1

Unit 8

CP and PRO, Raising, ECM, and more

Morfosintaxis Inglesa302

2

Types of sentences

� Sentences come in several types. We’ve mainly seen declarative clauses.� Horton heard a Who.

� But there are also questions (interrogative clauses)…� Did Horton hear a Who?

� Who did Horton hear?

� …exclamatives…� What a crazy elephant!

� …imperatives…� Pass me the salt.

2

3

Declaratives and interrogatives

� Our syntactic theory should allow us to distinguish between clause types.

� The basic content of…� Phil will bake a cake.

� …and…� Will Phil bake a cake?

� …is the same. Two DPs (Phil, nominative, and a cake, accusative), a modal (will), a transitive verb (bake) that assigns an Agent θ-role and a Theme θ-role. They are minimally different: one’s an interrogative, and one’s a declarative. One asserts that something is true, one requests a response about whether it is true.

4

Clause type

� Given this motivation, we seem to need one more category of lexical items, the clause typecategory.

� We’ll call this category C, which traditionally stands for complementizer.

� The hypothesis is that a declarative sentence has a declarative C in its structure, while an interrogative sentence (a question) has an interrogative C.

3

5

Complementizers

� The reason for calling this element a complementizer stems from viewing the problem from a different starting point.

� It is possible to embed a sentence within another sentence:� I heard [Lenny retired].

� And when you embed a declarative, you generally have the option of using the word that.� I heard that [Lenny retired].

� So what is that that?

6

What’s that?

� We can show that that “belongs” to the embedded sentence with constituency tests.� What I heard is that Lenny retired.

� *What I heard that is Lenny retired.

� There’s a demonstrative that, but that’s not what that is.� *I heard this Lenny retired.

� So, that is its own kind of thing. It’s an introducer of embedded clauses, a complementizer.

4

7

Complementizers

� There are a couple of different kinds of complementizer. That is for embedding declarative sentences.� I understand that Phil will bake a cake.

� It’s also possible to embed an interrogative sentence, like so:� I wonder if Phil will bake a cake.

� I wonder whether Phil will bake a cake.

� Here, if and whether serve as complementizers, introducing the embedded interrogative.� I wonder about the answer to Will Phil bake a cake?

8

Selection� Just like the verb bake takes the DP a cake as its object,

some verbs take whole clauses as their object.

� Some verbs specify what kind of clause they take:� I claimed that Phil will bake a cake.

� *I claimed if Phil will bake a cake.

� *I wondered that Phil will bake a cake.

� I wondered if Phil will bake a cake.

� This is a matter of selection. Some verbs select for declaratives, some verbs select for interrogatives. Some verbs can take either, some neither.� I know that Phil will bake a cake.

� I know if Phil will bake a cake.

� *I washed that Phil will bake a cake.

� *I washed if Phil will bake a cake.

5

9

C

� So, we have lexical items like that and if, which are complementizers (category: C), and have a value for clause type.

� that [C, clause-type:decl, …]

� if [C, clause-type:Q, …]

� Where is it structurally? We know it forms a constituent with the clause it introduces. We know that verbs can select for different kinds of C. The natural conclusion is that it is a sister to TP, at the top of the tree, which projects.

10

CP

� C is the head of CP.

� Saying this also provides a natural explanation of why in SOV languages, complementizers are generally on the right.

� Hanako-ga [Taroo-ga naita to] itta.H.- nom T. -nom cried that said‘Hanako said that Taro cried.’

vP

T′

T

TP

DPSubject

CP

Cthat

6

11

CP� C specifies the clause type; that indicates a

declarative clause.

� Why then can you say either of these?� Jack claimed that Jill fell.

� Jack claimed Jill fell.

� In Spanish and, probably, most other languages you don’t have the option to leave out the C.� He dicho que estaba enferma

� *He dicho estaba enferma

� Claim doesn’t embed interrogatives.� *Jack claimed if Jill fell.

� So Jill fell is declarative in Jack claimed Jill fell.

12

Ø

� Where does that leave us?� Jack claimed Jill fell

� Claim only takes declarative complements.

� Jill fell is declarative.

