lecture 2: constraints on movement. formal movement rules (called transformations) were first...

32
Lecture 2: Constraints on Movement

Upload: mckenzie-skilton

Post on 13-Dec-2015

222 views

Category:

Documents


0 download

TRANSCRIPT

Lecture 2: Constraints on Movement

Formal movement rules (called Transformations) were first introduced in the late 1950s

During the 1960s a lot of work was done to see what could be achieved by using transformations

Enthusiastic linguists used transformations to describe phenomena in: Syntax Morphology

Semantics PragmaticsStylistics PoeticsLiterary Analysis

By the end of the 1960s it was clear that you could do anything with a transformation!

If transformations can do anything: they can describe everything but explain nothing

Suppose we observe some phrase X moving from position Y to position Z Can we describe this?

Of course – transformations can describe anything

Can we explain why it happens? NO! – Anything could have happened

It therefore became clear that transformations were too powerful

At the same time linguists didn’t want to get rid of transformations altogether

The only way forward was to maintain transformations but attempt to limit their power

There must be things that you can’t do with a transformation

The first observation of situations in which things couldn’t move concerned examples such as: The train [VP came [PP out [PP from [DP Paris]]]] [PP out from where]1 did the train [VP come t1 ] [DP where]1 did the train come out [PP from t1 ] * [ from where]1 did the train come [PP out t1 ]

So: A PP can move out of a VP A DP can move out of a PP But a PP cannot move out of a PP

Similarly: They [VP liked [DP [DP John]’s picture]]

[DP whose picture]1 did they [VP like t1 ]

* [DP whose]1 did they like [DP t1 picture]

So: A DP can move out of a VP But a DP cannot move out of a DP

Similarly: They [VP made [VP the ice melt]] I thought they would make the ice melt and

[VP make the ice melt]1 [IP they did t1 ] * [VP the ice melt]1 they [VP made

t1 ]

So: A VP can move out of an IP But a VP cannot move out of a VP

It looks as though the movement of a phrase cannot be out of a phrase of the same category An XP cannot move out of an XP

There are cases where a phrase can move out of a phrase of the same category: He painted [DP a picture of [DP a lake]]

[DP what]1 did he paint [DP a picture of t1 ]

A DP can move out of a DP

Sometimes an XP cannot move out of a YP, where XP and YP are not the same category John knows [DP the woman you sold [DP your car] to]

John likes [DP the car you sold [PP to Mary]]

* [DP Which car]1 does John know [DP the woman you sold t1 to]

* [PP to whom]1 does John like [DP the car you sold t1 ]

Ross (1967) identified a number of constituents from which he claimed it was impossible to move anything

He called these constituents ‘Islands’

No phrase can be moved out of a clause that begins with a wh-element Who1 can you guess [that Mary likes t1]

* Who1 can you guess [why Mary likes t1]

This constraint also covers relative clauses as they begin with a wh-element We like the present [CP (which) Mary gave to Bill]

* who1 do we like the present [CP Mary gave to t1 ]

No phrase can be moved out of a clause which is contained inside a DP (also covers relative clauses!) He denied [DP the allegation [CP that he

murdered his wife]] * Who1 did he deny [DP the allegation [CP

that he murdered t1 ]]

No phrase can be moved out of a CP that is the subject of another clause [CP that he hid the drugs] was proof of his

guilt * what1 was [CP that he hid t1 ] proof of his

guilt

No phrase can move out of a phrase that is coordinated: John [VP likes beer] but [VP hates wine]

* what1 does John [VP like t1 ] but [VP hates wine]

* what1 does John [VP like beer] but [VP hates t1 ]

Islands constrain transformations and so increase their explanatory power

But what explains Islands? Why are wh-clauses, complex DPs, sentential

subjects and coordinated constituents Islands? Just proposing that there are Islands does

not give any clues to why there are islands or what constituents will be islands

In response to this problem, Chomsky (1973) proposed one general restriction on movement

This restriction predicted most of the Islands and so offered an explanation for why certain constructions are Islands

Subjacency works on the assumption that certain categories are hurdles that have to be jumped over by moving phrases

These hurdles were called Bounding Nodes

In English: IP and DP are bounding nodes

No single movement can cross more than one bounding node

Consider the following Who1 did [IP you think [IP Mary thought [IP Bill liked

t1]]]

It appears that three bounding nodes (IPs) are crossed by the wh-element

But, if we assume: the wh-element doesn’t move in one go it moves to each vacant specifier of CP in turn then each movement crosses only one IP Who1 did [IP you think [CP t1 [IP Mary thought [CP t1 [IP Bill liked t1]]

The wh-Island Who1 can [IP you guess [CP why [IP Mary likes t1]

Movement to the first CP specifier is blocked by the wh-element why

So only long distance movement is possible This crosses two bounding nodes So it violates subjacency

The complex DP Island Who1 did [IP he deny [DP the allegation [CP t1 that [IP he

murdered t1 ]]]

The first movement is fine as it only crosses one bounding node

The second movement however crosses both DP and IP and so violates subjacency

Head movements are always short They never cross clause boundaries Therefore they never get anywhere

near violating subjacency So what makes sure they are short?

V can move to I John love1-s [VP Mary t1]

I can move to C can1 [IP you t1 make the tea]

V cannot move directly to C * make1 [IP you can [VP the tea t1]]

V can move to C only if it first moves to I [CP who2 made1 [IP t2 t1 [VP the tea t1 ]]]

In other words: Heads must

move to the their closest head position

Subjacency constrains phrase movement

The head movement constraint constrains head movement

But both constraints prevent long distance movement

A wh-phrase can move to the specifier of CP [CP who1 did [IP you see t1]]

It cannot move to the CP more than one clause away: * [CP who1 did they ask [CP why [IP you can’t see

t1]]]

The only way a wh-phrase can get to the higher CP is if it goes through the lower CP: [CP who1 did they think [CP t1 [IP you can see t1 ]]

In other words: A wh-phrase

must move to its closest CP specifier

This is very similar to the head movement constraint

A DP can move to the specifier of its own clause (subject) [IP he1 was [VP seen t1]]

It cannot move directly to the subject position of the next clause up: * [IP he1 seems [IP it was [VP seen t1]]]

The only way it can get to this position is by going through the lower subject position: [IP he1 seems [IP t1 to have been [VP seen t1]]]

In other words: A DP must move

to its nearest subject position

Again, this is similar to the head movement constraint

It seems that there is one main restriction on all movements: Movements must be as short as possible

But what is ‘possible’ for a movement depends on what is moving: A head can only move to a head position A wh-phrase can only move to a specifier of CP A DP can only move to a specifier of IP (subject)

Thus the constraint is: An element must move to its nearest relevant

position

Constraints have developed from specific to general ideas

A-over-A and Islands specific constructions

are barriers to movement

Subjacency specific nodes in a

structure are barriers Relativised Minimality

all movements are short

Constraints