Ethnolinguistic Notes

Series 4, Number 21

COLLATERAL DAMAGE FROM LINGUISTICS?
1.THE POST-CHOMSKYAN PARADIGM AND ITS UNDERLYING ASSUMPTIONS

"The problem which initiates inquiry"
"The problem which initiates inquiry" (revised)
A further look at the assumptions that underlie the Post-Chomskyan paradigm
The underlying picture of language
Confusion: Two very different conceptions of the nature of language

The use of the term "collateral damage" by the military to refer to incidental damage, damage inflicted on anything other than the intended target, has become familiar. My concern here is that linguistics may itself be causing a kind of collateral damage with its central research strategy--that is, with its strategy for research directed toward solving what is taken to be its central problem.

I believe the prevailing conception of that problem rests in large part on false assumptions about the very nature of human language, and that the promulgation of these false assumptions has a deleterious effect on other kinds of research and other areas of human concern. These false assumptions lead to a conception of the nature of language in which the products of more recent cultural evolution are lumped indiscriminately with the original products of our biological evolution.

To take one example of the potential for collateral damage, how can the attempt to reconstruct how language originally evolved not be severely hampered by false assumptions about the nature of language--about just what it was that was evolving? The same question arises about using false assumptions concerning the nature of human linguistic competence as the basis for specifying the similarities and (more especially?) the dissimilarities between that competence and what other species are capable of. Moreover, there are numerous other fields of research whose strategies are dependent to some degree on what is known (or assumed to be known) about language. Among those that have been of particular interest to me personally are language "contact" (indeed, linguistic change generally) and translation. But the research strategies of many other areas seem likely to be affected--the study of aphasia, for example, or language learning (of both first and second language). It seems inevitable that any such research program (and more, see Grace 1995, 1998) must be hampered to some extent if it starts from false assumptions about the nature of language.

I'm blaming these false assumptions on the prevailing conception of what linguistics takes to be its central problem--of what F.S.C Northrop, in his book The logic of the sciences and the humanities (Northrop 1947), referred to as "the problem which initiates inquiry".

"The problem which initiates inquiry"

Of course, disciplines such as linguistics usually don't make a great point of defining their inquiry-initiating problems. And, of course, contemporary linguistics has a number of different subspecialties, each of which perhaps should be described as having its own particular problem that it's focused on solving. However, what I have in mind as the central problem of linguistics as a whole is "the problem which initiates inquiry" for what we may refer to as the "Post-Chomskyan paradigm"-- the problem that was at the heart of the revolution led by Noam Chomsky in the late 1950s and early 1960s and its continuation in contemporary linguistic theory. I've always understood that problem to be something like this:

Although, this is the problem the Post-Chomskyan paradigm set out to solve, there would have been no way to design practical research on a problem so broadly conceived. Further hypotheses or assumptions were necessary to focus the research, and the paradigm supplied them.

There is, of course, already an assumption underlying the general question: the assumption that it is in fact true that, given any reasonable access at all, all "normal" children do "acquire" the ambient language. (Although I presume this assumption is essentially valid, it's worth pointing out that I'm not aware of any attempt actually to test it or even to formulate it as a testable hypothesis. Therefore, it might be hard to tell whether a particular case did constitute an actual counter example rather than simply involving an abnormal child or one who had been denied the requisite access to the language).

But there are further assumptions that are generally made. It appears to be generally assumed not only that children acquire the language with or without any guidance, but that any guidance that may be received has little effect on the nature or quality of the competence in the language that the child eventually attains. Chomsky, in particular, has made such assertions--e.g. (Chomsky 1975, 144), "There need be no explicit teaching or training, and when the latter does take place, it has only marginal effects on the final state achieved".

Moreover, it seems to be assumed that the quality of the examples of speech available to the acquirer is generally poor, but that that circumstance likewise has little or no effect. Again, a claim of Chomsky's--e.g. (1972, 27), "The native speaker has acquired a grammar on the basis of very restricted and degenerate evidence", or (1972, 78), "we cannot avoid being struck by the enormous disparity between knowledge and experience--in the case of language, between the generative grammar that expresses the linguistic competence of the native speaker and the meager and degenerate data on the basis of which he has constructed this grammar for himself."

