the latter being a phenomenon with multifarious realisations. According to the prevalent conception, polysemy is a semantic relation between terms which share one form and provenance

134

Chapter Four

and manifest different but necessarily related meanings. A polysemous meaning derives from the lexicalisation of a unit in result of associative processes. This is why polysemous words are interrelated in mental representation and form meaning chains (Taylor 1989). To reformulate, polysemy comes into being through the process of extension of similar meanings rather than historical coincidence, which is the case of homonymy (Sweetser 1990). Distinguishing between homonymy and polysemy of words, with their coinciding phonological realisations, one may also follow the premise advocated by Zhang (1998), based on the property of similarity between lexical items and semes, respectively. If a word corresponds to more than one distinct lexical item and has two different denotations, it is perceived as homonymous. Homonymous words are representations of different lemmas, i.e. lexical units (Levelt 1989). On the other hand, if a word has two meanings because its semes are semantically related, this is a case of polysemy (Zhang 1998). On the whole, two criteria have been proposed to distinguish between polysemy and homonymy, i.e. that of etymological derivation and (un)relatedness of meanings (Lyons 1977). Two different provenances and unrelatedness of meanings are the characteristics of homonymy, while the same provenance and related meanings correspond to polysemy. Homonymy originates from phonetic coincidence, the juxtaposed lexemes bearing no similarity at the semantic level. This is why ambiguity arising thanks to homonymy is sometimes conceptualised at the phonological level, while that couched in polysemy at the lexical level. Both the phenomena can, however, be regarded as yielding lexical ambiguity, i.e. ambiguity of lexical meanings, whether or not related in origin. In addition, as Lyons (1977) observes, neither of the criteria is infallible, because historical derivation may be uncertain, while relatedness vs. unrelatedness of meaning may not act only as two extremes, being more of points on the continuum. Therefore, sometimes two polysemous meanings may be perceived as entirely divergent as if they did not share the same origin and function as distinct lexical entries just as homonyms do (Williams 1992). Despite this provenance-related dichotomisation, which is of utmost importance from a historical diachronic perspective, the two phenomena are of lexical nature, yielding ambiguity of lexemes. Polysemy is a more multifarious phenomenon than homonymy. Polysemous senses originate from innovations concerning primary encyclopaedic knowledge and salient associations, or from already existing polysemies with their own cognitive foundations (Blank 2003). Not all derivational forms of polysemy are significant here but only those which

Ambiguity-Based Types of GPs

135

have retained the alternative meanings in the process of diachronic change. As regards single-word polysemy, the focus is on synchronic polysemy consequent upon lexical semantic changes (Blank 2003), with the exclusion of diachronic polysemy, which yields a single contemporary application of a lexeme, whose provenance and so polysemousness cannot be appreciated by ordinary language users. Even if all polysemies derive from diachronic changes, only those displaying synchronic relations can serve ambiguity-based humorous purposes. This constraint is not pertinent to multi-constituent lexemes and phrases, i.e. phrasemes (Mel’cuk 1995, 1998), carrying only one conventionalised interpretation. Very rarely do two independent idiomatic interpretations of one phraseme exist side by side (e.g. “take something for granted”: “neglect”/“assume as given”). However, phrasemes may be deemed as polysemous if they are also interpreted at the literal level. Phrasemes appear to show varied degrees of proneness to polysemous use reliant upon literalisation and inversely proportional to the (varied) degrees of the literal interpretation’s logicality and communicational applicability. It is not always easy for the literal meaning to take precedence over the idiomatic meaning, cf. “push up the daisies” (e.g. “a mole is pushing up the daisies”) vs. “once in a blue moon”, the latter being far more difficult to decompose (unless there is, for instance, a drawing showing a blue moon). In the case of GP use, the more unexpected the interpretation of the chunk, the better for the humorous surprise. Therefore, the majority of, if not all, multi-word polysemies can be used as the foundation for GP humour. The only prerequisite is that at least one element of the multi-constituent chunk should have a literal meaning which does not directly participate in the construction of the phrasemic sense. This category is amply exemplified and discussed in detail in the next section. What is significant, a polysemous item may come into being thanks to metaphoric or metonymic (and synecdochal) transfer, thereby earning the epithet of figurative language. Metaphors, metonymy and synecdoche are regarded as established senses of polysemous words (Nerlich and Clarke 2001). Blank (2003) concurs that these are the best known forms of polysemy, which are grounded in a salient similarity between two concepts representing distant conceptual domains (metaphors) or contiguity between the domains (metonymies), yielding metaphor polysemy and metonymy polysemy, respectively (Apresjan 1973). Metaphorically motivated polysemy is unconstrained regarding the relatedness of meanings, whose similarity can be bigger (“foot”: “part of the body”/“mountain”) or smaller (e.g. “an eye”: “part of the body”/“part

136

Chapter Four

of a needle”). The metonymic type presupposes contiguity or connectedness between the two meanings (e.g. “chicken”: “a bird”/“bird’s meat”). Nota bene, Apresjan (1973) argues that metaphorically motivated polysemy is closer to homonymy, while metonymy is on the opposite pole. The distinction between two polysemy types can also be approached from a different perspective, depending on whether new senses originate from arbitrary meaning shifts (e.g. “a mouse: “a rodent”/“a computer device”) or from clear semantic relations (e.g. “a book”: “a printed volume”/“its content”). Transparently, the results of this dichotomisation correspond to those generated on the basis of the metonymy vs. polysemy differentiation. Deane (1988) dubs the unpredictable polysemy lexical polysemy, while the predictable type arising from overt application of conventional patterns is called regular polysemy. Rightly, Blank (2003) supports the dichotomy but prefers to rename the types as idiosyncratic and rule-based polysemy, respectively, on the grounds that both the types are lexical by nature. Relevant to GP humour will be only the idiosyncratic or metaphor type, owing to the requisite discrepancy between the two senses, the sine qua non for humorous incongruity emergent at the stage of the switch. Due to the closeness of the senses, regular or metonymy-based polysemy used as a punning element would not manifest sufficient surprising incongruity with the switch. On the whole, various manifestations of lexical homonymy and polysemy can be exploited in GPs. It is also important that in GPs pivoting on lexemes with two legitimate co-existing meanings, it does not matter which of the meanings is, historically, the basic one and which is its development. Their order of their activation (on the first reading of the lead-up or after the switch’s emergence) is governed, however, by their salience (see chapter 6). No, you can’t join me. I’m not coming apart! [polysemy of “join”: “accompany”/“put together”, NB The meaning depends on whether the verb complementation is an inanimate noun or a person] (the speaker does not want to be kept company ĺ the speaker does not want to be put together) When my wife said she was seeing another man, I told her to scrub her eyes. [polysemy of “see”: “perceive”/“meet”; NB The sentence manifests a grammatical mistake, since “see” as “perceive” does not operate with continuous tenses.]

Ambiguity-Based Types of GPs

137

(the speaker’s wife said she was meeting another man ĺ believing her to be experiencing an optical illusion, the speaker told his wife to scrub her eyes) I don’t drink any more but I don’t drink any less either. [polysemy of “any more”: “any longer”/“not more by any degree”] (the speaker has stopped drinking ĺ the speaker drinks the same amount, not more or less) Children would laugh at my son when he started wearing braces on his teeth. They couldn’t understand that otherwise his trousers would fall down. [homonymy of “braces”: “wire frames worn to straighten teeth”/“a pair of elastic cloth bands used to hold trousers up, AmE ‘suspenders’”] (the speaker’s son was laughed at when he started wearing metal frames to straighten his teeth ĺ the speaker’s son was laughed at when he started wearing suspenders on his teeth)

Interestingly, in the case of polysemy, the salient meaning of a verb may depend on whether it is used intransitively or transitively. This phenomenon is exploited in GPs where the intransitive use fosters one meaning to be cancelled when the complementation, coinciding with the switch, emerges. If you go abroad, you must steal …the views with your eyes. [polysemy of a transitive/intransitive verb “steal”: “take without any right”/“get quickly”] (if someone leaves their country, they must take people’s possessions ĺ if someone leaves their country, they must absorb the views)

In the context of the present study, another peculiar category needs to be distinguished, which may be dubbed colloquial shift of meaning with or without semantic overlap. This phenomenon is observable when an already existing word is granted another, entirely unrelated, meaning (e.g. “cheesy”: “smelling of cheese”/“cheap, unpleasant, insincere”, “wicked”: “bad, mischievous”/“super”), or a new, albeit somehow related meaning, whose derivation can be appreciated even by a layman (e.g. “big brown eyes”: “female breasts”, “a birdcage”: “a cell”). The latter case is clearly a manifestation of polysemy, since some semantic overlaps do exist, given that the two denotations share a few characteristics. The other case of colloquial meaning shift entailing a total change of meaning is more problematic. On the one hand, the new term is derived from, or simply coincides formally with, an already existing word but, on the other hand, it has an entirely different, unrelated meaning. Rather intuitively, it could be postulated that this case of shift is the epitome of homonymy, even though

138

Chapter Four

the new word formally emerges from the already existing one, borrowing its form. What speaks in favour of the heuristic step of categorising such forms as homonymous is the fact that the two meanings appear to be unrelated from the standpoint a language user oblivious to the derivation process. After a swilling session, I wake up and crave for an eye-opener, so I look into the mirror. [homonymy via colloquial shift of meaning of “eye-opener”: “an alcoholic drink usually in the morning”/“a shocking revelation, sight”] (having a hangover, the speaker needs do drink some alcohol in the morning ĺ having a hangover, the speaker needs to wake himself/herself up by looking at something shocking, so he/she takes a look in the mirror)

A most peculiar homonymous mechanism will pertain to the humorous use of an acronym normally bearing one conventional meaning but given a new sense in the process of acronym decomposition. This new sense results from “wrong” development of the first letters into words. My boss gave me an IOU…a note saying “I overpaid you.” [IOU: “a statement saying than one owes something to someone”] (the speaker’s manager gave him/her a note saying that he owed him money ĺ the speaker’s manager gave him/her a note saying that he/she had been overpaid)

Yet another phenomenon of the same type is decomposition of a lexeme (also in the form of a compound), which is partitioned and reinterpreted, often contrary to its morphological structure. This entails the attribution of a new meaning to an already existing lexeme, single-word or a compound. To be liable to this humorous process, a lexeme must be comprised of composite homonymous chunks which constitute separate lexemes. Their meanings, different to the meaning of the lexeme they form, are humorously foregrounded by the switch. I got a divorce because of stalemate. My husband got really uninteresting. [ “stalemate”: “deadlock”/decomposition “a mate that is stale”] (the speaker got divorced because her marriage reached deadlock ĺ the speaker got divorced because her husband became uninteresting) My wife is a customs officer. She keeps checking if I’m behaving myself in the pub. [decomposition of the compound lexeme “customs officer”: “customs employee”/“a person who controls someone else’s behaviour”]

Ambiguity-Based Types of GPs

139

(the speaker’s wife is a customs employee ĺ the speaker’s wife will check his behaviour when he’s in a pub)

Admittedly, cases of standard homonymy and polysemy (of various types) have the greatest number of representatives in the data, the two decomposition categories presented above being intermittent. Decomposition processes appear to be far more frequent in the case of longer lexicalised units, i.e. phrasemes.

5.4. Literalisation of phrasemes: phrasemic polysemies Many a lexically ambiguous GP will revolve around a punning element coinciding not only with a single word or a compound but also with a polysemous string of words constituting one conventionalised phrase. This is hardly surprising, since language abounds in fixed expressions, i.e. phrasemes (Mel’cuk 1995, 1998). Phrasemes can be dichotomised into pragmatic phrasemes comprising sayings, proverbs and clichés, and semantic phrasemes, i.e. idioms, conventional (dead) metaphors and (set) collocations, which can be conceived of as multi-word lexemes (cf. Mel’cuk 1995, 1998). The categories within either of the three-member sets are by no means mutually exclusive but can overlap, primarily sayings with proverbs, clichés with sayings; or dead metaphors with idioms, and collocations (recurrent word combinations) with idioms, which is why idioms may be considered the superordinate category combining the other two. Both the phrasemic types, pragmatic and semantic, are fixed, conventionalised expressions bearing conventionalised meanings, which are the normally generated by default. While in ordinary circumstances lexicalised chunks are treated as entities manifesting monosemy, for humorous reasons, they can be deemed as polysemous, gaining their alternative meanings on the basis of literalisation. The denotation of this term is similar to that of deidiomatisation (Fill 1992) or relexicalisation of preconstructed phrases with its subtype in the form of delexicalisation (Partington 1998, 2006). However, it is broader in scope, as it embraces both semantic and pragmatic phrasemes. A preconstructed semantic phraseme exploited in a GP is processed first on the conventionalised non-literal level (admittedly corresponding to the most salient or default meaning). At the stage of the switch, however, the speaker causes the interpreter to realise that the chunk needs to be backtracked and interpreted at the lexical compositional level, each constituent member being understood according to its literal meaning.

140

Chapter Four I’m not a heavy smoker. I smoke only 10 packets of lights a day. [collocation “heavy smoker”: “someone who smokes a lot of cigarettes”] (the speaker does not smoke a lot ĺ the speaker smokes 10 packets of cigarettes, which are light, rather than strong) I used to think you’re a pain in the neck. Now I have a much lower opinion of you. [idiom, conventional metaphor “a pain in the neck”: “a nuisance”; NB This example capitalises also on the word “lower” in the switch and the resultant allusion to the tabooed part of the body and hyperbolisation consequent upon the use of “used to”] (the speaker no longer thinks that the hearer is a nuisance ĺ the speaker thinks the hearer is a great nuisance) Never keep up with the Joneses. It’s easier to drag them to your level. [idiom “keeping up with the Joneses”: “trying to have all the possessions friends or neighbours have in order not to be inferior”] (one should keep to one’s own standards, rather than aspiring to others’ ĺ one should not aspire to somebody else’s standards but lower theirs to one’s own)

The humorous force of such GPs relies on a double reading of each semantic phraseme on the idiomatic level, which accounts for the more salient meaning, and on the lexical compositional level, where the idiom’s elements are ascribed their respective literal meanings. It is only after the recipient has appreciated the salient meaning of the lead-up that the punster invalidates it with the switch and foregrounds the so far covert interpretation of the multi-constituent chunk consequent upon literalisation, springing a surprise upon the interpreter. As a result, the latter must retrace and reanalyse the preceding chunk of the text to acknowledge the existence of a nonsalient meaning generated upon compositional words’ meanings and dismiss the initially salient meaning as inadequate. Humour-oriented decomposition of idioms violates prevalent norms of language use. An idiom is characterised as an expression of figurative meaning which cannot be computed solely on the basis of the meanings of its parts and is hence noncompositional (Van de Voort and Vonk 1995). This is because an idiom is a combination of words with one lexicalised meaning, which is historically based on the computation of the meanings of the elements but cannot be reduced to them. Idioms are also syntactically frozen and cannot be internally modified. Only minor alterations are legitimate so that the idiomatic processing is not impaired (Van de Voort and Vonk 1995). In addition, there purports to be an overlap between idioms and stable collocations, which are also

Ambiguity-Based Types of GPs

141

characterised by varied degrees of conventionality. It could then be claimed that certain stable combinations of words can be regarded as idioms, especially if it is difficult to appreciate their meanings thanks to the analysis of their components (cf. Veisbergs 1997). In essence, gradability is a crucial quality of idiomatisation, which explains why any conventionalised juxtaposition of words is, at least partly, idiomatic (cf. Fill 1992). It should also be highlighted that what differentiates set collocations from single-word polysemies with their complementations (e.g. “join someone/two elements”) is that the former are well-established lexically and their elements cannot be freely substituted (vs. “join her/the children/two pieces etc.”). A question arises whether GP exploitation of idioms is affected by the latter’s nature. According to the Decomposition Hypothesis (Gibbs and Nayak 1989, Gibbs et al. 1989), meanings of individual words can contribute to the figurative meaning. The advocates of this approach propose a tripartite division of idioms as regards their compositionality, i.e. decomposable (e.g. “pop the question”6) or abnormally decomposable idioms (e.g. “carry a torch for somebody”, where only “torch” cannot be treated literally) and nondecomposable idioms (e.g. “kick the bucket”). Accordingly, while the first two categories embrace idioms whose constitutive members provide meanings which, when gathered, will yield, or at least contribute to, the overall idiomatic meaning, the third one concerns examples when the scrutiny of the elements’ meanings does not bring satisfying results in the sense that the idiomatic meaning can only be recovered from mental lexicon (unless one knows the provenance of the phraseme). However, despite their variability, all the idiom types can be used to produce GP humour. This testifies to the view that each idiom bears a lexicalised and salient reading, which is not computed anew each time the idiom is encountered (see chapter 6, section 4). I have a drinking problem. I keep spilling tea. [decomposable] (the speaker is addicted to alcohol ĺ the speaker’s problems with drinking tea result from his/her clumsiness) I’ve never carried a torch for her! It’s the 21st century; electricity is everywhere! [abnormally decomposable]

6 Whether this idiom can indeed be understood on the basis of a literal understanding of its subordinate elements, as the authors argue, is debatable.

142

Chapter Four (the speaker has never been in love with a given woman ĺ the speaker has not lit the way for the woman) Take life with a pinch of salt, a slice of lemon and a bottle of tequila. [nondecomposable] (one should be doubtful about one’s life ĺ one should drink tequila with salt and lemon all the time)

By the same token, GPs tend to derive from the enforced polysemy of pragmatic phrasemes in the form of proverbs, sayings (or clichés), many of which may be considered to be proverb distortions or a particular class of what Mieder and Litovkina (1999, 2006) dub antiproverbs. Very often, such GPs subvert the import of original proverbs. The primary operation is that a pragmatically conventionalised chunk undergoes decomposition into literal meanings of its constituent elements. Sometimes, if the literalisation effect is not particularly strong, as the meaning of a conventionalised chunk corresponds directly to the computation of the meanings of its constituent elements, the result may be considered a red-light joke, rather than a GP (e.g. “A fool and his money are soon elected.”). On the other hand, pragmatic phrasemes tend to include homonymous/polysemous elements, which strengthen the surprise effect, since their alternative meanings (irrelevant to phrasemes) are eventually activated. There are two other formal mechanisms underlying saying/proverbbased GPs. Firstly, the final element of the original expression can be deleted to be substituted with a different one, compatible with the literal interpretation of the preceding elements. Encountering a familiar expression, the interpreter will, in all likelihood, understand it as an entity and complete it by default according to the standardised pattern only to have to withdraw this inference and backtrack the preceding chunk in the light of the surprising switch. This gives rise to the experience of defeated expectations (cf. section 8, chapter 2). A bird in hand shits on your wrist. [proverb “A bird in hand is worth two in the bush”] (it’s better to be satisfied with what one has ĺ a bird held in hand leaves its waist on one’s wrist) Where there’s a will, I want to be in it. [saying “When there is a will, there is a way.”; polysemy of “will”: “testament”/“determination”] (if someone wants something, they will find the means to achieve it ĺ whenever somebody dies and leaves a testament, the speaker wants to be the heir)

Ambiguity-Based Types of GPs

143

Secondly, as in the case of semantic phrasemes, the whole proverb or saying can be verbalised, while an additional element is added at the end of the text, shedding new light on the meaning of the preceding conventionalised chunk, which needs to be reanalysed along the literal meanings of its constituents. This is, admittedly, the most widespread manifestation of the multi-word GP mechanism holding for all phrasemic polysemies, whether pragmatic or semantic (described earlier). Hard work never killed anybody, but why take the chance. [saying “Hard work never killed anybody.”] (hard work is not detrimental; it is beneficial ĺ hard work may not have actually killed anybody, but it is best not to work and check the veracity of this statement) Beauty comes from within…bottles, jars and tubes. [proverb “Beauty comes from within.”; also polysemy of “within”] (beauty is in one’s personality, not one’s appearance ĺ beauty is the result of using cosmetics) Early to bed, early to rise indicates unskilled labour. [proverb “Early to bed, early to rise.”] (if someone goes to bed early, he/she gets up early ĺ if someone goes to bed early and gets up early, he/she must be a blue-collar worker)

As will be argued in chapter 6, Giora’s (1997, 1999, 2003) salienceoriented approach is admittedly the one which best explains the comprehension process of the humour-oriented phrasemic decomposition process. In brief, semantic and pragmatic phrasemes are instantly interpreted on the non-literal level, for which a sine qua non is the language user’s knowledge of these chunks. It is then the non-literal sense that is subject to faster evocation than the senses of composite elements. Humorous speakers intuitively make use of this phenomenon, exploiting the literal meanings of the elementary constituent members of a familiar phraseme, thereby invalidating the most salient meaning. After the latter has been activated, the next part of the co-textual material causes backtracking and reanalysis of the phraseme according to the meanings of its compositional parts.

6. Pragmatic ambiguity and GPs The concept of pragmatic ambiguity does not appear to have been widely appreciated in literature, even though it is by no means a rare one and certainly crucial to many forms of humour. However, it happens to be

144

Chapter Four

mentioned in passing (e.g. Attardo et al. 2002: 15). Also, Ritchie (2002, 2004), discussing linguistic ambiguity, does recognise that some jokes centre on a peculiar type of ambiguity such as the famous doctor-patient joke discussed by Raskin (1985), which does not involve linguistic ambiguity or two “meanings” in the normal linguistic sense of the term; rather, it involves two ways that the audience (and one of the characters in the joke) could make sense of the other character’s actions. It is just about possible to relabel this as “pragmatic ambiguity”, by claiming that there is ambiguity about the perlocutionary effect (Levinson 1983) of the query Is the doctor at home?, but this is a rather contrived argument, since it would class any action which could have more than outcome as “ambiguous”. (Ritchie 2004: 91)

Dismissing the notion of pragmatic ambiguity as unfeasible, Ritchie (2004) chooses to posit that some jokes entail alternative interpretations of whole sequences of events. Ritchie’s rejection of the term in question is regarded here as unfounded. Indeed, in pragmatic terms, almost any utterance can potentially be ambiguous, with one locution carrying various illocutions, depending on the context and the author’s intention as regards the inferences to be drawn. Although such ambiguities normally pass unnoticed, given the contextual import and background assumptions, they can be employed for humorous purposes. In addition, analysing a garden-path novella featuring a couple whose sexes are not revealed until the end of the story, Mey (1991) appears to be discussing the case of pragmatic ambiguity. The latent ambiguity and the default (but ultimately cancelled) inference that the reader makes about. the couple being heterosexual is possible thanks to the Spanish writer’s avoidance of morphological references and attributives (Mey 1991). Even though a number of lexically ambiguous innuendos reverberate throughout the story, the default inference is of pragmatic, not lexical, nature. The underlying ambiguity could hence be labelled as pragmatic, although Mey does not do thus. As Mey (1991: 245) wittily puts it, “Pragmatic gardens, unlike real ones, are maintained by default. In particular, the famous ‘garden path’ owes its existence to what is not said or written.” Pragmatic ambiguity will then revolve around some form of implicitness. Tentatively, three subtypes of pragmatic ambiguity can be distinguished, according to whether the inference is generated upon the lead-up as a whole (and what is not stated); a particular chunk determining the ambiguity of the whole lead-up; or a presupposed belief and some background knowledge. All of the pragmatic ambiguities can, however, be

Ambiguity-Based Types of GPs

145

perceived as generalised (context-independent) implicatures, as will be argued in the next two chapters.

6.1. GPs based on purely inferential ambiguity Firstly, pragmatic ambiguity in GPs hinges on twofold inferences which are not couched in any particular element of the verbalisation’s surface structure. Instead, the ambiguity appears only at the level of inferences generated from the lead-up, which is why it can be regarded as purely inferential ambiguity. The inference coincides with additional information constituting the hearer’s default development upon what is explicitly said, according to conventional patterns of thinking and communicating. The comprehension process of pragmatically ambiguous GPs does not differ from that typical of lexically ambiguous ones. The interpreter makes a default inference (see chapters 5 and 6) of the lead-up, which needs to be later annulled, owing to its incongruity with the import of the switch. As a result, a second interpretation compatible with the switch is found, thereby proving the lead-up pragmatically ambiguous, i.e. conducive to two competitive inferences, one of which is generated by default, while the other is obtained through an effortful mental process triggered by the incongruity of the first interpretation and the switch. Such inferences can reside in various subordinate mechanisms mirroring common assumptions, factual knowledge and standard patterns of reasoning. Smoking makes me calm. I know I’ll die from it before something worse gets me. (smoking is relaxing, as it affects the processes in the brain ĺ thanks to smoking, the speaker knows the reason of his/her future death and needn’t fear other diseases) Honesty is the key to relationships. If you can’t fake it, you’re lost. (being honest is a sine qua not for building relationships ĺ it is the pretence of honesty that renders relationships successful) I saw the play under adverse conditions; the curtain was up. (while watching the play, the speaker was being disturbed by somebody or something ĺ the speaker finds the source of disturbance the fact that he/she could actually watch it) If you don’t feel good having woken up after 10 a.m., you should try getting up at 5 a.m. I always do and I feel… awful.

146

Chapter Four (sympathising with the hearer, the speaker suggests getting up at 5 a.m., as he/she does, rather than at 10 a.m., which may make the hearer feel better ĺ the speaker suggests that the hearer should get up at 5 a.m. to feel that it has even worse consequences) You should say no to drugs. That will drive the prices down. (people should never buy or take drugs, as they are harmful ĺ people should not buy drugs for some time, which will make them cheaper to buy in the future) We ought never to do wrong when people are looking. (one should not do wrong, irrespective of anything ĺ only when there are witnesses, should one not do wrong)

A distinct subtype of the purely inferentially ambiguous lead-up manifests itself in the incremental development of information, yielding a cluster of compatible inferences of the same type, e.g. conditions or factors, until the switch adds another fact, thereby shedding new light on the global inference garnered so far. I can afford a first-class plane ticket although I’m a teacher. I eat free dinners at the canteen, I walk to school, I’ve got a few shopping discounts and I’ve got a rich lover. (although the speaker earns little as a teacher, she can afford an expensive first-class plane ticket, since she does not spend much due to three factors ĺ she can buy the ticket not because she lives economically, but because she has it bought by a rich lover) I love children as long as they are clean, not hungry, quiet and shown on television. (the speaker likes keeping children company, as long as they are neither dirty nor hungry ĺ the speaker likes children on TV, but does not like children)

The default inferences facilitating GPs presented in this section are obviously generated upon verbalisations. However, they are not directly prompted by any particular lexemes or language structures carrying some default content but by pragmatic inferences based on what is not explicated.

6.2. GPs based on linguistically-motivated inferential ambiguity Another major category of pragmatically ambiguous GP is directly associated with conventional understanding of language structures or

Ambiguity-Based Types of GPs

147

vocabulary elements, which can be pinpointed in the surface-level realisation. Nevertheless, they are not, strictly speaking, lexically ambiguous and, consequently, not conducive to puns. Such formulations stereotypically engender rich, pragmatically motivated interpretations, which are exploited in GPs, where the conventional implications are cancelled and superseded by unconventional and unexpected ones. This category embraces, for example, a common understanding of referents or conventional grammatical formulae. I was happy with my wife for 20 years. Then we met. [“somebody with somebody”: acting together] (the speaker had a happy 20-year marriage with his wife ĺ the speaker was happy before he met his wife at the age of 20 and so was she) I resigned because of illness and fatigue. Everybody was sick and tired of me. [lack of possessive preposition, thus subject co-reference; NB this example additionally hinges on lexical ambiguity in the switch] (the speaker quit because he was ill and tired ĺ the speaker resigned because everybody else had had enough of him) I’ve been going to the “Adonis” Gym for the past three months. Tomorrow I may even start working out. [the “go to N” (school/church/hospital) formula engenders the meaning “to do the associated activity”; NB Such a default interpretation is blocked by the insertion of an article, unless the noun is somehow described or defined, as in this example] (the speaker has been working out in the gym for the past three months ĺ the speaker has only been visiting the gym but has not exercising there) You’ve got the brain of a four-year-old boy, and I’ll bet he was glad to get rid of it. [“noun of somebody”: typical of somebody] (the hearer has a mental capacity typical of a four-year-old ĺ the hearer has the brain donated by a four-year-old) They look really intelligent but actually they’re dumb. [“look”: “give an impression”] (the speaker considers the people intelligent ĺ the speaker admits that they give the impression of being intelligent, which is not the case)

Apart from grammatical structures, at the heart of linguistically-based pragmatic GPs, there tend to lie conventional formulae, whose standard pragmatically motivated import is humorously subverted. Such formulae should not be equated with phrasemes, because they are not lexicalised

148

Chapter Four

and do not result in punning. The ambiguity emerges not from a double lexical meaning of an ambiguous textual chunk but from the twofold inferences (conventional and non-conventional) that are computed on its basis. The little girl is just like her mum. If she keeps growing up like this, things will be pretty bad. [conventional compliment “the little girl/boy is just like her/his mum/dad”] (the speaker makes a routine compliment, praising the likeness between the girl and her mother ĺ the speaker suggests that the daughter resembles her mother, which is nothing praiseworthy, as the girl may end up like her mother) In an emergency, put your head between your knees and kiss your ass goodbye. [formula “put your head between your knees”: the position protects one against head injuries] (if there is an emergency, the head-between-knees position should be assumed as a safety measure ĺ if there is an emergency, one should assume the position to comfort oneself before death) I’m like Achilles, but he had only a heel and I have the whole body. [simile “like Achilles”: very strong like the mythical hero] (the speaker is as strong as Achilles ĺ the speaker is similar to Achilles in the sense that he has a weakness, but its not only his heel but his whole body)

Another distinct subclass of this category is comprised of GPs which are contingent on monosemic words with different specialisations. Ruhl (1989) observes that multi-purpose words, such as common verbs (e.g. “get”, “hit, put” or take) or prepositions (“at”, “in”, “on” and “to”), need to be interpreted in the light of informational extralinguistic import invoked in a particular linguistic environment and conversational context. Ruhl (1989) adopts a monosemic bias, averring that words with potentially many related meanings have essentially one meaning, irrespective of what dictionary entries may indicate. The proposal of monosemic bias seems far-fetched, for Ruhl (1989) deems polysemy or homonymy as a last resort, while the phenomena are by no means uncommon. On the other hand, it cannot be denied that multiplication of meanings due to overspecification is sometimes unnecessary, even if useful for non-native speakers of a given language. Various lexemes such as those enumerated above manifest one meaning but, simultaneously, various realisations achieved via pragmatic specialisation (Ruhl 1989) depending on the collocations used. They cannot be regarded as lexically ambiguous,

Ambiguity-Based Types of GPs

149

notably polysemous. Nevertheless, the monosemic word and its neighbouring co-text, e.g. a verb or prepositional complementation, can be considered one collocation instantly interpreted by default. In GPs, this pragmatic inference is later overridden and, while the core meaning of the element remains unaltered, the specialisation changes into a non-default one, e.g. the verb “love”. I love my work; it fascinates me. I can sit and look for hours. (the speaker loves working in his/her profession and finds it fascinating ĺ the speaker is fascinated by looking at work piling up/other people working) I love cats; they taste like chicken. (the speaker loves the company of cats ĺ the speaker loves cat meat) Finally, I met this guy who said he loved children. Then I found out he was on parole for it. (the speaker finally met a man who was fond of children ĺ the speaker met a man who sexually abused children and had served time in prison)

A GP can also hinge on a switch introducing unexpected hyperbolisation, rather than negation, of a feature introduced in the leadup. Contrary to what may be felt upon the first reading of the lead-up, the speaker is not asserting the negative proposition but objects to the method of verbalisation and reiterates the proposition with greater emphasis. This phenomenon subsumable under pragmatic ambiguity is dubbed metalinguistic negation (Horn 1989). From the perspective of syntax studies, some of the instances could also be regarded as a sudden shift in the scope of negation. You’re not silly…. You’re simply remarkably stupid. (the hearer is quite clever ĺ the hearer is not only silly but also very stupid) The dress doesn’t look expensive. It is expensive. (judging by the looks of the dress, it cannot be expensive ĺ the dress not only seems to be but also is expensive.) Lying is not easy. It is simplistic. (lying is difficult ĺ it is not the case that lying is easy, it is very easy)

Yet another category of GP operates on the metalinguistic use of an element within the lead-up. In essence, the hearer makes a default

150

Chapter Four

inference, not presupposing that a given element is a term, rather than the concept denoted, and should be taken in quotation marks if transcribed. I know karate and a few other words of Japanese origin. (the speaker can perform karate ĺ the speaker knows the word “karate” and a few other words in Japanese)

The above appear to be a few of the most frequently recurring mechanisms of GPs, i.e. couched in pragmatic ambiguity of the lead-up. More second-order mechanisms corresponding to particular verbal vehicles are presented in chapter 6, section 3.3.

6.3. GPs based on presuppositional ambiguity A significant GP class may also be viewed as motivated by ambiguity consequent upon default or salient assumptions, conceived of as presuppositions, which are ultimately cancelled. The nature of default interpretations rooted in presupposition can be appreciated on the basis of witty riddles which respondents solve by asking closed questions. James Smith is lying in his bed wide awake. There are burglars rummaging around in his room. Although he can see them and hear them, he doesn’t even try to stop them or call the police. Why? John and Mary are dead. They are lying on a wet carpet among pieces of glass. The window is open and it is windy. How did John and Mary die?

Interpreters, focused on arriving at answers, normally presume some information, which turns out to be crucial to the solution. The answer to the first riddle comes with a simple realisation that James Smith is a baby, which is very difficult to observe, as formal reference by one’s name and surname is virtually restricted to adults. As regards the second example, after a long series of questions, interpreters normally infer that it must have been wind that thrust the window open and caused a water container, an aquarium to be exact, to fall down and break into pieces. However, very rarely do they conclude that John and Mary are fish, taking for granted the fact that the two proverbial human names must refer to humans. Upon first hearing the riddle, interpreters visualise the two as dead people, which precludes them from solving the riddle. Striving to solve the conundrums by making inferences, hearers often neglect some backgrounded pieces of information, which form presuppositions.

Ambiguity-Based Types of GPs

151

Presupposition is widely discussed in literature as a semantic phenomenon (Strawson 1952, Frege 1892/1984, Keenan 1971, inter alia) and as a pragmatic phenomenon (Stalnaker 1972, 1973, 1974; Kempson 1975, Burton-Roberts 1989; Karttunen 1974, Gazdar 1979, Karttunen and Peters 1979, inter alia). Presupposition, as defined in logical terms, should be true so that the whole statement can be evaluated for truth value (Strawson 1952, Frege 1892/1984). The contemporary conception of presupposition is largely credited to Stalnaker (1972, 1973, 1974), who conceptualised it as constraints on contexts in the form of dictionary meanings of particular words or constructions and general conversational rules, thereby championing the pragmatic perspective on presupposition. The dichotomy between semantic and pragmatic presupposition is nowadays viewed in the context of conventional vs. inferential distinction. The two approaches to presupposition are not in conflict, since a semantic presupposition coincides with given interlocutors’ pragmatic presupposition in a context, while the converse does not hold. Semantic presuppositions are conventionally dependent on lexical items or constructions. On the other hand, pragmatic presuppositions do originate from them but entail propositional inferences as well. To recapitulate, a pragmatic presupposition of an utterance is a background belief which must be mutually known or assumed by the speaker and the addressee for the utterance to be considered appropriate in this context and which can be associated with a specific lexical item or a grammatical feature dubbed presupposition trigger, such as “again”, “too”, “stop” or a cleft sentence (Levinson 1983). However, there may also be pragmatic presuppositions without semantic sources (e.g. von Fintel 2001). What is also significant, both affirmative and negative forms of a statement share one presupposition (Strawson 1952), which is why negation can be used as a test for presupposition and as a way of drawing a distinction between presupposition, entailment and implication (e.g. Gazdar 1979, Levinson 1983). Pragmatic presupposition is normally conceived of as “a background of beliefs or assumptions which are shared by the speaker and his audience, and which are recognised as such” (Stalnaker 1974: 199), against which communication takes place. In addition, Stalnaker (1972, 1973) proposes that if the speaker presupposes something, he/she takes it for granted and presumes that the hearer does the same. Later Stalnaker (1974) adds the premise of mutual acknowledgement of these assumptions. It is also a common contention, credited to Stalnaker, that presupposition is normally a part of common ground, i.e. what is mutually known (e.g. Levinson 1983). Nonetheless, it is not so that the listening interlocutor must already have the knowledge.

152

Chapter Four (…) a speaker may act as if certain propositions are part of the common ground when he knows that they are not. He may want to communicate a proposition indirectly, and do this by presupposing it in such a way that the auditor will be able to infer that it is presupposed. (Stalnaker 1974: 202)

It emerges that at least some presuppositions must be inferred, which accounts for why some presuppositions merge with implicatures (Grice 1975/1989b, see chapter 5). Grice (1981/1989b) equates presupposition with conventional implicature (cf. Karttunen and Peters 1979), and even conversational implicature (e.g. in a statement entailing negation), as the latter can be cancelled. Grice (1981/1989b) thus redefines presupposition as implicature, given that it is not common knowledge but “it is noncontroversial, in the sense that it is something that you would expect the hearer to take from you (if he does not already know)” (Grice 1981: 190). Some may protest that any presupposition as a conversational inference should not be equated with conversational implicature, the latter being the focus of the speaker’s communicative intention. In other words, the strength of the communicative import of presupposition tends to be much less vital in comparison to the strength of implicature, whose conveyance is the main focus of the speaker’s utterance. On the other hand, the speaker need not, even if may, have communicative intentions concerning the presupposition of an utterance, if backgrounded and taken for granted (e.g. Stalnaker 1973; Simons 2003, 2006). Just like conversational implicatures, presuppositions are defeasible and nondetachable (e.g. Gazdar 1979, Levinson 1983, Simons 2006). These two parameters are of utmost significance to the GP humorous mechanism operating on the cancellation of presuppositions derived from particular lexical triggers. Presuppositions are backgrounded inferences, less rigidly associated with the speaker’s main communicative intention. As a result, it might be felt that it is primarily the hearer that takes responsibility for their arising. In other words, although presuppositions are conceived of as mutual, i.e. speaker-intended and, consequently, hearer-derived, the speaker finds it easy to prove them wrong on the grounds that the interlocutor must have drawn ill-advised inferences, regardless of the speaker’s intent. Obviously, the initial inferences are exactly the ones that the speaker intends to communicate with a view to cancelling them for the sake of GP humour. A peculiar class of GP capitalises on presuppositions couched in lexical triggers manifesting covert ambiguity contingent on common knowledge, viz. social facts (e.g. gender stereotypes), conventional understandings of concepts and the choice of prototypical class members. Sometimes, presuppositions can also be equated with dictionary definitions

Ambiguity-Based Types of GPs

153

of lexemes, which can hardly be conceived of as being understood differently, unless context demands this (e.g. “a book”). In accordance with the communicator’s wish, the hearer draws backgrounded inferences, following the standard train of thought motivated by communal common ground (cf. Bubel 2008), which proves to be wrong at the stage of the switch. Also, as the last two examples show, GPs may exploit the speakerintended lack of personal common ground between the addressee and the communicator, which allows for default presuppositions to arise. It’s difficult to find single men who are sensitive, caring and good-looking because they already have boyfriends. [presupposition: men are heterosexual] (it is difficult to find tender and handsome men who haven’t got wives ĺ such men are gay and have boyfriends) The book I like the most is my husband’s cheque book. [presupposition: a book is a novel/collection of stories, etc.] (the speaker’s favourite piece of writing to read ĺ the speaker’s favourite book is not one to read but to spend money with) I’ve been in love with the same woman for forty years – if my wife finds out she’ll kill me. [presupposition: married couples enjoy long-lasting (e.g. forty-year) love] (the speaker has been in a love and enjoyed a stable relationship with one woman for forty years ĺ the speaker’s long-lasting love is not his wife) I’ve betrayed my husband only once during our whole marriage. We’ve been married for two days. [presupposition: the whole period of marriage is at least a few years] (the speaker has betrayed her husband only once during their long marriage ĺ the speaker is remarkably unfaithful having betrayed her husband once during their two-day marriage)

As all the examples show, presupposition-based GPs are different from the other two types presented earlier, where it is the foregrounded conjecture that is eventually defeated. Here, any primary default inferences, constituting the utterance’s main import, which hearers may draw do not embrace those which are most crucial in GPs, due to their being humorously cancelled. Presupposition triggers which take the form of lexical items can be explained as generalised conversational implicatures (Grice 1975/1989b), salient interpretations (Giora 1997, 1999, 2003) or default/presumptive meanings (e.g. Levinson 2000) (see chapters 5 and 6).

154

Chapter Four

7. Final comments The discussion above was concentrated on basic types of GPs according to the underlying ambiguity types. GPs were dichotomised into linguistic and non-linguistic types depending on whether they rely on puns based on lexical (rarely coupled with syntactic) ambiguity or on pragmatic ambiguity, respectively. Several sub-mechanisms within either of the two primary categories were also distinguished to familiarise the reader with the nature of GP phenomena. The next two chapters are meant to account for the two features determining the operation of GPs, i.e. the default/salient interpretation (rendering the ambiguity covert) and the cancellability of this interpretation at the stage of the switch. The discussions will be conducted in the light of relevant cognitive and pragmatic approaches. It will be shown how pragmatic and cognitive proposals can explain GPs, both punning and non-punning. In his analysis of forced-reinterpretation jokes, coinciding with GP humour discussed here, Ritchie (1999, 2002, 2004) introduces the parameter of “obviousness” as the quality of the first interpretation of the set-up. Nevertheless, nowhere does he elucidate the notion of obviousness, taking it for granted and treating it rather intuitively. Ritchie also (2002) suggests that the inference process will always yield a set of interpretations ordered by the degree of obviousness, with one maximal element, i.e. the most obvious interpretation. As earlier suggested, it is, nevertheless, open to doubt whether a set is indeed activated or whether the interpreter is initially satisfied with one understanding and searches for alternatives only if necessary, i.e. on the strength of the following co-text. Admittedly, only one interpretation is initially evoked, but as the on-line processing continues, the oncoming text invalidates the meaning observed, necessitating the activation of another one. To reformulate, the following co-text clashes with the salient meaning and demands reinterpretation of the preceding chunk, in result of which the salient/default meaning is annulled and the nonsalient/non-default one is foregrounded.

CHAPTER FIVE GRICE’S PHILOSOPHY AND GARDEN-PATH HUMOUR

The publication of Grice’s lecture entitled “Logic and conversation” in 1975 is regarded as a milestone in the development of the studies on communication and the discipline of pragmatics, in particular. The philosopher’s perspectives on language and communication never cease to garner scholarly interest, giving rise to neo-Gricean studies as well as the polemic strand, post-Gricean studies. As Davis (1998) observes, while Grice viewed his philosophical ideas as tentative and exploratory, contemporary researchers regard them as rigid, and thus many heuristic problems emerge causing heated debates, even among advocates of the Gricean framework of communication. Indeed, it contains a host of underdetermined and even contradictory points, which can be interpreted and resolved in a variety of ways. Despite any potential problems the model may cause when used as a heuristic tool for the analysis of GPs, its value cannot be denied. Grice’s lectures presented over a few decades in articles (Grice 1957, 1961, 1968, 1969, 19751, 1978, 1981, 1982, 1989a) were posthumously published in a book (Grice 1989b)2, which boasts a variety of strands, each of which presents a number of topics, often intertwining. In the context of the present work, the most significant of those are the speaker’s meaning, and implicatures, together with the Cooperative Principle and the subordinate maxims. Due to the complexity and multifariousness of Grice’s philosophy, which developed over the years, it is hardly possible 1

The lecture on “Logic and Conversation” was published in two different volumes in 1975, but had existed in spoken form (together with the 1978 article) already in 1967. 2 Most verbatim quotes come from the volume of collected lectures (Grice 1989b), while only the year of the original publication is provided. It cannot be ruled out that in some cases the quotes may not mirror the extracts from the original publications, given the corrections that Grice introduced prior to the 1989 publication.

156

Chapter Five

to revisit all the relevant points in a clear and orderly manner. Nor is it the objective here. The goal is to expound on the Gricean conceptualisations as seen heuristically feasible to explain the relationship between humour and the Cooperative Principle as well as the applicability of the phenomenon of the generalised conversational implicature in the analysis of GP humour. The aim of this chapter is to conduct two, almost independent but equally important, discussions on humour in the light of Grice’s philosophy. The chapter opens with the analysis of GPs, and humour in general, within the framework of communication anchored in the Cooperative Principle with a view to proving that, contrary to the wellentrenched approach, the Gricean model of communication does hold for humorous interactions. The second sub-chapter addresses the GP mechanism with regard to the postulate of inferences in the form of implicatures.3 A question may appear in the reader’s mind as to the rationale underpinning the choice of Grice’s implicatures in the study of GPs. The answer is that the classic concept of the generalised conversational implicature is known to manifest two features fundamental to the operation of GPs, i.e. it arises irrespective of co-text and context, on the whole, and is defeasible/cancellable.

1. Humour and the CP The interdependence between humour and the Cooperative Principle (CP) (Grice 1975/1989b, 1978/1989b, 1989a) appears to be a most problematic issue in pragmatic studies on verbal humour. The wellentrenched approach advocated by Raskin and Attardo is that jokes (and also other forms of intentionally produced humour) constitute the nonbona-fide mode of communication standing vis-à-vis the Gricean mode, the supposed bona-fide mode, and being governed by a humour-CP (Raskin 1985, 1987, 1998; Raskin and Attardo 1994; Attardo 1990, 1993, 1994, 1996a, 2006), and that they violate, not merely flout, the maxims and even the CP (Attardo 1990, 1993, 1994, 1996a, 2006). The aim here is to discuss the relation between humour and the CP in order to substantiate that the authors who regard humour as an independent communicative mode and as an intrinsic violation of maxims 3 In pragmatic literature “implicatures” and “inferences” are used almost interchangeably, even though the former represent the speaker’s perspective, while the latter the interpreter’s. They are used synonymously, as the speaker’s implicatures will normally coincide with the implicatures or inferences drawn by the hearer.

Grice’s Philosophy and Garden-Path Humour

157

and the CP appear to labour under a serious misapprehension. It will hence be argued that the Gricean model does allow for humorous expression. Firstly, although humorous forms infringe maxims, they do thus overtly, i.e. flout/exploit them, which the CP model of communication presupposes. Secondly, humour does not violate the CP as long as the humour-oriented communication proceeds as planned. The CP obtains in any interaction between the speaker who rationally produces an utterance and the hearer who draws implicatures along rational lines, presuming the interlocutor to be rational. Accordingly, rationality, which lies at the heart of Grice’s model, is also inherent to intentionally produced humorous verbalisations, be it jokes or conversational humour such as witticisms, teases, etc. It is also GP humour that is compatible with the Gricean model of cooperation and rationality, even if the speaker leads the interpreter up the garden path and deliberately misleads the latter. Finally, even if humorous keying does exist, the concept of the non-bona-fide mode standing in opposition to the Gricean model is unfounded.

