CAS LX 522 Syntax I the Y Model

Home , Binding (linguistics)

CAS LX 522 The Y model ¥ We’re now ready to tackle the most abstract branch of Syntax I the Y-model, the mapping from SS to LF. Here is where we have “movement that you can’t see”.

θ Theory Week 10. LF Overt movement, DS Subcategorization Expletive insertion X-bar theory Case theory, EPP SS Covert movement Phonology/ Morphology PF LF Binding theory

Derivations Derivations

¥ We think of what we’re doing when we ¥ The steps are not necessarily a reflection of what construct abstract structures of sentences we are doing online as we speak—what we are this way as being a sequence of steps. doing is characterizing our knowledge of language, and it turns out that we can predict our Ð We start with DS intuitions about what sentences are good and bad Ð We do some movements and what different sentences mean by Ð We arrive at SS characterizing the relationship between underlying Ð We do some more movements thematic relations, surface form, and interpretation in terms of movements in an order with constraints Ð We arrive at LF on what movements are possible.

Derivations Derivations

¥ It seems that the simplest explanation for the ¥ Concerning SS, under this view, languages pick a complex facts of grammar is in terms of several point to focus on between DS and LF and small modifications to the DS that each are subject pronounce that structure. This is (the basis for) SS. to certain constraints, sometimes even things which ¥ There are also certain restrictions on the form SS seem to indicate that one operation has to occur has (e.g., Case, EPP have to be satisfied). before another could.

DS LF DS SS LF

1 Derivations Derivations ¥ Although speaking sloppily we might say that ¥ It’s just that parts that happen between SS and LF movements that happen in the part of the derivation are invisible to the pronunciation because all of the between SS and LF happen “after pronunciation” changes (movements, etc.) that occur between DS this doesn’t imply that in time we arrived at SS, and SS are reflected in the SS representation that we pronounced, and then did further syntactic focus on, and none of the changes that occur computation. between SS and LF are.

DS SS LF DS SS LF

Derivations Quantifiers

¥ Because we can’t see (hear) them, the things that ¥ We interpret Bill saw everyone as happen between SS and LF are more difficult to ¥ For every person x, Bill saw x. detect—we have to rely on somewhat indirect evidence. That’s what we’ll be focusing on today. ¥ This is the meaning. This is the logical form of the sentence Bill saw everyone. In the notation of formal logic, this is written as ∀x. Bill saw x ‘For all x (x a person), Bill saw x.’ DS SS LF

Quantifiers Quantifiers

¥ Every boy hates his roommate. ¥ These phrases which don’t refer to specific ¥ Notice that each boy hates a different people/things in the world but rather seem to do roommate, the roommates are specific to each things to sets of people/things are quantifiers. boy. Examples include: ¥ For every boy x, x hates x’s roommate. Ð most students Ð twelve angry men ¥ This means that every boy doesn’t just mean Ð fewer than half of the members the group of boys; rather it goes through the Ð some custodian set of boys and says something about each of Ð nobody in their right mind them individually.

2 Quantifiers Binding

¥ To write the logical form (meaning) of a sentence ¥ A quantifier is said to bind its variable. That is, the with one of these, you put the quantifier first, and reference of the variable is assigned by the replace where it came from with a variable: quantifier. ¥ Most students eat at Taco Bell. ¥ Bill read every book. For most students x, x eats at Taco Bell For every book x, Bill read x ¥ No administrators eat at Taco Bell. For no administrator x, x eats at Taco Bell ¥ Is this true? Well, let’s go through the books. ¥ Mary likes every flavor of ice cream. Moby Dick. Did Bill read Moby Dick? Yes. Ok, For every flavor of ice cream x, Mary likes x War and Peace. Did Bill read War and Peace? Yes. Ok, …

Scope Scope

¥ A student read every book. ¥ A student read every book ¥ When is this true? There is a student x such that Ð Mary, it turns out, has read all of the books. for every book y, x read y Ð Nobody has read everything, but Mary read or half of the books and Bill read the other half. For every book y, there is a student x Every book was read by a student. such that x read y ¥ There are two meanings here, the sentence ¥ It matters which quantifier comes first in the is ambiguous between two logical forms. logical form.

LF LF

¥ We think about this in much the same way ¥ LF is the structure that we interpret, LF is we think about the structure that represents the logical Mary heard a dog bark in the house. form. Ð (either Mary was in the house or the dog was) ¥ This is a syntactic ambiguity, depending on where the PP in the house is attached. ¥ As syntacticians, we think of this structure ¥ If there are two different interpretations, in the same way we think of DS and SS—as there are two different structures. a tree, with constituents and c-command.

