Author: Prof. Ellen Thompson
        ethomps@upracd.upr.clu.edu
Title: The Syntax and Semantics of Gerundive Relatives
Section 2


The Syntax and Semantics of Gerundive Relatives



1.0  Introduction
Recent work on the syntax of tense shows that there is an principled
relationship between the meaning and the phrase structure representation of
temporal information (Hornstein 1977, 1981, 1990, Zagona 1988, 1990, Giorgi
and Pianesi 1991, Stowell 1993). I contribute to this discussion here by
arguing that syntactic locality at LF constrains the interpretation of
temporal relations; temporal dependency between times requires the times to
be in the same checking domain at LF.
	I adopt the semantic analysis of Reichenbach (1947), whereby tenses are
composed of different times, and I claim that syntactically, Event time
information is associated with VP, while Speech time information is
associated with an inflectional projection. Based on this syntax and the
claim that temporal dependency requires times to be in the same checking
domain, I present an analysis of the syntax and semantics of gerundive
relative clauses. I argue that gerundive relatives are temporally dependent
on the main clause tense and thus are required to be within the checking
domain of a matrix time at LF.
	This analysis of gerundive relatives has implications for the temporal
syntax of arguments: I argue that reflexes of the syntax of tense are
evident in the interpretation site of subjects at LF. Since the Event time
is associated with VP, a gerundive relative in subject position that is
interpretted in VP internal position receives a reading where the gerund is
temporally dependent on the Event time. Because the Speech time is
associated with IP, a gerundive relative in subject position that is
interpreted in IP position receives a reading where the gerund is
temporally dependent on the Speech time.
	An example which illustrates the interpretive effect of the position of
the subject on a gerundive relative in subject position is provided in
(1a-b). The example in (1a) may have a reading where the waiting is
interpreted as occuring at the time of the Event of the matrix clause, the
storming into the room, or a reading where the waiting is interpreted as
occuring at the Speech time, the time of utterance. However, the
existential construction version of (1a) in (1b) does not permit the Speech
time reading; the waiting here is necessarily interpreted as occuring at
the time of the event of storming into the room.

(1)	a.	Three passengers waiting for the flight stormed into the room.
	b.	There stormed into the room three passengers waiting for the flight.

	Assuming that existential constructions involve interpretation of the
associate of the expletive within VP (den Dikken (1995) and Groat (1995)),
we can explain why only the Event time reading is permitted in the
existential construction; when the subject is located within VP, it is in
the checking domain of the Event time.
	This paper is organized as follows: in section 2, I outline the aproach 
to tense adopted here. I assume the semantic analysis of Reichenbach (1947),
whereby tenses are composed of different times, and I claim that
syntactically, Event time information is associated with VP, while Speech
time information is associated with an inflectional projection.
	In section 3, I show that gerundive relatives have a reduced tense
structure, and that the temporal interpretation of gerundive relatives is
dependent on the tense of the main clause. Assuming the syntactic
representation of tense developed here, I claim that the reduced tense
structure of gerundive relatives correlates with a reduced clause
structure, in section 4.
	Section 5 shows that the tense of gerundive relatives is ambiguous 
between
being dependent on the Event time or the Speech time of the main clause.
Assuming that the Event time is represented in VP, as well as the proposal
that temporal dependency requires syntactic locality, I claim that a
gerundive relative in subject position which is interpreted with respect to
the Event time correlates with VP-internal position of the subject at LF.
Since the Speech time is associated with IP, when the gerundive relative is
interpreted with respect to the Speech time, the subject is located in
Spec, IP at LF.
	Evidence for this analysis of the syntax of subjects is discussed from
constructions involving coordination, existential there, scope of
quantificational and cardinality adverbials, extraposition, and ECM
constructions, in section 6. I argue that in constructions where the
subject is located inside VP at LF, the gerundive relative is interpreted
with respect to the Event time of the clause, whereas when the subject is
located in IP at LF, the gerundive relative receives a Speech time reading.
I also discuss temporally dependent readings of full relatives, showing
that these interpretations are predicted by the present analysis.
	Section 7 discusses the interaction between gerundive relatives and
binding-theoretic reconstruction effects. Unlike full relatives, gerundive
relatives do not circumvent reconstruction effects with WH-movement. I
argue that their exceptional behavior is explained by the structural
consequences of their requirement for temporal dependency.
	Section 8 presents a discussion of extraposition with gerundive 
relatives. In contrast to full relatives, gerundive relatives do not undergo
extraposition. I claim that in extraposed position, gerundive relatives are
not syntactically close enough to the matrix tense to be temporally
dependent on this tense. They are therefore not permitted in extraposed
position.

