+ All Categories
Home > Documents > Chapter 10

Chapter 10

Date post: 11-Nov-2014
Category:
Upload: thatha-baptista
View: 107 times
Download: 5 times
Share this document with a friend
Popular Tags:
38
253 Chapter 10 WH-MOVEMENT 0. INTRODUCTION In the last chapter we looked at situations where NPs were appearing in positions where they didn’t get theta roles. Instead, the were in a derived position. The trigger for this movement was the requirement that NPs bear a case. This case can only be assigned in specific structural positions. In this chapter, we turn to another kind of phrasal movement, one where NPs already have case. NPs (and other phrases) move for a different reason. This transformation is found in Questions. There are several different kinds of questions, only two of which we will be concerned with in this book. The first kind is the familiar yes/no question that we looked at in the chapter on head-movement: 1) Are you going to eat that bagel? 2) Do you smoke Camels? 3) Have you seen the spectrograph for that phoneme?
Transcript
Page 1: Chapter 10

253

Chapter 10WH-MOVEMENT

0. INTRODUCTION

In the last chapter we looked at situations where NPs were appearing in

positions where they didn’t get theta roles. Instead, the were in a derivedposition. The trigger for this movement was the requirement that NPs bear acase. This case can only be assigned in specific structural positions. In thischapter, we turn to another kind of phrasal movement, one where NPs already

have case. NPs (and other phrases) move for a different reason. Thistransformation is found in Questions.

There are several different kinds of questions, only two of which wewill be concerned with in this book. The first kind is the familiar yes/noquestion that we looked at in the chapter on head-movement:

1) Are you going to eat that bagel?2) Do you smoke Camels?3) Have you seen the spectrograph for that phoneme?

Page 2: Chapter 10

254 Sentence Structure: A Generative Introduction

© Andrew Carnie

The answers to these questions cannot be a phrase other than Yes or No. Anyother response is infelicitous:

1’) #Pizza/ 3yes2’) #Marlboros/ 3no3’) #Syntactic tree./ 3 no

The other kind of question we’re interested in are called Wh-questions. Theytake this name from the fact that the words that introduce them (mostly) beginwith <wh> in English: who, what, when, where, why, which, and how. The

responses to these kind of questions cannot be yes or no. Instead they must beinformative phrases.

4) When did you do your syntax homework? #Yes / 3Yesterday

5) What are you eating? #No/ 3 A bagel6) How is Louise feeling? #yes/3much better

It's the syntax of this kind of question that is the focus of this chapter.

Who and WhomIn traditional prescriptive grammar there are two wh-words that refer

to people: who and whom. Who is used when the wh-word originates

in subject position and gets nominative case. Whom is the accusative

version. In most spoken dialects of Canadian and American English

this distinction no long exists, and who is used in all environments.

For the sake of clarity, I use who(m) to indicate that the wh-word

originated in object position, but you should note that from a

descriptive point of view who is perfectly acceptable in object

position.

Page 3: Chapter 10

Chapter10 – Wh-Movement 255

© Andrew Carnie

1. MOVEMENT IN WH-QUESTIONS

If you look closely at a statement and a related wh-question, you'll see that the

wh-word appears in a position far away from the position where its theta role isassigned. Take for example:

7) Becki bought the syntax book8) What did Becki buy?

The verb 'buy' in English takes two theta roles, an external agent and an internal

theme. In sentence (7), Becki is the agent, and [the syntax book] is the theme. Insentence (8), Becki is the agent and what is the theme. In the first sentence, thetheme is the object of the verb, in the second the theme is at the beginning of theclause. The situation becomes even more mysterious when we look at sentences

like (9)

9) What did Stacy say Becki bought?

In this sentence what is still the theme of bought, yet it appears way up at thebeginning of the main clause. This would appear to be a violation of the localitycondition on theta role assignment introduced in the last chapter.

The situation becomes even more murky when we look at case. Recallthat accusative case is assigned when an NP is the sister to a V:

10) Matt kissed her

But in wh-questions the accusative form (like whom) is not a sister to V:

11) Whom did Matt kiss?

So it appears as if not only are these wh-words not in their theta positions, they

aren't in their case positions either!

