Asa Kasher

Pragmatics and Chomsky's Research Programme

I. INTRODUCTION

The growth of any science involves a process of maturation. While "immature science" consists of a mere uneven pattern of trial and error, a "mature science" consists of research programmes, continuous series of theories.

This continuity, which welds a succession of theories into a research programme, is created by several threads: a conception of objectives, a sensible methodology, a philosophical "hard core", as well as an evolving conceptual framework of description and explanation.

Generative linguistics, as it has been depicted by Chomsky since its inception, is a research programme. The continuity of its series of theories should be manifest to every careful reader of Chomsky's books about language. The Logical Structure of Linguistic Theory (1955) and Aspects of the Theory of Syntax (1965) share their basic conception of objectives and their sensible methodology with, say, Reflections on Language (1975) and Knowledge of Language (1986). Similarly, the philosophical hard core of Language and Mind (1968) is the same as that of Rules and Representations (1980) or Language and Problems of Knowledge(1988a).

The continuous series of theories that the research programme of generative linguistics has comprised so far has had syntax in focus. However, the methodological and philosophical threads of this research programme clearly transcend its succession of studies of "linguistic form".

Chomsky has expressed this transcendence on several occasions. In a discussion of the study of language (Chomsky, 1968b) he says that one of the reasons why he thinks "the present-day study of language is most interesting is that it suggests a way in which other cognitive systems could be studied." (p. 113) More recently, in the preface to his Knowledge of Language, we find the same view being expressed:

"If we can discover something about the principles that enter into the construction of this particular cognitive system... [w]e can then ask whether an approach that meets with a degree of explanatory success in the case of human language can at least serve as a suggestive model for similar inquiries in other cognitive domains." (p. xxvi).

Chomsky's belief "is that the principles do not generalize, that they are in crucial respects specific to the language faculty, but that the approach may indeed be suggestive elsewhere..." (ibid.) A somewhat stronger position would be that, indeed, syntactic principles do not generalize, because they are specific to a certain language faculty, that is to say, because they form part of the emerging content of the evolving parts of the research programme of generative linguistics and as such are not supposed to be applicable outside the language faculty dedicated to linguistic form. However, "the approach", the general framework of the research programme, including its basic conception of objectives, as well as its underlying methodology and philosophy, is applicable to related cognitive domains. The purpose of the present paper is to show how this approach actually applies to the case of pragmatics.

II. PRAGMATICS IN CHOMSKY'S WRITINGS

A certain notion of pragmatics has been present in Chomsky's writings since the very beginning of generative linguistics. It makes its first published appearance in the paper "Explanatory Models in Linguistics" (Chomsky, 1962). Knowledge a speaker of a language has acquired constitutes "an implicit theory of the language that he has mastered, a theory that predicts the grammatical structure of each of an infinite class of potential physical events, and the conditions for the appropriate use of each of these items." (p. 528) Chomsky's 1973 introduction to The Logical Structure of Linguistic Theory portrays "the overarching semiotic theory in which the theory of linguistic form is embedded" in the same vein. Towards the end of the '70s this "thin" notion of pragmatics is replaced by a richer conception, that of "pragmatic competence". In "Language and Unconscious Knowledge" (Chomsky, 1978) a distinction is drawn between "grammatical competence", which involves knowledge of form and meaning, and "pragmatic competence", restricted "to knowledge of conditions and manner of appropriate use, in conformity with various purposes." (p. 224) The pragmatic competence "places language in the institutional setting of its use, relating intentions and purposes to the linguistic means at hand." (p. 225) The same conception appears in the main part of the book Rules and Representations This move, from the notion of conditions for appropriate use to the notion of pragmatic competence, is most significant, because it clearly suggests the application to pragmatics of the general approach of the research programme of generative linguistics. The introduction into pragmatics of the very idea of competence marks the adoption of the underlying methodology and philosophy of generative linguistics. Thus incorporated into the study of language use is a variety of abstractions, such as those brought forward by the distinction between competence and performance.

The formative influence of this move has not been confined to the methodological and philosophical ingredients of the research programme of generative linguistics. Once the suggestion is being made that a pragmatic competence exists and can be studied on a par with the grammatical competence, new research avenues come to the fore. For example, the notion of an innate "universal grammar", as used in the research programme of generative linguistics, will play a similar, major role in the research programme of pragmatics, viz. in the basic conception of the objectives of pragmatics. Each of these research programmes is a pursuit of an explanatory specification of the initial state of certain competence.

Although appropriate use of language is, according to the present view, the subject matter of the knowledge embodied in the pragmatic competence, not every reference to language use in Chomsky's writings is an allusion to this competence. Hence, some clarification of the use of "use" seems in place. In Syntactic Structures (1957), for example, it is pointed out that "[w]e can judge formal theories in terms of their ability to explain and clarify a variety of facts about the way in which sentences are used and understood." (p. 102) This reference to uses to which sentences are put has, however, nothing to do with conditions of appropriate use or the pragmatic competence. This becomes clear when one reads the next sentence: "In other words, we should like the syntactic framework of the language that is isolated and exhibited by the grammar to be able to support semantic description...". Here, to use a sentence is to use a syntactic device.

Accordingly, semantics is characterized in the 1973 introduction to The Logical Structure of Linguistic Theory as "concerned with problems of meaning and reference, and with the systematic use of the syntactic devices available in the language." (p. 18) In Chomsky's recent books, Knowledge of Language (1986) and Language and Problems of Knowledge(1988a), certain issues of language use are discussed in some detail, but there is no simple way of drawing from the discussions of use of language in these two books strong conclusions as to the nature of knowledge embodied in the pragmatic competence and our understanding of it. Three "basic questions" that arise in the study of generative grammar, are discussed in these books, viz. "(i) What constitutes knowledge of language? (ii) How is knowledge of language acquired? (iii) How is knowledge of language put to use?" (Chomsky, 1986, p. 3; see also Chomsky, 1988a, p. 3). The third basic question "breaks down into two parts: a "perception problem" and a "production problem". Let us have a closer look at each of these parts of the question.

