EDITORIAL BOARD General Editor JERROLD M. $ap0c% Department of Linguistics University of Chicago Advisory Editors Groxct D. Beoett 1 PRank W, Hey University of California University of Groningen Los Angeles Lawnence Hors Tuomas G. BEVER sual University of Southern California Columbia University Jean Cssonanve Gronce Laxorr University of Florida University of California, Berkeley Eve V. CLank Stanford University Perer W, Cuticover University of California, Iivine Daviv Dowty JOHN R. Ro Ohio University Massachusetts Institute Groncts M. Green of Technolowy University of linois Rosin Lakort University of California, Berkeley James D. MeCawuey University of Chicago SaNpFORD A. StaNe Jone Haskastoen University of California, Harvard University San Diego SYNTAX and SEMANTICS , VOLUME 9 Pragmatics Edited by PETER COLE University of Mlinois Urbana, Mlinois Acapeic Press New York San Francisco London A Subsidiary of Harcourt Brace Jovanovich, PublishersCorvaicitt © 1978, ay Acaoe se Press, Ine NC United Kingdom Edition published by ACADEMIC BRESS, INC: (LONDON) LTD. Lsaaty oF Concnss Caratoc Cano Nunn: 72-9123 ISBN 0-12-613500-6 CONTENTS List of Contributors Preface On the Origins of Referential Opacity Peter CoLe = Introduction Autributive and Referential Descriptions Referential Opacity ‘The Etiology of Opacity ‘The Relative Scope Hypothesis Cardinality of Readings Quotational Complements and Propositional Complements Evidence for a Quotational Analysis The Fall of the Scope Hypothesis, Conclusions and Unresolved Issues References Negative Scope and Rules of Conversation: Evidence from an OV Aulce Davison NP Specification Permutation Emphatic Particles Context Multiple Clause Relationships ‘Semantic Representation Conclusion References B 30 x 3 35 35 38 4a 4“TWO TYPES OF CONVENTION IN INDIRECT SPEECH ACTS J. L. MORGAN University of tins INTRODUCTION In this paper I want (o take up the problem of “indirect speech acts." as exemplified by the infamous case. Cun you pass the salt?, with the goal of reaching an understanding of its apparently paradoxical nature. In consider ing the competing analyses of Gordon and Lakoff (1975), Sadock (1974) and Searle (1975). my initial inclination was 10 reject Searle's discussion as missing the point, in favor of one of the other two. But I have gradually come around to Searle's position, or perhaps I have only constructed a misin- tetpretation of it that appeals to me. At any rate, in this paper 1 will be attempting an elaboration of my interpretation of Searle's remarks. T will argue for an account of Can you pats te salt? and similar expressions which treats them as CONVENTIONAL but not 1p1OMS. by establishing the necessity for distinguishing two kinds of language-related convention: CONVENTIONS OF LANGUAGE, that jointly give rise to the literal meanings of sentences: and CONVENTIONS OF USAGE, that govern the use of sentences. with their literal meanings, for certain purposes." I will suggest. in short, that Can you pass the salt? is indeed conventional in some sense. but not an idiom. Rather, it is conventional to use it (with its literal meaning) for certain “aman eater version ofthis paper I had use the term “convention AROU' language,” 38 ‘opposed to “convention OF langage, “sine I wasted 19 aod the meaning suse controversy Hot the original terminlogy proved so confining hat have hereused the phrase "convention oF sat. Ieshould hectear hat these conventions ce distinct rom conventions itera meaning my “conventions or language repirdles of whether eral meaning sdeseribed in erms of truth contin oc rules a use 261262 4 Morgan purposes. Part of my tisk will be to dissipate the fog of initial implausibitiny by establishing on independent grounds the need for this kind of convention T hope to end up with a framework that gives a reasonable picture of the diachronic transition from indirectly conveyed to literal meaning and allows, the possibility of intermediate points on the natural-conventional scale. wil also argue, contra Searle, for the notion “conversational postulate’, which | hhave recently argued against (Morgan 1977) Twill proceed as follows: First { will briefly review the nature of 1 problem involved in expressions like Cun you pass the salt? This will be followed by a discussion of the role of pragmatics in linguistics. leading up 1 a discussion of “natural” ay opposed to “conventional” and pointing out the difference hetween two kinds of linguistic convention, { will then offer a schema for describing the less familiar kind of convention, and an account of Can you pass the salt? in terms of this kind of convention. [ will end with a number of examples of various subtypes of conventionalization STATEMENT OF THE PROBLEM Why are expressions like Cam you pass the salt? a problem? Why do { sy that this expression is apparently paradoxical? The basic faet is this: One can use a sentence like Can you pass the salt? to convey a request, though it ims at first glance we woukl not want to consider the literal meaning of the sentence to be that of a request for the salt. Grice’ y (1975) notion of “conversational implicature’ and accompanying maxims offer a potential explanation for this fact (ef. Gordon and Lakoff 1975), but how can we be sure this is the correct analysis? In faet. there are Two ways to go about giving an account of such cases. and one can construct a case for euch that has a certain amount of initial plausibility. The first way, which [ will call the NATURAL approach, is to argue that even when mean to make # REQUEST in uttering Can you puss the salt?, Lam using the sentence with its literal meuning of a yes/no question: the fact that, by asking this yes/no question, I can manage to convey what amounts to a request is not a matter of knowledge of English. but a consequence of Grice’s maxims. ‘which are, roughly, a set of rules for inferring the intentions behind speech acts. or. from the speaker's viewpoint, for selecting one’s utterances s0 ats to convey one’s intentions. by exploiting the maxims. Given that the need for Grice’s maxims has already been clearly demonstrated and that we can show how the request nature of Can you pass the salt? is “calculable.” that is, can be derived from Grice’s maxims, then Occam's razor dictates that We take this as the correct analysis. kicking strong evidence to the contrary Further support might be derived from the admittedly vague intuition that i “just FEELS like"” one means it in its literal meaning even when using it make a request, a point