2. The Gricean model For the sake of clarity of presentation, the discussion on the relation between humour and the Cooperative Principle (CP), together with its subordinate maxims, must be opened with a recapitulation of the pertinent premises of Grice’s output on communication, with a view to gravitating towards a presentation of humour in the context of the model. Readers familiar with Grice’s writings may consider some, if not all, parts of this overview obvious. Nevertheless, this overview is not redundant, as it will be compared with Attardo’s and Raskin’s reading of Grice’s philosophy and their postulates concerning humour. A number of thorny issues need to be resolved as regards the formulation of the cooperation model. The analysis is not confined to the two publications (Grice 1975/1989b, 1978/1989b), most widely acknowledged in linguistic studies, but has a broader perspective on Grice’s work, which is crucial for adequate perception of Grice’s theories (cf. Davies 2000, 2007). Utmost attention will be paid to vital, albeit underestimated, aspects of the Gricean model which are of crucial significance to understanding conversational humour, a phenomenon grounded in interlocutors’ rationality.

2.1. Maxim nonfulfillment and inferences Undoubtedly, Grice’s model of communication will have garnered widespread support primarily thanks to the concept of implicature

158

Chapter Five

generated in the light of the CP and its subordinate maxims. The process of inferring conversational implicatures depends on the assumption that the speaker is adhering to the CP, whereas the subordinate maxims can be flouted, i.e. exploited, to yield implicatures. Interestingly, Levinson (1983) appreciates that Grice presents implicatures as associated with maxim flouts but suggests that implicatures can also originate from observing the maxims, i.e. “standard implicatures are derived from the assumption that the speaker is observing the maxims” (Levinson 1983: 126). The formula of the CP can by no means be unfamiliar to any language researcher. Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged. (Grice 1975: 26/1989b: 26)

Grice (1975/1989b) introduces Quality, Quantity, Relation and Manner as categories (not maxims, as they tend to be called in literature), each of which comprises maxim(s), i.e. maxims of Quality, maxims of Quantity, maxims of Manner and the maxim of Relation. Grice (1975/1989b) does allow for the fact that other maxims (aesthetic, social or moral such as “Be polite”) may also exist, but the conversational maxims are central to conversational implicatures, being connected with particular purposes of verbal exchanges. Quantity (i) Make your contribution as informative as is required for the current purposes of the exchange. (ii) Do not make your contribution more informative than is required.4 Quality Supermaxim: Try to make your contribution one that is true. (i) Do not say what you believe to be false. (ii) Do not say that for which you lack adequate evidence. Relation5 Be relevant. Manner 4 Grice (1975/1989b) makes a reservation that overinformativeness may not be a transgression but only uneconomical use of time. On the other hand, he admits that overinformativeness may also mislead hearers. 5 Contrary to what may often be found in literature, Grice employs the term “maxim of relevance” only in passing (1969/1989b: 87), whereas when focusing on the maxims, Grice (1975/1989b, 1978/1989b) does not introduce it at all. He does use the word “relevance” in the literal sense, though.

Grice’s Philosophy and Garden-Path Humour

159

Supermaxim: Be perspicuous. (i) Avoid obscurity of expression. (ii) Avoid ambiguity. (iii) Be brief (avoid unnecessary prolixity). (iv) Be orderly. (Grice 1975/1989b: 26-27)

The maxims are presented as subordinate to the CP and should not be (although they often are in literature) analysed in isolation from the metalevel of the CP. As Taylor and Cameron rightly note, “few commentators pause to consider Grice’s avowed motive for introducing the CP. Instead they rush on to consider the various maxims which are subordinate to it” (Taylor and Cameron 1987: 83). The CP is proposed as unchallengeable common knowledge and a tacit mutual agreement, which interlocutors take as a covert communicative sine qua non enabling them to observe implicatures manifesting a great variety of forms. The process of computing conversational implicatures depends on the assumption that the speaker is adhering to the CP, while the subordinate maxims may be either observed or flouted, i.e. exploited, to yield implicatures. Even if all the four categories of maxims (Quality, Quantity, Relation and Manner) should be exploited, the CP still holds. Many misinterpretations of the Gricean model arise due to the inconsistencies of the nomenclature researchers apply when revisiting the theory, which sometimes results in equating maxim violation with maxim flouting. Even Grice himself happens to be inconsistent in the use of the term “violation” (e.g. Grice 1975/1989b: 32), which he appears to apply as a superordinate term in reference to two main types of failures to fulfil maxims (cf. Thomas 1995). Grice suggests that a flout is “a justifiable violation” or “only a seeming, not a real, violation; the spirit, though perhaps not the letter, of the maxim is respected” (Grice 1989a: 370). Irrespective of such unfortunate formulations, neither the two phenomena, nor the terms denoting them can be equated. Most rational is Mooney’s (2004) suggestion to use the term nonfulfilment, rather than “violation”, as a blanket category covering cases when the participant will “fail to fulfil a maxim in various ways” (Grice 1978/1989b: 49). Grice distinguishes two other ways of failing to fulfil a maxim, apart from flouts and violations, i.e. clashes and opting out. Thus, all the forms of nonfulfilment proposed are (Grice 1975/1989b: 30): 1. Unostentatious violation of a maxim, which is liable to mislead.

160

Chapter Five

2. Opting out both from a maxim and the CP, which the speaker manifests by verbally or otherwise indicating that he is unwilling to co-operate, e.g. by saying “My lips are sealed”.6 3. Clash between two maxims when the participant violates one maxim in order to fulfil another, e.g. being informative but lacking evidence. 4. Flout/exploitation (or infringement, cf. Grice 1975/1989b: 33) a blatant failure to fulfil a maxim, which is reconciled with the supposition that the entire CP being observed. Apart from listing examples which hinge on implicature-yielding maxim flouts, Grice (1975/1989b: 32) distinguishes two other categories, one in which “no maxim is violated” and one in which a maxim is “violated” but it is redeemed by the clash of maxims. Upon closer inspection, the former appears to be a clear instance of the Relation maxim flout, while the latter is an exploitation of the first maxim of Quantity. It is difficult to explain why Grice chose to distinguish those instances as different from flouts. The most vital and the most problematic, at the same time, are unostentatious violations and flouts, differentiated by their covertness and overtness, respectively. Flouting is a legitimate phenomenon guaranteeing the fulfilment of the CP and generating conversational implicatures, whereas violation takes place unbeknown to the interpreter and misleads him/her, albeit doing thus in accordance with the speaker’s intention. The majority of Grice’s lectures concerning the CP and implicatures, and, likewise, his followers’ research, are focused on flouts, whereas violations appear to be marginalised. Besides the widely acknowledged CP and its subordinate maxims, Grice distinguishes four other important factors which influence the inferential process and which the hearer must take into account when observing conversational implicatures: 1. the conventional meaning of the words used, together with the identity of any references that may be involved; 2. the context, linguistic and otherwise, of the utterance; 3. other items of background knowledge; 4. the fact (or supposed fact) that all relevant items falling under the previous headings are available to both participants and both participants know or assume this to be the case (Grice 1975/1989b: 31)

6

On a global level, the speaker does prove cooperative by explicating his/her intentions and thus by not being deliberately uncooperative.

Grice’s Philosophy and Garden-Path Humour

161

Yet another factor that Grice contributes later is “the psychological state or attitude which needs to be attributed to a speaker” (Grice 1989a: 370). Therefore, it transpires that Grice does allow for background assumptions in utterance interpretation, thereby completing the picture of communicative processes. It should also be noted that these assumptions will greatly influence the interpretation of maxims or their flouts. Before those are discussed, the notion of maxim violation must be revisited, given its prominence in humour studies. 2.1.1. Maxim violations Bearing in mind the intrinsic covertness of violations (from the hearer’s vantagepoint), it could be hypothesised that violation pertains only to the act of lying, understood as deliberate misleading, which invariably entails the violation of the first Quality maxim. Sometimes lying additionally involves being economical with the truth, i.e. violating the first Quantity maxim in order not to have to tell the whole truth. Apart from these two cases of premeditated deception, it is hardly conceivable that other maxims should be unostentatiously violated. Should they be (e.g. Manner maxims in doublespeak), they only accompany the violation of the first Quality or Quantity maxims. If present, the intention to violate a maxim, i.e. to mislead must be covert, while the inherent quality of a flout is its overtness. A question arises as to what happens if the hearer discovers the speaker’s aim to mislead, thereby putting paid to the latter’s communicative plan. In such a case, the hearer may actually draw certain inferences as to the speaker’s premises, different from the speaker’s intended implicatures. Normally, there can be no error recovery concerning violation, due to there being no indication of it (Mooney 2004). Mooney (2004) thus dubs lies successful violations, which produce no implicatures. On the other hand, successful violations may be regarded as fostering misleading implicatures (Thomas 1995). Allegedly, an untruthful proposition which remains a successful violation is simply interpreted as genuinely meant by the speaker and, as a result, emerges as most cooperative, which is on a par with the communicator’s expectation. Non-detected lies would then be subsumable under the CP. This is, however, what Grice does not allow for, acknowledging the privileged status of the first maxim of Quality, writing that it appears to spell out the difference between something being, and (strictly speaking) failing to be, any kind of contribution at all. False information is not an inferior kind of information; it just is not information. (Grice 1989a: 371)

162

Chapter Five

Grice hence maintains that the importance of, at least, the first maxim of Quality is such that it transcends the whole model and “other maxims come into operation only on the assumption that this [the first] maxim of Quality is satisfied” (Grice 1975/1989b: 27), which many concede (e.g. Levinson 2000, Horn 1984). Grice suggests that the CP holds only for truthful communication, with deliberate lies excluded from the model. This postulate, albeit put forward hesitantly, appears to carry the moral overtones of the model. Admittedly, this may be responsible for misinterpretations of the model and benevolence attributed to the speaker (see section 2.3). It is probably because of the illegitimacy of the Quality violation, which is here argued to underpin other violations, that this type of nonfulfilment is given short shrift in Grice’s writings, while flouts are of primary importance to the model and account for the generation of conversational implicatures, coinciding with the prototypical realisation of the CP. 2.1.2. Maxim flouts Among the core properties of implicatures, Grice (1975/1989b, 1978/1989b) lists calculability, which entails hearer’s making inferences, following from the premise of rational conversational activity. Conversational implicatures are calculable, because of rationality that underlies the line of argumentation which leads the addressee from acknowledging a maxim flout to construing the speaker’s intended meaning. Presupposing the existence of the CP, the addressee is always capable of drawing the conversational implicature of a particular utterance. To reformulate, maxim flouts guide the discovery of non-conventional inferences, on the assumption that the speaker is rational. It is also crucial to observe that rational thought is non-monotonic and subject to change in the light of new information (e.g. Ginsberg 1987, Elio 2002). This means that implicatures can be cancelled if they are falsified by new information, which will be elaborated in the second part of this chapter. Davies (2000, 2007) reports that the implementation of the CP and the subordinate maxims is sometimes claimed to coincide with being direct and explicit, while failure to do so necessitates clarification and repair (e.g. Bernsen et al. 1996). By the same token, Chapman (2005: 190) states that some commentators unduly accuse Grice of utopian idealisation of conversations, which hardly ever progress as envisaged, i.e. with all the maxims fulfilled. It needs to be highlighted that the model may be idealistic in the sense that it does not presuppose any miscommunication or deceit, but maxim flouts are presupposed. They are most legitimate in

Grice’s Philosophy and Garden-Path Humour

163

the model. Grice is preoccupied with the conveyance of implicit meanings and hence the discrepancy between saying and meaning. Rather than concentrate on miscommunication avoidance, as some suggest he does (e.g. Bernsen et al. 1996), Grice is concerned with how the communicator forms his/her utterance, employing implicit language, i.e. flouting maxims, so that the addressee recognises his/her intention. Conversational implicatures recuperate seemingly inappropriate conversational contributions, since they exploit the speaker’s background knowledge of the maxims in a given situation, which the speaker also expects to be the hearer’s presumption. A maxim which the hearer observes to have been flouted sets in motion the process of argumentation in order to safeguard the underlying presumption of cooperation. Therefore, cooperation does not invariably correlate with directness, i.e. conformity with the maxims. Nor is it so that observing the maxims enhances a conversation (Davies 2000, 2007). Moreover, as Lakoff (1995) rightly notes, in some situations, the lack of flout generating implicatures may be conducive to misunderstanding. To recapitulate, maxims are to be understood as conversational assumptions, which arise “from basic, rational considerations and may be formulated as guidelines for efficient and effective use of language” (Levinson 1983: 101). On the other hand, efficiency and effectiveness may rest on maxim flouting, which is thus equally legitimate as maxim observance. A claim could even be ventured that maxim flouts and implicatures consequent upon them are more frequent and unmarked, whereas maxim observance is the marked case (Davies 2007). The maxims are indeed advanced mostly to be acted against for the sake of specific communicative goals. Also, some authors question Grice’s model on the grounds that it does not apply universally across all speech encounters, e.g. those where interactionists are not equals. This strand is pursued, e.g. in critical discourse analysis (cf. Chapman 2005: 193). However, there is nothing in the Gricean model of communication that prevents it from being applied also to such exchanges, if need be, with some of the maxims being suspended. Yet another point of criticism, not fully substantiated, though, is cultural variability in the applicability of maxims, for which Grice’s model does not allow (e.g. Keenan 1976). Nevertheless, Grice simply does not take the cultural parameter into account. By no means does he claim that the maxims will apply in all cultures in the same form. Also, the inoperability of particular maxims should not invalidate the entire idea of rational communication and so the CP, which appears to be universal.

164

Chapter Five

2.2. Rationality and intentionality at the heart of cooperation Introducing the notion of the CP, Grice presents communicational cooperation, stating, “our talk exchanges do not normally consist of a succession of disconnected remarks, and would not be rational if they did. They are, to some degree at least, cooperative efforts” (Grice 1975/1989b: 26). The import of this quotation is transparently that “cooperation” is to be understood not as a joint communicative effort towards a common goal but, primarily, as the rationality of coherent exchanges. It is noteworthy that the term “cooperation” is marginal in the philosopher’s writings, discussed in only two lectures on communication (Grice 1975/1989b, 1978/1989b), whereas the prevailing and prevalent notions are intentionality, the speaker’s meaning, rationality, as well as the calculability of meanings (Davies 2000, 2007), all of which form the requisite basis for the CP, motivating the operation of conversations. Grice’s conceptualisation of cooperation pivots on the notion of rationality. In the first lecture on the CP, Grice states, “one of my avowed aims is to see talking as a special case or variety of purposive, indeed rational behaviour” (Grice 1975/1989b: 26). Also, in the retrospective epilogue, Grice emphasises that the use of language is a rational activity which involves “rational cooperation” (Grice 1989a: 341). Clearly, the notion of cooperation as proposed by Grice should not be equated with the folk meaning of the term (Davies 2000, 2007). Grice’s cooperation resides in logicality, providing the rational and intention-based foundations of meaning, maxims and the CP. It is the rationality or irrationality of conversational conduct which I have been concerned to track down rather than any more general characterization of conversational adequacy. So we may expect principles of conversational rationality to abstract from special character of conversational interests. (Grice 1989a: 369)

Davies (2000, 2007) even argues that the only reason why the Principle of Cooperation is not dubbed the “Principle of Rationality” is because Grice addresses logic at the level of discourse, where rationality is realised between interlocutors. This underpins the postulate (expounded later) that humorous verbalisations, if rational and intentional, do conform to the CP mode. The notion of rationality as a communicative foundation reverberates throughout pragmatic literature. Recanati writes, “Pragmatic interpretation is possible only if we presuppose that the agent is rational” (Recanati 2001: 82), while Elster states, “Communication and discussion rest on the

Grice’s Philosophy and Garden-Path Humour

165

tacit premise that each interlocutor believes in the rationality of the others, since otherwise there would be no point to the exchange” (Elster 1986: 27). It is also Attardo (2003) that acknowledges the fact that the Gricean model is rooted in the rationality principle, which he equates with the interactants’ decision to cooperate unless they decide otherwise. This is the essence of the Gricean CP. 2.2.1. Intentionality, i.e. the speaker’s meaning Crucial in Grice’s writings is the factor of intentionality, the prism through which the notion of meaning is explicated (Grice 1957/1989b, 1969/1989b, 1982/1989b). Even before formulating the CP, Grice appears to have emphasised the aspect of the speaker’s intentionality underpinning communicated meanings. Nota bene, Grice’s interest in linguistics began with the analysis of the speaker’s meaning (to be discussed in the second part of this chapter). Grice maintained that linguistic meaning derives from the speaker’s meaning explained by the cognitive phenomenon of intention. However, faced with severe criticism voiced by other philosophers, in his later lectures published in 1989, he changed his views, admitting that what an utterer intends to convey can be constrained by the conventional meaning of an utterance (Chapman and Routledge 2005). Grice (1957/1989b) advocates nonnatural meaning (meaningNN) in conversation (see also section 6 below) as the meaning uttered by the speaker who intends to induce a belief in the hearer. Grice (1961/1989b) develops this, contending that what the speaker means is what he/she Mintends to convey through a particular utterance. Since Grice presupposes intentionality on the speaker’s part, unintentional communicative stimuli (e.g. hiccupping, slips of the tongue, or unintentional humour production) would not count in the model of communication.7 Communication occurs when the hearer appreciates the speaker’s intention, which itself is the cause of the belief (Grice 1957/1989b: 219). Later, Grice (1969/1989b) sustains the notion of intentionality but adds that the utterer means something if he/she intends to produce a particular response in the audience and wants the audience to be aware of this intention. The fulfilment of the latter is the sine qua non for the fulfilment of the former intention. Admittedly, the speaker’s intention may be geared towards generating humour and eliciting the hearer’s humorous response. 7

Some authors (Sperber and Wilson 1986/1995, Carston 2002, Bach 1994) discredit Grice’s model on these grounds.

166

Chapter Five U meant something by uttering x is true iff, for some audience A, U uttered x intending: (1) A to produce a particular response r (2) A to think (recognize) that U intends (1) (3) A to fulfill (1) on the basis of the fulfillment of (2) (Grice 1969/1989b: 92)

Whether a belief is to be induced or a response produced, Grice (1957/1989b) emphasises the intentionality of nonnatural meaning as realised by both the speaker and the audience. Accordingly, if the speaker nonnaturally means something, he/she intends to produce some effect upon the hearer and to have this intention recognised by the latter. Therefore, the utterance will have the nonnatural meaning which is codependent on the audience. The audience’s appreciation of the speaker’s intention is the reason for the meaning surfacing. As Levinson (2000: 29) puts it, “communication involves the inferential recovery of speakers’ intentions: it is the recognition by the addressee of the speaker’s intention to get the addressee to think such-and-such that essentially constitutes communication.” In other words, communicative intention is reflexive, not iterative, and so the communicative success consists in the hearer’s recovery of the speaker’s intentions, which need not occur at the level of the recipient’s awareness. Arguably, the intention will be consciously observed only in the case of utterances which pose problems in the inferential process and provoke the interpreter’s deeper and so conscious analysis of the speaker’s intent (as is the case of genuinely friendly teasing realised via pretended putdowns). 2.2.2. Rationality, reasoning and mutual observance of the CP As already stated, within Grice’s model, the communicative and inferential processes hinge on the presumption of the CP, which is realised by what is said and what is implicated (Grice 1969/1989b). Calculating conversational implicatures entails inferring what is necessary to preserve the supposition that the CP is being observed, which revolves around specific factors peculiar to a given situation (Grice 1975/1989b). The immanent presumption is that the interlocutor is rational and that a logical interpretation of his/her utterance is to be sought, even if it flouts maxims. Reasoning and thus rational communicative behaviour are crucial for both producing utterances and generating inferences (Grice 1982/1989b). On the assumption that the CP holds, the hearer computes implicatures, i.e. rational inferences based on the recognition of the speaker’s intentions.

Grice’s Philosophy and Garden-Path Humour

167

Since, to calculate a conversational implicature is to calculate what has to be supposed in order to preserve the supposition that the Cooperative Principle is being observed, and since there may be various possible explanations, a list of which may be open, the conversational implicatum in such cases will be the disjunction of such specific explanations; and if the list of these is open, the implicatum will have just the kind of indeterminacy that many actual implicata do in fact seem to possess. (Grice 1975/1989b: 39-40)

The CP is then upheld, while the maxims may be either observed or flouted. Grice defines maxims as “principles or axioms dependent on an overall super-principle enjoying conversational cooperation” (Grice 1981/1989b: 268) and as “desiderata that normally would be accepted by any rational discourses” (Grice 1981: 185). Many misunderstandings concerning the status of the CP and the subordinate maxims may originate from their formulation, which grants them a prescriptive appearance (Chapman 2005). The maxims are speaker-oriented, since they are formed to resemble imperatives directed at the speaker. Likewise, the formulation of the CP, i.e. “Make your conversational contribution” (Grice 1975/1989b: 26), appears to be targeted at the speaker. However, the CP and the subordinate maxims are designed neither as prescriptions nor as observations (as some claim) but rather as default settings or presumptions (Horn 2005). It should be appreciated that the ultimate framework of communication and the Cooperative Principle predict that although it is the speaker that is responsible for the production of meanings, the hearer will be able to draw adequate inferences, as intended by the speaker, and to preserve the default assumption of rationality and the resultant cooperation. Hence, the concept of implicature, even if originally proposed as the speaker’s meaning, can be viewed from the hearer’s perspective, encompassing inferences that he/she is capable of drawing in the light of the speaker’s rationality. He has said that q; there is no reason to suppose that he is not observing the maxims, or at least the Cooperative Principle; he could not be doing this unless he thought that p; he knows (and knows that I know that he knows) that I can see that the supposition that he thinks that p is required; he has done nothing to stop me thinking that p; he intends me to think, or is at least willing to allow me to think, that p; and so he has implicated that p. (Grice 1975/1989b: 31)

To conclude, the underlying presumption of the CP is that interlocutors cooperate in the sense that they are communicatively rational (Grandy and

168

Chapter Five

Warner 1986). As already discussed, this does not mean that they must be literal to accomplish this goal. Cooperation does not intrinsically entail the speaker’s benevolence to the interpreter manifesting itself in the avoidance of any processing effort to be made by the latter, as some commentators suggest (see Davies 2007). On the contrary, within the CP model, such difficulties will arise only to be overcome by the hearer who presupposes the speaker’s communicative rationality. As Davies (2007) rightly posits, it is the presumption of the CP that allows the speaker to impose a greater inferential effort on the hearer, who must invest his/her cognitive resources in order to comprehend a given utterance. This observation is most pertinent to humour, the understanding of which is often intellectually demanding. Also, cooperation, as conceived by Grice, immanently carries no interlocutors’ benevolence manifesting itself in skirting difficult or debatable topics or in cooperative striving towards a common goal. The communicative goal will now be the focus of attention.

2.3. The purpose of communication The Gricean model allows for different communicative purposes and requirements, granting speakers considerable leeway in the choice of communicative goals. The CP offers the rationale for any type of communication, whether formal or informal, adversarial or friendly, taskoriented or ostensibly purposeless. It cannot be argued that the Gricean rationality model is reliant on the stipulation that the sole reason for communication is the exchange of informative content. Researchers often neglect one of Grice’s statements found in the first lecture on the CP. The conversational maxims, (...) and the conversational implicatures connected with them, are specially connected (I hope) with the particular purposes that talk (and so, talk exchange) is adapted to serve and is primarily employed to serve. I have stated my maxims as if this purpose were a maximally effective exchange of information; this specification is, of course, too narrow, and the scheme needs to be generalised to allow for such general purposes as influencing or directing the actions of others. (Grice 1975/1989b: 28)

Additionally, in the epilogue to his collection of writings, Grice (1989a: 369) acknowledges the fact that some analysts will have misunderstood his idea of cooperativeness as being restricted to purposeful conversations, with the exclusion of disputes or seemingly purposeless chats. Grice (1989a) clarifies that irrespective of whether the aim of an exchange is specified or if it is indeterminate, whether the

Grice’s Philosophy and Garden-Path Humour

169

interlocutors concur or at cross purposes, the same principle will hold. Grice hence reiterates his postulate on communicative goals, which appears to have been clear enough even without this explication. Goals are very broadly conceptualised and may also be second-order ones, as in the case of a casual chitchat, in which “each party should, for the time being, identify himself with the transitory interests of the other” (Grice 1975/1989b: 29). The model could be extended even further to cover interactions produced as a form of the Jakobsonian phatic communion, which embrace forms of humour devoid of communicative content, e.g. jokes produced solely for the sake of a humorous effect. Phatic utterances are commonly defined as linguistic devices aimed at establishing or maintaining the interactive contact between interlocutors (e.g. Abercrombie 1998, Laver 1975). In other words, they facilitate the creation of a feeling of involvement, agreement or relational solidarity between interlocutors. However, it must be borne in mind that humour, even canned jokes, can also carry relevant informative content, as argued in chapter 1. Additionally, one of the fundamental premises here is that even ostensibly uninformative contributions such as humorous acts do convey meanings, producing social implications (cf. Mooney 2004) and affect interpersonal relations (see e.g. Martin 2007 and references within). What is of crucial importance, discussing an intentionally conveyed utterer’s occasion-meaning, Grice (1969/1989b: 92) does allow for the audience’s amusement as the primary goal of the speaker. A major query concerns the way the goal is chosen and pursued. The CP entails the presumption that the speaker and the hearer interact rationally to reach a common goal. The use of Passive Voice in “as is required” in the formulation of the CP does not specify by whom or from where the requirements and acceptance of the communicative goal are issued. A question appears whether the requirement and purpose of the interaction comes from the speaker, the hearer or perhaps widely accepted norms. Grice elucidates this issue, stating that human conversations are “cooperative efforts; and each participant recognizes in them, to some extent, a common purpose or set of purposes, or at least a mutually accepted direction” (Grice 1975/1989b: 26), which suggests underlying, tacit negotiations between communicators. Nevertheless, Grice explicates that although participants in an interaction have a common immediate aim, their ultimate respective aims may be “independent and even in conflict” (Grice 1975/1989b: 29). Moreover, conversationalists may manifest “a high degree of reserve, hostility, and chicanery and with a high degree of diversity in the motivations underlying quite meager common objectives” (Grice 1989a:

170

Chapter Five

369). Therefore, one may gather that quarrels or adversarial humour, such as exchanges of brickbats, by means of which interlocutors wish to outwit one another, will also be easily accommodated by the CP model as long as two conditions are met. The first condition is that the CP transactions must be voluntary and concerted exchanges, rather than solitary ones, (Grice 1989a) and they need to be formed by two interlocutors, the speaker and the hearer, with their respective goals. The second condition is that verbal encounters continue in an appropriate style until the interactants wish to terminate them (Grice 1975/1989b). In essence, Grice (1989a) affirms that irrespective of whether the common goals are apparent or even nonexistent, the CP still holds.

2.4. Can the CP be violated? The CP should be regarded as a communicative presumption, since people converse with a communicative intention (Bach and Harnish 1987). Only if the speaker acts within the rationality model and presumes that the hearer does likewise and vice versa, can communication succeed. Arguably, if the CP were not followed and the speaker were not guided by rationality and the resultant cooperativeness, all communicative interactions would grind to a halt. One of the most intriguing queries is then whether the entire CP can possibly be (covertly) violated or overtly nonfulfilled just as the subordinate maxims can be. Grice (1978/1989b) contributes to this conundrum by stating, “the Cooperative Principle and some subordinate maxims are standardly (though not invariably) observed by participants in a talk exchange” (Grice 1978/1989b: 41). One may infer, admittedly wrongly, that the CP may sometimes be acted against. The reservation “not invariably” probably pertains mainly to maxims and their giving rise to flouts. Although the CP may not always be operative (cf. the act of lying), it should be acknowledged that Grice (1975/1989b, 1978/1989b) takes as his departure point the ideal model of perfectly formed communication which does not allow for any communicational failures, where the CP unremittingly obtains, and where there is no room for misunderstandings or deception. Grice philosophises on the issue of communicative practices, not verifying his claims against empirical material (Chapman 2005), which may explain why some of the theses appear fallible when applied to real-life discourse. In his notes for the lecture on “Logic and conversation”, eventually not included in the published version, Grice writes,

Grice’s Philosophy and Garden-Path Humour

171

I am here considering what is (or may be) only an ideal case, one which is artificially simplified by abstracting from all considerations other than those involved in the pursuance of a certain sort of conversational cooperation. I do not claim that there actually occur any conversations of this artificially simplified kind; it might even be these could not be (cf frictionless pulleys). (Grice’s notes, quoted in Chapman 2005: 192)

This ideal Gricean model of frictionless communication lays the foundations for pragmatics, the field of linguistics focused on the mechanisms underpinning language in use. This is hardly surprising, since the aptness of many of Grice’s observations on rational communication, even if purely theoretical, is undeniable. There are, however, exceptional cases when the CP does not hold in discourse. Thomas (1995) dubs the CP nonfulfilment infringement (different from Grice’s (1975/1989: 32-33) use of the term) arising owing to imperfect command of language consequent upon low competence (of a child or a non-native speaker) or state of mind such as drunkenness or nervousness. Undoubtedly, infringement is a case of communicative failure, but the CP is not violated in actual fact, given that the speaker fails communicatively what with a linguistic or mental incapability he/she cannot control and is even unaware of. If rational and competent, the hearer does realise that the former does not, or rather cannot, conform to the CP. In accordance, the hearer observes the interlocutor’s uncooperativeness or linguistic irrationality as unintended, unless he/she cannot be cooperative either on the same grounds. “Infringement” means that the CP is not operative, and thus suspended, due to the fact that the condition of the speaker’s rationality and intentionality is not met. This postulate could also be extended to cover situations when it is the hearer that is incapable of acting in accordance with the cooperative rationality provision or simply misunderstands the speaker’s intention. Expounding on this ideal model of communication, Grice is hardly preoccupied with misunderstandings or miscommunication (Davies 2007). Nota bene, this type of idealised approach is also adopted in the present work on GPs, which are interpreted according to one prototypical pattern, yielding a humorous result. Although Grice (1978/1989b) mentions in passing the possibility of implicatures being misunderstood, he does not conceive of any reasons for, or the consequences of, communicative failures. Nor are such equated with the CP violation. Needless to say, reallife communication is prone to misunderstandings, which can arise, for example, when the hearer takes it for granted that the speaker is observing a maxim, while the latter is not, in result of which the former will miss the implicature (Hunter 1983). Hunter (1983) proposes reformulation of

172

Chapter Five

Grice’s maxims so that the threat of misunderstanding should be minimised. The author restates the maxims, emphasising the conditions on which those can be flouted. To meet this goal, she introduces the parameter X, constituting “the hearer’s information (as conceived by the speaker), his inferencing skills and his willingness to process cooperatively” (Hunter 1983: 201). The maxims are reformulated with a view to vindicating speakers. However, the tripartite stipulation is present, even if inexplicitly, in Grice’s notion of rational cooperativeness. Hunter’s (1983) addition is, therefore, superfluous. In essence, unintentional communicative failures committed by interactants (either by the incompetent speaker or the hearer who draws wrong inferences or is not able to do thus), for which the model does not account, cannot be equated with violation of the CP. If at all, the latter could be the case of two interlocutors who are incapable of intentional and rational expression of their thoughts and of appreciation of each other’s irrationality. Otherwise, no absolute CP violation is possible, as normally interlocutors have certain intentions and consciously pursue them in all social encounters. The nonfulfilment of the CP can also be analysed from a different perspective. According to Grice’s (1975/1989b) conceptualisation, the CP does not hold if the first maxim of Quality is violated, which is the case of deliberate deception, i.e. lying. What is of crucial importance, that the CP does not hold does not mean that it is violated. If a communicative act is completed, albeit revolving around false premises introduced by the speaker, it still cannot be said that the CP has been violated. The CP is simply inoperative, as Grice (1975/1989b) himself suggests. In real-life discourse of lying, which Grice’s model does not capture, the hearer who does not detect a lie and interprets it as the truth will perceive no maxim violation. The speaker, on the other hand, will achieve his/her communicative goal by violating the Quality maxims and producing a “misleading implicature” (Thomas 1995: 73). To conclude, in the Gricean view, the possibility of violation of the CP, whether overt or covert, is simply ruled out and the operation of the CP is assumed by default. The goal itself does not need to be rational, while the speaker’s pursuit of it must be thus (Attardo 2003). Once again it ought to be emphasised that the CP is inherently contingent on the concept of the rationality of conversational participants, who ought to know, or rather who ought to have internalised the norm that it is reasonable to follow the CP and, therefore, that it should not be abandoned either by themselves or by their interlocutors. In other words, the speaker produces an utterance which prompts inferential processes on the part of the hearer, who, presupposing rational conduct of argumentation, can draw

Grice’s Philosophy and Garden-Path Humour

173

conversational implicatures. This carries the overtones of Grice’s idealist beliefs about the pre-established mutual trust of communicators and the importance of the first Quality maxim, betraying the moral embedding of the conception, which might also arouse the critique that the model is utopian.

3. The Gricean model and humor This part of the chapter will aim to revisit and criticise the postulates advocated by Attardo and Raskin as regards the correspondence between the Gricean model and humour (Raskin 1985, 1987, 1998; Raskin and Attardo 1994; Attardo 1990, 1993, 1994, 1996a, 2006). The arguments raised against the two authors’ approach to intentional verbal humour are grounded in Grice’s communication model, whose most fundamental premises have been discussed above and will be assumed as being familiar to the reader. Attardo boldly states, “a consensus has been built within humour research that humorous texts violate one or several of the maxims” (Attardo 1993: 103), which is an immodest claim, insofar as Attardo is the main proponent of this approach. Also, there are authors who reject the view of maxim violation and the CP violation (e.g. Morreall 2004, Kotthoff 2006). Unfortunately, such competitive lines of argumentation manifest varied degrees of plausibility. For example, Kotthoff (2006) unnecessarily raises the issue of politeness, a complex multifaceted issue (cf. Brown and Levinson 1978, 1987), in the discussion on humour, laying her contention open to criticism (see Attardo 2006).8 It cannot be denied, however, that numerous authors tend to take for granted and repeat, even if only in passing, Raskin’s notion of the non-bona-fide mode with its own humour-CP and Attardo’s claims about humour violating Grice’s maxims and the CP. The four strands within the study on humour and the Gricean model intertwine and are mutually reliant. By necessity, such will be the sections of the discussion below, since the explanation of one of the issues will entail another to be tackled later. Raskin and Attardo’s theory aims to hold primarily for canned jokes but is also presented as pertinent to other forms of humour (Attardo 1996a). The discussion below will present a broader picture of intentionally produced verbal humour, both jokes and conversational humour, whether spoken or written. Naturally, the focus here is primarily 8 Grice (1975/1989b) mentions that Politeness maxims may need to be added to the set of four maxim categories and be embraced by the CP.

174

Chapter Five

on GP one-liners and witticisms, which can use frames typical of canned jokes (e.g. if told by stand-up comedians) or can occur spontaneously in conversations, bearing relevance to the ongoing interaction. In the latter case, the humour frame will hence merge with the non-humorous mode of communication. The fundamental assumption here is that even if the speaker’s intention is not initially available to the interpreter, i.e. the hearer is oblivious to the fact that the speaker wants to produce humour, this cannot be the case of maxim violation or the CP violation, because the true intentions are eventually revealed and humour-oriented communication develops as planned by the speaker. In other words, it is here argued that producing humour, the speaker flouts maxims but abides by the CP, perfectly realising the Gricean model of communication. Raskin and Attardo’s approach will emerge as nothing but a superfluous methodological complication.

3.1. Other discussions of the Gricean model and humour interface Raskin and Attardo are not the first to have expressed their viewpoint on the relation between the Gricean model and humour (cf. Attardo 1994 for further references). For instance, Hancher assumes that many jokes “involve violating the felicity conditions on illocutionary acts described by Austin (1962) and Searle (1969), and the related conversational maxims proposed by Grice” (Hancher 1980: 20). First of all, Grice’s philosophy as such is independent from the theory of speech acts. As Chapman (2005: 76) reports, Grice objected to having his theory of meaning discussed as reliant on Austin’s “How to Do Things with Words”, affirming that his “Meaning” had been almost finished by the time the latter work was publicised. Secondly and more importantly, the violation of Grice’s maxims does not appear to be feasible, thanks to the inherent covertness of real violations (see the discussion below). Moreover, Hancher (1980: 21) posits that the exploitation of linguistic ambiguity “would be one way for the hearer to ‘opt out’ from the rule of the ‘Conversational Principle’ that organizes conversation.” The author discusses examples of what nowadays may be called “trumping” (Veale et al. 2006, Brône et al. 2007, Brône forth.). The second speaker brings to light the ambiguity of the first speaker’s utterance, responding to the meaning the first speaker did not intend to convey. From the perspective of the meta-recipient, such humour can also be viewed as realisations of the GP mechanism. It transpires that Hancher (1980) understands cooperation according to the literal sense of the word, which contrasts with Grice’s principle of interactional

Grice’s Philosophy and Garden-Path Humour

175

rationality, and regards such humorous interactional pairs as the suspension of the CP. However, an adequate understanding of the principle allows for the inclusion of the humorous verbalisations described in the rationality model. The genre of witty conversational humour, some centred on trumping (Veale et al. 2006, Brône forth.), is also discussed by Hunter (1983) in her analysis of witty responses. Hunter’s (1983) examples include: “Ben isn’t the smartest guy in the world” “Not when the world has you in it.” (ironically); “I consider Frank brilliant as he reads Nietzsche.” “But he sure is a moron the rest of the time.”; “My shoe hurts!” “Take it off, and I’ll rub it for you.” Hunter (1983) avers that witty responders pretend to misunderstand the intentions of their interlocutors, or need not even pretend if they can capitalise on the first and second Manner maxims, which make the speaker’s meaning less certain. Additionally, maxims are claimed to entail two levels of meaning, i.e. sentence-meaning or literal meaning and utterer’s meaning. Such a dichotomy appears to be misguided, given Grice’s parlance and the postulate that implicatures are built upon what is said, both of which are classified as the speaker’s meaning. Hunter (1983) might be excused, since she addresses witty responses to utterances which yield two meanings, one intended and one unintended by the speaker. Nevertheless, it might be better to reformulate the bifurcation into meanings contingent on maxim fulfilment (what is said) and maxim flouts (implicatures). Indeed, Hunter (1983) proposes that wit operates on the respondent’s reference to the preceding utterance as if it relied on the adherence to the maxims, which it actually flouts. Nota bene, one verbalisation may actually show maxim flouts on both readings, the humorous speaker’s aim being to refer to the meaning which is not intended by the first speaker. On the other hand, as Hunter (1983: 196) also observes, an utterance can have two literal meanings, of which only one is intended by the speaker, e.g. “Can you make out these letters?” after which the responder starts kissing the newspaper. Nota bene, this instance manifests conventional indirectness couched in the use of the modal verb. In essence, Hunter (1983) posits that witty respondents misapply maxims to produce wit and are uncooperative in their endeavour. Neither of these claims, even if palatable at first glance, appears to be entirely plausible. “Misapplication” is not a term to be found in Grice’s writings. A question arises if its introduction is necessary. First of all, this misapplication produced by the humorous speaker pertains not to the witty response he/she produces but to the way he/she interprets the preceding verbal contribution. Albeit aware of the first speaker’s intention, the responder rationally chooses to refer to the alternative unintended

176

Chapter Five

meaning, not the intended one, in expectation of producing humour. The responder, now the speaker, redirects the purpose of the communicative exchange, which Grice does allow for, and utters a response relevant to the alternative meaning, which is unintended by the first speaker. The response entails maxim flouts, primarily Relation and Quality, and is still cooperative in Grice’s sense, for the speaker is most rational and rationally pursues a goal, which the interlocutor will also appreciate. After the humorous interaction has been completed, the conversationalists may again redirect their exchange, returning to serious communication, or may engage in a longer humorous exchange, i.e. mutual teasing or banter. A humorous response may follow the preceding humorous utterance. Hunter (1983: 197) also mentions such a case of a dean who walks in, sporting a new hair transplant and says, “Hello, ladies! I’m your new dean,” to which a secretary responds, “Are we glad to see you! The guy you’re replacing is a real jerk.” and aptly analyses the first humorous utterance as reliant on a flout, i.e. “overt violation of a maxim” (Hunter 1983: 197). The response hinges on the pretence that the first speaker is not flouting, i.e. overtly violating, the Quality maxims. Essentially, the responder enters the humorous frame, in which the maxims can be freely flouted to meet a humour-oriented objective. It must be emphasised that this is not a covert violation, because the parties involved know each other’s real intentions and share the humour-oriented communicative goal. Yet another paper tackling humour and the Gricean model is Yamaguchi’s (1988) proposal concerning GP humour. Given the focus of the present work, this paper should be treated with utmost attention. Yamaguchi posits that the maxims are not “deceptively violated in any other jokes but the garden-path joke” (Yamaguchi 1988: 327), which is partly at odds with Raskin and Attardo’s view described below, according to which all humorous forms perform such violation. Nota bene, criticising Yamaguchi’s (1988) approach, Attardo (1993, 1994) fails to observe that it is restricted to one class of joke. As already discussed in chapter 1, Yamaguchi (1988) differentiates between punning and non-punning garden-path jokes, claiming the latter to be subject to analysis as pivoting on the violation of Grice’s maxims. However, it is here asserted that all GPs, irrespective of the type of ambiguity in which they reside, can be analysed within the Gricean paradigm, manifesting maxim flouts, rather than (covert) violations. Yamaguchi (1988) holds a view that maxims are deceptively violated so that the addressee is misled. In particular, it is the first maxim of Quality (i.e. saying what is false) and the first maxim of Quantity (i.e. saying too little) that are conducive to garden-path jokes, while the

Grice’s Philosophy and Garden-Path Humour

177

violation of other maxims is of secondary importance, with the exception of the second maxim of Quantity and third and fourth maxims of Manner, which bear little significance for garden-path jokes, since they appear incapable of producing ambiguity. Despite supporting the notion of maxim violation inherent to garden-path jokes, Yamaguchi (1988) argues in favour of the Character-Did-It Hypothesis, with a view to indicating a cooperative aspect of jokes. According to the hypothesis, characters are entitled to violate maxims in order to produce the necessary ambiguity, while the narrator must avoid, or at least minimise, his/her responsibility for maxim violation. As Yamaguchi wittily puts it, the narrator who has the writer’s voice can not be a leg-puller himself. Instead, he remains a string-puller and lets the characters dance on the textual stage, saying, ‘The character did it. The character did it.’ (Yamaguchi 1988: 336)

Yamaguchi (1988) lists a number of strategies which allow the narrator not to shoulder the responsibility for maxim violation, i.e. direct speech and indirect speech conveying the character’s words attributed solely to himself/herself, as well as three strategies occurring in the narrator’s report of the event, i.e. viewpoint projection (the violation/misunderstanding is projected to the character’s mind), evasion (some crucial detail is left unaccounted for, e.g. thanks to the formulation) and backgrounding (a particular misleading expression the narrator produces recedes to the background and holds no interest for the interpreter). Nota bene, Attardo’s (1994) line of argumentation against Yamaguchi’s (1988) postulate is that if the narrator needs to minimise the responsibility, the violation is still present. Rightly, Yamaguchi observes, “When communication takes place (even if for fun), transmission of information should be realized in largely cooperative circumstances” (Yamaguchi 1988: 324) and adds that maxims “are not violated in the garden-path joke for any other purpose than that of making the joke potentially ambiguous” (Yamaguchi 1988: 327). It emerges, therefore, that garden-path jokes can be analysed at two levels, i.e. the in-joke level, at which maxims are violated, and the realcommunication level, at which the violations are treated as conducive to the success of ambiguity-based jokes. Ultimately, no genuine violations are present, while those which do exist are rationally applied and, what is crucial, revealed. A question, therefore, emerges as to whether the notion of maxim violation in jokes should be supported at all. Accordingly, the violation occurs in the fictional world only, while the rational joke-teller only exploits the maxims, sometimes using joke characters as mouthpieces,

178

Chapter Five

having every intention of amusing the hearer. It is here averred that actually no genuine violations exist, inasmuch as the hearer is not left oblivious to maxim nonfulfilment and appreciates the global intent of the speaker. Irrespective of its import, Yamaguchi’s (1988) Character-Did-It Hypothesis is inapplicable in the context of GPs, insofar as the responsibility for the production of conversational witticisms or one-liners lies solely on the speaker and there is nobody else to be held accountable. Indeed, GPs may be viewed as failing to fulfil the maxims. However, the nature of this nonfulfilment is not that of violation but a flout, since the mischief is produced rationally and instantly revealed by the speaker. The problem of GP humour will be raised again at the end of this subchapter after Raskin and Attardo’s perspectives have been criticised.