"The perception problem would be dealt with by construction of a parser that incorporates the rules of the I-language along with other elements: a certain organization of memory and access..., certain heuristics and so on" (Chomsky, 1986, p. 25). Construction of such a parser could not be part of an inquiry into the pragmatic competence, because a solution of this "perception problem", which is a natural objective of a grammatical performance theory, as is clear from its incorporation of principles of organization of memory and access, heuristics and so on, could not be plausibly held to be part of the study of the conditions for appropriate use of sentences in contexts, which is what a theory of pragmatic competence is all about. In other words, the "third question" in the study of the grammatical competence - how is grammatical knowledge put to use? - should not be confused with the "first question" in the study of the pragmatic competence - what constitutes knowledge of appropriate use of language in context?

Indeed, in pursuing a broad inquiry into pragmatics, we are going to face "basic questions" on a par with those that have arisen in such an inquiry into grammar. In addition to that "first question" of the nature of pragmatic knowledge, we shall have the "second question" - how is pragmatic knowledge acquired? - as well as a "third question" - how is the pragmatic knowledge put to use? - and a "fourth question" - "[w]hat are the physical mechanisms that serve as the material basis for this system of knowledge and for the use of this knowledge", to use Chomsky's wording in Language and Problems of Knowledge (p. 3).

The second part of the "third basic question" which arises in the study of generative grammar is the "production problem". This problem is of an utterly different nature. Actually, it is the problem of the nature of "the creative aspect of language use" (CALU), as discussed by Chomsky, for example, in the paper "Current Issues in Linguistic Theory" (1964), in the book Cartesian Linguistics (1966), then in his Knowledge of Language and most recently, in much more detail, in Language and Problems of Knowledge, where it is naturally dubbed "Descartes's problem". This problem "has to do with what we say and why we say it" (Chomsky, 1988a, p. 5). The question, then, arises of the relations between the creative aspect of language use and the pragmatic competence and we would like to answer it by making three observations.

First, notice that the creativity under consideration is not some creative aspect of language, but rather the creative aspect of language use:

"The central fact to which any significant linguistic theory must address itself is this: a mature speaker can produce a new sentence of his language on the appropriate occasion, and other speakers can understand it immediately, though it is equally new to them... [I]t is clear that a theory of language that neglects this 'creative` aspect of language is of only marginal interest.... Clearly the description of intrinsic competence provided by the grammar is not to be confused with an account of actual performance.... Nor is it to be confused with an account of potential performance. The actual use of language obviously involves a complex interplay of many factors of the most disparate sort, of which the grammatical processes constitute only one. It seems natural to suppose that the study of actual linguistic performance can be seriously persued only to the extent that we have a good understanding of the generative grammars that are acquired by the learner and put to use by the speaker or hearer." (Chomsky, 1964, p. 50-52)

The creative use of language is attributed to grammatical performance. In response to Drach, 1981, Chomsky makes this point explicitly, referring to the passage we have just quoted and to additional ones:

"In fact, the passages are perfectly explicit in attributing the CALU to performance, not competence; the CALU is an aspect of the "use of language" (i.e. performance), which is "not to be confused" with competence. (Note also that the passage stresses at once the relevance of "appropriateness" to the CALU...). The passage is explicit that it is language use that is being discussed, and stresses the importance of distinguishing the study of competence from the study of language use, which obviously involves "factors of the most disparate sort" apart from the mechanisms of grammar."

Strictly speaking, the present distinction is between grammatical competence and grammatical performance, that is to say, between the grammar itself, on the one hand, and all the human systems the operation of which involves the grammar as well as other factors of various sorts. Attributing the creative aspect of language use to grammatical performance is compatible with a general theory of language which specifies and explains a grammatical competence, a pragmatic competence, as well as several other components, including psychological mechanisms and neurological systems. According to this conception, the pragmatic competence is, actually, a component of grammatical performance.

Secondly, "Descartes's problem" of the creative use of language is a problem which rests, in a sense, not only beyond the grammatical competence, but beyond the pragmatic competence too. To see that, recall what Chomsky describes as "[t]he central fact to which any significant linguistic theory must address itself", viz. "a mature speaker can produce a new sentence of his language on the approriate occasion, and other speakers can understand it immediately, though it is equally new to them." Thus, the mature speaker's ability involves at least the following ingredients: (a) a grammar, according to the rules of which, what the (ideal) speaker utters are well-formed sentences; (b) a pragmatic system, according to the rules of which, what the (ideal) speaker utters at some context is appropriate; and (c) a third element, which enables the speaker to create speech events of new types, violating neither the grammatical rules nor the pragmatic ones. All creative uses of language are, therefore, performed within the limits set by the rules of these two competences. "Descartes's problem" of the nature of the creative use of language is the problem of understanding element (c), given an adequate understanding of (a) and (b).

