CAS LX 522 Syntax I Week 10b. P shells Small clauses Last time we talked about small clauses like: I find [ intolerable]. I consider [ incompetent]. I want [ off this ship]. (Immediately!) Let s talk about a few more aspects of small clauses and infinitial complements that might make the discussion more conincing. Small clauses s. infinitial complements All of the small clause examples from last time seem to be able to be paraphrased as examples with infinitial (to be) complements as well. I find [ AP intolerable]. I find [ to be intolerable]. So, we might wonder if small clauses are really just infinitial complements with a silent to be. o be or not to be Suggestie against the idea that small clauses are really camouflaged s, it turns out that there s a difference in meaning between the small clauses and infinitial complements. I found the table to be three-legged. I found the table three-legged. I found him to be 6-feet tall. I found him 6-feet tall. I found her to be amused by cartoons. I found her amused by cartoons. o be or not to be I found the chair comfortable. I found the table three-legged. Small clauses Also, not all small clauses are of this sort, though. Consider: he semantic distinction is ery subtle, but it feels like I found AP means that the only eidence for being AP is a subjectie judgment, nondependently determinable. Still, there is a difference. I saw [him fall]. I saw [them upset]. I saw [her in the garden]. here s no to be missing in any of these. 1
Small clauses So, I saw her in the garden would look like this. j I I P see P her P in the garden ECM finds me to be intolerable. j I P [pres] find i me I P to AP be A intolerable ECM wants for me to eat cake. j I P [pres] CP want C for i me I P to eat cake Passies again Another thing that argues in faor of the ECM analysis of how embedded subjects of small clauses and infinitial subjects check Case features: Recall how the actie sentence ate the sandwich relates to the passie sentence he sandwich was eaten. I P q q eat the sandwich Passies again he sandwich was eaten. In the passie, the erb has had its external q-role remoed (and with it, the ability to check objectie Case), so the heme argument moes into Spec, satisfying the E (and checking Case). j the sandwich i +I was ti P P q eaten tj And, speaking of dolphins I consider them to be intelligent. Consider assigns two q-roles, the external Experiencer q-role, and the internal Proposition q-role. I checks Nom Case with the subject, consider checks Obj Case with them. j I I P [pres] consider i them I P to AP be A intelligent 2
And, speaking of dolphins hey are considered to be intelligent. Passiizing consider remoes the external Experiencer q-role, and the ability to check Obj Case. j they j +I are P P considered ti I P to AP be A intelligent Giing trees to ditransities You may recall our discussion of q-theory, where we triumphantly classified erbs as coming in three types: Intransitie (1 q-role) ransitie (2 q-roles) Ditransitie (3 q-roles) heta roles go to obligatory arguments, not to adjuncts. Giing trees to ditransities You may also recall that we beliee that trees are binary branching, where: Syntactic objects are formed by Merge. here s just one complement and one specifier. And our discoery that subjects should start out within the projection of their predicate, so that q-role assignmens strictly local (assigned to either a complement or a specifier). Giing trees to ditransities Problems continue Fantastic, except that these things just don t fit together. We know what to do with transitie erbs. But what do we do with ditransitie erbs? We re out of space! P I showed Mary to herself. *I showed herself to Mary. I introduced nobody to anybody. *I introduced anybody to nobody. his tells us something about the relationship between the direct and indirect objecn the structure. (What?) 3
Problems continue Some clues from idioms he c-commands the. But how could we draw a tree like that? Een if we allowed adjuncts to get q-roles, the most natural structure would be to make the an adjunct, like this, but that doesn t meet the c-command requirements. * P Often idiomatic meanings are associated with the erb+object complex the meaning deries both from the erb and the object together. We take this as due to the fact that the erb and object are sisters at DS. threw a baseball. threw his support behind the candidate. threw the boxing match. Idioms in ditransities In ditransities, it seems like this happens with the. Beethoen gae the Fifth Symphony to the world. Beethoen gae the Fifth Symphony to his patron. Lasorda sent his starting pitcher to the showers. Lasorda sent his starting pitcher to Amsterdam. Mary took Felix to task. Mary took Felix to the cleaners. Mary took Felix to his doctor s appointment. So and are sisters Larson (1988) took this as eidence that the is a sister to the at DS. Yet, we see that on the surface the comes between the erb and the. Mary sent a letter to. Where is the? It must c-command the, remember. Why is the to the left of the at SS? Where s the? Where s the? Where s the? Where s the? We already know how to deal with this kind of question if what we re talking abous the erb coming before the subjecn Irish, or the erb coming before aderbs in French he answer: he erb moes oer the. But to where? Larson s answer to this is obious, in retrospect. If we re going to hae binary branching and three positions for argument XPs (,, ), we need to hae another XP aboe the P. Since the subjecs in the specifier of the higher XP, that must be a P too. Ditransitie erbs really come in two parts. hey are in a P shell structure. P 4