that gains some support from the frequently noted | | ‘Two Types of Convention in Indirect Speech Acts 263 fact that the class of possible responses to Catt rou fone would expect from ity literal meaning (Or one can take a conventional approach. saying that Can you pass the salt? is an idiom that wears its history on ity sleeve. as idioms often do, s0 that what the expression formerly had as implicature. it now has as literal meaning. As a consequence. Can you pass the sult? is now genuinely ambiguous between the literal meaning of a yes/no question and the literal ‘meaning of « request, One can support such an analysis by observing first that Ca vou pass tie salt? has some of the grammatical marks of dire: requests—the possibility of preverbal please, for instance—that not all cases of genuinely indirect requests have. Second, although Cam you pass the salt? is indeed CALCULARLE, itis not in fact calculated: rather, one gets the point mote of less directly, without any inferential processing. which is what we would expect if it has become an idiom, thereby part of knowledge of language. Third, Can you pass the salt? is intuitively more direct than its apparent close paraphrases, like Are you able to... and Is it possible for vou to... Which do not have the grammatical properties of direct requests, like preverbal please, but can, ronetheless. be used to convey indirect requests. Fourth, this kind of conventionalization of indirectly conveyed ‘meaning is in fact clearly attested. which at least increases the plausibility of the idiom approach For instance, as Robin Lakoff (1973) has observed. the typical history of euphemisms, expressions the speaker uses to merely hint at what he wants to avoid mentioning directly. is that they eventually take on as literal mean- ing the very thing they were originally used to avoid. One can see a clear example of this in the expression 10 go t0 the bathroom, which obviously originated as « euphemism. having. literal meaning like “to transport oneself to the bathing room’, with the conversational implicature that one actually went there with the purpose of excretion, but at the same time avoiding direct mention of such revolting matters. But now. in at least some American dialects, the implicature has been conventionalized as literal meaning. $0 that go 10 the bathroom is now an idiom with the meaning “to excrete" speakers of these dialects thus can say, nonmetaphorically, The dag went 10 the bathroom on the living room rug. Cole (1975) presents. a persuasive discussion of another kind of granmaticalization of implicature, focusing in particular on this as the most reasonable treatment of the expression ler's ‘Then we have the apparent paradox that the expression Can vou . . is in some ways natural, in some ways conventional. How can we have both at the same time? 1 will argue that the answer lies in the following quotation from Searle: “It is by now. | hope, uncontroversial that there is a distinction to be made between meaning and use, but what is less genetully recognized is that there can be conventions of usage that are not meaning conventions (1975-76) Before exploring the idea in this quotation, T need to discuss convention and pragmatics a bit, is just about what2h ba PRAGMATICS AND LINGUISTIC: To decide between the “natural” and “conventional” approaches. itis necessary to make clear what these terms mean. To do this. I must begin with a general discussion of pragmatics. As faras [know. the term wats until recently applied to the analysis of expressions like indexicals. whose mean. ing can be fully specified only relative to context of utterance, Recently, though. the term huts been extended to cover matters like Grice’s conversa tional implicature that are not part of the literal meaning of sentences. As a result, “pragmatics” may be in danger of hecoming a useless eateh-all term, But there may be a grain of truth in this lumping together of conversational implicature with the interpretation of indexicals and the like. 1 think a ‘moment's reflection will show that there is a natural connection, and that the problem of indexicaly is naturally subsumed under the problem of the in- terpretation of intentions behind use of linguistic expressions. If we mean that a pragmatic treatment of demonstrative pronouns and of deictic terms Tike here and mone should include a recapitulation of the principles we use in determining referents for these terms. then itis clear that itis the Same sort ‘of problem, depending on stich matters of context as our interpretation of the speuiker’s goals in the conversation, his intentions, interests, and so on, For example, imagine a jar of sugar with a ghiss Tid. on which the word sugar is, puinted in blue: and imagine that someone puts her fingertip just under the letter w of the word sugar and says, What's that? Our answer might be, among other things. the letter u, the word sugar, paint, blue paint, blue, English, a lid, glass, a glass lid, a jar, sugar. a jar of sugar. and so on, depending on our interpretation of the person's interests—is she learning English. the use of seasoning, physics. or what? ICS clear that there is a natural connection between an account of indexical expressions and the interpretation of intentions. But there is occasional confusion, it seems to about the nature of pragmatic principles, so a brief discussion of their nature is in order ‘A central question for the study of language is this: How do people ‘understand what's said to then? Linguistics must eventually provide at least ‘4 partial answer to this question by saying how much and in what ways knowledge of language per se contributes to the ability to comprehend. It has. become fairly obvious in the past few years that a good part of comprehen: sion must be ascribed not to the rules of language that assign meanings 10 sentences as a function of the meanings of the parts. but {0 our ability to somehow infer what the speaker's intentions were in saying what he said. with the literal meaning it has. But this ability is not, in general, a strietly linguistic ability—in fact, [think often not a linguistic ability at all, but the ‘application to linguistic problems of very general common-sense strategies for inferring intentions behind all kinds of acts, which may or may not be Two Types of Convention in Indirect Spcch Acts 265 different in different cultures. And to eall them rules of conversation is misleading in the same way that itis misleading to refer to rules of driving as rules