Thirdly, Chomsky has recently argued, with respect to "Descartes's problem", that "despite much thought and often penetrating analysis... this problem still remains unsolved, much in the way Descartes formulated it" (1988a, p. 147), and suggested a Cartesian explanation of this puzzling failure, in terms of the nature and limitations of human intellectual grasp. "Descartes's problem" of the creative aspects of "what we say and why we say it" is directly related to our pragmatic competence, because what we say is constrained by conditions for appropriate use of sentences in contexts, and why we say what we say is part of the context of language use and should, therefore, be expected to play a role in some rules which govern appropriate use of sentences in context. A full-fledged pragmatic theory would specify our knowledge of rules which govern appropriate use of sentences in contexts and explain it in terms of the innate, initial state of the pragmatic competence and the specific conditions of its growth in a certain pragmatic environment. However, there is no reason to assume that our research programme of pragmatics is bound to lead us to a futile attempt at reaching beyond the limits of our understanding. Some progress has already been made within the present research programme of pragmatics, as a result of which we do have gained some understanding of the nature of the linguistic appropriateness of sentences to contexts.

To be sure, pursuit of understanding the pragmatic competence does not depend on our ability to solve "Descartes's problem". Knowledge of the conditions which constrain "what we say" in a context is different from an understanding of the creative aspects of "what we say and why we say it".

To see that, consider, first, the simpler case of the relation between a move in a game of Chess and its "context", the latter being determined, say, by all the previous moves of both players in the same game. Knowledge of the conditions which constrain "what we do" in the course of such a game would not provide us with an understanding of the creative aspects of Chess playing, of "what we do and why we do it".

More generally, knowledge of the conditions an act or an object should satisfy for it to be a solution of a given problem does not amount to an understanding of any of the creative aspects of the situation. Knowledge of the rules which govern the pairing of a problem and a solution would explain neither the creative aspects of raising that problem, nor the creative aspects of suggesting that solution, no matter how highly rule-governed is the conceptual framework in terms of which both the problem and its solution are couched. Here too, the "pragmatic problem" of understanding appropriateness conditions is clearly different from "Descartes's problem" of understanding creativity as involved in case satisfaction of such appropriate conditions is being sought or attained.

III. LANGUAGE AND COMMUNICATION: A NEGATIVE HEURISTIC

Research programmes, such as that of generative linguistics, include characteristic methodological rules. Some of these rules "tell us what paths of research to avoid", to quote Lakatos, who called such rules "negative heuristics" (1970, p. 132.) Methodological rules which tell us what paths of research to avoid fall under two heads. Lakatos himself was interested in negative heuristics which redirect research from the conceptual "hard core" to some "protective belt" of "auxiliary hypotheses" which "has to bear the brunts of tests and get adjusted and re-adjusted, or even completely replaced" to defend the core (ibid., p. 133). Negative heuristics of another type tell us to avoid certain paths of research for different reasons: do not be led astray by alternative research programmes, friendly as these may seem in disguise.

Within the research programme of generative linguistics, a prevailing negative heuristic of the latter type tells us to shun according communication linguistic prominence. This methodological directive flies in the face of what used to be the wisdom of students of language. For many linguists, psychologists and philosophers it is still a truism to say that natural language is a system of communication. Roman Jakobson, for example, in a 1970 paper entitled "Language in relation to other communication systems" (our italics), considers the relation between natural language and communication so basic as to warrant a delimitation of linguistics:

"The science of language investigates the makeup of verbal messages and of their underlying code. The structural characteristics of language are interpreted in the light of the tasks which they fulfill in the various processes of communication, and thus linguistics may be briefly defined as an inquiry into the communication of verbal messages." (p. 3) Similarly, a recent introduction to the philosophy of language has on the top of the list of "the most salient features of language" the observation that "[a] language is a system for communicating information between ourselves and others..." (Devitt and Sterelny, 1987, p. 4).

These philosophers have not been guilty of disregarding the obvious: Third on their list of "the most salient features of language" is what they call "private uses": "we talk to ourselves, we write for our own future benefit." (ibid.) These examples of "private use" of language involve two points of view, albeit of the same person. It would be reasonable to assume that according to the suggested depiction of language as "a system for communicating information", those "private uses" of language should also involve communication, even if it is of a peculiar type: In a "private use" of language, a person talks to oneself, as if one talked to another participant in a dialogue, who happened to be oneself.

Subscription to the view that natural language is a system of communication is often nothing more than an expression of some crude intuition, hardly meant to be of any theoretical significance. However, sometimes this view has underlain subtle theoretical moves. A famous example is the Gricean theory of meaning, whether in his own version of it or in Strawson's or McDowell's.

An interesting example of another nature is the case of a certain trend within developmental psychology. Consider, for instance, the following seemingly innocent definitions used by Elizabeth Bates and her associates, in a study of "intentionality prior to and at the onset of speech"

"...[F]or present purposes the imperative is defined as the use of the adult as the means to a desired object. Conversly, the declarative is defined as the use of an object (through pointing, showing, giving, etc.) as the means to obtaining adult attention." (Bates, Camaioni and Volterra, 1975, p. 115; cf. Ochs's remarks in Ochs and Schieffelin, 1983, p. 189).

Those definitions paved the way to the discovery of what have been called "proto-imperatives" and "proto-declaratives", prior to the onset of speech. Such apparent discoveries would not have been possible, had it not been assumed that both natural language and pre-verbal gestures are related systems of communication. Actually, this assumption played a major role in the study of language development, during "the fifth day" of its creation, to put it in terms of a parable used by R. M. Golinkoff and L. Gordon, in their 1983 paper "In the Beginning Was the Word: A History of the Study of Language Acquisition." During the fifth day, some notions of pragmatics were introduced into the field, by Jerome Bruner and others, causing

"language acqisition to be thought of as embedded in a social and cultural context... Increasingly, as researchers moved back earlier and earlier in the child's life in an attempt to bridge the transition from preverbal to verbal communication, the study of language merged into the study of communication." (ibid., p. 7) Consequently, the sixth day is being devoted by some to an attempt at reinstating a clear distinction between language and communication and then reshaping the study of language acquisition.