Given what we've seen in the previous two chapters this looks like

another case of movement. This one with different triggers again. Let's start withthe question of where wh-words move to. One position that we've had for a

Page 4: Chapter 10

256 Sentence Structure: A Generative Introduction

© Andrew Carnie

while, but have never used is the specifier of CP. It's to this position that we'llmove wh-words:

12) CP

C'

C IP

NP I'

Matt INFL VPdid V'

V NP

whom

Notice that what moves here is an entire phrase. You can see this if you look atslightly more complex wh-questions:

13) [To whom] did Michael give the book14) [Which book] did Michael give to Millie

When you move an entire phrase, it cannot be an instance of head-to-headmovement (by definition), so this must be movement to a position other than ahead, in this case we have the empty specifier of CP.

Page 5: Chapter 10

Chapter10 – Wh-Movement 257

© Andrew Carnie

.Notice that the movement to the specifier of CP accounts for another

fact about the word order of wh-questions: they also involve I→C movement (in

main clauses):

15) Who(m) are you meeting?

16) *Who(m) you are meeting?

This means that the wh-word must raise to a position higher than C. The onlyone available to us is the specifier of CP:

A Mystery: No INFLàC movement in embedded clauses.

In the main text, we've noticed that wh-movement and INFLàC

movement often go hand in hand. One surprising fact about English is

that this is not true of embedded wh-questions. When a wh-question is

embedded the subject does not invert with the auxiliary (i.e. no

INFLàC movement):

i) I wonder what he has done?

ii) *I wonder what has he done?

Why you don't get INFLàC movement in embedded wh-clauses is a

mystery. We don't have a good explanation for it

Page 6: Chapter 10

258 Sentence Structure: A Generative Introduction

© Andrew Carnie

17) CP

C'

C IP

NP I'

Matt INFL VPdid

V'

V NP

whom

The same fact can be seen in languages that allow both a wh-word and an overt

complementizer:

18) Cé aL bhí sa seomra?Who that-wh was in-the room

Who was in the room (Irish)

Page 7: Chapter 10

Chapter10 – Wh-Movement 259

© Andrew Carnie

Before turning to the constraints on wh-movement let's briefly considerwhat would trigger wh-movement. One possibility is to motivate it with aspecial form of the question feature. Let's call this feature [+WH], this resides in

the C of a Wh sentence. A wh-word moves to the specifier of CP to be "near"

The Doubly filled CP FilterWhile CP's containing Wh-phrases allow complementizers to contain

an inverted auxiliary: ([CP What did [IP she kiss ]]?) They don't allow

any other kind of complementizer:

i) *I asked what that she kissed?

ii) *I asked what whether she kissed?

Multiple wh-phrases are also not allowed:

iii) *I asked what how she kissed?

There thus seems to be a constraint that disallows the CP having

multiple pronounced contents. This constraint is called the doubly

filled CP filter:

iv) * [CP WH that ]

This constraint holds of English only. Other languages allow both a

complementizer and the wh-phrase (Data from Bavarian German,

Bayer 1984):

v) I woass ned wann dass da Xavea kummt.

I know not when that the Xavea comes

"I don't know when Xavea is coming'

See also example (18) from Irish in the main text.

Page 8: Chapter 10

260 Sentence Structure: A Generative Introduction

© Andrew Carnie

the [+WH] feature. Another way to phrase this is to say that wh-words move into the specifier of CP to "check" the wh-feature. We can formalize wh-movement the following way:

19) Wh Movement

Move a wh-word to the specifier of CP to check a wh-feature in C.

The existence of a [+WH] feature on C is not completely unmotivated. In somelanguages (such as Irish), there are special forms of complementizers thatrepresent these features:

20)

[-Q, -WH] go

[+Q, -WH] an

[+Q, +WH] aL

Traces - Wanna contraction.In the last chapter, we obliquely suggested that movement leaves a

'trace' behind which we represented with a t (coindexed with word it

replaces). Here too we are going to claim that wh-movement leaves a

trace at the base position of movement. Later in this chapter we'll see

that traces are required to block certain kinds of illicit movement. But

an important question is whether is any reality behind the notion

'trace'. This is especially important in a theory like generative

grammar which claims psychological underpinnings. Finding

evidence for something that isn't pronounced is remarkably difficult.

