An On-Line Interview with Noam Chomsky: On the nature of pragmatics and related issues

Brigitte Stemmer

Centre de Recherche, Institut universitaire de gériatrie de Montréal, Canada
and
Lurija Institute for Rehabilitation and Health Sciences at the University of Konstanz,
Kliniken Schmieder, Allensbach, Germany
e-mail: brigitte.stemmer@uni-konstanz.de

When the idea of the special issue on pragmatics was born, so was the idea to invite two scientists who have had a tremendous impact on the fields of linguistics and pragmatics: Noam Chomsky and John R. Searle. Both assured me that the project sounded extremely interesting and that they would love to be part of the project if it weren’t it for the fact that one of them was, meanwhile, diving into different spheres and the other one“committed for a long time to come, with deadlines looming and sometimes passing at a depressing rate.” I was not able to compete with the spheres but thanks to modern technology, looming deadlines could be beaten: some authors, the editor of the special issue as well as the editor of Brain and Language framed some questions which were sent to and readily discussed by Noam Chomsky via e-mail.
 

QUESTION 1: How do you see contemporary clinical neuropsychological research influencing your views on language processing?

N.C.: The question is hard to answer because I have never had any particular views of language processing beyond what seems fairly obvious: that if Jones has the language L, then Jones's language processing accesses L; thus in processing language, I access a variety of English, not Japanese. Here I understand L (for me, some variety of English) to be an attained state of a genetically-determined faculty of language FL (I overlook here irrelevant real world complexities, e.g., the fact that no one is "monolingual" in this sense).
    I follow the processing literature to see what I can learn from it about language processing and about the nature of language, of course with particular questions in mind: e.g., how apparent properties of language are "apportioned" among processing systems and the system L that they access.
 

QUESTION 2: (a) Do you think that language is a module in a strictly Fodorian sense of the term, i.e., as a "stupid" satellite of a general-purpose central machinery, where "real intelligence" would lie?

    (b) Do you think that communication is, or at least could be, a competence in his (or Marr's) terms?

    Both questions relate to the position you might take on the domain-generality vs. domain-specificity debate. It seems to me that your idea of "mental organs" resembles the latter rather than the former position, but you seem to me rather obscure on the whole topic of the architecture of the mind. If I am right, then when you (in a marginal remark in one of your books) talk of "communicative competence", you might mean something similar to what I mean: a mental organ, but an inferential, non-modular one. Else, you would take a Fodorian position, and should therefore conceive of an intrinsically inferential process like communication as performance (i.e., as a task of the central system) rather than competence.