Thus, the methodological shunning of according linguistic prominence to communication is neither self-evident nor insignificant. It would, therefore, be interesting to compare the research programme of generative linguistics with competing ones on that score.

We turn, then, to some observations on the independence of language in some respects. However, before we do that in any detail we have to briefly clarify the notion of "independence" of language from communication. It is not our intention to defend a view which could be naturally interpreted as dissuading anybody from a proper inquiry into communication as done with words. Indeed, the objectives of some overarching theory of language will be to show and explain how several linguistic competences are exercised, for instance, in thought and in communication.The objectives of another theory are to specify and explain talk-in-interaction, including communication in a strict sense as one form of talk-in-interaction.®FN1¯ However, the existence of neither of these theories would render language dependent on communication in a the following sense of the term. A thesis of "independence" of some competence from communication is a negative answer to the question whether primitive notions of a theory of communication are indispensable to adequate description and explanation of that competence.

Such independence of grammatical competence from communication has been defended within the research programme of generative linguistics in different ways. Chomsky's Reflections on Language defends "the thesis of autonomy of formal grammar" against claims made by John Searle and others, in the context of the philosophical tradition of analyzing meaning in terms of communication-intentions. Frederick Newmeyer's Grammatical Theory includes a defense of the same thesis against attempts made by some linguists to show that grammar is grounded in communication- or discourse-based principles.

An additional argument rests on two intriguing case studies. John was 3 years old. His referral to a certain medical center

"was based upon his total failure to speak in the nursery school either to teachers or to other children. His manner, though, was pleasant, and he smiled readily and appropriately when something pleased him... [H]e spontaneously displayed normal levels of skill in the non-verbal sphere." (Blank, Gessner and Esposito, 1979, p. 330)

A careful analysis of John's behaviour yielded the following results: First, John's verbal productions were "within the range of his peers" (ibid., p. 338), but, secondly, most of John's (recorded) utterances

"offered inadequate responses... He seemed to show no interest or ability in accommodating his comments to the other participant's utterance. Rather he simply used the verbalization as a stimilus to impose his own -- generally unrelated -- utterance." (ibid., p. 344).

Thirdly, "John appeared to evidence little if any interest or skill in preverbal or a-verbal communication... His troubles were graphically demonstrated with the pointing gesture. He himself never pointed when referring to an object, and he appeared totally bewildered if his parents pointed when trying to draw his attention to an object." (ibid., p. 347).

John's parents "could not recall his ever having responded to peek-a-boo or pat-a-cake, nor ever having pointed or waved bye-bye. (ibid., p. 331). The case of John is, indeed, a case of a child in whom syntax and communication "are markedly disparate" (ibid., p. 330), to quote the authors who aptly entitles their report of this case study "Language without Communication". Manifestation of a capacity for verbal communication is, then, not a necessary condition of the presence of grammatical competence. Is the presence of verbal communication a sufficient condition of the presence of grammatical competence?

A clear negative answer emerges from the famous case of Genie, as described by Susan Curtiss (in her 1977 and 1982). Genie started acquiring her first language as a teenager, undergoing extreme isolation and deprivation until she was 13.5 years old. Her ensuing language development resulted in a "markedly uneven [linguistic] profile" (Curtiss, 1982, p. 290). Her grammar is very poor: "Sick people lady driving ambulance" would be the expression of the observation that a lady is driving sick people in the ambulance. Moreover, Ginie is very limited in using certain linguistic forms as communicative devices. For example, she produces no vocatives or grammatically marked questions and has no topicalization or focusing devices save repetition (ibid., p. 286). On the other hand, Genie is a powerful "nonlinguistic communicator. She has, for example, well-developed use of gesture, facial expression, eye gaze, attention-getting devices, and turn-taking knowledge." (ibid., p. 286f). Thus, Genie is an example of a person in whom grammatical competence and communication are markedly disparate in another way. The cases of John and Genie show us that the grammatical competence is dissociated from manifestation of a capacity for verbal communication. We have presented these cases in some detail, not only because of their contribution to the defense of the autonomy thesis of grammatical competence, but also because of their possible import in a related context, viz. in examining the extent to which the pragmatic competence is also independent from communication. We turn now to a discussion of this independence problem.

IV. PRAGMATICS AND COMMUNICATION

We take it for granted that there is a pragmatic competence, which is a system of knowledge related to acts of sentence-use. It is assumed that by having acquired a pragmatic competence a speaker has mastered a family of sub-systems, each governing acts of a certain type, such as assertion, request or advice. We also assume that each of these sub-systems takes the form of rules which constitute conditions for appropriate use of certain sentences. Different systems of rules govern "speech acts" of different types.

Independence from communication of the pragmatic competence requires independence from communication of that family of sub-systems, each governing speech acts of a certain type. Since various types of speech act seem independent of each other, for example - asking and advising, it would be only natural to raise a separate problem of independence from communication with respect to each type of speech act.

It is not clear how to go about establishing dependence of some type of speech act upon communication. A constitutive requirement that an appropriate addressee should be present at the context of use seems a necessary condition of such dependence. Let us call this condition "the addressee condition".

Notice, first, that this necessary condition is clearly not a sufficient one. Thus, in a certain "semantical analysis of English illocutionary verbs", presence of an appropriate "hearer" is required by the rules which govern assuring, threatening, urging, congratulating and a few other acts. However, communication, in the pretheoretical sense of passing on information, is hardly the point of any of these acts.

Secondly, since the addressee condition is a necessary condition of dependence of a speech act type upon communication, its denial is, of course, a sufficient condition of independence of that speech act type from communication. Hence, if we show that for a certain type of speech act the addressee condition does not hold, we thereby show that our notion of speech act in general is independent of communication, and consequently, that the pragmatic competence is independent of communication.