However, quite surprisingly there is some straightforward evidence for

the existence of traces. First a little background: In spoken varieties

of English (both standard and non-standard) function words often

contract with nearby words. One such contraction takes non finite

INFL ('to') and contracts it with a preceding verb like want:

i) I want to leave → I wanna leave

This phenomenon is called wanna contraction. Now consider what

Page 9: Chapter 10

Chapter10 – Wh-Movement 261

© Andrew Carnie

Let's do a derivation for the following sentence:

21) Whom is Matt kissing?

The D-structure of this sentence will look like (22):

happens when you have wh-movement and wanna contraction going

on at the same time. Wanna contraction is permitted when the wh-

movement applies to an object:

ii) Who(m) do you wanna kiss

But look what happens when you try to wanna contraction, when wh-

movement targets the subject:

iii) Who do you want to kiss the puppy?

iv) *Who do you wanna kiss the puppy?

English speakers have very strong judgements that wanna

contraction is impossible when the subject is wh-questioned. Why

should this be the case? If we have traces, the explanation is simple:

the trace intervenes between the 'to' and the verb. It blocks the strict

adjacency between the verb and the to, thus blocks contraction:

v)Whoi do you want ti to kiss the puppy?

Without the theory of traces, we wouldn't have any explanation for

this fact.

Page 10: Chapter 10

262 Sentence Structure: A Generative Introduction

© Andrew Carnie

22) CP

C'

C IP Ø[+Q,+WH]

NP I'

Matt INFL VPis

V'

V NP kissing

whomMatt and whom both get their theta roles in these positions1. The both also getcase in these positions. What is lacking in this D-structure is the unpronounced

[+Q] feature and the unchecked [+WH] feature. We get INFLàC movement tosolve the first problem:

23) CP

C'

C IP

Ø[+Q,+WH]

NP I'

Matt INFL VPis V'

V NP

1 For simplicity's sake I'm abstracting away from the VP internal subjecthypothesis. If we assumed the VP internal subject hypothesis we'd have to moveMatt from the specifier of VP to the specifier of IP for case.

Page 11: Chapter 10

Chapter10 – Wh-Movement 263

© Andrew Carnie

kissingwhom

Wh-movement applies to solve the second:

24) CP

C'

C IP

Ø[+Q,+WH]

NP I'

Matt INFL VP

is V'

V NP kissing

whom

This results in the surface string:

25) Whom is Matt kissing?

Consider a more complicated example, this one involves, NP movement,Wh-movement and INFLà C movement:

26) Who was kissed?

The D-structure of this sentence is in (27)

Page 12: Chapter 10

264 Sentence Structure: A Generative Introduction

© Andrew Carnie

27) CP

C'

C IP Ø[+Q,+WH]

I'

INFL VPwas

V'

V NP kissed

who

Who is the only argument in the sentence (a theme) and it is assigned it's theta

role in object position. Since this is a passive, however, it cannot get case in thisposition. It must move to the specifier of IP to get Nominative case:

28) CP

C'

C IP Nominative Case

Ø[+Q,+WH]

I'

INFL VPwas V'

V NP kissed

who

Page 13: Chapter 10

Chapter10 – Wh-Movement 265

© Andrew Carnie

Once this NP has been licensed by case, it can move on to the specifier of IP.The auxiliary also undergoes INFL→C movement.

29) CP Wh-feature checking

C'

C IP Nominative Case Ø[+Q,+WH]

I'

INFL VP Theta Rolewas V'

V NP kissed

who

All in all there are three movements in this structure, NP movement, I→C

movement and Wh-movement. Wh-movement only moves NPs that are case-

marked. If we were simply to skip the NP movement as in the ill-formed tree in(30):

Page 14: Chapter 10

266 Sentence Structure: A Generative Introduction

© Andrew Carnie

30) *CP Wh-feature checking

C'

C IP Nominative Case Ø[+Q,+WH]

I'

INFL VP Theta Rolewas

V'

V NP kissed

who

The NP who would never get case, and thus would violate the case filter, (30) is

not an acceptable derivation.

Wh-movement can also apply across clauses. Now, we'll do aderivation of a sentence where the Wh-phrase moves from an embedded clause

to the specifier of a main clause CP.

31) Who(m) did you think Jim kissed?

