Summary of Chomsky 1977[*]
Emi Mukai (v4, 03/13/05)
1.Overall picture of the model assumed in the paper
1.1.The model assumed in the paper
(1)a.[A] grammar is a theory of competence[.] [U]niversal grammar (UG) is in essence a system of principles specifying the nature of linguistic representations and the rules that apply to them, and the manner in which these rules apply. (p. 71)
b.A grammar (strongly) generates a set of structural descriptions and (weakly) generates a language, assigning one or more structural descriptions to each sentence of the language. (p. 71)
---The model assumed in the paper is in (2).
(2)
Base (categorical component & Lexicon; see (3))
Base rules ((4))
Deep Structure
Transformational rules (see (5)), constrained by Cycle; (13).
Surface Structure (see (6))
Phonological rulesInterpretive rules ((7)), constrained by SSC & PIC; (15)(16).
PR (see (9))LF (see (8))
The rule of predication (see Appendix)
prlf
(These are a pair of representations assigned to each sentence by a grammar.)
(3)About Base (p. 71):
A categorical component should satisfy the principles of some version of the X-bar theory.
(4)Mapping between a base and Deep Structure (p. 71):
Thematic relations in the sense of Jackendoff (1972) and related work are determined by interaction of lexical properties and configurations of deep structure. (p. 72)
(5)Mapping between Deep Structure and Surface Structure (p. 72):
Derivations D = (K1, ..., Kn) are generated by the transformational component of the grammar as shown below.[1]
K1 (Deep Structure) K2 K3 ... Ki Ki + 1 ... Kn (Surface Structure[2])
(Some transformational rule is applied between Ki and Ki + 1.)
(6)About Surface Structure (p. 72):
Thematic relations are properly expressed in Kn [(i.e., in Surface Structure; EM)], though determined at K1 [(i.e., at Deep Structure; EM)].
(7)Mapping between Surface Structure and LF (p. 72)
[I]nterpretive rules extend the derivation D, carrying Kn to a representation in LF.[3]
(8)About LF (p. 71):
LF is the level which expresses whatever aspects of semantic representation are determined by properties of sentence-grammar.
(9)About PR's dependency on the side of the mapping between Surface Structure and LF (p. 72):
Some crucial aspects of PR may be determined by the extended derivation from Kn [(i.e., Surface Structure; EM)] to LF. Thus, as noted first by Lees (1960), deletion seems sensitive to some aspect of semantic representation, and under the present theory that means that the possibilities of deletion are in part fixed by properties of representation at LF or between Kn [(Surface Structure; EM)] and LF.
1.2.Rules and Conditions
Two transformational rules in (10) and three interpretive rules in (11) are discussed in the paper. Note that the assumption in (12) is important for(10) and(11).
(10)Transformational rules (=Chomsky 1977: (1)[4])
a.Move NP
b.Move wh-phrase
Both leave a trace (p. 76). As for trace theory, see section 1.3 below.
(11)Interpretive rules (=(C2))
a.Reciprocal rule: assign to each other the feature [+anaphoric to i] in a structure containing NPi
b.Bound anaphora: assign to a pronoun the feature [+anaphoric to i] in a structure containing NP. in the context [NP ---Possessive---Nx]
c.Disjoint reference: assign to a pronoun the feature [-anaphoric to i] in a structure containing NPi.
(12)Every non-terminal(e.g., NP) is assigned an index at Deep Structure (see p. 72 right below his (2)).
The rules in (10)are assumed to meet the conditions in (13), while those in (11) are subject to(15) and (16). Note that the opposite combination is not the case.
(13)Cycle (=(C3))
a.Transformational rules, e.g., [(10)], meet the condition of the (strict) cycle; the subjacency condition is a property of cyclic rules, i.e., part of the definition of the cycle.
b.[On the other hand, interpretive rules, e.g., (11), or postcyclic transformational rules[5] are not cyclic, and thus are not subject to the subjacency condition; EM]
(14)Subjacency Condition (=(C6), also see p. 73)
"A cyclic rule cannot move a phrase from position Y to position X (or conversely) in [(i)]."
(i)... X ... [ ... [ ... Y ... ] ... ] ... X ..., where and are cyclic nodes.