� Clause type is a feature of C.

� Thus: There is a declarative C.You just can’t hear it.

� English has two declarative complementizers. One is that, one is Ø. In most cases, either one works equally well.

7

13

Jill fell is a declarative

� But hold on a minute. Jill fell, just as its own sentence (not embedded) is also declarative.� Cf. Did Jill fall?

� So, we’ll suppose that since the function of C is to mark clause type, there’s a C in simple sentences as well.

� The C that heads the whole structure has somewhat special properties. Declarative C in that position is never pronounced. Interrogative C is not pronounced as a word, but makes its presence known by causing movement.

14

SAI in YNQs

� In yes-no questions, the subject and auxiliary “invert” (Subject-Auxiliary Inversion):� Scully will perform the autopsy.

� Will Scully perform the autopsy?

� Assuming everything we’ve got so far:� T (will) has a [uD*] (EPP) feature to check,

so Scully is in SpecTP.

� The question is an interrogative.

� (Unpronounced) C is to the left of TP.

� So what must be happening in yes-no questions?

8

15

T-to-C

� A natural way tolook at this: T is moving to C.

� Just like V moves to v,or like Aux (Perf, Prog, orPass) moves to T, or like N moves to n.

� In (main clause) questions, T moves to C.

vP

T′

Twill

TP

DPScully

CP

C[clause-type:Q]

16

T-to-C

� Specifically:

� Suppose T has anuninterpretable featurethat matches a featureof C: [uclause-type:].

� Suppose that when C values [uclause-type:] as Q, the uninterpretable feature is strong.� Cf. When T values [uInfl:] on Aux (Prog, Perf, Pass),

the feature is strong, and Aux moves to T.

vP

T′

Twill

TP

DPScully

CP

C[clause-type:Q]

9

17

Declarative� A simple declarative clause

would look like this.

� The YNQ would be formed by replacing the declarative C with an interrogative C.

Vperform

VP

DPthe autopsy

v′

v

vP

<DP>

T′

Twill

TP

DPScully

v<V>

CP

CØ

[clause-type:Decl]

18

YNQ� In a YNQ, the

[Q] feature of C matches and values the [uclause-type] feature of T as strong ([Q*]).

� T moves up to adjoin to C, checking the feature.

Vperform

VP

DPthe autopsy

v′

v

vP

<DP>

T′

<T>

TP

DPScully

v<V>

CP

CØ

[Q]

C

Twill

[Q*]

Abbreviations:[Q] = [clause-type:Q][Q*] = [uclause-type:Q*][uclause-type] = [uclause-type:]

10

19

YNQ� If T isn’t a

modal, but just a past or present tense marker, vis no longer the head of T’s sister.So we pronounce do: Did Scully perform the autopsy?

Vperform

VP

DPthe autopsy

v′

v

vP

<DP>

T′

<T>

TP

DPScully

v<V>

CP

CØ

[Q]

C

T[tense:past]

[Q*]did

20

Embedding questions� So, you can embed declaratives…

� I heard (that) Jill fell.

� …and you can embed questions…� I asked if Jill fell.

� Notice that the main clause is different:� If the topmost C is interrogative, we get SAI. If the

topmost C is declarative, it is pronounced Ø.

� If an embedded C is declarative, it can be pronounced either as Ø or as that. If an embedded C is interrogative, C is audible (if) and no SAI.

� So, T moves to C only in main clause interrogatives. [uclause-type:] is strong only when valued as Q by a main clause C.

11

21

Nonfinite clauses

� Some verbs embed finite declaratives, as we have seen: I heard (that) Jill fell.

� There are other verbs that embed nonfinite clauses. These come in a few types, but we’ll start with the try type.

� Scully tried to perform the autopsy.

� This is two clauses: Scully tried something, and what it was was to perform the autopsy.

22

θθθθ-roles

� Scully performed the autopsy.

� Scully tried to perform the autopsy.

� The verb perform has an Agent and a Theme, here Scully and the autopsy, respectively.

� The verb try also has two θ-roles, an Agent (the one trying) and a Theme (the thing attempted). Suppose that the Theme of tryis [to perform the autopsy] here.

