X-bar parameters CAS LX 522 n Many (most?) languages of the world have I something like a basic , an order in which words come in in “neutral” sentences. n English: SVO n Akira ate an apple. Week 5. Head movement n Japanese: SOV n John wa ringo o tabeta. n John top apple acc ate n ‘John ate an apple.’

X-bar parameters X-bar parameters n These two word n Notice that in English, orders work nicely TP TP both V and T are TP TP with X-bar theory head-initial, and in as it stands; the DP T¢ DP T¢ Japanese, both V and DP T¢ DP T¢ difference can be T are head-final. In Akira T VP John VP T Akira T VP John VP T stated in terms of a -ed -ta fact, languages tend to -ed -ta simple parameter V¢ V¢ be consistent in their V¢ V¢ which headedness: differentiates V DP DP V n Japanese has V DP DP V languages as to eat tabe postpositions, C comes eat tabe an apple ringo o after TP in embedded an apple ringo o whether they are clauses… head-initial or n English has prepositions; head-final. C comes before TP in embedded clauses…

X-bar parameters X-bar parameters

n By changing the order of the TP specifier and the X¢, we can get n There are also languages in which the VOS order, and by changing the T¢ DP basic word order is VOS, although they order of both (with respect to are few in number. English) we can get OVS order. T VP ny vehi- [PAST] vavay TP n Malagasy: VOS V¢ n Malagasy: VOS T¢ DP n Nahita ny mpianatra ny vehivavay. n Nahita ny mpianatra ny vehivavay. V DP VP T bïrye- n saw the student the woman n saw the student the woman nahita [PAST] komo n ‘The woman saw the student.’ ny mpia- n ‘The woman saw the student.’ natra V¢ n Hixkaryana: OVS n See how we might generate an X-bar n Kana yanïmno bïryekomo DP V structure of this? n Fish caught boy yanïmno n ‘The boy caught a fish’ kana

1 The problem of VSO X-bar parameters languages n There are quite a number of languages, n So by changing the parameters of head- however, for which the basic word order complement order and specifier-X¢ is VSO. Irish, Welsh, and Arabic are order we can generate the following basic word orders: among them. n SVO (spec-initial, head-initial) (English) n Try as we might, there is no way to set the n SOV (spec-initial, head-final) (Japanese) X-bar parameters to get VSO order—we n VOS (spec-final, head-initial) (Malagasy) have a specifier (the subject) between the n OVS (spec-final, head-final) (Hixkaryana) verb and its complement. n And that’s all…

French French

n In French the sentence is n TP TP French presents a n Je mange souvent des pommes. similar problem; DP T¢ n I eat often of.the apples DP T¢ consider the English n ‘I often eat apples.’ I I T VP n T VP sentence [PRES] The adverb souvent appears [PRES] between the verb and its n I often eat apples. V¢ complement; there is no place to V¢ AdvP V¢ put it in this tree. AdvP V¢ n The adverb often is an n Moreover, it should be basically often V DP in the same place as in English, often V DP adjunct, attached at V¢, eat eat apples given the structural similarity apples as seen here. and the sameness of meaning.

X-bar theory: A sham? Movement

n Consider English yes-no questions… n So is X-bar theory not up to the task of being a universal principle of n To form a question from a statement like: structure, despite its initial promise in n Bill should eat his peas. English (and Japanese and Malagasy n We prepose the modal should to the front and Hixkaryana)? Should we scrap it of the sentence, before the subject. and start over? n Should Bill eat his peas? n Where is should in this sentence? n Answer: No… There is a way we can salvage all the good stuff we’ve gotten from X-bar theory so far…

2 Movement Movement

n All of this suggests that the way to look at n Should Bill eat his peas? this is that we start with the sentence… n There is one position in our sentence structures so far that is to the left of the subject, the one where the n Bill should eat his peas complementizer that goes (C): n …as usual, and if we’re forming a yes-no n I said that Bill should eat his peas. question, we follow this up by moving n This is where we expect should to be. It is, after all, a modal, of category T. It is not a complementizer. should to the position of C. If we can’t move it (in an embedded question, there’s n Also notice that if we embed this question, should stays after the subject, and if is in C: already something in C: if), it stays put. n I wonder if Bill should eat his peas.