of getting to the grocery store and back. It should be clear upon reflection that. unless we are in solitary confinement. we spend most of our waking hours interpreting observed events involving other people in terms of intentions and related notions like purpose and interest—not consciously, of course. but we do it nonetheless. As long ay we are able to do it with ease, and to pigeonhole these events in terms of nonthreatening intentions, the matter does not occupy our thoughts, But if a case arises that is not easily classified—we don’t understand the intentions invelved—it catches our at= tention. and we say expend some effort t0 resolve the matter, even if the outcome is of no consequence to the conduct of our affairs. For example, if while studying in the library I notice the person at the opposite carrel slowly and quietly removing pages from a notebook, wadding them up. and putting them in a wastebasket. I probably would ignore him and continue my work But if he repeatedly removed a sheet of paper, wadded it up, unwadded it, replaced it in the notebook, removed it again, and so on, [ would be unable to work until T had provided myself with an explanation of his behavior. Less bizarre cases confront us constantly. Lopen the door to find a person standing there who holds out a package, and instantly 1 interpret her be: havior as motivated by the intention that I take the puckage. Many everyday eases fit Grice’s (1957) characterization of “non-natural meaning" of an uticrance: “A meantyy something by x" is (roughly) equivalent to °A intended the utterance of x to produce some effect in an audience by means of the recognition of this intention’ ” (p. 385). For example, such everyday acts as holding open a door for somebody or looking daggers at somebody who is on the point of revealing a secret are quite analogous to meaningful utterances under Grice’s characterization. And the notion “conversational implicature’ can be naturally extended to nonlinguistic acts. If upon being asked my opinion of a spinach soulle [ have been served. T shovel the contents of my plate into the dog's dish, have rendered my judgement as clearly as if Fhad said Irs aveful, though less directly. In interpreting my action the questioner must invoke Crice's maxims just as if [ had responded by saving something In short. then. conversational pragmatics of the sort Grice discusses is not really conversational at bottom, but the application of general principles of interpreting acts. applied to the subcase of communicative acts, and more particularly, verbal communicative acts. Unless I have misinterpreted him, | ‘am following Grice in this. Well then, one might object. this is not linguistics, at least not if we narrowly limit the subject matter of linguistics to those abilities that are tuniquely linguistic abilities. And the only answer is. of course it’s not. But even if we accept this narrow definition of the scope of the field, we are stuck266 JL Morgan with pragmatics on methodological grounds, Semantics is now irretrievably part of linguistics. But our data about semantics are not direct, but really Gata of comprehension. Comprehension is demonstrably a mixture of prag- matic and semantic matters, and introspection supplies us no simple clue to what's semantic and what's pragmatic in a given case. Our only methodolog- ical tool consists of the tests for implicature discussed by Grice. which Sadock (this volume) shows to be difficult to apply. A major problem for linguists and psycholinguists is the devising of reliable empirical tests for distinguishing semantic properties from matiers of implicature in com. prehension, But it must be kept constantly in mind that pragmatic “rules” have to do not with linguistic abilities, but with more general ones. so that if it can be shown that a linguistic theory of meaning like Montague grammar ‘or generative semantics can give a unified account of semantics and prag- matics (especially ifthe account isin terms of formal properties of sentences), ‘we should suspect that there is something wrong with the theory. unless we want to give up the position that there is a difference between the two. NATURAL VERSUS CONVENTIONAL (One basis of difference between semantics and pragmatics is the distine tion between conventional and natural. By NATURAL | mean that kind of “information” that one can reasonably infer as (part of) what the speaker intended to convey, but where the inference is not based DIRECTLY on any kind of linguistic convention but on assumptions about what counts as rational behavior, knowledge of the world, and so on, Let me give a couple of examples of natural inferences, to make clear what I mean. First a nonlinguistie exsimple. Imagine that { approach a classroom door and turn the knob, The door does not open. I continue turning the knob back ‘and forth, but the door still does not open. A person whe has been watching ime (and who T have noticed watching me. and who f observe has seen that 1 have noticed him) approaches and hands me a key. I thank the person. insert the Key in the lock on the door, unlock it, and so on. Now there are a number of inferences here, none of them based on any convention save for the conventions involved in the use of rhunk you. The inferences | have in mind are these: The other person inferred from my behavior that I was trying t ‘open the door, aind that I was having no success. Notice this is not the sort of inference one would want to consider a matter of COMMUNICATION: it was not my intention that the person make this inference (cf. Grice’s definition of “non-natural meaning”), But the next inference is indeed communicative. The person hands me « key, and | am justified in inferring that 1 am being tziven the key so that I can open the door with it (thus that itis in faet a key (0 the door Lam trying to open). {am justified in assuming this in that (leaving 7 {wo TyEes of Convention in Indirect Speech Acts 267 ‘out many steps), given that the person is tational, and knowing that he has seen me vainly trying the door. and that he knows that [ know he saw me, then the most likely interpretation 0” his behavior is that he is giving me the right key so that can open the door. Moreover. itis fairly clear that he must be aware that Iam very likely to make this inference. and he has done nothing « stop me from making it. so he must intend for me to make it. No doubt this description will call 1o the reader's mind various points in Grice 1975. It is intended to. Notice that in