12

23

θθθθ-roles

� Scully performed the autopsy.

� Scully tried to perform the autopsy.

� In the second sentence, Scully is both the one trying and, if you think about it, the one performing the autopsy. The same individual is the Agent of both.

� Agent θ-roles are assigned to the DP that is Merged into SpecvP.

� However: You are not allowed to assign two different θ-roles to the same DP. Otherwise, it should be possible for Scully admires to mean Scully admires herself.

24

PRO

� Scully tried to perform the autopsy.

� So, we have something of a problem here. We need an Agent DP in the vP for perform, and an Agent DP in the vP for try. But it appears as if there is only one DP around, Scully.

� What to do? Once again gritting our teeth, we resolve ourselves to the fact that we need two DPs and can only see one— therefore, there must be a DP we can’t see.

� The DP we can’t see, we call PRO.

13

25

Control� Scully tried [PRO to perform the autopsy].

� PRO is a DP that is the Agent of perform, Scully is a DP that is the Agent of try.

� It is impossible to actually pronounce an Agent for perform.� *Scully tried [Mulder to perform the autopsy].

� The PRO Agent of perform must be interpreted as being the same person as the Agent of try.� PRO is a little bit like an anaphor in this respect; this fact is

similar to the fact that herself in Scully admires herself must refer to Scully.

� This obligatory co-reference goes by the name control. Scully controls PRO. Sentences with PRO in them are often called control clauses.

26

PRO

� So why is it impossible to say this?� *Scully tried [Mulder to perform the autopsy].

� The answer we’ll give is that nonfinite T (to) does not have a case feature.

� Finite T has a [nom] feature which matches, values, and checks the [case] feature of the subject, checking itself in the process.

� Nonfinite T has no case feature at all, so Mulderwould be left with its case unchecked.

14

27

Null case

� As for PRO, it is a DP so it has a [case] feature. If Mulder can’t get its case checked by the nonfinite T, how does PRO get its case checked?

� A standard (and perhaps less than completely elegant) way to look at this:� PRO is special, it can only “show up” with “null

case” (ucase:null).

� Null case is special, it is only allowed on PRO.

� Control clauses are special, they are introduced by a null C that has a [null] case feature, which can check the [case] feature on PRO.

28

Try

� So, try embeds a nonfinite CP, headed by the special null C with the [null] case feature.

� In turn, the subject must be PRO, in order to successfully check that feature of C.� If the [case] feature of any other DP is valued and

checked as [null], the derivation crashes: only PRO can have null case.

� The embedded clause must be nonfinite (T can’t itself have a [nom] feature).� If the [nom] feature of T checks the [case] feature of

the subject, nothing is left to check C’s [null] feature.

15

29

Try

� Here, the [null] feature of C will match, value, and check the [case] feature of PRO, checking itself in the process.

Vperform

VP

DPthe autopsy

v′

v

vP

<DP>

T′

Tto

TP

DPPRO

[case]

v<V>

CP

CØ

[null]

VP

Vtry

30

Believe

� Another place where nonfinite clauses can be embedded is under the verb believe.� I believe [him to be innocent].

� Here, we have an accusative subject, and a nonfinite T that is not capable of checking case.

� How is the (accusative) case of him checked?

� This relates to the fact that believe can also simply take a DP object:� I believe him.

� So, how is the accusative case of him checked here?

16

31

ECM

� The idea is that believe (actually the v that combines with the V believe) has an [acc] feature that can check the case of him in I believe him.

� Suppose that believe can either have a DP or a TP as its complement.

� What do we expect?

32

ECM

� Nonfinite T cannot check the case feature of him. But the higher v of believecan.

� Checking the case of a subject “from above” like this goes by the name Exceptional Case Marking (ECM).

Vbelieve

v′

v

vPbe innocent

DPI

T′

Tto

TP

DPhim

[case]

v[acc]

VP

<V>

vP

17

33

Arranging to leave

� A somewhat similar phenomenon occurs with verbs like arrange.� Harry arranged for Tom to leave MI-5.

� Here, we have:� Nonfinite T (to), which cannot check case.

� An overt subject (Tom) in the accusative.