3 QR QR

¥ Sue read every book. ¥ Sue read every book. For every book x, Sue read x. For every book x, Sue read x.

¥ Between SS and LF, the quantifier moves to ¥ [every book]i [TP Sue read ti ]. a position above the sentence, so there is then a direct mapping between the structure and the logical form. ¥ Notice that the trace is the variable at logical form—moving quantifiers is a way ¥ [every book]i [TP Sue read ti ]. to establish a quantifier-variable structure.

QR Quantifiers and binding

¥ Sue read every book. ¥ Every girl aced her exams. For every book x, Sue read x. ¥ [Every girl]i [ ti aced heri exams] ¥ [every book]i [TP Sue read ti ]. ¥ For every girl x, x aced x’s exams ¥ Not only the trace of QR, but also pronouns, can be bound by the quantifier, their ¥ This movement is called Quantifier Raising referent determined by the quantifier. (QR), and it happens to every quantifier between SS and LF.

Quantifiers and binding Why believe in QR?

¥ [Every girl]i [ ti aced heri exams] ¥ The thing is: QR is invisible, it is supposed ¥ Binding (assigning reference) is subject to c- to happen between SS and LF, and since SS command. A quantifier can only assign is (the basis for) the structure we hear, we reference to a variable (its trace and possibly other pronouns) which it c-commands. can’t hear its effects—the effect shows up only in the meaning. ¥ Her brother said that every girl aced her exams. ¥ The things which a quantifier c-commands are ¥ So, is QR just gratuitous formalization? said to be in its scope. Syntacticians just not knowing when to stop ¥ Quantifiers can only bind variables in their already? scope.

4 ∃QR ∃QR

¥ A doctor assisted every patient. ¥ Here’s one reason to think that QR is real, ¥ Which patient did a doctor assist t? that QR is actually syntactic movement and not just figments of the imagination of overeager syntacticians: QR acts like ¥ A doctor spread the rumor that Bill immunized every patient. movement. ¥ *Which patient did a doctor spread the rumor that Bill ¥ We can detect QR by meanings in sentences immunized t ? with multiple quantifiers—i.e. in someone ¥ A doctor wondered who immunized every patient. likes everyone. ¥ Which patient did a doctor wonder who immunized t ?

∃QR ∃QR

¥ LF: *[Every patient]i [TP a doctor wondered ¥ Another reason to believe in QR: [ who [ t immunized t ]]] CP j TP j i ¥ What did Sue buy? ¥ Pick the x such that Sue bought x. ¥ LF: *[Every patient] [ a doctor spread i TP ¥ The interpretation of wh-words has the same [DP the rumor [CP that Bill immunized ti ]]] kind of operator-variable structure that ¥ QR obeys island constraints—like wh-movement, it quantifiers do. can’t get out of a complex noun phrase, it can’t get out ¥ The difference is that wh-movement happens of a wh-island. where we can see it, but it still leaves a variable behind, bound by the quantifier (wh-word).

WCO WCO

¥ There is an interesting property of this kind ¥ But now consider this: of operator-variable formation, which we ¥ Who does his roommate like? can see in wh-movement. ¥ Can this mean the same thing as Whose ¥ Who likes his roommate? roommate likes him? ¥ Pick the x such that x likes x’s roommate. ¥ *Whoi does hisi roommate like ti ? ¥ Whoi [TP ti likes hisi roommate] ¥ How is this different from ¥ Like with quantifiers, it is possible to have a ¥ Whoi ti likes hisi roommate? pronoun bound by a wh-word.

5 WCO WCO

¥ *Whoi does hisi roommate like ti ? ¥ Now, let’s look at quantifiers again. ¥ Who t likes his roommate? i i i ¥ Every girl likes her roommate.

¥ The difference lies in the fact that the wh-phrase ¥ For every girl x, x likes x’s roommate. had to cross over the coindexed pronoun on its ¥ Her roommate likes every girl. way to SpecCP. This appears to be impossible, and we can state this as follows: ¥ For every girl x, x’s roommate likes x. ¥ Weak Crossover (WCO): A coindexed pronoun ¥ Why can’t the second sentence have this cannot intervene between a quantifier and its meaning? variable.

