2 A not-quite-argument for X-bar structure in noun phrases

Transcription

1 CAS LX 321 / GRS LX 621 Syntax: Introduction to Sentential Structure ovember 16, and pronouns (1) he linguists yodel. (2) We linguists yodel. (3) hey looked at us linguists. (4) hey looked at linguists. C C C C /0 decl /0 decl we yodel yodel linguists linguists hese constructions are kind of rare; normally, pronouns are just containing no s. we you m RO who 2 A not-quite-argument for X-bar structure in noun phrases he canonical X-bar structure of early 1980s looked like this: X Y specifier Z adjunct X X head X W complement 1

2 We almost had an argument for an,, structure. If we say determiners go in specifier of, n one must replace, and is recursive based on ability to attach arbitrarily many adjectives between determiner and head noun. hat is: re must be a projection (node) between and that is recursive. So,. big red book of poetry We lost that argument, though, when we decided that determiners head a higher. big red book of poetry he difference is where adjuncts (modifiers, like adjectives and adverbs) attach. o y attach (recursively) to X or to X? We don t really have any evidence yet for a X level from this, nor really any evidence for adjoining adjuncts anywhere by to X level. Where I think we re going to end up: here are X nodes (and refore specifiers), but adjunction is to X and not to X. 3 ossessors and subjects and X here s at least an argument from simplicity to suppose that re is a specifier, and an X level, of. And from re, it s a short leap to suppose that re s a in, and we ll see one for C as well. 2

3 his was mostly in previous handout. But basic point is that: seems like a head, and re s an X to its left that can be eir a possessor or subject of nominalized clause. And if that s where subject is in, seems reasonable that it s in same place in. Also, to a certain extent we can reveal a constituent within that excludes subject but includes head and complement. (5) at always eat lunch and might never eat dinner. (6) Each boy eat his lunch and might eat his dinner. at Adv always eat lunch Maybe slightly more convincing is fact that re seems to be a parameterization of order of specifiers and complements. his makes more sense if initial really refers to a constituent containing head. Head-initial (English, French) vs. Head-final (Japanese) Spec-initial (everything except maybe ASL C) 4 and measure phrases Most of s that we have seen so far are just a head and a complement. o real evidence of a node. into woods Measure phrases might arguably go into specifier position of a, though. (7) at drove two miles into woods and through a puddle. (8) he car rolled 300 feet down road and up hill. (9) he illness began three days after festival and before cruise. 3

4 Meas 300 feet down road (10) he drone flew 10 feet above fence. (11) he drone flew right above fence. (12) * he drone flew 10 feet right above fence. (13) * he drone flew right 10 feet above fence. 5 X-bar ory We have to make a call about X policy. We will adopt following: hrases (X) have a unique head (X). Complement (Y) is sister to head (X). Specifier (Z) is sister to X and daughter of X. hrases (W) can be adjoined to phrases (X).? hrases (W) can be adjoined to X.? here is always an X regardless of wher re is a specifier.? Branching is maximally binary. olicy: O, adjuncts only to X olicy: O, X can be omitted olicy: O, ternary allowed Binary branching is not required, but it seems to be at least close to a true generalization. wo places that we can t do binary branching yet (without more thought at least): coordination with and, and ditransitive verbs (like put). Why might we want to say that re is always an X? If we re defining phrase structure rules, we need an rule as well as and rules. So, anything without a specifier is structurally ambiguous. And what is simpler? Fewer rules or smaller trees? articularly if we don t have any evidence of being able to adjoin at X, we don t really need X to be re. Except kind of incidentally. It s not a simple choice. In interest of keeping trees readable, we ll go with X -only-when-needed version, though we can also think of it as a shorthand (that is, if you ever write, re is an implict re as well, just not written). 4

5 6 Auxiliaries (14) I eat. (15) I have eaten. (16) I be eating. (17) I have been eating. How do we do this? his is actually a kind of hard problem. (18) I eat. (19) * I have eaten. (20) * I have en eat. (21) * I be eating. (22) * I be ing eat. hey have to be in a specific order, but y don t always have to be re. C-selection? Clumsy, baroque, seems wrong. How do we do it? here seems to be an ordering that we can t capture with existing technology. Wait, we had anor one too. (23) he big red hat. (24) * he red big hat. We don t know why those have an order eir, it just seems to be smoething like: (25) value > Size > emperature > Age > Shape > Color > origin > Material (26) his is made from beautiful small cold 3-year-old square blue Italian silver coins. Something like. So, let s say re s a similar kind of externally imposed ordering here; (27) modal/to > have+en > be+ing > be+en > verb (28) modal/to > perfective > progressive > passive > verb (29) Sprouts have been being eaten. 7 egation 8 Yes-no questions 9 Wh-questions 5