There are, broadly speaking, three competing frameworks for answering the foundational questions of linguistic theory—cognitivism (e.g., Chomsky 1995, 2000), platonism (e.g., Katz 1981, 2000), and nominalism (e.g., Devitt 2006, 2008).
Platonism is the view that the subject matter of linguistics is an uncountable set of abstracta—entities that are located outside of spacetime and enter into no causal interactions. On this view, the purpose of a grammar is to lay bare the essential properties of such entities and the metaphysically necessary relations between them, in roughly the way that mathematicians do with numbers and functions. The question of which grammar a speaker cognizes is to be settled afterward, by psychologists, using methods that are quite different from the nonempirical methods of linguistic inquiry.
The nominalist, too, denies that grammars are psychological hypotheses. But she takes the subject matter of linguistics to consist in concrete physical tokens—inscriptions, acoustic blasts, bodily movements, and the like. Taken together, these entities comprise public systems of communication, governed by social conventions. The purpose of a grammar, on this view, is to explain why some of these entities are, e.g., grammatical, co-referential, or contradictory, and why some entail, bind, or c-command others.
Cognitivism, by contrast, is the view that linguistics is a branch of psychology—i.e., that grammars are hypotheses about the language faculty, an aspect of the human mind/brain. A true grammar would be psychologically real, in the sense that it would correctly describe the tacit knowledge that every competent speaker has—a system of psychological states that is causally implicated in the use and acquisition of language.
The epistemological side of the platonist position faces a challenge from the Quinean attack on the distinction between empirical and nonempirical modes of inquiry. Katz (2000) argues that Quine’s epistemology is inconsistent, because it entails that the principles of reasoning are simultaneously revisable and unrevisable. But this argument is fallacious. It overlooks a distinction between our principles of reasoning and our theory of those principles. Drawing this distinction eliminates the threat of inconsistency.
As regards the ontology of linguistics, Katz (1985) argues that the optimal grammar for natural language generates nondenumerably many sentences, and, hence, that linguistics cannot be about any aspect of the natural world. But I maintain that linguists’ claims concerning the infinitude of language need not signal an ontological commitment to abstract entities. Rather, they reflect the lawlike, counterfactual-supporting character of linguistic generalizations, as well as a principled idealization away from mortality, memory constraints, and motivational factors, inter alia. As both Chomsky (2001) and Devitt (2006) agree, there is no reason to take as literally true the claim that there exist an uncountable infinity of sentences. The nearest truth in the vicinity is that there is no principled reason to set an upper bound on the length of a sentence that an English speaker (whether human, future-human, or artificial) can process.
All in all, then, platonism provides a radically mistaken view of the methodology and ontology of linguistics. Let’s now examine the remaining interpretations of linguistic theory: the dominant cognitivist position and the sophisticated nominalist rival. As both Chomsky and Devitt point out, if one begins with the cognitivist conception, then the question of whether a syntactic rule or principle is psychologically real reduces to the question of whether we have grounds to believe the grammar that posits it. If Chomsky’s cognitive conception is right, then, given that we typically do have such grounds, the psychological reality issue is already settled. By contrast, if one begins with Devitt’s nominalist conception of grammars, then the ascription of psychological reality to one or another syntactic principle requires, in addition, powerful psychological assumptions.
Though I am ultimately neutral on which of these views provides a more satisfying conception of linguistic inquiry, I side with Devitt in thinking that the psychological reality debate cannot be settled without direct appeal to the results of psycholinguistic experiments. Thus, in an effort to block the trivialization of the psychological reality issue, let me briefly cast doubt on two standard arguments for Chomsky’s cognitivist conception. If the cognitivist conception is not the only game in town, then the psychological reality claim becomes interesting; far from being trivially true, it stands in need of a sustained defense. I go on to develop such a defense in later posts.
One common motivation for cognitivism is that, unlike its rivals, it has the resources to motivate the search for universal linguistic principles—a Universal Grammar (UG). On the cognitivist conception of syntax, a specification of UG tells us about the innate resources that a child brings to bear in the acquisition process. Given that language acquisition is an independently interesting phenomenon, the more help we get from syntax in theorizing about it, the more credibility accrues to the syntactic proposals. What analogous motivation can the nominalist provide?