no way is there any convention involved in this inferential chain, unless one would want (© say that there is some cultural convention like “be helpful” involved. At any rate. itis clear that most of the steps i the inference are natural, rather than convention based. ‘There is an inference involved in the interpretation of my use of thank vou, fon the other hand, that could be described its involving both convention and natural inference. The inference | have in mind is the justified inference by the other person that in saying shank you, [ mean to thank him for giving me the key. The inference here is in part conventional, in that it is based on knowledge of the English phrase shank yor. and on knowledge of the conventions concerning when one thanks and for what kind of thing. But it also involves natural inference in is figuring out just what itis I'm thanking him for As far as communication is concemed, then, Luse the term NATURAL ina way that would be appropriately applied to meaning that is conveyed, or at least can be conveyed, via inferences about intentions behind communica: tive acts, as in the ease of conversational implicature. In such cases, the relation between what is said and whit is conveyed as natural meaning is not arbitrary. as itis in the case of the literal meanings of words. but can be reasoned out from the literal meaning taken together with the facts surround ing the utterance (i.e. “context” By CONVENTIONAL, on the other hand, is usually meant the relation between linguistic form and literal meaning, which is arbitrary. « matter of knowledge of language. One cannot reason out from the word dog that it is uused to refer to a certain kind of animal; one just knows it (or not) as a synchronically arbitrary fact of English. Such knowledge is knowledge of the conventions of English, which jointly constitute all or part of knowledge of language per se. But as Searle points out in the passage quoted earlier, there is another sort of languuge-related convention, conventions of usage: “It is by now, Fhope. uncontroversial that there is a distinction to be made between meaning and use, but what is less generally recogrized is that there can be conventions of that are not meaning conventions” (1975:76), arle has in mind as a case but I think Now it is not crystal clear in this passage what of “conventions of usage that are not menning conventions,268 JL. Morgan there are cases that can be perspicuously described in these terms: in particular. conventions that are. strictly speaking. not conventions of kan, Buage, but conventions of usage of language. properly considered conven. tions of the culture that uses the language. For example. just as in our own culture it is conventional to greet someone by inquiring after the other person’s health, so [am told that in some cultures itis conventional to greet by asking after the other person's gastronomic welfare, most likely (but not exclusively) by saying something like Have you euten?, ie. its direct transl tion. Now on the one hand Have you eaten? is by virtue of its semantics a natural way of greeting someone by conveying concern for his well-being, tziven the right conditions in the culture. as opposed to Seven is prime or Your hair is missing; but at the same time itis entirely arbitrary whether or not a given culture uses Have you eaten? as a CONVENTIONAL way of greeting. And 1 think we would not want to say even when itis a conven: tional greeting that the expression Have you eaten? means the same as 1 _greet sou though indeed that kind of linguistic change does oceur now and then. Rather, the convention involved here is cultural convention about the use of language, not part of the language itvelf—though that is not to say a good language teacher would not teach it. ‘Another case: according to Webster and Webster (1968), the customary way of opening a conversation among Eskimos is by saying (the direct translation of) You are obviously ——. where the blank is filled according to ‘what the hearer is doing at the time of the utterance—for example. You are obviously reading Kant ot You are obviously skinning a seat. Again. | think. we would not want to say that the conventional literal meaning of the expression is merely a statement of intended effect. namely, to open a conversation, Still there is a convention of some sort here, to the effect that itis customary or conventional to say a certain sentence AND MEAN 11 under certain circumstances, with certain purposes. Still another example: According to Wolff (1966). in Cebu cu not knock at a door but says something in the way of greeting. like good morning. Both ways—knocking and greeting—would seem 10 be equally effective as natural ways of getting the attention of the inhabitants and provoking them to open the door. But one way is conventional in Cebu Culture. the other in ours. We might be tempted to assign something like request For opening’ as the literal meaning. so to speak. of the knock. since its use for that purpose is indeed conventional. and seems not to stem from any other ““meining”” associated with knocking. But the temptation is less tereat t0 say in the Cebu case that the expression translated ais “good morn: ing” is ambiguous between the literal meaning of a morning greeting and that ‘ofa request to open the door. Rather. it would scem more appropriate (0 say that there is u convention to the effect that one announces one’s presence it jure one does “Two Types of Convention ia Indirect Speech Acts 269 the door, ete, by issuing a greeting to the inhabitants, This is not a conven tion of the language, but about its use. In sum, then, Lam proposing that there are at least two distinet kinds of convention involved in speech acts: conventions of language (for example, the meaning of dog, the fact that in English the subject of a passive sentence is interpreted as (roughly? patient, and so forth] and conventions in a cultuse of usage of language in certain cases [for example, the fuct that to start an auto race one says Gentlemen, start your engines (and MEANS it), the fact that one is expected to say something in the Way of consolation at funerals. and so or}. sometimes (but sometimes not) with particular expression (with their meanings) mentioned in the convention ‘The former. conventions OF the kinguage, are what make up the lan- guage. at least in part. The latter, conventions of usage. are a matter of culture (manners, religion, law, . i not knowledge of language perse. And I propose that by looking litle