(Cyclic nodes = S' and NP)
(15)(=(C4))
Propositional-island condition (PIC) ("the "tensed-S condition" of the references cited." (p. 74))
"No rule can "involve" X and Y [(of (17); EM)] where is a finite clause (tensed-S)" (p. 74) and "is the cyclic node immediately dominating the category of Y" (p. 75). <==But, with my revision, we must assume that S' counts as a clause, which may be okay.
(16)(=(C5), also on p. 74)
Specified Subject Condition (SSC)
No rule can "involve" X and Y[(of (17); EM)]where contains a specific subject, i.e., a subject not containing Y and not controlled by X. ([It is assumed that Y contains Y. <== What are the effects/consequences of this? I.e., what does this give us/Chomsky?]))
(17)(=(C11))
... X ... [ ... Y ... ] ... X ...
The definition of 'involve' is as in (18).
(18)a.[A] transformational rule involves X and Y when it moves a phrase from position X to position Y (p. 75).
b.[A] rule of construal involves X and Y when it assigns Y the feature [+/- anaphoric to i], where X has the index i (p. 75).
1.3.Trace theory & the treatment of PRO
As we have seen in (12), every NP should get indexed at Deep Structure.
(12)Every non-terminal (e.g., NP)is assigned an index at Deep Structure (see p. 72 right below his (2)).
It is also assumed that movement leaves a trace. A trace comes to have an index which is assigned to the moved NP (as discussed on page 83 to 84). The trace left by movement is assumed to be a nonterminal, not a terminal.
(19)Before the movement; ... i ...
After the movement; ... i ... [NP it] ...(* ti )
The reason is the variable in Semantics is not always assumed to be identical to the trace of movement.
(20)a.Who did Mary say that John kissed t(=(C32))
b.for which x,x a person, Mary said that John kissed [x](=(C37))(variable x = trace)
(21)a.Whose book did Mary read t(=(C33))
b.for which x, x a person, Mary read [x's book](=(C35))(variable x trace)
PRO, on the other hand, is assumed as a non-terminal NP and as being base-generated with an index x, x a variable. The fixed index is assigned by a rule of control, which Chomsky does not explicitly define.
1.4.The distinction between move NP and move wh
Chomsky claims that the interpretation of the relation between the moved NP and its trace should be assigned through one of the interpretive rule, namely, through (11b).
(11)b.Bound anaphora: assign to a pronoun the feature [+anaphoric to i] in a structure containing NP. in the context [NP ---Possessive---Nx]
Therefore the relation between the moved NP and its trace is constrained by SSC and PIC. (See the discussion in the last paragraph of page 76.)
He does not explicitly suggest that the relation between the moved NP and its trace should be distinguished from that between the moved wh and its trace. However, it should be so, and the latter relation should not be constrained by SSC or PIC. Otherwise the sentence in (22) would be predicted as unacceptable, for the movement of who crosses the tensed S as well as the specified subject, John.
(22)(=(C30))
Who did [S John see t ]?
See the actual derivations in (27) and (28).
1.5.The illustrations of the rules and conditions
The illustrations of the rules and conditions are as follows.[6], [7]
(23)We want very much [S' for [NP picture of each other] to be on sale] (=(C7))
Base
(The NPs are indexed; (12))
DS: [NP1We] want very much [S' for [NP2 picture of [NP3each other] ] to be on sale]
SS: [NP1 We] want very much [S' for [NP2 picture of [NP 3each other] ] to be on sale]
(each other comes to have the feature [+anaphoric to 1]; (11a))[8]
LF: [NP1 We] want very much [S' for [NP 2 picture of [NP 3each other][+anaphoric to 1]] to be on sale]
=>each other is interpreted as anaphoric to we.
A key point is that (11a) is insensitive to subjacency.
(24)The men expected [S' that [NP pictures of each other] would be on sale] (=(C8))
Base
(The NPs are indexed; (12))
DS:[NP 1The men] expected [S' that [NP 2 pictures of [NP 3each other]] would be on sale]
SS: [NP 1The men] expected [S' that [NP 2 pictures of [NP 3each other] ] would be on sale]
(each other comes to have the feature [+anaphoric to 1]; (11a); no violation of SSC/PIC.)
LF: [NP 1The men] expected [S' that [NP 2 pictures of [NP 3each other][+anaphoric to 1] ] would be on sale]
=>each other is interpreted as anaphoric to the men.
Key pointsare(i) that (11a) is insensitive to subjacency, and (ii) there is no in between the men and each other which is a tensed-S and is the cyclic node immediately dominating the category of each other.