3.2. Non-bona-fide mode? Proposing the non-bona-fide mode for humorous cooperation, Raskin (1985) develops upon the premise that Grice’s CP, together with the maxims, yield the bona-fide mode, which he ascribes to Grice (1975) and defines as the “ordinary” mode, where there is no room for lying, acting or joking (Raskin 1985: 89). In other words, the motivation for the alternative communicative model is that the bona-fide mode accounts for effective and truthful conveyance of meanings, i.e. for abiding by the CP, to which humour does not conform. Bona-fide communication is governed by the ‘co-operative principle’ introduced by Grice (1975). According to this principle, the speaker is committed to the truth and relevance of his text, the hearer is aware of this commitment and perceives the uttered text as true and relevant by virtue of his recognition of the speaker’s commitment to its truth and relevance. (Raskin 1985: 100-101)

First of all, it is interesting to observe that nowhere in his writings does Grice (1989b) advocate the “bona-fide mode” as synonymous to the realisation of the CP. In the whole collection of lectures, Grice (1989b) mentions the term “bona fide” only once in a non-technical sense and certainly not in reference to the CP (Grice 1961/1989b: 244). It is Raskin (1985) that is responsible for the term’s formation and ill-advised attribution to Grice. Secondly, Raskin (1985) interprets the bona-fide mode, supposedly based on the CP, as the speaker’s being relevant and truthful. As discussed earlier, the primary concept in Grice’s writings is rationality and not relevance, with the Relation maxim bearing the

Grice’s Philosophy and Garden-Path Humour

179

potential for being legitimately flouted as long as this serves some communicative purpose. Transparently, Raskin’s relevance claim is not well-founded. The issue of the truthfulness maxim is a problematic one. In Raskin’s view, the Gricean framework presupposes that the speaker is “absolutely and unexceptionally committed to the truth of what is being said” (Raskin 1998: 99), while in the non-bona-fide mode, the speaker’s commitment to truth is waived (Raskin 1985, 1987, 1998). Strangely enough, Raskin (1998) laments that the truth-based communication approach, credited to 20th-century logicians, fails to embrace and handicaps humour. (The judiciousness of Raskin’s negative evaluation of the truth-oriented perspective on communication shall not be commented upon here.) Raskin (1985, 1998) develops his alternative model on the assumption that Grice’s commitment to truth is replaced by the speaker’s commitment to humor, so the new maxims (see the next section) serve as a formula for an efficient joke. Indeed, Grice (1975/1989b) leans towards placing the first maxim of Quality above all others, but this should not be used as an argument against humour being subsumed under the CP, since, producing humour, the speaker does not normally mean to be untruthful. It is doubtful whether humour can be conceived as a communicative phenomenon standing vis-à-vis truth. Raskin actually admits that humor is “a very close extension of the completely truthful” (Raskin 1998: 100), suggesting that people will find truth in the fictional world of jokes standing in opposition to the real world. The crux of the matter is that jokes and forms of conversational humour (with the exception of cases when the speaker malevolently wishes to deceive and poke fun at the target) cannot be equated with not being truthful, i.e. lying. Producing a joke or a humorous utterance relevant to the ongoing interaction, or not being serious and not supporting the import of one’s utterances, e.g. in teasing and banter, does not mean that the speaker has any intention of lying. The speaker’s objective is not to lie, and thus to deceive, but to amuse the hearer, which appears to be most legitimate in the Gricean model, because a given verbalisation does not covertly violate Quality maxims. Raskin’s (1995, 1998) understanding of the Gricean model of communication (in Raskin’s terminology, the bona-fide mode) is fallacious also on different grounds. Raskin (1998) explicitly states that the model requires that all the maxims be infallibly fulfilled, which leads him to the conclusion that real-life implementation of the bona-fide mode is the language of a technical manual. Thereby, Raskin rejects legitimate flouts and implicatures stemming from the former, which are the most

180

Chapter Five

fundamental tenets in Grice’s writings on communication, the tenets which have given rise to extensive pragmatic research. In addition, Raskin (1985: 146) claims that deliberate ambiguity is excluded from the Gricean CP. This is certainly another misguided belief. Grice does allow for the flout of the second Manner maxim, which yields deliberate ambiguity underlying, among others, many humorous forms. Beyond a shadow of a doubt, Raskin’s interpretation of the Gricean model and the role of maxims and the CP by no means coincides with the philosopher’s conceptualisation. In the same vein, Raskin and Attardo (1994: 32) state that “the main venue” for bona-fide communication is using literal language. The latter stands vis-à-vis non-literal language, which the authors equate with figures such as: metaphor, irony, understatement and exaggeration or metonymy; indirect speech acts; implicature and humour. (Needless to say, the list provokes misgivings, since the categories are not distinguished along one criterion and are not mutually exclusive.) Even if the authors admit that these non-literal phenomena “stand in various relations to Grice’s category of bona-fide communication” (Raskin and Attardo 1994: 32), they appear to be of the opinion that literalness is the prototypical realisation of communication within the CP, which is not the case, inasmuch as most significant in Grice’s writings on communication are maxim flouts and resultant implicatures. Ironically, Raskin and Attardo (1994: 32) regard flouts/exploitations yielding implicatures as “uncomfortably accommodated” in the bona-fide model. Transparently misreading Grice’s view of speaker’s maxim flouts and their relation to the CP, the authors put forward other dubious claims. When the speaker is making a joke, Grice’s cooperative principle is not honored, and therefore, the speaker is not trying to communicate as simply as possible, or in other words, the speaker is engaged in non-bona-fide communication, one set of non-literal modes. (Raskin and Attardo 1994: 34) Raskin distinguishes, as does Grice, between a bona-fide type of communication, in which the speaker is committed to communicating in the most effective way, as clearly as possible, etc., in short follows the CP. (Attardo 1996a: 5)

At this stage, it hardly needs to be explained that no mode bifurcation is to be found in Grice’s lectures, while the fulfilment of the CP does not entail being explicit so that the hearer’s processing costs are minimised. In addition, the authors maintain that the bona-fide mode is “informationconveying” (Raskin 1985: 89), “fact-conveying”, “no-nonsense” (Raskin

Grice’s Philosophy and Garden-Path Humour

181

and Attardo 1994: 32) and “precisely serious” (Attardo 2006: 353), whereas the purpose of the non-bona-fide joke-telling mode is not to convey any information contained in the text he is uttering but rather to create a special effect with the help of the text, namely to make the hearer laugh (Raskin 1985: 101)

As earlier reported, Grice (1975/1989b, 1978/1989b, 1989a) explicates that his model does capture communicative practices which are not intended to convey information as such. The amusement of the hearer, solidarity enhancement or the release of tension in a social encounter may constitute the primary purposes of a communicative act, which is realised in accordance with the Gricean model. In essence, the CP model, albeit not coinciding with the fallacious concept of the bona-fide mode, will successfully hold for intentionally produced humour. Raskin (1985) asserts that this separate humour-cooperative mode has to be introduced, and, to guarantee the hearer’s humour-oriented cooperation, the speaker forewarns the former, either verbally or by means of non-verbal cues, that a joke is intended, which obviously does not pertain to spontaneous jokes. The interpreter may appreciate this intent without having had it signalled. Alternatively, the hearer may not know that the speaker is trying to produce humour and only when he/she fails to analyse the text in bona-fide communication, will he/she backtrack the text of a joke as a humorous one (Raskin 1985, Raskin and Attardo 1994). Analysing a sample joke, Raskin (1985) suggests that it is only at the stage of the trigger that the switch into non-bona-fide mode occurs. (…) the combinatorial rules will have to reject the first default value of the mode of communication adopted earlier and switch to the next default value, namely that of joke telling, which is of course, non-bona-fide. (Raskin 1985: 125)

Admittedly, this situation appears to be highly unlikely in reality. A joke is characterised by distinctive features in structuring, register or topic, rendering it instantly recognisable. However, initial mistaking a humorous chunk for a serious utterance, though instantly cleared, may be the case of some conversational humour forms, e.g. contextually adjusted one-liners. Moreover, the postulate of the non-informative non-bona-fide mode may also be found inappropriate, given that jokes and, particularly, conversational humour can carry informative import relevant to the ongoing non-humorous exchange. One-liners and witticisms, as argued in chapter 1, may function, for example, as witty comments germane to a

182

Chapter Five

conversation. Attardo himself addresses the problem but claims to solve the paradox, by stating “jokes communicate on the basis of the presuppositions that the text may have independently of its humorous nature” (Attardo 1996a: 6). The fact that humorous verbalisations communicate meanings, whether via presuppositions or via implicatures, appears to be yet another reason why humour cannot be excluded from the Gricean model. Finally, the authors posit that the hearer who realises that the speaker has abandoned bona-fide communication (…), no longer has any clue as to the speaker’s commitment to the truth or relevance of anything in his or her utterances. In other words, if the speaker is not bound by Grice’s cooperative principle, then the hearer is limited to the processing of the literal meaning of the speaker utterances while being in the dark about their speech act-status and thus barred from making any inferences from these meanings. Even the existential presuppositions are then called into question (…) The effect of this situation should clearly be devastating to the process of communication. (Raskin and Attardo 1994: 34)

Indeed, having apprehended that the speaker is being jocular, the hearer needs to appreciate the former’s intention and, with the exception of conversational humour conveying relevant meanings, sense that he/she is not referring to genuine situations, feelings, beliefs, etc. However, it is not so that the hearer perceives the speaker to be irrational, which would entail suspending the CP, and discontinues the process of implicature computation, e.g. to understand a joke. The world of the joke may indeed be different from the “real world”, violating common sense assumptions, but the communicative process occurs unobstructed. To account for this fact, Raskin and Attardo (1994) emphasise that the non-bona-fide mode is indeed cooperative and shares some features with the “bona-fide” mode. The reason bona-fide communication extends so naturally into humor is that both are cooperative. In fact, we tend to think that the distinction between the cooperative and non-cooperative modes of communication will prove more essential than that between bona-fide communication and non-bona-fide communication, one of the set of non-literal modes in which Grice showed no interest whatsoever. (Raskin and Attardo 1994: 34)

This quotation reveals that the authors will have had some doubts about the distinction, at least at some point of their studies. Nota bene, discussing the non-cooperative mode, Raskin and Attardo (1994) focus on lying, which is indeed uncooperative and falls outside Grice’s Cooperative Principle. By the same token, Raskin does observe that the two modes

Grice’s Philosophy and Garden-Path Humour

183

could actually be merged, even if his premises for this statement are dubious. In fact, it would be quite reasonable to postulate an extended form of bonafide communication which would include humor since ordinary discourse abounds in facetiousness. (Raskin 1985: 104)

Such strands as the two above are not pursued, while the authors (particularly Attardo 1990, 1993, 1994, 1996a, 2006) are adamant that the two distinct modes are absolutely indispensable and are governed by two separate principles and sets of maxims (see the next section). It is Raskin’s multi-faceted misunderstanding of the Gricean CP and the role of the maxims will have lead him to the conclusion that humour falls outside the CP model and that an alternative mode of communication must be generated for it. Raskin’s (1985) doubt-provoking bona-fide vs. non-bona-fide distinction is frequently repeated in humour literature. On the whole, the dichotomy between the humorous mode and the serious mode is certainly an important one, as it helps to distinguish purely humorous units (jokes, teasing, etc.) from serious discourse, with a reservation that some forms may lie at the border between the two. A vexing question arises whether the serious vs. humorous mode distinction must be introduced in the context of the CP. The notion of a humorous mode is essential but should not be contrasted with the supposed Gricean mode if such a move entails dissociation from the CP as well. This is why the notion of humorous keying (Goffman 1974; Kotthoff 1999, 2007) or framing (Bateson 1953, Gumperz 1982) appears to be preferable, as already argued in chapter 1. Needless to say, contrary to Raskin’s claim about exclusive modes, keying/framing allows for purely humorous verbal activities to be described within the Gricean model. In essence, according to Grice, communicators engage in communicative behaviour, presupposing a mutual tacit agreement that some rules are followed. This guides the recipient to the interpretation intended by the speaker. The same pertains to communication of a humorous character. Additionally, it must be highlighted that the two modes of communication can merge. They converge in units of conversational humour, such as witticisms, which are amusing but do convey meanings relevant to the interaction and, therefore, cannot be enclosed solely in the jocular mode (cf. Morreall 2004).

184

Chapter Five

3.3. Humour-CP? Claiming that humour does not conform to the CP and its maxims, Raskin (1985) postulates that humorous discourse and serious communication are mutually exclusive, while the former necessitates a cooperative principle peculiar to itself, viz. the humour-CP. Thus, four maxims anchored in the humour-CP for the non-bona-fide mode of joke telling are advocated. First and foremost, it must be appreciated that the proposal explicitly accounts only for jokes and does not cover forms of verbal conversational humour (e.g. teasing or witticisms), which appear to be unjustly marginalised as if they were captured by neither the CP nor the humour-CP. (i) Maxim of Quantity: Give as much information as is necessary for the joke (ii) Maxim of Quality: Say only what is compatible with the world of the joke (iii) Maxim of Relation: Say only what is relevant to the joke (iv) Maxim of Manner: Tell the joke efficiently (Raskin 1985: 103, Raskin and Attardo 1994: 37)

Raskin (1985) and Raskin and Attardo (1994) also draw some parallelisms between the Gricean model and their own framework. Just as bona-fide communication can fail if the speaker does not have full control of the maxims [i.e. Grice’s], humor fails if the maxims [Raskin’s] are not abided by. Similarly, the hearer can fail the speaker in bona-fide communication even if the speaker does everything right and the hearer of the joke can fail to get it even if the speaker provides all the necessary ingredients and follows all the maxims. (Raskin 1985: 103)

Indeed, the nonfulfilment of Raskin’s maxims may result in a joke’s failure. However, it should be emphasised that, in Grice’s model, the maxims need not, and very often will not, be abided by but flouted, in which case the speaker still has full control of the maxims, while the hearer infers the implicated meaning. Secondly, Grice does not allow in his idealised model for miscommunication consequent upon the hearer’s misunderstanding the speaker’s intention, especially if the speaker verbalises it adequately. Both the similarities are, therefore, unfounded. Surprisingly enough, Raskin (Aymone 2007) admits that his humourCP was actually “a spoof”, which appears to have deceived many. As Raskin (Aymone 2007), provocatively explains, his motivation in the 1985 book was to ridicule the idea of the CP and to show its methodological

Grice’s Philosophy and Garden-Path Humour

185

vacuity for humour analysis. Whatever his motivation may be, Raskin appears to have had a change of heart and no longer fully advocates his earlier proposal, the proposal which has garnered so much support. On the other hand, Raskin (Aymone 2007) still upholds the notion of the nonbona-fide mode for humour, admittedly divorced from the main original premise as being opposed to the Gricean model. Among the deceived, the most prominent researcher is Attardo (1990, 1993, 1994, 1996a, 2006), who zealously supports Raskin’s (1985) thesis of a separate CP with its set of maxims forming a different mode of communication, adding that humour actually violates the Gricean maxims and the CP. The lowest common denominator is the original CP, but then a humour-CP is introduced which can accommodate the original CP, but can also allow violations of the CP as long as they are eventually redeemed by an ulterior intent. (Attardo 1994: 286-287)

The postulate of the humour-CP being hierarchically superior to the Gricean CP does not appear to be wise. Even if one accepts that the two principles do exist, a reverse situation would more likely be the case, humour being a subtype of communication. However, the most rational choice would be to retain only the original CP.

3.4. Maxim violation? Attardo’s (1993, 1994) major contribution to Raskin’s proposal is that humour actually violates the Gricean CP and its maxims, as a result, constituting the non-bona-fide mode governed by a humour-CP. Attardo (1993, 1994) acknowledges the difference between maxim violation and flouting, in accord with Grice’s view that flouting is an overt violation, but uses the terms inconsistently, as many quotations will show. Attardo champions a claim that humour, primarily jokes, violate the maxims and that they “do not flout or exploit the maxims, but that they violate them, i.e. they fail to conform to their ‘recommendations’” (Attardo 1993: 542, Attardo 1994: 273). It is self-evident that this clarification does not elucidate much, as it fails to appreciate that flouting also entails “failing to conform” but with a particular intention on the speaker’s part. Attardo finds some motivation for the violation claim in Grice’s writings. (…) flouting, i.e. their patent (Grice has ‘blatant’) violation that allows the hearer to infer that a given maxim is being violated only insofar as another maxim is being obeyed. (Attardo 1993: 543)

186

Chapter Five (…) Grice’s original suggestion that irony and humor might be connected to violations of the conversational maxims. (Attardo 1993: 541) All jokes involve violations of one or more of Grice’s maxims. The claim that jokes could be viewed in terms of flouting of conversational maxims dates back to Grice himself, who considers irony as an example of implicature, and a (complex) pun. (Attardo 1996a: 5)

Firstly, Grice (1975/1989b: 30) clearly does not suggest any correlation between one maxim’s flout and another’s observance but discusses maxim flouts in the light of the CP. There is nothing in the Gricean proposal that disallows the simultaneous flouts even of all maxims. Secondly, Grice (1975/1989b: 34) discusses the case of irony as a means of verbalisation which hinges on the Quality maxim flout, certainly not maxim violation, which produces a type of implicature. Irony can sometimes coincide with humour but certainly cannot be equated with a joke or a pun, merely because it is couched in some form of ambiguity. Grice does not explicitly account for the interdependence between humour and maxim flouts, let alone violations. However, Grice’s (1975/1989b: 36) discussion of an example centring on non-humorous ambiguity (I have sinned/I have Sind), which may be perceived as a pun (a common manifestation of humour), corroborates that a communicator may wish to flout at least one of the Manner maxims to convey an implicature. Attardo’s argumentation adduces no evidence to his violation proposal, and even subverts it in the light of Grice’s original view of flouts. It is here maintained that the humorous speaker flouts maxims, aiming to amuse the hearer. Indeed, jokes and other forms of humour may fail to fulfil maxims, temporarily misinforming or perplexing the hearer, for example, by providing him/her with too little information or, in the case of garden-path humour, allowing “wrong” inferences to exist until the punchline/switch is reached. However, all this occurs according to the speaker’s plan, of which the hearer becomes fully aware, whether or not warned in advance that the verbalisation is of the humorous type. In either case, the hearer is led to the proper interpretation of the whole text at the end of the interpretation process. Hence, on no account can this nonfulfilment be a matter of violation, because real violation is inherently covert to the interpreter. What is of utmost importance, producing humour, including GPs, the speaker has no intention of genuinely deceiving the hearer, which is the aim underlying maxim violation. Moreover, in many humorous forms, such as jokes or teasing exchanges, maxim flouts are even expected. Contrary to Attardo’s (1993) claim, whether a joke (or any other humorous form) is preceded by clues or whether it is rendered with a

Grice’s Philosophy and Garden-Path Humour

187

deadpan facial expression, the rational interpreter will still know that a verbalisation is oriented towards his/her amusement. Even if the recipient, whether prepared for a jocular twist (in canned jokes) or not (in shorter units of conversational humour), does not instantly realise that a maxim is not fulfilled, he/she acknowledges this thanks to the on-line processing of the consecutive part of the utterance. The assumption that in the initial stage of interpretation, the nonfulfilment is covert, i.e. that it is a violation, is unfounded, since whole utterances are analysed, not elements thereof, while emergent implicatures are subject to cancellation in the light of new information, as will be discussed in the second part of the chapter. Just as in the case of, for instance, figurative language, maxim nonfulfilment cannot be perceived as violation but as flouting, because the speaker expects the hearer to be able to appreciate maxim nonfulfilment and to draw adequate inferences. Attardo is, however, adamant at refuting the argument about the hearer’s awareness of maxim nonfulfilment. The fact that one is aware of the speaker’s violations does not change the status of the violation, but merely makes them obvious. (Attardo 1993: 545)

This explanation is hardly convincing. In Gricean terms, an obvious, i.e. detected, violation, or rather non-fulfilment, can be nothing but a flout, which the speaker commits so as to gain a communicative aim, which may be to amuse the hearer. On the other hand, genuine violations are those realisations of maxim nonfulfilment which are covert and pass undetected.

3.5. CP violation?: Rationality and cooperative goals The problem of maxim flouting/violation has a bearing on the postulate of the CP violation. Attardo’s reasoning is the result of his misreading of yet another of Grice’s tenets, viz. the convolution of maxim violations and flouts, and the wrong inference that the CP is momentarily violated if a maxim is flouted. (…) when a maxim is being flouted, the violation of the CP is only superficial and temporary, so much as, that the hearer assumes that while the speaker is violating one maxim he/she is still fulfilling the other three. (Attardo 1993: 539)

Firstly, the quotation above testifies to Attardo’s inconsistent use of the terms “flout” and “violation”, which he appears to treat as synonymous.

188

Chapter Five

Secondly, as already observed, that a maxim is flouted does not mean that the remaining maxims (Nota bene, not only three, as Attardo claims, mistaking maxims for maxim categories) must obligatorily be fulfilled. Most importantly, according to the original conceptualisation, flouts occur only because the CP is unfailingly upheld. On no account can the CP be even momentarily violated. Attardo (1993, 1994) wrongly assumes that the CP is normally violated when a maxim flout occurs and claims that in jokes (and humour, on the whole), which operate on maxim violations, there is “no ulterior interpretation of the text that can salvage it from the violation of the maxim” (Attardo 1993: 543, Attardo 1994: 273), i.e. that the CP must be violated as well. It emerges, therefore, that the contention of the CP violation in humour is specious, being contingent on fallacious premises. This is also corroborated by the example the author discusses. Analysing a well-trodden humorous form of interchange (“Do you know what time it is?” “Yes”), Attardo (1993) posits that it relies on the responder’s misinterpretation of the indirect speech act of requesting, which causes him/her to violate the Quantity maxim, while there is hardly any maxim that is followed to meet the request. The affirmative answer reliant on the pretended misinterpretation may not be cooperative in the literal sense of the word, but violates neither the maxim nor the CP. The response is rationally produced and has a communicative humorous purpose, which the interlocutor should normally appreciate. As Grice (1975/1989b, 1989a) emphasises, the CP holds not only in exchanges aimed at conveying informative content. Also, subscribing to Raskin’s (1985, 1998) ideas presented earlier, Attardo avers that the bona-fide mode, i.e. the CP, presupposes “the speaker’s commitment to truth, relevance, clarity, and to providing the right quantity of information at any give time” (Attardo 1994: 274). Given the notion of legitimate flouts of Quality, Quantity, Relation and Manner maxims, this is transparently not what Grice must have intended as the prototypical implicature-yielding realisation of the CP. Understanding the CP in this manner, however, Attardo purports to regard humour, which need not provide true, relevant, clear and quantitatively exact information, as a phenomenon violating the CP. On the other hand, deflecting Kotthoff’s (2006) criticism of his work, similar to this presented above, Attardo (2006: 353-354) affirms that both the observance of maxims (as if they were instructions) and maxim flouts generating implicatures are omnipresent in communication within the CP, which is indeed the essence of the Gricean model. Nonetheless, Attardo (2006: 354) maintains that humorous communication falls outside this model, based on “unredeemed violation to CP” not conducive to implicatures. However, this argument

Grice’s Philosophy and Garden-Path Humour

189

can easily be overthrown, since humorous verbalisations very often do yield implicatures (the meta-implicature invariably being “I want to amuse you”) and will then not violate the CP. Another point adducing evidence against his postulate of the CP violation is Attardo’s (1993, 1994, 2006) misguided understanding of Grice’s “cooperation” as standing vis-à-vis “non-cooperation” understood in the literal sense. In Attardo’s view, humour violates the CP and manifests “non-cooperative behaviour” (Attardo 1994: 275). Following Raskin, Attardo fails to appreciate the foundations of Grice’s conceptualisation of cooperation as rationality. On the other hand, Attardo (2006) emphasises the aspect of goal assumption in cooperation but denies that this phenomenon is also inherent to humorous communication. Cooperation is an adoption of goals. It can happen at the linguistic level (comprehension) or at the extra-linguistic level (cooperation toward goals that are non-linguistic). Neither has any allowance for special ‘humor’ cooperation. (Attardo 2006: 350)

It is difficult to comprehend the underlying premise that humour cannot be subsumed under either of the categories of cooperation. Producing a humorous verbalisation, the speaker normally wants to be understood and has an aim, which is shared by the interlocutor, even if the original utterance is disruptive of the conversation (e.g. trumping or interactional puns) or entails competition (e.g. mutual adversarial teasing). Interestingly enough, outside his humour studies, Attardo (1997b,c) employs a different approach, i.e. that Grice’s cooperation depends on the common purpose or direction of the conversation, as proposed in the basic formulation of the CP. Attardo (1997c) rightly emphasises that cooperation need not exclude competition, and thus conflicting goals, which is why the cooperative behaviour also embraces quarrels, in which conversationalists pursue different goals but have one mutually accepted direction, i.e. to hold an argument. All this is true, but Attardo suggests “broadening the scope of the CP to include a host of competitive situations which it failed to apply previously” (Attardo 1997c: 32), which yields a conclusion that in Attardo’s understanding, the Gricean model, as originally proposed, does not account for competitiveness and different goals, whether or not openly stated. As earlier reported, Grice (1975/1989b, 1978/1989b, 1989a) does allow for such communicative goals. Therefore, no broadening is necessary. Non-cooperative behaviour, according to Attardo (1997c), manifests itself in misleading the hearer, while the latter cannot possibly have a compatibly cooperative goal to want to be misled or lied to. This is indeed

190

Chapter Five

what Grice (1975/1989b) appears to have suggested, emphasising the importance of the first Quality maxim. However, Attardo (1997c) mentions that the second case of non-cooperativeness is joking. Although Attardo acknowledges the pivotal importance of cooperation (Attardo 1997b,c) and rationality (Attardo 2006, 2003, 1997b,c) on the part of the communicators, he fails to ascribe their due significance to the CP model, when applied to humour. Attardo (2006: 259) even highlights, “I never connected my work on pragmatics on the NCP, rationality, the PCP, etc. to my work on humour, since the latter needs only Grice’s CP.” This quotation testifies to Attardo’s miscomprehension of Grice’s principal premise of cooperative rationality. Attardo (1993, 1994, 2006) is oblivious to critiques of commentators asserting that jokes (and other forms of humour) cannot possibly violate the CP, insofar as communication does occur, or that the hearer generates implicatures in accordance with the speaker’s intentions (Kotthoff 2006). Attardo maintains that humour immanently violates the CP. Motivating this claim, Attardo (2006) equates humour with lies (already discussed in the section on the non-bona-fide mode), which are transparently divergent. Lying, the speaker intends the hearer to make an inference without any access to some information, while in humour the whole speaker’s meaning and intent become overt at some point. There is thus little motivation for the statement, “the CP is violated without the intention to let H arrive at an implicature whenever S finds it useful” (Attardo 2006: 350). In humorous communication, the speaker may wish the hearer to appreciate some implicatures (and not other implicatures), the global aim being to allow the latter to comprehend and enjoy a humorous stimulus (and possibly contextually relevant meanings). The same will also be germane to nonhumorous communication. According to Grice’s (1975/1989b, 1978/1989b) model, implicatures are cancellable (to be discussed in the second subchapter) and can be defeated to make way for other implicatures, which is a feature facilitating the occurrence of many forms of humour, especially the garden-path type. Notwithstanding all the above, Raskin and Attardo consent to humour’s rationality and cooperation. Raskin (1985: 104, 1998) does admit that humour is a form of play which obeys mutually accepted rules and is cooperative. Additionally, Raskin and Attardo (1994) do admit that jokes are successful communicative exchanges anchored in cooperativeness and rationality. The speaker and the hearer in a joke context are both actively, consciously – and cooperatively – engaged in joketelling non-bona-fide communication mode (…) hearers perceive the

Grice’s Philosophy and Garden-Path Humour

191

intention of the speaker as an attempt to make them laugh. (Raskin and Attardo 1994: 37)

This means that the hearer acknowledges the speaker’s intention to amuse him/her and willingly follows the train of thought initiated by the speaker, to the effect that the communicative process is perfectly cooperative. Raskin and Attardo (1994) and Attardo (1990, 1993, 1994) do observe the paradox of the CP violation, which should entail “breakdown of communication” (Attardo 1993: 537), and the communicative success of jokes, i.e. their ultimate cooperativeness. Attardo (1990, 1993, 1994) endeavours to explain this paradox by resorting to Raskin’s (1985) earlier postulate of non-bona-fide communication and humour-CP (problematic conceptualisations, as earlier discussed), by which means he aims to avoid admitting that humorous acts are cooperative in Grice’s sense or that they do not violate the CP. Accordingly, the non-bona-fide mode and the humour-CP are claimed to presuppose speakers’ cooperation, which explains why jokes usually do not give rise to any misunderstanding and are appreciated. Attardo (1990: 358) hence maintains that jokes “violate” the maxims and the CP to “achieve a socially desirable effect.” Jokes (and humour, on the whole) violate the CP, because they are not produced within the bona-fide mode of communication, which (supposedly) represents the realisation of the CP as construed by Grice. On the other hand, “What takes one into NBF [non-bona-fide] is the unredeemed violation of the CP” (Attardo 2006: 354). This reasoning is a clear manifestation of circular logic, with one postulate testifying to the other and vice versa. It emerges that the introduction of the second (redundant) mode of cooperative and successful communication involves another (redundant) CP holding for humour, and the explanation of its communicative feasibility. A conclusion can be drawn that the additional mode of communication and another CP for humour are based on dubious and even ill-advised premises and are superfluous. In a humour-oriented communication based on the CP, the maxims may be, and usually are, flouted to meet the intended communicative purpose. The CP obtains in humorous communication as long as the interchange does take place in accordance with a plan rationally pursued by the speaker and the hearer, who fulfil their respective roles. The hearer some point observes the speaker’s humour-oriented intention and lets himself/herself be amused. Like other communicative tasks, humour is a rational goal which the speaker and the hearer try to attain, having their respective complementarily opposite sub-

192

Chapter Five

aims. Both the claim of the CP violation and the claim on which it is based, i.e. that humour is contingent on maxim violation, are ill-conceived.

4. Concluding remarks: the CP and maxim flouts in humour In conclusion, the humour-CP and the non-bona-fide mode of communication appear to be redundant terminological complications, since jocular communication nicely falls into the Gricean model. It transpires that upholding the postulate of the violation of maxims and of the CP in humour validates the introduction of the non-bona-fide mode, and thus dissociation from the CP. The latter easily accommodates humour as long as it is rationally produced and does not entail covert mendacity. The parallel framework for humour is not divorced from the original one, also pivoting on rationality and communicativeness. The major difference is that it is reconstructed to provide a formula for how to “tell an effective joke” (Attardo 1993: 544). In the light of the discussion above, it should be evident that Raskin and Attardo misinterpret the Gricean model of communication in a considerable number of ways. As a result, they launch the superfluous model burdened with various theoretical complications, following circular logic. The concepts of the CP and maxim violations together with the nonbona-fide and its separate CP motivate each other but contribute nothing new to the model of rational communication. Ultimately, what governs interactants, whether humorous or serious, is conversational rationality. Even if humorous verbalisations may entail genre-specific illogicalities, absurdity, ostensible deceits, etc., they are still rationally produced, the underpinning aim being to amuse the hearer. Humour then conforms to the CP but exists thanks to Maxim flouts, which are legitimate in the original Gricean model and inherent to communication, also humorous in nature. The CP framework presupposes the abundance of flouts, since they help obtain implicatures on the assumption of the conversationalists’ intention to preserve the CP, understood as the intentionality and rationality of communicative behaviour. The speaker employs flouts in humorous verbalisations for the sake of the rational meta-aim of amusing the hearer and its further benefits for interpersonal communication and, optionally, the conveyance of informative content. Staying within the CP model and making rational inferences, the hearer also presupposes that the speaker is rational and cooperative, thus compliant with the CP. So does the hearer.

Grice’s Philosophy and Garden-Path Humour

193

It is not the aim here to engage in a detailed analysis of various types of humour, pointing to particular maxim flouts, as they have an inexhaustible number of manifestations. For example, Manner and Quantity flouts are inherent to humour, for it is very often couched in semantic ambiguity or lack of clarity of expression, whether a short humorous chunk, such as a neological lexeme, a comparison or a conversational witticism. The Relation maxim flout always occurs in interactional puns, while absurd humour or friendly teasing resembling putdown humour will often flout the first Quality maxim. Below, three random examples of conversational humour are discussed to show that it violates neither the CP nor the maxims but only flouts the maxims to achieve various communicative effects. Asked about the fortune he has, an actor by the name of George Raft said wittily but evasively, “Part of the $10 million I spent on gambling, part on booze and part on women. The rest I spent foolishly.”9

In his answer, the actor flouts the maxim of Relation (given the question), the first maxim of Quality, and the maxims of Manner, ultimately implicating his unwillingness to discuss his properties and simultaneously entertaining the audience. The maxim of Quality is exploited in two ways, i.e. the actor is transparently not telling the truth, saying he has no fortune, and, secondly, is overtly untruthful, pretending to genuinely believe that spending money on alcohol, gambling and women is not foolish. Its inferential complexity notwithstanding, this witticism can still be interpreted within the Gricean model. An SMS interchange between two young men, who engage in banter and try to outwit one another. A: I’ve just heard that tonight a UFO will kidnap all handsome men, but you needn’t worry, obviously. I just want to say good-bye. B: Well, I’m already in their flying saucer. Yesterday they took those with a very high IQ.

The exchange is initiated with an ostensibly hurtful remark hinged on Quality maxim and Manner (“Be orderly”) maxim flouts. Basing his utterance on a transparently false premise, i.e. a UFO wishing to kidnap someone, the first interlocutor initially purports to offer a warning (the 9

A similar quote can be found in the 2007 Oscar winning “No Country for Old Men” by the Coen brothers. Asked for the money he took, the protagonist says, “It’s too late. I spent it. About a million and a half on whores and whiskey and the rest of it I just sort of blew it here.”

194

Chapter Five

supposed Quantity flout), only to implicate that the other is not handsome enough to be kidnapped. Thanks to the maxim flouts, one of the men communicates his willingness to amuse the other, who engages in the humorous frame and produces a parallel verbalisation, also flouting the first maxim of Quality. “Banning the bra would be a big flop,” says a woman during a conversation on lingerie.

The pun above flouts one maxim of Manner (“Avoid ambiguity”) in order to communicate two relevant meanings based on the polysemy of the word “flop” via a witty comment. Likewise, canned jokes capitalise on egregious flouts of all the maxims. Unless the jokester signals that a joke is to come by saying, “Do you know this joke?” or “Have you heard the one about…?”, the joke will also flout the maxim of Relation. Nota bene, signalling the joke does not entail suspending the maxims, as Perlmutter (2002) asserts, but explicit delineation of the current purpose of the exchange. Returning to the main focus of interest, the creation and perception of GP humour are also rationally cooperative activities, as the listener participates in the humorous process and draws inferences, in accordance with the speaker’s plan, in order to experience humour. The hearer finds himself/herself having made a “wrong” inference and having been led to do this by the speaker but appreciates the latter’s global aim, which is benevolent (with the exception of disparaging/putdown humour based on genuine deception of the addressee, the butt, to the amusement of the meta-recipient, who can appreciate the real import of the verbalisation). In the case of canned jokes, whether signalled by the speaker or not, the speaker’s intentions are almost instantly recognised, due to the specificity of this humorous form. The act of GP joke telling can be metaphorically compared to a walk in an uneven seaside landscape (Paulos 1980).10 Despite knowing a safe path, the speaker guides the hearer towards the sea on a trajectory which leads the latter over the cliff and ensures a surprising discontinuity. What Paulos (1980) fails to observe is that listening to a joke, the hearer tacitly expects the guide to be deceitful. Moreover, it is both the speaker and the hearer that are willing conspirators. The hearer participates of his/her own accord in taking the

10

As discussed in chapter 3, the author presents the theory as if it was applicable to all types of jokes. However, it is here maintained that it accounts only for the GP type.

Grice’s Philosophy and Garden-Path Humour

195

path of a joke, even if presupposing that he/she will encounter a dangerous cliff (cf. Veale 2004). The situation will be different in the case of the two categories of conversational humour discussed here, i.e. witticisms and one-liners, which are not normally signalled so that the humorous mode and the GP mechanism can take the recipient by surprise. However, the hearer will take maxim nonfulfilment, eventually emerging as overt, in his/her stride and analyse the text (just like other indirect realisations flouting maxims, e.g. metaphors), presupposing that the speaker is rational, and thus cooperative in the Gricean sense. The hearer acknowledges that the ostensible deception is still to his/her own benefit, appreciates the positive humorous surprise and recognises the speaker’s jocular intention, which may so far have been covert, there being no cues, verbal or non-verbal. In other words, although the addressee is somehow manipulated in the process of inferencing and led up the garden path, he/she is not genuinely a victim, as the whole procedure is employed to his/her benefit, i.e. humorous appreciation. On the other hand, the speaker rationally instigates the GP process, expecting the hearer to draw “wrong” inferences, letting himself/herself be deluded, only to arrive at the “right” meaning of a verbalisation, ambiguous, as it turns out to be in on-line interpretation. In essence, GP phenomena rely on “the active participation of the readers or hearers, and their willingness to let themselves be manipulated” (Mey 1991: 242), while the outcome is most benevolent, given that “the hearer is left with the feeling of having been taken in - but not necessarily in an unpleasant way, since it was, after all, a joke (and not even a practical one at that)” (Mey 1991: 236). For example, in a conversation on looks held by middle-aged female friends, one interactionist may say, I still have the body of an 18-year-old. It’s in my trunk.

What may initially sound like a boastful remark is turned into a humorous comment, which is couched in ambiguity (Manner maxim flout) and transparently flouts Quality maxims, inasmuch as it is hardly imaginable that one of the women should be a murderess. It emerges that the witty speaker must have produced this utterance with a view to amusing the conversationalist (and possibly also to convey implicitly her willingness to comfort the latter or change the topic of conversation). What is significant, discussing the developing stages of intentional meaning production and reception, Grice (1982/1989b) lists the stage of joined play or games, at which both participants are conscious that they are engaged not in deception but in faked deception, since the communicator makes it clear that a deceptive action has taken place.

196

Chapter Five

Appreciating this fact, the interpreter joins in the play with the speaker. To illustrate this stage, Grice (1982/1989b) makes use of behaviours which, if produced nonvoluntarily, epitomise natural meanings. However, for the sake of the present discussions, the behaviours could be moved to the level of nonnatural meanings, and hence language games used for humorous purposes. Postulating a game as a rational activity between interlocutors, Grice cannot have been averse to the idea of humorous communication, which may consist in deliberate GP deception of the hearer, who lets himself/herself be deceived in order to experience humorous effects. The game may be differently structured. It can be a complex exchange of humorous verbalisations, as is the case of multi-turn teasing, one longer verbalisation in the form of a canned joke, listening to which the hearer presupposes a surprising punchline, or a short phrase/witticism, which may be most surprising if occurring in non-humorous discourse. In conclusion, the process of GP humour comprehension is a complex one but does take place along rational lines, as argued by Grice, and conforms to the model of rationality manifesting itself in humour-oriented cooperativeness. The next part of the chapter will aim to analyse the humorous GP process in detail, accounting for the cancellable inferences in the form of Grice’s generalised conversational implicatures and their explanatory potential.

5. Generalised conversational implicatures in GPs The first part of the chapter aimed to show that the Gricean model of communication operating on the CP, together with direct or implicated meanings, i.e. implicatures consequent upon maxim flouts, is by no means divorced from humorous communication. This subchapter aims to discuss the workings of the GP mechanism within the Gricean model of communicated meanings, whose fundamentals must be carefully revisited. This analysis of GPs takes as the departure point the premise that pragmatic inference in the form of the generalised conversational implicature (GCI) is initially generated irrespective of contextual factors, simultaneously being cancellable, i.e. defeasible, in the light of clashing contextual (here, only co-textual) factors. This is why the initial interpretation of an utterance can be overridden by new co-textual evidence, which is the sine qua non for the operation of GPs. It will also be borne out that Grice’s categorisation of meanings and the conceptualisation of ambiguity give rise to several problems, which must be addressed before the revised GCI is applied to the analysis of GPs. Although Grice’s theory of meaning and the theory of conversation are

Grice’s Philosophy and Garden-Path Humour

197

often treated as if they were divorced from each other, they are evidently interdependent (Neale 1992).

6. Meaning One of the primary concepts introduced by Grice is the nonnatural meaning (meaningNN), juxtaposed with natural meaning (Grice 1957/1989b, 1982/1989b). Grice (1982/1989b) uses the latter in reference to any stimulus which, inherently and by nature, conveys some information, e.g. black clouds mean forthcoming rain, some symptoms mean a heart disease, groans mean pain, etc. Other meanings, with language as their epitome, are nonnatural meanings, meanings which are intentional but not naturally conveyed, e.g. crossed fingers mean wishing someone good luck. Nonnatural meanings conveyed by words serve intentional and rational communicative purposes and the realisation of the Cooperative Principle (Grice 1975/1989b). As already discussed in the previous subchapter, speakers engage in communication in anticipation of achieving certain goals, rationally following communicative strategies and expecting those to be followed by their interlocutors. Therefore, the effect of an utterance depends on what the speaker wishes to communicate, while the hearer’s recognition of this wish is immanently prior to the arrival at the nonnatural meaning. In Grice’s own words, “‘A’ uttered x with the intention of inducing a belief by means of the recognition of this intention” (Grice 1957/1989b: 219). Also, Grice (1969/1989b) distinguishes several levels of meaningspecification, namely timeless meaning, applied timeless meaning, utterance-type occasion-meaning and utterer’s occasion-meaning. The speaker’s meaning, which appears to be Grice’s (1975/1989b, 1978/1989b) focus of interest, is conceptualised in terms of the utterer’s intention to produce a particular response in the hearer on the strength of the hearer’s recognition of the speaker’s intention. There is thus no possibility of unintentional meaning occurring in the Gricean model of communication. The most significant division of meaning is that between what is said (p) and what is implicated (q), i.e. implicatures, further divided into what is conventionally implicated and what is nonconventionally implicated, which may be conversationally implicated (Grice 1968/1989b, 1978/1989b). Grice does not discuss any cases of nonconventional implicature other than the conversational one, which leaves a lacuna in the model.

198

Chapter Five

The underlying premise of this meaning categorisation is that an utterance produced literally and intentionally, i.e. what is said, does not need to exhaust the speaker’s intended meaning. Therefore, the starting point is the observation that what people say and what they mean do not invariably coincide, and meanings are often implicated. Both what is said and what is implicated contribute to the realisation of the CP and convey the speaker’s meaning (Grice 1969/1989b: 86). However, Davis (1998) argues that the parameter of intentionality cannot be found in any of Grice’s discussions on implicatures per se. Even if this is the case, the idea of intentionality does lie at the heart of the entire model of communication propounded by Grice, because implicatures are generated from maxim flouts and the CP, which are rooted in intentionality. It should also be remembered that conventional meanings help determine what is said and, as a result, what is implicated as well (Grice 1975/1989b). SPEAKER’S MEANING (total signification of an utterance) p p SAID IMPLICATED p p CONVENTIONALLY NON-CONVENTIONALLY p p CONVERSATIONALLY otherwise? p p GENERALISED PARTICULARISED IMPLICATURES IMPLICATURES (cf. Grice 1969/1989b, 1975/1989b; Levinson 1983)

6.1. What is said What is said (1969/1989b, 1975/1989b, 1989a), or the said, is not to be interpreted in the folk sense of the term, i.e. verbalising or uttering a string of sounds. What Grice means is the import which the speaker intends to communicate but which stands vis-à-vis what is conveyed via implicatures. It may be inferred that what is said is the coincidence of the speaker’s meaning with the conventional linguistic meaning. It could hence be conceptualised as the intended meaning compatible with semantic meaning or what Grice calls standard meaning (Grice 1957/1989b: 216) or most frequently conventional meaning (Grice 1975/1989b, 1978/1989b). Also, Grice (1975/1989b: 25) appears to equate

Grice’s Philosophy and Garden-Path Humour

199

conventional meaning with literal meaning. What is crucial, however, non-literal meaning can also be conventional, which is the case of dead metaphors or other idioms. Those, on the other hand, do not yield implicatures (Grice 1978/1989b: 50). Should the said not suffice as the total signification of the speaker’s meaning, it serves as the input for implicatures, which come into play subsequently. Many authors (e.g. Levinson 2000, Carston 2002) differentiate between what is said and implicatures of all kinds along the parameter of truth-value, which the latter need not manifest. This dichotomy corresponds to the semantics vs. pragmatics distinction. The “said” can be taken to be truth-conditional content, the proposition expressed, the output of the process, of semantic interpretation; the proper domain of a theory of linguistic meaning. The “implicated” can be taken … to include all the processes of pragmatic inference; it is the proper domain of a theory of communication. (Levinson 2000: 186-187)

It should be remembered, however, that Grice’s what is said and what is implicated are subordinate to the notion of the speaker’s meaning. Given that the concept of intentional meaning is central to his concept of communication, Grice is a theorist of communication-intention and strongly opposes the “truth-conditional approach” pursued, for example, by Frege (cf. Grice 1957/1989b, 1982/1989b). Many authors argue that the status of what is said is somewhat vague (e.g. Bach 1994, Levinson 2000). Carston (2002, 2004) observes that at some points in his discussion of non-literal language, Grice associates the level of what is said with what the speaker means beyond saying, i.e. makes as if to say. Indeed, Grice mentions that “Nothing may be said, though there is something which a speaker makes as if to say” (Grice 1978/1989b: 41). Making as if to say corresponds to two tropes (rhetorical figures), viz. irony and litotes. Carston (2002) also extends making as if to say over metaphor and hyperbole only to find this concept problematic, claiming that nothing is said at all and hence no implicatures can even be generated, since they normally rely on what is said. However, this is certainly not what Grice (1975/1989b: 32-33) conceives of, presenting tropes as dependent on maxim flouts. Also as Grice (1975/1989b: 34-35) states, irony and litotes (and other figures) reside in implicatures consequent upon maxim flouts. The phenomenon of making as if to say should be embraced by the notion of implicature, which must have been Grice’s intention from the outset. A greater problem is the alleged intrusion of implicatures into what is said. Grice lists three conditions for determining what is said, namely

200

Chapter Five

reference resolution (identifying the referents), indexical fixing (fixing deictic parameters) and the disambiguation of a semantic string. Levinson (2000) also expands the list of what is said’s determinants by ellipses unpacking and generality narrowing. Many researchers subscribe to the opinion that what is said can actually be determined by implicatures and, consequently, that the same inferential process which underlies implicature computation underlies the computation of what is said (e.g. Levinson 1983, 2000; Carston 1988, 1993; Recanati 1989, 1993). Levinson (1983, 2000) dubs this phenomenon Grice’s circle. Grice’s account makes implicature dependent on prior determination of ‘the said’. The said in turn depends on implicature: it depends on disambiguation, reference fixing, …[etc.]. But each of these processes, which are prerequisites to determining the proposition expressed themselves, depend crucially on implicatures. Thus what is said seems both to determine and to be determined by implicature. Let us call this ‘Grice’s circle’… the theory of linguistic meaning is dependent on, not independent of, the theory of communication. (Levinson 2000: 186-187)

Other researchers also appreciate the vague status of what is said and put forward their own definitions. Bach (1994, 2001), for instance, resolves the problem by postulating a three-fold distinction between what is said (as purely semantic meaning), the impliciture and the implicature (see section 2, chapter 6), while Recanati (1989) holds a view that pragmatic resolution and strengthening should count as what is said, thus equating it with the generalised conversational implicature. Probably, the most prominent contention is that of explicature (Sperber and Wilson 1986/1995), the informational enrichment of semantic representation. Nota bene, Levinson (2000) observes that generalised conversational implicatures would count as explicatures, which would then entirely undermine the Gricean perspective. Levinson (2000) squarely criticises relevance-theoretic argumentation, proving that the introduction of the notion of explicature provokes nothing but terminological multiplication, since both explicatures and implicatures are pragmatic inferences affected by contextual factors and are subject to cancellation.