at the STRUCTURE. so to speak, of this second kind of convention, we can derive first an account of the apparent paradox involved in cases like Can you pass the salt?. in which they are treated as simultancously conventional and natural, just as Searle says. Second, we will see that there is a range of pissihilities for conventions intermediate between naturalness und conventions of the language. Third, along the way, swe will construct a plausible picture of at least one way that expressions can cchange their status diachronically, by passing from the status of convention bout kinguage t the status of convention of language CONVENTIONS OF USAGE As an initial approximation, 1 think conventions of usage can be considered to contain three kinds of elements: occasion, purpose, and means. AS the statement of means become more and more specific, the convention approaches i convention of the language. a statement about literal meaning. As the connections between purpose and means become obscured, the relation between them is ripe for reinterpretation as entirely arbitrary. at which point the convention of usage is reinterpreted as ‘a convention of the kinguage. As an illustration, we might consider various hypothetical versions of a convention concerning departure salutations, proceeding from less to more specific versions. As a rather nonspecific version, we might start with something like a statement of occasion (which not all conventions will have), and purpose, as in (2) (1) Upon parting. one expresses one's regard for the other person.x0 Jot Mr Then (1) might be made more specific in a given cult ° eans specifications. as in (2a) or (2b): . a Oa by expressing a concer é 1g « concern for the welfare of the other person by expressing a des 1g a desire or intention to see the other person agai is of ons cin i tm ind veh tn a by expres aby expressing @ wish for x a been or goud health BA okie te sod of Gad vend eae Ww expressing a wish for peace ne PON Each of these com conventions. of course, ee course, could be mad fing considerable freedom nade more specific si eae Tne ene cae Nae etc nel . ranialei ee eee jointly defined by (1). 2a). and Qb) by uttering pire comets “ May God be with you. God be with you. Toray 0 God that He L hope God will be good to - mn but one fer conn er type of specificity leads to a qual cen: i ion i HERI ee eter pression to be used in conforming 10 the s specification of eee ae es ING it in its literal meaning, a! y Sorte tte ee to the greeting convention. Oa oe is ten meaning; one says it an ci rea tam ater ely su Types of Convention in tars Speech AC m itis amatter of convention that one SAYS cession). Ye' imstances. choose some other eNPr sre means it ora east purparts to mea i) under certain eiteus for certain purposes rheve is 4 naaainess co the convention #98 tive chain between the (enost general) PAPO” wecifeation of the means in term of 9 NTS sean of the kanguage earner i to infer the of this chain. that is, the Beare petween the occasion of USAEE 8nd the ex: Peis no fully reconstructed. the connec” Come arbitrariness arises. the rela Thus the original con- earners to fully recon in the following hat there is a natural connec ‘iyo the convention and the iar English sentence. Part of purpose-meaning COP! Insofaras this ch Prcames arbitrary 40 some degrees On pee Mes pe for reinterpretation as entirely PATE! ee ncab).. through failure of ange vention (1) enti ge oven expression cht might BE terre pression used. ways. cane invokes the gourd will of God toward 1 (6) Upon parting. he Enclish sentence God be with you ther person by saving 1 cone expresses one’s regard Jor the other Person (1) Upon partie: God be with you by saying the English sensence (8) Upon parting, ome sass the Ents sentence God be with you (8), where the convention between not via « purpose-means eh. 1 vaio no longer plays. direct ole ‘expression has a certain rat meaning, has to do she most arbitrary version, of COUFSE:f deeasion and expression is stated dices Oech eases. the meaning. of the Titeral NPR soc caevention speakers may be sware (hak the ial mening. But may be entirely Unaware what t with parting, Te eae ofthe expression brea eg to Win a performance’ is an especially imteresing or of a Perens headed for eventual Salts Of conventio that gt this expression originated de 10 (NF superstition to tek to wih someone Bood lwck—there Fe ES wis dood luck by wishin him bad Tacks a broken leg. But set pression is Figily fixed as partof the ‘convention: none of Sraaphrases below will do forthe sme purpose: svperformer good luck before fa convention of usage wm of language. {'m wn that it would be hed a fellow actor ‘by now the choice the plausible Fracture « tibia: ® uo Break your les an T hope you break a leem J. Ls Morgan ] It is likely that newcomers to the theater subculture will not be aware of the history of the expression, so that the connection between purpose and means will be direct and arbitrary: Before a performance, t0 wish a per: former good luck in his performance. say break « leg. But the expression is not theteby ain idiom: if it were we would expect to find it used as if it were aan idiom whose literal meaning was “have good luck’, as in (12) as a way of saying (13), But the expression cannot be used this way (12) John really broke a leg last night (13) John realy had good luckidid well in his performance last night ‘The schema [ have argued for seems to fit nicely in this ease: an occasion, a purpose. and a means. the means specified as the ulterance of a particular expression. But the original natural connection between purpose and means has now been lost Given this view of conventions of usage. the language learner's task is to discover or reconstruct the details of the connection between occasion and purpose. on the one hand, and linguistic means—the sentence used—on the other. In the case of the literal, nonformulaic use of language. the ated in a natural way, with the literal meaning of the sentence as one of the links in the connecting chain, as in some of the parting conventions discussed earlier. But these connections. where they are not trivial (e.g. saying [1's raining to convey that its raining), must be worked out by the language learner, whose only immediate data are inferences in context about the occasion and/or purpose of the utterance and the expres: sion employed. It may take some time for the language learner to fil in all the missing links in the chain. Accordingly. we might expect to find that chil dren’ linguistic competence has typically more of this arbitrary connection than does an adult's. But even in the case of adults there will probably be interpersonal variation on some expressions. describable in terms of the number of missing links in the knowledge of use of the expression. For example, we might find that knowledge about Gesundheit is best captured by (14) for some adults, by (15) oF (16) for others: connection is me (14) When someone sneezes, 10 express concern for his health, say the German word for health, Gesundheit (15) When someone sneez say Gesundheit. 