Assertion is our prime example of a speech act type which does not satisfy the addressee condition. Events of putting sentences in the indicative to an assertoric use do not have to involve an addressee. Right now, when I am trying to write this paper, I am trying to express my own defensible thoughts, to present my justifiable views, to put forward some of my supportable representations, and I am doing it by using various sentences. However, I am not addressing anybody else, nor do I address myself, in the sense of bearing to myself the relations I would bear to some interlocutor. There seems to be no reason to analyze 'I put forward some of my (supportable) representations by using [sentence] S`, 'I present my (justifiable) views by using S` or 'I express my (defensible) thoughts by using S` in terms of 'I address S to person X` and 'X = me`. Many cases of the so-called "private uses" of language do, indeed, involve making assertions addressed to nobody.

This observation seems fairly obvious, though it has been often disregarded. Among those who failed to notice it one finds Russell, who claimed in An Inquiry into Meaning and Truth, that

"[i]n adult life, all speech... is, in intention, in the imperative mood. When it seems to be a mere statement, it should be prefaced by the words 'know that'. We know many things, and assert only some of them; those that we assert are those that we desire our hearers to know." (1940; 1962, p. 24)

The most interesting counter train of thought was suggested by R. G. Collingwood. In his autobiography he mentions "a principle of logic" which he found it "necessary to state", viz.

"the principle that a body of knowledge consists not of 'propositions', 'statements', 'judgments'..., but of these together with the questions they are meant to answer; and that a logic in which the answers are attended to and the questions neglected is a false logic." (Collingwood, 1939, p. 30).

Moreover, in order to find out what a person has meant, having "spoken or written with perfect command of language and perfectly truthful intention", says Collingwood, "you must also know what the question was (a question in his own mind, and presumed by him to be in yours) to which the thing he has said or written was meant as an answer." (ibid., 31)

According to this view, questions are, in a sense, more basic than assertions. An adequate analysis of an assertion renders it an answer to a certain question. If this is correct, it eliminates our prime example of a speech act which is governed by rules which make no allusion to an addressee: If every assertion rests on some question and every question constitutively requires the presence of an addressee, then every assertion involves the presence of an addressee.

Our present problem is, then, which are more basic, questions or assertions? Formal analyses of questions have not settled this problem. On the one hand, some logicians and linguists have analyzed questions in terms of assertions. For example, within the framework of Hamblin, 1973, an answer is, indeed, a proposition, and questions are defined as sets of possible answers, i.e. as sets of propositions. Similarly, in Karttunen, 1977, a question is a property of propositions, the property of those propositions which correspond to true answers.

On the other hand, some completely different analyses have also been proposed. Peirce suggested that questions are requests for information, and requests are imperatives of a special kind. More recently, Aqvist, since his 1965, and Hintikka, since his 1974, have tried to show how to capture this idea in an appropriate formal way. From the present point of view, imperatives seem to be on a par with interrogatives, because they also seem to constitutively require the presence of an addressee. However, such formal analyses of questions in terms of imperatives could not settle the issue of whether questions are more basic than assertions or not. To see why Collingwood's suggestion to consider questions as more basic than assertions should be rejected, compare it with the Aristotelian approach (in Topics), according to which what has been said has to be seen against the background of the question under discussion and the answers already given. Whereas for Collingwood the question in the background determinesthe meaning of what has been said, according to the Aristotelian conception, that question is related to the purpose which has been intended to be served by uttering what has been said. The distinction is significant: according to the former view a sentence has any meaning, only if it is considered against the background of a certain question; without such a question in the background it cannot be understood at all. According to the latter view a sentence has its meaning in the langauge to which it belongs and can be understood as such. It is, rather, an act of utterance that cannot be understood unless its purpose is reckoned. A question already asked and answers already given enable one to clarify the issue at hand, as well as the purpose for which the sentence has been used.

The Aristotelian view seems compelling. The alternative view rests on a confusion of understanding an act of using a sentence and understanding this sentence itself. Once this distinction is noted there seem to be no grounds for the suggestion that questions are more basic than assertions.

Notice, furthermore, that assertions can be felicitously performed against the background of no question at all. An unprompted utterance of a sentence, used in an appropriate assertoric way, can introduce into a conversation a new topic, thus serving a new purpose. And then, sometimes question arise only against the background of some assertions. A process of argumentation, for example, is usually initiated by a challenge to some proposition. Such challenges often take the form of a question and they "may be motivated by a variety of propositional attitudes: puzzlement, doubt, skepticism, rejection, devil's advocacy,..." (Blair and Johnson, 1987, p. 45), all directly related to some background assertions.

Having rejected Collingwood's view of assertion, there seems to be no reason to assume that the speech act type of assertion involves the addressee condition. It would, therefore, be reasonable to consider assertion as conceptually independent of communication. Consequently, the pragmatic competence, as such, could also be reasonably held to be independent of communication. Is assertion the only type of speech act which does not involve the addressee condition? Somewhat surprisingly, perhaps, both questions and commands seem to be explainable in terms of systems of governing rules which also do not satisfy the addressee condition. In a nutshell, the idea is to analyze an act of posing a question as an act of raising a problem and to analyze an act of issuing a command as an act of invoking a norm.