� The word for, which we classify as C.

� For, as a P, checks accusative case (He baked a cake for her). If the C for also has an [acc] feature, it could check the [case] feature on Tom.

34

Arranging to leave� Harry arranged for Tom to leave MI-5.

� So, arrange-type verbs can take a CP complement.

� Notice that it is also possible to say� Tom arranged PRO to leave MI-5.

� But this is expected.� Nonfinite T (to), cannot check case.

� The null C with [null] case can check the case of PRO.

� An overt subject can’t get null case: *Harry arranged Tom to leave MI-5.

� PRO cannot get anything but null case: *Tom arranged for to leave MI-5.

18

35

arrange

� Here, the [acc] feature of C will match, value, and check the [case] feature of Tom, checking itself in the process.

Vleave

VP

DPMI-5

v′

v

vP

<DP>

T′

Tto

TP

DPTom

[case]

v<V>

CP

Cfor

[acc]

VP

Varrange

36

Summary

� Complementizers indicate clause type (that/Øfor declaratives, if/whether for interrogatives).

� Some verbs embed clauses. Finite clauses are always CPs.

� Some verbs can embed nonfinite clauses, some embedding TP and others embedding CP.� Believe (expect, …) embed TP and check accusative

case (ECM verbs).

� Try (want, …) embed CP. This can either be:� C[null], checking null case on PRO.

� for[acc], checking acc case on an overt subject. Not all verbs allow this option (want does, try doesn’t).

19

Raising, etc

38

Sentences inside sentences

� So far, we have looked at embedded sentences.

� Embedded sentences can be finite:� Shannon claimed [CP that she could catch a

fish].

� Or nonfinite:� Michael wants [CP PRO to leave].

� Jin wants [TP Michael to return the watch].

� Sun arranged [CP for him to return the watch].

20

39

Embedded clauses� Embedded finite clauses are CPs, with a

complementizer (that or Ø).� Shannon claimed [CP that she could catch a fish].

� Shannon claimed [CP Ø she could catch a fish].

� Embedded nonfinite clauses have to as T, and can be CPs or bare TPs— the distinction is determined by case properties of the verb.� Michael wants [CP ØNULL PRONULL to leave]

� Jin wantsACC [TP MichaelACC to return the watch].

� Sun arranged [CP forACC himACC to return the watch].

� Nonfinite to does not assign case, so the subject must get case (have its [case] feature checked) in some other way.

40

Seems

� Now, we’ll turn to another kind of embedded nonfinite clause.� Charlie seems [to dislike bees].

� This looks a little bit like:� Charlie tried [to sneak away].

� Which is really:� Charlie tried [PRO to sneak away].

� Charlie is the Agent of try.

� PRO (=Charlie) is the Agent of sneak.

� So, what about Charlie seems to dislike bees? What θ-roles go to Charlie?

21

41

Charlie seems to receive(just) one θθθθ-role

� Seems can also embed a finite clause, so consider the pair:� Charlie seems to dislike bees.

� It seems that Charlie dislikes bees.

� The it in the second sentence is the same it we find in It rained. It does not get a θ-role, because rain doesn’t have any θ-roles. We only have itthere because sentences need subjects (EPP: T has a [uD*] feature).

� So what θ-roles does seem assign?

42

Seem seems to assign(just) one θθθθ-role.

� What seem (and appear) mean when paired with an embedded sentence is that the proposition expressed by the embedded sentence appears true.

� There’s only one participant in a seeming, the Proposition.� It seems [that seem assigns one θ-role].

� So, seem assigns a Proposition θ-role (structurally, to its sister, the CP daughter of V′), and nothing else (hence, it is needed to check the EPP feature).

22

43

Back to Charlie

� It seems [that Charlie dislikes bees].

� Charlie seems [to dislike bees].

� These two sentences mean basically the same thing.

� Dislike assigns two θ-roles, we might say Experiencer and Theme.

� It’s the same verb dislike in both sentences. So, we presume that the bottom of both trees will look the same…

44

Disliking bees� Starting with It seems that Charlie dislikes bees, we

would build a vP that looks like this:� V (dislike) assigns a Theme θ-role to the DP bees.

