Sunteți pe pagina 1din 8

Georgetown University 3/26/02

Tense, Case and the Nature of Syntactic Categories


David Pesetsky, MIT
Esther Torrego, UMass/Boston
Part One: that is T, Nominative is uT [Pesetsky & Torrego 2001]
1. T-to-C, English that
Nature of that
(1) T-to-C asymmetry in matrix questions (Koopman 1983)
[non-subject wh --> "optional" T-to-C]
a. What a nice book Mary read __!
b. What did Mary read __?
[subject wh --> no T-to-C]
c. Who __ read the book?
d.*Who did __ read the book?/*What a nice person did read the book!
(2) Belfast English: T-to-C asymmetry in embedded declaratives
(Henry 1995, 108-9; p.c.)
[non-subject wh --> (optional) T-to-C movement]
a. Who did John say [did Mary claim [had John feared [would Bill attack __]?
[subject wh --> no T-to-C movement]
c. Who did John say [ __ went to school]
d. *Who did John say [did __ go to school]? (bad unless do is emphatic)
(3) "That-trace effect" (Perlmutter 1971)
[non-subject wh --> optional that ]
a. What do you think [Mary read __]?
b. What do you think [that Mary read __]?
[subject wh --> no that]
c. Who do you think [__ read the book]?
d. *Who do you think [that __ read the book]?
Pesetsky & Torrego (2001): The T-to-C asymmetry and the that-trace effect are
instantiations of the same phenomenon.
(4) Nature of English that
That is not C, but an instance of T moved to C.
(5) Nature of English C
C is a null morpheme.
The relation between that and T in its original position is akin to "clitic doubling" and
"partial wh-movement".
The source of the T-to-C asymmetry
Agree
(6) Probe/Agree
1. An uninterpretable feature uF on acts as a probe (Chomsky 2000) whose goals are
constituents bearing F that are c-commanded by and closest to [see (7) below].
2. For each probe-goal pair, if one member of the pair bears an uninterpretable feature
uF and the other member bears an instance of F, Agree takes place between the two
instances of the feature and uF is marked for deletion [notated uF].
(7) Accessible Closest F (ACF) [simplified from Chomsky 1995, p. 296]
A probe uF on takes as a goal if
(i) bears F; and
(ii) is c-commanded by ; and
(iii) no that also bears F c-commands and is c-commanded by .
Move
(8) EPP
If uF on is also [+EPP], one of its goals G
i
moves to (" attracts G
i
").
In (1)-(3): C bears [uT, +EPP] and [uWh, +EPP].
Why does T move to C, rather than TP?
Idea: When uF on is [+EPP] and the goal of uF that undergoes movement is the
complement of , the result is head movement.
(9) Head Movement Generalization [Travis's (1984) Head Movement Constraint of ]
Suppose a head H attracts a feature of XP as part of a movement operation.
(i) If XP is the complement of H, copy the head of XP into the local domain of H.
(ii) Otherwise, copy XP into the local domain of H.
It is thus the closest bearer of T and the closest bearer of wh with which the uT and uWh
features of C agree.
Why are (1c), (3c) and (2c) acceptable despite the absence of T-to-C?
(10) The nature of nominative case
Nominative case is uT on D. [cf. Haeberli (1999);Williams (1994)]
Thus, the nominative wh-phrase is one of the goals probed by uT on C, and
movement of the nominative phrase to Spec,CP can satisfy the EPP property of
this feature.
(11) uT feature on subject is marked for deletion by T before it moves to Spec,CP...
[
TP
[
DP
subject, uT, ]
i
[T, u] [ t-subject
i
bought the book] ]
.... but is not deleted until at least CP is complete.
-2-
(12) The T-to-C asymmetry
a. [C, uT, uWh] [
TP
[Mary, uT] T [
VP
read [what, +wh]] ]
closest T: subject Mary and T
closest wh: object what
wh C'
C+T TP
uT [+EPP]
uWh [+EPP] DP T'

T VP
..wh...
b. [C, uT, uWh] [
TP
[who, uT, uWh] T [
VP
read the book] ]
closest T: subject who and T
closest wh: subject who
wh C'
uT
C TP
uT [+EPP]
uWh [+EPP] wh T'
uT
T VP
Why is T-to-C not merely unnecessary, but impossible in (1c-d),(3c-d),(2c-d)?
uWh on C probes the nominative wh-phrase, and satisfies its EPP feature by wh-
movement of the subject to Spec,CP.
uT on C takes as goals two equally close phrases: TP and Spec,TP. In principle, either
phrase could move and satisfy the EPP feature of uT:
1. If TP moves, EPP satisfaction for uT on C will require T-to-C movement in
addition to independently necessary wh-movement.