6.2. Implicature types and their features The implicature is Grice’s neological term for implying, while implicatum stands for the meaning implied (Grice 1975/1989b). “‘Implicature’ is a blanket word to avoid having to make choices between words like ‘imply’, ‘suggest’, ‘indicate’, and ‘mean’” (Grice 1969/1989b:

Grice’s Philosophy and Garden-Path Humour

201

86). Before coining the concept, Grice (1961/1989b) will have already conceptualised the idea, speaking of something that is implied but different to something stated. In essence, by implicatures, Grice (1981/1989b) understands intended meanings beyond the said. Although, according to Grice’s original parlance, “implicature” refers to the process of implicating leading to implicatum, the latter term appears to have been neglected in literature, with the former term standing for the inference made. This is hardly surprising, because, in Grice’s ideal model of communication, the hearer’s inference should coincide with the speaker’s implied meaning. An implicature is what the speaker “implies, suggests, hints or in some other less than fully direct manner conveys” (Grice 1989a: 360). Although Grice omits to state it explicitly, it is at the level of what is said that maxim flouts take place, offering foundations for emergent implicatures. In other words, what is implicated transcends, but is normally “built” (Grice 1978/1989b) on, the level of what is said. Presuming the speaker to be cooperative, the hearer makes inferences contingent on what is said to reach an interpretation which will prove rational. However, it is not invariably so that the implicature is built directly upon what is said (Hawley 2002). Even Grice (1975/1989b: 35) provides an example in which a moment of stunned silence after a putdown carries the implicature that a faux pas has been committed. An implicature can be generated on the basis of silence, the said being non-existent. However, some of Hawley’s (2002) examples manifesting implicatures independent from what is said are not entirely plausible, because what is said must be computed first, e.g. “I can’t see you! Are you still there?” said by someone who wants his friend to turn on the light. Undoubtedly, in result of the inadequacy of the said generated on the basis of the utterance, it needs to be reassessed in pursuit of an implicature, on the understanding that all the categories of maxims have been flouted. This is what Grice (1975/1989b: 39) must mean as an implicature not being carried “by what is said but only by the saying of what is said” (see the discussion on centrality at the end of section 6.2.2). Most importantly, implicatures are further divided into the conventional implicature, which coincides with the conventional context-free meaning intended by the speaker, and the conversational implicature. 6.2.1. Conventional implicature The conventional implicature (CI), just like what is said, derives from conventional (encoded) meanings. All the same, Grice argues that the CI is

202

Chapter Five

determined by the yield of what is said but cannot be equated with it. In other words, CI refers to what is coded linguistically but transcends what is said. To illustrate this category, Grice provides a few examples: “moreover”, “but” and “therefore” (1968/1989b: 120-122), as well as “on the other hand” and “so” (1989a: 361-362). Other representatives of this category are particles such as “even” or “too” (Horn 2004). Obviously, the list can be further expanded (see Levinson 1983). CIs are thus dictated by the lexical, conventional meanings of the words used, which are impervious to conversational circumstances. Also, similarly to conversational implicatures, CIs do not have a truth-conditional content. On the whole, the concept of the CI is not very well developed and its status never ceases to be questioned. Many (e.g. Bach 1999, Levinson 2000) contend that Grice’s distinction between what is said and what is implicated clarifies the boundary between semantics and pragmatics. A question arises, therefore, as to the status of the conventional implicature. Bach (1999) criticises this notion and even casts doubt upon the existence of such a phenomenon, postulating that it unnecessarily complicates the distinction between what is said and what is implicated. Attempts have been made to reduce it to entailment and truth-conditional meaning (Sperber and Wilson 1986/1995, Carston 2002), presupposition or generalised conversational implicatures (e.g. Kempson 1975, Bach 1999, Chierchia and McConnell-Ginet 2000, Karttunen and Peters 1979). If it exists at all, the phenomenon of conventional implicature appears to be relatively narrow in scope, covering a limited number of lexemes carrying their conventional meanings via implication. Inasmuch as conventional implicature is not subject to cancellation, the primary feature of inferences exploited in GPs, it is not further elaborated here. 6.2.2. Conversational implicatures It is worth noting that Grice (1975/1989b) introduces the conversational implicature as a subclass of the nonconventional implicature, which suggests that there are also other types of nonconventional implicatures besides the conversational type. No discussion of any other nonconventional type is provided, though. A man who, by (in, when) saying (or making as if to say) that p has implicated that q, may be said to have conversationally implicated that q, provided that (1) he is to be presumed to be observing the conversational maxims, or at least the Cooperative Principle; (2) the supposition that he is aware that, or thinks that, q is required in order to make his saying or making as if to say p (or doing so in those terms) consistent with this

Grice’s Philosophy and Garden-Path Humour

203

presumption; and (3) the speaker thinks (and would expect the hearer to think that the speaker thinks) that it is within the competence of the hearer to work out, or to grasp intuitively, that the supposition mentioned in (2) is required. (Grice 1975/1989b: 30-31)

Conversational implicatures fall outside the scope of conventional meanings of the words used, emerging as nonce inferences capitalising on the CP and maxim flouts, as opposed to CIs, which are determined by the conventional use of language. Nevertheless, Grice (1978/1989b) does concede that nonconventional implicatures are reliant on what is said, which is built on the conventional meanings of the words used. Grice observes the emergent paradox. If we, as speakers, have the requisite knowledge of the conventional meaning of sentences we employ to implicate, when uttering them, something the implication of which depends on the conventional meaning in question, how can we, as theorists, have difficulty with respect to just those cases in deciding where conventional meaning ends and implicature begins? (Grice 1978/1989b: 49)

Grice appears to be voicing his doubt as to whether he can substantiate his claim convincingly enough but argues that if words do not bear senses which they are supposed to, the result is an implicature or an idiom (Grice 1978/1989b: 50). The latter case could also include such linguistic phenomena as set phrases or collocations, sayings or proverbs, which do not mean directly what they say but should not be regarded as implicatures. Admittedly, all those will have come into being as nonconventional implicatures only to become conventionalised expressions, which Grice (1975/1989b: 39) also mentions. This leads to a vital observation that the meanings of phrasemes, very often recruited in GPs, should not be looked upon as (generalised) conversational implicatures. Conversational implicatures have two manifestations, i.e. the generalised conversational implicature (GCI) and the particularised conversational implicature (PCI) (Grice (1975/1989b). These are distinguished on the basis of the criterion of the presence/absence of context interference during the process of meaning comprehension. The latter subtype comes into being irrespective of contextual factors. A GCI is normally carried by particular words unless special circumstances arise. (…) one can say that the use of a certain form of words in an utterance would normally (in the absence of special circumstances) carry such-andsuch an implicature or type of implicature. (Grice 1975/1989b: 37)

204

Chapter Five

A PCI must always be weighed by the speaker against particular contextual factors of a given speech event. Therefore, it arises in cases in which an implicature is carried by saying that p on a particular occasion in virtue of special features of the context, cases in which there is no room for the idea that an implicature of this sort is normally carried by saying that p. (Grice 1975/1989b: 37)

To reformulate, a PCI emerging from an utterance operates on specific contextual assumptions, which do not obtain across all contexts, while a GCI is immanently evoked by a particular formulation, unless contextual factors point to the contrary. It should be mentioned that PCIs are better expounded on and amply exemplified, since the bulk of Grice’s discussion of implicatures generated upon maxim flouts in the section “Examples of Conversational Implicature” (Grice 1975/1989b: 31-37) is devoted to PCIs. The discussion on GCIs (1975/1989b: 37-38), which are in focus here, is much more succinct, leaving them underdefined. Some authors also acknowledge the insufficient theoretical motivation for generalised and particularised conversational implicatures (Carston 2002, Hirschberg 1991, Geurts 1998, Neale 1992), pointing out that both the types rely, to a certain extent, on contextual factors. There are also attempts to unify the two types of conversational implicatures (e.g. Carston 1998, 2004; Sperber and Wilson 1986/1995). These critics advocate the context-driven approach, according to which both generalised and particularised implicatures are generated with context support, arousing strong opposition of other authors (e.g. Levinson 2000) who champion the original dichotomy. The latter belief is also espoused here. Even if, from a pragmatic perspective, context may be important for both the types of conversational implicatures, the particularised one cannot exist without it and is determined by it. By contrast, the generalised type is generated across contexts. Among other developments of the Gricean approach, there is the contention that the PCI is necessarily the speaker’s implicature, while the GCI is a sentence implicature (Davis 1998). However, such a view invalidates Grice’s rudimentary premise concerning the speaker’s intentionality and places GCIs dangerously close to conventional meanings. Also, any GCI is entirely dictated by the speaker’s intention, even if it is ascribable to many speakers and is relevant across contexts. This observation is particularly vital in the case of GPs repeated verbatim, which will yield the same inferences. Admittedly, as a form of development upon his original model of meaning types, Grice (1989a: 361) advances the parameter of centrality of

Grice’s Philosophy and Garden-Path Humour

205

meaning, which might be interpreted as contributing to the analysis of implicatures, although it appears to be neglected in pragmatic literature. The parameter in question manifests itself in formality (the generated meaning partakes in the conventional meaning of the signifying expression) and dictiveness (the generated meaning is part of what the signifying expression says, not implies), differentiating between primary and nonprimary ranges of signification. Grice hence foregrounds another problem relevant to the notion of implicated meanings. Namely, the speaker may produce a statement which can be described by both, either or even neither parameter. Meanings can be formal or informal, dictive and non-dictive. These four types can obviously be intersected. Admittedly, the combination of formality and dictiveness yields what is said, while all the other combinations of the two parameters generate implicatures. Formality without dictiveness will pertain to cases when “part of what the words conventionally mean without being part of what the words say” (Grice 1989a: 361), e.g. “He is a prominent politician. On the other hand, he is exceptionally honest.” implies that politicians are normally dishonest. Conversely, words may be dictive, but their content may be nonformal and absent from their conventional meaning, which yields dictiveness without formality, e.g. “He is a prominent businessman.” to mean that he is wicked and dishonest. Finally, there are cases of nondictiveness coupled with informality, e.g. “Yes, madam,” said to a capricious businessman who makes his discussants comply with all his demands. Nota bene, some authors (Haugh 2002, Davis 1998) claim that all of Grice’s (1975/1989b) examples of implicatures entail indirection, i.e. “the act of meaning or implying something by saying something else” (Davis 1998: 5). Arguably, this indirection, as meant by the authors, could be equated with Grice’s conceptualisation of nondictiveness. However, not only such examples can be found in this lecture, for dictive implicatures are also discussed (e.g. Grice 1975/1989b 37-38). A most important division of implicatures is then into those which contribute to and enrich what is said, (partly) retaining its import (e.g. “Do you think biting your nails helps you?” to mean “Stop biting your nails.”), and those which communicate assumptions entirely independent of what is said (e.g. “Venus de Milo must have started like this.” to mean “Stop biting your nails.”). In his two lectures on logic and conversation, Grice (1975/1989b, 1978/1989b) illustrates GCIs with the former type and PCIs with the latter.

206

Chapter Five

6.2.3. Features of implicatures Grice (1975/1989b, 1978/1989b, 1989a) proposes a number of features of conventional and conversational implicatures. The properties attributed to conversational implicatures are non-conventionality (not being part of the conventional meaning of linguistic expressions), calculability (the potential for being inferred), non-detachability (the potential for being generated even if words used originally are substituted with different ones, as implicatures are not attached to particular words but correspond to conversational rules) and cancellability (the potential for being annulled by additional premises, in neo-Gricean studies equated with defeasibility) (Grice 1975/1989b, 1978/1989b). By contrast, conventional implicatures are conveyed by standardised means of expression and, therefore, are motivated by convention, i.e. they are conventional. They are also noncalculable, either in the light of pragmatic principles or contextual knowledge, and non-cancellable, because they do not rely on cancellable contextual assumptions. Finally, they are detachable, since they are couched in particular linguistic items which can be omitted, whereby their meanings are no longer carried. Upon closer examination, one may discover that the features are not actually the determinants of implicatures (Sadock 1978/1991). Grice (1975/1989, 1978/1989b, 1981/1989b) also does harbour his reservations about the methodological status of the parameters. He is unsure whether “it is possible, in terms of some or all of these features, to devise a decisive test to settle the question whether a conversational implicature is present” (Grice 1978/1989b: 42), yet he hypothesises, “at least some of them are useful as providing a more or less prima facie case in favour of the presence of a conversational implicature” (Grice 1978/1989b: 43). In essence, the features typical of either type of implicature do not serve as infallible tests for its occurrence, but they do act in favour of its presence (Grice 1978/1989b). Grice (1981/1989b) admits that the final test is the investigation into the presence and understanding of, as well as the rationale for, an implicature. Sceptics would undermine such an argument, claiming that the key to proving a presumed implicature to be a given implicature is only one’s wish and willingness to prove it. The calculability of conversational implicatures refers to their potential for being generated in the light of the linguistically coded content of the utterance, the Cooperative Principle and its maxims as well as the verbal and non-verbal context of the utterance, background knowledge and the mutual assumption that all the parameters are available to both participants of a verbal exchange (Grice 1975/1989b). As already discussed in the

Grice’s Philosophy and Garden-Path Humour

207

preceding part of the chapter, the maxims subsumed under the four categories are the desiderata serving as the fundament of cooperative talk. Grice emphasises that conversational implicatures characteristically originate from maxim flouts and revolve around the presumption that the CP is not violated (Grice 1975/1989b, 1978/1989b, 1981/1989b). The CP should be observed both at the level of what is said and at the level of what is implicated (Grice 1978/1989b). Irrespective of whether the speaker’s utterances do or do not proceed according to the maxim specifications, the interpreter presupposes that the CP invariably holds, while flouts are employed deliberately to produce implicatures. (Non)detachability is propounded as another feature distinguishing between conventional and conversational implicatures (Grice 1978/1989b: 44). Specifically, conventional implicatures are detachable, while conversational implicatures are non-detachable. As Grice argues, “insofar as the manner of expression plays no role in the calculation, it will not be possible to find another way of saying the same thing, which simply lacks the implicature in question” (Grice 1975/1989b: 39). The nondetachability of the conversational implicature is a property enabling it to emerge in the same form, even after the replacement of the words used in the original utterance with synonyms, i.e. expressions with the same linguistically coded content. Moreover, it is impossible to find another way of saying the same thing without the implicature in question (e.g. “try to do it” carries the same implicature as “endeavour”/“attempt to do it”). Grice allows exceptions to the rule, though, “where some special feature of the substituted version is itself relevant to the determination of an implicature (in virtue of one of the maxims of Manner)” (Grice 1975/1989b: 39). Grice then explicitly negates the applicability of the test to Manner maxim implicatures, which is the case of the present analysis of GP humour, which flouts the second maxim of Manner, yielding ambiguity. Also, “there may be no alternative way of saying what is said, or no way other than one which will introduce peculiarities of manner, such as by being artificial or long-winded” (Grice 1978/1989b: 43). On the whole then, nondetachability is not an infallible test for a conversational implicature. Cancellability is the most important feature in the context of the present work on GPs, which appear to be anchored in this feature. Grice (1987/1989b) proposes that any conversational implicature, whether generalised or particularised, can be contextually cancelled without giving rise to contradiction, when the incoming text appears incompatible with it or when the contextual factors, later revealed, clash with it. By contrast, conventional implicatures are not cancellable without causing a contradiction. The cancellability test appears to be most reliable, even if

208

Chapter Five

Grice (1978/1989b) admits that it is not a sufficient test for an implicature. The exception Grice discusses is “loose or relaxed” (1978/1989b: 44) use of words yielding conventional implicatures. Sadock (1978) perceives cancellability as the most (but not perfectly) reliable parameter differentiating between conversational implicatures and conventional ones, making only one reservation. Sadock (1978: 290) adds that the shortcoming of the cancellation test is that it cannot distinguish between implicatures and homonymous words (and thus perhaps also polysemous words). These two phenomena are the foundation for many GPs, which leads to the conclusion that the process of cancellation underlying such humorous units cannot be explained within the Gricean model of GCIs. However, Grice (1975/1989b) explicitly discusses two cases of ambiguity (one polysemous, the other homophonous) which engenders implicatures (see the discussion in the next section). It cannot be denied, however, that lexical ambiguities carry two conventional meanings. Therefore, their status as implicatures is questionable. To summarise, conventional implicatures are neither calculable nor cancellable but are, however, detachable. Conversely, conversational implicatures are calculable and cancellable but not detachable, except for those induced by Manner maxims’ flouts. The detachability parameter will then be inapplicable to GP humour which intrinsically entails ambiguity of expression. The most important parameter for the present analysis is cancellability, which is inherent to all GPs. The pending question is whether all cancelled inferences coincide with conversational implicatures as conceived by Grice. Before this issue is addressed, Grice’s view of ambiguity, the second prerequisite for GPs, needs to be revisited.

7. Ambiguity Central to the present analysis of GPs is the notion of ambiguity, which reappears on various occasions in Grice’s (1957/1989b, 1975/1989b) writings. It is crucial to observe that in his analyses of linguistic ambiguity, Grice focuses only on homonymy, polysemy and homophony of single words. Grice appears to have held a twofold view of the phenomenon, analysing it at the levels of what is said and what is implicated. Admittedly, the incompatibility, if not mutual exclusiveness, of the two approaches to ambiguity could be reconciled by the postulate that one occurs prior to determining the speaker’s meaning, i.e. before the level of what is said is established, while the other occurs at the stage when the speaker’s intention is already deciphered. However, in the light of Grice’s contention, it is impossible to provide a theoretical rule

Grice’s Philosophy and Garden-Path Humour

209

discriminating between the two types, other than analysing particular examples and determining the speaker’s intent.

7.1. Disambiguation before determining what is said In his discussion on nonnatural meaning, Grice (1957/1989b) allows for the fact that doubts may arise as to what the speaker aims to convey, due to the existence of competitive conventional meanings, in which case the context, whether linguistic (co-text) or not, determines the alternative germane to what the speaker is saying, viz. what his/her intention is. Context is the determinant of the choice between the two lexical senses of the word “pump” (Grice 1957/1989b: 222). Again, in cases where there is doubt, say, about which of two or more things an utterer intends to convey, we tend to refer to the context (linguistic or otherwise) of the utterance and ask which of the alternatives would be relevant to other things he is saying and doing, or which intention in a particular situation would fit in with some purpose he obviously has (e.g. a man who calls for a pump at a fire would not want a bicycle pump). (Grice 1957/1989b: 222)

In Grice’s view, this disambiguation takes place prior to determining what is said. Neale (1992) questions such a reading of Grice’s contention, pointing to his mention of a “relevant” interpretation, which suggests the impact of the Relation maxim, and hence implicatures, on the determination of what is said. This corresponds to Grice’s circle (Levinson 2000), i.e. the interdependence between linguistic meaning and pragmatic inferences in the process of utterance comprehension. Clearly, from Grice’s (1975/1989b, 1978/1989b) perspective, however, disambiguation, i.e. sense selection, takes place at the level of what is said, being determined by conventional linguistic meaning and context. Grice (1975/1989b) dissociates the derivation processes of what is said from the force of conversational maxims. Those yield implicatures when applied subsequently to the already determined what is said. Therefore, disambiguation (and reference assignment) cannot be guided by the conversational maxims responsible for implicatures. This type of ambiguity is resolved prior to the determination of the speaker’s meaning, viz. the meaning intended by the speaker. Consequently, such ambiguities must be unintentional, while the speaker means only one of the competitive meanings. Moreover, Grice’s (1957/1989b) early use of the term “relevant” cannot be equated with what he later discusses in reference to the Relation

210

Chapter Five

maxim (Grice 1975/1989b), which is substantiated by the exploration of a similar ambiguity issue, when maxims are already fully-fledged. Accordingly, analysing the ambiguous phraseme “in the grip of vice”, Grice (1975/1989b: 25) appreciates its two competitive meanings, which need to be elucidated before determining what is said, without any impact of maxims yielding implicatures. To identify what the speaker has said by means of such an ambiguous phrase, one needs to know, among other things, “the meaning, on the particular occasion of utterance, of the phrase” (Grice 1975/1989b: 25). It emerges that disambiguation occurs prior to determining what is said (and optionally what is later implicated) and proceeds in order to determine the speaker’s intention with the help of the context. It is noteworthy that GP-inducing ambiguity will not surface at the initial stage of on-line processing. Both meanings are intended but arrive consecutively, which is not compatible with Grice’s postulates on the process of conscious disambiguation. In addition, Grice (1957/1989b, 1975/1989b) avers that two meanings of an ambiguous chunk will reach the interpreter’s awareness before he/she chooses the adequate meaning at the level of what is said. One might observe that if the interpreter arrives at two competitive meanings, he/she could attribute them to the speaker’s intentionality, hence perceiving Manner maxims’ flouts. Grice (1957/1989b, 1975/1989b), however, assumes that it is not the speaker’s intention to convey the two meanings and that he/she is not to blame for the ambiguity, since its raison d’être is the speaker-independent coincidence of two timeless meanings in one form. As contemporary research shows, however, ambiguity need not become transparent, by instantly reaching the interpreter’s threshold of awareness. As in the case of GPs, only one meaning will be perceived as the speaker’s meaning, which may need to be cancelled eventually. In conclusion, the analysis of ambiguity at the level of what is said raises serious misgivings. It is reasonable to suppose that lexical ambiguities associated with particular words must be resolved to yield the speaker’s meaning. So conceptualised, ambiguity appears to be inapplicable in the analysis of GP humour. First of all, it is not intended by the speaker, which renders it irrelevant to intentional ambiguity-based humour discussed here. Also, in GPs, the hearer is initially oblivious to the fact that there is ambiguity in the lead-up, i.e. there exists another interpretation intended by the speaker, apart from the one inferred. Last but not least, in GPs, the first interpretation of an ambiguous chunk, and hence the entire inference based on the lead-up needs to be cancelled, which the Gricean model does not allow at the level of what is said. The

Grice’s Philosophy and Garden-Path Humour

211

process of ambiguity emergence and resolution must be reconceptualised to serve the purpose of the present analysis. After the meaning of an ambiguous chunk is determined, the entire inference needs to be classified as a conversational implicature, thanks to which it passes the cancellability test.

7.2. Manner maxim flouts – the level of PCIs Grice (1975/1989b) analyses deliberately produced ambiguity as a case of flouting the Quality maxims within the “Be perspicuous” supermaxim. Intentional ambiguity is a legitimate phenomenon within the CP model and yields, according to Grice, particularised conversational implicatures. The interpreter’s role is to infer why the speaker has chosen to produce an utterance in an ambiguous way. Grice distinguishes two cases of ambiguity. The first type embraces situations when there is no difference, or striking difference, between two interpretations of an utterance with respect to straightforwardness; neither interpretation is notably more sophisticated, less standard, more recondite or more farfetched than the other. (Grice 1975/1989b: 35)

Admittedly, what Grice means by this is the situation when the interpreter appreciates two different but relevant meanings and oscillates between them, as in the poetic example taken from Blake’s poem “I sought to tell my love, love that never told can be”, where the word “love” may refer either to the state or to the object of affection, while the second clause may mean “love that cannot be told” or “love that if told cannot continue to exist” (Grice 1975/1989b: 35). This is thus the type of ambiguity with two meanings deliberately conveyed by the message sender, in this case, for a poetic effect. However, a question arises whether this type of verbalisation should be indeed considered a particularised, rather than generalised, conversational implicature. In the verbalisation quoted, there is no necessity to account for any contextual factors so that all the inferences can be drawn. In discourse where the maxim “avoid ambiguity” is flouted with a view to achieving a humorous effect, this type of ambiguity will manifest itself in puns with two relevant meanings, e.g. “You are stuck with your debt if you can’t budge it.” This is, however, the type of ambiguity which will not be conducive to GP humour, as the latter is reliant on a sudden switch from one interpretation of an ambiguous chunk to another. The other type of ambiguity distinguished by Grice (1975/1989b), which is most significant for the present work, occurs when one

212

Chapter Five

interpretation is more “straightforward” than the other. Admittedly, what Grice means by this is that one meaning is more transparent, obvious or more easily accessible than the other. It is difficult to judge, however, how exactly such qualities can be measured. Admittedly, Grice’s approach appears to be an intuitive one, as the premises for the distinction between straightforward and nonstraightforward interpretations are not presented. Grice (1975/1989b: 36) illustrates this type of ambiguity with a single example of homophone produced by a general who, having captured the province of Sind, sent a message in Latin saying, “Peccavi” (i.e. a confession of sin), which is itself unambiguous but, when translated, may be read as “I have sinned”/“I have Sind”, the former of which is the straightforward translation, while the latter reading is its homophonous pun. It may be extrapolated that it is the nonstraightforward interpretation that the speaker aims to convey, while the confession of having sinned would not be the intended one, unless in special circumstances, i.e. if the speaker is simultaneously guilty of some transgression. Nota bene, although Grice does explain it, this is indeed the case in this example, given the prior orders not to seize the territory, which the general disobeyed due to little resistance in the province. Grice (1975/1989b) stipulates that it is either both or only the less obvious interpretation that is meant by the speaker. He does not allow for cases when the most easily accessible meaning is actually the intended one, while the second activated one must be suppressed. This may be because Grice believes that ambiguity would not or should not arise in such a case at all. Accordingly, the speaker need not subject the hearer to undue mental effort, which the hearer may expect, finding it unnecessary to search for another meaning which would turn out irrelevant anyway. There may be stylistic reasons for conveying by a sentence merely its nonstraightforward interpretant, but it would be pointless, and perhaps also stylistically objectionable, to go to the trouble of finding an expression that nonstraightforwardly conveys that p, thus imposing on an audience the effort involved in finding this interpretation, if this interpretant were otiose so far as communication was concerned. (Grice 1975/1989b: 36)

To summarise, in the two independent discussions, Grice presents two lexical ambiguity types, occurring at the stage of determining what is said and what is implicated in a given context. Even though Grice does not explicate it, their distinguishing feature appears to be the speaker’s intention to convey the alternative meanings, whether ultimately supported or not. However, there is no theoretical differentiation between the two types of ambiguity. It emerges that each example of an ambiguous

Grice’s Philosophy and Garden-Path Humour

213

verbalisation must be rationally evaluated in the light of the speaker’s intent to realise whether the speaker wants the ambiguity to be present. The focus of attention here will be the intentional type of ambiguity, which yields implicatures. It is argued that the original conceptualisation of GCIs needs to be extended to cover a broader range of phenomena, both linguistic and pragmatic ambiguities. In addition, the second maxim of Quality can be flouted for humorous purposes, hinging on GCIs, rather than PCIs, as Grice posits. It is worth noting that while Grice’s “peccavi” example is indeed anchored in situational factors, the other example capitalising on the use of the polysemous lexeme “love” could actually be regarded as a GCI. Even if Grice does not appear to have voiced any opinion on humorous ambiguity, not to mention covert ambiguity conducive to GP humour, the phenomenon cannot be claimed to fall outside the model of intentional and rational communication. That Grice simply did not consider this peculiar case of covert ambiguity does not mean that it cannot be interpreted in the light of his model. GPs capitalise on the hearer’s initial obliviousness to one interpretation (the nonstraightforward one, as he would call it), the ultimate emergence of which gives rise to the perception of ambiguity and cancels the first meaning at the end of the utterance processing. Why polysemy-based ambiguity should not instantly enter the interpreter’s mind finds its partial explanation in the notion of Modified Occam’s Razor (see section 8), which motivates the existence of some GCIs.

8. Conventional meanings vs. GCIs exploited in GPs As already summarised, GCIs derive from maxim flouts and implicate certain propositions, in the absence of defeating features of the context (Grice 1975/1989b). GCIs arise regardless of the contextual support and are thus pragmatically salient readings generated from linguistic constructions across various contexts. Neat and plausible as this definition may be, it is not easily verifiable, inasmuch as “noncontroversial examples are perhaps hard to find, since it is all too easy to treat a generalized conversational implicature as if it were a conventional implicature” (Grice 1975/1989b: 37). What is significant, even though GCIs are presented as being couched in particular lexical items, they appear post-propositionally, i.e. after a clause or a sentence has been completed. Grice’s examples of GCIs are contingent on the indefinite article. “X is meeting a woman this evening” (i.e. The woman X is meeting is not his wife, mother or sister; flout of the first maxim of Quantity) (Grice 1975/1989b: 37)

214

Chapter Five “X went to a house yesterday and found a tortoise inside the front door” (i.e. The house was not X’s own; flout of the first maxim of Quantity) (Grice 1975/1989b: 37) “I broke a finger yesterday” (i.e. The finger was my own; flout of the first maxim of Quantity) (Grice 1975/1989b: 38)

Having presented the three examples, Grice (1975/1989b) rules out the possibility of multiple meanings of the expression “an X”. This exemplification, however, testifies that the use of an indefinite pronoun, rather than a definite or a possessive one, may or may not indicate an intimate relation between the subject and the thing indicated by the noun phrase. Similarly, Grice (1981/1989b) provides the example of the word “and”, which carries the generalised implicature of temporal sequencing of events, being simultaneously subject to cancellation. The GCI as a default inference eradicates the necessity to allow for disparate meanings of one expression, which may otherwise be regarded as polysemous. As Grice (1975/1989b, 1981/1989b, 1989a) emphasises on many occasions, many mistakenly perceive the GCI as conventional meaning, while “what is conversationally implicated is not to be thought of as part of the meaning of the expressions that are used to get over the implication” (Grice 1981: 185).11 To substantiate the distinction between conventional meanings and GCIs, Grice (1978/1989b, 1981/1989b) champions Modified Occam’s Razor,12 according to which “senses are not to be multiplied beyond necessity” (1978/1989b: 47). What is meant by necessity is rational evaluation that another sense is of importance to a given usage of a word. Grice’s underlying aim in advocating Modified Occam’s Razor is to highlight the heuristic advantage of generalised conversational implicatures over conventional ones or timeless meanings. Given that the Cooperative Principle acts as the motivation for implicatures, Grice propounds a thesis that can be dubbed Grice’s Razor (Davis 1998). This is a principle of parsimony supporting a preference for linguistic explanations in terms of conversational implicatures over semantic explanations. This is a feasible postulate, which motivates, for example, the crucial feature of cancellability.

11

This extract cannot be found in the 1989 publication. Although Grice does not explicate it, the original principle bearing this name and attributed to the 14th century logician, Francis of Ockham, holds that all things being equal, the first and simplest solution tends to be the right one. 12

Grice’s Philosophy and Garden-Path Humour

215

(…) one should not suppose what a speaker would mean when he used a word in a certain range of cases to count as a special sense of the word, if it should be predictable, independently of any supposition that there is such a sense, that he would use the word (or the sentence containing it) with just that meaning. (…) one should suppose a word to have a less restrictive rather than more restrictive meaning, where a choice is possible. (Grice 1978/1989b: 47-48)

Referring to “a special sense of the word”, Grice (1978/1989b: 47) also uses the epithet “further”, which may suggest that it is not the original non-restrictive sense, but the more restrictive sense achieved via “a superimposed implicature” (Grice 1978/1989b: 48). In addition, Grice distinguishes two types of such special senses, i.e. transferred senses (e.g. “loose” as collocating with “liver”/“life”) and derivative senses, in which “the original sense becomes obsolete (like car meaning ‘wheeled vehicle’), or the specificatory condition takes over” (Grice 1978/1989b: 48). However, Grice (1978/1989b) makes a provision concerning the derivative sense. Neither process (obsolescence or specification) can be observed, e.g. if the word “animal” means “a member of the animal kingdom” (more “predictably”, as Grice argues) or carries the further sense “beast” (Grice 1978/1989b: 49). All of these phenomena might be associated with that of polysemy, and thus polysemous ambiguity. On the whole, one may conclude that Grice favours a view that the interpreter should choose the most easily accessible meaning of a word, without searching for another one, unless particular circumstances demand this, i.e. “on the assumption that the words were to have a specificatory further sense” (Grice 1978/1989b: 49). This nicely ties in with the concept of covert ambiguity in GPs and the fact that not more than one meaning is observed initially. Nevertheless, in the light of contemporary research on human cognition, a reservation needs to be made concerning the default meaning. Contrary to what Grice appears to suggest, it is also the special/further sense that may actually be the prominent one, the one generated by default, e.g. “a mouse” as “a computer device”, being more familiar to the young urban generation than “a rodent”. Finally, it is crucial that modified Occam’s Razor accounts only for one GCI type, i.e. the one which may be regarded as centring on polysemy or words with pragmatic specialisation (Ruhl 1989). The emergent query is whether GCIs will embrace inferences generated upon homonymous lexemes with two conventional meanings. Defending the notion of the GCI from the criticism garnered since its introduction, Grice (1981/1989b) emphasises that some analysts fail to acknowledge the difference between the conventional implicature and the

216

Chapter Five

nonconventional generalised implicature, which is normally carried by the use of a given form but cannot be conceived of as part of the conventional meaning. Grice’s perseverance in his support of non-conventionality does not exclude the fact that what is said (and thus conventional meanings of words, as long as speaker-intended) do play a significant role in the generation of implicatures. In Grice’s (1975/1989b) analysis of the two examples entailing ambiguity, both competitive conventional meanings of either punning element (“love” and “sinned/Sind”) do contribute to the emergent PCIs (as shown is section 7.2). Lexical ambiguities of conventional meanings participate in implicature formation. The problem of conventional meanings looms large in the analysis of punning GP humour, as it is sometimes difficult to differentiate between GCIs and conventional meanings. On the one hand, the inference computed from the lead-up appears to be dependent on a lexical (timeless) meaning of a given word or a string of words, i.e. a phraseme. On the other hand, it does turn out to be cancellable. The initially privileged meaning, is subject to backtracking and reanalysis in the light of the context, here understood as the final part of the GP text (i.e. co-text in the form of the switch) yielding incongruous information. Consequently, the chosen conventional meaning of the ambiguous lexeme contributes to the post-propositional implicature of the whole utterance, which conforms to Grice’s approach towards ambiguity-based implicatures. It seems that inferences which might be regarded as what is said need to be relabelled as generalised conversational implicatures owing to the parameter of cancellability. Additionally, problems in the differentiation between conventional and conversational implicatures originate from the fact that conversational implicatures transform into conventional implicatures, which Grice (1978/1989b: 43) also appreciates, i.e. “what was originally a conversational implicature has become conventionalised.” Levinson (1983) rightly observes a lacuna in the studies on the interaction between implicatures and conventionalised language structures. In some cases, this conventionality includes indirectness, which gives rise to GCIs, such as speech act rituals (e.g. “Here you are.”), and indirect speech act conventions (e.g. “Could you please?”), serving the purpose of successful communication of information, politeness norms, adequate style, and efficiency (cf. Davis 1998). A vexing question arises as to whether the import of such formulations or other phrasemes (primarily idioms, set collocations and sayings) should be deemed as conventional or conversational implicatures. Davis (1998) dubs implicatures rooted in formulae “conventional conversational implicatures”, which is a paradox,

Grice’s Philosophy and Garden-Path Humour

217

given the Gricean postulates. It appears reasonable to argue that all such formulations and phrasemes will have been initially introduced as nonconventional language to follow the path of conventionalisation. Repeated and eventually conventionalised usage attaches a distinctive sense to a particular phraseme (e.g. a conventional metaphor or an idiom) with noncompositional meaning garnered via the process of semantic shift. As already reported, Grice (1978/1989b: 50) chooses not to view such constructions as yielding implicatures. Admittedly, such phrasemes (including those exploited in GPs) could be classed as conventional implicatures, as they rely on the standardised meanings of elements used. However, it may be assumed that the very fact that they are cancellable determines their conversational nature. Furthermore, the scope of GPs could be extended to cover also other forms of default inferencing, which do not depend on particular lexical items per se but capture “our intuitions about a preferred or normal interpretation” (Levinson 2000: 11). Such a preferred interpretation may not necessarily operate on the use of one particular lexeme or phraseme but on the interpretation of a chunk of text. In his analysis of PCIs, Grice focuses on verbalisations, regarding them as entities bearing special nonce senses motivated contextually and communicating something entirely different from what is said. A similar procedure could be applied to the GCI on the provision that this type of inference will emerge irrespective of the co-text or situational context. Referring to Grice’s (1989a) conceptualisation of centrality, such implicatures usually manifest dictiveness without formality.

9. Conclusion: Defeasible GCIs conducive to GPs Broadly conceptualised, GCIs manifest two qualities necessary for the operation of GPs, viz. cancellability and imperviousness to contextual factors, which could be dubbed defaultness (see chapter 6). It must be emphasised once again that Grice’s GCIs are generated postpropositionally, rather than depend solely on particular lexemes or phrasemes. GCIs act as default interpretations which are subject to cancellation at the second stage of processing, when the following text or contextual factors are computed (Horn 1984, Levinson 2000, Breheny et al. 2006). As Capone (2006) rightly observes, context (here understood as co-text) plays a role in the appreciation of GCIs only in the case of their cancellation but not promotion. Grice (1975/1989b) introduces cancellability as an instrument which serves to avoid misunderstandings. However, here cancellability is the facilitator of GPs. The humour-inducing cancellability

218

Chapter Five

of GCIs is necessitated in on-line processing as the co-text in the form of the switch turns out to be incompatible with the implicature computed on the basis of the preceding part of the humorous utterance, i.e. the lead-up. The hearer processes the lead-up, arriving at one meaning and taking it as the meaning the speaker intends to communicate. Indeed, this is the interpretation that the speaker wants the hearer to make initially only to have to cancel it. It is the non-default and hitherto covert meaning that is ultimately enforced, as the cancellation process is instigated. It needs to be highlighted that both readings of the lead-up, emergent consecutively, are intended by the speaker, who rationally evokes them in the interpreter’s mind in order that the latter should experience a humorous effect. The first (default) meaning is incongruous with the switch, which proves the leadup ambiguous and foregrounds the alternative meaning. It is at this moment that the speaker realises that the “Avoid ambiguity” maxim has just been exploited (flouted). In the section above, a few categories of GCIs were distinguished, not all explicitly discussed by Grice but conforming to his model. Grice’s GCI perfectly encompasses implicatures facilitating the generation of GPs couched in pragmatic, linguistically-motivated inferential ambiguity, which capitalises on inferences motivated by the use of particular words or formulations. Additionally, the principle of Modified Occam’s Razor can be taken to explain why only one meaning of each such ambiguity is appreciated first. Initially, as the speaker envisages, the interpreter will choose the meaning best fitting the co-text within the lead-up, not expecting any other derivative meaning to become relevant, as it does at the stage of the switch. I like kids but I don’t think I could eat a whole one. [“liking kids” implicates being fond of them] (the speaker has a positive attitude to children ĺ the speaker likes eating children) The secret of our long marriage: we go to a nice restaurant and to the cinema and then we have sex. She goes Fridays, I go Saturdays. [“we” implicates two people acting together] (the secret of the speaker’s long marriage with his wife is that they spend time together ĺ the secret of the speaker’s long marriage with his wife is that they have a nice time separately)

Likewise, GCIs will also coincide with presuppositional inferences yielding GPs. It must be remembered that Grice appears to have been averse to the concept of presupposition (Grice 1981/1989b), considering it

Grice’s Philosophy and Garden-Path Humour

219

subsumable under his notion of implicature. Indeed, the example below can be regarded as explainable by Grice’s Modified Occam’s Razor. The best way to drag a genuine man out of a bath is to turn on the tap. [a bath: a basin filled with water] (the most successful way of stopping a man indulging in a bath ĺ the most successful way to drag a man from an empty bathtub is to let in some water)

All the remaining categories of GCIs are beyond Grice’s original conceptualisation. Admittedly, the least problematic one exploited in GPs but not explicitly discussed by Grice is that embracing utterances manifesting what is here called purely inferential pragmatic ambiguity, not being hinged on any lexical element in isolation but arising postpropositionally. These are generalised counterparts of particularised implicatures Grice (1975/1989b) discusses, but they do not entail a complete diversion from what is said. Referring to Grice’s (1989a) conceptualisation of centrality, such implicatures manifest dictiveness without formality. In other words, hearers make inferences completing propositions with meanings which are not conveyed by the said per se and are not prompted by any particular words therein. Such meanings are, however, predictable/standard extensions of/contributions to what is said. Don’t spend your money on drugs. Become a pop star and get them free. (do not buy drugs and do not take them, because they are detrimental to your health ĺ you do not have to spend your money, as you can get drugs free when you are a pop star) If I want to hear the pitter-patter of little feet, I certainly will. I will put shoes on my cat. (if the speaker wants to have a baby, she will get pregnant/adopt a childĺ if the speaker wants to hear the sound of little feet, she will make her cat wear shoes) TV is very educational. Every time it is on, I go to another room and read a book. (in the speaker’s view, watching TV carries educational value. ĺ in the speaker’s view, TV programs are so poor that he/she prefers reading books to watching TV, which carries more educational value)

Although Grice does not explicitly address the category presented above, it poses little problem for the model. GPs pivoting on various forms of lexical ambiguity are a thornier issue. If one assumes that GPs are reliant on cancellable GCIs, a problem arises as regards homonymous and

220

Chapter Five

polysemous (even if motivated by Modified Occam’s Razor) ambiguities, which immanently rely on conventional meanings of particular lexemes, as those cannot be treated as generating conversational implicatures. However, it is the interpretation of the whole lead-up, not the ambiguous lexeme alone, that must be regarded as a GCI, invariably computed postpropositionally. This is exactly the phenomenon Grice (1975/1998b) discusses as an implicature (cf. the “sinned/Sind” and “love” examples). What additionally testifies in favour of such a view is the successful cancellability of the inference upon the speaker’s recognition of having made a “wrong” one and fallen into the trap intentionally set by the speaker. Back up my hard drive? How do I put it in reverse? [homonymy of “back up”: “make copies”/“reverse (up something)”; the polysemy of “drive”: “computer disk”/“road for vehicles”] (the speaker is dubious about making copies of the computer’s disk ĺ the speaker does not know how to change the gear, and hence to back a car up the drive near his/her house) It’s not MP. It’s YP, your problem. [acronym “MP”: “Member of Parliament”/non-conventionalised acronym for “my problem”] (it is not a Member of Parliament ĺ it’s not the speaker’s problem but the hearer’s)

Equally problematic are inferences generated upon lexically ambiguous phrasemes, which (in the Gricean sense) should not be treated as implicatures but which cannot be placed at the level of what is said, either. However, by analogy to the previous category, the fact that the inferred sense can be cancelled to pave the way for the alternative meaning, interpreted according to the meanings of the phrasemic member constituents, suggests the inference’s being a GCI. Life is not all beer and skittles. There are also girls. [saying “not all beer and skittles”: “not just sheer pleasure”] (life is not only pleasure ĺ life is not all about drinking beer and playing skittles but also spending time with girls) Marriage is when a man and woman become as one; the trouble starts when they try to decide which one. [idiom “become as one”: “become similar, act together upon mutual agreement”]

Grice’s Philosophy and Garden-Path Humour

221

(in marriage two people start acting and thinking together, e.g. by making compromises ĺ in marriage, one person needs to change and adjust to the other, which causes problems)

Although, beyond a shadow of a doubt, Grice’s GCIs account for pragmatically based GPs, subsuming the last two categories of lexically ambiguous GP lead-ups under the label of GCIs might be considered farfetched, given the conventionality of meanings subject to cancellation. Critics may then conclude that the discussion on GPs (specifically, the first inference made) transcends Grice’s (1975/1989, 1978/1989b, 1989a) GCI, with all lexically ambiguous (i.e. punning) GPs being a lacuna in the model. Even if so, this does not mean that the entire analysis was in vain. The Gricean model of communication determined by rationality can indeed incorporate humour, including the GP category, in which maxims are flouted for the sake of humorous effects. The problem emerges only at the lower level, at the level of meanings and pertains only to some of those. The GCI is a recurrent topic in pragmatic literature. Many researchers (e.g. Gazdar 1979; Horn, 1984, 1989, 2004; Levinson 1995, 1998, 2000; Recanati 2003, 2004; Jaszczolt, e.g., 1999, 2004, 2005)13 hold a view that GCIs are presumed meanings occurring independently of context and coincide with default meanings which normally emerge unless blocked by contextual factors. The most significant for the present analysis are the developments of the GCI proposed by neo-Griceans, primarily Levinson (1983, 1995, 1998, 2000), according to whom, GCIs are inferred on the basis of a set of default heuristics.14 The next chapter will aim to shed new light on the GP mechanism, drawing on the recent theories of defaults as well as on cognitive studies on salient meanings, which together capture GPs, not arousing queries, as do Grice’s GCIs.

13

RT supporters reject defaults and subsume salient meanings under the category of context-dependent pragmatic inference, viz. explicatures (Sperber and Wilson 1986/1995; Carston, e.g. 1988, 2002). 14 A competitive view is espoused, for example, by Carston (1988, 2002), who conceives GCIs as contents that the speaker directly communicates, rather than conversationally implicates. GCIs are thus not the products of default inferences, since what is communicated depends heavily on the specific context, and not on the use of lexical items.

CHAPTER SIX DEFAULT AND SALIENT MEANINGS IN GPS

The present chapter will focus on two different contemporary approaches to salient or default but defeasible meanings, which are a sine qua non for the successful realisation of GPs. Processing a GP, the hearer makes one interpretation of the lead-up, based on the salient/default meaning of the ambiguity, to the effect that the latter remains covert. The second stage of interpretation hinges on the cancellation of this inference and the evocation of another meaning in the light of the final part of the co-text, i.e. the switch. It is here postulated that the processes underlying the perception of pragmatically and lexically ambiguous GPs are best explained from two different theoretical standpoints. Pragmatic defaults anchored in the neo-Gricean rational communicative behaviour (Levinson 1995, 2000) best account for GPs centring on all manifestations of pragmatic ambiguity and lexical phrasemic ambiguity and can also be argued to hold for other lexically-based GPs. On the other hand, salient meanings conceptualised by Giora (1997, 1999, 2003) offer explanation for GPs couched in all forms of lexical ambiguity and in presuppositionbased ambiguity. The chapter is thus divided into two parts, one pragmatic, devoted to neo-Gricean studies, a development upon Grice’s model described in the previous chapter, the other cognitive, which focuses on the processes of lexical interpretation. It should be highlighted that the two perspectives assumed to explain the workings of GPs revolve around discrepant key premises, i.e. Levinson focuses on pragmatic inferences arising irrespective of context, while Giora is preoccupied with cognitive processes underpinning meaning evocation and the post-lexical impact of contextual support. Nevertheless, both the approaches focus on pre-propositional/local inferences of ambiguous chunks, with context playing its role later in the interpretation process.

1. Neo-Gricean and post-Gricean studies Grice’s output is acknowledged to have inspired two competitive strands of studies, viz. the post-Gricean one coinciding mainly with the

Default and Salient Meanings in GPs

223

relevance-theoretic approach represented, inter alia, by Dan Sperber, Deirdre Wilson, Robyn Carston and Francois Recanati and the neoGricean one, whose prominent advocates are, for example, Stephen Levinson, Kent Bach, Jay Atlas and Larry Horn. Both of the groups of theories discuss, among other things, the phenomenon of generalised conversational implicatures (GCIs) subsumable under the so-called Underspecified Model and the Default Model, respectively (Bezuidenhout and Morris 2004). The epithet “neo-Gricean” is attributed to a number of British and American linguists and their work produced since the early 1980s (Chapman 2005). Neo-Griceans have been trying to reduce the number of maxims and to elucidate the levels of meaning, as those distinguished by Grice give rise to doubts, a few of which were discussed in chapter 5. Authors thus use different terminologies to dub the layers of meaning, instead of following the distinction between what is said and what is implicated. The most crucial notion here is the development on the issue of context-insensitive default inferences, which will account for GPs. Also Sperber and Wilson (1987/1995) and their followers take issue with the levels of meaning championed by Grice, arguing in favour of explicature, thereby rejecting the notion of what is said (which is a matter of not merely literal content but disambiguation and reference assignment) and GCIs, which are close to conventionally conveyed, albeit implicit, meanings. An explicature occurs if and only if “it is a development of a logical form encoded by the utterance” (Sperber and Wilson 1986: 182). Its recovery involves three pragmatic processes, viz. disambiguation, reference assignment (both tackled by Grice’s what is said), and enrichment. GCIs are entirely ousted from Relevance Theory (Sperber and Wilson 1986/1995), since they take little account of context, which, in contrast, is crucial in Sperber and Wilson’s model, inasmuch as it is focused on the matters of nonce inference and the speaker’s intention governed by the single principle of relevance. Therefore, for example, Sperber and Wilson (1986/1995), Carston (e.g. 1988, 2002), and Recanati (1993, 2002, 2004) reconceptualise GCIs as directly communicated messages and pragmatic developments of semantically underspecified logical forms, which occur in result of enrichment and loosening of semantic content to generate ad hoc concepts. Accordingly, analysing the speaker’s utterance, the interpreter will use semantically encoded information, together with the information mutually manifest in the cognitive context. Even if this theoretical standpoint may appear palatable to many, it is not supported here, given that it would not facilitate the present analysis of

224

Chapter Six

GPs. Firstly, the model is lacking in context-independent GCIs different from nonce inferences, which are crucial for the workings of GPs. In addition, the inferential model favoured by relevance theoreticians is deduction, which is inherently monotonic and does not allow for defeasibility (Levinson 2000), which is the other cardinal feature of GCIs exploited in GPs. The present discussion leans on the alternative approach, for it redefines the GCI as a form which appears to be a useful interpretative tool in the analysis of GPs. Neo-Griceans, primarily Levinson (1983, 1998, 1995, 2000) conceptualise GCIs as default inferences engendered (unless blocked by peculiar contextual factors) on the basis of default heuristics, which are triggered by the presence of certain lexical items.