An act of raising a problem does not require for its felicitousness the presence of another person. Raising problems in soliloquy is very common. To be sure, one's raising a problem in the presence of some other person may result in the latter's providing the former with a solution of the raised problem. Social prudence has encouraged cooperation under the circumstances of problem raising. Actually, cooperation under such circumstances has been entrenched, to the extent that most often it is being taken for granted that it is part of a one's linguistic knowledge that questions have to be answered, if possible. This commonly held view about the linguistic nature of questions and answers is, however, wrong. Linguistic knowledge determines only the problem raised by an utterance of a question and the induced space of possible answers. Understanding a question amounts to figuring out the related problem and the induced space of possible answers. Answering a question is basically pointing out one of the solutions of the raised problem. Having to answer a question, in case one knows how, means having to point out a solution of a problem that has been raised in one's presence, in case one knows how. Since answering a question is also making an appropriate assertion, answering a question is pointing out what one takes to be a solution of the problem raised by posing the question. Having to answer a question, in case one knows how, means, therefore, having to solve the raised problem and to point out what is held to be a solution of it. However, having to solve problems that have been raised in one's presence is, clearly, not a matter of linguistic rules, but rather of social policy or social norms of cooperation in general.

Similarly, an act of invoking a norm does not require for its felicitousness the presence of an another person. Invoking norms in soliloquy, for instance, when one is driving alone, is trying to solve a Chess problem in the presence of nobody else or is praying, is not a rarity either. Of course, one's invoking a norm in the presence of some other person may, under appropriate circumstances, result in the latter's following of the norm the former invoked. Certain social institutions require that, under appropriate circumstances, norms invoked by a speaker be followed by the addresssee. Again, obedience under such circumstances has been entrenched, to the extent that most often it is being taken for granted that it is part of a one's linguistic knowledge that commands should be obeyed, if possible. And again, this commonly held view about the linguistic nature of commands is, however, wrong. Linguistic knowledge determines only the norm invoked by an utterance of a command. Understanding a command amounts to identifying the related norm. Now, obeying a command is following the norm invoked by it. Having to obey a command, in case one knows how, means having to follow a norm invoked in one's presence, in case one knows how. However, having to follow norms, as invoked under certain circumstances, is also not a matter of linguistic rules, but rather of some general social policy or certain institutional norms.

Raising a problem with words and invoking a norm with words do not require the presence of another person. Do they require the presence of oneself as both a speaker and an addressee? As long as one carefully avoids begging the question of language and communication, it seems one has no reason to analyze 'X raises a problem` as derived from 'X addresses a problem to Y` and 'Y = X`. Such an analysis would conflate characterization of an act with characterization of common reasons for performing it. The reasons for raising a certain problem or invoking a certain norm, under some circumstances, may well involve an addressee, either someone else or oneself, but specifying those reasons goes beyond an adequate description of the act itself.

We have, then, argued that assertions, questions and commands do not involve for their linguistically appropriate performance, neither the presence of another person nor oneself playing the role of an interlocutor. These observations lead us to the conclusion that the pragmatic competence, as such, is independent of communication.

Dummett has recently argued that "cases in which we employ language with no direct, and sometimes no remote, communicative purpose... prove nothing", because it is generally agreed that language is both a vehicle of thought and an instrument of communication (1989, p. 201). However, our arguments to the effect that assertions, questions and commands are not, as such, acts of communication involve more than examples of language used with no communicative purpose. The suggested general analyses of assertion, question and command, as not involving the addressee condition, explain why language is not primarily an instrument of communication, though it can be put to use both in thought and in communication.

On the grounds of our conclusion, that it is reasonable to hold the pragmatic competence, as such, independent of communication, it would be natural to predict the existence of a psychological or even a neuropsychological counterpart of this dissociation. In other words, it would be plausible to predict that a certain brain damage would result in a loss of the capacity to communicate, leaving intact, say, the capacity to assert, or vice versa.

Actually, the above-mentioned cases of John, the child with "language without communication", and of Genie, the girl deprived of a linguistic environment until she was 13.5 years old, seem to provide us with the expected examples, but a thorough examination of the data is still required, on grounds of some articulated theories of the pragmatic competence and of communication, before that prediction could be deemed fully confirmed by the evidence.

This concludes our discussion of the "negative heuristics" of the research programme of generative linguistics. In the next parts of this paper some "positive heuristics" will be considered.

V. MODULARITY OF PRAGMATICS

A research programme tells us not only what paths of research to avoid, but of course also what paths to pursue. The research programme of generative linguistics tells us to try to explain cognitive phenomena in terms of "modules", separate systems of knowledge, each with its own properties. The principles of a module specify its content as a system of knowledge, but they also enter into the processes of acquisition of this knowledge and of its use. This modular approach seems to be a distinguishing feature of the research programme of generative linguistics (see Newmeyer 1983, pp. 2-3), and it would, therefore, be interesting to see what happens when we pursue this modular path in our study of the pragmatic competence.

The question whether there is a pragmatic module seems a natural one to raise and answer on grounds of our present knowledge of language use, but the issue is actually far from being settled. The reason is simple: one should not hasten to accept a distinct answer to an indistinct question. We cannot set ourselves to look for a "pragmatic module" without having in mind some notion of "pragmatics" and some notion of a "module". Apparently conflicting answers that have been given to this question involve, as a matter of fact, different delimitations of pragmatics or different characterizations of modularity.

Leaving aside, for a moment, the question whether pragmatics is embodied in a separate module or not, consider, first, claims that have been made to the effect that pragmatics is embodied in the right hemisphere of the brain (of right-handed persons). Sheila Blumstein compares the role of the right hemisphere processing to that of the left one:

"While the left hemisphere appears to be specialized for processing the linguistic grammar per se... the right hemisphere contributes to the "pragmatics" of language use, on the one hand, and the integration of the linguistic grammar with cognitive (nonverbal) processes, on the other" (1981, p. 250).