2. If the nominative Spec,TP is picked, EPP satisfaction for uT on C is accomplished
by movement of the wh-phrase that must move anyway.
(13) Economy of movement
The EPP properties of uF on a head H are satisfied by the smallest possible number
of movement operations.
2. The that-omission asymmetry in finite CP
Finite clauses
(14) "That-omission" asymmetry (Stowell 1981; Kayne 1980b)
[non-subject CP--> optional that]
a. Mary thinks [that Sue left].
b. Mary thinks [Sue left].
[subject CP--> obligatory that]
c. [That Sue left] is obvious.
d. *[Sue left] is obvious.
Probe uT on C finds two goals: TP and Spec,TP. In principle, either can satisfy
the EPP property of uT on C:
1. Move T to C (that)...
(15) Mary expects [
CP
[
T
that]
j
+[C, uT] [
IP
Sue will buy the book.]]
2. ... or move the nominative subject to C (phrasal movement or feature movement):
(16) Mary expects [
CP
[Sue, uT]
j
[C, uT] [
IP
t-Sue
j
will buy the book.].
Both options satisfy Economy of Movement equally well, since no wh-feature is at
issue. That is why we see two variants, unlike in the case of the T-to-C
asymmetry.
(17) a. ... [
CP
[
T
that]
j
+[C, uT] [
IP
Sue will
j
buy the book.]] ...
"...that Sue will buy the book..."
b. ... [
CP
[Sue, uT]
j
[C, uT] [
IP
t-Sue
j
will buy the book.]...
"...Sue will buy the book..."
(18) An account of the that-omission asymmetry (14)
CP in (17a) [(14a,c)] has "real" T in its head.
CP in (17b) [(14b,d)] has uT in its head -- which is deleted at a later point in the
derivation.
Once uT is deleted, C still bears T-features in (17a), but not in (17b).
(19) Match Condition on EPP satisfaction
If [+EPP] features of a head H attract a set of phrases K, each syntactic feature of
H must be present on some member of K.
Assume that uT on the head of a CP is deleted before EPP of u on T
s
triggers
movement. The Match Condition (19) prevents from moving so as to satisfy this
EPP property. This yields (14).
-3-
3. T, for, to
Infinitival clauses: unpronounced T-to-C movement
(20) Realis infinitive asymmetry (Stowell 1982, Pesetsky 1989)
[non-subject CP--> realis or irrealis infinitive]
a. Mary wanted [PRO to learn the election results] [irrealis]
Mary would hate [PRO to lose the game]
Mary hates [PRO to lose games] [generic]
b. Mary hated [PRO to learn the election results. [realis: factive]
Mary managed [PRO to lose the game]. [realis: implicative]
[subject CP--> only irrealis/generic infinitive]
c. [PRO to learn the election results] would shock me. [irrealis]
[PRO to lose the game] would prove they are idiots.
[PRO to learn the election results early] is a crime. [generic]
[PRO to lose games like this] annoys the public.
d. ??[PRO to learn the election results] shocked me. [realis: factive]
??[PRO to lose the game] proved they were idiots.
*[PRO to lose the game] was managed by the team. [realis: implicative]
(21) Stowell's Conjecture:
Irrealis/generic infinitives, unlike realis infinitives, have T in C.
Conclusion: T of irrealis -- but not realis -- infinitives moves to C.
for as overt T moved to C in infinitives
Infinitives with for share the semantics of irrealis infinitives without for. [also a
link to the semantics of if-clauses; Bresnan, Carstairs-McCarthy, Pesetsky 1989]
(22) a. ??We hated for Mary to learn she'd won the lottery.
b. We would hate for Mary to learn she'd won the lottery.
(23) a. We arranged for there to be sushi at the reception.
b. I would prefer for there to be sushi at the reception.
(24) a. %I tried for Mary to get elected chairman.
b. *I managed for Mary to get elected chairman.
For raises from T to C (cf. Rosenbaum 1967), yielding a for-trace effect:
(25) The for-trace effect
*Who would you like [for __ to buy the book]?
