In his March 2014 MIT lectures, Chomsky continues to claim that 'internal Merge', which yields the traditionally problematic 'displacement' property, is, in fact, the simplest and most economical structure-building operation possible, since it minimises 'search'. Of course, the latter observation is true to the extent that the 'target' of 'internal Merge' is locally available in a limited domain of the computational system's 'working space' (a previously built syntactic object {X, Y} that functions as the current active 'locus' of the computational operations), whereas in the case of 'external Merge' the 'target' must be searched for, at best in the whole 'working area', or even worse, in the whole of the speaker's lexicon, if initial one-step lexical insertion via 'numerations'/'lexical arrays' is considered stipulative and dispensed with.

However, this 'elegant' argument of Chomsky's has always made me uneasy, because I cannot help asking myself two very naïve questions:

1) Why should it be 'economical' to make a 'copy' of Y if the 'original' Y is already there (in the object {X, Y}) and whatever the computational system may have had to do to Y (possibly nil, if 'no tampering' holds, a highly dubious hypothesis) has already been done to it when the {X, Y} object was formed at the previous 'external Merge' stage? Or, in other words, why doesn't 'internal Merge' violate Economy?

And 2), since a) 'internal Merge' occurs only when some other head Z is attached to {X, Y} by 'external Merge' yielding a new syntactic object {Z, {X, Y}}, b) is triggered by some feature of Z that requires 'licensing' (checking, valuation,...), and c) entails 'looking into' {X, Y}, a previously built syntactic object, and replicating Y, to what extent is it coherent to claim that 'internal Merge' respects 'strict cyclicity'? Or, in other words: Why shouldn't the minimal computational 'cycle' (i.e., a single application of 'external Merge' to two objects X and Y) count as a 'cycle' when it is the 'basic principle' of the Language Faculty that is at stake? [I know that only certain computation 'cycles', i.e., 'phases', are supposed to 'count' for this purpose, but, of course, that is a stipulation, and, as Chomsky insists, any stipulation is a step backwards on our way to genuine science].

Question: Is this reasoning of mine in any way unfair to, or, simply, the result of misinterpretation/misrepresentation of, Chomsky's position?

  • I think the answer is already in your question, cf. "the 'original' Y is already there," "the 'target' of 'internal Merge' is locally available" vs. "in the case of 'external Merge' the 'target' must be searched for, at best in the whole 'working area', or even worse, in the whole of the speaker's lexicon." – Alex B. Mar 29 '15 at 15:55
  • As for 2) - for some people, e.g., Kayne, we CAN use the minimal cycle. – Viridianus Dec 7 '18 at 9:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.