(25)Bill was killed t by John(=(C27))
Base
(The NPs are indexed; (12))
DS: [NP 1John] [VP be+en kill [NP 2Bill] by [NP 3e] ]
(Move NP; (10a), no subjacency violation)
[NP 1t] [VP be+en kill [NP 2Bill] by [NP 1John] ]
(Move NP; (10a), no subjacency violation)
SS: [NP 2Bill] [VP be+en kill [NP 2t] by [NP 1John] ]
(No violation of SSC/PIC).
LF: [NP 2Bill] [VP be+en kill [NP 2t] by [NP 1John] ]
A key point is that there is no subjacency violation.[9]
(26)John seems [S' to certain [S't to win] ](=(C9))
Base
(The NPs are indexed; (12))
DS: [NP 1e] seems [S' [NP 2e] to be certain [S' [NP 3John] to win] ]
(Move NP; (10a), no subjacency violation)
[NP 1e] seems [S' [NP 3John] to be certain [S' [NP 3t] to win] ]
(Move NP; (10a), no subjacency violation)
SS:[NP 3John] seems [S' [NP 3t] to be certain [S' [NP 3t] to win] ]
(No violation of SSC/PIC).
LF:[NP 3John] seems [S' [NP 3t] to be certain [S' [NP 3t] to win] ]
A key point is that the movement is successive cyclic and thus no subjacency violation.[10]
(27)*Bill seems [John to like t](=(C13a))
Base
(The NPs are indexed; (12))
DS: [NP 1e] seems [ [NP 2John] to like [NP 3Bill] ]
(Move NP; (10a), no subjacency violation)
SS: [NP 3Bill] seems [ [NP 2John] to like [NP 3t] ]
(No violation of PIC. However, the relation between Bill and its trace violates SSC.)
LF: [NP 3Bill] seems [ [NP 2John] to like [NP 3t] ]
=> the output representation crashes and thus the sentence cannot be accetable.
A key point is that the relation between an NP and its trace is subject to SSC/PIC.
(28)Who did John see? (=(C30))
Base
(The NPs are indexed; (12))
DS: [S' [S [NP 1 John] did see [NP 2 who] ] ]
(Move wh; (10a), no subjacency violation)
SS: [S' [NP 2 who] [S [NP 1 John] did see [NP 2t ] ] ]
(No violation of SSC and PIC.[11])
LF: [NP 3Bill] seems [ [NP 2John] to like [NP 3t] ]
=> the output representation crashes and thus the sentence cannot be accetable.
A key point is that the relation between an NP and its trace is subject to SSC/PIC.
(29)who did Mary hope [S' that Tom would tell Bill [S' that he should visit t] ](=(C10))[12]
Base
DS: [S' [S Mary did hope [S' that Tom would tell Bill [S' that he should visit who] ]
(Move wh-phrase; (10b), no subjacency violation)
[S' [S Mary did hope [S' that Tom would tell Bill [S'who that he should visit t] ]
(Move wh-phrase; (10b), no subjacency violation)
[S' [S Mary did hope [S'who that Tom would tell Bill [S't that he should visit t] ]
(Move wh-phrase; (10b), no subjacency violation)
SS: [S'who [S Mary did hope [S't that Tom would tell Bill [S't that he should visit t] ]
(No violation of SSC/PIC).
LF: [S'who [S Mary did hope [S't that Tom would tell Bill [S't that he should visit t] ]
A key point is that the movement is successive cyclic and thus no subjacency violation.
(30)I persuaded the men PRO to leave.(=(C17b))
Base
(The NPs are indexed; (12))
DS: [NP 1 I] persuaded [NP 2 the men] [PROx to leave]
SS: [NP 1 I] persuaded [NP 2 the men] [PROx to leave]
(The rule of control is applied; see 1.3 above.
Since the verb is persuade and the men is NP2, the infixed index x becomes 2.)
LF: [NP 1 I] persuaded [NP 2 the men] [PRO2 to leave]
=>the men is interpreted as the subject of leave.
(31)I promised the men PRO to leave.(=(C17b))
Base
(The NPs are indexed; (12))
DS: [NP 1 I] promised [NP 2 the men] [PROx to leave]
SS: [NP 1 I] promised [NP 2 the men] [PROx to leave]
(The rule of control is applied; see 1.3 above.
Since the verb is promise and I is NP1, the infixed index x becomes 1.)
