Copyright © 1999 Massachusetts Institute of Technology
Although the idea has a large philosophical tradition (especially in the work of the Continental rationalists; see RATIONALISM VS. EMPIRICISM), modern ideas concerning the innateness of language originated in the work of Chomsky (1965 etc.) and the concomitant development of GENERATIVE GRAMMAR. Chomsky's hypothesis is that many aspects of the formal structure of language are encoded in the genome. The hypothesis then becomes an empirical hypothesis, to be accepted or validated according to standard empirical methods.
As with any other hypothesis in the natural sciences, the innateness hypothesis (that there exist genetically specified aspects of language) has to be evaluated alongside competing hypotheses. Clearly, the competing hypothesis is that there are no genetically specified aspects of language. If one accepts that any genetically specified aspects of language exist, then there is no more debate about a general innateness hypothesis, but only a debate about exactly which aspects of language are innate. This debate, in fact, is central to current research in linguistics and PSYCHOLINGUISTICS.
There are many arguments for the innateness hypothesis. But the most significant one in Chomsky's writings, and the one that has most affected the field, is the argument from the POVERTY OF THE STIMULUS (APS; see also Wexler 1991). As Chomsky points out, this argument in the study of language is a modern version of DESCARTES's argument concerning human knowledge of CONCEPTS. The basic thrust of the argument goes as follows:
(1) Human language has the following complex form: G (for Grammar)
(2) The nature of the information about G available to the learner is the following: I (for Input/Information, called Primary Linguistic Data in Chomsky 1965).
(3) No learner could take the information in I and transform it into G.
In other words, the argument from the poverty of the stimulus is that the information in the environment is not rich enough to allow a human learner to attain adult competence.
The arguments to support APS in linguistic theory usually involve linguistic structures that do not seem to be encoded in environmental events. Thus (4a,b) seem to have identical surface structures, yet in (4a) Mary is the subject of please (she will do the pleasing) and in (4b) Mary is the object of please (she will be pleased). How will a learner learn this, inasmuch as it seems that the information is not directly provided to the learner in the surface form of the sentence?
(4) a. Mary is eager to please.
b. Mary is easy to please.
The field of learnability theory (Wexler and Hamburger 1973; Wexler and Culicover 1980) developed as an attempt to provide mathematical preciseness to the APS, and to derive exact consequences. Learnability theory provides an exact characterization of the class of possible grammars, the nature of the input information, and the learning mechanisms, and asks whether the learning mechanism can in fact learn any possible grammar. The basic results of the field include the formal, mathematical demonstration that without serious constraints on the nature of human grammar, no possible learning mechanism can in fact learn the class of human grammars. In some cases it is possible to derive from learnability considerations the existence of specific constraints on the nature of human grammar. These predictions can then be tested empirically.
The strongest, most central arguments for innateness thus continue to be the arguments from APS and learnability theory. This is recognized by critics of the Innateness Hypothesis (e.g., Elman et al. 1996; Quartz and Senjowski 1997). The latter write (section 4.1):
The best known characterization of a developing system's learning properties comes from language acquisition -- what syntactic properties a child could learn, what in the environment could serve as evidence for that learning, and ultimately, what must be prespecified by the child's genetic endowment. From these questions, 30 years of research have provided mainly negative results. . . . In the end, theorists concluded that the child must bring most if its syntactic knowledge, in the form of a universal grammar, to the problem in advance. . . . The perception that this striking view of syntax acquisition is based primarily on rigorous results in formal learning theory makes it especially compelling. Indeed, above all, it is this formal feature that has prompted its generalization from syntax to the view of the entire mind as a collection of innately specified, specialized modules. . . . It is probably no overstatement to suggest that much of cognitive science is still dominated by Chomsky's nativist view of the mind.
In addition to the APS, there are a number of other arguments for the innateness hypothesis. These include (1) the similarity of languages around the world on a wide array of abstract features, even when the languages are not in contact, and the features do not have an obvious functional motivation; (2) the rapid and uniform acquisition of language by most children, without instruction (see PHONOLOGY, ACQUISITION OF; SEMANTICS, ACQUISITION OF; SYNTAX, ACQUISITION OF) whereas many other tasks (e.g., problem solving of various kinds) need instruction and are not uniformly attained by the entire population.
Over the years there have been many attempts (currently fashionable ones include Elman et al. 1996; Quartz and Senjowski 1996) to suggest that perhaps learning could explain LANGUAGE ACQUISITION after all. However, no arguments have been given that overcome the central learnability argumentation for the innateness hypothesis. For example, neither Elman et al. nor Quartz and Senjowski explain via "learning" any of the properties of universal grammar or how they are attained. Quartz and Sejnowski attempt to critique learnability theory, but their critique does not apply to actual studies in learnability theory. For example, they characterize learnability theory as assuming that learners must enumerate every possible language in the class as part of the learning procedure. This is false of Gold (1967) and explicitly argued against in Wexler and Culicover (1980). Wexler and Culicover, who are greatly concerned with the psychological plausibility of the learning procedure, derive their results under some quite severe restrictions, making their learning procedure much more empirically adequate on psychological grounds than the procedures that are considered in so-called learning accounts. (See Gibson and Wexler 1994 for an analysis of psychologically plausible learning mechanisms in the principles and parameters framework in which there is much innate knowledge.) One simply has to say that CONNECTIONIST APPROACHES TO LANGUAGE and its acquisition are simply programmatic statements without any kind of theoretical or empirical support. To be taken seriously as a competitor to the innateness hypothesis, these approaches will have to attain real results .
Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Elman, J. L., E. A. Bates, M. H. Johnson, A. Karmiloff-Smith, D. Parisi, and K. Plunkett. (1996). Rethinking Innateness: A Connectionist Perspective on Development. Cambridge, MA: MIT Press.
Gibson, E., and K. Wexler. (1994). Triggers. Linguistic Inquiry 25(3):407-454.
Gold, E. M. (1967). Language identification in the limit. Information and Control 10:447-474.
Quartz, S. R., and T. J. Sejnowski. (1997). The neural basis of cognitive development: a constructivist manifesto. Behavioral and Brain Sciences 20(4):537-596.
Wexler, K. (1991). On the argument from the poverty of the stimulus. In A. Kasher, Ed., The Chomskyan Turn. Cambridge: Blackwell, pp. 253-270.
Wexler, K., and P. Culicover. (1980). Formal Principles of Language Acquisition. Cambridge, MA: MIT Press.
Wexler, K., and H. Hamburger. (1973). On the insufficiency of surface data for the learning of transformational languages. In K. J. J. Hintikka, J. M. E. Moravcsik, and P. Suppes, Eds., Approaches to Natural Language. Proceedings of the 1970 Standard Workshop on Grammar and Semantics. Dordrecht: Reidel, pp. 167-179 .
Copyright © 1999 Massachusetts Institute of Technology