1.1. Default reasoning and meanings Inferences, implicatures included, can be divided into those which demand costly processing, and those which are drawn effortlessly and automatically irrespective of contextual factors, i.e. default meanings. The default is what is preferred, everything else being the same; it is the “null-solution”, the unmarked alternative. It is, concretely speaking, what one does “normally”, when no other directives are given. (Mey 1991: 238)

It is now widely acknowledged that default interpretations are inherent in human communication and that humans use default defeasible logic in the absence of contextual evidence (Jaszczolt 2005, 2006a,b; Asher and Lascarides 2003). Default interpretations of the speaker’s utterance are deemed as the meanings intended or presumed to be intended by the speaker and are characterised by a number of properties. Firstly, they are easily available, and thus activated instantly, not demanding the interpreter’s conscious inferential effort or decision-making processing of alternatives, which yields a shorter processing time in comparison to consciously inferred meanings. Secondly, they are often entertained locally/pre-propositionally, i.e. before the completion of the entire verbalisation’s processing. By contrast, the Gricean GCIs (described in the previous chapter) emerge post-propositionally. Default meanings arise in the process of default reasoning, which is overridden only if there is some reason to the contrary (Bach 1984), i.e. if a special context is present (Horn 2004). Default reasoning is best defined as

Default and Salient Meanings in GPs

225

“inference to the first unchallenged alternative.” By this I mean the first option that comes to mind without being immediately followed by the thought of a reason against it or of an alternative to it. I don’t mean that we check for the occurrence of such a thought, for what is crucial is not our noticing the thought but its very occurrence. In this way, we jump to conclusions except when we look before we leap. (Bach 1984: 40)

Hearers make interpretative shortcuts, because speakers lead them to, “thereby overcoming the problem of the slow speed of speech production as compared with the speed of the recovery of meaning.” (Jaszczolt 2005: 46). However, human default reasoning is fallible and happens to produce interpretative mistakes, because interpreters jump to conclusions, i.e. make inferences, even in the absence of sufficient information, which can emerge in the light of the following part of the text. As a result, interpreters must invest additional cognitive resources to correct their inferences after the first meaning has proven to be contextually inappropriate, which is the price paid for the initial interpretative shortcut. The last fundamental characteristic of default meanings is their defeasibility, i.e. cancellability (Jaszczolt 2005). This feature allows for rectifying interpretative mistakes. In the case of default meanings in GP humour, the hearer’s jumping to “wrong” conclusions and ultimate cancellation of a default is motivated by the speaker’s humorous intention. The speaker deliberately leads the interpreter to effortlessly make a “wrong” inference, predicting and trusting that the latter will take a reasoning shortcut and choose the easiest available option. The prototypical process of interpretation of a GP entails obtaining the default meaning from the lead-up, which is then annulled, i.e. defeated, by the following co-text in the form of the switch. Default meanings are recognised in many approaches to utterance interpretation and are conceptualised in a variety of ways. There have been many accounts of default meanings in strictly pragmatic studies as well as in computational linguistics. Various views on default interpretations are presented within semantics (Kamp and Reyle 1993, Asher and Lascarides 1998, 2003). Asher and Lascarides’s (1998, 2003) Segmented Discourse Representation Theory is a model of conversational competence tackling the patterns of computational modelling of intersentential relations. Underspecified logical forms are combined with pragmatically preferred interpretations governed by rhetorical relations which link utterances’ contents. Pragmatically inferred information contributes to underspecified logical forms and can be thought of as the result of the defaultness of a discourse’s rhetorical structure. The relations between sentences include Narration, Elaboration, Background and

226

Chapter Six

Explanation and are regulated by axioms, which preclude the relations from overlapping. These relations are computed in the process of default non-monotonic reasoning, in which logical forms and other sources of information are added to yield strong probabilities of relations, which normally obtain but may be cancelled under special circumstances. So framed, default interpretations will rarely be exploited in GPs, given the latter’s textual succinctness, but may indeed be used in long jokes, where they account for pragmatic ambiguities. For example, in the joke about a businessman calling his house quoted in chapter 1, the default relation forged by default is the fact that if the man dialled a number and then started talking to the woman, he will have reached the right number. Obviously, this interpretation must later be cancelled for the sake of a humorous effect. Pragmatic studies also yield many definitions of defaults. Some, following Grice, consider such pragmatic contributions to utterance meaning to be a revised version of Grice’s GCIs (Levinson 1995, 2000), whereas others classify them as pragmatic input to what is said (Jaszczolt, e.g. 1999, 2004, 2005, 2006a,b), or even to explicatures (Sperber and Wilson 1986/1995 Carston 1988, 2002, Recanati 2003). Still others regard them as a separate level of what is implicit in what is said (Bach 1994, 2004, 2006; Horn 1984, 2004). Jaszczolt (2005) aptly summarises the nature of defaults. (…) there is no consensus as to the meaning of conversational defaults, neither is there much evidence concerning their sources. Defaults stem out of heuristics of rational behaviour, be it Gricean, neo-Gricean, pertaining to the logic of information structuring, or some other defeasible logic. Defaults operate due to the structure and form of language, the way the world is, the way societies are organized, or the way the operations of our brain make us perceive and comprehend all these things. These sources and types are not necessarily mutually exclusive. (Jaszczolt 2005: 46-47)

The present approach is rooted in neo-Gricean pragmatic studies on default inferences. The focus is on those which appear to be candidates for theories offering explanatory potential for the analysis of GPs, specifically Bach’s (1994, 1999) impliciture and primarily Levinson’s (1998, 1995, 2000) account of GCIs. The two perspectives are similar as regards the agreement on the default status of inferences, but Levinson’s centres on sounder premises and is more meticulously developed.

Default and Salient Meanings in GPs

227

2. Bach’s implicitures Writing about default meanings within neo-Gricean studies, one cannot fail to discuss Bach’s contribution (Bach 1984, 1994, 1995, 1998b, 1999). Although the articles appear to address different issues, those can be meshed to present a coherent picture of default meanings. Bach (1984) draws a distinction between (conscious) inference made in the light of contextual factors and the unconscious act of “taking a step” towards the enriched default interpretation. Hearers often interpret speakers’ utterances with recourse to default reasoning, jumping to conclusions on the basis of some context-independent meaning which is the first available and unchallenged alternative. It is arrived at unconsciously, regardless of whether this is the meaning intended by the speaker. This step is cancellable if the addressee realises that the emergent meaning is not what the speaker has intended. Obviously, in the case of GPs, the default meaning is exactly what the speaker has intended, with a view to producing humour. Bach (1994, 1999) considers some meanings to be derived from default inferences. What is crucial from the present vantagepoint, Bach (1995, 1998b) also stresses the role of standardisation, which gives rise to default meanings or enriched meanings. Standardisation, resulting from an earlier use of a particular expression, helps interpreters omit the process of inferencing, even unbeknownst to themselves. In other words, standardisation offers a shortcut for the inference process so that the hearer draws inferences without realising it. What is strictly and literally said can be bypassed in the interpretation process, yielding a default meaning. Only in an unusual context will such a default meaning not be drawn. Standardisation should be differentiated from conventionalisation, which corresponds to interlocutors’ mutual belief that an utterance has a certain illocutionary force (Bach 1995, 1998b). This explanation of standardised meanings perfectly encompasses phrasemes used in GPs but may not account for all cases of inferencing across contexts, i.e. pragmatic ambiguity. In Bach’s (1994) view, default standardised meanings are neither implicatures of any type nor what is said (or explicatures). In lieu of those, Bach (1994) proposes implicitures1, suggesting that Grice himself indirectly appreciates these phenomena, even though he officially distinguishes only implicatures and what is said, by mentioning 1

However, it must be noted that implicitures can be the result of a contextdependent process of inference and so cannot be conceived of as purely default meanings.

228

Chapter Six

“strengthening one’s meaning by achieving a superimposed implicature” (Grice 1989b: 48). Such strengthening, according to Bach (1994), may mean increasing the information content of what is explicitly conveyed by what is said, but not adding a separate proposition to it. Bach’s (1994) proposal hinges on the basic idea that the implied needs to be distinguished not only from what is said but also from the implicit. Implicitures go beyond what is said but stem from what is said unlike implicatures, which are additional propositions external to the said. In other words, implicitures cover the implicit in what is said, whereas implicatures are implied by (the saying of) what is said (Bach 1994). In implicature one says and communicates one thing and thereby communicates something else in addition. Impliciture, however, is a matter of saying something but communicating something else instead, something closely related to what is said. (Bach 1994: 126)

This definition is provocative, the reasons being several. Firstly, differentiating between implicitures and implicatures, Bach fails to recognise Grice’s categorisation of implicatures into conventional and non-conventional ones, but, admittedly, he means the latter, given that he deems the former redundant (cf. Bach 1999). Additionally, Bach appears not to differentiate between inferences generated in a context (PCIs) and irrespective of context (GCIs), as the examples quoted below will also show. Thirdly, Bach’s understanding of “implicature” appears not to coincide fully with Grice’s construct, in which it is the intended meaning beyond what is said, and not always “communicating something else in addition” (cf. section 6.2.2.). Bach rightly observes that an implicatum can be completely separate from what is said and from the saying of it, which he regards as the prototypical implicature, e.g. saying “It’s after 7” in order to implicate “Guests should be arriving.” Indeed, as already observed many of the examples Grice (1975/1989b) provides represent this category of implicature. In Grice’s (1975/1989b, 1978/1989b) terminology, implicatures refer to what the speaker actually means via implying, which may be entirely distinct from what the speaker says. Nevertheless, this does not mean that Grice (1975/1989b, 1978/1989b) does not allow for implicatures which, as Bach would put it, are directly built upon what is said. Introducing the notion of the GCI, Grice (1975/1989b) provides examples of lexical items which carry implicatures corresponding to Bach’s implicitures. Also, Bach fails to appreciate the notion of centrality and the dichotomies between dictiveness and nondictiveness as well as formality and informality (cf. Grice 1989a), which appear to give rise to subtypes of implicatures. Thus, implicitures

Default and Salient Meanings in GPs

229

correspond to Grice’s implicatures manifesting formality and a lack of dictiveness (cf. Grice 1989a). If this is the case, implicitures can be nothing but otiose. The conceptualisation of implicitures as such appears to be acceptable if they are considered a peculiar subtype of Grice’s implicature. It is also noteworthy that implicitures share many characteristics with implicatures. Bach (1994) explicates that implicitures are cancellable and that impliciture mechanisms (completion and expansion) “involve basically the same sort of pragmatic process as in implicature proper” (Bach 1994: 144). Implicitures are directly built upon the said, which Bach (1994) equates with explicit content of the utterance, thanks to conceptual strengthening. This process enables the recovery of any lexical material missing from the utterance. In the case of an impliciture, what the sentence means does not fully determine the speaker’s meaning, because of either sentence nonliterality or semantic underdetermination (Bach 1994). Bach’s contention is that implicitures manifest themselves in two versions. (…) what is communicated is only implicit in what is explicitly expressed, either because the utterance is semantically underdeterminate and completion is required or because what is being communicated is an expanded version of the proposition expressed. (Bach 1994: 126)

On the one hand, an utterance may be semantically underdeterminate and demand completion, yet, on the other, it may constitute a complete proposition but fail to coincide with the proposition meant by the speaker, thereby requiring expansion. The underdetermination of utterances, albeit syntactically well formed, manifests itself in semantic or conceptual incompleteness (pertaining to one conceptual element or a relation between conceptual elements). Bach (1994) provides various examples of this phenomenon throughout his article. That lamp is cheap. [relative to other lamps] Gregor was merely a bookkeeper. [as opposed to an accountant] Tipper is ready. [to dance] Al has finished. [speaking] Psychologists can also attend the meeting. [together with other scientific groups] Even accountants make mistakes. [accountants are the least likely to but they do] A few arsonists destroyed many buildings. [each/together]

Expansion involves conceptual strengthening (not necessarily lexical, for the interpreter may use words different from those the hearer would

230

Chapter Six

use) which renders the communicated content fully explicit, yet is not obligatory, insofar as the proposition is complete even without it. Bach (1994) also observes that the inclusion of an additional element (e.g. “at least”) can block logical strengthening (e.g. “exactly”). To quote a few of Bach’s (1994) examples: I have eaten breakfast. [today] I have eaten. [breakfast today] John has three cars. [at least] You’re not going to die. [from this cut] France is hexagonal. [roughly]

It emerges that both completion and expansion may entail the knowledge of particular circumstances and demand situational-dependent defaults, which excludes them as a basis for GPs, which must operate on cross-contextual defaults. On the other hand, there are implicitures which can be deemed as context-independent, and thus parallel to GCIs. These are cases of completion of semantically underdeterminate sentences, expansion of sentence-nonliteral utterances, lexical and phrasal completion, and extension entailing pragmatic specialisation or generalisation (cf. Ruhl 1989), metalinguistic negation and Saxon’s genitive, which are semantically underdeterminate, unless enriched. On the whole, although plausibly described, implicitures appear to be reducible to Grice’s implicatures, either generalised or particularised. Also, Levinson (2000) evaluates Bach’s conceptualisation negatively, observing that it offers hardly any improvement upon Grice’s model and his distinction between what is said and implicatures. Levinson’s (1983, 1998, 1995, 2000) proposal offers a better (more adequate and meticulous) model of default meaning generation. Another merit of Levinson’s approach is that defaults are normally constructed irrespective of contextual factors and are always subject to cancellation.

3. Levinson’s Presumptive Meanings As discussed earlier, GCIs advocated by Grice (1975/1989b, 1978/1989b) and exemplified primarily by means of a number of connectives appear to have provoked a heated debate on whether these inferences ought not to be considered conventional meanings. That some inferences are more often than not carried by particular lexemes does not mean that they are immanently and invariably carried. As Levinson (2000: 20) puts it, “It is this normal, general, default tendency of interpretation that makes it nonobvious that what we are dealing with here is mere

Default and Salient Meanings in GPs

231

pragmatic inference.” This type of inference must then be of pragmatic, rather than semantically conventional, nature. Also, GCIs conventionally ascribed to lexemes or phrasemes are potentially cancellable in the light of contradictory contextual factors. In various publications, Levinson advances a revised, if not entirely changed, version of Grice’s GCI deemed as a default interpretation pivoting on utterance-type meaning, i.e. statement-type meaning (1983, 1998, 1995, 2000), eventually dubbed presumptive meaning (2000).2 Levinson’s GCIs (1995, 2000) are a most useful heuristic tool for the analysis of GPs, given that the author explicitly attributes to it two features crucial to the phenomenon in focus, i.e. defaultness and defeasibility (cancellability), and allows for all kinds of lexically based but pragmatically motivated non-monotonic inferences to arise in the course of rational communicative behaviour. GCIs are inferences that appear to go through in the absence of information to the contrary; but additional information to the contrary may be quite sufficient to cause them to evaporate. Thus the mode of inference appears to have two properties: it is a default mode of reasoning, and it is defeasible. (Levinson 2000: 42)

Levinson (2000) contends that the nonmonotonic mode of inference is characterised by two properties, i.e. defaultness and defeasibility. Utterance-type meanings are independent of the discourse situation but can be cancelled should context, here regarded as a following co-text, demand it. Such an account is most plausible as regards GPs, which manifest a twofold structure. Namely, the initial chunk in the form of the lead-up offers the hearer input material to make a default inference, particularly on the strength of the I-heuristic embracing conventionalised expressions, which is necessarily cancelled by the following co-text. Interestingly, Levinson does appreciate that the non-humorous garden-path mechanism may participate in implicature cancellation, which is “like having taken the wrong horn of an ambiguity, and being jolted by a later context into backtracking and taking the other horn” (Levinson 2000: 384). In their collaborative article, Atlas and Levinson (1981) argue in favour of the existence of semantic structures above and beyond truth conditions which, in interaction with pragmatic principles, yield defeasible 2

Levinson’s output generated over more than two decades manifests insignificant changes in terminology and typographical conventions, especially the use of hyphens in many of the terms coined.

232

Chapter Six

implicata. Departing from this premise, Levinson revisits the Gricean GCI, i.e. “An implicature i is generalized iff utterance U implicates i unless there are unusual specific contextual assumptions that defeat it” (Levinson 2000: 16). The point of similarity between Gricean GCIs and GCIs as presumptive meanings (and the kernel of the present analysis of GPs) is that, in ordinary circumstances, both are obtained on the basis of verbalisations unless blocked by context and both are subject to cancellation in the light of emergent contextual factors. However, Levinson’s version of GCIs represents an ontological status divergent from Grice’s GCIs. Levinson (2000) explicitly differentiates between the speaker’s meaning and the meaning generated by default, regardless of the evaluation of the speaker’s intention and context, thereby transcending the traditional dichotomy. According to the standard line (more often presupposed than justified), there are just two levels to a theory of communication: a level of sentence meaning (to be explicated by the theory of grammar in the large sense) and a level of speaker-meaning (to be explicated by a theory of pragmatics, perhaps centrally employing Grice’s notion of meaningnn) (Levinson 2000: 22)

Levinson (1983, 1995, 2000) advocates a third level of meanings in communication, i.e. utterance-type meaning accounting for “aspects of meaning associated with the general, normal use of expressions” (Levinson 2000: 63). As Levinson suggests, we can expect the systematicity of inference that might be deeply interconnected to linguistic structure and meaning, to the extent that it can become problematical to decide which phenomena should be rendered unto semantic theory and which unto pragmatics. (Levinson 2000: 6)

In this view, GCIs are expectations about ordinary language use, representing utterance-type meaning, intermediate between sentence-type meaning and speaker-meaning. The third layer is a level of systematic pragmatic inference based not on direct computations about speaker-intention, but rather on general expectations about how language is normally used. These expectations give rise to presumptions, default inferences, about both content and force. (Levinson 1995: 93)

Levinson’s (1995, 2000) GCIs acting as defaults, i.e. presumptive meanings, can be reduced neither to semantics nor to pragmatics, since

Default and Salient Meanings in GPs

233

“they sit midway, systematically influencing grammar and semantics on the one hand and speaker-meaning on the other” (Levinson 2000: 25). Levinson’s (1983, 1995, 2000) postulate of the layers of meaning is plausible and most useful in the present analysis, but one crucial issue needs to be raised, given his main departure from Grice’s model, not only concerning the number of layers of meanings. Levinson (1983, 1995) appreciates the traditional dichotomy between sentence-meaning and speaker-meaning, which corresponds to Lyons’s (1977) semantics vs. pragmatics distinction. Sentence-meaning denotes the systematic combination of meanings of all the constituents, together with the constructions in which they occur, while the speaker/utterance-meaning represents the import of a sentence and the import carried by contextual factors, which is why one verbalisation may manifest different meanings in different contexts (Levinson 1995). Levinson (1983) also observes that it is very rarely so that sentence-meaning exhausts utterance-meaning, i.e. that the speaker means only what he/she says explicitly. Even though Levinson (1983, 1995) fully acknowledges Grice’s (1957/1989b, 1958/1989b, 1975/1989b) notion of the speaker’s meaning (Levinson 2000: 381), he ventures a claim which might be seen as distorting the division into the said, conventional implicatures and conversational implicatures. Accordingly, Levinson (1983, 1995, 2000) contrasts the speaker’s meaning with the meaning of a sentence, which coincides with the conventional content.3 (…) the coded content (roughly our sentence-meaning) was divided between “the said” and “the conventionally implicated” (…) while the inferred content (our utterance-meaning) was divisible between particularised and generalised conversational implicatures. (Levinson 1995: 91)

Such a conceptualisation of sentence-meaning encompasses Grice’s what is said, which is realised without any implications. However, it needs to be emphasised that in Grice’s (1968/1989b, 1969/1989b, 1975/1989b) model of nonnatural (i.e. intentional) meaning, what is said and what is 3

Levinson (1983: 17) mentions in passing that the latter should not be equated with literal meaning, since such an approach would fail to embrace non-literal but conventional meanings of idioms, which have two sentence-meanings, one nonliteral (idiomatic) and one literal (compositional, nonidiomatic). However, later (e.g. Levinson 1995, 2000) he considers idioms and other conventionalised formulae as subordinate to the level of utterance-meaning, given that they manifest one salient reading.

234

Chapter Six

implicated cannot be interpreted as independent semantic meaning vis-àvis the speaker’s meaning, respectively. This is because the speaker intends both the said (if the speaker is explicit) and the implicated (if the speaker’s meaning does not coincide with what is said). From Grice’s standpoint, the speaker’s meaning covers all meanings, both what is implicated (via GCIs, PCIs or CIs) and what is said, which is also, at least to some extent, pragmatically motivated. Nevertheless, in defence of Levinson’s reconceptualisation, it must be acknowledged that if an utterance can be understood at the level of what is said and does coincide with the conventional meanings of words, it may be conceived as a regular semantic meaning. Additionally, Levinson (1983, 1995, 2000) rightly observes that the Gricean notion of the speaker’s meaning is solely the case of nonce inferencing as to the speaker’s intention. What is most essential, even though Grice’s GCIs are context-independent inferences, they are to be deemed as nonce inferences, which the interpreter will arrive at singly on each occasion, necessarily deciphering the speaker’s intention on the presumption that the CP holds. Nota bene, despite his zealous support of the speaker’s meaning, Grice (1967/1989b) does appreciate a definitional circle related to the interdependence between timeless meaning and occasion meaning. If the former is described in terms of the latter, a reverse process may also take place, giving rise to “epistemic regress” (Grice 1967/1989b: 138). Regrettably, the strand is not expounded on further. Levinson’s reformulation of Grice’s division of meaning is in aid of proposing GCIs as the utterance-type-meaning which is neither conventional meaning nor the speaker’s meaning. GCIs are inferences pivoting not on the computation of the speaker’s intentions but on “general expectations about how language is normally used” (Levinson 1995: 93). Levinson (1995) construes GCIs as default interpretations which speed up the process of communication and hence “overcome the bottleneck” caused by slower production of speech in communication (1995: 95). It is a process much faster and more reliable than analysing the speaker’s intentions. Like Grice’s GCIs, Levinson’s are also defeasible. Nevertheless, contrary to the former, which derive from inductive inference, default meanings are non-monotonic and presumptive (Levinson 1995, 2000). Levinson (2000) explicitly rejects the view that inferences are invariably generated anew on each occasion, depending on a given speaker’s intention and context. Because GCIs are associated with utterance-types, they can be accessed without considering the utterer’s occasion-meaning, and it is thus

Default and Salient Meanings in GPs

235

legitimate to view them neutrally, both from the speaker’s and addressee’s perspectives. (Levinson 2000: 372)

This is certainly a most apt conceptualisation of GCIs, which emerge across contexts and speech situations. This is why they can be computed even if the utterances are extracted from original texts/conversations. This is also why many interpreters can appreciate isolated instances of GP humour, even if they lack any knowledge about the speaker, not to mention his/her intentions. However, whether the sender is known or not, his/her intention underlying the whole utterance, the presumptive element included, will be subconsciously presupposed. As Levinson (1995) concedes, “utterance-type-meaning will shade into speaker-meaning at the one end and sentence-meaning at the other” (Levinson 1995: 95). It is also worth noting that conventionalised, albeit pragmatically motivated, expressions must have taken “a diachronic path from speaker-meanings to utterance-type-meanings to sentence-meanings” (Levinson 1995: 95). The level of utterance-meaning will then cover, for example, clichés, idioms and conventionalised formulae (e.g. “Good luck” or “See you”), which are partially conventional and partially inferentially motivated. Nota bene, their status was not fully determined by Grice. Also, for instance, generality narrowing coincides with some processes in semantic change and conventionalisation (e.g. “a drink” as “an alcoholic drink”, “a doctor” as a man, rather than a woman) (Horn 1984). These observations are extremely vital, inasmuch as such phrasemes and conventionally interpreted lexemes are very often exploited in GPs. Another undisputable merit of Levinson’s approach in the context of the present work is that presumptive meanings can emerge at any stage of the comprehension process, even before an entire utterance has been processed. Given that discourse processing progresses incrementally (Levinson’s term for “on-line”), segments are also processed incrementally. Default meanings are thus not necessarily sentential/postpropositional but also, if not mainly, pre-propositional/local and yet simultaneously cancellable by the preceding or following context (here cotext within the switch). In this respect, Levinson’s GCIs are considerably different from Grice’s. Ordered incrementation4 of the context by the content C of an utterance U

4

Levinson builds the notion of incrementation on the basis of Gazdar’s (1979) postulates, proposing also the ordering of meaning elements (entailments> Qimplicatures> M-implicatures> I-implicatures).

236

Chapter Six Each aspect of utterance content should be added to C only if it is consistent with what has already been added to C by U of preceding utterances, or is part of C by virtue of being assumptions taken for granted. (Levinson 2000: 162)

Despite the postulate that new incoming elements are analysed incrementally in the light of the preceding information, Levinson’s affirmation reverberating throughout his work is that GCIs are contextinsensitive. This juxtaposition is by no means contradictory. Levinson (2000) does appreciate the impact of the preceding context understood as common ground but simultaneously proposes that some elements will be interpreted by default only to be weighed against their context and, possibly, defeated. “The important concept of defeasibility is thus modelled on terms of filtering by inconsistent, higher order assumption” (Levinson 2000: 163). Levinson thereby opposes the nonce-inference view upheld, e.g. by relevance theorists. Incremental addition progresses unobstructed if the consistency check is satisfied. It can be gathered that the default interpretation initially drawn may sometimes clash with the preceding context and mutual ground, in which case it will be instantly defeated. The simple device of an ordered incrementation together with a consistency check suffices to account for many aspects of the defeasibility in implicatures. And such a consistency filter is consonant with the possibility of modelling implicature generation with default rules. (Levinson 2000: 51)

Surprisingly enough, Levinson (2000) states that one statement-type meaning may yield many default inferences, e.g. “John’s book” to mean “book John owns/bought/borrowed/read/wrote” (Levinson 2000: 37) or “I ate some of the cookies” to mean “I did not eat all/most/many of the cookies” (Levinson 2000: 77). Bezuidenhout and Morris (2004) and Jaszczolt (2005) observe that such a view subverts the very essence of default meanings, which cannot originate from the choice-making process (see section 3.3). Admittedly, Levinson might be defended on the grounds that he merely provides some potential interpretations, while in real-life situations only one will be chosen (Bezuidenhout and Morris 2004), depending on the common ground built earlier. This, however, poses another problem, as it might suggest that default inferences are not always default but must be evaluated in the light of contextual factors (yet not necessarily a direct verbal context) pertinent to a speech encounter or to the mutual knowledge of two particular speakers. This is indeed a major

Default and Salient Meanings in GPs

237

objection against some presumptive meanings, which is irrelevant here, inasmuch as GPs immanently operate on clear cases of defaults, with context being reduced to the GP text and no idiosyncratic common ground built between the speaker and the recipient. Nevertheless, the criticism can be overcome. Some defaults, single meanings activated, can heavily rely on the common ground/mutual knowledge but this is the effect of priming, which cannot be equated with contextual effects (cf. sections 3.3 and 4.3).

3.1. Heuristics and the resultant implicatures Grice’s GCIs are generated on the basis of maxim flouts. In a similar manner, Levinson (1998, 1991, 1995, 2000, inter alia) develops a set of heuristics, i.e. default inference patterns, which generate default GCIs, strengthening the coded content of utterances (unless there is explicit information to the contrary) within the model of rational communicative behaviour. The heuristics invoke and filter through stereotypical knowledge of the world or metalinguistic knowledge, i.e. information about semantically related expressions (Levinson 1995). The mutual acceptance of these heuristics by interlocutors facilitates the recovery of meanings without much content having to be verbalised. The heuristics ought to be regarded not as rules but as inferential patterns which motivate generalised behavioural norms and implicatures (Levinson 2000). The heuristics developed over time, being reformulated and differently labelled. Q1, Q2 and M (Levinson 1995) correspond to Q, I and M (Levinson 1998, 1991, 2000), respectively. Q1: What is not said is not the case Q2: What is simply described is stereotypically and specifically exemplified a. unmarked expressions warrant rich interpretations to the stereotype b. minimal forms warrant minimal interpretations. M: Marked descriptions warn “marked situation” (Levinson 1995: 97) (1) Q-heuristic ‘What isn’t said, isn’t’; (2) I-heuristic ‘What is expressed simply is stereotypically exemplified’, and (3) M-heuristic ‘What’s said in an abnormal way isn’t normal’. (Levinson 2000: 31-33)

The heuristics are Levinson’s development upon the Gricean maxims, acting as inference shortcuts stemming from the mutual appreciation of interlocutors’ rationality and cooperativeness. In the light of the respective heuristics, Levinson (1995, 2000) distinguishes (Q1)Q-implicatures, the epitome of which is the scalar implicature, (Q2)I-implicatures and M-

238

Chapter Six

implicatures. Also, Levinson (1995, 2000) posits that Q(Q1)- and I(Q2)heuristics correspond to the first and the second Quantity maxims, respectively, while the M-heuristic refers to the first and fourth maxims of Manner.5 However, a number of divergences can be pinpointed between the two models. Firstly, Grice’s maxims are written from the speaker’s vantage point, while Levinson’s from the hearers, which is why the compatibilities between the two sets are mirror reflections. Secondly, contrary to the Gricean paradigm, where implicatures derive from maxim flouts, in Levinson’s model, it is the observance of the heuristics that produces implicatures. Thirdly, what Levinson also fails to explicate is that, while M-implicatures correspond to the Manner maxims’ flouts, Qimplicatures and I-implicatures rely on the observance of the counterpart Gricean maxims, showing the result of their application, as it were. In Grice’s terms, the fulfilment of a maxim would not yield an implicature but the level of what is said, rendering the inference non-cancellable. Levinson additionally reconceptualises the original maxims of Quantity. As earlier reported, Grice was uncertain about the importance of the second Quantity maxim, but he clearly thought of it as the avoidance of informational surplus, and hence of redundant overinformativeness. Levinson, on the other hand, includes in the I-heuristic means of verbalisation and so the maxim of Manner. Therefore, in Levinson’s version, the Gricean “Do not make your contribution more informative than required” might be extended by a stipulation, “if the hearer can infer the information resorting to stereotypicality.” It is evident that inferences generated on the basis of the three heuristics cannot be identical with, or sometimes even similar to, Grice’s GCIs, which will become even more obvious when examples are provided. The Q-heuristic relies on contrast sets such as entailment scales comprising the strong member (e.g. “all”) and the weak member (e.g. “some”). Levinson (1995: 99) enumerates a number of scalar sets including cardinal numbers, truth-functional connectives (e.g. “and” and “or”), gradable properties, morphemes describing grammatical meanings (“a”, “the”) and modal adjectives (e.g. “necessary” and “possible”), observing that Q-implicatures will be attached to logical vocabulary 5

Levinson’s (2000) model is also based on Horn’s division of pragmatic labour. Levinson’s Q- and M-Principles are compared to Horn’s (1989) Q Principle, and IPrinciple to R Principle. Horn (1989) holds a view that conversations are anchored in two conflicting forces, i.e. the drive to clarity (Q, i.e. quantity), which makes speakers responsible for verbalising themselves in a comprehensible manner, and the drive towards economy (R, i.e. relation), which renders hearers responsible for interpreting utterances as adequately as necessary.

Default and Salient Meanings in GPs

239

elements. Levinson (1995) also broadens the scope of this implicature, pointing to the use of one lexeme from a set, e.g. of colours, or the use of a superordinate to denote the lack of knowledge as to the subordinate category. Q-heuristic and the resultant implicatures are metalinguistic, i.e. they make references to contrast sets and the negative, i.e. they entail the inapplicability of the contrasting alternative (Levinson 1995). Generally, as regards the Q-heuristic, Levinson (2000) pays a lot of attention to the scalar (e.g. “some, hence not all” or “three, hence at least three”) and clausal implicatures (weak and strong verbs with their complements, e.g. “claim” or “divulge”). Those are not essential to the present analysis of defaults’ cancellation, though. This may be because the use of a weaker form, which is then strengthened via addition, does not annul the initial interpretation. For example, “I ate some cakes, in fact all.” does not negate the initial inference that the speaker did eat a few cakes. Because there is no complete cancellation, scalar and clausal implicatures are not normally conducive to humour, unless they are combined with other defaults, e.g. negative rising. (Q2) I-implicatures and M-implicatures stand vis-à-vis each other (Levinson 1995, 2000). I-implicatures depend on the hearer’s realisation of the speaker’s having used an unmarked minimal expression, which then licenses the former to resort to stereotypical information available to enrich the content of the speaker’s utterance. M-implicatures come into being when the speaker uses a marked or prolix form of expression. In other words, contrary to the I-heuristic, which guides inferences from unmarked brief counterparts, the M-heuristic yields implicatures contingent on markedness, which corresponds to formal prolixity and infrequent expressions, inviting non-stereotypical extensions (Levinson 1995). According to Levinson (1995), these phenomena can best be appreciated on the basis of lexical items, which would not count as GCIs in Grice’s view. - lexical doublets (e.g. Q2: “knife”/M: “cutter”, Q2: “house”/M: “residence”) - word formation patterns (e.g. Q2: “informer”/M: “informant”, Q2: “unscientific”/M: “non-scientific”) - simple lexemes vs. derived forms, the former of which refer to more extreme propensity (e.g. Q2: “rude”/M: “impolite”) - periphrastic alternatives (modals, causatives, double negations) vs. simple lexicalisations Q2: “She could +> and she did”/M: “She had the ability +> but she didn’t” Q2: “stop the car”/M: “cause the car to stop” Q2: “it’s possible+> to a certain degree n”/M: “it’s not impossible +> to degree less than n” (Levinson 1995: 104-105)

240

Chapter Six

What is also crucial, the I-heuristic (expounded on in the next section) is not compatible with the other two. As already mentioned, I-heuristic stands vis-à-vis the M-heuristic. Moreover, the I-heuristic is antagonistic towards the Q-heuristic. While the former invites inferences to be more specific, the latter prohibits enriching the interpretation beyond what the speaker has said (Levinson 1995). To resolve the problem of mutual exclusion of inferences, Levinson (1998, 1995, 2000) champions the hierarchy of implicatures’ importance corresponding negatively to their proneness to defeasibility in the case of conflict. Q-implicatures are of the highest priority, M-implicatures come in between, while I-implicatures are the most liable to cancellation. The defeasibility of default inferences from minimal descriptions appears to be the driving force of many GPs, where conventional means of expression and inferencing reliant on the I-heuristic are acted against. It must be emphasised that heuristic-based inferences emerge if there is no information to the contrary and can be cancelled if such information does arise. In GPs, this will be a matter of incremental addition of information in the most crucial part of the following co-text, i.e. in the switch.

3.2. The I-heuristic The I-heuristic, the most significant of the three in view of the present analysis, is developed upon the Principle of Informativeness (Atlas and Levinson 1981, Levinson 1998, 2000), which itself is a redefinition of Grice’s second Quantity maxim. Speaker’s maxim: the maxim of Minimization. “Say as little as necessary”; that is, produce the minimal linguistic information sufficient to achieve your communicational ends (bearing Q in mind). Recipient’s corollary: the Enrichment Rule. Amplify the informational content of the speaker’s utterance, by finding the most specific interpretation, up to what you judge to be the speaker’s m-intended point, unless the speaker has broken the maxim of Minimization by using a marked or prolix expression. Specifically: a. Assume the richest temporal, causal and referential connections between described situations or events, consistent with what is taken for granted. b. Assume that stereotypical relations obtain between referents or events, unless this is inconsistent with (a) c. Avoid interpretations that multiply entities referred to (assume referential parsimony); specifically, prefer coreferential readings of reduced NPs (pronouns or zeros).

Default and Salient Meanings in GPs

d.

241

Assume the existence or actuality of what a sentence is about if that is consistent with what is taken for granted (Levinson 2000: 114115)

According to the basic formulation of the I(Q2)-heuristic, the speaker need not say what can be taken for granted, i.e. what the hearer can easily infer. I-implicatures, generated under the I-heuristic trigger complementation to extensions of the unmarked counterparts and take the specification of meaning from the level of what is said (Levinson 1995, 2000). Levinson (1995) also considers them iconic, as minimal verbal forms correspond to rich interpretations. Verbal elaborations used instead could be burdensome, redundant or even socially undesirable, which is, for example, the case of euphemisms (Levinson 2000). A crucial property of the I-heuristic is that it allows the interpreter to resort to background knowledge from various domains to produce an extensive interpretation of a minimal description (Levinson 2000), adjusting the interpretation to general, stereotypical extensions accepted by consensus. Needless to say, such stereotypes do not have to reflect the veracity or be corroborated by statistical tendencies of the phenomena to which they refer. Levinson (1995) motivates the Q2 (I)-heuristic with the assumption that, in accordance with a psychologically grounded tendency, perceiving an utterance, interpreters will form its stereotypical exemplification. However, this is not an idiosyncrasy of each individual representing his/her personal knowledge of the world but a well-established pattern of communication hinged on Grice’s meaningNN and particularly reflexive reasoning. It must be remembered, however, that Levinson does not espouse the idea of the GCI being the speaker’s meaning in Grice’s sense, i.e. the context-independent meaning that is, nevertheless, construed anew on each occasion. It is the mutual awareness of a stereotype that assures the speaker that the hearer will interpret an utterance according to the former’s wish. The speaker must be positive about the stereotyped meaning, which the interpreter will doubtlessly arrive at (Levinson 1995). This stipulation is simultaneously the pivot of the GP mechanism, for it helps explain how the speaker can be sure that the humorous verbalisation will come off, i.e. that the interpreter will unfailingly infer the default meaning only to experience surprise by having to abandon it in the light of incoming information. The theory of presumptive meanings carries predictive potential, accounting for why a given reading will be the default one. GCI theory in its current state is a set of generalisations about the interrelations of linguistic expressions, the extra inferences that these

242

Chapter Six interrelations give rise to, and the conditions under which these default inferences do not go through. These generalisations give us clear expectations about inferences that ought to arise from particular expressions, and in this sense the theory is predictive. (Levinson 2000: 368)

The model of GCIs presupposes a body of metalinguistic6 knowledge, albeit perhaps subconscious, on the language user’s part, such as the availability of alternative expressions, subjective assessment of frequency and markedness of specific expressions, stereotypical associations of linguistic concepts, and principles governing inferences (Levinson 2000). Speakers must be certain about, or take for granted, the nature of default inferences, which will be made first. (…) certain linguistic expressions will tend to be associated with specific pragmatic inferences across a broad range of contexts, so that these associated inferences can be predicted in a systematic way, and play a systematic role in shaping patterns of lexicalisation or grammaticalisation. (Levinson 1995: 93)

In addition, as already signalled, defeasible presumptive meanings aptly encapsulate the level of conventionalised readings of semantic and pragmatic phrasemes, which pose difficulties when analysed within the Gricean paradigm, falling under the umbrella of the non-cancellable conventional meaning. Levinson’s framework encompasses those preferred interpretations of conventional expressions at the level of GCIs, which serve as “a corollary, a theory accounting for preferred interpretations” (Levinson 1995: 94). It emerges, therefore, that Levinson’s GCIs will also perfectly capture GPs based on lexical ambiguities couched in literalised phrasemes. The I-heuristic manifests itself in the inference to the stereotype and so phenomena associated with semantic generality and interpretative tendencies. Levinson (1995: 102-103, 2000: 37-38, 117-118)7 enumerates the following linguistic manifestations of this heuristic:

6

Levinson (1995), however, posits that I-inferencing is not metalinguistic, as it does not evoke a particular element that was supposed to have been used. No conscious metalinguistic knowledge is, therefore, necessary. 7 Many of the types appear in both publications with different exemplification. In addition, the 2000 publication does not explicitly divide I-inferences into two types, i.e. stereotype- and coherence-based.

Default and Salient Meanings in GPs

243

- the use of possessive to determine different relations (e.g. “John’s book” +> “the one he read/wrote/borrowed”; “Wendy’s children” vs. “Wendy’s house”) - generality-narrowing inferences drawn from lexemes (e.g. “a secretary” +> a female, “a drink” +> an alcoholic drink) - specialisation of spatial terms and the use of prepositions (e.g. “the nail is in the wood” +> buried, “a toothbrush is in the mug” vs. “tea is in the mug”) - noun-noun compounds, in which the two nouns enjoy various relationships, based on stereotypical assumptions and thus the compounds carry conventionally induced interpretations (e.g. “the oil compressor gauge” +> a gauge used to measure the state of compressor that compresses oil; “a steel knife” +> made of steel vs. “a bread knife” +> for cutting bread)

The second class of inference, apart from the reference to a stereotype, embraces cases of “the maximisation of coherence, the minimisation of postulated entities and the presumptive enrichment of relations” (Levinson 1995: 101), examples of which are (Levinson 1995: 101-102, 2000: 37-38, 117-118): - conjunction-buttressing, the enrichment of conjunctions by the assumption of temporal sequence, teology and causality (e.g. “and” +> “and then/therefore/in order to”) - conditional perfection (e.g. “If you do it, I’ll help you” +> “If you don’t do it, I won’t help you”) - local (proximal) coreference (e.g. He came and he sat down +> he is the same person) - negative-raising (e.g. “I don’t think this is so” +> “I think this is not so”) - strengthening negations from contradictories to contraries (e.g. “don’t like” +> “dislike”) - conjunction and parataxis: the assumption of conjoined subjects acting together (e.g. “John and Jenny bought a piano” +> “together”) - bridging, e.g. finding local antecedents (“The picnic was awful. The beer was warm.” +> “the beer is part of the picnic”)

All the constructions enumerated above as well as conventionalised phrasemes mentioned earlier can be conceived as conducive to covert ambiguity exploited in GPs. The inferences drawn on the basis of phrasemes and compounds can be thought of as lexically ambiguous chunks giving rise to puns. On the other hand, the remaining structures are, on the whole, instances of what can yield pragmatic ambiguity. It is important from the present perspective that the category of generality narrowing be divided into generality-narrowing consequent upon purely

244

Chapter Six

stereotypical inferences understood as presuppositions (e.g. secretary will be a female) and polysemy-based shifts of meaning (e.g. “a drink” being “an alcoholic drink”). Before presumptive meanings generated upon the Iheuristic are presented in detail as a methodological tool helping explain the workings of GPs, criticism against Levinson’s proposal must be rebutted.

3.3. Criticism of the model Jaszczolt (2005) criticises Levinson’s presumptive meanings and considers them a “waste basket” (Jaszczolt 2005: 26), due to their being too eclectic in origin and susceptible to contextual effects. Jaszczolt (2005, 2006a,b) observes that Levinson’s default interpretations are diversified and escape a coherent taxonomy. Some hinge on mental lexicon or grammar, whereas others depend on the way humans think or construct their social and cultural reality. Also, Carston (1995) disapproves of the heterogeneity of presumptive meanings, pointing out that some of them contribute to what is said, which corresponds to the relevance-theoretic concept of enrichment, while others refer to independent import an utterance communicates. Levinson, however, considers the multifariousness of presumptive meanings a merit, highlighting that they arise at different levels of generalisation, i.e. morphemes, words, constraints on non-lexicalised items, and syntactic patterns. As regards Carston’s argument, it should be highlighted that the two categories of implicatures are most legitimate and can be found in Grice’s discussions of implicatures and the notion of centrality, as already reported. Jaszczolt’s (2005) main reservation against Levinson’s approach is that while some meanings are uncontroversially made by default, others do entail some processing. She observes that some default meanings are context-free, because they stem from words and structures, while others are context-dependent, which is the case of pronouns or deixis. Jaszczolt emphasises that genuine defaults should arise prior to context processing or the speaker’s intention deciphering, although their intentionality is presupposed. Thus, default-generating jumping to conclusions is most effective when it proceeds along the path of preconceived beliefs in the form of defaults that do not require pragmatic inference. And this default reasoning is both intended and recognized as being intended. (Jaszczolt 2005: 46)

Indeed, defaults should emerge irrespective of context, which participates in meaning computation later in the inferential process.