The existence and properties of for provide support for the claim that irrealis infinitives
involve T-to-C movement, even when T is not pronounced.
Part 2: Accusative case is T
4. The object question
(10) The nature of nominative case
Nominative case is uT on D.
Naturally suggests:
(26) The nature of structural case
All "structural case" is actually uT on D.
"Case Filter phenomena"
Contrast in distribution of DPs vs. CPs (and PPs).
We will not explain the existence of a "Case Filter", but take as our point of departure
the restatement of the Case Filter as the Argument-Tense Condition of (27).
(27) Argument-Tense Condition (ATC) [Case Filter]
An argument (whatever its category) must bear T (uT or iT).
(28) Internal syntax of DP vs. CP arguments (P&T 2001)
D bears uT which cannot be checked internal to DP.
[DP is not self-sufficient w.r.t. T.]
C bears uT which can (and must) be checked internal to CP.
[CP is self-sufficient w.r.t. T]
(29) The object question
What checks uT on an (accusative) DP object?
5. Complement to A: only self-sufficient arguments
Before answering the object question as stated in (29), let us make sure that it is the
right question.
Prediction: Suppose there were a syntactic environment E such that nothing could check
uT on an argument in E (a "non-case position"). The argument could be a
CP, but could not be a DP. The argument would need to be "self-sufficient".
Proposal: Complement of AP is such an environment.
(30) DP complement to A: impossible
*Bill was afraid the storm.
(31) CP complement to A: acceptable (all types)
a. Bill was afraid that the storm will be destructive.
b. Bill was afraid the storm will be destructive.
c. Bill was careful to read the instructions.
d. Bill was eager to read the instructions.
-4-
6. Why PPs are self-sufficient: P is T
(32) PP complement to A
Bill was afraid of the storm.
If PP is an an argument of afraid, P must be self-sufficient, like CP.
P must bear T-features.
(33) The P-trace effect (Kayne 1979, "Two Notes on the NIC")
a. *John is the one who I'm counting on __ marrying her.
b. *The only one who we're in favor of __ studying linguistics is John.
(34) a. the student that he resented __ reading his unpublished paper
b. *the student that he apologized for __ reading your unpublished paper
c. the unpublished paper that I apologized for the student reading __
(35) a. This much attention I don't remember __ being paid to John and Mary.
b. *This much attention I didn't talk about __ being paid to John and Mary.
c. This much attention I didn't talk about Bill paying __ to John and Mary.
(36) a. the headway that I anticipated __ being made on this project
b. *the headway that I thought about __ being made on this project
c. the headway that I thought about us making __ on this project
(37) P preceding gerund is T attracted by C. The P-trace effect follows as an
instance of the that-trace effect.
apologize CP
C TP
DP T'
the student
T VP
for
reading my paper
(38) The nature of P
P and T belong to the same supercategory (which we will call T).
This makes sense of the fact that elements of the prepositional vocabulary appear
in T positions in English (and other languages, e.g. Irish) --
(39) a. John considers there to be many reasons for this.
b. Mary kept there from being a riot.
-- and also appear as T moved to C (cf. (25))
(40) Mary would prefer [for+C there to be more than one option].
7. The nature of head movement
On the other hand, the result of T
prep
-to-C movement here cannot be a CP headed
by a C containing T
prep
. The external syntax of PPs differs from the external
syntax of CPs, e.g. in the impossibility of PP as subject of TP (vs. CP):
(41) a. [For there to be more than one option] would be preferred.
b. *[For Bill reading the paper] was apologized.
-- and the possibility of movement stranding T
prep
:
(42) a. *[There to be more than one option] Mary would prefer for __.
b. [There being more than one option] Mary already apologized for __.
Proposal: Movement of T
prep
to C differs from the standard picture of head
movement in two respects:
a. T
prep
is not adjoined to the head C, "morphologically merging"with it.
b. T
prep
, rather than C, projects after movement.
(43) P + Gerund
CP
C TP
prep ====>
DP
T
prep
VP
TP
prep
(=PP)
T
prep
CP
C TP
prep
DP
T
prep
VP
Puzzles of standard head movement [Fanselow (2002), Matushansky (2002)]:
Movement to a non-c-commanding position.
Target head and moving head form a "term"accessible to later operations.
EPP satisfied in two ways -- specifier formation and affixation.
Features of target head and moving head are both present in the label of the target head
[cf. (18)].