The D-structure of this sentence will look like (32). In this tree, who(m) is thetamarked by the verb kiss, and gets it's internal theme theta role in the objectposition of that verb. The two NPs also can get case in their D-structure

positions, so there is no NP movement:

Page 15: Chapter 10

Chapter10 – Wh-Movement 267

© Andrew Carnie

31) CP

C'

C IP NOMINATIVE CASE

Ø[+Q,+WH]

NP I'

you INFL VP-ed

V'

V CP Think

C'

C IP

Ø[+Q,+WH]

NP I' NOMINATIVE CASE

Jim INFL VP

-ed V'

V NP

kissACCUSATIVE CASE whom

Starting with the embedded clause, we can note that the INFL suffix -ed needssupport, so it undergoes affix lowering:

Page 16: Chapter 10

268 Sentence Structure: A Generative Introduction

© Andrew Carnie

32) CP

C'

C IP Ø[+Q,+WH]

NP I'

you INFL VP-ed

V'

V CP think

C'

C IP

Ø[+Q,+WH]

NP I'

Jim INFL VP

-ed V'

V NP

kisswhom

There is no NP movement in this sentence, but there is an unchecked [WH]feature in the main clause. This causes the phrase "whom" to move. In addition,there is the unpronounced [+Q] feature; this triggers movement of INFLàC.

However, since there isn't an auxiliary in INFL, the rule of do-insertion alsoapplies:

Page 17: Chapter 10

Chapter10 – Wh-Movement 269

© Andrew Carnie

33) CP

C'

C IP Ø[+Q,+WH]

NP I'

you INFL VP-ed

V'do

V CP think

C'

C IP

Ø[+Q,+WH]

NP I'

Jim INFL VP

-ed V'

V NP

kissedwhom

This derives the correct S-structure.

Page 18: Chapter 10

270 Sentence Structure: A Generative Introduction

© Andrew Carnie

The that-trace effect

In English there is a strange effect with respect to complementizers

found with wh-movement from an embedded clause. Wh-movement

of objects seems to be free, you can do it either when there is a

complementizer or when there is no complementizer:

i) Whati do you think Matt kissed ti ?

ii) Whati do you think that Matt kissed ti ?

Strangely, this is not the case with subjects. Wh-movement from

subject position is only possible when there is no overt that

complementizer:

iii) Whoi do you think ti kissed Matt?

iv) *Whoi do you think that ti kissed Matt?

This surprising effect is called the that trace effect, from the

constraint that is used to rule out sentences like (iv), the that trace

filter:

v) * that t

The that trace effect is not universal. Many languages (such as

Italian), don't have it:

vi) Chi credi che ti venga?

Who you-think that ti come

"Who do you think is coming (Rizzi 1982)

The explanation for the that trace effect consumed much of the

syntactic literature from the 1980s to about 1991, but lies beyond the

scope of this text.

Page 19: Chapter 10

Chapter10 – Wh-Movement 271

© Andrew Carnie

2. CONSTRAINTS ON WH-MOVEMENT: BOUNDING THEORY

Wh-movement isn't limited to just NPs. It can apply to a wide variety of phrasal

categories:

34) To whom did you give the book ? (PP)

35) Why did you dance the macarena? (AP)36) How did you cut your toe? (AP)37) When did you cut your toe? (AP)38) Which book did you drop off at the library (NP)

It isn't without some limitations however. These limitations hold on whatcategories you can move out of (the categories that contain the wh-word).Compare the following two sentences, one of which has wh-movement out of

simple complement CP (39). The other (40) moves a wh-word out of a clausethat is contained inside an NP (a relative clause).

39) Whati did Bill claim [CP that he read ti in the syntax book?]40) *What did Bill make [NP the claim [CP that he read ti in the syntax book?]]

In sentence (39), we see that wh-movement out of a complement clause is

grammatical. The nearly parallel sentence (40), on the other hand, shows thatnot all wh-movement out of a CP is acceptable. In particular it seems that youcan't move a wh-word out of a CP contained within an NP. This phenomenon,first observed by Ross (1967), has come to be known as the Complex NP islandphenomenon. The word Island here is meant to be iconic. Islands are places youcan't get off of (without special means like a plane), they are surrounded bywater, so you are limited in where you can move: you can only move about

within the confines of the island. Islands in syntax are the same. You cannotmove out of an island, but you can move around within it. Noun Phrases areislands.