Default and Salient Meanings in GPs

245

Admittedly, as already suggested, the most controversial example found in Levinson’s work is that of the Saxon’s genitive, the understanding of which, according to Jaszczolt (2005), entails background knowledge, a type of contextual effect. However, background knowledge should not be treated as context, which is the verbal or non-verbal surrounding of an utterance, but as a parameter affecting the salience/defaultness of some meanings on the hearer’s mind. What interlocutors have in their minds can be conceived of as the result of priming which affects the very defaultness of one interpretation (cf. section 4.3). Also, Jaszczolt (2005) denies the status of defaults to physical preposition-dependent presumptive meanings, which she claims to be context-dependent (e.g. “There is coffee is in the cup.” [i.e. the liquid fills the inside of the cup] vs. “There is a spoon is in the cup.” [i.e. the spoon protrudes from the cup and takes only a minor part of its inside]). Levinson’s standpoint may be defended on the grounds that the presumptive meaning trigger is not reduced to the preposition but is comprised of the whole prepositional phrase, which, in either of the instances provided, coincides also with a post-propositional inference. It is then not so that the noun within the prepositional phrase constitutes a context against which the meaning of the preposition is analysed. Additionally, Jaszczolt (2005) questions the defaultness of sentential connectives, given that they do not always carry default interpretations, and negative rising, which does not apply to all verbs of attitude. Although Jaszczolt’s observations are by no means entirely incorrect, it should be appreciated that Levinson says nothing of his list of phrases and structures being invariably conducive to default interpretations. Also Grice’s (1975/1989b) examples of GCIs originating from language use need not always be carried, being blocked by context. To circumvent the “problems” posed by Levinson’s intermediate level of meaning and the eclectic nature of defaults, Jaszczolt proposes to dichotomise them into social-cultural and cognitive defaults. Jaszczolt deems cognitive defaults as rooted in the human thinking process and social-cultural defaults as depending on the way society and culture are organised. Cognitive defaults are, in Jaszczolt’s (2005) view, stronger defaults and derive from the intentionality of mental states and their corresponding intentional linguistic expressions. However, they arise without any computation of the speaker’s intentions or contextual import. Social-cultural defaults are non-cognitive types of defaults (Jaszczolt 2005), which are also “pre-inferential defaults” subject to “instantaneous, automatic interpretation” (Jaszczolt 2005: 55) and appear on the level of propositional interpretation of implicatures. Social-cultural defaults include gender-related interpretations (e.g. a nurse being a female) and lie

246

Chapter Six

close to sociological and anthropological studies. This is why Jaszczolt chooses not to concentrate on them in detail. Also, in her opinion, they are similar to conscious pragmatic inference, while the caesura between them is very difficult, if not impossible, to delineate. Conversely, because of social norms and patterns (including those Jaszczolt mentions), some interpretations are taken for granted and not given any conscious thought at all. Although initially feasible, Jaszczolt’s dichotomisation of defaults may be considered less plausible when given a second thought. Could it not be so that socio-cultural defaults are actually cognitive by nature? Societal norms do determine the ways in which humans think. Therefore, social-cultural defaults are also grounded in human cognition although their provenance is partly societal norms and stereotypes. All in all, Jaszczolt’s perspective is not supported here, the reasons being several. In contrast to Levinson’s presumptive meanings, Jaszczolt’s defaults arise post-propositionally after an utterance has been completed. Also, what is most significant, Jaszczolt (2005) is sceptical about the cancellability of default inferences (even if she refrains from rejecting this postulate entirely), questioning the stage at which this should occur. Cancellability is, nonetheless, a crucial feature of default meanings exploited in GPs.

3.4. GCIs and ambiguity resolution In his short discussion of syntactic and lexical disambiguation, Levinson (2000) observes that a given chunk’s stereotypical reading and the maxim of Relevance (Sperber and Wilson 1986/1995) collaborate to foreground one reading of an ambiguous lexeme. It is worth noting that Levinson’s mention of relevance contradicts the primary neo-Gricean premise of non-monotonic inferencing and the secondary function of context in the interpretation of utterances. Only later in the interpretation process, after the disambiguation has taken place by default, will context become significant. Therefore, the extrapolation that ambiguity resolution is facilitated by co-textual relevance is not espoused here. The perspective supported here is that, since interpretation proceeds incrementally, the ambiguous element per se, together with the elements directly preceding and following it, may be seen as an entity (not a word and its context) manifesting its default interpretation, even though it does not constitute a set collocation. This is how the notion of Levinson’s default could be extended to cover more GPs (see the discussion in the next section). Most significantly, Levinson (2000) posits that GCIs can partake in the disambiguation process, especially through I-inferences to stereotypical

Default and Salient Meanings in GPs

247

extensions, because they are mutually accepted by the speaker and the hearer as the most predictable (salient and stereotypical) interpretations. (…) the speaker must presume that the addressee will use the same salient stereotype (regardless of its factual probabilities) to resolve the ambiguity (…) [while the recipient] works on the assumption that the speaker would so design his utterance that, other things being equal, the predictable GCI would go through. (Levinson 2000: 176)

Levinson (2000) thus perceives the resolution of lexical ambiguities as, at least partly, reliant on nonmonotonic and defeasible GCIs, i.e. utterancetype meanings. Disambiguation should also not be treated as a conscious inferential process. Also, only if proven contextually inappropriate, do default interpretations evaporate. This is a most significant postulate, given the interpretational processes underlying GPs, in which defaults are maintained by both the speaker and the hearer until the co-text in the form of the switch clashes with it, necessitating the enforcement of a second meaning. To recapitulate, Levinson’s model of GCIs as defaults resolving ambiguities is applicable primarily to pragmatic linguistically-based ambiguity, corresponding to some of the structures under the I-heuristic, expounded on below, and multi-word polysemies (pragmatic and semantic phrasemes), which reverberate in Levinson’s work. Conversely, presumptive meanings appear to correspond to single-word lexical ambiguities with two lexicalised meanings, only on condition that the words are computed as inextricably connected with adjacent segments of the text analysed on-line. Given that the model is not, and cannot be, equipped with the requisite postulates on the resolution of lexical ambiguities alone, one may assume that this type of ambiguity needs to be explained by a different theory, not necessarily pragmatic. This theory is Giora’s (1997, 2003) salience (see section 4).

3.5. GPs and presumptive meanings’ cancellation In the light of the above, it is extrapolated that the jocular speaker exploits presumptive meanings by purposefully communicating and then invalidating them in order to generate GPs. The interpreter appreciates the most stereotypical or otherwise obvious interpretation, as intended by the speaker, only to have to subvert it in the light of the switch, whereby the presumptive meaning is ousted and another non-default inference needs to be made.

248

Chapter Six

Several of the mechanisms presented as conducive to I-implicatures, as well as interpretations of semantic and pragmatic phrasemes may serve as a basis for the explanation of the workings of pragmatic GPs reliant on linguistically-motivated inferential ambiguity. The exception is generality narrowing, which will be responsible for what can be regarded as presuppositional ambiguity. All those forms yield defaults, which can be cancelled. Money can’t buy you everything. That’s what credit cards are for. [conventional reading of a pragmatic phraseme: cliché, saying] (not everything can be purchased with money, e.g. relationships, emotions ĺ money sometimes does not suffice, but one can use credit cards then) He’s not fine. Last week he started hitting the bottle, thinking it was his wife. [conventional reading of a semantic phraseme: idiom] (the man has problems and, wanting to forget about them, started drinking a lot of alcohol ĺ the man has psychiatric problems, as he started beating a bottle, thinking it was his wife) My wife made chicken soup. I was lucky that muck wasn’t for me. [noun-noun compound transformed into syntactic ambiguity] (the speaker’s wife prepared soup with chicken meat ĺ the speaker’s wife had made some food for their chicken, rather than soup with chicken meat for him) She can play with her kid all day long. Problems start when she bounces him too hard. [conjunction] (she plays games together with her child ĺ she uses her child as a ball) Every morning I exercise. Up and down, up and down and then the other eyelid. [generality narrowing] (the speaker exercises by making upward and downward movements, e.g. squatting or bending ĺ the speaker only tries to open his/her eyes, moving his/her eyelids upwards and downwards) I like long walks if they are taken by people who annoy me. [local coreference] (the speaker has pleasure in walking long distances ĺ the speaker likes annoying people to take long walks) If I turn into my parents, either I’ll be an alcoholic blonde chasing young boys or I’ll wind up like my mother.

Default and Salient Meanings in GPs

249

[generality narrowing, cf. presupposition] (if the speaker resembles her parents, she will be like her mother, an alcoholic blonde woman pursuing young boys ĺ the speaker will be like her father, who is a blond alcoholic pursuing boys, or she will be like her mother) It’s not that some women look for rich men, who will guarantee them life stability. All look for such men. [negative rising and scalar Q-heuristic] (it is not true that some percentage of women look for rich men ĺ not some but all look for such men) I don’t dislike my work. I practically loathe it. [strengthening negations from contradictories to contraries] (the speaker does not dislike, i.e. likes, his/her work ĺ the speaker not only dislikes his/her work but hates it)

Additionally, although Levinson does not elaborate on this, the model will successfully embrace purely pragmatic inferences, which might be dubbed post-propositional defaults, some of which also involve mechanisms which Levinson mentions, e.g. bridging. Post-propositional defaults are not prompted by any specific elements in the surface structure but emerge in result of the computation of all the utterance elements. They yield pragmatic GPs based on what is here called purely inferential ambiguity. I saw six men kicking and punching my mother-in-law. I wasn’t going to help. I thought six should be enough. [post-propositional default, bridging] (the speaker was not going to help his mother-in-law who was being kicked and punched ĺ the speaker wasn’t going to help the assailants, since there were as many as six of them) My French is getting better. Yesterday I ordered a meal in French in such a way that the waiter was astounded. I was in a Chinese restaurant. [post-propositional default, bridging] (the French waiter was astounded by the high quality of the non-native speaker’s verbal performance in French ĺ the waiter was astounded by the very fact of French being used, because it was a Chinese restaurant and probably because he was Chinese as well) Saving is worth your while. Save some money every month and at the end of the year you’ll be surprised how little you have. [post-propositional default]

250

Chapter Six (saving is beneficial because if someone saves money every month, at the end of the year, the amount they have garnered can be surprisingly high ĺ saving is beneficial, because it shows that if someone saves money every month, at the end of the year, he/she will discover that the amount he/she has garnered is surprisingly low)

A GP-related lacuna in Levinson’s model of GCIs, as in the case of Grice’s model, is single-word homonymous and polysemous ambiguities. Nevertheless, contrary to Grice’s, Levinson’s framework easily accommodates linguistic phenomena which arise at the strictly lexical level, entailing cancellability of perfectly conventional content, although not always entirely literal (e.g. metaphoric polysemies). However, although Levinson’s model does not rule out single-word lexical ambiguity as the pivot of a default interpretation, the sole argument that this is the consequence of conventional or stereotypical meaning of an ambiguous lexeme does not suffice, inasmuch as both meanings are such. As theorised in the previous section, if lexical ambiguities are to be considered defaults, they should be perceived as meanings generated incrementally in the light of the neighbouring (preceding/following) textual chunk. Within the Gricean model, this step cannot be regarded as the impact of context. Consequently, it can protested that there are ambiguities whose interpretation depends on the adjacent (collocated) elements, e.g. the meaning of a verb may rely on whether the subject and the object are animate or not. It is textual chunks containing single-word lexical ambiguities, sometimes even amounting to whole lead-ups, that may be regarded as units on which defaults are built. Why one and not the other meaning of an ambiguous lexeme alone is appreciated first can only be explained from the perspective of cognitive studies, which is done in the second part of this chapter. Nobody supports Clinton; he can still walk by himself. [conventional meaning of lexical ambiguity “someone supports someone”] (Clinton has no political supporters ĺ Clinton does not need to lean on anybody) They charged John with a Duracell battery. [conventional meaning of lexical ambiguity “somebody charges someone with a crime”] (John was accused in court of breaking the law ĺ John had a battery installed in him) I’ve had three husbands, apart from my own. [conventional meaning of lexical ambiguity “have a husband”]

Default and Salient Meanings in GPs

251

(the speaker has been married three times ĺ the speaker has been married once but has had affairs with three other married men) I was a chair at yesterday’s meeting. I was inert and I said nothing. [conventional meaning of lexical ambiguity “someone is a chair (at a meeting)”] (the speaker chaired the meeting ĺ the speaker behaved as if he/she were a chair, i.e. was motionless and said nothing]

In conclusion, Levinson’s GCIs, which constitute a peculiar level between what is said and what is implicated, are useful heuristic tools in the analysis of GPs capitalising on all pragmatic as well as phrasemic polysemous ambiguities. GCIs conceptualised by Levinson are generated by default, most often locally, i.e. pre-propositionally. Whether or not default interpretations arise from conventionalised lexemes, they cannot be deemed fully conventional, since they need not be invariably carried and can be defeated in the light of the context understood as the remainder of an utterance following the chunk subject to default interpretation. As regards their GP explanatory power, Levinson’s GCIs have a few advantages over Grice’s GCIs, in spite of a number of overlaps between them. Firstly, they may arise pre-propositionally. Secondly, presumptive meanings often correspond to inferences originating from conventional meanings of verbal constituents, which could not be classed as implicatures and would more likely partake in what is said in the Gricean model, hence being resistant to defeasibility. Pragmatic models are lacking in postulates explaining the workings of GPs couched in two conventional meanings of single-word lexemes. The aim is now to account for the remaining types of GPs, i.e. those centring on single-word ambiguous lexemes, which have two legitimate conventional meanings. Explaining why one of such interpretations gains priority over another is tenable thanks to Giora’s concept of salience, which will also account for the operation of lexeme-based presuppositional GPs.

4. Giora’s salient meanings8 Lexical ambiguity seems to play a central role in the cognitivist debate on whether language interpretation processes are interactive or modular, and thus whether the evocation of meaning is context-determined or not. 8

Given the extensive laboratory research on lexical processing, nearly all the references in section 4 are just a small sample of the topical literature.

252

Chapter Six

The bone of contention is then whether the meanings of an ambiguous lexeme are activated as affected by context (co-text) or whether they are activated regardless of it. Research data and theories in the field of processing of alternative meanings appear to vary, often giving rise to contradictory results. A number of tenets within these two approaches will be succinctly discussed with a view to arriving at the model of defaults realised as salient meanings (Giora 1997, 1999, 2003) explaining the workings of GPs contingent on lexical ambiguity and presuppositional ambiguity reliant on particular lexemes. The choice of the most salient meanings for GP pivots guarantees GPs’ humorousness. An additional provision is made regarding equibiased ambiguities which, outside any context, manifest two (almost) equally salient readings. The hearer will then activate one meaning after another, while the order of activation might be idiosyncratic. This does not occur in GPs, inasmuch as salience embraces also the parameter of givenness achieved via lexical/semantic priming. In most GPs, however, the context, primarily the co-text in the lead-up, will not be anyhow biasing and will concur with the salient interpretation, which is ultimately cancelled for humorous purposes at the stage of the switch, which brings out the nonsalient or less salient interpretation of the ambiguity.

4.1. Competitive models of lexical processing Many classical psycholinguistic models of language comprehension operate on the assumption that word-level (lexical) processing is the sine qua non of higher-level processing. Therefore, they divide the word comprehension process into two stages, lexical and post-lexical (e.g. Fodor 1983, Forster 1981). Lexical processing is the fast automatic stage of processing entailing, firstly, recognition of the phonemic or orthographic input, which is compared to templates in the mental lexicon, and, secondly, access, in which syntactic and semantic information associated with the word is activated. In the consecutive stages of post-lexical processing, much slower and controlled, linguistic information is combined with contextual and background knowledge in order to integrate the new word with the developing representation at the message level. The two stages, i.e. lexical and post-lexical processing, are determined by lexical access dependent on the ease with which a meaning is activated. This ease, on the other hand, is the ultimate result of a number of parameters, the primary of which are frequency and recency effects. The frequency effect means that less time is necessary to identify and access high frequency words, and thus their lexical information, than low

Default and Salient Meanings in GPs

253

frequency words (e.g. Balota and Chumbley 1984, Forster and Chambers 1973, Taft 1979, Whaley 1978). Secondly, tanks to the semantic priming effect, it is easier to interpret and respond to a word when it is preceded by a semantically or associatively related one (e.g. Neely 1977, Meyer and Schvaneveldt 1971, Meyer at al. 1975, Fischler 1977, Tweedy et al. 1977). A pertinent question is whether context does not actually come into play at the stage of accessing representations from the mental lexicon, rather than later during the processing of the entire message representation, especially the preceding co-text. There are two classes of hypotheses, both borne out with laboratory research, explaining how and when a biasing context aids the final selection of the relevant reading of an ambiguous lexeme. The fact that two competitive approaches garner support must be the outcome of the choice of methodologies and parameters which affect the results (Swinney 1979). The interactionist view (Simpson 1981, 1994) favours the impact of context on processing. Prior contextual information directs the access of a single relevant meaning and it is only this contextually compatible one that is activated thanks to the early interaction of heavily weighted contextual information. In other words, lexical access is influenced by context and the only interpretation generated is determined the prior sentence context (e.g. Glucksberg et al. 1986, Simpson 1981, see Giora 2003, Gorfein 1989, Simpson 1994, Small et al. 1988). The developments of the model are the selective access model (e.g. Simpson 1981, Simpson and Krueger 1991, Tabossi 1988) or the context-sensitive model (e.g. Martin et al.1999, Paul et al. 1992). Within the selective access model, a constraining context interacts with lexical processes, facilitating the interpreter’s arrival at the exclusive and adequate interpretation, without entailing any contextually inappropriate understandings at the initial stage. According to the contextsensitive model, the activation of meanings is selective, but either meaning frequency or a biasing context can be influential, depending on its strength. In essence, human cognition is governed by a single interactive mechanism benefiting from both linguistic and extra-linguistic information and activating exclusively contextually compatible interpretations, which is corroborated by experiments (e.g. Vu and Paul 1998, Vu et al. 2000). A question arises how this approach would account for what meaning is perceived first if the stimulus occurs in isolation or in a practically neutral context. Such phenomena will occur very rarely in real-life discourse but cannot be entirely ruled out, either. The exhaustive access model (e.g. Conrad 1974, Lucas 1987, Onifer and Swinney 1981, Swinney 1979) is a development upon the modularbased model (Fodor 1983, Forster 1979), according to which the initial

254

Chapter Six

stage of comprehension is stimulus-driven, and hence impervious to context or world knowledge. As a consequence, more than one, if not all, meanings of an ambiguous word or expression are observed. The exhaustive access theories maintain that context has no impact on the initial activation of a word’s meanings and that multiple meanings are acknowledged. In other words, lexical access is exhaustive and all the meanings of an ambiguous word are activated even irrespective of a contextual bias in favour of one of them. Context affects the output of the lexical module later in the inferential process (see Giora 2003, Gorfein 1989, Simpson 1994, Small et al. 1988). An important offshoot of this approach is the ordered access model (e.g. Hogaboam and Perfetti 1975), which presupposes that access to meanings is ordered and conditioned by the frequency of interpretations of ambiguous words. Analysing the research findings collectively, one may gather that a compromise is needed between pure context-determined and contextinsensitive models concerning the interpretation of ambiguities. Indeed, a few compromising theories are launched. The constraint satisfaction approach (MacDonald et al. 1994) hinges on the premise that meaning selection is governed by contextual factors and lexical factors limiting the former, such as an ambiguous word’s meaning frequency. The impact of context appears to be weaker than that of the bottom-up lexical information (MacDonald et al. 1994, Trueswell and Tanenhaus 1994). In essence, context can affect the interpretation of ambiguities, even if “it cannot overcome strong frequency biases to promote a subordinate interpretation over the frequency dominant alternative” (McDonald et al. 1994: 693). Similarly, lexical ambiguity processing is explained via the hybrid model, known as the reordered access model (Duffy et al. 1988, Rayner et al. 1994, cf. Simpson 1994), which suggests that all meanings can be activated, but the order of activation will be sensitive to the relative frequency of words, i.e. the comprehender’s earlier experience, context notwithstanding. On the other hand, prior context after which an ambiguous element occurs may activate the contextually appropriate meaning, even when less salient. Therefore, less frequent meanings are selected only when compatible with their context. Giora (2003) also conducts a methodical survey of these competing views, highlighting their weakness and proposes another model as reconciliation between them, i.e. the graded salience hypothesis (Giora 1997, 1999, 2003), verifying it with extensive research (Giora et al. 1998; Giora and Fein 1999; Peleg et al. 2001, 2004, 2008, inter alia). Giora embarks on her discussion focused on the literal language vs. figurative language distinction and competitive proposals of meaning evocation,

Default and Salient Meanings in GPs

255

championing salience as reconciliation between those mutually exclusive approaches. The approach is rooted in the thesis that it is coded meanings, rather than context or literal meanings, that govern human interpretation of language phenomena such as those entailing polysemous or homonymous ambiguity. The access of meanings is exhaustive, salience sensitive and unchangeable across contexts. Giora’s proposal is in many ways similar to the reordered access view. Both the views presuppose the exhaustive access of meanings, which is sensitive to lexical information (i.e. degree of salience) and contextual information. The recipient’s prior experience with an ambiguous word may render one meaning more salient than the other, which is why this meaning is accessed faster, regardless of context. On the other hand, prior semantic information may be strong enough to activate the contextually appropriate meaning, even if objectively less salient (Duffy et al. 1988, Rayner et al. 1994). However, the point of divergence between the two models is that the graded salience hypothesis does not allow contextual processes to interact with lexical access, which seems to be the case of the reordered access view (Giora 2003, Peleg et al. 2008). According to Giora (2003), comprehension involves two distinct mechanisms, linguistic and contextual. Contextual processes constitute the global top-down mechanism integrating linguistic and nonlinguistic information, while the linguistic mechanism is stimulus-driven, bottom-up and modular. The activation of meanings is automatic and impervious to context effects, which is why also irrelevant but salient ones may be activated. Giora (2003) thus posits that the stimulus-driven and data-driven (contextual) processes do not interact but run in parallel without interacting and distinguishes two stages in the process of comprehension, i.e. initial activation (i.e. lexical stage) and the subsequent integration processes. At the first stage, salient meaning(s) will be automatically (and consecutively) accessed in response to a linguistic stimulus, either from the lexicon (in response to stimuli) or, sometimes, as a result of a highly predictive context. Such predictive contextual support does not coincide with the priming effect (see section 4.3), which does not penetrate the lexicon and is an inherent component of salience. Meanings are activated one after another according to their salience, while strong prior semantic information may render one contextually adequate meaning the first available (Peleg et al. 2008). In the consecutive integration phase, the meanings activated are integrated with contextual information. The integration processes manifest themselves as retention or suppression. A salient meaning which has reached the activation threshold is rejected as soon as it is regarded as incompatible with its context, especially when it interferes with interpretation processes. Obviously, if the outputs of

256

Chapter Six

contextual and lexical processes accidentally match, the integration process is facilitated. This will normally be the case of salient meanings exploited in GPs, where the initial interpretation of lexical ambiguity is unquestionably the most salient one and gains support from the co-text within the lead-up. Giora’s fundamental extrapolation is that no context can entirely block the activation of the salient meaning, even if it is biasing towards the less salient one, in which case the more salient meaning is activated, albeit later than the contextually supported one. Therefore, while contextual factors can hasten the choice of an appropriate meaning at the integration stage, they are unlikely to hinder the earlier activation of incompatible salient meanings, which will be later rejected. Additionally, context supportive of a nonsalient (noncoded) or simply a less salient, meaning downplays the role of the contextually inappropriate salient meaning at the final stage of processing, yielding the ultimate interpretation of a verbalisation. It can be concluded that the graded salience hypothesis does not discredit a host of pragmatic theories or even cognitive postulates (e.g. Coulson 2001, Coulson et al. 2006) which highlight the on-line formation of meanings and the interdependence between context and inferences. Admittedly, such models focus on the ultimate effect of discourse interpretation, while the salience approach concentrates on the interpretation of meanings from a cognitive perspective, starting from the initial, and often, very short imperceptible stages. This is why Levinson’s framework, which may incorporate default inferences based on lexical ambiguities, cannot elucidate why one meaning is more easily accessible than the other, without seeking context support, thereby undermining the tenet of non-monotonic inferencing. By contrast, in Giora’s view, context, whether preceding or following, comes into play later, causing the retention or suppression of activated meanings depending on their relevance. Giora’s findings, on the whole, tie in with earlier research data (e.g. Swinney 1979, Gernsbacher 1990, Rayner et al. 1994, Duffy et al. 1988, MacDonald et al.1994, Trueswell and Tanenhaus 1994), which argue against the selective access/interaction view. Concurring evidence comes also from a number of other researchers. For instance, De Palma and Weiner (1992) and Weiner and de Palma (1993) also define meaning accessibility as the correlate of prototypicality, salience and parallelism. The underlying premise is that the units stored in human memory including the submeanings of polysemous words are not equal in accessibility or salience. In addition, “dominant” meanings of polysemous

Default and Salient Meanings in GPs

257

words are apt to emerge even in contexts biased towards the “subordinate” meanings (Durkin and Manning 1989). Similar findings are reported by Swaab et al. (2003), Tabossi et al. (1987) and Tabossi (1988), who substantiate that the “dominant” meaning of an ambiguous lexical item is always accessed irrespective of its context.

4.2. Salience9 The most salient meanings (Giora 1997, 1999, 2003) coded in the mental lexicon are normally accessed instantly, irrespective of any contextual information. In essence, Giora puts forward three theses: a. Salient interpretation has unconditional priority over less salient interpretation: The most salient meaning of a word or an utterance is always activated. b. A novel interpretation of a salient meaning involves a sequential process, whereby the salient meaning is rejected as the intended meaning and reinterpreted. The more salient the reinterpreted language, the more difficult it is to reject as the intended meaning. c. Novel interpretation must be more difficult to derive; it should require more and different contextual support for its derivation. (Giora 1997: 200)

Salient (lexicalised) meanings are those coded in the mental lexicon of an individual or community members, being thus “foremost on our mind” (Giora 2003: 10), owing to their higher degree of conventionality, frequency, familiarity and prototypicality (stereotypicality) (Giora 1999, 2003, see references therein). Each of the parameters increases the speed and ease with which meanings are activated. Conventionality is viewed as linguistic regularity which a population has tacitly agreed to conform to, for the sake of uniformity (Nunberg et al. 1994). Frequency, familiarity (e.g. Gernsbacher 1984) and prototypicality (Rosch 1975) are dependent

9

This conceptualisation of salience should not be mistaken for the salience hypothesis proposed by some psychologists (though, for example, Goldstein et al. 1972 hold a view similar to this presented here). In the strictly psychological perspective (see for example Kuhlman, 1985), a stimulus is salient when it is of importance for the subject, which means that it has a cognitive, emotional and motivational value (the property to elicit a response). On the other hand, in neuroscience, the salience of an item is its state or quality of standing out relative to neighbouring items. I am grateful to Giovannantonio Forabosco for pointing this out to me.

258

Chapter Six

on the objective occurrence of an element in the mental lexicon, but not necessarily in the referential or real word. As Giora defines it, salience relates to the entrenchment status of stored meanings at a given time in a given community or, more precisely, in the mind of a specific individual, affected by exposure – that is, by such factors as familiarity, conventionality, and frequency of occurrence. (Giora 2003: 55)

What is of utmost importance, the salience of an ambiguous lexeme’s meanings may be regarded in the same way by a whole language society, but there are also idiosyncratic salience perceptions. It might be argued that a social group’s (even society’s) identical interpretation of an ambiguous chunk’s salient meaning is only the consequence of the fact that individual perceptions concur. Unanimous activation of the same salient meaning is fundamental to the operation of GPs, which ought to be deemed equally by virtually all competent language users. It is also noteworthy that meanings of ambiguous lexemes may manifest various salience imbalance realisations, i.e. salient vs. less salient meanings and salient vs. nonsalient, i.e. noncoded meanings. Although both the meaning juxtapositions may be exploited in GPs, the latter may be considered “safer”, since the perception process will infallibly manifest one pattern, based on one conventional interpretation. Giora (1997, 1999, 2003) corroborates her postulates on the basis of clear instances manifesting salience imbalance for language users, salient and nonsalient meanings, represented by (dead/conventionalised or novel) metaphors, idioms and (conventionalised or non-conventionalised) irony, thus differentiating between conventional and unconventional chunks. The compositional parts of an unlexicalised phrase carry the potential for retaining their meanings if those are salient, which renders more difficult the non-literal understanding of the whole chunk treated as an entity. Conversely, the derivation of compositional meanings may be obstructed in the light of the standardised/lexicalised (and thus more salient) meaning of the entire conventionalised chunk, i.e. a phraseme. Therefore, when the most salient meaning (e.g. the figurative meaning of an idiom) is the one intended by the speaker, it is accessed directly, without unnecessary processing of the literal but considerably less salient meanings first. By way of contrast, when a nonsalient meaning is intended, such as the literal meanings of words constituting an idiom, the more salient idiomatic meaning is initially processed, before the intended meaning can be observed. For example, one may imagine such an interchange between two people: “Why is the water spilt all over the floor?” “I kicked the bucket.” In this case, the salient interpretation of the idiom will likely emerge first but will

Default and Salient Meanings in GPs

259

have to be cancelled in favour of the contextually relevant interpretation. The latter process is used in each GP which is determined by the hearer’s choice of the salient meaning of the lead-up dependent on a semantic phraseme (a set collocation, an idiom including a dead metaphor) or a pragmatic phraseme (a cliché, a proverb or a saying). This stage is followed by the rejection of the salient interpretation in favour of the nonsalient one, on the strength of the switch, which can be conceived as the following heavy context. Attention, animal lovers! What you are doing is illegal! [set collocation with one conventional/salient interpretation] (some information will be directed at people fond of animals ĺ the information, directed at people who have affairs/sex with animals, is that their actions are illegal) It’s not hard to meet expenses. They’re everywhere. [idiom/set collocation with one conventional/salient interpretation] (it is not difficult to pay money for certain purposes ĺ it is not difficult to encounter situations when one has to pay) Save the whales. Collect the whole set. [cliché with one conventional/salient interpretation] (prevent the whales from extinction ĺ collect whales)

Moreover, salience imbalance will characterise homonymous and polysemous meanings of single ambiguous words exploited in GPs. Even though two meanings of an ambiguous element can be salient and equally legitimate, one must show greater salience than the other to be observed effortlessly. It should also be remembered that in the case of polysemy, the basic meaning which is known to have given rise to another meanings need not be the most salient one. What matters is the familiarity and frequency of meanings. However, sometimes, if two are possible, one of the salient meanings may need to be strengthened by its givenness obtained via the priming of contextual effects.

4.3. Salience, priming and predictive prior co-text In her discussions on jokes, with their scope narrowed down to instances operating on what is here called the GP mechanism based on lexical ambiguity (revisited in section 7, chapter 3), Giora (1991, 2003) describes the effects of salience as being crucial for a humorous surprise. This is exactly the view advocated here. Nevertheless, Giora appears to focus only on ambiguous lexemes showing conspicuous salience imbalance

260

Chapter Six

determined by their conventionality and prototypicality. Indeed, in GPs, the ambiguous element is most often placed in the initial sentential position and is later supported by the neighbouring co-text within the leadup, while the preceding co-text, if present at all, is neither priming nor biasing/constraining/predictive. However, an ambiguous chunk with two (almost equally) salient meanings and no striking salience imbalance between them may also be the foundation for a GP. Still, even is such a case, only one of the meanings is the most salient and hence the only one that is initially available, guaranteeing a jocular text’s effectiveness. This is the effect of yet another correlate of salience, i.e. givenness resulting from the preceding co-text governing the interpretation of an equibiased ambiguous chunk. Hence, the context primes one of its salient readings so that both are not almost equally salient, which would put paid to the humorous effect. In essence, if only one meaning is available, it must be the most salient one, at least more salient than the alternative salient meaning (Giora, personal communication). The fact that the other, less salient interpretation is not computed at once is also motivated at the integration stage by the co-textual appropriateness of the first most salient meaning, which is sustained as relevant until the switch emerges. The notion of the priming effect derives from cognitive psychology, where it was originally proposed in reference to non-verbal stimuli perception. It is widely acknowledged in the field of cognitive psychology that human understanding is geared towards the reduction of multiple meanings and ambiguities, which holds for various stimuli. Leeper (1935), the forerunner of this strand of research, is known to have proven in his classic study how the perception of visual stimuli is determined by prior context, while outside such context, the two alternative meanings are equally likely to emerge. To study the role of a perceptual set, Leeper (1935) had the image of an old croon vs. young woman redrawn in two biased forms emphasising either interpretation. The results showed that, on the whole, primed subjects instantly perceived the ambiguous picture in conformity to the stimulus they had seen first.

Default and Salient Meanings in GPs

261

10

On the other hand, without any earlier bias, the order of salient interpretations’ activation may vary from subject to subject. This is also possible, according to the graded salience hypothesis, which allows for idiosyncrasies in the perception of one meaning’s priority (Giora 1995, 1999, 2003). Theoretically speaking, the competitive interpretations appear to be equally accessible, i.e. salient, and neither can be unanimously given precedence as more salient. Without priming, which interpretation comes first appears to be determined by the idiosyncrasies of a given individual (present state of mind, preferences, past experience, etc.). Notably, even if observed almost instantly, the two interpretations cannot be perceived simultaneously and one always takes precedence over the other, after which they can be appreciated alternately. The process of perception of two alternative meanings is dual and repetitive (e.g. Attneave 1971, Howard 1961, Orbach et al. 1963). Research on priming is, obviously, not restricted to pictorial stimuli. Bruner and Postman (1949) proved that particular situational contexts set up expectations in the observer. Also, in their famous studies, Bugelski and Alampay (1961) or Palmer (1975) attested that humans have a predisposition to perceive something in relation to prior perceptual experiences, while Higgins and King (1981) substantiated that a concept which has recently been primed is particularly likely to be applied to the interpretation of incoming information, even in unrelated contexts. In essence, it emerges that what is the most salient on the human mind can be affected by givenness, i.e. what the individual has been exposed to via priming before encountering the target stimulus. Priming can, therefore, affect salience imbalance of meanings. This is also easily appreciated on the basis of a common verbal game in which the respondent not only is biased in favour of some concept but also actively produces an answer, inadequate though it is.

10

This image juxtaposing a young woman with an old woman, originally created by a British cartoonist, W. E. Hill, in 1915, was introduced into the psychological literature by Boring (1930).

262

Chapter Six A: What is the colour of a clean sheet of paper? B: White. A: What is the colour of snow? B: White. A: What is the colour of albino rabbits? B: White. A: What is the colour of nice healthy teeth? B: White. A: What is the colour of a bride’s dress? B: White. A: What do cows drink? B: Milk. (not water)

Because of the preceding questions, the responder’s attention is focused on the primed white colour and he/she is very likely to keep responding in conformity to the already established pattern. The final wrong answer is also primed by the reference to the animal. This is the effect of a human tendency to introduce new items consistent with the activated schema (Brewer and Treyens 1981). In addition, studies on cognitive processing of meanings across sensory modalities which centre on the impact of lexical priming corroborate the facilitation of lexical decisions concerning ambiguities in both simultaneous and successive presentation of semantically related words (e.g. Neely 1977, Meyer and Schvaneveldt 1971, Meyer at al. 1975, Fischler 1977, Tweedy et al. 1977). On the whole, within the modular view, research on the semantic associative (lexical) priming effect points to the facilitated processing for a word or its associates if a relevant one has already been processed, for instance “a table” primes the meaning of “a chair” as “a piece of furniture”. Another example presenting this phenomenon is the following hackneyed one-liner: “The court bailiff found the defendant below the bar.” The word “bar” outside any context will, in all likelihood, be interpreted as “a counter in a pub” or “a pub”, rather than as a reference to the barrister profession, which might be the more easily available interpretation, on the strength of the preceding cotext. Such priming is “attributed to the automatic activation between related meanings in the lexicon” (Giora 2003: 22). A prior priming context is not regarded as inducing contextual effects, inasmuch as its impact is intralexical, i.e. operating inside the input system of the mental lexicon (cf. Fodor 1983). In other words, lexical priming is a not a contextual effect, because it occurs in the lexicon, which is why it is not a strong test for the salience approach. On the contrary, priming affects givenness and thus co-determines salience imbalance.

Default and Salient Meanings in GPs

263

What Giora considers a strong contextual effect, and a reliable test for salience is a predictive context facilitating a concept, a statement, or a (figurative) mode of speech, which should by no means be equated with a priming context. Giora and co-researchers’ theses on the impact of predictive context as not blocking irrelevant but salient meanings are perhaps not of utmost importance to the operation of GPs, where the (optional) preceding context only supports the most salient meaning at the integration stage. If another (less) salient meaning were activated, besides the most salient meaning, due to a predictive co-text, the humorous effect would fall through. There are many findings on the impact of preceding context substantiated by studies on ordinary on-line sentence comprehension, in which prior context also biases the interpretation of an ambiguity before the latter actually occurs (e.g. Lackner and Garrett 1972; Blank and Foss 1978; Morton and Long 1976; Swinney and Hakes 1976; Swinney et al. 1979; Swinney 1979; Fodor 1983, McClelland 1987; McRae et al. 1998; Peleg and Eviatar 2008; Peleg et al. 2001, 2004, 2008; Vu and Paul 1998; Vu et al. 2000). However, viewpoints differ as to whether the salient but contextually inadequate meanings will also be activated (Rayner et al. 1994; Peleg and Eviatar 2008; Peleg et al. 2001, 2004, 2008) or will not be activated at all (e.g. McClelland 1987; McRae et al. 1998; Vu and Paul 1998, Vu et al. 2000). Still other research findings indicate that in the case of ambiguous words presented at the beginning of sentences or clauses, salient but incompatible meanings surface immediately, despite prior disambiguating contextual information to the contrary (e.g. Gibbs 1990, Duffy et al. 1988, Dopkins et al. 1992). In various papers, Giora and her co-researchers ascertain that experiments on the contextual impact only show that context can facilitate one meaning but does not restrict lexical access, and the alternatives, if salient, will also be activated. Giora (1997) hence appreciates context’s function in speeding up meaning recognition (Grosjean 1980, MarslenWilson and Tyler 1980, Tyler and Wessels 1983, 1985), with the salient meaning being activated as well, in spite of the prior context bias to the contrary (Rayner et al. 1994, Peleg and Eviatar 2008, Peleg et al. 2001, 2004, 2008). In addition, Peleg et al. (2000, 2008) raise the issue of the predictive mechanism being less/more effective depending on the sentence position of an ambiguity, which would explain some of the discrepancies in research findings on the influence of contextual effects. Meanings of ambiguities in the sentence initial position appear to be activated primarily thanks to the lexical mechanism, irrespective of any preceding contextual

264

Chapter Six

information to the contrary, while the contextual effect may be stronger in the final sentential position. The expectation-driven mechanism is faster toward the end than at the beginning of sentences, where different types of constraints (pragmatic, semantic, and syntactic) enable it to better predict an upcoming concept and thus activate compatible meanings even before the relevant lexical stimulus is encountered (as shown by Peleg et al., 2001). However, in sentence initial position, only constraints from previous context can be operative, which makes it somewhat harder for the predictive mechanism to come up with the next concept before lexical accessing occurs. (Peleg et al. 2008: 23)

Research (Peleg et al. 2001, 2004, 2008) shows that in both the sentence initial and the sentence final ambiguity positions, preceding context biased towards one salient (even less salient) meaning of an ambiguous word will not block the activation of the other salient meaning(s), which will emerge, even if only after the contextually adequate sense has been activated. To reformulate, although the less salient but contextually compatible meaning is primed and activated faster than the salient but inappropriate meaning, the latter is activated as well, in spite of its contextual misfit. Also, even though the final position benefits contextual effects, those do not interact with lexical processes and do not block salient but inappropriate meanings. Simultaneously, Peleg et al. (2001, 2004, 2008) propose that findings testifying that only contextually compatible meanings are activated (e.g. McClelland 1987, McRae et al. 1998, Vu and Paul 1998, Vu et al. 2000) are not necessarily the outcome of the interactive mechanism at the stage of lexical access but the integrative expectation-driven mechanism. In other words, strongly biasing context does play a role in the sense that guesses and inferential processes take place even before the lexical stimulus is encountered. Nonetheless, such an interpretation is not the effect of direct lexical access, in which case the alternative would not emerge. The direct lexical access of single salient meanings, optionally affected by priming context influencing intralexical operations, is the key determinant in the initial processing of ambiguities, including those exploited in GPs.

4.4. Equibiased ambiguities In GPs, ambiguity emerges at the sentence-initial stage, while the strength of one salient interpretation must be superior for the sake of a strong surprise effect consequent upon its ultimate cancellation at the stage

Default and Salient Meanings in GPs

265

of the switch. It must be the most salient meaning that is acknowledged initially. This sense also gains support from its neighbouring co-text within the lead-up at the integration stage (if the ambiguous text does not exhaust the text of the lead-up), before the switch comes and annuls it, simultaneously evoking a nonsalient or considerably less salient interpretation. Each ambiguity resulting from the use of a collocation, an idiom (also discussed by Giora), or a saying/proverb carries one salient lexicalised interpretation, the other meaning being nonconventional, nonsalient and latent. Even if prior predictive context should support the latter meaning (which does not happen in GPs), in such phrasemes, the priority of the salient meaning over the other nonsalient one is unquestionable. The same pertains to single-word lexemes which are characterised by strong salience imbalance. On the other hand, there are also equibiased ambiguities (cf. Saab et al. 2003) carrying two (sometimes more) almost equally salient interpretations, and thus manifesting imperceptible salience imbalance if produced in isolation or in a neutral context (e.g. “calf”: “young cow”/“the back of a leg”). Giora’s theory does not assume that it is invariably so that only one meaning is salient enough to be accessed first. If some meanings are of similar salience, they “will all be accessed at once/initially” (Giora, personal communication). Admittedly, whether two meanings manifest the same degree of salience, or whether there is some salience imbalance between them cannot be verified without advanced psycholinguistic research. For the present purposes, suffice it to say that in some cases, alternative interpretations are felt to be of similar salience, on the grounds that they are both conventional and frequently used. Besides, as Giora (2003) emphasises, salience of a given lexeme need not be perceived in exactly the same way by all society members. There exist lexemes with two meanings whose salient manifestations may vary in the perception of individuals, e.g. “key”: “test solutions”/“an instrument to open the door” for a teacher vs. for a person who is not a teacher. Similarly, the first salient interpretation of “a bank” as “a financial institution” or “a riveredge” depends on a language user’s background, experience, life habits, etc. Therefore, when such lexical items are produced, both their meanings might be activated one after another or, alternatively, the choice of one interpretation may differ from subject to subject. Such equivocation of meanings is, nonetheless, what cannot happen in GPs. One interpretation must be more easily available and, for the sake of a humorous effect, eclipse the competitive meaning until the switch fosters the latter. Consequently, it may be extrapolated that the parameter affecting the choice of the most salient meaning in the case of equibiased ambiguity,

266

Chapter Six

apart from the interpreter’s past experience, is the meaning’s givenness contingent on the prior context. In such a case, preceding context will prime one salient contextually adequate meaning (within the lead-up), precluding the necessity to activate any other meaning, although another salient meaning will exist. If the co-text supports one meaning of an ambiguous item, this one is selected more rapidly (Seidenberg et al. 1982; Swinney 1979; Gernsbacher 1990; Rayner et al. 1994; Duffy et al. 1988; MacDonald et al.1994; Trueswell and Tanenhaus 1994; Giora 2003; Peleg et al. 2001, 2004, 2008). This is of crucial importance if the same meaning is to be unanimously appreciated by the majority of language users, i.e. GP humour recipients. Priming co-text may then play a significant role as salience determinant, the alternative meaning being less salient, and hence initially latent. Such a view can by no means be regarded as being at odds with Giora’s studies. Indeed, their leitmotiv is that salience is the paramount determinant in the process of language production and reception and that it is only post-lexically that context becomes influential, suppressing the salient but inappropriate meaning, providing it interferes with activating the appropriate one. Nevertheless, Giora (1997: 185) defines salience as a function of conventionality, familiarity, frequency and “givenness status in a certain (linguistic and nonlinguistic) context.” The last parameter is of crucial significance, since it does allow for the contextual impact on salience via priming. Additionally, albeit from a different perspective, Giora (2003) does acknowledge, adducing evidence (e.g. Keysar et al. 1998; Fodor 1983; Peleg et al. 2001, 2004, 2008), that the preceding material may support salience and so may fasten the activation of one salient meaning. Directly preceding co-text can affect the “predictability of the oncoming effect” (Giora 2003: 23) and the expectations as to the development of the text (Fodor 1983, Rayner et al. 1999). Moreover, it could be extrapolated that because this meaning is not only the most salient but also relevant in the integration phase, no other salient alternative needs to be consciously activated at the stage of the lead-up. To recapitulate, punning and lexically-triggered presuppositional GPs, which operate on a salient vs. nonsalient/less salient pairs of meanings manifesting salience imbalance, will testify to and be explained by Giora’s conceptualisation of salience without fail. The observation that salience imbalance is affected by a priming co-text, which helps the recipient recognise one meaning faster than the other, will dispel the problem of potentially equibiased ambiguities in GP lead-ups. It is always the single unquestionably most salient meaning that is chosen, while the lead-up co-

Default and Salient Meanings in GPs

267

text surrounding the ambiguous lexeme (if present) is later also supportive of it.

4.5. GPs and salient meanings’ cancellation Salience, as conceptualised by Giora (1997, 1999, 2003), is postulated to determine the creation and understanding of GPs couched in lexical and lexically-triggered presuppositional ambiguities. The initial evocation of only one salient meaning of an ambiguous lexeme or a phraseme is a sine qua non for the GP humorous effect. In other words, both the interlocutors (essentially, community of speakers interacting in a given language) must deem the same interpretation as the most salient so that the GP effect can arise. Needless to say, language users are oblivious to the phenomenon but intuitively feel which meaning is the salient one. If existent, the co-text preceding or following the ambiguous chunk, the GP pivot, within the lead-up is never biased towards the nonsalient/less salient interpretation but is neutral (and thus perfectly compatible with) but may favour the salient one, causing the retention of the salient meaning at the integration stage. In other words, the lead-up co-text is usually supportive of the salient meaning, which would be salient even in a different verbal neighbourhood. It is worth noting that, in her discussion on jokes, Giora also acknowledges the “collaborative function of a supportive context” (Giora 2003: 168). The GP humorous effect emerges later and operates on cancelling this salient interpretation in the light of the switch, i.e. the final part of the co-text. As a result, the less salient or even nonsalient meaning is activated and retained at the integration stage. The graded salience hypothesis perfectly embraces salient meanings prompted by semantic and pragmatic phrasemes (collocations; idioms, including conventionalised metaphors; proverbs; sayings; and clichés), which serve as covert ambiguity realisations. A phraseme has one lexicalised and unquestionably more salient meaning to be countered by the switch, in result of which the less salient/nonsalient one emerges. I’m not a complete idiot. Several parts are missing. [salient meaning of the collocation “complete idiot”: “somebody very stupid”] (the speaker is not very stupid and cannot be duped ĺ the speaker is not downright stupid, but only stupid to a certain extent) The recipe is silly. It says to separate two eggs but it doesn’t say how far to separate them.