-5-
Matushansky (2002): Standard picture conflates two distinct operations:
1. Moving head H
m
merges in the same fashion as a phrasal head.
a. All movement is to a c-commanding position.
b. EPP satisfiable in only one way -- specifier formation.
2. Optional: H
m
and probe head undergo "morphological merger" (m-merger), which
forms a single word out of H
m
and the target head H
t
.
c. H
m
and H
t
expected to function as a single term.
d. Possible to understand why features of both H
m
and H
t
co-project.
New possibilities available with Matushansky's proposal:
If m-merger fails to take place, "excorporation phenomena" are not unexpected.
H
m
might project under some circumstances, rather than H
t
.
Thus:
-->Clausal T-to-C movement as in Pesetsky and Torrego (2001) instantiates head-
movement plus m-merger.
-->T
prep
-to-C movement, forming a prepositional phrase, involves no m-merger and
involves projection of H
m
as in (43).
We propose the same analysis for prepositional phrases with a DP (rather than
gerund) object. (See Szabolcsi 1983, 1987 and Pesetsky and Torrego 2001 for the idea
that CP and DP are essentially identical categories.):
(44) P + DP
DP
D TP
prep ====>
T
prep
VP
TP
prep
(=PP)
T
prep
DP
D TP
prep
T
prep
VP
Consequences:
The fact that DP/CP can move, stranding T
prep
, follows from the fact that DP/CP
excluding T
prep
is a maximal projection.
The presence of a "T
prep
-trace" effect in gerunds follows from the fact that T
prep
moves
from inside DP/CP.
Prepositional phrases are thus "self-sufficient" in that they start as DPs (CPs) whose
head contains uT, which is marked for deletion by an instance of T (T
prep
) that enters an
Agree relation and satisfies the EPP property of uT on D (C).
When head movement yields an (English) "PP", it is not followed by morphological
merger, and involves projection by T
prep
rather than D (C).
A "T
prep
-trace" effect arises in part because wh-movement of the subject of a gerund
blocks T
prep
-to-C raising -- but also because in the relevant configurations, the absence
of T
prep
-to-C raising leaves the subcategorization property of the higher predicate (e.g.
apologize in (34b-c)) unsatisfied.
"Pure" DPs? Either:
1. lack T
prep
(i.e. D takes NP as its sister); or else
2. select a "defective" TP which enters an Agree relation with uT on D, but does
not mark that feature for deletion.
8. Complement to V: verbal T
o
(45) Mary T
S
read ... [the book, uT]
Since DP is not "self-sufficient", (45) must contain an instance of T that checks uT on
the DP the book.
The behavior of complements to A taught us inter alia that T
S
(the checker of uT on the
subject) cannot also check uT on the object.
Conclusion: (45) contains another instance of T:
(46) Structure of the verbal clause
SUBJ T
S
v T
O
V OBJ
(47) Particle in English is overt T
o
John T
S
[
v
drank] [
TPo
[
To
up] [
VP
V the wine]]]
The object of a transitive V may not freely be a PP (i.e. a T
prep
P).
This contrasts with the object of A or N:
(48) a. Bill was afraid of the storm.
b. Bill's fear of the storm
c. *Bill feared of the storm.
(49) Special property of verbal T
O
The goals of u on verbal T
O
must bear uT.
What goes wrong in (48c): of is an instance of iT, not uT.
-6-
9. When do features marked for deletion delete? Complements of V vs.
subjects
P&T 2001: uT in C is deleted at the end of the CP phase.
If so, how can we reconcile (49) with the acceptability of (50a-b)?
(50) a. Mary said [that Sue will buy the book].
b. Mary said [Sue will buy the book].
(51) a. Mary said T
o
V [
CP
[
T
that]
j
+[C, uT] [
IP
Sue will
j
buy the book.]]
"...that Sue will buy the book..."
b. Mary said T
o
V [
CP
[Sue, uT]
j
[C, uT] [
IP
t-Sue
j
will buy the book.]
"...Sue will buy the book..."
[A wrong answer: Verbal T
O
is optional. This would incorrectly let (48c) back in.]
An right answer:
uT in C is deleted eventually (to explain the behavior of CP subjects; cf. section 2), but
not as early as the end of the CP phase.
If the Match Condition explains the fact that T-to-C movement (e.g. that) is obligatory
in CP subjects, uT on CP must be deleted no later than the end of the vP phase (cf.