LF: [NP 1 I] persuaded [NP 2 the men] [PRO1 to leave]
=>I is interpreted as the subject of leave.
2.WH-movement---General characteristics and the aim of the paper
It is said that the rule of wh-movement has the general characteristics in (32), which is illustrated in (33), for example.
(32)(=(C49))[13]
a.It leaves a gap.
b.Where there is a bridge [verb], there is an apparent violation of subjacency, PIC and SSC.
c.It observes CNPC
d.It observes wh-island constraints
(33)a.Who did Mary meet t(=(C40))
b.Who did you tell Mary that she should meet t(=(C41); tell is an instance of bridge verbs.)
b'.What did John quip that Mary wore t(=(C42b); quip is not a bridge verb.)
c.*Who did John believe [the claim [that Bill would visit t] ]
d.*What do you wonder [when [ John ate t] ]
The aim of the rest of the paper is to see whether (34) is correct.
(34)[T]he configuration [(32)] serve[s] as a kind of "diagnostic" for wh-movement. (C: p. 86)
3.Comparatives
(35)Claim:
Comparatives involvewh-movement (where wh is either AP or DP in the current framework), and the deletion of the moved-wh (which may optional for some dialects; see (39)).[14]
The sentence John is taller than Mary is, for example, is assumed to be derived as in (36).
(36)John is taller than Mary is
Base
(The AP as well as NP is indexed; (12))
DS of the than clause:... [than [S' [S John 1 is [AP 2 what] ] ]
(wh-movement)
SS of the than clause:... [than [S' [AP2what] [S John is [AP 2t] ] ]
(deletion of what)
PR of the than clause: ... [than [S' [AP 2what] [S John is [AP 2t] ] ]
LF of the than clause: ... [than [S' [AP 2what] [S John is [AP 2t] ] ][15]
Evidence 1; Comparatives essentially have the properties in (32).
(37)(=(C52))
a.Mary isn't the same as [she was five years ago]
b.Mary isn't the same as [John believes [that Bill claimed [that she was five years ago] ] ]
c.*Mary isn't the same as [John believes [the claim [that she was five years ago] ] ]
d.* Mary isn't the same as [I wonder [whether she was five years ago] ]
(38)a.Mary isn't taller than [she was five years ago]
b.Mary isn't taller than [John believes [that Bill claimed [that she was five years ago] ] ]
c.*Mary isn't taller than [John believes [the claim [that she was five years ago] ] ]
d.* Mary isn't taller than [I wonder [whether she was five years ago] ]
Evidence 2; many dialects of American English allow to have a wh-word.[16]
(39)(=(C51))
a.John is taller than what Mary is.
b.John is taller than what Mary told us that Bill is.
Further support: Comparatives exhibit cross-over phenomena, just like their question counterparts do.
(40)Comparatives (=(C60), originally cited from Bresnan)
a.more studentsi flunked than ___ thought theyi would (flunk)
b.*more studentsi flunked than theyi thought ___ would (flunk)
(41)Questions (=(C62))
a.how many (students)i [t thought [theyi would flunk] ]
b.*how many (students)i [ they thought (did they think) [ t would flunk] ]
It should be noted here that Chomsky (1977) distinguishes the account for comparatives (which is called comparative deletion (CD) in Roumi's lecture (Sp. 2005)) from that for so called comparative subdeletion cases (CSD), as suggested on page 89.
(42)Chomsky's accounts;
a.CD involves wh-movement of the whole AP (or DP).
b.CSD and CD are not subcases of a single process; CSD does involve the different process from CD (i.e., from wh-movement).
4.Topicalization
(43)Claim: topicalization is the process containing;
a.the base-generation of a dislocated NP (=topic NP) in the sentence initial position.
b.the movement of a wh-word to the comp position the mother node of which is a sister of a topic NP.
c.the obligatory deletion of the moved-wh.
The following phrase structure rules are crucially assumed here.
(44)S" TOP S'
S' COMP {S/S"}
The sentence this book, I really have, for example, is assumed to be derived as in (45).