Movement French

n Given that things do seem to move around n Jean mange souvent des pommes. TP in the sentence (that is, they start where n Jean eats often of.the apples DS we’d expect them to but we hear them n ‘Jean often eat apples.’ DP T¢ somewhere else), this gives us a way we n If we suppose that the French sentence starts out just like the Jean T VP might “save X-bar theory” from Irish and English sentence, we have the [PRES] French. underlying DS (deep structure) V¢ representation shown here. n Let’s go back and look at French with this AdvP V¢ in mind… souvent PP n What needs to happen to get the V mange correct surface word order? des pommes

French Why does this happen?

n Jean mange souvent des pommes. n Jean eats often of.the apples n ‘Jean often eat apples.’ TP n Why would a language need to move its verb up to n Of course—the V (mange) moves SS tense? DP T¢ up to the T position. n In French, verbs are marked for tense and n This always happens in French Jean agreement—past tense verbs look different from Vi+T VP with a tensed/agreeing verb. mange+[PRES] present tense verbs, which look different from This generally doesn’t happen in V¢ future tense verbs. If the tense information is in T English. ([PRES]), and the verb reflects this, somehow the n Hence, the difference in “adverb AdvP V¢ verb needs to get together with T. position” (really, of course, it’s n French does this by moving the verb to T. verb position) souvent ti PP n English does this by moving T (-ed) to the verb. des pommes

3 Pondering about T Features

n In the DS of every matrix sentence (French or English or anything) there is a TP. n In general, as far as syntax is concerned, we can think of the things at the terminal nodes in our tree as being “bundles of features” or “collections of properties.” n In the example Jean mange souvent des pommes, n The T node has (by definition) the feature “is of the tense marked on the verb is present tense. So, category T” for one. Terminal nodes have categorial we suppose that T was present, which we can features, like [T]. mark as being [PRES], i.e. having the feature for n The T node also has features indicating what kind of “present” on tense. The present tense morpheme tense it is ([PRES], [FUT], [PAST], [-FIN]). in French is Ø, so writing the feature is clearer. n The V node has features indicating its theta-grid, and so forth.

Projections Projections

n When we say the category of the head of an X-bar phrase determines the category of the phrase as a n One consequence of modeling category this way whole (i.e. an N heads an NP, a D heads a DP, and is that an XP node doesn’t have a category so forth), we sometimes refer to this as projection of feature intrinsically, it essentially inherits it from the category feature (property). its head.

n A DP is a DP because the [D] feature of its head n For this reason, an XP (a phrase, a projection) projects up to the phrase level (and through the must always have a head. intermediate bar-levels as well). n For similar reasons, we also assume that an XP n For this reason, XPs are sometimes referred to as can’t have two heads—only one head projects its “projections (of their head)”. features to the XP.

What happens when What happens when V moves to T? V moves to T? n If we think that V moves to TP n The T and V must fuse in TP T in order to get the verb SS some way, retaining the SS together with the tense DP T¢ features of T, since the tense DP T¢ feature, then certainly V feature of T is why the verb Vi+T VP Vi+T VP cannot replace T. T must still mange+[PRES] needed to move up there. mange+[PRES] be there, with its tense V¢ V¢ feature. n In fact the features of T must AdvP V¢ AdvP V¢ still be primary, since a n Moreover, if T were replaced ti PP phrase cannot have two ti PP by V, the TP wouldn’t be a heads and it remains a TP TP any longer, would it? (not a VP).

4 What happens when What happens when V moves to T? V moves to T?

n To show that V attaches to T, TP n This structure that is formed in this but that T remains primary, SS way is a complex head. It’s a head (T) DP T¢ this is drawn in the tree with another head (V) attached to it. T structure like this. T VP n It’s still a T head, it still heads the TP. It just has a V attached to it. Vi T mange [PRES] n We say that V head-adjoins Vi T V¢ n The tree structure shown is the normal (attaches, head-to-head) to T. mange [PRES] convention for drawing this, so we will AdvP V¢ follow this convention. This will require a bit of concentration. This is ti PP one head, there is one T—with a V attached.

