Upload
nathan-barker
View
218
Download
1
Embed Size (px)
Citation preview
Ian Roberts
Generate well-formed structural descriptions
“create” trees/labelled bracketings More (X’) or less (PS-rules) abstract Can create infinite structures
Movement rules: Don’t create structures but manipulate
them Technically, they map phrase markers
into other phrase markers Informally, constituents “move” from
one place in the structure to another
Head-movement, as in English subject-auxiliary inversion in questions:John can leave. Can John t leave?
CP | C’ ru C TP Can ru NP T’ John ru T VP t | V win
as in the English passive:The policeman arrested the student The student was arrested t
TP ru NP T’ ru ruD N T VPThe studentwas ru
V NParrested t
The most important and interesting type of movement for various reasons.
Moves (almost) any XP to the beginning of the sentence to form, in the simplest case, a question.
(1) Which man will John see t ? -- object NP
(2) Who should John talk to t ? – indirect object NP (“preposition stranding”)
(3) To whom can John talk t ? – indirect object PP (“pied-piping”)
(4) How angry is Alex t ? -- AP(5) What does John believe t ? – CP TP and VP do not undergo WH-
movement
“Echo-questions” show where the wh-phrase originates:
Bill bought WHAT?! You talked to WHO?!
CP ru
NP C’ru ru
D N C TP Which man will ru
NP T’ | ru
N T VP John t ru
V NP see t
I wonder ..(1) which man John will see t ?-- object NP(2) who John talk should to t ? – indirect
object NP (“preposition stranding”)(3) to whom John can talk t ? – indirect
object PP (“pied-piping”)(4) how angry Alex is t ? -- AP(5) what John believes t ? – CP(6) Just like direct questions except no
subject-aux inversion. (7) C must be empty here (no
that/if/whether)
Who t saw John? “vacuous movement”:
CP ru
NP C’ | ru
N C TP Who ru
NP T’ t ru T VP
ru V NP saw John
I wonder who t saw John ? *I wonder who did t t see John ? *I wonder who that/if t saw John ? *I wonder that/if who saw John ?C must always be empty in indirect
subject questions like all other indirect questions.
Movement to SpecC’ always, accompanied by a zero [+wh] C.
(1) Which man did you say (that) John will see t ? -- object NP
(2) Who did you say (that) John will talk to t ? – indirect object NP (“preposition stranding”)
(3) To whom did you say (that) John will talk t ? – indirect object PP (“pied-piping”)
(4) How angry did you say (that) Alex is t ?-- AP
(5) What did you say (that) John believes t ? – CP
Who did you say [ that Mary believes [ that John saw t ]] ?
Who did you say [ that Mary believes [ that Fred knows [ that John saw t ]]] ?
Who did you say [ that Mary believes [ that Fred knows [ that I asserted [ that John saw t ]]] ?
.. and so on.
The unbounded nature of WH-islands poses a problem for PS-rules/X’theory because
PS-rules/X’-theory are local: they define little bits of the tree at a time, e.g.VP V CP
How does VP “know” it’s part of a wh-dependency as in:
Who did you say [ that Mary [VP believes [CP that John saw t ]] ?
Either we complicate the PS-rules/X’-theory hugely (this can be and has been done) or we have two relatively simple rule types:
PS-rules/X’-theory build structure (create phrase markers)
Movement/transformational rules manipulate structure (map phrase markers into other phrase markers)
WH-movement: Move a WH-phrase to the specifier of a [+wh] C.
“NP-movement”: move the object to the subject position (passive)
Head-movement: move T to C (subject-aux inversion)
(The last two can be generalised; WH-movement is already in quite a general form here).
Although WH-movement is unbounded, it doesn’t apply just anywhere, but is subject to stringent locality conditions, cf.:
(1) my guitar, John’s cat – possessor NP in Spec of higher NP:
NP1 ruNP2 N’1 | |N’2 N1| catN2 guitarJohn’smy
And similarly for whose cat
If you want to know whose cat you’re talking about:
(1) Whose cat did you feed t ? – object NP(2) *Whose did you feed [NP t cat ] ? –
can’t apply WH-movement directly to the possessor, but must “pied-pipe” the whole object.
The Left Branch Constraint (LBC):WH-movement can’t apply to a left-branch, or to part of a left branch.
So, whose can’t move on its own. Cf also:
How angry is Alex t ? -- AP-movement*How is Alex [AP t angry ] ? – no
movement just of Spec of AP.
Sometimes a pronoun appears where a gap could be:
John, I like (him). John, I like {him/t}. Pronouns don’t obey island constraints:*Whose did you say that you like [NP t
cat ]?Whose cat did you say that you like t ??Who did you say that you like his cat ?Marginally (in English) a resumptive
can link to a WH, but only in an island.
The LBC is one of several constraints on WH-movement called “island constraints” (islands are things that are difficult move off). In general, WH-movement:
Leaves a gap Is unbounded Obeys the LBC (and other island
constraints)
Recall the constituency test clefting:John wrote the book It was John that
wrote the book.
It leaves a gap*It was John that he wrote the letter. It can be unboundedIt was John that Mary said that Bill thinks t
wrote the letter. It obeys LBC:*It was John’s that I saw [NP t cat ].
We need movement rules alongside PS-rules/X’-theory
Three types of movement WH-movement involves unbounded
dependency WH-movement leaves a gap, obeys LBC Clefting is a type of WH-movement
(albeit a bit hidden)