1. In an interesting study, Kaminsky, J., Call, J. and Fischer, J. (2004) Word learning in a domestic dog: evidence for ‘fast mapping’. Science, 304, 1682–1683, the authors demonstrate that dogs are capable of associating phonetic signals to the representation of objects for a remarkable number of items, without training and even on the basis of ‘indirect learning’.CrossRefGoogle Scholar

2. A systematic account of this observation is developed in G. A. Miller, E. Galanter, K. H. Pribram (1960) Plans and the Structure of Behavior (New York: Holt, Rinehart & Winston).Google Scholar

3. The view sketched is presented in a slightly different way in N. Chomsky (1995) The Minimalist Program (Cambridge, Mass: MIT Press).Google Scholar

4. For example in Chomsky, N. (2002) On Nature and Language (Cambridge University Press).CrossRefGoogle Scholar

5. See Jackendoff, R. (2002) Foundations of Language (Oxford University Press) chapter 8.CrossRefGoogle Scholar

6. As, for example, Fuster, J. M. (2003) Cortex and Mind (Oxford University Press), p. 40, admits, without discussing alternatives or additions, however.Google Scholar

7. See Fuster, J. M. (2003) Cortex and Mind (Oxford University Press), pp. 208 f, where the example cherries turn red as they get ripe is discussed in some detail along these lines. In slightly different ways, and with technical details relying on context-free phrase structure, a similar view is advocated by Schnelle, this issue.Google Scholar

8.Hebb, D. O. (1949) The Organization of Behavior (New York: Wiley) proposed in fact two rules of synaptic modification, which have been shown to be reducible to one unified principle.Google Scholar

9. It is a non-trivial question of whether Hebbian Learning can be extended to cover the creation of new synaptic connections under appropriate conditions, or whether this would introduce a completely different type of learning. For the sake of argument, I will assume here that an advocate of Hebbian Learning would assume a conservative option to be indicated.Google Scholar

10. I will leave aside here the question whether there are also two occurrences of do in they didn’t do it, which would raise difficult problems about the identity of lexical items. A different problem to be ignored here is the representation of an interpretation of the ‘original’ coordinate structure, which must, of course, be realized in one way or the other.Google Scholar

11.Jackendoff, R. (2002) Foundations of Language (Oxford University Press) discusses this example and a wide range of related problems in much the same spirit as that of the present remarks.CrossRefGoogle Scholar

12. It is sometimes suggested that syntactic trees or configurations of the type in question might, as such, belong to long-term memory. Thus, J. M. Fuster (2003) Cortex and Mind (Oxford University Press) appears to assume that what he calls executive cognits are networks that account for syntactic structures. But even if a range of recurring syntactic configurations would enter long-term memory representations, it is the very nature of a recursive combination characterizing natural language that clearly contradicts the notion of fixed repertoire storage.Google Scholar

13. These are scattered examples of phenomena that have been explored in linguistics over the past four decades or so. A recent discussion can be found in Chomsky (1995) The Minimalist Program (Cambridge, Mass: MIT Press).Google Scholar

14. It might be worth noting that the condition of projective and non-projective mapping corresponds precisely to the effect of the two basic syntactic operations Merge and Move, assumed in Chomsky (2000) Minimalist inquiries: the framework. In R. Martin, D. Michaels and J. Uriagereka (eds) (2000) Step by Step (Cambridge, Mass: MIT Press), pp. 89–156. Actually, Merge creates projective hierarchies by combining lexical elements (more technically: elements from a ‘Lexical Numeration’), while Move creates departures by ‘recombining’ elements already combined in structures. As Chomsky shows, Move can be reduced to Merge, applied to already merged elements. Under this perspective, projective and non-projective correspondence derive from the same combinatorial principle, applying merely to different types of memory.Google Scholar

15. It is a frequently invoked, plausible, strategy to refer to the vast complexity of neuronal structures and connections, which ultimately accomplish the intricate processes of language acquisition and/or use. This consideration is, of course, correct in principle, but it must be emphasized that it does not explain the cerebral realization of language, as long as we do not understand how exactly the limits of spreading activation and learning by repetition will be overcome. It does not help, in particular, to construct formal mechanisms that tacitly rely on principles that are shown to be insufficient for the task at hand.Google Scholar