5, 10 express concern for his health, (16) When someone sneeses. say Gesundheit. ‘Two Type of Convention in Indirect Speech Acts 27 The third version, (16), if indeed if actually occurs. iy of a rare type: The | purpose has been entirely lost, so that the speaker knows only the oveasion Of using the expression, the only purpose for saying it being the purpose of conforming to the convention. It may be that this kind of case is more Frequent among children: When such-and-such happens. one is supposed to say so-and-so, Linguistic change arises when a speaker (or group. or entite generation of speakers) fails to reconstruct all the links of the chain, resulting in grcuter arbitrariness of the connection between purpose and expression. and poten: tially leading to use of the expression in situations incompatible with the ‘original literal meaning of the expression. An obvious kind of example is the use of expletives like for Christ's suke by non-Christians. or God damn it by atheists: but there are more interesting cases as well, ranging from the tutierance by German speakers of auf Wiederselen to people whom one Knows one will never see again.* to eventual change of literal meaning at the levical level. But this kind of change is inbibited when the expression transparently has a (relevant) literal meaning. When ity literal meaning is obscure (as in the ease of Gesundheit) or becomes obscure due to linguistic notice the archaic subjunctive in Ged he with you) speakers may not ze that the expression hats & literal meaning distinet from its purpose an the connection between purpose and form becomes arbitrary. Thus God be with you eventually becomes goodby by phonological change. In such a case the question arises whether iti the growing arbitrariness that makes the phonological change possible. or the other way around: or do the two reinforce each other? At any rate. it is clewr that a distinction must be made between conven: tions of langwage—matters of literal meaning—and conventions of usage. ‘And the descriptive schema I huive given for the latter. in terms of ‘oceasion-purpose-meins chains. allows a pkusible account of the change from, convention of usage to convention of language.’ It also gives a picture of thingy wherein some cases are more arbitrary than others. in that more purpose-means links have been lost in one case than in the other. Now given this kind of convention, how can it be extended to cases like Can you pass the salt? What’s needed is a description that says that in using Can xou pass the salt? to make & request, one is using the sentence with its Such cases ease questions about the rife of habit in this Kind of Hinguistie change. For ce\ample. Pave norive in my om hehsvite the ecent annoying habit of using se son later ‘shen taking leaves even in circumstances where is clear to everyone involved tha Iwill ot ty inerocotor Later. © As farasFcan see its ogclly poss for there to arse changes inthe ater direction 38 wwe, from convention of language to cnention of usage. But such a change would Teahive 2 far move complicated chain of events. an may Well never pre4 J. Ls Morgan literal meaning. with the intention of conveying a request via Grice’s maxims. but that in doing so one is following a convention about language use, the convention being. roughly: To request someone to do such-and-such indirectly. say the sentence Can sou (do such-and-such)?, with its literal sense, My proposal. then. goes like this: The expression Can sour. is not an idiom. but has only the obvious literal meaning of a question about the hearer’s abilities." One can readily see how the expression could have, via Grice’s maxims. the implicature of a request. In fact it has become conven. tional wo use the expression in this way. Thus speakers know not only that Can you... has a certain literal meaning (a convention of language): they know also that using Can you. . . isa standard way of indirectly making a request (a convention of usage). Both are involved in a full under- standing by the hearer of what is intended in the use of the expression, SHORT-CIRCUITED IMPLICATURE ct this will stike some readers as counterintuitive, in that the feel" of an implicature is lacking. One ean see that a request implicature is calculable via Grice’s maxims, but the subjective reaction is that the request nature of the speech act is conveyed without the sort of indirect feeling we attribute to the presence of inference: the literal meaning is in some way latent. rather than the basis for an inference. 1 think this intuition is correct, ‘and that we need a notion of “short-circuited” implicature to account for it Let me choose another. clearer, illustration to show what I mean by short- circuited implicature pppose [ have a stingy friend. One day when asked for a loan. he replies, Do I lovk like a rich man?. intending thereby the conversational implicature ‘of refusal. Now suppose my friend is not very imiginative. and, impressed by his own wit, he comes to use that sentence for refusing loans as a matter oof habit, Still, itis a habit of saying a certain sentence, with its literal meaning, intending thereby to convey a refusal indirectly via Grice's maxims. But in interpreting my friends’s utterance, I no longer have to make the inference—his habits are now part of my background knowledge, Upon hearing him say it in the right context, this buckground knowledge tells me immediately what he is doing. Now suppose my friend’s habit spreads. so that it is common throughout the community to refuse loans by saying Do 1 look like a rich man? To be a member oF a culture is to Some extent to be an observer of the culture loans are commonly refused (more specifically. perhaps commonly INDI T susp “1am ignoring othe reais of Can sou «(6 deontiy and epistemic readings of cam tht are relevant tothe present dcasion embers will thus observe that in this community * “Two Types of Convention in oaiect