WCO ACD

¥ [Every girl] [ t likes her roommate]. i TP i i ¥ Here’s another reason, antecedent contained ¥ For every girl x, x likes x’s roommate. deletion. This one’s kind of complicated, so hang on tight. ¥ [Every girl] [ her roommate likes t ]. i TP i i ¥ First, we need to talk about VP ellipsis. ¥ For every girl x, x’s roommate likes x. ¥ Mary bought a record, and Bill did too.

¥ Answer: WCO again. But WCO is about moving a ¥[TP Mary -ed [VP buy a record]] and quantifier over a variable—so if WCO rules out this [TP Bill -ed [VP buy a record]] too. meaning, there must have been movement. There must have been QR. A movement we couldn’t see.

VP ellipsis VP ellipsis

¥ Mary bought a record and Bill bought a tape. ¥ We will consider the process of VP ellipsis to be ≠ Mary bought a record and Bill did too. one of deletion under identity. ¥ VP ellipsis is allowed when a preceding VP ¥ DS: -ed [ Mary sleep] and -ed [ Bill sleep] too. is identical. VP VP ¥ SS: ¥ To interpret this, you need to use the Mary -ed [VP t sleep] and Bill -ed [VP t sleep] too content of the preceding VP. ¥ LF: ¥ Mary bought a record and Bill did (buy a Mary -ed [VP t sleep] and Bill -ed [VP t sleep] too record) too. ¥ just after SS on the way to PF: Mary -ed [VP t sleep] and Bill -ed [VP t sleep] too Mary slept and Bill did too

6 VP ellipsis VP ellipsis

¥ So, as long as two VPs in sequence look ¥ Note that identity is actually fairly abstract. identical (where traces of movement look ¥ John slept and Mary will too. identical to one another—they sound the ¥ John slept and Mary will sleep too. same), we are allowed to pronounce the ¥ SS:

second one very quietly. John -ed [VP t sleep] and Mary will [VP t sleep] too ¥ Like an extreme case of ¥ Further support for “affix hopping” being Mary bought a record and “phonological” (after SS, on the way to PF); the V doesn’t inherently have a tense suffix. Bill bought a record too.

ACD ACD ¥ Bill likes every book Mary does. ¥ Now, consider a DP with a relative clause: ¥ Bill [ likes every book Op Mary [ likes t ]]. ¥ the record [whichi Mary bought ti ]. VP i VP i ¥ VP: likes [every book Op Mary likes t ] ¥ Bill likes [the record which Mary bought]. ¥ VP: likes t ¥ Those aren’t the same. VP ellipsis shouldn’t ¥ Bill likes the record which Mary bought and Sue does too. work, but yet it does. ¥ Bill likes the record which Mary bought and Sue ¥ The deleted VP is contained in the antecedent does (like the record which Mary bought) too. VP (antecedent-contained deletion)

QR and ACD LF

¥ But now let’s consider what QR would do. ¥ It looks like there is a syntactic structure ¥ Every book that Mary does is a quantifier. that is crucial for interpretation—it seems to ¥ Quantifiers have to move up past the subject by LF. be derived by movement. LF is the syntactic ¥ Bill likes every book Mary does. representation from which we get the ¥ SS: logical form. Bill [VP likes [every book Opj Mary [VP likes tj ]]]. ¥ LF: ¥ So far, the only “covert” movement [every book Opj Mary [VP likes tj ]]i Bill [VP likes ti ]. ¥ But now the VPs are identical. operation we have seen is QR, but we’ll get ¥ So if we believe in QR, we can explain ACD to another one shortly. sentences in a natural way.

7 Where do quantifiers go? A new position

¥ Every student left. ¥ In order to accommodate this, we need to formulate a new position ¥ [Every student] [ t left ] TP i TP i to which quantifiers move. ¥ We need a variable in subject position, so ¥ This position is going to be QP TP QR must be moving the quantifier out of TP, adjoined to TP. spec T′ to somewhere higher then TP. ¥ This is an extension of the ¥ Believe me that it is also moving concept of adjunction to X-bar T VP somewhere lower than CP. and to X heads, and it looks like this.