Although it has already been established that the right hemisphere has considerable linguistic ability (see, e.g., Eran Zaidel's survey "Language in the right hemisphere", 1985), the idea that the right hemisphere controls the pragmatic competence is somewhat puzzling, if according to one's general conception of language, ordinary uses of sentences involve both the grammatical and the pragmatic competences. Common use of sentences by subjects with complete cerebral commissurotomy would involve mysterious, minute coordination on the part of these two informationally disconnected hemispheres. How could the left hemisphere produce a genuine speech act, that involves an utterance of a sentence, in a context held to be appropriate, if the right hemisphere controls the pragmatic aspects od a speech act, while the left hemisphere controls the syntactic ones? In order for the left hemisphere to be in a position to produce a felicitous utterance of a sentence, it must be informed that the context of utterance is held to be appropriate. However, if pragmatics is controlled by the right hemisphere, then the information that the context is held appropriate is at the right hemisphere. Hence, in order for the left hemisphere to be in a position to produce a felicitous utterance of a sentence it needs information which is at the right hemisphere. Under normal conditions information can flow from one hemisphere to the other, but in subjects with complete cerebral commissurotomy this is impossible. How could they, then, produce felicitous speech acts? This is "the puzzle of pragmatics in the right hemisphere". The puzzle disappears when the evidence is carefully scrutinized.

What is the main evidence? Howard Gardner, Hiram Brownell and their associates found that right-brain damaged subjects have deficits in interpreting metaphors and idiomatic expressions, in understanding sarcastic or humorous texts as well as cartoons, and in retelling stories and getting their point. Claus Heeschen, Nancy Foldi and others discovered that right-brain damaged subjects have severe deficits in understanding the so-called "indirect speech acts". For example, when such subjects encounter an interrogative sentence commonly understood as an "indirect request", they would consistently respond to the literal interpretation rather than fulfil the intended request.

All these are highly interesting phenomena and many of the experimental results seem to be of significance not only within neuro-psycholinguistics, but within linguistic and philosophical studies of pragmatics as well. However, with the view of defending the hypothesis that the pragmatic competence is embodied in the right hemishpere, perhaps in a right hemisphere module, the evidence seems rather weak in several respects. First, if the hypothesis of pragmatics in the right hemisphere were the best explanation of all those deficits, found in right-brain damaged subjects, one would have expected that in a case of complete cerberal commissurotomy the disconnected left hemisphere would manifest the same deficits and would perform on a chance level.

This is not what happened, when the Right Hemisphere Communication Battery of Gardner and Brownell, was administered to four split-brain subjects (Zaidel and Kasher, 1988; Spence, Zaidel and Kasher, 1989). This battery includes tests related to humor, prosody, indirect requests, metaphors, sarcasm and several other phenomena. Although these four subjects were performing significantly worse than controls, there was no test in which all the four performed on a chance level. This result suggests that none of the tests in the battery is related to an exclusively right hemisphere phenomenon.

Moreover, whereas right-brain damaged subjects show severe deficits in all the subtests, one of the split-brain subjects, LB, has shown severe deficit only in a few of them, viz. story retelling, interpretation of indirect requests and prosody. Only mild deficits have been found in LB's performance on other tests, including pictorial metaphors and understanding sarcasm. These results show that even on an understanding of "pragmatics" as being related to humor, emotion, non-literal language and integrative processes, the pragmatics in the right hemisphere hypothesis should be significantly revised or perhaps even replaced by an utterly different one.

Secondly, notice that most if not all of these deficits involve phenomena which are hardly understood on any abstract explanatory level. In the absence of adequate theories of metaphor, sarcasm and humour, for example, it is not clear what actually is being tested when any of the related tests is given to a subject. Moreover, without better theoretical understanding of each of these phenomena there seems to be no good reason for combining them to form a separate competence. Again, an hypothesis that a pragmatic competence operates in the right hemisphere cannot be grounded on results pertaining to such a heterogeneous cluster of such poorly understood phenomena.

Thirdly, even if we do cluster such phenomena under some notion of a pragmatic competence, the problem arises of the relation this competence bears to various families of phenomena which have often been investigated under the heading of "pragmatics", for instance, basic speech acts and performatives, conversational and conventional implicatures, politeness and the like.Are these phenomena additional members of the same cluster, or are they rather related to another competence, and, in any case, why? In other words, the suggested affirmative answer to the question whether a pragmatic competence exists, seems to rest on some diffused, unwarrantable notion of pragmatics. The question whether there is a pragmatic module, in a different, more justifiable sense of "pragmatics", remains, therefore, in want of an appropriate answer.

Notice that a sensible delimitation of what should be understood under "pragmatics" will bring us closer to such an answer mainly because it is going to render clearer the question itself. However, without some clarification of the notion of "module" as well, the question is going to remain hopelessly ambiguous. Consider, for example, Fodor's notion of "module" in his Modularity of Mind (1983, pp. 47-101). Input systems are modular, in Fodor's view, since they are domain specific, mandatory, informationally encapsulated, "exhibit characteristic and specific breakdown patterns" (ibid., p. 99) and have several other properties. It is a "main thesis of [Fodor's] work that the properties in virtue of which input systems are modular are ones which, in general, central cognitive processes do not share."

If a module is an input system which has the nine properties Fodor enumerated, then there is no pragmatic module, in the sense of a module which embodies the knowledge which we commonly find under "pragmatics". Conditions of appropriate use of sentences involve much more than a syntactic analysis of an input sentence and a perceptual analysis of some contextual features. They also involve institutional analysis and thereby allude to beliefs entertained by the user at the context of use. Many of these beliefs belong to some central cognitive system rather than to any input device. Recall, for example, beliefs which are mentioned in the rules constitutive of promising, such as "S believes H would prefer his doing A to his not doing A" (Searle, 1969, p. 58). Thus, conditions of appropriate use involve beliefs to which input systems do not have access, which shows that the pragmatic competence is not embodied in an input system.