� vExperiencer assigns an Experiencer θ-role to the DP Charlie.

Vdislike

VP

v′

v

vP

vExperiencer <V> DP

NPbees

DØindef

DP

NPCharlie

DØproper

23

45

Disliking bees� And then we add T and C to get that

Charlie dislikes bees…� The [case] feature of Charlie is valued and

checked by the [nom] feature of T.

� The [uInfl:] feature of v is valued and checked by T: [uInfl:pres3sg].

� The [uclause-type:] feature of T is valued and checked by the [clause-type:Decl] feature of C.

Vdislike

VP

DPbees

v′

v

vP

<DP>

T′

T[pres]

TP

DPCharlie

v<V>

CP

Cthat

46

Disliking bees� And then we add

the main clause (seem, v, T, it, C)

vP

T′

T[pres]

TP

DPCharlie

CP

Cthat

Vseem

VP

vP

v

v<V>

T′

T[pres]

TP

DPIt

CP

CØ

Vdislike

VP

DPbees

v′

v

vP

<DP>

v<V>

24

47

Disliking bees

vP

T′

T[pres]

TP

DPCharlie

CP

Cthat

Vseem

VP

vP

v

v<V>

T′

T[pres]

TP

DPIt

CP

CØ

Vdislike

VP

DPbees

v′

v

vP

<DP>

v<V>

� Now, consider Charlie seems to dislike bees.

� First, does Charlie get a θ-role from seem?

� Well, no. Seem only assigns the one θ-role.

� So, unlike in Charlie tried [PRO to elude the bees], we have as many DPs as we have θ-roles.

48

Disliking bees

vP

T′

T[pres]

TP

DPCharlie

CP

Cthat

Vseem

VP

vP

v

v<V>

T′

T[pres]

TP

DPIt

CP

CØ

Vdislike

VP

DPbees

v′

v

vP

<DP>

v<V>

� Charlie seems to dislike bees.

� So, what θ-role doesCharlie get?

� Still seems to be the Experiencer of dislike.

� So, suppose that Charliestarts out in the same place, SpecvP.

� But now, after building vP, we add a nonfinite T…

25

49

Disliking bees

� So, we have Charlie to dislike bees…� The [uInfl:] feature of v is valued and

checked by T: [uInfl:none].

� Nonfinite T has no [uclause-type:] feature.

� The [case] feature of Charlie is still unchecked, since nonfinite to has no case feature.

Vdislike

VP

DPbees

v′

v

vP

<DP>

T′

Tto

TP

DPCharlie

v<V>

50

Disliking bees

� Can we add a C to this?

� Let’s assume not, by the following reasoning:

� The only C that is compatible with a nonfinite T is ØNULL, that assigns null case to PRO. Charlie is not PRO, so it can’t get null case. So, this is just a TP, not a CP.

Vdislike

VP

DPbees

v′

v

vP

<DP>

T′

Tto

TP

DPCharlie

v<V>

26

51

Disliking bees

� So, we add seem, taking our TP (Charlie to dislike bees) as its Proposition complement.

Vdislike

VP

DPbees

v′

v

vP

<DP>

T′

Tto

TP

DPCharlie

v<V>

Vseem

VP

vP

v

v<V>

52

Disliking bees

� We add T…

� Charlie has [case] to check.� Checked ([nom]) by T

� T has [nom], [uD*], and [uφφφφ:] features to check.� [nom] checked valuing case on

Charlie. [uφφφφ:3sg] matches [φ:3sg] feature on Charlie. [uD*] remains.

� seem (v) has [uInfl:] to check� [uInfl:pres3sg], valued by

[tense:pres] and [uφφφφ:3sg] on T.

VP<V> bees

v′

V+vdislike

vP

<DP>

T′

Tto

TP

DPCharlie

Vseem

VP

vP

v

v<V>

T′

T[pres]

27

53

Disliking bees� Finally, we move

Charlie up to check the EPP ([uD*]) feature of T.