A new position A new position

¥ This QP is not in SpecTP, it is kind of ¥ One difference between QR on the edge of TP, hooked on top. (adjunction to TP) and movement to SpecTP is in the motivations. ¥ There is still only one SpecTP. TP TP ¥ Moving to SpecTP or moving to ¥ In fact there is really only one TP node. QP TP SpecCP is motivated by some need of QP TP ¥ QP is not dominated by TP because it is T (EPP: SpecTP must be filled) or C spec T′ spec T′ “partly inside” and “partly outside” of ([+WH] C needs a [+wh] in its TP (there is a segment of TP which T VP specifier). T VP doesn’t dominate QP, hence TP as a ¥ Moving a quantifier (QR) is required whole doesn’t dominate QP). because the quantifier needs to get ¥ QP c-commands the material inside TP, out of the TP (for interpretation). TP but TP does not c-command QP. itself has no need for quantifiers.

A new position The Y model ¥ So, we now have a case for there being some movement ¥ So, we could say that moving to Spec that happens between SS (surface form) and LF is something that happens if the (interpretation). Covert movement. Hidden from view. moving thing is pulled (T is pulling TP But still subject to the rules of syntax and movement. up a subject to satisfy its own needs, not the needs of the moving subject) QP TP or pushed (quantifiers move to satisfy θ Theory spec T′ Overt movement, DS Subcategorization their own needs, not the needs of the Expletive insertion T). X-bar theory T VP ¥ An XP which is pulled up goes into Case theory, EPP SS Covert movement Spec. Phonology/ ¥ An XP which is pushed up adjoins. Morphology PF LF Binding theory

8 Covert wh-movement Covert wh-movement

¥ There is another, somewhat related argument for ¥ Who bought what? covert movement of wh-phrase as well. ¥ Who gave what to whom? ¥ What did Sue buy? ¥ Pick the x and pick the y such that x bought y. ¥ Pick the x such that Sue bought x. ¥ Pick the x and pick the y and pick the z ¥ The interpretation of wh-words has the same kind of such that x gave y to z. operator-variable structure that quantifiers do. ¥ It seems that for interpretation, we need to create ¥ However, the difference is that it looks like this operator-variable chains for each wh-word, yet we movement happens visibly (“overtly”) between DS only create one before SS. and SS. Right?

Covert wh-movement The wh-typology

¥ This suggests that there is covert movement ¥ English: One wh-phrase moves to the front. of the other wh-phrases as well, like QR… Ð What did Bill give to whom? ¥ Japanese: No wh-words move to the front. Ð Taroo-ga dare-ni nani-o ageta no? ¥ LF: [ Who what [ t bought t ]] T-nom who-to what-acc gave Q CP i j TP i j ‘What did Taroo give to whom?’ ¥ Bulgarian: All wh-words move to the front. ¥ LF: someone everyone [ t met t ] Ð Kakvo na kogo Ivan dade? i j TP i j what to whom Ivan gave ¥ LF: everyonej someonei [TP ti met tj ] ‘What did Ivan give to whom?’

The wh-typology The wh-typology

¥ Yet in all of these languages, the meaning of ¥ In this sense, Bulgarian looks like its SS is quite What did Bill give to whom? is the same… close to the necessary LF. ¥ Pick the x, pick the y, such that ¥ Bulgarian: All wh-words move to the front. Bill gave x to y. Ð Kakvo na kogo Ivan dade? what to whom Ivan gave ¥ The meaning has a quantificational (operator- ‘What did Ivan give to whom?’ variable) structure, so if the meaning (logical ¥ Pick the x, pick the y, such that Ivan gave x to y. form) in all of these languages is derived from LF, these quantifier-variable chains must be ¥ We see the wh-phrases moving in Bulgarian. But we know from the meaning that all of these there at LF. movements have to happen in all languages.

9 The wh-typology The wh-typology

¥ We conclude then that languages differ not ¥ We can phrase this in terms of overt and in whether they move all of the wh-words to covert movement: the front of the sentence—they all do—but rather in when they move them there. ¥ Bulgarian: All wh-movement is overt. ¥ Bulgarian: All between DS and SS. ¥ Japanese: All wh-movement is covert. ¥ Japanese: All between SS and LF. ¥ English: One wh-phrase moves overtly, the ¥ English: One between DS and SS, the rest rest move covertly. between SS and LF.