268

Chapter Six [salient meaning of “separate” collocated with “an egg”: “separate the yolk from the white”] (the speaker finds the recipe silly because it says to separate the yolk from the white of two eggs ĺ the speaker finds the recipe silly because it does not provide any information as to how far two eggs must be separated) Question authority. Ask me anything. [salient meaning of the saying “Question authority”: “avoid the fallacy of the argument from authority”] (avoid the fallacy of the argument from authority ĺ pose questions to the speaker, who is an authority) All my life, I always wanted to be somebody. Now I see that I should have been more specific. [salient meaning of the idiom “to be somebody”: “to be considered an important, worthy person”] (the speaker always wanted to be a worthy person ĺ the speaker always wanted to be an undetermined person but should have specified who exactly he/she wanted to become)

Also, Giora’s model infallibly accounts for each GP resting on a single-word lexeme which manifests one salient interpretation and which can hardly be considered polysemous, unless wordplay is creatively introduced with recourse to false etymology (e.g. “a headmistress” as “the main lover”, “a housekeeper” as “a woman who keeps a house after she gets divorced”), or which is characterised by strong salience imbalance of its two meanings (e.g. “a bird” as “a creature with feathers and wings”/“a girl”). Only the most salient meaning is initially activated at the stage of the lead-up. I’m a wonderful housekeeper. Every time I get divorced, I keep the house. [salient meaning of the lexeme “housekeeper”: “a woman who looks after the house”/no other salient meaning] (the speaker is excellent at cooking and maintaining her house ĺ the speaker keeps the house after each of her divorces) When a bird dirties my windscreen, I never ask her out again. [salient meaning of the lexeme “bird”: “a creature with feathers and wings”/less salient: “a girl”] (when a flying creature leaves its waste on the windscreen ĺ when a young woman dirties the speaker’s windscreen, the speaker does not ask her out again)

The notion of salience also perfectly explains the default, stereotypical understanding of a phenomenon, a person or a thing evoked by a non-

Default and Salient Meanings in GPs

269

punning but covertly ambiguous lexical item, which carries a latent unconventional interpretation countering the interpreter’s presuppositions. The graded salience hypothesis will then account for GPs centred on presuppositional lexeme-triggered ambiguity. I’ve only been drunk once. And that was from 1971 to 1990. [salient (presupposed) meaning: “intoxication” is a temporary state lasting a few hours] (the speaker has been under the influence of alcohol only once and, deterred, has never drunk since ĺ the speaker was under the permanent influence of alcohol for over nineteen years) I fired two warning shots into his head. [salient (presupposed) meaning: “a warning shot” is “a shot in the air”] (the speaker fired a warning shot in the air ĺ the speaker shot a man in the head as a warning]

In the case of equibiased ambiguities with two (equally) salient interpretations, whose perception may vary depending on the individual choice of a given interpreter at a given time, the most salient meaning will be strengthened by givenness established by priming context (preceding text and, sometimes, the form of the ambiguous word). This increases the probability, if not guarantees, that the same salient meaning will be instantly entertained by all linguistically competent hearers, who might otherwise prefer different meanings of lexemes, as governed their idiosyncratic knowledge and experience, or activate two meanings one after another. The alternative salient meaning will not be initially perceived, because the most salient one appears to be sufficient, i.e. contextually relevant at the stage of the lead-up/before the switch is appreciated. In other words, the activation of the second salient meaning will be postponed, until the final part of the verbalisation, i.e. the following co-text, demands it. This extrapolation may need to be verified in the future with further experimental research, while the empirical proof for this claim is simply the success of many instances of GP humour. On the other hand, it cannot be ruled out that laboratory experiments may indicate greater salience imbalance between two interpretations of (at least) some of the ambiguous lexemes carrying two salient meanings, which would explain why only one interpretation is activated, the force of a priming context being less significant. In either case, GPs are formed in a way which guarantees no equivocation of meaning at the lead-up stage. I miss my wife but my aim is improving.

270

Chapter Six [the primed and most salient meaning of “miss”: “feel unhappy or sorry at the absence or loss”/less salient: “fail to hit”; the most salient meaning is supported by the personal subject and the tense of the ambiguous verb] (the speaker wishes his wife were with him ĺ the speaker fails to hit/shoot his wife) Sport? I love tennis. I participate in a few rounds a week. I buy a few rounds in a pub and we watch the telly. [the primed and most salient reading of the polysemy “round”: “a stage of competition or fight”/less salient: “a sandwich”/“a set of events”/“a regular journey”/“a shot from a gun”/“a number of alcoholic drinks”] (the speaker loves tennis and plays it a few times a week ĺ the speaker drinks alcohol and watches tennis on TV with his friends) The police arrested the prostitute for soliciting. She asked her clients for advice on money laundering. [the primed and most salient reading of the homonym “solicit”: “offer sex for money”/less salient: “ask for”] (the prostitute was arrested for performing her job ĺ the prostitute wanted to get illegal information) If a customs officer asks if you have anything to declare, tell him the trip sucked. [the primed and most salient reading of the polysemy “declare”: “make a statement of property”/ less salient: “state”] (asked by a customs officer to show any purchased goods which are subject to customs duty ĺ asked by a customs officer if you wish to make an official statement, you should say that the trip was awful)

The default/salient meaning of an equibiased ambiguous item necessary for the success of a GP corresponds to givenness and saliency11 (Prince 1981), which, in a way, mirror Giora’s salience. The former parameter refers to shared knowledge as to what the speaker assumes the hearer to be aware of or to be capable of inferring (Prince 1981). The latter refers to what the hearer has in his/her consciousness upon hearing an utterance. Prototypically, save for equibiased ambiguities, the salient reading of an ambiguous lexeme should be automatic and succeed even without the contribution of Prince’s saliency or speakers’ conscious assessment of shared knowledge. A generalisation can be made that for the success of a GP, the salient meaning of an ambiguous element occurring in the lead-up will be activated first. Additionally, it is supported at the integration stage by the 11

Prince’s terms are terminological opposites of Giora’s.

Default and Salient Meanings in GPs

271

present neighbouring verbal context or, at least, there must be nothing in the co-text within the lead-up that will suggest otherwise. However, in cases where both or either of the salient meanings may theoretically surface, the preceding co-text should also facilitate the choice of the most salient meaning, which is eventually invalidated for a humorous effect. To reformulate, the context within the lead-up must initially be supportive of, or neutral to, the most salient interpretation, which is thus considered adequate and sufficient, precluding further pursuit of another meaning. The first inference is sustained until the switch is introduced in the final portion of the co-text, cancelling the salient interpretation and necessitating retrospective activation of the second nonsalient or less salient meaning. The aim is to spring a surprise on the recipient, who is oblivious to any ambiguity at the beginning, but has the other interpretation suddenly thrust upon himself/herself. Nota bene, Giora (2003) also mentions the pertinence of her findings to the analysis of garden-path sentences, i.e. non-humorous verbalisations entailing syntactic ambiguity. It must be appreciated that there is an extensive body of research on the interpretation of garden-path sentences within a psycholinguistic approach (cf. Van Gompel et al. 2006).

5. Conclusion Taken together, the disparate (but not mutually exclusive) pragmatic and cognitive models revisited above help explain what meanings enjoy a default/salient status and are almost effortlessly inferred but can be cancelled to make way for alternative hitherto unobserved interpretations, thus serving as fertile ground for GPs. This interdisciplinary and eclectic approach facilitates a detailed description of comprehension processes underlying the workings of garden-path humour. Notably, it affords an explanation as to why each form of ambiguity exploited in GPs remains covert, yielding a single “obvious” interpretation, which is legitimately suppressed in the light of the switch. Grice’s GCIs developed by Levinson and Giora’s salient meanings coincide with interpretations of various ambiguous forms which will be generated by default across contexts and will also be subject to cancellation, should contextual factors demand this. These are the two intrinsic characteristics of ambiguous (pivots of) GP lead-ups. Postulating the existence of presumptive meanings, Levinson focuses on pragmatic processes and discusses inferences which will be made by default across contexts. Although Levinson addresses the problem of particular language structures or lexemes (causing linguistically-based inferential ambiguities),

272

Chapter Six

it is extrapolated that his proposal covers also post-propositional inferences entailing purely inferential ambiguities. Contrary to Levinson, Giora focuses on cognitive processes of ambiguity resolution. This may be why the graded salience approach accounts for different language phenomena, which is fortunate, inasmuch as it completes the picture of GPs’ manifestations. Giora’s salient meanings coincide with the most obvious and the most easily available interpretations of all lexemes and phrasemes manifesting punning ambiguity as well as words which can be viewed as presuppositionally ambiguous. Such meanings are also impervious to contextual factors (with the exception of priming intralexical context). To recapitulate, although the two authors subscribe to different disciplines, and thus different methodological approaches, one can draw a parallel between their postulates. Both concentrate on the meanings generated effortlessly irrespective of context. Because their foci of interest and heuristic apparatuses differ, the two theories complement each other.

FINAL COMMENTS

In a nutshell, the book aimed to shed light on one distinct humorous mechanism, the garden-path mechanism, focusing on its manifestations in two categories of conversational humour, i.e. witticisms and one-liners. The notion of the garden-path mechanism was presented on the basis of existing literature, within both humour and grammar studies. GPs were distinguished from other humorous forms and divided into categories depending, primarily, on the nature of underpinning ambiguity (lexical and pragmatic) and also on subordinate linguistic phenomena. Additionally, GP workings were discussed in view of relevant writings on verbal humour, which were critically revisited. Therefore, readers will have had an opportunity to acquaint themselves with multifarious linguistic theories and postulates on forms and workings of humour in general. GP humour was argued to represent a particular realisation of the incongruityresolution model. Another chapter examined several frameworks of humorous texts’ on-line interpretation, verifying them as regards their applicability in the analysis of GPs and humour in general. Attempts have been made to explain the workings (but not the funniness) of GPs from pragmatic and cognitive perspectives. Notably, the objective was to account for why only one meaning of (covert) ambiguity is initially observed, while the other emerges at the stage of the switch. This entailed an overview and assessment of a number of theories on human communication, predominantly Grice’s, Levinson’s and Giora’s works, in the context of the humorous phenomenon. The garden-path mechanism underlies a considerable part of verbal humour, whether in the form of jokes, witticisms, interactional puns or responses to pragmatically ambiguous verbalisations. However, contrary to what some researchers suggest in their descriptions of humour mechanisms, it is by no means inherent to all verbal humour. On the other hand, the mechanism is more ubiquitous, as it can be found not only in humour but also in anxiety-provoking stimuli. For example, it is a recurrent motif in thrillers, whose plots entail some mystery to be solved (e.g. “The Sixth Sense”, “The Others”, “Fight Club”, “The Machinist”, “Righteous Kill”, etc.). In the final scenes, the audiences discover that they have been drawing wrong inferences or have taken for granted something that should not have been, letting themselves be led up the garden-path.

274

Final Comments

Not to spoil the readers’ pleasure of being deceived if they decide to watch any of these films, no details will be disclosed. Suffice it to say that the thrilling effect is achieved by skilful structuring of the plot and cinematographic tactics, which allow for some information to remain latent but to be inferred by default. The garden-path phenomenon, whether humorous or not, is also by no means absent from everyday interactions. On a more personal note, I would like to recount an anecdote. As I was finishing the book in autumn, I went to a green-grocer’s, where I heard a dialogue between an elderly lady and a shop assistant. This interchange helped me realise the prevalence, and hence significance, of the mechanism I had been analysing. The elderly customer asked (in Polish), “Are there any worms in these plums? Yesterday, I found a worm in every second fruit.” The shop assistant spluttered that she was extremely sorry and that this time the plums came from a different source and must be worm-free, to which the lady replied, “I won’t buy any, then. They must contain a lot of pesticides if no warms want to eat them.” And she left, leaving the shop assistant flabbergasted. All the queuing clients, including myself, burst into laughter. I wanted to explain the operation of the garden-path mechanism the lady had just employed, pointing out that her utterance was not meant to be humorous, but appeared funny (even if not to the shop assistant, who must have felt awkward), inasmuch as it subverted our default inferences, not causing any negative emotions. I bit my tongue, though. As someone wittily observed, “Analysing humour is like dissecting a frog. Hardly anybody is interested and the frog dies.”

APPENDIX

List of GP examples used. A bird in hand shits on your wrist. A woman came to ask the doctor if a woman should have children after 35. I said 35 children is enough for any woman. After a swilling session, I wake up and crave for an eye-opener, so I look into the mirror. All my life, I always wanted to be somebody. Now I see that I should have been more specific. As you know, I don’t repeat gossip so listen carefully. Attention, animal lovers! What you are doing is illegal! Back up my hard drive? How do I put it in reverse? Beauty comes from within…bottles, jars and tubes. Children would laugh at my son when he started wearing braces on his teeth. They couldn’t understand that otherwise his trousers would fall down. Could you pass me the recipe for your ratatouille? So I can tear it up. Don’t count your chickens will do it for you. Don’t drink and drive. You might hit a bump and spill your drink. Don’t spend your money on drugs. Become a pop star and get them free. Early to bed and early to rise indicates unskilled labour. Every morning I exercise. Up and down, up and down and then the other eyelid. Finally, I met this guy who said he loved children. Then I found out he was on parole for it. Gambling? I once won a lot. I bought my first Reubens with my poker winnings. Two Reuben sandwiches. Get a new car for your spouse. It’ll be a great trade. Giving up smoking is easy. I’ve done it a thousand times. Hard work never killed anybody, but why take the chance. Having drunk once, he resolved never to move his eyeballs again. He told me I was in shape. It was nice with the exception that round is a shape as well.

276

Appendix

He’s not fine. Last week he started hitting the bottle, thinking it was his wife. Honesty is the key to relationships. If you can’t fake it, you’re lost. How I managed? Well, there are remarkable people in this world, but I’m not one of them. I can afford a first-class plane ticket although I’m a teacher. I eat free dinners at the canteen, I walk to school, I’ve got a few shopping discounts and I’ve got a rich lover. I don’t dislike my work. I practically loathe it. I don’t drink any more but I don’t drink any less either. I don’t suffer from insanity. I enjoy every minute of it. I fired two warning shots into his head. I got a divorce because of stalemate. My husband got uninteresting. I have a drinking problem. I keep spilling tea. I know karate and a few other words of Japanese origin. I like kids but I don’t think I could eat a whole one. I like long walks if they are taken by people who annoy me. I love cats; they taste like chicken. I love children as long as they are clean, not hungry, quiet and shown on television. I love my work; it fascinates me. I can sit and look for hours. I miss my wife but my aim is improving. I resigned because of illness and fatigue. Everybody was sick and tired of me. I saw six men kicking and punching my mother-in-law. I wasn’t going to help. I thought six should be enough. I saw the play under adverse conditions; the curtain was up. I still have the body of an 18-year-old. It’s in my trunk. I still miss my ex-wife but my aim is improving. I used to think you’re a pain in the neck. Now I have a much lower opinion of you. I want to pay my income tax with a smile but they demand cash. I was a chair at yesterday’s meeting. I was inert and I said nothing. I was happy with my wife for 20 years. Then we met. I’m a one-man woman. One at a time. I’m a wonderful housekeeper. Every time I get divorced, I keep the house. I’m addicted to chocolate. I snort cocoa. I’m like Achilles, but he had only a heel and I have the whole body. I’m not a complete idiot. Several parts are missing. I’m not a heavy smoker. I smoke only 10 packets of lights a day. I’m on a seafood diet. I see food and I eat it.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

277

I’ve been going to the “Adonis” Gym for the past three months. Tomorrow I may even start working out. I’ve been in love with the same woman for forty years – if my wife finds out she’ll kill me. I’ve betrayed my husband only once during our whole marriage. We’ve been married for two days. I’ve had three husbands, apart from my own. I’ve never carried a torch for her! It’s the 21st century; electricity is everywhere! I’ve only been drunk once. And that was from 1971 to 1990. If a customs officer asks if you have anything to declare, tell him the trip sucked. If God had intended us to fly, he would have made it easier to get to the airport. If I turn into my parents, either I’ll be an alcoholic blonde chasing young boys or I’ll wind up like my mother. If I want to hear the pitter-patter of little feet, I certainly will. I will put shoes on my cat. If you don’t feel good having woken up after 10 a.m., you should try getting up at 5 a.m. I always do and I feel… awful. If you go abroad, you must steal …the views with your eyes. If you have missed it the first time, here’s a golden opportunity – you can miss it again. If you want to see a comic strip, you should see me in the shower. In an emergency, put your head between your knees and kiss your ass goodbye. In case of fire, do not use elevators, use water. It is not a book to be tossed lightly aside. It should be thrown with great force. It’s difficult to find single men who are sensitive, caring and good-looking because they already have boyfriends. It’s not hard to meet expenses. They’re everywhere. It’s not MP. It’s YP, your problem. It’s not that I always see a glass half empty. I always see a glass half full of poison. It’s not that some women look for rich men, who will guarantee them life stability. All look for such men. Life is not all beer and skittles. There are also girls. Lying is not easy. It is simplistic. Marriage is when a man and woman become as one; the trouble starts when they try to decide which one.

278

Appendix

Money can’t buy you everything. That’s what credit cards are for. My boss gave me an IOU…a note saying “I overpaid you.” My favourite machine in the gym is the vending machine. My French is getting better. Yesterday I ordered a meal in French in such a way that the waiter was astounded. I was in a Chinese restaurant. My motto is: I’m a light eater. I start eating as soon as it’s light. My wife is a customs officer. She keeps checking if I’m behaving myself in the pub. My wife made chicken soup. I was lucky that muck wasn’t for me. Never hit a man with glasses. Use your fist! Never keep up with the Joneses. It’s easier to drag them to your level. No, you can’t join me. I’m not coming apart! Nobody supports Clinton; he can still walk by himself. Oh, I’ve been enjoying a ravishingly passionate, hot, steamy and exhausting but satisfying relationship with my course in logical semiotics. Oysters enhance your sexual performance but they don’t work for me. Maybe I put them on too soon. Question authority. Ask me anything. Save the whales. Collect the whole set. Saving is worth your while. Save some money every month and at the end of the year you’ll be surprised how little you have. Sex on television can’t hurt you unless you fall off. She can play with her kid all day long. Problems start when she bounces him too hard. She has her looks from her father. He’s a plastic surgeon. Smoking makes me calm. I know I’ll die from it before something worse gets me. Sport? I love tennis. I participate in a few rounds a week. I buy a few rounds in a pub and we watch the telly. Take life with a pinch of salt, a slice of lemon and a bottle of tequila. Talk is cheap until you hire a lawyer. The best way to drag a genuine man out of a bath is to turn on the tap. The book I like the most is my husband’s cheque book. The dress doesn’t look expensive. It is expensive. The little girl is just like her mum. If she keeps growing up like this, things will be pretty bad. The police arrested the prostitute for soliciting. She asked her clients for advice on money laundering. The recipe is silly. It says to separate two eggs but it doesn’t say how far to separate them.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

279

The secret of our long marriage: we go to a nice restaurant and to the cinema and then we have sex. She goes Fridays, I go Saturdays. They charged John with a Duracell battery. They look really intelligent but actually they’re dumb. This dress looks better on… fire. TV is very educational. Every time it is on, I go to another room and read a book. We ought never to do wrong when people are looking. When a bird dirties my windscreen, I never ask her out again. When I was a kid my parents moved a lot, but I always found them. When my wife said she was seeing another man, I told her to scrub her eyes. When you meet Mr Right, you don’t know his first name is Always. Where there’s a will, I want to be in it. Yes, darling, I know you have an open mind. I can feel the draught from where I’m sitting. You should say no to drugs. That will drive the prices down. You’re not silly…. You’re simply remarkably stupid. You’ve got the brain of a four-year-old boy, and I’ll bet he was glad to get rid of it.

REFERENCES

Abercrombie, David. “Phatic communion.” In Concise Encyclopaedia of Pragmatics, edited by Jacob Mey, 672-673. Amsterdam et al.: Elsevier, 1998. Alexander, Richard. Aspects of Verbal Humour in English. Tübingen: Gunter Narr Verlag, 1997. Apresjan, Juri. “Regular polysemy.” Linguistics 142 (1973): 5-32. Apte, Mahadev. Humor And Laughter. An Anthropological Approach. Ithaca and London: Cornell University Press, 1985. Apter, Michael. The Experience of Motivation: The Theory of Psychological Reversals. London: Academic Press, 1982. Archakis, Argiris and Villy Tsakona. “Analyzing conversational data in GTVH terms: A new approach to the issue of identity construction via humor.” Humor 18 (2005): 41-68. Asher, Nicholas and Alex Lascarides. “Bridging.” Journal of Semantics 15 (1998): 83-113. —. Logics of Conversation. Cambridge: Cambridge University Press, 2003. Atlas, Jay David and Stephen Levinson. “It-clefts, informativeness, and logical form.” In Radical Pragmatics, edited by Peter Cole, 1-51. New York: Academic Press, 1981. Attardo, Salvatore. “The violation of Grice’s maxims in jokes.” In Proceedings of the 16th Annual Meeting of the Berkeley Linguistics Society, edited by Kira Hall, Michael Meacham and Richard Shapiro, 355-362. Berkeley, CA: University of California Press, 1990. —. “Violation in conversational maxims and cooperation: The case of jokes.” Journal of Pragmatics 19 (1993): 532-558. —. Linguistic Theories of Humor. New York: Mouton, 1994. —. “Humor.” In Handbook of Pragmatics, edited by Jef Verschueren JanOla Östman and Jan Blommaert, 1-18. Amsterdam: John Benjamins, 1996a. —. “Humor theory beyond Jokes: The treatment of humorous texts at large.” In Automatic Interpretation and Generation of Verbal Humor, edited by Joris Hulstijn and Anton Nijholt, 81-94. Enschede, Netherlands: University of Twente, 1996b.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

281

—. “The semantic foundations of cognitive theories of humor.” Humor 10 (1997a): 395-420. —. “Locutionary and perlocutionary cooperation: The perlocutionary cooperative principle.” Journal of Pragmatics 27 (1997b): 753-779. —. “Competition and cooperation: Beyond Gricean pragmatics.” Pragmatics and Cognition 5 (1997c): 21-50. —. “The analysis of humorous narratives.” Humor 11 (1998): 231-260. —. “Irony as relevant inappropriateness.” Journal of Pragmatics 32 (2000): 793-826. —. Humorous Texts: A Semantic and Pragmatic Analysis. New York, NY: Mouton de Gruyter, 2001. —. “Translation and humour: An approach based on the General Theory of Verbal Humour (GTVH).” Translator 8 (2002): 173-194. —. “On the nature of rationality in (neo-Gricean) pragmatics.” International Journal of Pragmatics 14 (2003): 3-20. —. “Cognitive linguistics and humor.” Humor 19 (2006): 341-362. —. “On the GTVH and SSTH: Comment’s on Raskin’s interview.” In New Approaches to the Linguistics of Humour, edited by Diana Popa and Salvatore Attardo, 226-229. Galati: Editura Academica, 2007. Attardo, Salvatore and Jean Charles Chabanne. “Jokes as a text type.” Humor 5 (1992): 165-176. Attardo, Salvatore, Christian Hempelmann and Sara Di Maio. “Script oppositions and logical mechanisms: modeling incongruities and their resolutions.” Humor 15 (2002): 3-46. Attardo, Salvatore, Donalee Hughes Attardo, Paul Baltes and Marnie Jo Petray. “The linear organization of jokes: Analysis of two thousand texts.” Humor 7 (1994): 27-54. Attardo, Salvatore and Victor Raskin. “Script theory revis(it)ed: Joke similarity and joke representation model.” Humor 4 (1991): 293-348. Attneave, Fred. “Multistability in perception.” In Scientific American: Recent Progress in Perception, edited by Richard Held and Whitman Richards, 143-151. San Francisco, CA: W.H. Freeman, 1971. Austin, John. How To Do Things With Words. Oxford: Clarendon Press, 1962. Aubouin, Elie. Les Genres du Risible. Marseille: C.N.R.S., 1948. Aymone, Alessandro. “Interview with Victor Raskin.” In New Approaches to the Linguistics of Humour, edited by Diana Popa and Salvatore Attardo, 217-225. Galati: Editura Academica, 2007. Bach, Kent. “Default Reasoning: Jumping to Conclusions and Knowing When to Think Twice.” Pacific Philosophical Quarterly 65 (1984): 3758.

282

References

—. “Conversational impliciture.” Mind and Language 9 (1994): 124-162. —. “Standardization vs. conventionalization.” Linguistics and Philosophy 18 (1995): 677-686. —. “Ambiguity.” In Routledge Encyclopaedia of Philosophy Online, 1998a. —. “Standardization revisited.” In Pragmatics: Critical Assessment, edited by Asa Kasher, 712-720. London: Routledge, 1998b. —. “The myth of conventional implicature.” Linguistics and Philosophy 22 (1999): 327-366. —. “You don’t say?” Synthese 127 (2001): 11-31. —. “Pragmatics and the philosophy of language.” In Handbook of Pragmatics, edited by Larry Horn and Gregory Ward, 463-487. Oxford: Blackwell, 2004. —. “The excluded middle: Semantic minimalism without minimal propositions.” Unpublished paper, 2006. Bach, Kent and Robert Harnish. Linguistic Communication and Speech Acts. Cambridge Mass.: MIT Press, 1979. Balota, David and James Chumbley. “Are lexical decisions a good measure of lexical access? The role of word frequency in the neglected decision stage.” Journal of Experimental Psychology: Human Perception and Performance 10 (1984): 340-357. Bartlett, Frederic. Remembering: An Experimental and Social Study. Cambridge: Cambridge University Press, 1932. Bateson, Gregory. “The position of humor in human communication.” In Cybernetics, edited by Heinz von Foerster, 1-47. New York: Josiah Macy, Jr. Foundation, 1953. —. Steps to an Ecology of Mind: Collected Essays in Anthropology, Psychiatry, Evolution, and Epistemology. Chicago: University Of Chicago Press, 1972. Beattie, James. “An essay on laughter and ludicrous composition.” In Essays. Edinburgh: William Creech, 1776. Reprinted by New York: Garland, 1971. Berlyne, Daniel. Conflict, Arousal and Curiosity. New York: McGrawHill, 1960. —. “Laughter, humor, and play.” In The Handbook of Social Psychology, Vol. 3, edited by Gardner Lindzey and Elliot Aronson, 795-852. Reading, MA: Addison-Wesley, 1969. —. Aesthetics and Psychobiology. New York: Appleton, 1971. —. “Humor and its kin.” In The Psychology of Humor: Theoretical Perspectives and Empirical Issues, edited by Jeffrey Goldstein and Paul McGhee, 43-60. New York: Academic Press, 1972.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

283

Bernsen, Niels Ole, Laila Dybkjær and Hans Dybkjær. “Cooperativity in human-machine and human-human spoken dialogue.” Discourse Processes 21 (1996): 213-236. Bever, Thomas. “The cognitive basis for linguistic structures.” In Cognition and the Development of Language, edited by John Hayes, 279-361. London: Wiley, 1970. Bezuidenhout, Anne and Robin Morris. “Implicature, relevance and default inferences.” In Experimental Pragmatics, edited by Ira Noveck and Dan Sperber, 257-282. Basingstoke: Palgrave Macmillan, 2004. Blank, Andreas. “Polysemy in the lexicon and in discourse.” In: Polysemy: Flexible Patterns of Meaning in Mind and Language, edited by Brigitte Nerlich, Zaxie Todd, Vimala Herman and David Clarke, 267293. Berlin: Mouton de Gruyter, 2003. Blank, Michelle and Donald Foss. “Semantic facilitation and lexical access during sentence processing.” Memory and Cognition 6 (1978): 644-652. Boring, Edwin. “A new ambiguous figure.” American Journal of Psychology 42 (1930): 444-445. Boxer, Diana and Florencia Cortés-Conde. “From bonding and biting: Conversational joking and identity display.” Journal of Pragmatics 27 (1997): 275-295. Breheny, Richard, Napoleon Katsos and John Williams. “Are generalized scalar implicatures generated by default? An on-line investigation into the role of context in generating pragmatic inferences.” Cognition 100 (2006): 434-436. Brewer, William and James Treyens. “Role of schemata in memory for places.” Cognitive Psychology 13 (1981): 207-30. Brône, Geert. “Hyper - and misunderstanding in interactional humour.” Journal of Pragmatics. forth. Brône, Geert and Kurt Feyaerts. “Assessing the SSTH and GTVH: A view from Cognitive Linguistics.” Humor 17 (2004): 361-372. Brône, Geert, Kurt Feyaerts and Tony Veale. “Introduction: Cognitive linguistic approaches to humor.” Humor 19 (2006): 203-228. Brône, Geert, Jeoren Vandaele, Kurt Feyaerts. “Multiple viewpoints in humour.” Paper presented at 10th ICLC, Kraków, Poland July 15-20, 2007. Brown, James. “Eight types of puns.” Publications of the Modern Language Association of America 71 (1956): 14-28. Brown, Penelope and Stephen Levinson. “Universals in language use: Politeness phenomena.” In Questions and Politeness, edited by Esther Goody, 56-311. Cambridge: Cambridge University Press, 1978.

284

References

Brown, Penelope and Stephen Levinson. Politeness: Some Universals in Language Use. Cambridge: Cambridge University Press, 1987. Bruner, Jerome and Leo Postman. “On the perception of incongruity. A paradigm.” Journal of Personality 18 (1949): 206-223. Brzozowska, Dorota. O Dowcipach Polskich i Angielskich. Aspekty Jezykowo-kulturowe. Opole: Uniwersytet Opolski, Opolskie Towarzystwo Przyjacióá Nauk. Studia i monografie nr 281, 2000. Bubel, Claudia. “Film audiences as overhearers.” Journal of Pragmatics 40 (2008): 55-71. Bubel, Claudia and Alice Spitz. “‘One of the last vestiges of gender bias’: The characterization of women through the telling of dirty jokes in Ally McBeal.” Humor 19 (2006): 71-104. Bucaria, Chiara. “Lexical and syntactic ambiguity as a source of humor: The case of newspaper headlines.” Humor 17 (2004): 279-309. Bugelski, Bergen and Delia Alampay. “The role of frequency in developing perceptual sets.” Canadian Journal of Psychology 15 (1961): 201-11. Burton-Roberts, Noel. The Limits to Debate: A Revised Theory of Semantic Presupposition. Cambridge: Cambridge University Press, 1989. Buttler, Danuta. Polski Dowcip JĊzykowy. Warszawa: Wydawnictwo Naukowe PWN, 1968. Carrell, Amy. “Joke competence and humor competence.” Humor 10 (1997): 173-85. Carston, Robyn. “Implicature, explicature, and truth-theoretic semantics.” In Mental Representations: The Interface Between Language and Reality, edited by Ruth Kempson, 155-181. Cambridge: Cambridge University Press, 1988. —. “Conjunction, explanation and relevance”. Lingua 90 (1993): 27-48. —. “Quantity maxims and generalised implicatures.” Lingua 96 (1995): 213-24 —. Thoughts and Utterances: The Pragmatics of Explicit Communication. Oxford: Blackwell, 2002. —. “Relevance theory and the saying/implicating distinction.” In Handbook of Pragmatics, edited by Larry Horn and Gregory Ward, 633-656. Oxford: Blackwell, 2004. Chafe, Wallace. “Givenness, contrastiveness, definiteness, subjects, topics and point of view.” In Subject and Topic, edited by Charles Li, 25-56. New York: Academic Press, 1976. Channel, Joanna. Vague Language. Oxford: Oxford University Press, 1994.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

285

Chapman, Siobhan. Paul Grice: Philosopher and Linguist. New York, Houndsmill Basingstoke: Palgrave Macmillan, 2005. Chapman, Siobhan and Christopher Routledge. Key Thinkers in Linguistics and the Philosophy of Language. Edinburgh: Edinburgh University Press, 2005. Chiaro, Delia. The Language of Jokes: Analysing Verbal Play. London: Routledge, 1992. Chierchia, Gennaro and Sally McConnell-Ginet. Meaning and Grammar: An Introduction to Semantics 2nd edition. Cambridge, MA: The MIT Press, 2000. Coates, Jennifer. “Talk in a play frame: More on laughter and intimacy.” Journal of Pragmatics 39 (2007): 29-49. Conrad, Carol. “Context effects in sentence comprehension: A study of the participative lexicon.” Memory and Cognition 2 (1974): 130-138. Coulson, Seana. “The Menendez brothers virus: Analogical mapping in blended spaces.” In Conceptual Structure, Discourse, and Language, edited by Adele Goldberg, 67-81. Palo Alto: CSLI, 1996. —. Semantic Leaps. Frame-Shifting and Conceptual Blending in Meaning Construction. Cambridge: Cambridge University Press, 2001. —. “What’s so funny? Cognitive semantics and jokes.” Cognitive Psychopathology/Psicopatologia Cognitive 2 (2005a): 67-78. —. “Extemporaneous blending: Conceptual integration in humorous discourse from talk radio.” Style 39 (2005b): 107-122. Coulson, Seana and Marta Kutas. “Frame-shifting and sentential integration.” UCSD Cognitive Science Technical Report Postscript Version 98.03. La Jolla, CA: UCSD, 1998. —. “Getting it: Human event-related brain response to jokes in good and poor comprehenders.” Neuroscience Letters 316 (2001): 71-74. Coulson, Seana, Thomas Urbach and Marta Kutas. “Looking back: Joke comprehension and the space structuring model.” Humor 19 (2006): 229-250. Crystal, David. Language Play. London: Penguin, 1998. Curcó, Carmen. “Some observations on the pragmatics of humorous interpretations. A relevance-theoretic approach.” UCL Working Papers in Linguistics 7 (1995): 27-47. —. “The implicit expression of attitudes, mutual manifestness and verbal humour.” UCL Working Papers in Linguistics 8 (1996): 89-99. Dascal, Marcelo. “Language use in jokes and dreams: sociopragmatics vs. psychopragmatics.” Language and Communication 5 (1985): 95-106.

286

References

Davies, Bethan. “Grice’s Cooperative Principle: Getting the meaning across.” Leeds Working Papers in Linguistics and Phonetics 8 (2000): 1-26. —. “Grice’s Cooperative Principle: Meaning and rationality.” Journal of Pragmatics 39 (2007): 2308-2331. Davis, Wayne. Implicature: Intention, Convention, and Principle in the Failure of Gricean Theory. Cambridge: Cambridge University Press, 1998. Deckers, Lambert. “On the validity of a weight-judging paradigm for the study of humor.” Humor 6 (1993): 43-56. Deckers, Lambert, Steven Jenkins and Eric Gladfelter. “Incongruity versus tension relief: Hypotheses of humor.” Motivation and Emotion 1 (1977): 261-272. De Palma, Paul and Judith Weiner. “Riddles: Accessibility, parallelism and knowledge representation.” In Proceedings of CoLing-92: The Fourteenth International Conference on Computational Linguistics, Nantes, France, 1121-1125, 1992. Deane, Paul. “Polysemy and cognition.” Lingua 75 (1988): 325-361. Delabastita, Dirk. “Introduction.” In Traductio. Essays on Punning and Translation, edited by Dirk Delabastita, 1-22. Manchester: St. Jerome Publishing and P. U. de Namur, 1997. Dews, Shelly and Ellen Winner. “Muting the meaning: A social function of irony.” Metaphor and Symbolic Activity.” 10 (1995): 3-19. Dolitsky, Marlene. “Aspects of the unsaid in humour.” Humor 5 (1992): 33-43. Dopkins, Stephen, Robin Morris and Keith Rayner. “Lexical ambiguity and eye fixation in reading: A test of competing models of lexical ambiguity.” Journal of Memory and Language 31 (1992): 461-477. Drew, Paul. “Poe-faced receipts of teases.” Linguistics 25 (1987): 219253. Duffy, Susan, Robin Morris and Keith Rayner. “Lexical ambiguity and fixation times in reading.” Journal of Memory and Language 27 (1988): 429-446. Durkin, Kevin and Jocelyn Manning. “Polysemy and the subjective lexicon: Semantic relatedness and the salience of intraword senses.” Journal of Psycholinguistic Research 18 (1989): 577-612. Dynel, Marta. No aggression, only teasing: The pragmatics of teasing and banter. Lodz Papers in Pragmatics (2008): 4.2. (http://versita.com/science/socialsciences/lpp/) —. “How do puns bear relevance.” In proceedings of Interpreting for Relevance, forth.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

287

Dziemidok, Bohdan. The Comical. A Philosophical Analysis. Translated by Marek Janiak. Dordrecht, Boston, London: Kluwer Academic Publishers, 1993. Elio, Rene. Common Sense, Reasoning and Rationality. Oxford: Oxford University Press, 2002. Elster, Jon. Rational Choice. New York: New York University Press, 1986. Esar, Evan. The Humor of Humor. New York: Bramhall House, 1954. Fauconnier, Gilles. Mental Spaces. Cambridge: Cambridge University Press, 1994. Fauconnier, Gilles and Mark Turner. “Conceptual projection and middle spaces”. Cognitive Science Report 9401, UCSD, 1994. —. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books, 2002. Fill, Alvin. “Vision and revisions: A type of pun based on idiomatised complex words.” In Language and civilization: A Concerted Profusion on Essays and Studies in Honour of Otto Heitsch, Vol. 2, edited by Claudia Blank, Teresa Kirschner, Donald Gutch and Judith Gilbert, 551-563. Frunkfurt-on-Main: Peter Lang, 1992. Fischler, Ira. “Associative facilitation without expectancy in a lexical decision task.” Journal of Experimental Psychology: Human Perception and Performance 3 (1977): 18-26. Fish, Stanley. Doing What Comes Naturally: Change, Rhetoric, and the Practice of Theory in Literary and Legal Studies. Durham, NC: Duke UP, 1989. Fodor, Jerry. The Modularity of Mind. Cambridge: MIT Press, 1983. Fodor, Janet Dean and Atsu Inoue. “The diagnosis and cure of garden paths.” Journal of Psycholinguistic Research 23 (1994): 407-434. Forabosco, Giovannantonio. “Cognitive aspects of the humour process: the concept of incongruity.” Humor 5 (1992): 9-26. —. “Is the concept of incongruity still a useful construct for the advancement of humor research?” Lodz Papers in Pragmatics 4 (2008): 45-62. Forster, Kenneth “Levels of processing and the structure of the language processor.” In Sentence Processing: Psycholinguistic Studies Presented to Merrill Garrett, edited by William E. Cooper and Edward Walker, 27-65. Hillsdale, NJ: Lawrence Erlbaum Associates, 1979. —. “Priming and the effects of sentence and lexical contexts on naming time: Evidence for autonomous lexical processing.” Quarterly Journal of Experimental Psychology 33 (1981): 465-495.

288

References

Forster, Kenneth and Susan Chambers. “Lexical access and naming time.” Journal of Verbal Learning and Verbal Behavior, 12 (1973): 627-635. Franz, Alexander. Automatic ambiguity resolution in Natural Language Processing: An Empirical Approach. Berlin: Springer-Verlag, 1996. Frazier, Lyn. On Comprehending Sentences: Syntactic Parsing Strategies. PhD dissertation, University of Connecticut, 1978. Frazier, Lyn and Keith Rayner. “Making and correcting errors during sentence comprehension: Eye movement in the analysis of structurally ambiguous sentences.” Cognitive psychology 14 (1982): 178-210. Frege, Gottlob. “On sense and meaning.” In Frege: Collected Works, edited by Brian McGuinness, 157-177. Oxford: Basil Blackwell, 1984. Originally published in Uber Sinn und Bedeutung. Zeitschrift fur Philosophie und philosophische Kritik, Gottlob Frege, 1892. Freud, Sigmund. Jokes and Their Relation to the Unconscious. Translated by James Stranchey. New York: Norton, 1960. Originally published as Der Witz und seine Beziehung zum Unbewussten, 1905. Fry, William. Sweet Madness: A Study of Humor. Palo Alto: Pacific Books, 1963. Gazdar, Gerald. Pragmatics: Implicature, Presupposition, and Logical Form. New York: Academic Press, 1979. Gernsbacher, Morton. Language Comprehension As Structure Building. Hillsdale N.J.: Lawrence Erlbaum Associates, 1990. Geurts, Bart. “Scalars.” In Lexicalische Semantik Aus Kognitiver Sicht, edited by Petra Ludewig and Bart Geurts, 95-117. Tübingen: Narr Verlag. 1998. Gibbs, Raymond. “Comprehending figurative referential descriptions.” Journal of Experimental Psychology: Learning, Memory and Cognition 16 (1990): 56-66. Gibbs, Raymond and Nandini Nayak. “Psycholinguistic studies on the syntactic behavior of idioms.” Cognitive Psychology 21 (1989): 100138. Gibbs, Raymond, Nandini Nayak and Cooper Cutting. “How to kick the bucket and not decompose: Analyzability and idiom processing.” Journal of Memory and Language 28 (1989): 576-593. Gibson, Edward and Neal Pearlmutter. “A corpus-based analysis of psycholinguistic constraints on prepositional-phrase attachment.” In Perspectives on Sentence Processing, edited by Charles Clifton, Lyn Frazier, and Keith Rayner, 181-198. Hillsdale, NJ: Lawrence Erlbaum, 1994. Gillon, Brendan. “Ambiguity, generality, and indeterminacy: Tests and definitions.” Synthèse 85 (1990): 391-416.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

289

Ginsberg, Matthew. Reasoning in Nonmonotonic Reasoning. San Mateo, CA: Morgan Kaufmann, 1987. Giora, Rachel. “On the Informativeness Requirement.” Journal of Pragmatics 12 (1988): 547-565. —. “On the cognitive aspects of the joke.” Journal of Pragmatics 16 (1991): 465-485. —. “Understanding figurative and literal language: The graded salience hypothesis.” Cognitive Linguistics 7 (1997): 183-206. —. “On the priority of salient meanings: Studies of literal and figurative language.” Journal of Pragmatics 31 (1999): 919-929. —. On Our Mind: Salience, Context, and Figurative Language. New York: Oxford University Press, 2003. Giora, Rachel, Orna Fein and Tamir Schwartz. “Irony: Graded salience and indirect negation.” Metaphor and Symbol 13 (1998): 83-101. Giora, Rachel and Ofer Fein. “On understanding familiar and less-familiar figurative language.” Journal of Pragmatics 31 (1999): 1601-1618. Glucksberg, Sam, Roger Kreuz and Susan Rho. “Context can constrain lexical access: Implications for models of language comprehension.” Journal of Experimental Psychology, Learning, Memory, and Cognition 12 (1986): 323-335. Goffman, Erving. Frame Analysis. New York: Harper and Row, 1974. —. Forms of Talk. Pennsylvania: University of Pennsylvania Press, 1981. Goldstein, Jeffrey, Jerry Suls and Susan Anthony. “Enjoyment if specific types of humor content: Motivation or salience?” In The Psychology of Humor: Theoretical Perspectives and Empirical Issues, edited by Jeffrey Goldstein and Paul McGhee 159-171. New York: Academic Press, 1972. Gorfein, David (ed). Resolving Semantic Ambiguity. New York: Springer Verlag, 1989. Grandy, Richard and Richard Warner, “Paul Grice. A review of his work.” In Philosophical Grounds of Rationality, edited by Richard Grandy and Richard Warner, 1-44. Oxford: Claredon Press, 1986. Green, Thomas and William Pepicello. “Wit in riddling: A linguistic perspective.” Genre 11 (1978): 1-13. Greimas, Algirdas. Structural Semantics. Translated by Daniele McDowell, Lincoln: University of Nebraska Press, 1983. Originally published as Semantique Structurale. Paris: Larousse, 1966. Grice, Herbert Paul. “Meaning.” In Studies in the Way of Words, Paul Grice, 213-233. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in Philosophical Review 64 (1957): 377-88.

290

References

—. “The causal theory of perception, part I.” In Studies in the Way of Words, Paul Grice, 224-247. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in Proceedings of the Aristotelian Society 121 (1961): 121-152 —. “Utterer’s meaning, sentence meaning, and word-meaning.” In Studies in the Way of Words, Paul Grice, 117-137. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in Foundations of Language 4 (1968): 225-42. —. “Utterer’s meaning and intentions.” In Studies in the Way of Words, Paul Grice, 86-116. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in The Philosophical Review 78 (1969): 147-77. —. “Logic and conversation.” In Studies in the Way of Words, Paul Grice, 22-40. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in Syntax and Semantics, Vol.3: Speech acts, edited by Peter Cole and Henry Morgan, 41-58. New York: Academic Press, 1975; and in The Logic of Grammar, edited by Donald Davison and G. Harman. 64-75. Encino: Dickenson, 1975. —. “Further notes on logic and conversation.” In Studies in the Way of Words, Paul Grice, 41-57. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in Syntax and Semantics, Vol. 9, edited by Peter Cole, 113-127. New York: Academic Press, 1978. —. “Presupposition and conversational implicature.” In Studies in the Way of Words, Paul Grice, 269-282. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in Radical Pragmatics, edited by Peter Cole, 183-197. New York: Academic Press, 1981. —. “Meaning revisited.” In Studies in the Way of Words, Paul Grice, 283303. Cambridge, Mass.: Harvard University Press, 1989b. Originally published in Mutual Knowledge, edited by Neil Smith, 223-43. New York: Academic Press, 1982. —. “Retrospective epilogue.” In Studies in the Way of Words, Paul Grice, 339-386. Cambridge, Mass.: Harvard University Press, 1989a. —. Studies in the Way of Words. Cambridge, Mass.: Harvard University Press, 1989b. Grimes, Wilma. “A theory of humor for public address: the mirth experience.” Speech Monographs 22 (1955): 217-226. Grosjean, Francois. “Spoken word recognition processes and the gating paradigm.” Perception and Psychophysics 28 (1980): 267-283. Gruner, Charles. Understanding Laughter: The Workings of Wit and Humour. Chicago: Nelson-Hall, 1978.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

291

Gumperz, John. Discourse Strategies. Cambridge: Cambridge University Press, 1982. Halliday, Michael and Ruqaiya Hasan. Cohesion in English. London: Longman, 1976. Hammond, Paul and Patrick Hughes. Upon the Pun: Dual Meanings in Words and Pictures. London: W.H.Allen, 1978. Haugh, Michael. “The intuitive basis of implicature: relevance theoretic implicitness versus Gricean implying.” Pragmatics 12 (2002): 117134. Hausmann, Franz. Studien zu einer Linguistik des Wortspiels: Das Wortspiel im Canard Enchaîné. Tübingen: Max Niemeyer Verlag, 1974. Hawley, Patrick. “What is said.” Journal of Pragmatics 34 (2002): 969991. Hay, Jennifer. “Functions of humour in the conversations of men and women.” Journal of Pragmatics 32 (2000): 709-742. Heller, Louis. “Toward a general typology of the pun.” Language and Style 7 (1974): 271-282. Hempelmann, Christian. “Script opposition and logical mechanism in punning.” Humor 17 (2004): 381-392. Hempelmann, Christian and Andrea Samson. “Visual Puns and verbal puns: Descriptive analogy or false analogy.” In New Approaches to the Linguistics of Humour, edited by Diana Popa and Salvatore Attardo, 180-196. Galati : Editura Academica, 2007. Higgins, Tory and Gillian King. “Accessibility of social constructs: Information processing consequences of individual and contextual variability.” In Personality, Cognition and Social Interaction, edited by Nancy Cantor and John Kihlstrom, 161-197. Hillsdale, NJ: Erlbaum, 1981. Hirschberg, Julia. A Theory of Scalar Implicature. New York: Garland, 1991. Hirst, Graeme. Semantic Interpretation and the Resolution of Ambiguity. Cambridge, UK: Cambridge University Press, 1987. Hobbes, Thomas. Leviathan. London: Crooke, 1651. Hockett, Charles. “Grammar for the hearer.” In Structure of Language and its Mathematical Aspects, edited by Roman Jakobson 220-36. Providence: American Mathematical Society, 1961. —. “Jokes.” In The View From Language: Selected Essays, Charles Hockett, 257-289. Athens: University of Georgia Press, 1977. Originally published in Studies In Linguistics In Honor Of George L. Trager, edited by Estellie Smith, 153-178. The Hague: Mouton, 1972.