These last-cited quotations incidentally serve to bring out a further assumption of the Post-Chomskyan paradigm: the assumption that learning to speak a language is mainly a matter of "acquiring" its grammar. In short, knowing a language is knowing a grammar.(1)

How does this grammar acquisition take place? The assumption here is that the child actually "constructs" the grammar for him/herself on the basis of the available data (see the above-cited quotations from Chomsky).

How is it able to do this? The assumption in this case is that the reason that human infants can and do acquire language in the first place is that they are born with an innate "language acquisition device" (LAD). The LAD has been described as (among other things, presumably) defining the set of possible grammars. What this means, in fact, is that it contains a kind of universal grammar with a number of parameters waiting to be set. (What is meant by the set of "possible grammars", then, is simply the set of all of the possible combinations of settings for all of these parameters). The acquisition process is a matter of the child's using the evidence provided by the speech that it hears to set the parameters.

Determining just what is in this set of possible grammars--which is to say, what the properties of the universal grammar are--has become a major objective of theoretical linguistics. In fact, we may think of this problem--determining the properties of this universal grammar--as the latest conception of the "problem which initiated inquiry" for Post-Chomskyan linguistics. In other words, the original question--how it's possible for all normal children to acquire language--has been reinterpreted, with the aid of certain assumptions, as a question about the properties of a hypothesized universal grammar.

"The problem which initiates inquiry" (revised)

A further look at the assumptions that underlie the Post-Chomskyan paradigm

The assumptions being made by the Post-Chomskyan paradigm really deserve careful scrutiny, and I want to make my own attempt to show what some of these assumptions are. Of course, the following account makes no pretense of being definitive or complete; different analysts would surely come up with somewhat differently stated assumptions and somewhat different orderings among them. However, I don't think there should be any serious disagreements about their cumulative content.

There is some logical order among the assumptions; the following three seem the obvious place to start:

3. The quality of the examples of speech available to the acquirer is generally poor, but that circumstance has little or no effect.

Then, in explanation of these assumed facts, there is the assumption:

4. The reason that human infants can and do acquire language in the first place is that they have an innate "language acquisition device" (LAD).

Maybe I've formulated this badly--is this an assumption at all? However, this formulation does make the point that humans are born able to acquire language (in a way that no other species is) and it does introduce the concept of the LAD (with its suggestion that this ability is a unit--an innovation of kind rather than of degree). But beyond this suggestion, what do we know about this LAD? The next assumption provides an answer, although it raises its own questions.

5. The main component of the LAD (or at least the one of greatest linguistic interest) is a kind of universal grammar with a number of parameters waiting to be set. (As we saw, this universal grammar has been described as defining the set of possible grammars, which is to say the set of all of the possible combinations of settings for all of these parameters). The acquisition process is thus assumed to be a matter of the child's using the evidence provided by the speech that it hears and setting the parameters.

6. Linguistics has the potential to make a major contribution to cognitive science and to the understanding of the human brain by determining (or making progress toward determining) the properties of the supposed universal grammar (which is to say by making progress toward solving the inquiry-initiating problem of the Post-Chomskyan paradigm). This assumption has provided a great stimulus to and justification for linguistic research within the paradigm. But is the assumption that the main component of the LAD is a universal grammar indeed valid? It doesn't really make sense without other assumptions that would explicate and support it. These can be stated as follows:

7. Acquiring a language consists in essence in acquiring (constructing for oneself) a grammar. (Therefore, it makes sense for the main component of the LAD to be a universal grammar).

Again, this assumption doesn't make sense without this further assumption underlying it:

8. A language consists essentially of a syntax that provides the logical structure of its sentences and a lexicon (the word "grammar" is sometimes understood to include the lexical items as well as the logical rules that govern their combination). (Therefore, it makes sense for the acquisition of a grammar to be the essential part of the acquisition of a language).(2)

However, we're still dealing with assumptions whose basis is not self-evident. Their immediate source appears to be a particular picture of language--a particular understanding of what language is--that seems to underlie the whole Post-Chomskyan paradigm. It appears that the Post-Chomskyan paradigm's picture of the basic nature of human language can be described with something like the following propositions:

The underlying picture of language

I. The prototypical function of language is to communicate factual information.

(What I mean by saying that some function [or whatever] is assumed to be "prototypical" is that it's being treated as the key one--as one that can serve as a key to whatever others may exist. For example, understanding it first will presumably open the door to understanding any others that may exist [whereas understanding some other one first would presumably offer no such advantage].)(3)

VI. An "ideal speaker" of a language would be a person who had all the knowledge necessary to encode and decode every sentence permitted by the language. [NB this definition ignores knowledge or other abilities required for any functions that these assumptions haven't identified as prototypical].