Speech Acts 25 RECTLY refused) by saying. Do Took like a rich man? (though there may be other modes of refusal as well. Thus it becomes “common knowledge" in the community that people refuse loans by saying a certain sentence, with its literal meaning, intending thereby to convey an implicature of refusal. But the inference of the implicature is short-circuited: armed with this common knowledge. | know more or less immediately, WITHOUT CALCULATING THE INFERENCE, that an implicature 0” a refusal was intended. Nonetheless. a speaker of the language who lacks this bit of common knowledge will understand what is intended if he hears Do I look like a rich man’, by the original route of conversational implicature. SOME MORE CASES {suspect that the reader will not yet be entirely convinced by my (admit= tedly counterintuitive) claim that an expression can be conventionalized and at the same time keep its literal meaning. Let me therefore present some more examples, of two kinds: first, cases where a particular expression is wart of the convention. and secend, some cases where it is a “rule of conversation” that is conventionalized. Cases abound where it is conventional to use a particular form for a particular purpose, bat where the literal meaning of the expression is still involved. A simple example isthe forms used for identifying oneself over the telephone. It is conventional, at least in the midwestern United States, to use the forms illustrated in (17) or (18), or a few other expressions: a7? This is Edith Thornton. ro) Edith Thornton speaking. On the one hand, in using these expressions one means them literally. But on the other hand, it is purely a matter of convention that one uses these particular forms rather than any of (19-22), which are equally appropriate if considered on semantic grounds alone (their literal translations might well be used for this purpose in another language: (20), for example. is used in Hebrew], but just happen not to be conventional English means of identifica tion over the telephone. The slight difference between (18) and (21) is especially interesting as a demonstration of how form specific such conven tions can be 19) Here is Edith Thornton. re) Here is Edith Thornton speaking.276 JL. Morgan ] ‘Two Types of Convention in Indirect Speech Acts i en Edith Thornton iy speaking. i Lam Eatith Thornton, There are many cases that involve implicature, but as a matter of conven, tion, so that the implicature is short-circuited. For example, (23) is com. monly used to convey something like (24) 2H Uf you've seen one, you've seen them all 24) They're all alike, so it's a waste of time to examine them separately It is intuitively clear that (24) could be reasoned out as an implicature of but it is now conventional to use (23) to convey (24). Although one ha in mind the literal meaning of (23) in using it to convey (24). the form of the ‘expression is strictly part of the convention. Sentences having precisely the same literal meaning but even slightly different Form do not convey (24) with, the sume immediacy. If one manages to convey (24) by saying (25) or (26. it will be as a fresh implicature, without the short-circuiting that accompanies the conventional form (23): es If you've seen one, xou've seen all of them. Co You've seen them all if you've seen one Below are some more cases of conventionalized implicature, where a certain expression, with its literal meaning, is used more or less conventionally to convey a certain implicature: en mu can say that again. (ef, You can repeat tht.) ee) How many times have I told you. ef, Tell me how many times Pve told sou...) 9) Ie takes one to know one. (cf.. Ie requires one to recognize one.) It should be pointed out that another kind of knowledge of usage (not of Tanguage can play an important part in the short-circuiting of implicature: Knowledge of previous use of an expression. A kind of common knowledge am (not always directly rekited 10 literal meaning) about particular expressions can be exploited to bring about a conversational implicature. as in “cliches” like (30-33), 30) [policeman to motorist] Where's the fire? BD no questions asked, re) [spouse to spousel "re gor a headucke. oy Your place or mine? But these expressions are clearly not idioms, One uses them meaning them literally, though their use conveys much besides the literal meaning, In other cases it is knowledge about particulars of history of use that is exploited for the sake of implicature, One conveys more than literal meaning in saying (34-36) by virtue of the heurer’s knowledge of well-known previ- fous uses of these sentences: Ga) Am L my brother's keeper? 39) Fwant to be alone. (36) Fd rather be in Philadelphia. ‘The hearer will recognize that these are famous lines, will conclude reason- ably that the speaker must have known he will make the historical connec- tion but did nothing to stop him from making it, therefore must have in- tended it to be made. and so on. The allusion, and resulting implicature, are conveyed in the usual Gricean fashion. Finally, T come to the question of the conventionaliza conversation. Just above [ presented saves involving particular expressions and the conventionalization of their use for certain implicatures, as in the case of If you've seen one. yow've seen them all, of the original example, Can you pass the salt? 1 said in the latter case that it had become a convention of usage to use this expression, with its literal meaning, to convey an implicature of a request. The question now arises, can there be this kind of conventionalization of ruk’s of conversation? | think there can. For example, it iy more or less conventional to challenge the wisdom of a suggested course of action by questioring the mental health of the suggestor, by ANY appropriate linguistic means, a in: jon of rules of 67 Are you crazy?278 JL. Morgan 38) Have you fost your mind? 6» Are you our of your gourd? tand so on. Most Americans have tWo oF three stock expressions usable as answers 1 obvious questions. 