2.0  Framework
2.1  Semantics of Tense
Within the framework of Reichenbach (1947), tenses are composed of three
times: the Event time, the Reference time, and the Speech time. This system
is illustrated in (2), where the Event time is the time of Mary's leaving,
the Reference time is the time by which Mary's leaving takes place (2:00 in
this sentence), and the Speech time is the time at which the sentence is
uttered.
	
(2)	At 2:00, Mary had left.

	I assume Hornstein's (1990) Reichenbachian approach to tense, according to
which the structures of the basic tenses of English are as in (3), where
the linear order of the Speech, Reference and Event times reflects their
temporal order. If two times are separated by a line, the leftmost time is
interpreted as temporally preceding the other time. If two times are
separated by a comma, they are interpreted as cotemporal.

(3)		S , R, E	present			E _ S , R	present perfect
		E , R _ S	past			E _ R _ S	past perfect
		S _ R , E	future			S _ E _ R	future perfect


2.2  Syntax of Tense
In the spirit of much recent work on the syntax of tense, I assume that
times are the syntactic, as well as the semantic, primitives of tense
(Hornstein 1977, 1981, 1990; Zagona 1988, 1990; Giorgi and Pianesi 1991;
Stowell 1993).
	I claim that times are represented as semantic features of particular
lexical items and I adopt Hornstein's (1990) proposal that the Event time
is associated with VP, whereas the Speech and Reference times are
associated with Inflectional projections. The Event time is a feature
associated with the head of VP, the Reference time is associated with the
head of Asp(ect)P, and the Speech time with the head of T(ense)P, located
above AspP (see Thompson (1996) for further discussion).
	Evidence for the structural hierarchy of Event and Reference times 
comes from the distribution of temporal point adverbials. In clause-final
position, these adverbials may modify the Event time or the Reference time;
(4) may mean that the leaving event takes place at 3 p.m. (Event time
reading), or that the leaving event takes place sometime before 3 p.m.
(Reference time reading).

(4)	Al had left the store at 3 p.m.

However, when a temporal point adverbial is unambiguously associated with
VP, it modifies only the Event time; in the VP fronting construction in
(5), the leaving event takes place at 3 p.m., not sometime before 3 p.m.

(5)	Al claimed he had left the store at 3 p.m., and left the store at 3
p.m. he had.
	
In contrast, when the adverbial occurs above VP, as in (6), with a
clause-initial temporal point adverbial, this adverbial modifies the
Reference time only; the reading is that the leaving takes place sometime
before 3 p.m., and not at 3 p.m.

(6)	At 3 p.m., Al had left the store.

The discussion of the syntax and semantics of gerundive relatives presented
here will provide evidence that the Speech time is located in an
inflectional projection above the Reference time, while the Event time is
located in VP.

3.0  Interpretation of Gerundive Relatives
Enç (1987:645) notes that the temporal interpretation of finite relative
clauses is independent of the tense of the matrix clause (see also Hudson
1973, Ladusaw 1977, Dowty 1982, Abusch 1988 for discussion of the temporal
interpretation of relative clauses). This is illustrated by the example in
(7), where the events of complaining and waiting are both interpreted as
occurring in the past relative to the Speech time, but are temporally
independent of one another; the complaining or the waiting may take place
first, or they may take place at the same time.