41) *What did Bill make [NP the claim [CP that he read ti in the syntax book?]]Complex NP island

Page 20: Chapter 10

272 Sentence Structure: A Generative Introduction

© Andrew Carnie

There is another kind of island. First, observe that it is possible to movea wh-word to the specifier of an embedded CP2:

42) I wonder [CP what [IP John bought t with the $20 bill]]

It is also possible to move (another) wh-word to the specifier of the main CP

:43) [CP How do [IP you think [ John bought the sweater?]]].

However, look at what happens when you try to do both (move one wh-word to

the embedded specifier, and the other to the main CP specifier:

44) *[CP Howj do [IP you think [CP whati [IP John bought ti tj]]]]

This sentence is wildly ungrammatical -- even though we have only done twootherwise legitimate transformations. Now this isn't a constraint on having twowh-words in a sentence. That is perfectly acceptable3:

45) How do you think John bought what?46) I wonder what John bought how?

It seems then to be a constraint on moving both of them. The same kind ofexample is seen in (47)-(48):

47) I wonder whati John kissed ti

48) Whoj did you think tj kissed the gorilla

2 There is actually a restriction on this. Wh movement to an embedded CP canonly take place when the main verb selects (theta marks) a question clause.

Verbs that do this are verbs like wonder and ask. Verbs like 'think' don't allowembedded wh-questions. Since there will be occasions where we want tocompare embedded wh, and non-embedded wh, I'm going to play fast and loosewith the choice of main verbs. I'll use wonder when I want an embedded

question, but think when I want an embedded statement (with no [+WHfeature]).3 If you have trouble with this judgement, try stressing the word what in (45) andhow in (46).

Page 21: Chapter 10

Chapter10 – Wh-Movement 273

© Andrew Carnie

Movement of either the subject (47) or the object (48) to the specifiers of CP isacceptable. However movement of both results in terrible ungrammaticality:

49) *Whoj did you think whati tj kissed tj

The central intuition underlying an account of these facts is that once you move

a wh-word into the specifier of a CP that CP becomes an island (normally calleda wh-island) for further extraction:

50) I asked [CP whati John kissed ti ] wh-island

Movement out of this wh-island results in ungrammaticality.

What is lacking from what we have said so far has been a formal

description of the fact that you can't extract out of islands. In the ideal situation,we want our formal description to account for both types of islands (complexNP islands and wh-islands.) In the mid-1980s a theory of these facts was

developed, called bounding theory (this is not to be confused with bindingtheory which accounts for the distribution of anaphors). Although currently amatter of great controversy, we will explore the most widely accepted version ofbounding theory.

The central idea underlying bounding theory is that certain kinds ofnodes are boundaries for movement. These nodes are called bounding nodes.There are two bounding nodes: IP and NP.

51) Bounding nodes: NP and IP.

Now it is clear that we can't simply say "Don't cross a bounding node". A simplewh question like (52) violates this off the bat:

52) [CP whoi did [IP John see ti ]]

yet sentence (52) is completely grammatical. The account has to be morecomplicated. The answer lies in the Subjacency condition of Chomsky (1973):

Page 22: Chapter 10

274 Sentence Structure: A Generative Introduction

© Andrew Carnie

53) Subjacency Condition:Wh-movement may not cross more than one bounding node (but it maycross one).

Let's see how this works. First consider the grammatical sentence in (52). Thewh-movement there only crosses one bounding node. So the sentence meets the

subjacency condition.

As an aside note that the subjacency condition forces us to make astipulation on wh-movement across clauses:

54) [CP whati did [IP you think [CP that [IP Millie said ti ?]]]]

Movement of this kind will necessarily cross two IPs, and thus should be aviolation of the subjacency condition. However, the sentence is grammatical!

Remember, wh-movement targets the specifier of CP. There is an unfilledspecifier of CP between the two IPs

55) [CP whati did [IP you think [CP that [IP Millie said ti ?]]]]