What happens when What happens when V moves to T? V moves to T?

n In past years, I experimented with n We should also consider TP introducing a convention of drawing what happens to the VP SS DP T¢ the connection between the “two T’s” T from which the V moved. in the diagram differently to help n This too is still a VP, it must T VP reinforce the idea that it’s not a normal Vi T still have a head. mother-daughter relationship. mange [PRES] n We notate the original Vi T V¢ n I will try to continue this tradition this location of the V by writing t mange [PRES] year, using a double-line (to evoke the (standing for “trace” left AdvP V¢ idea of an extra-strong connection), behind by the original V), ti PP although outside this class you will and we co-index the V and almost always see it drawn as a trace to indicate their regular line. relationship.

What happens when What happens when V moves to T? V moves to T? n Since the VP is still a VP, it TP n TP still gets a [V] category feature What has changed is that the SS original verb is now related SS projected up from its head. DP T¢ to a higher position in the DP T¢ n So the trace is still a verb. tree, and for many purposes, T VP T VP n In fact, there’s no reason to the top copy in the tree is considered to be primary. suppose that any of the Vi T V¢ Vi T V¢ features of the original verb mange [PRES] mange [PRES] AdvP V AdvP V have been removed given that ¢ n What we have created by ¢ moving the verb is a chain of [V] is still there. ti PP ti PP positions in the tree that the n We write it as t, but its content verb has occupied. has not really changed.

5 What happens when Why does V move to T? V moves to T? n When we think of moved n “The verb and tense have to get together” TP elements in SS and LF SS is what I said before, but we can focus this structures, we will often DP T¢ question a little bit more. need to consider the chain of n Think about the English past tense positions; this is usually T VP written like: morpheme, generated in (originating in, at DS) T, which we’ve written as -ed. Vi T V¢ mange [PRES] n ( Vi , ti ) We wrote it this way because it isn’t a AdvP V¢ whole word, it is the regular past tense t PP n referring to the two i suffix that appears attached to verbs. positions held by Vi and ti in the structure here.

Why does V move to T? Why does V move to T? n Similarly in French, regular tense is realized as a suffix on the verb. n In English, the tense affix (e.g., -ed) moves down to the verb rather than the verb moving up to T. n One productive way of thinking about why the verb and tense need to get together is that tense is a verbal n However, the negative marker not blocks this suffix. movement—for reasons that are controversial, n By definition, a verbal suffix can’t stand on its own, but we can state the fact as a stipulation (not it needs a verb to attach to. otherwise derived from our system) like so: n That is, the “need” for the verb and tense to get together isn’t something that the verb needs, it’s n Affix lowering is blocked by the presence of not something that tense needs. A verbal suffix needs a verb to attach to. in English. n If tense is “stranded” with no verb, the result is morphologically ill-formed = ungrammatical.

Why does V move to T? Why does V move to T?

n What happens in negative sentences in English, n We can state the rule like this: then, is that the tense affix is “stranded” up in T; n Do-insertion it can’t lower to the verb because not is “in the When there is no other way to support way”. inflectional affixes in T, insert the dummy verb n Bill -ed not buy cheese. (DS) do into T. n As a “last resort”, English has a rule which salvages this situation by inserting the n Bill did not buy cheese. meaningless verb do to “support the tense n In this sentence, the verb has not moved up to T affix”— do is only there to provide something nor has T moved down to V. And we see no for -ed to affix to. tense suffix on the verb as a result. n Bill did not buy cheese. (SS)

6 Why does V move to T? Why does V move to T?

n English has two special verbs which do move to T, the auxiliary verbs have and be in English. n Notice that if there is something in T already, like a n Bill is sloppily eating apples. modal, then the verb doesn’t move up to T. n Bill is not eating apples. n John might not be eating apples. n *Bill sloppily is eating apples. n And moreover, the verb has no tense inflection. n *Bill not is eating apples. n This all suggests that the view that it is the affix in T n *Bill did not be eating apples. which causes V to move to T. The verb is happy not n Bill has not eaten the apples. to move, but will move when it can in order to help n *Bill not has eaten the apples. T out. n There are requirements on T, not on V.

