Stephen M. Wechsler (wechsler@mail.utexas.edu)
Fri, 25 Aug 1995 16:19:17 -0500
I am wondering about the proper phrase structure of sentences like the following: 1. Kim left. 2. She left. A priori there seem to be at least four possibilities: (i) phrase / \ word word (ii) phrase / \ phrase phrase (iii) phrase / \ phrase phrase | | word word (iv) phrase / \ word phrase | word Comments: (i) phrase / \ word word (i) violates the ID Principle: by Schema 1, the VP should be a phrase. (ii) phrase / \ phrase phrase An odd consequence of this analysis is that the lexicon must contain lots of phrases. E.g., 'Kim' must be listed in the lexicon as a phrase (or perhaps as a sign, to allow for 'the Kim I know', 'a certain Kim', etc). Also, the sort 'phrase' declares the feature [DAUGHTERS con-struc], so the structure in (ii) seems to be ill-formed, since the lower phrases should each have a HEAD-DTR. Similar considerations rule out (ii'): (ii') phrase / \ phrase phrase | word (iii) phrase / \ phrase phrase | | word word Analysis (iii) seems to be assumed in P&S 95 (see p. 32, fn.32). However, the problem with (iii) is that while Schema 2 licenses the VP, there is no ID schema to license the subject subtree: phrase | word Actually Schema 3 might do here, if we take the COMP-DTRS list to be empty. If so then it seems to be a lucky coincidence, since Schema 3 was introduced to handle VSO order, SAI, scrambling, Warlpiri, etc. On the other hand, it might be considered a nice consequence to have a common and mundane application for Schema 3-- namely every fully saturated phrase which lacks complement daughters (adjuncts, bare plurals, names, pronouns, imperatives ('Leave!'), etc). A last option is: (iv) phrase / \ word phrase | word This seems to work, as far as I can tell, without Schema 3. It is not what P&S 95 assume (see p. 32). One consequence is that 'left' must subcategorize for a sign, not a phrase. Am I missing something here? Comments? --Steve
This archive was generated by hypermail 2.0b3 on Fri Dec 18 1998 - 20:33:03 PST