292

References

Horn, Larry. “Toward a new taxonomy for pragmatic inference: Q-based and R-based implicature.” In Georgetown University Round Table on Languages and Linguistics, edited by Deborah Schriffin, 11-42. Washington, D.C.: Georgetown University Press, 1984. —. A Natural History of Negation. Chicago: University of Chicago Press, 1989. —. “Implicature.” In The Handbook of Pragmatics, edited by Larry Horn and Gregory Ward, 3-28. Oxford: Blackwell, 2004. —. “Current issues in neo-Gricean pragmatics.” Intercultural Pragmatics 2 (2005): 191-204. Howard, Ian. “An investigation of a satiation process in the reversible perspective of revolving skeletal shapes.” Quarterly Journal of Experimental Psychology 13 (1961): 19-33. Hunter, Lynne. “On misapplying the maxims: a Gricean look at wit.” Chicago Linguistic Society 19 (1983): 195-204. Hutcheson, Francis. Reflections on Laughter and Remarks upon the Fable of the Bees. Glasgow: Baxter, 1750. Jahn, Manfred. “’Speak, friend, and enter’: Garden paths, Artificial Intelligence, and cognitive narratology.” In Narratologies: New Perspectives on Narrative Analysis, edited by David Herman, 167-194. Ohio: Ohio State UP,1999. Jaszczolt, Katarzyna. Discourse, Beliefs, and Intentions: Semantic Defaults and Propositional Attitude Ascription. Oxford: Elsevier Science, 1999. —. “Prolegomena to default semantics.” Research in Language 2. Special Issue on Developments in Pragmatics (2004): 7-32. —. Default Semantics: Foundations of a Compositional Theory of Acts of Communication. Oxford: Oxford University Press, 2005. —. “Meaning merger: Pragmatic inference, defaults, and compositionality.” Intercultural Pragmatics 3 (2006a): 195-212. —. “Defaults in semantics and pragmatics.” In Stanford Encyclopaedia of Philosophy, edited by Edward Zalta, 2006b, http://plato.stanford.edu/contents.html (accessed May 10, 2007). Johnson, Ragnar. “Two realms and a joke: Bisociation theories of joking.” Semiotica 16 (1976): 195-221. Jones, James. Cognitive Factors in the Appreciation of Humor: A Theoretical and Experimental Analysis. Unpublished doctoral thesis, Yale University, 1970. Just, Marcel and Patricia Carpenter. “A theory of reading: From eye fixations to comprehension.” Psychological Review 87 (1980): 329354.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

293

Kamp, Hans. “A theory of truth and semantic representation.” In Formal Methods in the Study of Language, edited by Jeroen Groenendijk, Theo Janssen and Martin Stokhof, 277-322. Amsterdam: Mathematical Centre Tracts 135, 1981. Kamp, Hans and Uwe Reyle. From Discourse to Logic: Introduction to Modeltheoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory. Dordrecht, Holland: Kluwer Academic, 1993. Kant, Immanuel. Critique Of Judgment. Translated by John Bernard. New York: Hafner Publishing Company, 1951. Originally published as Kritik der Urteilkraft. Berlin: Lagrade, 1790. Karttunen, Lauri. “Presupposition and linguistic context.” Theoretical Linguistics 1 (1974): 3-44. Karttunen, Lauri and Stanley Peters. “Conventional implicature.” In Syntax and Semantics 11: Presupposition, edited by Oh Choon-Kyu and David Dinnen. New York: Academic Press, 1979. Kasher, Asa. “Conversational maxims and rationality.” In Language in Focus, edited by Kasher Asa, 197-216. Dordrecht: D. Reidel, 1976. Katz, Bruce. “A neural resolution of the incongruity-resolution and incongruity theories of humour.” Connection Science: Journal of Neural Computing, Artificial Intelligence and Cognitive Research 5 (1993): 59-75. Kecskes, Istvan, “Dueling contexts: A dynamic model of meaning.” Journal of Pragmatics 40 (2008): 385-406. Keenan, Edward. “Two kinds of presupposition in natural language.” In Studies in Linguistic Semantics, edited by Charles Fillmore and D. Terence Langendoen, 45-54. New York: Holt, Reinhart and Winston, 1971. Keenan, (Ochs) Elinor. “The universality of conversational postulates.” Language in Society 5 (1976): 67-80. Keith-Spiegel, Patricia. “Early conceptions of humour: Varieties and issues.” In The Psychology of Humour, edited by Jerry Goldstein and Paul McGhee, 3-39. New York: Academic Press, 1972. Kempson, Ruth. Presupposition and the Delimitation of Semantics. Cambridge: Cambridge University Press, 1975. Keysar, Boaz, Dale Barr, Jennifer Balin and Timothy Paek. “Definite reference and mutual knowledge: Process models of common ground in comprehension.” Journal of Memory and Language 39 (1998): 1-20. Kittay, Eva. Metaphor: It’s Cognitive Force and Linguistic Structure. Oxford: Clarendon, 1987. Koestler Arthur. The Act of Creation. London: Hutchinson, 1964.

294

References

Kotthoff, Helga. “Impoliteness and conversational humor.” Folia Linguistica 30 (1996): 299-327. —. “Coherent keying in conversational humour: Contextualising joint fictionalisation.” In Coherence in Spoken and Written Discourse, edited by Wolfram Bublitz, Uta Lenk and Eija Ventola, 125-150. Amsterdam: Benjamins, 1999. —. “Pragmatics of performance and the analysis of conversational humor.” Humor 19 (2006): 271-304. —. “Oral genres of humor. On the dialectic of genre knowledge and creative authoring.” Pragmatics 12 (2007): 263-296. Kulhman, Thomas. “A study of salience on motivational theories of humor.” Journal of Personality and Social Psychology 49 (1985): 281286. Lackner, James and Merrill Garrett. “Resolving ambiguity: Effects of biasing context in the unattended ear.” Cognition 1 (1972): 359-372. Lakoff, Robin. “Conversational logic.” In Handbook of Pragmatics, edited by Jef Verschueren, Jan-Ola Östman and Jan Blommaert, 190-198. Amsterdam: John Benjamins, 1995. Leeper, Robert. “A study of a neglected portion of the field of learning the development of sensory organization.” Journal of Genetic Psychology 46 (1935): 41-75. Lampert, Martin and Susan Ervin-Tripp. “Risky laughter: Teasing and self-directed joking among male and female friends.” Journal of Pragmatics 38 (2006): 51-72. Langacker, Ronald. Foundations of Cognitive Grammar: Theoretical Prerequisites. Stanford, CA: Stanford University Press, 1987. Latta, Robert. The Basic Humor Process : A Cognitive-Shift Theory and the Case Against Incongruity. Berlin: Walter de Gruyter, 1999. Laver, John. “Communicative functions of phatic communion.” In Organisation of Behaviour in face-to-face interaction, edited by Adam Kendon, Richard Harris and Marie Key. The Hague: Mouton, 1975. Leeper, Robert. “A study of a neglected portion of the field of learning: The development of sensory organization.” Journal of Genetic Psychology 46 (1935): 41-75. Levelt, Willem. Speaking. From Intention to Articulation. Cambridge, MA: MIT Press, 1989. Levesque, Hector. Knowledge Representation. Cambridge, MA: MIT Press, 1992. Levinson, Stephen. Pragmatics. Cambridge: Cambridge University Press, 1983.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

295

—. “Pragmatic reduction of the binding conditions revisited.” Journal of Linguistics 27 (1991): 107-161. —. “Three levels of meaning.” In Grammar and Meaning. Essays in Honour of Sir John Lyons, edited by Frank Palmer, 90-115. Cambridge: Cambridge University Press, 1995. —. “Minimization and conversational inference.” In Pragmatics 4, edited by Asa Kasher, 545-614. London: Routledge, 1998. —. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, Mass.: MIT Press, 2000. Lipps, Theodor. Komik and Humor. Hamburg: L.Voss, 1898. Long, Debra and Arthur Graesser. “Wit and humor in discourse processing.” Discourse Processes 11 (1998): 35-60. Lucas, Margery. “Frequency effects on the processing of ambiguous words in sentence contexts.” Language and Speech (1987) 30: 25-46. Lyons, John. Semantics, Vol. II. Cambridge: Cambridge University Press, 1977. —. Language, Meaning and Context. London: Fontana, 1981. MacDonald, Maryellen, Neal Pearlmutter and Mark Seidenberg. “Lexical nature of syntactic ambiguity resolution.” Psychological Review 101 (1994): 676-703. Maier, Norman. “A Gestalt theory of humour.” British Journal of Psychology 23 (1932): 69-74. Marcus, Mitch. A Theory of Syntactic Recognition for Natural Language. Cambridge, Mass.: MIT Press, 1980. Marslen-Wilson, William and Lorraine Tyler. “The temporal structure of spoken language understanding.” Cognition 8 (1980): 1-71. Martin, Charles, Hoang Vu, George Kellas, and Kimberly Metcalf. “Strength of discourse context as a determinant of the subordinate bias effect.” The Quarterly Journal of Experimental Psychology 52 (1999): 813-839. Martin, Rod. The Psychology of Humor. An Integrative Approach. Burlington, MA: Elsevier, 2007. McClelland, James “The case for interactionism in language processing.” In Attention and Performance Xll. The Psychology of Reading, edited by Max Coltheart, 3-36. London: Erlbaum, 1987. McGhee, Paul. “The role of operational thinking in children’s comprehension and appreciation of humor.” Child Development 42 (1971): 733-734. —. Humor, Its Origin and Development. San Francisco: Freeman, 1979.

296

References

McGhee, Paul, Willibald Ruch and Franz-Josef Hehl. “A personalitybased model of humor development during adulthood.” Humor 3 (1990): 119-146. McRae, Ken, Michael Spivey-Knowlton and Michael Tanenhaus. “Modeling the influence of thematic fit (and other constraints) in on line sentence comprehension.” Journal of Memory and Language 38 (1998): 283-312. Mel’cuk, Igor. “Phrasemes in language and phraseology in linguistics.” In Idioms: Structural and Psychological Perspectives, edited by Martin Everaert, Erik-Jan van der Linden, André Schenk, and Rob Schreuder, Hillsdale, 167-232. NJ: Lawrence Erlbaum Associates, 1995. —. “Collocations and lexical functions.” In Phraseology: Theory, Analysis, and Applications, edited by Anthony P. Cowie, 23-53. Oxford: Oxford University Press, 1998. Mey, Jacob. “Pragmatic gardens and their magic.” Poetics 20 (1991): 233245. —. Pragmatics: An Introduction. Oxford: Blackwell, 1993. Meyer, David and Roger Schvaneveldt. “Facilitation in recognizing pairs of words: Evidence of a dependence between retrieval operations.” Journal of Experimental Psychology 90 (1971): 227-234. Meyer, David, Roger Schvaneveldt, and Margaret Ruddy. “Loci of contextual effects on visual word-recognition.” In Attention and Performance V, edited by Patrick Rabbitt and Stanislav Dornic, 98118. New York: Academic Press, 1975. Mieder, Wolfgang and Anna Tóthné Litovkina. Twisted Wisdom. Modern Anti-Proverbs. Burlington, Vermont: The University of Vermont, 1999. —. Old Proverbs Never Die, They Just Diversify: A Collection of AntiProverbs. Veszprém: University of Veszprém Press, 2006. Milne, Robert. “Predicting garden path sentences.” Cognitive Science: Artificial Intelligence, Psychology and Language 6 (1982): 349-373. Milner, George. “Homo ridens: Towards a semiotic theory of humour and laughter.” Semiotica 6 (1972): 1-30. Minsky, Marvin. “A framework for representing knowledge.” In The Psychology of Computer Vision, edited by Patrick Winston, 211-277. New York: McGraw-Hill, 1975. —. “Computer science and the representation of knowledge.” In The Computer Age: A Twenty-Year View, edited by Michael Dertouzos and Joel Moses, 392-421. Cambridge, MA: MIT Press, 1979. —. Jokes and their Relation to the Cognitive Unconscious. Al Memo 603, MIT-AI Lab., November, 1980.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

297

Monro, David. Argument Of Laughter. Melbourne: Melbourne University Press, 1951. Mooney, Annabelle. “Co-operation, violation and making sense.” Journal of Pragmatics 36 (2004): 899-920. Morin, Violette. “L’historie drole.” Communications 8 (1966): 102-119. Morreall, John. Taking Laughter Seriously. Albany: State University of New York Press, 1983. —. (ed.) The Philosophy of Laughter and Humor. Albany: State University of New York Press, 1987. —. “Enjoying incongruity.” Humor 2 (1989): 1-18. —. “Verbal humor without switching scripts and without non-bona fide communication.” Humor 17 (2004): 393-400. Morton, John and John Long. “Effect of word transitional probability on phoneme identification.” Journal of Verbal Learning and Verbal Behavior 15 (1976): 43-51. Mulkay, Michel. On Humour: Its Nature And Its Place In Society. Cambridge: Polity Press, 1988. Murch, Gerld. Visual and Auditory Perception. Indianapolis, IN: BobbsMerrill, 1973. Nash, Walter. The Language of Humor: Style and Technique in Comic Discourse. London: Longman, 1985. Neale, Stephen. “Paul Grice and the philosophy of language.” Linguistics and Philosophy 15 (1992): 509-59. Neely, James. “Semantic priming and retrieval from lexical memory.” Journal of Experimental Psychology: General 196 (1977): 226-254. Nerhardt, Göran. “Humor and inclinations of humor: Emotional reactions to stimuli of different divergence from a range of expectancy.” Scandinavian Journal of Psychology 11 (1970): 185-195. —. “Incongruity and funniness: Towards an new descriptive model.” In Humour and Laughter: Theory, Research and Applications, edited by Anthony Chapman and Hugh Foot, 55-62. London: Transaction Publishers, 1976. Nerlich Brigitte and David Clarke. “Ambiguities we live by: Towards a pragmatics of polysemy.” Journal of Pragmatics 33 (2001): 1-20. Neubert, Albrecht and Gregory Shreve. Translation as Text. Kent, Ohio/London: The Kent State University Press, 1992. Nilsen, Don and Alleen Pace Nilsen. Language Play: An Introduction to Linguistics. Rowley, MA: Newbury House, 1978. Norrick, Neal. “Stock conversational witticisms.” Journal of Pragmatics 8 (1984): 195-209.

298

References

—. “A frame-theoretical analysis of verbal humor: Bisociation as schema conflict.” Semiotica 60 (1986): 225-45. —. “From wit to comedy: Bisociation and intertextuality.” Semiotica 61 (1987): 113-125. —. “Intertextuality in humor.” Humor 2 (1989): 117-139. —. Conversational Joking: Humor In Everyday Talk. Bloomington: Indiana University Press, 1993. —. “Involvement and joking in conversation.” Journal of Pragmatics 22 (1994): 409-430. —. “On the conversational performance of narrative jokes: toward an account of timing.” Humor 14 (2001): 255-274. —. “Issues in conversational joking.” Journal of Pragmatics 35 (2003): 1333-1359. —. “Non-verbal humor and joke performance.” Humor 17 (2004): 401409. Nunberg, Geoffrey, Ivan Sag and Thomas Wasow. “Idioms.” Language 70 (1994): 491-538. Oaks, Dallin. “Creating structural ambiguities in humor: Getting English grammar to cooperate.” Humor 7 (1994): 377-402. Olbrechts-Tyteca, Lucie. Le Comique du Discours. Bruxelles, Belgium: Editions de l’Universite de Bruxelles, 1974. Onifer, William and David Swinney. “Accessing lexical ambiguities during sentence comprehension: Effects of frequency of meaning and contextual bias.” Memory And Cognition 9 (1981): 225-236. Orbach, Joseph, Dan, Ehrlich, and John Vainstein. “Reversibility of the Necker Cube: III. Effects of interpolation on reversal rate of the cube presented repetitively.” Perceptual and Motor Skills 17 (1963): 459464. Oring, Elliott. Jokes And Their Relations. Lexington: The University Press of Kentucky, 1992. —. Review of The Basic Humor Process: A Cognitive Shift Theory And a Case Against Incongruity by Robert Latta. Berlin: Mouton de Gruyter, 1999. Humor 12 (1999): 457-464. —. Engaging Humor. Urbana: University of Illinois, 2003. Palmer, Steven. “The effects of contextual scenes on the identification of objects.” Memory and Cognition 3 (1975): 519-526. Partington, Alan. Patterns and Meanings. Amsterdam and Philadelphia: Benjamins, 1998. —. The Linguistics of Laughter. A Corpus-assisted Study of LaughterTalk. Oxon: Routledge Studies in Linguistics, 2006.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

299

Paul, Stephen, George Kellas, Michael Martin, and Matthew Clark. “Influence of contextual features on the activation of ambiguous word meanings” Journal of Experimental Psychology 18 (1992): 703-717. Paulos, John. Mathematics and Humour. Chicago: Chicago University Press, 1980. Peleg, Orna and Zohar Eviatar. “Hemispheric sensitivities to lexical and contextual information : Evidence from lexical ambiguity resolution.” Brain and Language 105 (2008): 71-82. Peleg, Orna, Rachel Giora and Ofer Fein. “Salience and context effects: Two are better than one.” Metaphor and Symbol 16 (2001): 173-192. —. “Contextual strength: The whens and hows of context effects.” In Experimental Pragmatics, edited by Ira Noveck and Deirdre Sperber, 172-186. Basingstoke and New York: Palgrave Macmillan, 2004. —. “Resisting contextual information: You can’t put a salient meaning down.” Lodz Papers in Pragmatics 4.1.(2008):13-44. Pepicello, William, and Robert Weisberg. “Linguistics and humor.” Handbook of Humor Research, Vol. 1, edited by Paul McGhee and Jeffrey Goldstein, 59-84. NY: Springer-Verlag, 1983. Pepicello, William and Thomas Green. The Language of Riddles. Ohio: Ohio State University, 1984. Perlmutter, Daniel. “Incongruities and logical inconsistencies in humor: The delicate balance.” Humor 15 (2002): 155-168. Pien, Diana and Mary Rothbart. “Incongruity and resolution in children’s humor: A reexamination.” Child Development 47 (1976): 966-971. Pinkal, Manfred. Logic and Lexicon. London: Oxford, 1995. Poesio, Massimo. “Semantic ambiguity and perceived ambiguity.” Semantic Ambiguity and Underspecification, edited by Kees van Deemter and Stanley Peters, 159-198. Stanford: CSLI Publications, 1996. Prince, Ellen. “Towards a Taxonomy of Given - New Information.” In Radical Pragmatics, edited by Peter Cole, 223-255. New York: Academic, 1981. Pritchett, Bradley. Grammatical Competence and Parsing Performance. Chicago: The University of Chicago Press, 1992. Raskin, Victor. Semantic Mechanisms of Humor. Dordrecht: D. Reidel Publishing Company, 1985. —. “Linguistic heuristics of humor: A script-based semantic approach.” International Journal of the Sociology of Language 65 (1987): 11-25. —. “The sense of humor and the truth.” In The Sense of Humor: Explorations of a Personality Characteristics, edited Willibald Ruch, 95-108. Berlin: Mouton de Gruyter, 1998.

300

References

—. “Book review. Engaging Humour, Elliot Oring.” Journal of Pragmatics 39 (2007): 221-225. Raskin, Victor, and Salvatore Attardo. “Non-literalness and non-bona-fide in language: An approach to formal and computational treatments of humor.” Pragmatics and Cognition 2 (1994): 31-69. Rayner, Keith, Jeremy Pacht and Susan Duffy. “Effects of prior encounter and global discourse bias on the processing of lexically ambiguous words: Evidence from eye fixations.” Journal of Memory and Language 33 (1994): 527-544. Rayner, Keith, Katherine Binder and Susan Duffy. “Contextual strength and the subordinate bias effect: Comment on Martin, Vu, Kellas, and Metcalf.” Quarterly Journal of Experimental Psychology 52 (1999): 841-852. Recanati, François. “The pragmatics of what is said.” Mind and Language 4 (1989): 295-329. —. Direct Reference: From Language to Thought. Oxford: Blackwell, 1993. —. “What is said.” Synthese 128 (2001): 75-91. —. “Does linguistic communication rest on inference?” Mind and Language 17 (2002): 105-126. —. Literal Meaning. Cambridge: Cambridge University Press, 2004. Ritchie, Graeme. “Developing the incongruity-resolution theory.” In Proceedings of AISB Symposium on Creative Language: Stories and Humour, 78-85. Edinburgh, April 1999. http://www.csd.abdn.ac.uk/~gritchie/papers/aisb99.pdf (accessed May 10, 2007). —. “The structure of forced reinterpretation jokes.” In Proceedings of April Fools’ Day Workshop on Computational Humour, edited by Oliviero Stock, Carlo Strapparava and Anton Nijholt, 47-56. University of Twente, Enschede, Netherlands, April 2002. —. The Linguistic Analysis of Jokes. London: Routledge, 2004. —. “Reinterpretation and viewpoints.” Humor 19 (2006): 251-270. Rosch, Eleonor. “On the internal structure of semantic and perceptual categories.” In Cognitive Development and the Acquisition of Language, edited by Timothy Moore, 111-144. Academic Press: New York, 1973. Rosch, Eleonor. “Cognitive reference points.” Cognitive Psychology 7 (1975): 532-47. Rothbart, Mary. “Incongruity, problem-solving and laughter.” In Humour and Laughter: Theory, Research and Applications, edited by Anthony Chapman and Hugh Foot, 37-54. London: Wiley, 1976.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

301

Rothbart, Mary and Diana Pien. “Elephants and marshmallows: a theoretical synthesis of incongruity-resolution and arousal theories.” In It’s a Funny Thing, Humor, edited by Anthony Chapman and Hugh Foot, 37-40. New York: Pergamon, 1977. Ruch, Willibald. “Assessment of appreciation of humor: studies with the 3WD humor test.” In Advances in Personality Assessment, Vol. 9, edited by Charles Spielberger and James Butcher, 27-75. Hillsdale, NJ: Erlbaum, 1992. —. (ed). The Sense of Humor: Explorations of a Personality Characteristic. Berlin/New York : Mouton de Gruyter, 1998 Ruch, Willibald, Salvatore Attardo and Victor Raskin. “Toward an empirical verification of the General Theory of Verbal Humor.” Humor 6 (1993): 123-136. Ruch, Willibald and Franz-Joseph Hehl. “A two-mode model of humor appreciation: Its relation to aesthetic appreciation and simplicitycomplexity of personality.” In The Sense of Humor: Explorations of a Personality Characteristic, edited by Willibald Ruch, 109-142. Berlin/New York : Mouton de Gruyter, 1998. Ruhl, Charles. On Monosemy: A Study in Linguistic Semantics. Albany: State University of New York Press, 1989. Rumelhart, David and Andrew Ortony. “The representation of knowledge in memory.” In Schooling and the Acquisition of Knowledge, edited by Richard Anderson and Rand Spiro, 99-135. Hillsdale, NJ: Erlbaum, 1977. Sadock, Jerrold. “On testing for conversational implicature.” In Syntax and Semantics 9: Pragmatics, edited by Peter Cole, 281-297. New York: Academic Press, 1978. (Reprinted in Pragmatics: A Reader, edited by Steven Davis, 365-376. Oxford: Oxford University Press, 1991.) Sanford, Anthony and Simon Garrod. Understanding Written Language. Chichester: Wiley, 1981. Schank, Roger and Robert Abelson. Scripts, Plans, Goals and Understanding. Hillsdale, NJ: Erlbaum, 1977. Schiller, Paul. “A configurational theory of puzzles and jokes.” Journal of Genetic Psychology 18 (1938): 217-234. Schopenhauer, Arthur. Die Welt als Wille und Vorstellung, Leipzig: Brockhaus, 1819. Searle, John. Speech Acts. Cambridge: Cambridge University Press, 1969. Seidenberg, Mark, Michael Tanenhaus, James Leiman and Marie Bienkowski. “Automatic access of the meaning of ambiguous words in context: Some limitations of knowledge based processing.” Cognitive Psychology 14 (1982): 489-537.

302

References

Sherzer, Joel. “Oh! That’s A Pun And I Didn’t Mean It.” Semiotica 22 (1978): 335-50. —. “Puns and jokes.” In Handbook of Discourse Analysis, Vol. 3, edited by Teun van Dijk, 213-221. London: Academic, 1985. Shultz, Thomas. “The role of incongruity and resolution in children’s appreciation of cartoon humor.” Journal of Experimental Child Psychology 13 (1972): 456-477. —. “Development of the appreciation of riddles.” Child Development 45 (1974): 100-105. —. “A cognitive-development analysis of humour.” In Humour and Laughter: Theory, Research and Applications, edited by Anthony Chapman and Hugh Foot, 11-36. London: Wiley, 1976. Shultz, Thomas and Judith Robillard. “The development of linguistic humour in children: Incongruity through rule violation.” In Children’s Humour, edited by Paul McGhee and Anthony Chapman, 55-96. New York: Wiley, 1980. Simons, Mandy. “Presupposition and accommodation: Understanding the Stalnakerian picture.” Philosophical Studies 112 (2003): 251-278. —. “Foundational issues in presupposition.” Philosophy Compass 1 (2006): 357-372. Simpson, Greg. “Meaning dominance and semantic context in the processing of lexical ambiguity.” Journal of Verbal Learning and Verbal Behavior 20 (1981): 120-136. —. “Context and the processing of ambiguous words.” In Handbook of Psycholinguistics, edited by Morton Ann Gernsbacher, 359-374. San Diego, CA: Academic Press, 1994. Simpson, Greg and Merilee Krueger. “Selective access of homograph meanings in sentence context.” Journal of Memory and Language 30 (1991): 627-643. Small, Steven, Garrison W. Cottrell and Michael Tanenhaus. Lexical Ambiguity Resolution: Perspectives from Psycholinguistics, Neuropsychology, and Artificial Intelligence. San Mateo, CA: Morgan Kaufmann, 1988. Spencer, Herbert. “The physiology of laughter.” Macmillan’s Magazine 1 (1860): 395-402. Sperber, Dan and Deirdre Wilson. Relevance: Communication and Cognition. Oxford: Blackwell, 1986/1995. Stageberg, Norman. An Introductory English Grammar (Second Edition). New York: Holt, Rinehart and Winston, Inc., 1971. Staley, Rosemary and Peter Derks. “Structural incongruity and humor appreciation.” Humor 8 (1995): 97-134.

Humorous Garden-Paths: A Pragmatic-Cognitive Study

303

Stalnaker, Robert. “Pragmatics.” In Semantics of Natural Language, edited by Donald Davidson and Gilbert Harman, 380-97. Dordrecht: Reidel, 1972. —. “Presuppositions.” Journal of Philosophical Logic 2 (1973): 447-457. —. “Pragmatic presuppositions.” In Semantics and Philosophy, edited by Milton Munitzand and Peter Unger, 197-214. New York: New York University Press, 1974. Stanford, William. Ambiguity in Greek Literature: Studies in Theory and Practice. New York: Johnson, 1972. Strawson, Peter. Introduction to Logical Theory. London: Methuen, 1952. Suls, Jerry. “A two-stage model for the appreciation of jokes and cartoons: An information processing analysis.” In The Psychology of Humor, edited by Jeffrey Goldstein and Paul McGhee, 81-100. New York: Academic Press, 1972. —. “Cognitive processes in humor appreciation.” In Handbook of Humor Research, Vol.1, edited by Paul McGhee and Jeffrey Goldstein, 39-57. New York: Springer Verlag, 1983. Swaab, Tamara, Colin Brown and Peter Hagoort. “Understanding ambiguous words in sentence context: Electrophysiological evidence for delayed contextual se1ection in Broca’s aphasia.” Neuropsychologia 36 (1998): 737-761. —. “Understanding words in sentence contests: The time course of ambiguity resolution.” Brain and Language 86 (2003): 326-343. Sweetser, Eve. From Etymology to Pragmatics. Cambridge: Cambridge University Press, 1990. Swinney, David. “Lexical access during sentence comprehension (re)consideration of context effects.” Journal of Verbal Learning and Verbal Behavior 18 (1979): 645-659. Swinney, David and David Hakes. “Effects of prior context upon lexical access during sentence comprehension.” Journal of Verbal Learning and Verbal Behaviour 15 (1976): 681-689. Swinney, David, William Onifer, Penny Prather and Max Hirskowitz. “Semantic facilitation across sensory modalities in the processing of individual words and sentences.” Memory and Cognition 7 (1979): 159-165. Tabossi, Patrizia. “Accessing lexical ambiguity in different types of sentential contexts.” Journal of Memory and Language 27 (1988): 324340. Tabossi, Patrizia, Colombo Lucia and Remo Job. “Accessing lexical ambiguity: Effects of context and dominance.” Journal of Memory and Language 32 (1987): 359-372.

304

References

Taft, Marcus. “Lexical access via an orthographic code: The Basic Orthographic Syllabic Structure (BOSS).” Journal of Verbal Learning and Verbal Behavior 18 (1979): 21-39. Tannen, Deborah and Cynthia Wallat. “Interactive frames and knowledge schemas in interaction: Examples from a medical examination interview.” In Framing in Discourse, edited by Deborah Tannen, 5776. New York: Oxford University Press, 1993. Taylor, John. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press, 1989. Taylor, Talbot and Deborah Cameron. Analysing Conversation. Oxford: Pergamon Press, 1987. Thom Rene. Structural Stability and Morphogenesis. Translated by David Fowler. Reading, Mass.: Benjamin, 1975. Thomas, Jenny. Meaning in Interaction: An Introduction to Pragmatics. Harlow: Pearson Education, 1995. Todorov, Tzvetan. Theories of the Symbol. Ithaca, NY: Cornell Press, 1982. Toury, Gideon. In Search of a Theory of Translation. Tel Aviv: The Porter Institute for Poetics and Semiotics, 1980. Trueswell, John and Michael Tanenhaus. “Toward a lexicalist framework for constraint-based syntactic ambiguity resolution.” In Perspectives in Sentence Processing, edited by Charles Clifton, Lyn Frazier and Keith Rayner, 155-179. New Jersey, Hillsdale: Erlbaum, 1994. Tweedy, James, Robert Lapinski and Roger Schvaneveldt. “Semanticcontext effects on word recognition: Influence of varying the proportion of items presented in appropriate context.” Memory and Cognition 5 (1977): 84-89. Tyler, Lorraine and Jeanine Wessels. “Quantifying contextual contributions to word recognition processes.” Perception and Psychophysics 34 (1983): 409-420. —. “Is gating an on-line task? Evidence from naming latency data.” Perception and Psychophysics 38 (1985): 217-222. Vaid, Jyotsna, Rachel Hulla, Roberto Heredia, David Gerkens and Francisco Martinez. “Getting a joke: The time course of meaning activation in verbal humor.” Journal of Pragmatics 35 (2003): 14311449. Van de Voort, Marlies and Wietske Vonk. “You don’t die immediately when you kick an empty bucket: A processing view on semantic and syntactic characteristics of idioms.” In Idioms. Structural and Psychological Perspectives, edited by Martin Everaert, Erik-Jan Van

Humorous Garden-Paths: A Pragmatic-Cognitive Study

305

der Linden and André Schenk, 238-299. Hillsdale: Lawrence Erlbaum Associates, 1995. Van Dijk, Teun A. and Walter Kintsch. Strategies of Discourse Comprehension. New York: Harcourt Brace, 1983. Van Gompel, Roger, Martin Pickering, Jamie Pearson, and Jacob Gunnar. “The activation of inappropriate analyses in garden-path sentences: Evidence from structural priming.” Journal of Memory and Language 55 (2006): 335-362. Van Mulken, Margot, Renske van Enschot-van Dijk, and Hans Hoeken. Puns, relevance and appreciation in advertisements. Journal of Pragmatics 37 (2005): 707-721. Veale, Tony. “Incongruity in humour: Root cause or epiphenomenon.” Humor 17 (2004): 419-428. Veale, Tony, Kurt Feyaerts and Geert Brône. “The cognitive mechanisms of adversarial humor.” Humor 19 (2006): 305-340. Veisbergs, Andrejs. “The contextual use of idioms, wordplay and translation.” In Traductio. Essays on Punning and Translation, edited by Dirk Delabastita, 155-176. Manchester: St. Jerome Publishing and P. U. de Namur, 1997. Von Fintel, Kai. “Would you believe it? The King of France is Back! Presuppositions and Truth-Value Intuitions.” In Descriptions and Beyond, edited by Anne Bezuidenhout and Marga Reimer, 315-341. Oxford: Oxford University Press, 2001. Vu, Hoang and Stephen T. Paul. “Sources of sentence constraint in lexical ambiguity resolution.” Memory and Cognition 26 (1998): 979-1001. Vu, Hoang, George Kellas, Kimberly Metcalf and Ruth Herman. “The influence of global discourse on lexical ambiguity resolution.” Memory and Cognition 28 (2000): 236-252. Wardhaugh, Ronald. An Introduction to Linguistics. 2nd edition. Oxford: Blackwell, 1992. Weiner, Judith and Paul De Palma. “Some pragmatic features of lexical ambiguity and simple riddles.” Language and Communication 13 (1993): 183-93. Wilson, Christopher. Jokes: Form, Content, Use, And Function. London: Academic Press, 1979. Winograd, Terry. Language as a Cognitive Process, Vol. 1: Syntax. Reading, MA: Addison-Wesley, 1983. Wyer Robert and James Collins. “A theory of humor elicitation.” Psychological Review 99(1992): 663-688.

306

References

Yamaguchi, Haruhiko. “How to pull strings with words: Deceptive violations in the garden-path joke.” Journal of Pragmatics 12 (1988): 323-337. Yus, Francisco. “Humor and the search for relevance.” Journal of Pragmatics 35 (2003): 1295-1331. Zajdman, Anat. “Contextualisation of canned jokes in discourse.” Humor 4 (1991): 23-40. Zajonc, Robert. “Attitudinal effects of mere exposure.” Journal of Personality and Social Psychology 9 (1968): 1-27. Zhang, Qiao. “Fuzziness – vagueness – generality – ambiguity.” Journal Of Pragmatics 29 (1998): 13-31. Ziv, Avner. Personality and Sense of Humour. New York: Springer Publishing Company, 1984. Zillman, Dolf. “Disparagement humor.” In Handbook of Humor Research, Vol.1, edited by Paul McGhee and Jeffrey Goldstein, 85-107. New York: Springer Verlag, 1983. Zwicky, Arnold, and Elizabeth Zwicky. “Imperfect puns, markedness and phonological similarity. With fronds like these, who needs anemones.” Folia Linguistica 20 (1986): 493-503.

Sources of GPs Published volumes: Arnott, Stephen and Mike Haskins. Man Walks into a Bar. The Ultimate Collection of Jokes and One-liners. London: Random House, 2004. Jarski, Rosemarie. The Funniest Thing You Never Said. The Ultimate Collection of Humorous Quotations. London: Random House, 2004. Internet pages: http://www.dacton.com/humor2.html http://www.funnyjunksite.com http://www.angelfire.com/ok/ssplasher http://www.oneliners-and-proverbs.com http://www.jokesandhumor.com http://www.workinghumor.com Film scripts; quotes available at http://www. imdb.com The author’s personal communication (both written and spoken) with her English-speaking friends and students.

INDEX

ambiguity, ambiguous meanings covert 1, 3, 7, 20-22, 25, 33, 3739, 61, 67, 70, 72, 74, 9093, 96, 100, 101, 105, 106, 111, 112, 116, 124, 133, 210, 215, 218, 222, 231, 243, 267, 269, 271, 273 equibiased 252, 264-267, 269, 270 Gricean 159, 180, 186, 193-196, 200, 207-213, 215, 216, 218 lexical, punning 19, 22, 24, 30, 41, 70, 86, 89, 91, 95, 97, 99, 101, 105, 106, 111, 112, 115-117, 124, 125, 127, 128, 130-143, 153, 216, 219-222, 242, 243, 246, 247, 250-273 linguistic 2, 70, 75, 84, 113, 115, 117, 123, 126, 128, 131, 143, 154, 174, 213 linguistically-based inferential 128, 146-150, 218, 247, 248, 272 phonetic, phonological 70, 122, 126, 127 pragmatic 2, 23, 24, 34, 36, 75, 91, 101, 105, 112, 113, 115117, 122, 124, 128, 143153, 213, 218, 219, 222, 226, 227, 243, 247, 273 presuppositional 105, 112, 128, 150-153, 222, 248, 252, 267, 269, 272 purely inferential 128, 145-146, 219, 249, 272 resolution of 18, 109, 123, 246, 272

(semantic) ambiguity 2, 3, 7, 18, 23, 24, 26, 27, 29, 32, 42, 47, 51, 52, 56, 58, 68, 70, 73, 74, 77, 89-91, 93, 97101, 108, 109, 111, 116118, 123-129, 131, 176, 177, 222, 247 syntactic 2, 18-19, 23, 91, 99, 120, 124, 125, 127-131, 271 bisociation 48-51, 55-57, 93-94 cancellability, cancellable meaning 2, 123, 152, 154, 156, 190, 196, 206-208, 211, 214, 216, 217, 219, 220, 222, 224-227, 229, 231, 234-236, 240, 242, 246, 247, 250, 251 centrality 201, 204, 217, 219, 228, 244 cliché 66, 139, 142, 235, 248, 259, 267 collocation 139, 140, 141, 148, 149, 203, 216, 246, 259, 265, 267 conventional implicature (CI) 152, 197, 198, 201-203, 206-208, 213, 215-217, 233, 234 conversational humour 2, 8, 9, 1113, 15-17, 25, 33, 52, 82, 94, 96, 157, 173, 175, 179, 181-184, 187, 193, 195, 273 Cooperative Principle, CP 3, 155175, 178-198, 202, 203, 206, 207, 211, 214, 234 crossroads mechanism 25, 27, 2933, 35, 72, 76, 92, 93, 100, 102, 107, 112 co-text 2, 7, 21, 38-40, 111, 112, 143, 149, 154, 156, 160, 196, 209, 216-218, 222, 225, 231, 235, 240, 246, 247, 252, 253,

308 256, 259, 262, 263, 265, 266, 267, 269, 271 default meaning, defaultness 1-3, 7, 8, 18, 22, 24, 29-31, 33-35, 38, 40, 47, 55, 66, 68, 74, 78, 81, 101-105, 112, 115, 116, 123, 139, 142, 144, 145-147, 149, 150, 153, 154, 167, 172, 181, 214, 215, 217, 218, 221-228, 230-237, 239-242, 244-252, 256, 268, 270, 271, 274 defeasibility, defeasible meanings, see cancellability expect, expectation (lack or violation) 21, 24-26, 31, 33, 4346, 48, 49, 53, 54, 61, 63-68, 71, 72, 75, 79, 94, 95, 99, 100, 103, 104, 106, 112, 114, 135, 142, 147, 149, 194, 218 flout, exploitation of a maxim 156163, 166, 167, 170, 172, 174176, 178-180, 184-188, 191196, 198, 199, 201, 203, 204, 207, 208, 210, 211, 213, 214, 218, 221, 237, 238 garden-path(s), GP(s) 1-4, 7, 14, 15, 23, 24, 26, 27, 33, 34, 37-41, 51, 56, 63, 65, 66, 68, 73, 84, 91, 99, 100, 101, 105, 106, 112, 115-117, 123, 124, 127-156, 171, 176, 178, 186, 196, 202204, 207, 208, 210, 213, 215, 217-221, 222-227, 230-232, 235, 237, 240, 242-244, 246252, 258-260, 263-273 garden-path (GP) humour 1, 5, 16, 18, 20, 23, 27, 33, 40, 61, 76, 91, 106, 114, 123, 124, 130, 135, 136, 141, 152, 154, 155, 157, 158, 176, 178, 186, 190, 194, 196, 207, 208, 210, 211, 213, 216, 225, 235, 266, 269, 271, 273, 274 garden-path (GP) joke 21-23, 29, 33, 35, 57, 73, 91, 93, 100, 112, 194

Index garden-path (GP) mechanism 1-4, 8, 20-25, 29-31, 33-40, 51, 55, 67, 72, 73, 85, 86, 92, 93, 96, 98, 99, 105, 106, 111, 113, 115, 116, 129, 143, 156, 174, 195, 196, 221, 231, 241, 259, 273, 274 garden-path (GP) pun 23, 131-133, 134-143 (described) garden-path sentence 18-20, 23, 107, 129, 130, 271 generalised conversational implicature (GCI) 2, 3, 145, 153, 156, 196, 198, 200, 202208, 211, 213-221 , 223-224, 226, 228, 230-239, 241, 245247, 250, 251, 271 givenness 55, 252, 259-262, 266, 269, 270 homonymy, homonymous meaning 19, 75, 120, 124, 125, 127-131, 133, 134, 136-139, 142, 148, 208, 215, 219, 220, 250, 255, 259, 270 humorous keying/frame/mode 15, 16, 17, 31, 157, 183, 174, 194 idiom 18, 72, 106, 126, 129, 135, 139-141, 199, 203, 216, 217, 220, 235, 248, 258, 259, 265, 267, 268 incongruity, incongruous meanings 10, 24, 25, 27-33, 40-66, 68-74, 79, 83, 85-87, 92, 94, 97, 100, 101, 105, 106, 108, 110, 111, 114, 116, 124, 130, 132, 136, 145, 216, 218 incongruity resolution, incongruityresolution model (I-R) 1, 2, 25, 40-42, 51-62, 65, 68-76, 85-88, 95, 97, 116, 273 interactional pun 95, 96, 107, 189, 193, 273 irony 12, 14, 118, 180, 186, 199, 258 joke 2, 4-12, 14, 15, 17, 20-22, 24, 25, 28, 29, 31-37, 40, 47, 48,

Humorous Garden-Paths: A Pragmatic-Cognitive Study 51-54, 56, 57, 59-116, 118, 119, 121, 122, 124, 126, 127, 142, 144, 154, 156, 157, 169, 173, 174, 176, 177, 179-192, 194196, 226, 259, 267, 273 lead-up 2, 3, 24, 27, 33, 37-40, 47, 51, 56, 61, 66-68, 71, 73, 93, 99, 113, 116, 117, 124, 130-133, 136, 140, 144-146, 149, 150, 210, 216, 218, 220-222, 225, 231, 250, 252, 256, 259, 260, 265-271 lexeme 13, 31, 95, 112, 121,127, 128, 134-136, 138, 139, 146, 148, 152, 193, 202, 213, 215217, 220, 230, 231, 235, 239, 243, 246, 250-253, 258, 259, 265, 267-270, 272 linguistic humour, linguistic jokes 8, 13, 84, 94, 97, 99, 101, 106, 117-123, 124, 127, 128 literalisation (of phrasemes) 106, 129, 130, 135, 139-143, 242 (non-)bona-fide mode 3, 77, 156, 157, 173, 178-185, 188, 190192 one-liner 1, 2, 4-6, 9, 11, 13, 15, 21, 23-25, 32, 33, 37, 47, 54, 56, 60, 61, 63, 66, 73, 76, 94, 95, 96, 98, 99, 101, 102, 105, 107, 108, 11, 113, 115, 116, 129, 174, 178, 181, 195, 262, 273 on-line interpretation , incremental, linear interpretation 1, 2, 17-18, 30, 50, 65, 68-71, 76, 81, 88, 89, 91, 93-97, 100-102, 107-109, 123, 124, 131-133, 146, 154, 187, 195, 210, 218, 235, 236, 240, 246, 247, 250, 256, 263, 273 particularised conversational implicature (PCI) 198, 203-208, 211-213, 216, 217, 219, 228, 230, 233, 234 phraseme 38, 106, 128, 131, 135, 139-143, 147, 203, 210, 216,

309

217, 220, 222, 227, 231, 235, 242, 243, 247, 248, 251, 258, 259, 265, 267, 272 polysemy, polysemous meaning, 30, 37, 55, 124-131, 133-137, 139143, 148, 184, 208, 213-215, 219, 220, 244, 247, 250, 251, 255-257, 259, 268, 270 presumptive meaning, Levinson’s GCIs 3, 153, 230-251, 271 presupposition 16, 28, 96, 103, 105, 112, 114, 150-153, 182, 202, 218, 222, 244, 248, 249, 251, 282, 266, 267, 269, 272 priming 20, 37, 237, 245, 252, 253, 255, 259-264, 266, 269, 270, 272 proverb 66, 117, 129, 130, 139, 142, 143, 203, 259, 265, 267 pun, punning 2, 3, 11, 12, 21, 22, 27, 31, 36, 47, 56, 57, 72, 76, 79, 94, 107, 108, 111, 115, 117, 119, 120, 122-124, 126-139, 146, 147, 153, 154, 176, 186, 194, 211, 216, 221, 243, 266, 269, 272, 273 punchline 10, 11, 21, 22, 24, 28-35, 37, 47, 54, 56, 58, 60, 62-75, 79, 83-85, 90-95, 97, 98, 100, 101, 108-115, 186, 196 putdown 14, 26, 35, 166, 193, 194, 201 rationality 157, 162, 164,-168, 170172, 175, 178, 187, 189, 190, 192, 196, 221, 237 red-light mechanism 24-30, 32, 66, 72, 76, 92, 93, 100, 106, 112, 142 salient meaning, salience 1-3, 7, 17, 17, 24, 27, 29, 33, 38, 40, 47, 55, 61-63, 68, 74, 78, 81, 102, 108-112, 116, 123, 124, 134, 136, 137, 139-141, 143, 150, 153, 154, 213, 221, 222, 245, 247, 251-272

310 saying 66, 82, 129, 130, 139, 142, 143, 203, 216, 220, 248, 259, 265, 267, 260 self-deprecating humour, selfdenigrating humour 14-15, 26, 76 surprise 16, 25, 27, 29, 43-45, 54, 59, 61-66, 71, 72, 95, 97, 99, 105, 110, 112, 135, 140, 142, 195, 241, 259, 264, 271 switch 23-26, 30, 32, 33, 38-40, 47, 51, 56, 63, 66, 68, 73, 97, 113, 116, 124, 130-133, 136-140, 142, 145-149, 153, 154, 186, 216, 218, 222, 225, 235, 240, 247, 252, 259, 260, 265, 267, 269, 271, 273 teasing 11, 12, 14, 52, 82, 107, 119, 166, 176, 179, 183, 184, 186, 189, 193, 196 trumping 12, 36, 174-175, 189

Index verbal humour, verbal jokes 1, 8, 11, 42, 45, 46, 52-53, 60, 62, 72, 74, 82, 96, 107, 117-123, 156, 173, 184, 273 violating maxims, maxim violation 3, 22, 156-162, 172-174, 176179, 185-188, 191-192 violation of the CP 156, 170-174, 185, 187-193, 207 what is said 166, 175, 197-203, 205, 207-210, 212, 216, 217, 219, 220, 223, 226-230, 233, 234, 238, 241, 244, 251 wit 58, 111, 175 witticism 1, 2, 4-6, 11, 13-15, 17, 24, 25, 28, 30, 32, 33, 37, 47, 54, 60, 61, 63, 66, 82, 96, 114, 115, 157, 174, 178, 181, 183, 184, 193, 195, 196, 273 wordplay 12, 13, 99, 117-122, 268