If we allow the wh-word to stop off in this specifier on the way to the higher CP,we won't violate the subjacency constraint. We will have two movements, eachof which will cross only one bounding node:

56) [CP whati did [IP you think [CP that [IP Millie said ti ?]]]]

move # 1

57) [CP whati did [IP you think [CP ti that [IP Millie said ti ?]]]]

move # 2

By allowing the wh-word to stop off in the intermediate specifier of CP, we

have an ‘out’ for the (grammatical) movement of wh-words across clause

Page 23: Chapter 10

Chapter10 – Wh-Movement 275

© Andrew Carnie

boundaries. This property of stopping off intermediate specifiers of CPs issometimes called cyclicity. There are two cycles of movement in deriving asentence like (57). The first cycle moves to the embedded CP specifier, the

second to the higher one.

Now that we've seen that we can generate the grammatical sentences,

let's see if we can account for the island conditions with the SubjacencyCondition. First, a complex NP island.

58) *Whati did Bill make the claim that he read ti in the syntax book?]

A partial D-structure of this sentence is given in (59) (this isn't the real D-structure -- I haven't bothered to undo any affix lowering or INFLàC, I've onlyundone the relevant wh-movement). I’ve marked the potential landing sites for

the wh-word with an underline and the bounding nodes with a circle:

Page 24: Chapter 10

276 Sentence Structure: A Generative Introduction

© Andrew Carnie

59) CP

_____ C'

C IP did

Bill I'

INFL VP

V'

V NP make

the N'

N' CP

N ____ C' claim

C IP

he I'

INFL VP

V'

V' PP

V NP in the syntax book read

what

There are three bounding nodes in this tree: the matrix IP, the object NP, and theembedded IP. (Note that there are other NP nodes in this tree, but they do not

count as bounding nodes as we are only concerned with what nodes you can

Page 25: Chapter 10

Chapter10 – Wh-Movement 277

© Andrew Carnie

move out of. So the only relevant NP nodes are those that contain (dominate)the wh phrase. The first step in the movement is to move the wh-phrase into thespecifier of the embedded CP. This is a legitimate move as it only crosses one

bounding node.

Page 26: Chapter 10

278 Sentence Structure: A Generative Introduction

© Andrew Carnie

60) CP

_____ C'

C IP did

Bill I'

INFL VP

V'

V NP make

the N'

N' CP

N ____ C' claim

C IP

he I'

INFL VP

V'

V' PP

V NP in the syntax book read

what

Next let's try to move to the next available node: the specifier of the higher CP:

Page 27: Chapter 10

Chapter10 – Wh-Movement 279

© Andrew Carnie

61) *CP

_____ C'

C IP did

Bill I'

INFL VP

V'

V NP make

the N'

N' CP

N ____ C' claim

C IP

he I'

INFL VP

V'

V' PP

V NP in the syntax book read

what

This move is illegitimate. It violates the subjacency condition by crossing twobounding nodes: the NP and the highest IP. This is a good result, since the

sentence is ungrammatical. The subjacency condition correctly predicts that this

Page 28: Chapter 10

280 Sentence Structure: A Generative Introduction

© Andrew Carnie

sentence will be bad. Notice that we can't use the escape hatch we used with thegrammatical sentence. There is no specifier of CP between the highest IP andthe NP, so there is no stopping off place. For completeness sake, here is the

tree for the comparable sentence without an NP island (What did Bill claim thathe read in the syntax book):62) *CP

_____ C'

C IP

did Bill I'

INFL VP

V'

V CP claim

_____ C’

C IP

he I'

INFL VP

V'

V' PP

V NP in the syntax book

read what

Page 29: Chapter 10

Chapter10 – Wh-Movement 281

© Andrew Carnie

Without the containing NP, each wh-movement only crosses one boundingnode, so the sentence is grammatical.

Now, we'll do a wh-island sentence:

63) *Whoj did you wonder whati tj kissed t

The D-structure for this sentence (again ignoring all extraneous transformationslike INFLàC, etc) is shown in (64).

64) CP

C'

C IP did

you I'

INFL VP

V'

V CP wonder

____ C'

C IP

NP I'

who INFL VP

V'

V NP kissed

what

Page 30: Chapter 10

282 Sentence Structure: A Generative Introduction

© Andrew Carnie

Here there are only two bounding nodes, but there are also two wh-phrases tocontend with. First we'll move what to the lower CP specifier:

65) CP

C'

C IP did

you I'

INFL VP

V'

V CP

wonder ____ C'

C IP

NP I'

who INFL VP

V'

V NP kissed what

This movement does not violate the subjacency constraint. It only crosses onebounding node. Now , we move the other wh-phrase.