� (Subject (-to-subject)) Raising

VP<V> bees

v′

V+vdislike

vP

<DP>

T′

Tto

TP

<DP>

Vseem

VP

vP

v

v<V>

T′

T[pres]

TP

DPCharlie

54

Idioms

� Recall our idea about idioms: For something to have an idiomatic interpretation (an interpretation not literally derivable from its component words), the pieces need to be very close together when initially Merged.� Ortega took a dive.

� Now, we have idiomatic interpretations here:� It seems that the jig is up.

� It seems that the cat is out of the bag.

� It seems that the cat has your tongue.

28

55

Idioms� If pieces of the idiom move away after the original

Merge, we can still get the idiomatic interpretation:� [The cat]i seems ti to have your tongue.

� [The cat]i seems ti to be out of the bag.

� [The jig]i seems ti to be up.

� The important thing is that they be originally Merged together (the θ-role needs to be assigned by the predicate to the noun). Compare:� [The cat] tried to have your tongue.

� [The cat] arranged to be out of the bag.

56

Other raising verbs

� So far, we’ve only talked about seem, but there are a couple of other raising verbs as well.� [The cat]i is likely [TP ti to be out of the bag].

� [The cat]i appears [TP ti to have his tongue].

� [The jig]i proved [TP ti to be up].

� [The cat]i began [TP ti to get his tongue].

� What these verbs (in this use, anyway) have in common is that they have no external θ-role and an internal Proposition θ-role.

29

57

There seems…

� We also find seem with there.� There. The other expletive subject.

� Vincent seems to be lost.

� It seems that Vincent is lost.

� There seems to be a dog in the woods.

� It is an expletive subject that checks both the EPP and case features of T. There checks only the EPP feature of T (a dog checks T’s case feature).

58

*There seems a man to bein the garden.

� There seems to be a man in the garden.

� There appears in SpecTP, satisfying the EPP feature.

� There are two TPs here, and each TP has/had an EPP feature.� [TP There seems [TP to be a man in…]]

� So, there must have first Merged into the lower SpecTP and then moved to the upper SpecTP.� [TP There seems [TP <there> to be a man in…]]

30

59

*There seems a man to bein the garden

� [TP There seems [TP <there> to be a man in…]]

� This makes sense, both EPP features are satisfied, a man gets case from (the higher, finite) T.

� But think back to when we were building the structure and had reached this point:

� [T′′′′ to be a man in the garden]

� We now have to satisfy the [uD*] feature of T. We have there lying around in our numeration. But if we didn’t, we could have just moved a man to SpecTP to satisfy the EPP.

� [TP a man to be <a man> in the garden]

60

*There seems a man to bein the garden

� [TP a man to be <a man> in the garden]

� After doing this, we can continue to add on seem, v, T, and then insert there into the higher SpecTP, yielding:� [TP there seems [TP a man to be <a man> in…]]

� But this is ungrammatical. So what goes wrong?

� The difference between There seems a man to be in the garden and There seems to be a man in the garden is at the point where we’ve got [T′′′′ to be a man in the garden]. Here there’s a choice: Move a man or Merge there.

� The usual approach here is to say Merge is preferred to Move, so if you have the choice, you always Merge (it’s “easier”).

31

61

Object control� One last type of nonfinite complement, those that

appear with verbs like persuade.� Sayid persuaded Kate to stay.

� Once again, we think through the “participants” to get a handle on whether we have enough DPs for the θ-roles.� Stay has only one participant, Kate.

� Persuade has three—the one doing the persuading (Sayid), the one being persuaded (Kate), and the proposition in question ( [TP Kate to stay]).

� So we don’t have enough DPs for the job— Kate appears to be playing two roles (one from stay, one from persuade). This is a job for PRO.

62

Object control

� Sayid persuaded Kate to stay.

� Sayid persuaded Kate [CP ØNULL PRONULL to stay]

� Again we have PRO, as we do in� Kate tried [CP ØNULL PRONULL to see]

� But in Sayid persuaded Kate to stay, what “controls” PRO?

32

63

PRO� Here’s one last argument for PRO being there.