3s in ao Is the Pope Catholic? ap Do bagels wear bikinis? Bur for some speakers the convention does not specify a particular expres. sion, and new ones are manufactured as they are needed. It seems that here 4a schema for implicature has been conventionalized: Answer an obvious yevino question by replying with another question whose answer is very obvious and the same ingwer you intend to convey. In similar way. most speakers have a small number of expressions usuble as replies to assertions. with the implicature that the assertion i transpar- cently false—(42). for example: the a Yes, and U'm Marie the Queen of Romania: But again. for some speakers the convention specifies only a general strategy. rather than at particular expression: To convey that an assertion is transparently false, reply with another assertion even more transparently false. Hearers unfamiliar with the convention will take longer. having to calculate as conversational implicature what most Americans (at least) will recognize immediately. But itis clear that this conventional strategy could have arisen (and probably did arise) as a conversational implicature that became conventionalized. What was formerly u matter of natural inference becmes « convention about language. The result is the hypostaticization of 4 particular strategy of conversational implicature that one might call a conversational postulate’. In Morgan (1977) 1 criticized Gordon and Lakotf’s (1977) exposition of the notion “conversational postulate’ on the grounds that the “postulates” they proposed had no independent status, but could be (or ought to be) derived as consequences of general principles of conversation of the soit proposed by Grice, But the notion “convention of usage” as I have argued for it here allows for an interpretation of Gordon and Lakoff's proposal in which conversational postulates. would have independent status—namely. where implicuture strategies become in. situtionalized as conventions of ussige.* > Given this ew seme of onverstionl postulte” as consentontized strategy of imc ture, most of Gordon and Lako’s analyses partici eases will ove to he ecomshtee Fo “Two Types of Convention in Indirect Speech Acts 279 CONCLUSION Here is the moral: There is more to knowing “how to do things with words” than just knowledge of literal meaning. Besides knowledge of the conventions of word meanings and the semantic rules of combination, lan- guage users also have knowledge about the use of particular expressions or classes of expressions. This second kind of knowledge sometimes involves convention, but conventions of usage, conventions governing the use of meaning-bearing expressions on certain occasions. for certain pur: poses. These two kinds of knowledge are not mutually exclusive. They are involved simultaneously in the full understanding of many utterances. have left a couple of tough nuts uncracked, First. there is the methodo- logical problem of setting out empirical criteria by which the linguist can termine the status of a given expression vis-a-vis the distinctions T have discussed. Here 1 have relied heavily on the reader’s intuition that the expressions I have discussed work the way I say they do. H expect that psychologists will find it uifficult to construct simple relevant experiments. Second, I have followed hallowec linguistic tradition in carefully avoiding saying what I meant by “convention.” Some of the things | have called convention might seem more peispicuously described by phrases like ‘knowledge of shared habit" or “common knowledge of the way things are done.” T think a clearer understanding of these matters will probably strengthen my case. ACKNOWLEDGMENTS ‘The research reported herein wis support in prt by the National Institute of Education lune contract NSNHF-C-400-76 0116 and rat HEW-NIE-C-74 0007, REFERENCES Cole. B.1978)-Th sin Monga (1978, Cole, Pind Mesa, eds. 97S) Sons and Sevaanten: Speech Avis. sol. Academic Press. New York inate ynehonic and Dichromic Status of Conversational Implicatere ‘ekumple, thei analysis of Cun you as an instance of a conversational postulate does not ‘mention directly any particular expression, thin predicting incorrectly that hteralysymony- mous expressions tke Ave yom uble shu work the sme way 8 Co sim Under Ue amass I presented erie. iti just the vse of Cas.» tha a heen conventionalized 1s at indirect request. Synonyinous expressions work as genuine implicatures not shor ‘rete a6 with Ca ow atl ave this subjectively anne ingest280 JL. Morgan Gordon. D, nd G. Lakolf (1975) “Conversational postales. in Cole and Morgan 11978, Grice. H. P1987) “Meaning.” Philosophical Review 66. 177-388 Grice. H. P. (1975) “Logic and conversation.” in Cole and Morgan (1975), Lakoff, R973) “Language and Woman's Pace.” Language and Society 1 no. 2 Morgan. JL. (1977) "Conversational Postlates Revisited.” Language $3, 277-284, Sadock, JL, 1978) Toward a finenstictheare of speech acts. Acadensc Pees. Ne York Searle. J. (1978) “Indirect speech acts.” im Cole and Morgan (1979, Webster, D. and T. Webster (1968) Le's Learn Eskimo. Sunn Fairbanks, Alaska, Wolff J. (1966) Becinning Cebuano Yale Universiy Press, New Haven, Connecticut Fnsttae of Linguists, ON TESTING FOR CONVERSATIONAL IMPLICATURE JERROLD M. SADOCK University of Chicago H. P. Grice’s (1975) suggestions concerning the relationship between natural language and logic provide the outline of a system for explaining certain aspects of what utterances convey without claiming that they are part of the conventional force of the uttered sentence. The notion of conversa- tional implicature makes it possible to claim that a sentence with two quite distinct effects is nevertheless unambiguous from the point of view of its conventional content, and that (wo sentences that can convey practically the same thing are nevertheless not logically or linguistically equivalent. There is, then, a serious methodological problem that confronts the advo: cate of linguistic pragmatics. Given some aspect of what a sentence conveys in a particular context, is that aspect part of what the sentence conveys in irtue of its meaning (in the generative semanticist’s sense) or should it be forked out” on the basis of Gricean principles of conversation from the rest of the meaning of the sentence and relevant facts of the context of, utterance? Obviously, the problem of deciding whether a certain bit of conveyed information is attributable to the grammar or to pragmatics can be attacked from either direction. Either we can try to decide how one recog: nizes essentially grammatical facts and establish a rigorous methodology leading from surface structure down to meaning, or we could establish a pragmatic methodology that leads irom what is conveyed in contexts up to meaning. The first approach has been followed fairly extensively, e.g. in Zwicky and Sadock 1975. But at present a rigorous pragmatic methodology is lacking. This chapter examines the problems pervading the methodology of linguistic pragmatics, 281