N.C.: Little can be said with any confidence about "the architecture of the mind," and for that reason I (purposely) remain "rather obscure on the whole topic." I have expressed my own (of course tentative) assumptions many times, for example in the opening parts of _Reflections on Language_ (1975, RL). The assumption here is that the architecture of the mind is modular in the sense that Randy Gallistel recently described as "the norm these days in neuroscience." In Gallistel's words, in all animals learning is based on specialized "learning mechanisms," "instincts to learn" in specific ways, these being essentially "organs within the brain [that] are neural circuits whose structure enables them to perform one particular kind of computation," as they do more or less reflexively apart from "extremely hostile environments"; human language acquisition is instinctive in this sense, based on a specialized "language organ." In _RL_, the "learning mechanisms" are called LT(O,D) ("learning theories" for organism O in domain D); one of them -- the one that particularly concerns me -- is LT(Human, Language), the initial state of the human faculty of language FL, the specialized "language organ" that attains various states under the triggering and shaping effect of experience, e.g., the language L that Jones has.
    We can describe the initial state of LT(Human, Language) as a device that maps experience into state L attained, hence a "language acquistion device" (LAD). The existence of such LAD is sometimes regarded as controversial, or even as having been disproven. If terms are being used in their technical sense, these conclusions amount to saying that there is no dedicated "language module," in which case it remains a mystery why my granddaughter's pet kitten (or chimpanzee, or whatever) doesn't acquire a particular language just as she does, given essentially the same experience. I doubt that anyone believes that; when investigated at all closely, proposals about "general learning mechanisms" that apply to language seem to presuppose extremely rich innate language-dedicated structure, often richer than proposed in familiar linguistic theories (e.g., "irregular verb modules"). I know of no approach to these matters, including the most extreme "radical behaviorist" approaches, that does not presuppose (often tacitly) that a child can somehow distinguish linguistic materials from the rest of the confusion, again presupposing FL, hence LAD.
    The way to make the general assumptions less obscure is to discover the nature of the various specialized "learning mechanisms" -- the systems LT(O,D), in my terminology -- among them the "language organ" FL, the states it can in principle attain, the "neural circuits" involved, etc. That is also the way to arrive at one or another "position...in the domain-generality vs. domain-specificity debate," a very tentative position I would think, given the limits of current understanding. I concede that I don't really understand what this debate is about in the way it is usually waged (without my participation). There are very interesting questions about just what might be specific to human language (part of LT(Human, Language), the dedicated "learning mechanism" that is the "language organ"). These are the topics of inquiry in all study of language and other cognitive systems that I know of. But I do not understand the more general "debate" that seems to arouse much passion.
    It could be that one of the systems that develops, either as a distinct module or a component of others, is the kind of "communicative competence" that enables us to use language coherently and in ways that are appropriate to situations. This seems rather likely, for one reason, because of dissociations that have been discovered (limited communicative competence along with rich language competence, etc.). Whether this system, if it exists, is an "inferential, non-modular one" depends on the facts of the matter: e.g., are the properties of "communicative competence" similar to those of finding our way home when we come upon a detour? does "communicative competence" function independently (or partly independently) of general inferential capacities (if such exist)? There is evidence on these matters, and interesting efforts to organize it systematically, but far too little is understood, as far as I am aware, to take a confident stand. My own personal impression, for what it is worth, is that talk of "general inferential or problem-solving capacities" tends to be rather empty, and that when we investigate actual cases in one or another organism, we find that specific mechanisms are assumed. But that's a matter for discovery, not pronouncements.
    It's perhaps worth adding that one should be careful not to be misled by idiosyncratic informal usages. In particular, in English one uses the locutions "know a language," "knowledge of language," where other (even similar) linguistic systems use such terms as "have a language," "speak a language," etc. That may be one reason why it is commonly supposed (by English speakers) that some sort of cognitive relation holds between Jones and his language, which is somehow "external" to Jones; or that Jones has a "theory of his language," a theory that he "knows" or "partially knows." The systematic ambiguity of the term "grammar" in technical linguistics, though constantly emphasized, may also have contributed to such conclusions, along with the fact that the term "language" of English (and related though typically not identical concepts of other languages) is used in ways that involve sociopolitical, teleological, and normative elements, colors on maps and stability of empires, etc. One should not expect such concepts to play a role in systematic inquiry into the nature, use, and acquisition of language, and related matters, any more than one expects such informal notions as "heat" or "element" or "life" to survive beyond rudimentary stages of the natural sciences.