A word on auxiliaries A word on auxiliaries

n English has two auxiliary (“helping”) verbs have and be, which are not the main verbs of a n The DS of a sentence with an TP sentence but generally serve to indicate auxiliary verb would be DS differences in verbal aspect (progressive, past something like this, where DP T¢ perfect, …). the auxiliary verb heads a T VP VP, and takes the main -ed n These auxiliary verbs are verbs, but they have verb’s VP as its complement. V¢ special properties. Among these properties: they n Notice that we are treating V VP move to T, and they have no theta-roles to the past participle eaten as have just a special kind of verb. assign. V¢ This is good enough for present purposes. V … eaten

A word on auxiliaries English yes-no questions

n Now, let’s go back and think about English yes- n The DS of a sentence with an TP no questions, which we took originally to be auxiliary verb would be SS motivation that movement occurs. DP T¢ something like this, where n Bill will buy cheese. the auxiliary verb heads a Vi+T VP n Will Bill buy cheese? VP, and takes the main have+-ed verb’s VP as its complement. V¢ n What’s happening here? Well, we saw earlier n Notice that we are treating that it is reasonable to think that the modal will, the past participle eaten as ti VP which starts out in T, moves to C in questions. just a special kind of verb. n Willi Bill ti buy cheese? V¢ This is good enough for present purposes. V … eaten

7 English yes-no questions Ø+Q n Why does this movement n Incidentally, lots of languages have an happen? SS overt question morpheme, which adds CP plausibility to our assumption that English n By analogy with the motivation C¢ has a question morpheme in C that is just for V-to-T movement, we will null. take C to hold a special (this Ti+C TP n time silent, or perhaps prosodic) will+Ø+Q Akira ga hon o kaimasita ka? (Japanese) affix that must be joined up DP T¢ n Akira top book acc bought Q with T. This affix is the Bill ti n ‘Did Akira buy the book?’ “question” morpheme, of VP category C, which we can write buy cheese as Ø+Q.

English yes-no questions T to C

n Also notice that if there is an overt n In English, anything that would be in T question morpheme there in English moves to C. So, modals and auxiliaries all (which happens in embedded questions), “invert” around the subject: there is no need to move T to C: n Will Bill buy cheese? n Is Bill buying cheese? n I asked if Bill will buy cheese. n Has Bill bought cheese? n *I asked (if) will Bill buy cheese. n But main verbs never raise to T in English. Consider then: n Did Bill buy cheese?

T to C Negation

n We’ve used negation as a test, like (in fact n Did Bill buy cheese? usually better than) adverbs to see if the verb n Why is there a do there? Before, we only saw do appears to the left (suggesting it has raised, in a in sentences with not, inserted because the tense head-initial language) or to the right (suggesting affix couldn’t “reach” the verb, blocked by not. it has not raised). n What seems to be the case is that if T moves to C n Negation acts a little bit different from adverbs (that is, the past tense suffix -ed in this case), it in a few ways. One way negation acts different also gets too far away from the verb (now Bill is is that negation blocks affix lowering in English but adverbs don’t (in the tree, both come between the suffix and the verb), and Do- between T and V at DS): insertion is required. n Bill did not buy cheese. n Bill never buys cheese. n Bill quickly bought cheese.

8 NegP French negation

n In standard French, the negation of a sentence n A common view of negation is that it has generally involves a morpheme ne placed before its own projection, a NegP, headed by a NegP negative morpheme. For example, the tensed verb and a morpheme pas placed after it, as in: something like this. Neg¢ n Jean ne mange pas des pommes. n Interestingly, negation sometimes comes “in two parts”, with two morphemes Neg n Jean NE eats NOT of.the apples implicated in negation. NegP has in n ‘Jean doesn’t eat apples.’ principle two positions available for n However, English gives us reason to believe negative morphemes, its specifier and its (assuming NegP is in the same place in the tree in head. both languages) that NegP comes between TP and VP: n Standard French ne…pas is an example of this which we’ll look at now. n Bill will not eat apples.