(7)	A passenger who was waiting for flight #307 complained to the flight
attendant.

	In contrast to finite relative clauses, gerundive relatives are
interpreted as temporally dependent on the main clause. The event of
waiting in (8) may take place at the time of the event of complaining,
where the meaning is A passenger complained to the flight attendant while
he was waiting for flight #307. The event of waiting may also take place at
the Speech time, where the meaning is A passenger who is now waiting for
flight #307 complained to the flight attendant. However, a temporally
independent reading, for example with the waiting taking place after the
complaining, is not possible.

(8)	A passenger waiting for flight #307 complained to the flight attendant.
	
3.1  Temporal Dependency of Gerundive Relatives
In order to explain the temporal dependency of gerundive relatives, in this
section I analyze their tense structures. Following Hornstein
(1990:115-117), I assume that gerunds have a reduced tense structure with
no Speech time, consisting of only Reference and Event times, as in (9).

(9)	R , E

	Assuming that gerundive relatives lack a Speech time explains why they 
do not permit tense markers, as shown in (10). Since the tense morpheme orders
the Reference time with respect to the Speech time (see Hornstein 1990),
given that there is no Speech time, there can be no tense marker.

(10)	*The passengers were waiting for flight #307 left the room.

	A tense structure may be interpreted by being temporally linked to the
time of the event of  utterance, the Speech time, or by being linked to
another time which is in turn linked to the Speech time. Because the tense
structure of a gerund cannot be anchored to a Speech time within its own
clause, it must be interpreted by being linked to the matrix tense.
	I claim that the two readings of (8) (repeated in (11)) are due to
interpretation with respect to different times of the matrix tense. The
reading of (11) where the waiting is interpreted as occurring at the time
of complaining results from the tense structure of the gerund linking to
the Event time of the main clause. In contrast, the reading where the
waiting occurs at the time of Speech results from linking to the Speech
time of the main clause. From here on, I refer to these readings as the
Event time reading and the Speech time reading, respectively.

(11)A passenger waiting for flight #307 complained to the flight attendant.

(12)a. Event time reading - event of gerund (waiting) is interpreted as
occurring at time of matrix Event (complaining)
    b. Speech time reading - event of gerund (waiting) is interpreted 
as occurring at time of Speech

	The tense structures of the Event and Speech time readings in (12a) and
(12b) are as in (13a) and (13b), respectively. The tense structure in (13a)
represents the reading where the event of waiting occurs at the time of
complaining, with the tense structure of the gerund linked to the Event
time of the matrix clause. (13b) shows thetense structure of the reading
with the waiting occurring at the Speech time, where the tense structure of
the gerund is linked to the matrix Speech time.

(13)a.	E , R _ S		Event time reading
 		 |
		R , E


    b.	E , R _ S		Speech time reading
	        |
	       R , E

	The examples in (14) with temporal adverbs support the claim that
gerundive relatives are temporally dependent on the tense of the main
clause. The waiting of the gerund in (14a) may be interpreted as occurring
in the past, and hence is possible with yesterday, but not in the future,
and therefore is not possible with tomorrow, because the matrix tense is
past. However, the waiting in (14b) may be interpreted as occurring in the
future, but not the past, and hence can appear with tomorrow but not
yesterday, since the matrix tense is future.

(14) a. A passenger waiting (yesterday/*tomorrow) for flight #307 complained
to the flight attendant
     b. A passenger waiting  (*yesterday/tomorrow) for flight #307 will
complain to	the flight attendant

	In this section, I have argued that when a gerundive event is interpreted
as occurring at the time of the matrix event, the reduced tense structure
of the gerundive relative links to the Event time of the matrix clause, and
when the gerundive event is interpreted as occurring at the time of
utterance, it links to the Speech time of the matrix clause.