    Let's turn to questions (a) and (b).
    With regard to (a), the notion of modularity discussed in _RL_ and elsewhere, and explored in studies of FL and particular languages (states of FL, from my point of view), is not inconsistent with Fodorian modularity but is a different notion. Fodorian modularity is concerned primarily with input systems. In contrast, _RL_ modularity is concerned with cognitive systems: specialized "learning mechanisms," their initial states and states attained; and how input (perceptual) and output (language use) systems access the states attained. Whether these input and output systems are modular in Fodor's sense is a distinct question. Though related, the topics and conceptions of modularity are different.
    As Fodor puts the matter in his _Modularity of Mind_, "the perceptual system for a language comes to be viewed as containing quite an elaborate theory of the objects in its domain; perhaps a theory couched in terms of a grammar of the language." I would prefer to rephrase this by saying that the perceptual system for Jones's language L (a state of FL) accesses L. Theories of L (and FL) are what the linguist seeks to discover; adapting traditional terms, the linguist's theory of Jones's L can be called "a grammar of L," and the theory of FL can be called "universal grammar," but it is the linguist, not Jones, who has a theory of L and FL. Jones has L, but no theory of L (except what he may believe about the language he has, which has no privileged status, any more than what he may believe about his visual system or problem-solving capacities.
    When we look more closely, we see that more is involved here than choice of terminology, but let us put that aside. Clearly the notions of modularity are different, as are the questions raised, though they are not incompatible, except perhaps in one sense: FL and L appear to be "central systems" in Fodor's framework, but the assumption of _RL_ (and Gallistel's "norm of neuroscience") is that they are distinctive components of the central "architecture of mind," so that the "central systems" would not be unstructured -- what Fodor calls "Quinean and isotropic," andwhat Tirassa, if I understand him, calls "inferential, non-modular."
    In any event, the concept of modularity I have discussed for the past 40 years (e.g., in _RL_) is definitely _not_ one that takes language (FL and its states) to be ´a "stupid" satellite of a general-purpose central machinery, where "real intelligence" would lie.' That would be a category mistake, conflating the perceptual system (the Fodorian input module) with the cognitive system L it accesses (what Fodor calls the "theory of the objects in its domain." I think this is explicit in _RL_ and elsewhere, and I hope it is put as clearly as our limited understanding warrants.
    Turning to (b), communication is an action, not a competence (whether the term "competence" is being used in its technical or its informal sense). It's possible that a form of "communicative competence" underlies the use of language for communication (one of its many uses). As for Marr's famous three levels of analysis, he was concerned with input-output systems (e.g., the mapping of retinal images to internal representations). Language is not an input-output system. Accordingly, Marr's levels do not apply to the study of language, though one could adapt them to the very different problem of characterizing cognitive systems accessed in processing and production.

QUESTION 3: What do you think of neurolinguistics these days?

N.C.: I share the general hope that new non-invasive technologies will make it possible to circumvent, at least in part, the barriers to direct experiment. These include the obvious ethical barriers, and the apparent biological isolation of the human faculty of language in essential respects, which has rendered animal experimentation pretty much irrelevant to the study of language (even putting aside ethical barriers); to quote Gallistel again, "we may never be able to study the neural basis of language perception in animals [let alone language as a cognitive system accessed in language perception] for the same reason that you cannot use a mammal to study the neural basis of polarization perception." New technologies hold considerable promise, one hopes. That aside, more traditional neurolinguistic inquiries continue to yield intriguing results. To mention one recently published example, Yossi Grodzinsky and Lisa Finkel report finding a distinction between dislocation of phrasal and of lexical categories in agrammatic Broca's aphasia, a result that could relate to (still unpublished) ideas proposed at the borders of inquiry into the nature of language. That's just one case, which happens to interest me particularly.
    More generally, I don't see any principled way to distinguish linguistics (meaning, the branch of the study of language we are discussing here) from neurolinguistics, any more than one can distinguish chemistry from physical chemistry in principle. These may be useful distinctions for temporary purposes, but one looks forward to erosion of such boundaries as understanding progresses. My own view has always been that the part of the study of language relevant here is in principle part of human biology: "biolinguistics," as some have called it.
 