French negation French negation

n A common view of how French n How do we get the correct word TP TP negation looks at DS is like this, DS order? DS with ne being a morpheme of category Neg, heading a NegP DP T¢ DP T¢ n We know that V needs to move to with pas in its specifier. T NegP T, but wouldn’t this yield: T NegP [PRES] n Jean mange pas ne des pommes. [PRES] n For the moment, we won’t pas Neg¢ ? pas Neg¢ concern ourselves with the n You’d think so, yet the facts tell categorial status of pas; clearly it Neg VP us that we actually get: Neg VP must be an XP of some kind itself, ne n Jean ne mange pas des pommes. ne maybe also of category Neg, but V¢ V¢ it never heads the main NegP in a sentence. I’ll write it just as pas in the specifier. V PP V PP

French negation French negation

n Note that we take ne to be a n Suppose, however, that the TP TP verb moves first to Neg, and DS prefix (not a suffix), which means when we create the then moves up to T… T T DP ¢ complex head, the verb DP ¢ T NegP adjoins on the right. T NegP n What will happen first is [PRES] n Now, the verb still needs to [PRES] that the V will head-adjoin pas Neg¢ move to T, but it is attached pas Neg¢ to Neg, creating a complex to the Neg now… so the Neg Neg head… Neg VP moves to T. VP ne n Neg V V¢ Complex heads move as a ne i V¢ unit. You can’t “dis-attach” a

V PP head from a complex head. ti PP

9 French negation French negation n This final movement ends n So, we see that assuming that ne is the head of up with the verb close TP SS NegP in French (with pas in the specifier), and enough to the tense suffix assuming that the verb “stops off” to attach to T to satisfy the requirement DP ¢ Neg before moving (now as a part of the complex Neg head) up to T, we get the right that tense have a verbal T NegP host, while at the same time word order. “taking ne along” to get us Negj T pas Neg¢ [PRES] the right word order. n Note that, since *Jean mange pas ne des pommes is Neg V t n Jean ne mange pas… ne i j VP ungrammatical, we also know that the verb has V¢ to stop off at Neg on the way up.

ti PP

Head Movement Constraint Colloquial French?

n This is an example which motivated the n It turns out that the negation morpheme ne that hypothesis that head movement is constrained we suppose is the head of the NegP projection is by the Head Movement Constraint (or HMC) actually generally optional (or even which says that when a head moves to another preferentially omitted in colloquial French)—yet head, it cannot “skip” over a head inbetween. So, pas doesn’t act any differently (i.e. it doesn’t get the reason the verb stops at Neg is because Neg “picked up” by the verb on the way up to T is between where V began and T. instead of ne). n What this suggests is that colloquial French has n Head Movement Constraint a null morpheme which is the head of A head cannot move over another head. NegP—that pas is still in SpecNegP, but the head is Ø instead of ne.

English negation English negation n A common view of English negation is actually an n extension of this: Many people consider not to be Not doesn’t act this way, though—and in the specifier of NegP, with a null head. often sounds a bit archaic: n Has Bill not bought cheese yet? n However, sometimes English negation does appear to be the head of NegP, when it’s n Hasn’t Bill bought cheese yet? “contracted” as -n’t. n There are lots of interesting questions n Isn’t Bill hungry? about negation in English and other n Notice that when the verb moved to T and then to languages—we can’t pursue them here C, it seems to have carried negation along. any further, but this is a good first approximation to how negation works.

10 Back to VSO Irish n In support of verb movement, consider: n Now, let’s return to the question of VSO order in n Phóg Máire an lucharachán. languages like Irish (remember that?). Recall n kissed Mary the leprechaun that we started off with the observation that n ‘Mary kissed the leprechaun.’ there isn’t any way to “generate VSO order” at DS using X-bar rules because V and O are sisters n Tá Máire ag-pógáil an lucharachán. at DS. n Is Mary ing-kiss the leprechaun n However, now that we have verb movement at n ‘Mary is kissing the leprechaun.’ our disposal, we could certainly derive VSO like n We find that if an auxiliary occupies the verb slot at this: the beginning of the sentence, the main verb appears n DS: Subject Verb Object between the subject and verb—it remains, unmoved. n n SS: Verbi Subject ti Object This suggests that deriving VSO from SVO is on the right track.