Where s the? Where s the? he higher erb is a light erb (we ll write it as to signify that) its contribution is to assign the q-role to the subject. he lower erb assigns the q-roles to the and the. Sending a letter to So that coers Mary sent a letter to, by saying there are two Ps, send head-moes from the lower one to the upper one, oer the : gae a book to Mary and a record to Sue. Mary sen a letter to. gae i [ P a book to Mary] and [ P a record to Sue]. P Note: You can also say Mary sent a letter, which is one of the major things Larson was concerned about. Radford proides an analysis of this in the book, we ll come back to it. little Unaccusaties s. unergaties So this is the structure that we came up with to get the word order righn a binary-branching tree. John gae a book to Mary. We determined there must be a little, a light erb, to which the moes. his little assigns the Agent q-role. So English has a in its lexicon that assigns the Agent q-role. P Recall from a week ago that there are two types of single-argument (intransitie) erbs in terms of their theta grid with respect to whether they hae an external q-role to assign or not. Unaccusaties: Hae one, internal q-role. (Sometimes called ergatie too) Fall, sink, break, close Unergaties: Hae one, external q-role. Walk, dance, laugh Last week s tree for fell P shells Let s go back and consider P shells a bin connection with unaccusaties. I P q fall Finite I can check Case Unaccusatie cannot check Case i I P fall he ice melted. he boat sank. he door closed. melt the ice he ice, the boat, the door are all hemes, suggesting that the erbs are unaccusatie the argument starts in object position. P 5
melt P the ice P shells So far, so good. Now, melted the ice. he ice is still heme. he erb is still melt. Uniform heta Assignment Hypothesis (UAH): wo arguments which fulfill the same thematic function with respect to a gien predicate must occupy the same underlying position in the syntax. So the ice must still be a complement of the erb underlyingly. melt P the ice P shells In melted the ice what hae we done? We e added a causer, an agent. caused [the ice to melt]. was the agent of an ice-melting. We e already supposed that the light erb assigns the Agent q-role in ditransities Isn t much of a jump to think of as actually haing a contribution to the meaning, something like CAUSE. P melt the ice P shells melted the ice. So, something like this, where the main erb moes up to the light erb (which we had eidence for in ditransities). Later, will moe to Spec, to check Case and satisfy the E. Why does moe to? We ll assume that does this for a reason analogous to why moes to I (for French erbs, say). Might be uniersal, actually. needs a to moe to it. P shells Note. Een though may carry a causatie meaning, this does not mean thas synonymous with the English word cause. here is a difference in the directness of the causal connection. What really seems closest to is Agent. he water boiled. boiled the water i I +boil the water caused the water to boil cause P Aoiding redundancy So, we hae, which assigns an Agent q-role. We hae Agent q-roles in clauses other than sank the boat and gae a boat to Edward. We also hae an Agent q-role in sentences like ate the sandwich. Are there two ways to assign the Agent q-role? Whaf is the way the Agent q-role is assigned? What would ate the sandwich look like? ate the sandwich P eat the sandwich Well, we already saw essentially what would look like. It looks just like melted the ice. assigns Agent to, (eat) assigns heme to the sandwich. Also note: he subjecs still in SpecP except that we e sharpened our picture of what P is. A P with an Agens really a and a P. 6
lied. lied? P lie In fact, things get weirder Consider lied. hat s got an Agent, so it s got a. So, it could look like this. But lie is really (also?) a noun, right? Is this a coincidence? (How about danced, walked, sneezed, ) NP lie One proposal out there about this kind of erb is that really is built from the noun. has, we would hae +N, which would come out to mean something like was the agent of a lie. If that s right, it means really is its own thing, and moreoer, it s responsible for giing these erbs their erby nature. Radford rolling the ball down the hill Radford introduces the idea of with the sentences We rolled the ball down the hill and he ball rolled down the hill. Roll is an unaccusatie erb in he ball rolled. Like he ice melted or fell. i he ball I P roll Radford rolling the ball down the hill For he ball rolled down the hill, we need to add a, which Radford does this way. he ball is still getting the only q-role that roll has. But note thas now in the specifier of P. i he ball I P roll down the hill Radford rolling the ball down the hill If we allow the ball to get the heme q-role in the complement of in he ball rolled and the specifier of P in he ball rolled down the hill, then we hae to think about about what UAH is really supposed to mean. i he ball I P roll down the hill Radford rolling the ball down the hill Let s think of it this way: For a that assigns a heme q-role, the first Merged with it gets the heme q-role. he down the hill is Merged before the ball, busn t a. i he ball I P roll down the hill 7
UAH? UAH In fact, we also supposed that the heme q-role was assigned to the specifier of P in gae a book to Mary. Uniform heta Assignment Hypothesis (UAH): wo arguments which fulfill the same thematic function with respect to a gien predicate must occupy the same underlying position in the syntax. he limited eidence we hae so far seems to suggest that s are Merged firsf there is a to be Merged. (Notice too than gae a book to Mary, the is getting a q-role, bus still Merged first) P So, the same underlying position in the syntax is really to be understood as being the same from the perspectie of Merge e.g., the first Merged with the predicate. Giing Mary a book In this connection, we might now be able to understand a little bit about gae Mary a book. Notice that here, there s no, and the order of the arguments appears to be reersed. But maybe it still satisfies the UAH with respect to the heme q-role, since the heme is still the first Merged with. (What to say about the Goal is less clear not the first?) P IO 8