Page 31: Chapter 10

Chapter10 – Wh-Movement 283

© Andrew Carnie

66) CP

___ C'

C IP did

you I'

INFL VP

V'

V CP wonder

NPi C'

what C IP

NP I'

who INFL VP

V'

V ti

kissed

It is movement of this second wh-phrase that causes the violation. It can't stopoff in the lower CP, because this position is already occupied by the other wh-

phrase! Who is then forced to cross two bounding nodes, resulting in asubjacency violation. You might think that we could rescue this tree by simply

ordering the movements differently, by moving the who-phrase first (thusallowing it to stop in the lower CP specifier.) then moving the what-phrase. Dueto the existence of traces, however, this is also ruled out. Using the same D-structure as before (64), let's move the who-phrase first. and lets do it in two

steps so that it doesn't violate subjacency:

Page 32: Chapter 10

284 Sentence Structure: A Generative Introduction

© Andrew Carnie

67) CP

____ C'

C IP did

you I'

INFL VP

V'

V CP wonder

____ C'

C IP

NP I'

who INFL VP

V'

V NP

kissed what

Since we only cross one bounding node at a time, each of these moves isacceptable. The sentence runs into trouble however when we try to move theother wh-phrase:

Page 33: Chapter 10

Chapter10 – Wh-Movement 285

© Andrew Carnie

68) CP

NP C'

who C IP did

you I'

INFL VP

V'

V CP wonder

twho C'

C IP

twho I'

INFL VP

V'

V NP

kissed what

There is nowhere to move this NP to. The specifier of the lower CP is filled bythe trace of the who-phrase, and the higher CP is filled by the who-phrase itself!There is no landing site for this NP. Moving it into the trace would result inhaving a doubly filled specifier (containing both the trace and the what-phrase)

for X-bar theoretic reasons we want to rule out this possibility. Just as you can'tmove an NP into another NP, you can't move a Wh-phrase into a trace. The netresult of all of this is that you can't save the ungrammatical sentence *[who did

you wonder what kissed] by moving the who-phrase first. No matter what you

?

Page 34: Chapter 10

286 Sentence Structure: A Generative Introduction

© Andrew Carnie

do, this sentence is going to be ungrammatical. Again, this is correctly predictedby the subjacency constraint.

The arguments presented here are very complex, and the facts evenmore so. However, in the end a very simple pattern emerges with respect toislands. With the subjacency constraint, the range in which a wh-word can move

is restricted. Apparent long-distance moves are the result of cyclic shortmovements. Cases where you can't make a short hop result in terribleungrammaticality (such as the island violations).

3. CONCLUSION.

In this chapter, we looked at a third kind of movement transformation: Wh-movement. This process targets wh-words and moves them to the specifier of

CPs. This movement is triggered by the presence of a [+WH] feature in C.Movement is always from a case position to the specifier of CP (a non-caseposition). Wh-movement is not totally unrestricted; there is a locality constraint

on the movement: subjacency. Movement must be local, where local is definedas maximally crossing only one bounding node. In the boxes, we also saw thatin English (and English alone) there are two additional constraints on Wh-movement. One is the doubly filled CP filter: In English you can't have both a

Wh-word and an overt complementizer (although I→C movement is allowed).

The other is the that-trace filter, which rules out wh-movement from a subjectposition adjacent to an overt complementizer.

Now that we have looked at three different kinds of movement, in thenext chapter we are going to compare them and see what similarities anddifference exist, and see if there is a way to further simplify the system.

IDEAS, RULES AND CONSTRAINTS INTRODUCED IN THIS CHAPTER

i) Wh MovementMove a wh-word to the specifier of CP to check a wh-feature in C.

ii) Island:

Page 35: Chapter 10

Chapter10 – Wh-Movement 287

© Andrew Carnie

A phrase that contains you, and that you may not move out of.

iii) Bounding nodes:NP and IP.

iv) Subjacency Condition:Wh-movement may not cross more than one bounding node (but it may

cross one).

v) Doubly filled CP filter (English only):* [CP WH that ]

vi) That trace filter (English only)* That tracewh

FURTHER READING: (to be completed later)

Chomsky 81

Rizzi 82 & others

PROBLEM SETS

(TO BE COMPLETED)

1) ENGLISH TRANSFORMATIONS

For each of the following sentences, give the D-structure and annotate it witharrows indicating what transformations have applied. Hint: be careful, sincesome of these sentences might have PRO in them.