QUESTION 4: (a) Levinson, in his book on pragmatics (1983), discusses the role of pragmatics within linguistic theory. He argues that a general linguistic theory must incorporate pragmatics as a component or level in the overall integrated theory: "In order to construct an integrated theory of lingusitic competence, it is essential to discover the logical ordering of compoentns or levels. For example, Chomsky has elegantly argued that syntax is logically prior to phonology, in that phonological descriptions require reference to syntactic categories, but not vice versa; syntax is thus AUTONOMOUS with respect to phonology, and phonology (non-autonomous with respect syntax) can be envisaged as taking a syntactic input, on the basis of which phonological representations can be built up. Accepting for a moment this kind of argument, the question is, is it possible to argue that there is some accepted component of grammar that is non-autonomous with respect to pragmatics (i.e. some component requiring pragmatic input)? If so, pragmatics must be logically prior to that component, and so must be included in an overall theory of linguistic competence." (p. 34).

 What is your view on Levinson's argument?

 

    (b) Do you think it is worthwhile to postulate a "neuro" aspect of pragmatics?

N.C.: As for (a), perhaps I should begin with some terminological/conceptual clarification. My own view has always been stronger than what you quote from Levinson: "a general linguistic theory must incorporate pragmatics" not only "as a component or level in the overall integrated theory," but as a central and crucial component (for irrelevant terminological reasons, I wouldn't call it a "level" in the technical sense).
    My first (non-)publication on these topics is a very long manuscript called _Logical Structure of Linguistic Theory_ (1955, revised 1956 version published in part in 1975). Here, and in everything else I've written as far as I can recall, I used the term "syntax" in its traditional (Peircean, Fregean, Carnapian, etc.) sense, and assumed a kind of "use theory of meaning" (influenced at the time by the later Wittgenstein and John Austin particularly). Thus by "syntax," I mean the study of symbolic systems, including whatever computational/representational systems we take to be internal to the mind/brain. That includes formal relations among the elements of these systems (e.g., rhyme and entailment, insofar as these are formal relations among internal symbolic objects), model-theoretic semantics (insofar as the models are considered to be internal objects, i.e., "mental models" -- as in practice they are, in my opinion, contrary to what is often asserted), formal semantics based on a relation R (sometimes called "reference") holding between symbolic objects (e.g., between "London" and its "semantic value," not an entity in the world, or even the world as we conceive it to be, but of some internal system of thought that is itself related to the world), etc. Continuing to take Jones's language L to be a state of FL, Jones uses the syntactic objects of L in a variety of ways: internal thought (statistically by far the most prevalent use, I suppose), maintaining personal relations, communication, telling stories, etc. The study of such uses of language is "pragmatics," in a conventional terminology. Whether there is also a semantics of natural language in the traditional sense, based on a true reference relation between symbolic objects of L and external objects (in the real world, or the external world as conceived and imagined) seems to me an open question. I have always been rather skeptical, for reasons discussed elsewhere. In any event, it is a matter for discovery, not stipulation.
    From this point of view, "pragmatics" must be a central component of any linguistic theory that aims to be comprehensive.
    To continue with terminology, I've used "syntax" in both a narrow and a broad sense. The broad sense is the traditional one, just mentioned. The narrow sense is theory-internal. Suppose we postulate that oe part -- call it NS -- of the internal syntactic computational- representational) system of FL constructs syntactic objects that are mapped to "phonetic form" (accessible to sensorimotor systems) and to "logical form" (accessible to the systems of thought, conceptual-intentional systems). Then "narrow syntax" is the study of NS. That's one usage, which (purposely) leaves open many important questions, to be answered by empirical inquiry: e.g., where does determination of quantifier scope or anaphoric relations enter into the broad syntax? These empirical assumptions do hold that narrow syntax is "prior to phonology," in that the objects constructed independently by NS are mapped to phonetic form by a component of the syntax that we may call "phonology" (in a very general sense, including morphological processes, and -- in my opinion -- operations that have the properties of grammatical transformations). But these are empirical questions, which are quite lively and controversial. No one can sensibly regard them as matters of doctrine.