� Jack hoped [PRO to see Kim ]

� Jack hoped [ that Kim would be safe ]

� *Jack hoped [ that Kim would find himself ]

� Jack hoped [ that Kim would exonerate herself ]

� Principle A: An anaphor must be bound in its binding domain.� Jack hoped [ PRO to exonerate himself ]

� Jack hoped [ that Chase would exonerate him ]

� Principle B: A pronoun must be free in its binding domain.� Jack hoped [ PRO to exonerate him ]

64

Subject control v. object control

� Subject control verbs take a nonfinite complement, with PRO as the subject, and PRO must refer to the higher subject.

� Gael tried [ PRO to disarm the bomb ]

� Object control verbs are ditransitives that take an object and a nonfinite complement, with PRO as the subject, and PRO must refer to the higher object.

� David persuaded Sherry [ PRO to leave ]

33

65

Persuasion and promises� Not all ditransitive control verbs are object control

verbs.� Though all object control verbs are ditransitives.

� David persuaded Sherry [ PRO to leave ]

� David promised Sherry [ PRO to run for office ]

� Chase asked Jack [ PRO to be allowed to continue ]

� Chase asked Jack [ PRO to get off his case ]

� Whether a verb is a subject control verb or an object control verb is an individual property of the verb. Promise is recorded in our lexicon as a subject control verb, persuade as an object control verb.

66

ECM verbs

� ECM verbs also take infinitive complements, but with an overt subject (that checks accusative case with the ECM verb).

� Tony found [ Michelle to be charming ]� Tony found [ that Michelle was charming ]

� Jack expected [ Tony to take the day off ]� Jack expected [ that Tony would take the day off ]

34

67

Raising verbs

� Raising verbs have no Agent/Experiencer in SpecvP, and take a nonfinite complement. The subject of the embedded complement moves into their subject position:

� Jack seems [ <Jack> to be tired ]� It seems [ that Jack is tired ]

� The time appears [ <the time> to have expired ]� It appears [ that the time has expired ]

� The President happened [ <the P.> to have a pen ]� It happened [ that the President had a pen ]

68

Verb classes

� ECM verbs, e.g., believe, find� I believe [TP him to have told the truth].

� We find [TP these truths to be self-evident ]. (or hold)

� Subject control verbs, e.g., attempt, promise� Kimk promised [Jack] [CP ØNULL PROk to avoid kidnappers ].

� Kimk will try [CP ØNULL PROk to avoid kidnappers ].

� Object control verbs, e.g., convince, ask� I convinced [herk] [CP ØNULL PROk to drive to work].

� Jack asked [Kimk] [CP ØNULL PROk to avoid kidnappers ].

� Raising verbs, e.g., appear, seem� I appear [TP <I> to have missed the bus].

� Jack seems [TP <Jack> to need a nap].

35

69

Before we finishembedded clauses…

� Another place we find embedded clauses is as modificational adjuncts.� Pat ate lunch [PP on the hill ]

[PP by the tree ] [PP in the rain ].

� To express reasons and times, we also find whole CPs adjoined to our clause:� We discussed adjuncts [CP before we finished our

discussion of embedded clauses]

� There’s nothing really new here, except the observation that before can have category C.� Just like after, while, during, etc.

70

Adjunct clauses:where do they go?

� Pat cleaned poorly yesterday.

� #Pat cleaned yesterday poorly.

� Pat cleaned poorly [before Chris arrived].

� #Pat cleaned [before Chris arrived] poorly.

� Pat cleaned [before Chris arrived] yesterday.

� Pat cleaned yesterday [before Chris arrived].

� Pat heard that [before Chris arrived][Tracy cleaned the sink].

� Pat heard [before Chris arrived] that[Tracy cleaned the sink].

36

71

Because clauses

� Reason clauses are also clausal adjuncts.

� Because I lost the game, I left.

� I left because I lost the game. vP

lose the game

T′

T[past]

TP

DPI

CP

Cbecause

vPleave

T′

T[past]

TP

DPI

TP

CP

CØ

72

If clauses

� If clauses are like because clauses.

� If he loses the game, I will leave.

� I will leave if he loses the game.

vPlose the game

T′

T[pres]

TP

DPhe

CP

Cif

vPleave

T′

Twill

TP

DPI

TP

CP

CØ