VSO order in Irish VSO order in Irish n Where is the verb moving to, though? n Except, consider these: n An bhfaca tú an madra? n The verb ends up to the left of the subject, which in English we took to be movement n Q See you the dog n ‘Did you see the dog?’ to C:

n Will Bill buy cheese? n Duirt mé gur phóg Máire an lucharachán. n A natural thing to suppose is that the verb n Said I that kissed Mary the leprechaun moves to T and then to C in Irish to get n ‘I said that Mary kissed the leprechaun.’ VSO order. n If the verb moves to C, where are an and gur?

VSO order in Irish A VP-internal subject?

n One possibility that this suggests n In English (and German and other languages) if CP is that the verb is only moving to SS there is something in C, the verb doesn’t move T, but the subject is actually lower C¢ there (it doesn’t need to): than T—and we have a place in n Is Bill hungry? our tree which hasn’t been used C TP n Should Bill be hungry? yet, the specifier of VP. T¢ n I wonder if Bill is hungry. n But what about English? We expect that DS looks pretty much n But in Irish, we see an overt complementizer T+Vi VP followed by VSO. the same across languages, so why does the subject seem to start DP V¢ in different places in Irish and English? ti …

11 A VP-internal subject? A VP-internal subject?

n All the students will leave. n Actually, though, there’s some CP CP reason to think that in English the DS n The students will all leave. DS subject originates in SpecVP too, C¢ n *The students will leave all. C¢ contrary to what we’ve been n First of all, all the students looks assuming—and moves to SpecTP. C TP like the basic form—this is what C TP n One argument for this concerns the second sentence means, but the T¢ T¢ the “floating quantifier” all. all has somehow “floated off”. n All the students will leave. T VP n However, if the subject moves T VP n The students will all leave. from SpecVP to SpecTP and if the n *The students will leave all. DP V¢ students can move, leaving all behind, DP V¢ n Where can all be found? then all got left behind in SpecVP. V … V …

A VP-internal subject? The Italian DP n The movement of DPs (like subjects) n Remember earlier (not so long ago, really), will be the topic of next week’s class, CP DS we supposed that proper names could be but this idea the subject appears in SpecVP in Irish (and indeed in C¢ of category D, but yet we observed that in English) is not implausible. some languages, it is possible (even C TP obligatory) to say the Bill (rather than Bill, n Note: For this week’s homework, feel T¢ as we say in English). free to continue drawing your DS as if the subject originates in SpecTP. T VP n Let’s take a look at Italian, which has this property. Since we haven’t talked about the DP V¢ details of “NP” (DP) movement, you need not concern yourself with it… V … yet.

The Italian DP The Italian DP n In Italian, in many cases, there is simply an option n However, there is a difference with respect to (stylistically governed) as to whether you say The the order of adjectives and the noun depending Gianni or just Gianni: on which one you use. n L’ antica Roma n Gianni mi ha telefonato. E’venuto il vecchio Cameresi. n the ancient Rome n Gianni me has telephoned came the older Cameresi n ‘Ancient Rome’ n ‘Gianni called me up.’ n *Antica Roma *E’venuto vecchio Cameresi. came older Cameresi n Il Gianni mi ha telefonato. n ancient Rome E’venuto Cameresi vecchio. n the Gianni me has telephoned n Roma antica came Cameresi older n ‘Gianni called me up.’ n Rome ancient

12 The Italian DP And English? n But this makes perfect sense, if what is happening in the cases where n So, in Italian, there seems to be pretty good there is no is that the N is SS evidence that the N raises to D. moving up to D (just like V moves n In English, adjectives can sometimes be found up to T in the main clause), and DP with proper names, and they precede it: when there is a determiner, the N D¢ n Good old John stays put. n Ancient Rome n L’ antica Roma D+Ni NP n However, in English, we can never have a n the ancient Rome determiner with a proper name (*The Mary). N¢ n For now, all we can conclude is that English lacks n Roma antica *Antica Roma AdjP N¢ a (null affixal) determiner that causes raising, but Italian has it. Later, we might be able to revise n Rome ancient ancient Rome ti … this in light of further discussion.

Wrapup T

n So, what we’ve seen is basically that there T T is an operation of head movement which can take the head of an XP and attach it (head- T adjoin) it to a higher head. T T n This kind of movement cannot skip over T T intervening heads in the structure (HMC). n We’ve seen V-to-T movement, T-to-C T movement, and N-to-D movement as examples of this. T

13