    Suppose (as the question suggests) that we tentatively adopt empirical assumptions of this nature. Then let us consider the question raised: "is it possible to argue that there is some accepted component of grammar that is non-autonomous with respect to pragmatics (i.e. some component requiring pragmatic input)," in which case "pragmatics must be logically prior to that component, and so must be included in an overall theory of linguistic competence"? I suppose it is possible to argue that the computational/representational system accesses features of language use, though what such a system would look like, I have no idea. Suppose, for example, we consider the (plainly correct) fact that in a linguistic interchange, new/old information is a matter of background that participants assume to be shared (what is sometimes misleadingly called "discourse"; there need be no discourse in any significant sense of that term). Suppose further (as appears to be correct) that old/new information relates to "displacement effects" in narrow syntax. And suppose further (merely for concreteness) that we take these displacement effects to be expressed in narrow syntax by transformational operations. Should we then say that the operations of object-shift, topicalization, etc., literally access shared background information? This seems close to incoherent; any clarification of these intuitive ideas that I can think of yields computational systems of hopeless scope, compelling us to try to formulate what amount to "theories of everything" that cannot possibly be the topic of rational inquiry (NB: not TOE in the technical sense of physics, which is a totally different matter). A more reasonable approach, I think, is to take the operations to be "autonomous," i.e., syntax in the broad sense, and to understand pragmatics to be a theory concerned with the ways properties of expressions (such as displacement) are interpreted by language-external (but person-internal) systems in terms of old/new information. That leaves us with manageable and coherent questions. The conclusion generalizes to other such matters, in my opinion. If that's correct, then syntax (broad or narrow) will be "autonomous" of pragmatics -- though I might add that "autonomy of syntax" is a term I do not think I have ever used, except in reaction to references to some alleged "autonomy of syntax" thesis, sometimes attributed to me. There is a one-sided debate about "autonomy of syntax," one-sided in that only critics of the alleged thesis take part. There are a number of such debates, including the debate over what critics call the "innateness hypothesis" (also often attributed to me, Fodor, and others). I have no idea what the phrase is supposed to mean, and correspondingly have never advocated any such hypothesis -- beyond the truism that there is some language-relevant distinction, to be discovered, between my granddaughter and her pet kitten (monkey, rock, etc.).
    Should pragmatics "be included in an overall theory of linguistic competence"? We seem to be back to terminology. As noted, I've always assumed that pragmatics is a central part of general linguistic theory; my own view is rather extreme in this regard, as just discussed. As for the term "competence," it has the informal meaning of ordinary usage, and also a technical meaning, which is whatever its users assign to it (as in the case of "tensor," "undecidability" in the technical sense, etc.). I am familiar with only one technical sense, the one I've used for many years to mark the conceptual distinction between what Jones knows and what Jones does (competence vs. performance), a distinction that is not controversial. I introduced the technical term in a (probably vain) effort to avoid pointless debate engendered by uncritical use of the informal English notion "knowledge," or use of technical concepts of knowledge introduced in the philosophical literature that diverge considerably from informal usage (which is fine, as long as it is recognized clearly, in which case the debates dissolve).
    If we are using the term "competence" in my technical sense, then pragmatics is not part of a theory of linguistic competence, for uninteresting terminological reasons. If we are using the term "competence" in its ordinary English sense, then I suppose one might say that pragmatics is part of linguistic competence, but the conclusion is again uninteresting, merely a matter of terminology. I don't see any other way to interpret the question. Perhaps I am missing the point.

    Let's turn to (b): is it ´worthwhile to postulate a "neuro" aspect of pragmatics?' If pragmatics is the study of use of language, then it has some basis in the human organism, partly the brain (it presumably also includes much else, including gesture, etc.). If that much is agreed, then we are postulating that brain mechanisms are involved in pragmatics, including those that enter into organization of motor action, perception, etc. One assumes that these include neural mechanisms crucially, though perhaps there are other relevant brain mechanisms -- "brain" is an informal term. Again, I don't really understand what it is at issue.

END of interview
 

Copyright © 1999 by Academic Press