S LX 522 Syntax I Week 4a. The P and geometry of trees Previously, in LX522 Sentences are made of constituents, mselves possibly made of constituents, etc. [The kid [who [dropped [ cookie]]]] [ate [it]] ny account of syntactic knowledge will need to provide this kind of hierarchical structure. Overall idea: The syntactic system we have in our head builds trees, such that all and only trees that system can build are grammatical. To extent that grammatical and ungrammatical sentences are distinguished in terms of hierarchical structure, it must be part of system. Minimalism X -ory s we try to determine what properties of this grammatical system are, we should assume as little as we can get away with. ny language-like system that is going to create hierarchical structure is going to need something that takes two (or more, but let s say that two is simpler than any or number ) things and puts m toger into something eligible for furr combinations. So, machine that builds trees has at least operation Merge. phrase is a syntactic object formed by combining (merging) two syntactic objects, with properties inherited from one of m ( head of phrase). word is a syntactic object. minimal maximal XP YP X specifier X head intermediate ZP complement X -ory In 70s and 80s, se ideas went by name X -ory. maximal very XP has exactly one: head (a lexical item) intermediate complement (anor XP) specifier (anor XP) for any X (,,, P, I, etc.) XP minimal YP X specifier X head ZP complement Radford and X(P) To forestall confusion: is both a minimal and a maximal. It functions as a phrase, an XP, but it has nothing in it but a head, an X. Since you need to write something, Radford generally opts to write X for se X/XPs. eat 1
Radford and X(P) In this class, and on my overheads, I will usually write X/XP as XP. You should do same, but you should be aware that Radford does it differently. In general, this will depend on wher properties we are focusing on are those of phrases (XPs) or heads (Xs). In se ambiguous cases, it will almost invariably turn out that y act like phrases with respect to what we are focusing on. Radford and X(P) nor similar comment pertains to status of IP below. It is an IP. It is not an I. It s true that it will be an I after we combine Pat with IP, but it isn t yet. f. Radford p. 120. Pat I will IP eat eat X, X, XP X, X, XP In nglish, head and complement always seem to come in that order: head-complement. at (P = PP) eat ( = ) will eat (I = IP) ut here, languages differ. nglish is a head-first (or head-initial) language. P at PP In Japanese, head follows complement. Japanese is head-final. ringo-o tabeta ( = ) apple ate toshokan de ( P = PP) library at This seems to be a parameter that distinguishes languages ( head parameter) toshokan PP P de X, X, XP Wher specifier comes before X or after is independent of wher head comes before complement. Specifiers are overwhelmingly initial, although a few languages may be best analyzed as having final specifiers (sometimes). IP.g., Japanese, which is head-final, neverless has initial specifiers. I Ringo-ga ringo-o tabe- I ta arrowing in:? Traditionally, a phrase like is called a noun phrase and written as. What does this imply about structure? What category is? What category is? Which one is head? Where is or one? 2
arrowing in:? There are a couple of problems with this. There s headedness problem The syntactic object that combines with head is complement, not specifier. (ote: There is a way out of this, we ll see it later) Supposing that is a whole P is suspicious, because it can never be modified by anything. Modifiability is a signature? property of phrases. P P! If is not an, it must be a P. It s head-initial, like nglish should be. The can of course be modified (happy ). There are several reasons to think that is a P and not an, even better than se two, which we ll see in due course ( at appropriate P! juncture, in fullness of time). P onsider genitive (possessive) s in nglish: John s hat The student s sandwich The man from ustralia s book The man on hill by tree s binoculars otice that s attaches to whole possessor phrase in last two examples, it isn t even attached to head noun (it s man s book and binoculars, not ustralia s or tree s, after all). This is not a noun suffix. It seems more like a little word that signals possession, standing between possessor and possessee. (Recall, it s a clitic). P It seems to be impossible to have both a s and a determiner. *The building s roof f. The roof of building *The tiger s eye eterminers like and possession marker s seem to be in complementary distribution if one appears, or cannot. You may recall a similar pattern from a couple of weeks ago. What was explanation for *s? The big fluffy pink rabbit *The that rabbit *The my rabbit *very my rabbit Possessors Recursion This suggests a structure like this for possession phrases: The possessor P is in specifier of P. nd of course, this can be as complex a P as we like, e.g., very hungry linguistics student by tree with purple flowers over re. The possessed is complement of. P P student s book nor noteworthy aspect of possessor phrase is its recursive property. The possessor is a P in specifier of P. That means that P possessor could have a possessor too The student s far s book The student s mor s bror s roommate 3
X-bar ory: P X-bar ory: P The student s mor s bror s roommate P P P What do we do with apparently simple s like John or (e.g., Students in class complained bitterly)? re se s or Ps? P s student s mor s roommate bror Well, re are two options One: They are s. Pro: Just as y appear. on: Subjects, objects, etc. can be eir s or Ps. Two: They are Ps. Pro: Subject, objects, etc. are always Ps. on: ot obvious from surface pronunciation. Pronouns X-bar ory: Pronouns onsider: me, you, him (or I, you, he) Since a pronoun can be subject of a sentence (e.g., I left), a pronoun must be part of a P. For pronouns, however, re s some reason to believe that y actually head P. That is, that pronoun I is a. That is, not PR as we d called it up until now, but actually in same category as and s. onsider following: You politicians are all alike. We linguists need to stick toger. The media always mocks us academics. These seem to have a pronoun followed by a noun inside P; we can make sense of this if pronoun is a which can optionally take an complement. we P linguists X-bar ory: are nouns and proper names X-bar ory: are nouns and proper names How about something like (in Students poured out of auditorium at noon) or John (in John went for a walk)? For, we want to believe that it is an instance of category (in order to make sense of or we or John s. ut if this is contained in a P ( complement of a head), where is? In order to maintain consistency, we ll suppose that in bare nouns is present but null (it has no phonological representation; we write this as Ø). So for bare noun, we have a structure like that shown here. s for proper names like Pat, we will assume that y are essentially like. Why? Why not like I, me, m? Ø Ø P P John 4
Trees Trees The joints of tree are nodes. The nodes here are labeled (with node labels). odes are connected by branches. The node at top of tree (with no branches above it) is called root node. is root node. odes with no branches beneath m are called terminal nodes.,, are terminal nodes. odes with branches beneath m are called nonterminal nodes., are nonterminal nodes. node X dominates nodes below it on tree; se are nodes which would be pulled along if you grabbed node X and pulled it off of page. dominates and. set of terminal nodes is a constituent if y are all dominated by same node and no or terminal nodes are dominated by that node. is a constituent. is not. node X immediately dominates a node Y if X dominates Y and is connected by only one branch. immediately dominates and. is also sometimes called mor of and. node which shares same mor as a node X is sometimes called sister of X. is sister of. is sister of. is sister of. 5
X-bar configurations node X c-commands its sisters and nodes dominated by its sisters. XP YP X The complement is sister of head. X ZP The specifier is sister of X that is a daughter of XP. Precedence Precedence That is, is pronounced before, meaning is pronounced before all of terminal nodes dominated by. ven if tree is drawn sloppily, nothing changes (everything dominated by) is pronounced before (everything dominated by). This is still saw. saw saw o line crossing One of implications of this is that you cannot draw a wellformed tree with lines that cross. The can t be pronounced before because The is part of P and has to be pronounced before all of P. meet P 6