Chomsky 2000; Nissenbaum 2000). No evidence indicates deletion earlier than this.
Since VP (unlike vP) is not a phase, T
o
still sees uT on C in both (51a) and (51b).
Why are PPs not (generally) found as specifiers of T
S
?
(52) Special property of [verbal] T
S
[compare (49)]
The goals of u on [verbal] T
S
must bear uT.
(53) Four types of argument
a. PP c. that/for-CP
Head is iT; u Head is bimorphemic as the result of m-merger
(i) C has uT, i
(ii) T in C has iT, u
T XP
T XP
... TP
prep
(=PP)
T DP
... CP ...
D T
prep
P
uT TP

NP

b. DP d. that-less (for-less) CP
Head has uT, i Head has uT, i
T XP T XP
... DP ... CP ...

D NP DP C'
uT uT
C TP
uT
a. PP goal: Condition (52) excludes a PP as goal of T
S
, because it bears only iT on its head.
b. DP goal: Condition (52) permits a DP with unchecked uT as goal of T
S
,.
c. that/for-CP goal: Condition (52) permits a that-clause (or irrealis/generic infinitival CP)
as goal of T
S
, because it bears uT (in addition to iT) on its head. Crucially, (52) holds
before deletion of uT on the goal of T
o
.
d. that-less (for-less) CP goal: If condition (52) holds before deletion of uT in the vP
phase, it also permits a finite clause without that (or a realis infinitive) as goal of T
S
. The
Match Condition applies after uT is deleted and excludes these CPs as specifiers of TP
S
.
(54) Timing of T
S
-related featural requirements
step 1. vP is built.
step 2. Merge T
S
and vP.
step 3. u on T
S
acts as a probe finds goal(s) within vP.
step 4. Any instances of uF that were marked for deletion within the vP phase delete.
step 5a. EPP property of u on T
S
triggers movement, subject to the Match
condition.
step 5b. Mark uF for deletion if has undergone agree.
(55) Timing of uF -deletion: the general scenario
1. Deletion: uF (uF marked for deletion; i.e. "checked") in the domain of a phase head H
is deleted once HP is maximal.
2. EPP: EPP properties of the features of are satisfied after step 1. deletion.
3. Mark for deletion:
Newly agreeing uninterpretable features are marked for deletion after step 1.
(56) Maximality of a phase
A phase is maximal once
(i) has merged with , projecting (rather than ), and
(ii) all possible checking has taken place.
C
T T u
that
i

+
-7-
10. Complements to N: nominal T
o
PP/DP
(57) PP vs. *DP as complement of N
a. *Bill's fear the storm
b. Bill's fear of the storm
The contrast in (57) is consistent with an absence of T
O
in nominals, i.e. with the
analysis we accorded to APs.
But the behavior of CP complements to N (discussed below) suggests the opposite:
the presence of a T
O
that requires a goal that is not merely self-sufficient, but
actually bears iT.
Finite CP
(58) That obligatory in finite CP complement of N (Stowell 1980, 1981)
a. I liked your proof that Mary could not have committed the crime.
b.* I liked your proof Mary could not have committed the crime.
c. My demonstration that Sue was insane was accepted by the court.
d. *My demonstration Sue was insane was accepted by the court.
(46) Structure of the verbal system (clause)
SUBJ T
S
v T
OV
V OBJ
(59) Structure of the nominal system (derived nominals)
... T
ON
N OBJ
(49) Special property of verbal T
O
The goals of u on verbal T
O
must bear uT.
(60) Special property of nominal T
O
The goals of u on nominal T
O
must bear iT.
Infinitival CP
Generalization (60) extends to infinitival complements, if "irrealis" interpretation is a
sign that T has moved to C.