This is where the trouble starts. Two very different conceptions of the nature of language are being confused. On the one hand we have language as described by these propositions. Such a language is designed for expository prose, for communicating factual information by encoding and decoding propositions, for the closest approximation possible to what has been called "autonomous text".

"Autonomous text" in the intended sense would be text that could stand alone. Its entire meaning would be encoded in the text itself so that no other information was needed for its interpretation. Ordinarily, in interpreting an oral speech act one has many kinds of information other than the words themselves that contribute to the interpretation. One will surely take into account whatever one knows about who the speaker is, whom s/he is addressing (and who else may be listening), and how the speaker relates to any and all members of this audience. Further, one's interpretation would generally be affected by the apparent mood of the speaker and by the knowledge and assumptions that s/he seems to presume are shared between speaker and audience (both permanent sharings and sharings of what is perceptible in the immediate environment). And, of course, there may be much more. However, if the text itself is truly to be treated as autonomous, none of these kinds of information can be taken into account--only the text itself can be considered.

For convenience, let's call the kind of language designed for this purpose (which is to say, the kind defined by the above propositions) "autonomous text (AT) language". Although the nearest approximations to AT language can be and are spoken, its primary medium is writing. The spoken forms are derivative. These spoken derivatives appear in their purest form in contexts where existing or potential written documents form part of the background--"papers" presented at scientific conferences, legal arguments, business negotiations, etc. However, their use (especially when use in somewhat attenuated form is included) extends far beyond such contexts into every aspect of the lives of many of us (who often have no real competence in any other form of language--cf. Syder 1983).

Thus, we have one conception of human language in which it is seen as consisting of a great number of distinct units ("languages") that may appropriately be analyzed as codes for expository prose--that is, as AT languages. The widespread acceptance by linguists of this conception of language goes a long way toward explaining the success of the claim that knowing a language is knowing a grammar--no doubt competence in writing expository prose does require a considerable knowledge of grammar (although certainly not that alone).

But, we also have another conception, defined by the assumption with which we began--the assumption that all normal children acquire language even without any guidance and even when the evidence available to them is "very restricted and degenerate". Let's call the language defined by the linguistic abilities ("competence") of people whose language acquisition took place under these most unfavorable circumstances "SA (for 'spontaneously acquired') language". To assume AT-language- type linguistic performance to be the natural outcome of SA-language-type acquisition processes is an invitation to serious error.

We said that the central problem of linguistics--"the problem which initiated inquiry"--is to determine the properties of a supposed universal grammar that is the key part of the innate "language acquisition device" (LAD). It would seem, then, that our first order of business should be to determine what kind of linguistic competence the LAD itself is capable of producing. That would seem to mean studying the linguistic competence of individuals who acquired their language under as near an approximation as possible to the specified conditions, viz., "on the basis of very restricted and degenerate evidence" without any "explicit teaching or training". Thus, there should be no guidance from caregivers (or anyone else).

In fact, this would mean acquisition in an environment where no kind of linguistic ideology is in any way communicated to the child. It goes without saying that the environment wouldn't include such conspicuous kinds of intervention as the recitation of conjugations of French irregular verbs for the child while it was still in utero, but the requirements go far beyond that. For example, one requirement would be that the caregivers of the child not be believers in the grammar-lexicon model of language structure who would try to guide and assist its acquisition by isolating and repeating individual words for it.

Although it's probably still possible to find places in the world where children's caregivers are largely free of language ideology of the modern Western variety, it's difficult to imagine a language-acquisition environment that's completely free of all cultural influences (even the language samples that would constitute the evidence used by the LAD could hardly be purified of all content that reflected ideology concerning language). In fact, even if such human experimentation were permitted, it would be next to impossible to design laboratory conditions that could produce children whose language acquisition was entirely free of relevant enculturative influences. In short, it's unrealistic to suppose that we'll ever be able to find native speakers of "natural languages" (if "natural language" is really to mean SA language--language as defined by the LAD alone).

Then, what can realistically be done? What kind of approach could linguistics use to overcome this problem--to try to pin down what's actually acquired through the operation of those parts of the brain that constitute the so-called "language acquisition device"?