To reply, I note that evidence for any claim about the structure of UG invariably rests on the putative existence of a linguistic universal. But we have no way of determining what linguistic universals there are except by constructing grammars for a variety of languages and checking whether the constructs employed by the grammar of L are applicable to another language, L*. The methodology of devising and comparing the grammars of various languages does not presuppose or require cognitivism. In particular, the conclusion of the following inference is a non sequitur.
“Since general linguistic theory describes the common resources of the grammars and there must be something common to all humans as acquirers of language, it looks as if general linguistic theory was all along an account of a universal human cognitive feature, that is, UG” (Collins, 2008: 86).
It’s of course true that “general linguistic theory describes the common resources of grammars,” and it’s likewise true that “there must be something common to all humans as acquirers of language.” But it does not follow, and probably is not true, that every linguistic universal that we ever discover—if, indeed, we discover any (Evans and Levinson, 2009)—must automatically be seen as encoded in the human genome, represented in the minds of competent speakers, or involved in language acquisition. For any putative linguistic universal, there are numerous possible explanations. Methodologically, we must first get clear about what universals there are—or even whether there are any—and then examine them, one by one, proposing and (dis)confirming competing genetic, environmental, social, and psychological explanations. An a priori commitment to the cognitive conception is out of place in an empirical discipline.
Still, the project of “general linguistic theory” is a worthy one. Syntacticians are right to borrow the resources of a grammar of one language in theorizing about another. And, in accordance with general principles of theory choice—the desirability of explanatory unification and maximum generality—they are right to prefer grammars that fit well with what is known about other languages. For this reason, the explanatory adequacy of a grammar is best construed as a successful fit with the maximally general, simple, and unified theoretical coverage of all human languages. Construed in this way, explanatory adequacy is desirable whether or not the cognitivist conception of linguistics is true.
Another common argument in favor of the cognitivist conception is that the notion of a “public language” is irrelevant to scientific inquiry. On behalf of the opposition, I argue that the notion of a public language is indispensable in the study of language acquisition. The data and explananda of acquisition theory are routinely couched in terms that make ineliminable reference to public languages. The empirical findings that animate the poverty of the stimulus argument, and inquiry in acquisition theory more generally, typically concern quantitatively described patterns of error. It is very difficult to see how such findings might be formally recast as claims about the relation between the child’s usage and some specific I-language. Against whose I-language would a child’s usage be quantitatively compared?
One might reply that the acquisition theorist is comparing the child’s grammar to an idealized I-language. But it’s not at all clear what import the appeal to idealization has in this context. If we press on the notion of an idealized I-language, we find that it amounts to no more than a consistent setting of parameters (assuming a Principles-and-Parameters grammar). But there are many such settings, most of which fail to match the language of the child’s linguistic community. Thus, reference to the grammar of a public language seems unavoidable in singling out the language that the theorist identifies as the child’s “target grammar.” If there is an idealization in the vicinity, it is one that abstracts away from the variation within the community—i.e., the differences between individual speakers—and yields an idealized speech community as the theoretical object of interest.
References
Chomsky, N. (1995). The Minimalist Program, Cambridge, Mass.: MIT Press.
Chomsky, N. (2000). New Horizons in the Study of Language and Mind, Cambridge, Mass.: Cambridge University Press.
Chomsky, N. (2001). “Derivation by Phase,” in Ken Hale: A Life in Language (Current Studies in Linguistics 36), Michael Kenstowicz (ed.), pp. 1–52. Cambridge, MA: MIT Press.
Collins, J. (2008). Chomsky: A Guide for the Perplexed, Continuum Press.
Devitt, M. (2006). Ignorance of Language. Oxford: Oxford University Press.
Devitt, M. (2008). “Explanation and Reality in Linguistics,” Croatian Journal of Philosophy, Vol. VIII, No. 23, pp. 203-231.
Evans, N. and Levinson, S. C. (2009). “The myth of language universals: Language diversity and its importance for cognitive science,” in Behavioral and Brain Sciences, 32, pp. 429-92.
Katz, J. J. (1981). Language and Other Abstract Objects. Totowa, N. J.: Rowman & Littlefield.
Katz. J. J. (2000). Realistic Rationalism. Cambridge: MIT Press.