(61) Only irrealis allowed in complement of N
a. nominalizations of realis-selecting predicates
*Mary's hate/hatred to ride in the back seat *John's dislike to go home
*Sue's love to solve problems *Bill's bother to leave early
*Bill's luckiness to win the lottery *Bill's condescension to leave
*Bill's dare to leave *Bill's disdain to leave early
*Bill's help to leave early *Bill's management to leave early
*Bill's neglect to leave early *Bill's omission to leave early
*Bill's venture to leave at midday *Bill's scorn to leave early
b. nominalizations of realis-selecting predicates
Mary's desire to win Harry's need to be accepted
Bill's agreement to leave early Bill's arrangement to leave early
John's wish to be allowed to school Sue's eagerness to win the lottery
Bill's attempt to leave early Bill's choice to leave early
Bill's consent to undergo the operation Bill's decision to leave early
Bill's demand to leave early Bill's endeavor(s) to leave early
Bill's hope to leave early Bill's intention to leave early
Bill's learning to leave early Bill's offer to leave early
Bill's plan to leave early Bill's preparation(s) to leave early
Bill's promise to leave early Bill's proposal to leave early
Bill's refusal to leave early Bill's request to be allowed to leave early
Bill's resolution to leave early Bill's struggle to leave early
Bill's undertaking to leave early. Bill's vow to leave early
[Only known exception: Bill's failure to leave early. Fail is an implicative verb.]
(62) Crosscheck: complement to N may be a for-infinitive
Mary's desire for Bill to win. [etc.]
Since nominal and verbal T
o
have opposite requirements, why are V-complements
and N-complements not in complementary distribution?
(63) Non-complementary distribution of V- and N-complements
a. Mary said
V
that Sue will buy the book.
b. Mary's proof
N
that Sue will buy the book.
c. Mary prefers
V
to leave early.
d. Mary's preference
N
to leave early.
T-to-C movement creates a C containing both uT and iT. Such CPs satisfy the
requirements of both verbal and nominal T
o
.
Cross-check: any kind of CP may be an A-complement
Historical perspective on "non-case positions":
1981 GB Theory: Complement to N and complement to A are both non-case positions.
Result: *DP, ok CP, ok PP.
Chomsky (1986): Complement to N and complement to A are both inherent case
positions. Result: *DP, ok CP, ok PP.
Our perspective:
Two possible reasons for a restriction "*DP / ok CP / ok PP" on an argument:
(i) There is no category that can check uT on the argument, therefore the head of the
argument must either bear iT or else must check uT internally.
(ii) There is a category that could in principle check uT on the argument, but it
requires iT on the element that it probes.
Case (i) is instantiated by complements of A (cf. 1981 GB theory).
Case (ii) is instantiated by complements of N (cf. Chomsky (1986).
-8-
A difference between complements of A and N is predicted when the complement is
a CP.
(i) A should allow any sort of CP complement, since all can check uT on C
internally.
(ii) N should allow only a CP complement in which T has raised to C.
Prediction (ii) was already demonstrated in (61):
a finite CP complement to N must show that; and
an infinitival CP complement to N must be irrealis.
Prediction (i) is demonstrated by the fact that:
a finite CP complement to A shows that optionally; and
an infinitival CP complement to A may be irrealis or realis.
(64) Optional that in CP complement to A
a. Mary is certain (that) the world is round.
(cf. *Mary's certainty the world is round)
b. Bill is aware (that) Mary left early.
(cf. *Bill's awareness Mary left early)
(65) Realis infinitival complements to A
a. Mary was happy to learn the election results.
(cf. *Mary's happiness to learn the election results)
b. Sue was lucky to pick a topic that no one had worked on.
(cf. *Sue's luck to pick a topic that...)
c. John was very clever to figure this out.
(cf. *John's cleverness to figure this out)
d. Bill was careful to drive on the left in England.
(cf. *Bill's care to drive on the left in England)
11. What distinguishes N, V, and A?
a. PP b. DP c. that-clause
or irrealis
infinitve
d. that-less finite
clause or realis
infinitive
1. A-complement
2. V-complement
3. N-complement
4. subject of TP
1. A-complement no T
o
Must bear iT or uT, and must be self-sufficient (excludes b. and d.).
2. V-complement verbal To
Must bear uT (excludes a.).
3. N-complement nominal To
Must bear iT (excludes b. and d.).
4. subject of TP verbalTs whose goals occupy specifier of vP
Must bear uT before deletion of checked uninterpretable features (excludes a.).
Must bear iT or uT deletion (excludes d.)
The nature of syntactic categories
The fact that the categories N, V, and A exhaust a natural space of possibilities related to T
and case suggests that these differences might be definitional. For example, we might take
these results as evidence for the following proposal:
Predicates are category-neutral (Borer 1988, 1991, 1993; Marantz 1997).
An adjective is a predicate not combined with T
o
.
A verb is a predicate combined with verbal T
o
.
A noun is a predicate combined with nominal T
o
.

S-ar putea să vă placă și