(45)this book, I really have
Base
(The NPs are indexed; (12))
DS:[S" [NP(=TOP) 1 this book] [S' [COMP ] [S [NP 2 I] really have [NP3what] ] ]
(wh-movement)
SS:[S" [NP(=TOP) 1 this book] [S' [COMP[NP 3what] ] [S [NP 2I] really have [NP 3t] ] ]
(deletion of what)[17]
PR: [S" [NP(=TOP) 1 this book] [S' [COMP[NP 3what] ] [S I really have [NP 3t] ] ]
LF: [S" [NP(=TOP) 1 this book] [S' [COMP[NP 3what] ] [S I really have [NP 3t] ] ][18]
Due to the deletion of what, the embedded sentence is now an open S. Chomsky assumes (though not implicitly argues) that the interpretation of a sentence which contains an open S involves "rules of predication" (p. 81), which are not sensitive to SSC/PIC. See also the last paragraph of Chomsky 1977: 100.
Evidence; Topicalization does yield the configuration in (32).
(46)(=(C63))
a.this book, I really like
b.this book, I asked Bill to get his students to read
c.*this book, I accept the argument that John should read
d.*this book, I wonder who read
5.Left-dislocation
The sentences named Left-dislocationsare minimally different from topic sentences in the way (i) that they have a pronoun in the place that a trace is assumed in the case of topic sentences, and (ii) that the dislocated (topic) NP is optionally accompanied by the expressions as for ... or as far as ... is concerned.
(47)Claim:
Left-dislocations do not requirewh-movement.[19]
Evidence; the sentence is fine even in the complex NP island configuration.
(48)as far as John is concerned, I will never believe the claims that have been made about him(=(C24))
6.Cleft sentences
Cleft sentences are assumed to have the underlying structure shown in (49). In other words, Cleft sentences involve topic constructions.[20]
(49)it - is- [S" TOP S'](=(C88))
(50)Claim: Cleft sentences involve;
a.the base-generation of a dislocated phrase (=focused phrase) in the top position of S" in (49).
b.the movement of a wh-word to the comp position the mother node of which is a sister of a topic NP.
c.the obligatory deletion of the moved-wh.
The sentence it is this book that I really have, for example, is assumed to be derived as in (51).
(51)it is this book that I really have
Base
(The NPs are indexed; (12))
DS:it is [S" [NP(=TOP) 1 this book] [S' [COMP that ] [S [NP 2 I] really have [NP3what] ] ]
(wh-movement)
SS:it is [S" [NP(=TOP) 1 this book] [S' [COMP that-[NP 3what] ] [S [NP 2I] really have [NP 3t] ] ]
(deletion of what)
PR: it is [S" [NP(=TOP) 1 this book] [S' [COMP that-[NP 3what] ] [S I really have [NP 3t] ] ]
LF: it is [S" [NP(=TOP) 1 this book] [S' [COMP that-[NP 3what] ] [S I really have [NP 3t] ] ][21]
Evidence; Cleft sentences yield the configuration in (32).
(52)(=(C86))
a.it is this book that I really like
b.it is this book that I asked Bill to get his students to read
c.*it is this book that I accept the argument that John should read
d.*it is this book that I wonder who read
7.Indirect questions
(53)Claim:
"[A] rule of wh-movement is involved, analogous to direct questions." (p. 97)
Evidence 1; Finite indirect questions exhibit the contrast suggested in (32).
(54)(=(C94))
a.I wonder [who John saw]
b.I wonder [who John believed [that Mary would claim [that Bill would visit] ] ]
c.*I wonder [who John believed [the claim [that Bill would visit] ] ]
e.*who2 did you wonder [who1t1 to saw t2]
Evidence 2;The infinitive cases also exhibit the contrast.
(55)(=(C97))
a.I wonder [who to see]
b-1.I wonder [who to order Mary [to promise [to visit] ] ]
b-2.I wonder [who to persuade Mary [that she should promise [to visit] ] ]
c.*I wonder [who to insist on [the principle [that Bill should visit] ] ]
d.*who2 [ [to whom]1 to give t2 t1 ]
(cf. I wonder (don't remember) [what to give t to whom]; I wonder [to whom to give what t])
8.Infinitival relatives
(56)Claim (p. 98):
a.Relative head noun is base-generated.
b.Infinitival relatives involve wh-movement.
c.The deletion of the moved wh is obligatory unless Recoverability of deletion prevents it from applying.
8.1.Basic properties
Before going into the discussion of the claim, let us first briefly look at the basic properties / assumptions discussed in Chomsky 1977.
The movement of wh phrase has been assumed in the cases of finite relatives, for they can optionally have a wh-phrase (though they are not discussed in detail in the paper). It seems that there is one general rule for Modern English; "sequences of the form wh-phrase + complementizer are not permitted." (p. 86)