Until our understanding of the relevant neurology is very much more advanced than it is now, the only answer seems to lie in a more critical evaluation of the speakers from whom we obtain our information on grammars. This evaluation should be an attempt to determine what cultural influences might have contributed to their knowledge and use of language. The initial objective of such evaluation would be simply to make us more sophisticated about the kinds of cultural influences that affect linguistic performance.

I plan, in my next Note, to discuss how some of the enormous cultural evolution that's occurred since the time of earliest language seems to have had effects on language as well--to have led to cultural evolution in language.

1. To put it differently, this paradigm subscribes to what Paul Hopper has called "the A Priori Grammar Postulate"--it makes "the initial assumption that a grammar is a discrete set of rules which are logically and mentally presupposed by discourse; that is, that grammar is logically detachable from discourse and precedes discourse" (Hopper 1988, 118).(Back up)

2.We should probably also be conscious of a few additional assumptions that guide research within the Post-Chomskyan paradigm--at least the research most directly related to the inquiry-initiating problem. I believe all of the following are accepted quite generally, if tacitly, within the paradigm.

1. All "natural" languages must conform to the limits defined by the universal grammar.

2. Therefore, the grammars of natural languages are a (the main) source of evidence on the properties of the universal grammar.

3. Standard languages such as Standard English, French, Japanese, Mandarin, etc. are "natural" languages, and they, therefore, provide direct evidence for determining the properties of the universal grammar.

4. The data supplied by "intuition", that is, private data supplemented with private judgments of grammaticality, are entirely adequate as evidence concerning specific grammatical rules.

(As far as I know, none of these assumptions has ever been proposed as a hypothesis that should be tested, nor do I know of any attempt to formulate any of them in testable form.) (Back up)

3.I should say a little more about what is implied, or at least suggested, by the notion of "prototypicality" that I'm resorting to a good bit here. Recognizing--however tacitly-- a particular function as prototypical at least suggests that it's more important than any other function. The assumption that it's the key to the understanding of the others implies that it's logically prior to them--that they most likely presuppose it. Furthermore, its being logically prior would at least suggest that it's chronologically prior--that the others have actually evolved from it.

Therefore, to say that Post-Chomskyan linguistics assumes the encoding and decoding of propositions expressing factual information to be the prototypical function of language is to say that it treats this function as if it were the most representative one, the best example to start with. And it seems clearly implied that it must be the original function of language--the main function that guided the evolution of language.

It should be pointed out that assuming this to be the prototypical function of language in no way denies that we can and do use the mechanism which (by this assumption) constitutes the governing system for other purposes: to ask questions, give orders, make statements whose truth value is not determinate, and much else. What it does do is relegate these uses to derivative status. According to this view the main function of the supposed governing system is still to permit the formulation of propositions expressing factual information, and--for some purposes at least--any subsidiary functions can safely be disregarded.(Back up)

Chomsky, Noam. 1972. Language and Mind, Enlarged edition. New York, etc.: Harcourt Brace Jovanovich, Inc.(Back up)

Chomsky, Noam. 1975. Reflections on Language. New York: Pantheon Books. (Back up)

Grace, George W. 1995. Why I don't believe that language acquisition involves the construction of a grammar. Ethnolinguistic Notes, Series 4, Number 1. Internet World Web page at http://www2.hawaii.edu/~grace/elniv1.html. (Back up)

Grace, George W. 1998. Some puzzles that arise from the assumption that to learn a language is to construct a grammar. In Mark Janse (ed.)(with the assistance of An Verlinden). Productivity and creativity: Studies in general and descriptive linguistics in honor of E. M. Uhlenbeck. Berlin and New York: Mouton de Gruyter, pp. 69-81. (Back up)

Hopper, Paul. 1988. Emergent grammar and the A Priori Grammar Postulate. In Deborah Tannen (ed.). Linguistics in context: Connecting observation and understanding. Norwood NJ: Ablex Pub. Co., pp. 117-34. (Back up)

Northrop, F[ilmer] S. C. 1947. The logic of the sciences and the humanities. New York: Macmillan Co. (Back up)

Syder, Frances Hodgetts. 1983. The fourth R: Spoken language, English teaching and social competence. Auckland, New Zealand: Talanoa Press.(Back up)