Page 36: Chapter 10

288 Sentence Structure: A Generative Introduction

© Andrew Carnie

1) How was the plot discovered by the authorities?2) Which animals appear to have lost their collars?3) Alan told me who wanted to seem to be invincible.

2) BOUNDING THEORY

Why is the following sentence ungrammatical?

*Whoj did [IP George try to find out [CP whati [IP tj wanted ti].

Draw a tree showing exactly the problem with this sentence.

3) IRISH

Some dialects of English allow a kind of Wh-construction, where the baseposition of the wh-word is filled with what is called a "resumptive pronoun":

This is the booki that the police are arresting everyone who reads iti

In Modern Irish, this kind of construction is very very common. Modern Irishhas two different wh-complementizers (notice that these are NOT wh-words,which go in spec,CP, these are C°):

aL, aN. The complementizer aL is found in sentences like (1). Sentence (a)shows a simple sentence without wh-movement, sentences (b) and (c) show twopossible forms of the question (b) has the question moved only to anintermediate CP specifier. (c) has the wh-word moved to the topmost specifier.

1) a) Bíonn fios agat i gconaí go bhuailfidh an píobaire an t-amhrán] be.HAB know at.2.S always that play.FUT the piper the song 'You always know that the bagpiper will play the song'

a) Bíonn fios agat i gconaí [CP caidéi aL bhuailfidh an píobaire ti] be.HAB know at.2.S always whati COMP play.FUT the piper ti 'You always know what the bagpiper will play'

b) [CP Cáidéi [IP aL bhíonn fios agat i gconaí [CP ti aL bhuailfidh an píobaire ti]] What comp be.hab know at.2.s always comp play.fut the piper

"what do you always know the piper will play"

Page 37: Chapter 10

Chapter10 – Wh-Movement 289

© Andrew Carnie

Now the distribution of the complementizer aN seems to be linked to thepresence of a resumptive pronoun. Consider the following sentences (2b) and(3b) both show resumptive pronouns and the complementizer aN:

2) a) Bíonn fios agat i gconaí [CP caidéi aL bhuailfidh an píobaire ti] be.HAB know at.2.S always whati COMP play.FUT the piper ti 'You always know what the bagpiper will play'

b) [CP Cén Píobaire j aN [IP mbíonn fios agat i gconaí [CPcaidéi aL bhuailfidh séjti]] Which piper COMP be.HAB know at.2.S always whati COMP play.FUT he 'Which bagpiper do you always know what he will play'

3) a) Tá máthair an fhir san otharlann Be.PRES mother the man.GEN in.the hospital 'The man’s mother is in the hospital'

b) Cé aN bhfuil ai mháthair san otharlann who COMP be.PRES his mother in.the hospital

'Who is (his) mother in the hospital'

The aN complementizer and the resumptive pronouns are boldfaced in the aboveexamples. Where precisely does the aN-resumptive strategy appear? In whatenvironment do you get this construction

4)BINDING THEORY.

In chapter 4, you were asked why this sentence causes a problem for the bindingtheory. Remind yourself of your answer, and then explain how the model ofgrammar we have suggested accounts for this fact.

Which pictures of himself does Bill hate?

Page 38: Chapter 10

290 Sentence Structure: A Generative Introduction

© Andrew Carnie

5. ENGLISH

Do derivations for each of the following sentences. They may involve Head tohead Movement, do-insertion, expletive insertion, NP Movement and Wh-movement.1) Car sales have surprised the stockbrokers.2) Have you seen my model airplane collection?3) Can you find the lightbulb store?4) John was bitten by an advertizing executive.6) It is likely that Siobhán will leave New York.7) Siobhán is likely to leave New York8) It seems that Susy was mugged.9) Susy seems to have been mugged.10) What did you buy at the supermarket?11) You bought WHAT at the supermarket?!?12) I asked what John bought at the supermarket.13) What is likely for John to have bought at the supermarket14) What is likely to have been bought at the supermarket?


Recommended