The beetle in the box: Universal grammar and the private language problem

This paper take sup the theory put forward by Gunther Grewendorf (1995) tha t Wittgenstein's conception of language as a publicly constituted 'life for m' is compatible with Chomsky's conception of language as a human 'organ', although this conception could undoubtedly be characterized to a certain ex tent as a private-language conception in Wittgenstein's sense. In a first s tep, it is shown how Chomsky's model of an autonomous syntax module or a un iversal grammar can indeed by characterized as a 'private-language' concept ion. This is essentially related to the construction of this module as a ma terial, neural state (Chomsky 1995), as a linguistic 'knowledge', whose man ner of functioning cannot become apparent to the speaker. In a second step, the major points of Grewendorf's argumentation are summarized. In spite of a very precise and accurate reconstruction of Wittgenstein's argument, the reconstruction ultimately follows Kripke's interpretation, which, as is we ll known, trivializes this argument. The decisive aspect for Grewendorf's a rgumentation is his theory - which here undoubtedly contradicts Wittgenstei n's text and logic - that the differentiation between know and believe to k now is irrelevant for Wittgenstein's refutation of the possibility of a pri vate language. It is, thirdly, demonstrated that Chomsky's theory would be tenable, or not affected by Wittgenstein's argument, if the existence of th e syntax module could be demonstrated irrespective of the reference to perf ormance phenomena. Fourthly, and finally, it is then shown by using grammat ical categories as theoretical predicates that this is not possible for pur ely logical reasons. Chomsky can only give these predicates a second extens ion, i.e. use them as descriptors of neural states, if he simultaneously de fines them extensionally on the first level by deriving them in a generativ e grammar which culminates in assigning lexicon elements to terminal catego ries. Otherwise it would be simply impossible to correlate statements on th e forms of the syntax module with statements on surface forms. The autonomo us syntax module is thus, on the one hand, identified as a private-language construction par excellence and with Wittgenstein it is shown why such a c onstruction is logically untenable. In parallel to this , attention is draw n to recent developments in the theory of neural networks, in which Chomsky 's basic argument from his Skinner review that without 'innate ideas' a chi ld would not be in a position to develop a perfect grammar from imperfect l inguistic data is factually and definitively refuted. The human brain is ce rtainly capable of learning inductively or analogically on the basis of sim ilarity relations. Fodor's theory of the autonomy of brain modules in essen ce contradicts recent findings on the interlinkage of modules enabling the brain to become self-organizing.