Derivations Derivations

¥ This allows us a fairly uniform view of languages. ¥ So in terms of wh-movement, we can say that the ¥ Across languages, DS is basically the same (theta languages pick different parts of the derivation to requirements, functional projections). focus on as SS. ¥ Across languages, LF is basically the same (meanings, Bulgarian scope, etc.). Japanese English ¥ This means that across languages, the movements that happen between DS and LF are basically the same. ¥ What differs is the timing—whether the movements happen before SS or after SS on the way to LF. LF DS move second move first wh-word wh-word

Scope and wh-movement Superiority

¥ In a wh-question, you have a [+Q], [+WH] C and ¥ Who did Bill persuade to buy what? some wh-words. Ð (Bill persuaded Mary to buy a book, he ¥ The wh-words have to move up to the [+WH] C. persuaded Larry to buy a coffee table, he From here they have scope over the rest of the question. persuaded Sue to buy a futon, …) ¥ Where the [+WH] C is determines the scope of the ¥ Pick the x, pick the y, such that wh-phrases. Bill persuaded x to buy y. ¥ What C did John say t′ that he wanted to buy t ? i i i ¥ So both wh-words (who and what) take ¥ John knows what C he wanted to buy t . i i scope at the matrix clause SpecCP.

10 Superiority Superiority

¥ Who did Bill persuade to buy what? ¥ Superiority: The shortest wh-movements ¥ *What did Bill persuade who to buy? have to happen first. (Wh-movement isn’t possible if there was a shorter one). ¥ It seems that we can’t just choose just any wh-word to move to SpecCP—one works, ¥ Whoi did Bill persuade ti to buy what? one doesn’t. What’s the difference between the two? ¥ *Whati did Bill persuade who to buy ti ?

Where have all Superiority the wh-words gone? ¥ There is also a strict ordering for languages where ¥ This brings up an interesting question which we all wh-words move to the front. We take this to also haven’t addressed yet—if all of the wh-words are be due to Superiority: moving (in all languages, some overtly like in Ð cine ce a vazut Romanian Bulgarian and Romanian), where are they going? who whom has seen ‘Who saw whom?’ Ð *ce cine a vazut ¥ English moves its one wh-word to SpecCP. whom who has seen ‘Whom did who see?’ ¥ There is only one SpecCP. ¥ The higher wh-word has to move first (by ¥ Are the other wh-words not moving to SpecCP? Superiority) and shows up first.

Where have all Where have all the wh-words gone? the wh-words gone? ¥ Following an influential proposal by Rudin ¥ This way, there is still one specifier of SpecCP, but the (1988), many people assume that the other wh-words are still all in the specifier of SpecCP, one attached to the others. (Things are actually more wh-phrases move up and adjoin to the wh- complicated than this, but this is a good approximation phrase already in the specifier. of how one class of languages works—cf. Syntax II!)

CP CP CP CP ′ ′ DP1 C DP1 C who who DP1 C′ DP1 C′ C … C … DP2 DP DP DP2 DP DP what 1 2 C what 1 2 C who what who what

11 Subjacency revisited Subjacency revisited

¥ Who knows where we bought what? ¥ The same as the problem with Ð (Bill does—he knows all about things bought ¥ ??Whati does Bill know where we bought ti ? and places those things were bought from). ¥ Moving what here to the matrix SpecCP Ð (Bill knows where we bought the coffee table, violates Subjacency, so the sentence sounds Mary knows where we bought the futon, …) somewhat off. Yet: ¥ The second reading presents a problem. ¥ Who knows where we bought what? Both who and what take scope at the matrix clause’s SpecCP. What’s the problem? ¥ …sounds fine, even on the “list” reading.

Subjacency revisited Subjacency revisited

¥ These sentences would suggest that covert wh-movement ¥ What’s different? Why is the reading with is not sensitive to wh-islands. A very widely adopted what associated with (moved to) the matrix assumption about Subjacency is made to explain this: SpecCP allowed in one case and not the other? ¥ Subjacency only holds for overt movement.

¥ ??Whati does Bill know where we bought ti ? ¥ Because what in Who knows where we bought what? ¥ Who knows where we bought what? moves covertly, it is no longer subject to Subjacency and can just move directly into the matrix SpecCP.

Subjacency revisited Wh-in-situ languages ¥ Ni xiang-zhidao [shei mai-le sheme]? ¥ This leaves open some loose ends, and you wonder who bought what people have not reached their final ‘What do you wonder who bought?’ conclusions on whether Subjacency does or does not hold of covert movement. ¥ Mary-wa [John-ni nani-o ageta hito]-ni atta no? M-top J-to what-acc gave man-dat met Q Ð For one thing, QR seemed to obey Subjacency. ‘What did Mary meet [the man who gave t to John]?’ Ð On the other hand, it appears that in wh-in-situ (no overt wh-movement) languages, wh-words ¥ These are sentences which are possible in Chinese, are allowed in islands. Japanese, but not in English. They have wh-words inside islands.