Perhaps the most traditional way of framing the innateness hypothesis would be in terms of an opposition between rationalism and empiricism. This is an opposition that is frequently encountered in philosophical debates over language acquisition, with the one side arguing that language acquisition is a phenomenon associated with the maturation of a language faculty or "organ," while the other side argues that language acquisition is instead a process of generalization from experience. Underlying these two positions are specific hypotheses regarding the resources the child brings to the task of acquiring language. The rationalist postulates the existence of innate ideas and/or principles, while the empiricist does not. Chomsky's position, as he consistently has acknowledged for decades, is rationalist. As he put it not long after Aspects appeared, his is a theory of " psychological a priori principles that bears a striking resemblance to the classical doctrine of innate ideas " (Searle 1977 p. 121). This general description is as valid for the current Government Binding (GB) theory as it was for the various Transformational-Generative (TG) programs of the 1960s and 1970s.
Does anything consequential follow from this opposition? I believe it does. Prima facie, of course, there is a significant difference between the assertion that there are innate ideas and principles and the assertion that there are not, and it is reasonable to expect that a research program based on either position would hypothesize explanatory agents, frame questions, and categorize evidence in a way that would differ significantly from the other.
Of perhaps more importance, however, is the fact that rationalist and empiricist hypotheses are addressed to fairly different explananda. In terms of developmental phenomena, the generative theory's rationalism can be seen as largely concerned with explaining the activation of the a priori principles determining the adult's attained language state, whereas more or less empiricist theories -- connectionist-inspired theories come immediately to mind -- have been concerned with explaining the dynamic stages understood to be manifested in the child's linguistic development. Consequently, the rationalist and empiricist explanatory programs are aimed at different fundamental questions regarding language acquisition. Whereas the generative program asks
the empiricist program asks
The object of explanation aimed at in the first question is the maturation of a hypothesized pregiven language faculty, while the corresponding object of explanation aimed at in the second question is the " bootstrapping " by which language is learned. (It is worth noting that the generative program appears to be less concerned with developmental issues per se than with articulating the abstract principles of the universal grammar held to underlie language development.) Given this difference in explananda, we find corresponding differences in approach to the empirical data -- for example, the bits of language that serve as the child's input stimuli. For a theory of maturation, the relationship between the input stimuli and resulting language competence consists in the triggering of a set of pregiven constraints, while for a theory of learning the relationship is instead considered to be an inductive one. (We might expect that the relative importance each approach attributes to the input stimuli in explaining language competence will differ accordingly as well.)
So to the extent that the above distinctions drawn between explanatory programs have any validity -- and in broad outline I believe they do -- we can say that something of consequence does follow when we frame the debate over language acquisition in terms of rationalism versus empiricism. But rationalism and empiricism by no means exhaust the possible frameworks within which to discuss innateness, for not all proposed innate mechanisms must be described in terms of a priori principles. We might note that the empiricist appeal to inductive learning from environmental features does not necessarily rule out the functioning of innate mechanisms and constraints, and indeed may even be seen to require it. Certainly Chomsky has acknowledged this (Aspects p. 51). There is thus no necessary contradiction between holding that language is learned inductively and holding that the learning mechanisms that come into play are shaped by innate constraints on conceptual form and association, perceptual input and stimulus field organization, and so forth. Some theories of language learning based on PDP-influenced models of the brain in fact do hold something like this (e.g., Arbib and Hill).
So it would appear that reference to innate capacities does not necessarily entail rationalism, and that we can have a predication of innateness with our empiricism. For this reason, the best way to begin may be to acknowledge that there must be some innate factors operating in natural language. There is good evidence for this: because only humans acquire true languages -- that is, languages that are characterized by recursiveness -- it seems reasonable to hold that human languages are at least partly the product of constraints or mechanisms innate to humans and not present in other animals. If this is so, then any theory of natural language will necessarily involve the postulation or assumption of an innateness hypothesis of one sort or another. (Chomsky is thus right to observe that there is no innateness hypothesis per se, but instead there are a plurality of innateness hypotheses (Kasher 1991 p. 15).) This means that rather than asking whether or not the human capacity for language is innate, we must ask which aspects of language are plausibly explained by appeal to innate mechanisms and constraints, and then go on to suggest what these mechanisms and constraints may be.
It is possible of course for one to hypothesize that these mechanisms and constraints consist in a set of a priori principles -- and thus to take up a rationalist position. But under the terms of discussion proposed, the debate would not then revert to one of rationalism versus empiricism, but rather would pivot on a change of focus by virtue of which we would find ourselves considering one variety of innate constraint -- a rationalist one -- in relation to other, non-rationalist varieties.
Rather than framing an innateness hypothesis, then, we would instead distinguish among a number of innateness hypotheses. Accordingly, we can say that the way to frame the innateness question is to attempt to distinguish among the different constraints and mechanisms that can be attributed to the human biological endowment for language.
We might tentatively posit innate constraints on:
The first and third hypotheses here naturally raise questions regarding the relationship between purely linguistic mechanisms and the cognitive mechanisms associated with generalized learning structures as well as with non-linguistic representational systems. Such questions may lead to the formulation of
and
Of possible relation to item 5 would be
Other hypotheses may address themselves to the phylogenetic and/or ontogenetic aspects of the biological endowment for language, as would
An additional hypothesis to consider might be:
which posits the existence of innate principles, mechanisms, or faculties proper to a protolanguage and separate from those associated with a true language.
One final candidate for an innateness hypothesis merits mention. Following Jackendoff (1995), we might hypothesize
This faculty, which is claimed to encode a set of basic, universal cultural paramaters, presumably would be responsible for the language user's pragmatic competence. (Interestingly, such a hypothesis would open up a potentially fascinating point of contact between the generative program and Wittgensteinian intuitions regarding forms of life and apparently innate human tendencies to agree.)
This list of possible innateness hypotheses is far from exhaustive, but it should give an idea of the potential range such hypotheses may take.
Despite their diversity, all of the above hypotheses represent attempts to carve the language phenomenon at convenient explanatory joints. Some of the analytical frameworks assumed may overlap, some of the hypotheses may be seen to be complementary or at least compatible, while some may be seen to describe the same phenomenon from different angles, i.e., with different explanatory ends in mind. Note too that not all of these hypotheses are addressed strictly to the core aspects of language. What is common to all is that they are at least in principle empirically decidable hypotheses specifying the diverse innate constraints constituting or affecting the language capacity.
This essay is a response to How should the Innatenes Hypothesis be reframed?
Back up to Chomsky for Philosophers.
Arbib, Michael A., and Jane C. Hill. 1988 " Language Acquisition: Schemas Replace Universal Grammar." In Hawkins,, pp. 56-72.
Chomsky, Noam. Recent Contributions to the Theory of Innate Ideas: Summary of Oral Presentation. In Chomsky, Noam 1991 "Linguistics and Adjacent Fields: A Personal View" In Kasher, pp. 3-25.
Hawkins, John A. ed. 1988 Explaining Language Universals Blackwell, Oxford.
Jackendoff, Ray. Languages of the Mind: Essays on Mental Representation (MIT, 1995).
Kasher, Asa ed. The Chomskyan Turn (Blackwell, 1991).
Searle, John, ed. Philosophy of Language (Oxford University Press, 1977).