If a module is a system which has the properties Fodor enumerated, whether it is an input system or not, there still is no pragmatic module, in the sense of a module which embodies the knowledge used in deriving conversational implicatures. I have argued elsewhere that Grice's maxims follow from some rationality "most effective, least effort" principle (Kasher, 1976 and 1982), which applies to all intentional acts, including acts of language use as well as artistic acts. Accordingly, a cognitive system which derives conversational implicatures involves application of a general, most probably central principle to the output of some linguistic system. Hence, it would be implausible to assume that some domain-specific cognitive system produces conversational implicatures. Moreover, as Grice demonstrated, such implicatures can be cancelled ("John arrived and Bill left, but not necessarily in this order"), which makes it implausible to assume that some mandatory cognitive system produces them.

Similarly, understanding a so-called "indirect speech act" involves an attempt to understand a use of sentence as meant to attain more than one end. First, there is the "literal end", the one determined by the rules governing the literal use, but then, this end is understood as an intermediate one, a sub-goal, and an attempt is made to find another end, the attainment of which would be served by attaining the former, literal one. Indeed, under quite ordinary circumstances, such an attempt requires access to one's general system of beliefs. Thus, no informationally encapsulated cognitive system could embody the knowledge required for understanding indirect speech acts.

Since the property of being informationally encapsulated seems to be one of the major ingredients of the idea of modularity, the so-called indirect speech acts are not understood by a pragmatic module, in any interesting sense of the term "pragmatic module".

We conclude this discussion by making three general remarks about these observations.

First, the modular approach to the study of mind in general and language in particular has given rise to a new form of delineation problem. Individuation of a cognitive domain will often involve much more than finding the "right" category in some arbitrary or useless taxonomy. It will rather rest on an explanatory characterization of a related module, a cognitive system which has some required properties, including being domain-dedicated and informationally encapsulated.

Secondly, as a result of applying this modular approach to the study of language use, a new conception of pragmatics seems to have emerged. There seems to be room for a division of whatever has been labelled "pragmatics" into the following parts. (a) What we would like to dub "core-pragmatics", which consists of the family of systems of rules which govern basic speech acts. Core-pragmatics is autonomous, though whether it should be described as a module or not, depends on the notion of "module" one would apply to it. It is reasonable to assume that core-pragmatics is domain-dedicated, given the special nature of the systems of rules which govern speech acts, namely their being constitutive systems of rules. Whether core-pragmatics is informationally encapsulated or not seems to depend on the delimitation of the class of speech acts. If it is confined to speech acts which are syntactically marked (in ways yet to be understood), core-pragmatics seems to be encapsulated. Being both dedicated and encapsulated it is, in some interesting sense, a pragmatic module. However, if core-pragmatics is taken to include, say, promises, then, as has been pointed out earlier, one's general system of beliefs turns out to be invloved in understanding and production of speech acts of core-pragmatics. This would mean that core-pragmatics is not encapsulated.

(b) "Amplified core-pragmatics", which consists of all the systems of rules governing "things done with words" which are not basic speech acts. "Basic speech acts", such as assertion, questions and commands are distinguished from "things done with words", many of which, such as baptizations and acquittals, are not basic speech acts. Whereas knowledge of the former is part of knowledge of language, knowledge of the latter is social or institutional rather than linguistic in nature. Clearly, amplified core-pragmatics is not informationally encapsulated. The "central" system of one's beliefs often "penetrates" processes of understanding non-basic speech acts. For instance, whether production of a certain sentence in a certain context of utterance is understood as a speech act of acquittal or not depends on one's institutional knowledge in a way one's understanding a similar event as a speech act of assertion does not.

(c) Then, there is a "central pragmatic" system, which involves the application of some general rules and strategies, such as those governing rational intentional action in general, to cases of speech activity. This is how (and, in a sense, where) conversational implicatures are generated.This is also where politeness considerations participate in speech activity. Whether such a "central pragmatic" system is a module, in some interesting sense, is an interesting open question.

(d) There is reason to assume that alongside the "central pragmatic" system, which involves principles of intentional action in general, there a special system of rules governing basic aspects of "talk-in-interaction", such as organization of turn-taking, organization of sequences and organization of repair. Studies of these types of action seem to lend plausibility to the hypothesis that the processes involved constitute a separate module, even in a rich sense of this term.

(e) In addition to these parts of pragmatics, there seems to exist an additional class of various "interface" features, to be dubbed "interface pragmatics". Understanding indexicals, for example, involves the integration of the output of a language module with the output of some perception module, both serving as input for some "central" unit which produces the integrated understanding of what has been said in a context of utterance.

(f) Finally, one may introduce "extended pragmatics" which would include, for instance, some of the areas tested by the above-mentioned Right Hemisphere Communication Battery. Thirdly, the idea of a "central pragmatic" system bears an interesting relation with "Descartes's problem", of "how language is used in the normal creative fashion" (Chomsky, 1988a, p. 138), a problem which, to use Chomsky's arguments and terms, is unsolvable, a mystery.

The more we explain pragmatic facts in terms of a general intentional action theory as applied to instances of language use, the more we become closer to solving parts of "Descartes's problem". Creative use of language can be factored into (a) creative choice of ends and (b) rational pursuit of those ends. Factor (b), of the rational pursuit of given ends, seems to be amenable to explanations in terms of general rationality principles, which are parts of a general intentional action theory. However, factor (a), of the creative choice of ends, does perhaps constitute an unsolvable problem, a mystery.