3.2  Speech Time Readings
I have claimed that the Speech time reading of gerundive relatives is
derived by temporal linking of the tense structure of the gerund to the
Speech time of the main clause. Evidence for this view comes from the
distribution of gerundive relatives in environments in which the Speech
time is interpreted as a time other than the utterance time: historical
present and Sequence of Tense (SOT) constructions.

3.2.1  Historical Present Contexts
As noted by Hornstein (1990:11), the Speech time generally acts as a
deictic element that is interpretively anchored within the speech
situation. However, if a time is made sufficiently salient in the
discourse, it may reorient the value of the Speech time to a time other
than the utterance time. An example of this is given in (15), where the
first sentence reorients the Speech time to the seventeenth century.
	
(15)	Imagine that we are back in the seventeenth century. Erin is planting
potatoes in the 	field and Sean is gathering hay for the barn. One of the
little kids helping Sean is 	bothering the sheepdog.

      The event of helping in the gerundive relative of the last sentence of (15) is interpreted as occurring at the reoriented Speech time of the
seventeenth century, and not at the utterance time of the discourse. This
supports the claim made here that on the reading of a gerundive relative
where the event of the relative is interpretted as occurring at the time of
the utterance, the tense of the gerund is dependent on the matrix Speech
time.

3.2.2  Sequence of Tense Constructions
Sequence of Tense constructions give another example of a tense structure
where the Speech time links to a time other than the utterance time, and
therefore the readings of gerundive relatives with this construction
provide further evidence that the tense structure of gerundive relatives
may link to the Speech time.
	An embedded event in English may be temporally evaluated with respect to
the Speech time (temporally independent reading), or with respect to the
Event time of the subcategorizing verb (temporally dependent, or Sequence
of Tense, reading). The temporally independent reading is illustrated by
(16a), where the event of crying is interpreted as future with respect to
the Speech time. The Sequence of Tense, or SOT, reading is exemplified in
(16b), where the event of crying may be interpreted as past with respect to
the saying event, with the crying taking place before the saying event, or
present with respect to the saying event, where the crying takes place at
the same time as saying.

(16)	a.	Mary said that John will cry.
	b.	Mary said that John was crying.

The temporally dependent reading of (16b) which is relevent for our
purposes is the "present with respect to a past event" reading. In order to
account for this interpretation, Hornstein claims that the embedded clause
has present tense structure, and is only morphologically past tense. The
temporal dependency of SOT readings is accounted for on Hornstein's
analysis by linking the embedded Speech time to the matrix Event time,
yielding (17) for the "present with respect to a past event" reading of
(16b).

(17) 		E , R _ S
		 |
		S , R , E	

	A gerundive relative that occurs within a SOT clause does not show the
ambiguity between an Event time reading and a Speech time reading that it
does in temporally independent clauses. Consider the gerundive relative of
(18a), occuring within a temporally independent embedded clause. Here, the
event of waiting may occur at the Speech time, with the reading, John said
that the passenger who is now waiting for flight #307 will cry. The event
of waiting may also occur at the embedded Event time, with the reading,
John said that the passenger who will be waiting for flight #307 will cry
while he is waiting for the flight. The tense structure of the Speech time
reading is in (18b), where the tense structure of the gerund is linked to
the embedded Speech time, which is temporally independent of the matrix
clause. The structure of the Event time reading, where the waiting takes
place at the time of crying, is in (18c), where the tense structure of the
gerund is linked to the embedded Event time.

(18)	a.	John said that the passenger waiting for flight #307 will cry.
	b.	E , R _ S
		
		S _ R , E
		 |
		R , E
	c.	E , R _ S
	
		S _ R , E
	              |
	            R , E

Since in temporally independent clauses, the Speech time is not linked to
the matrix tense structure, it receives the utterance time value which the
gerund can inherit by linking to this time, deriving the Speech time
reading.
	In contrast, a gerundive relative within a SOT clause cannot receive a
Speech time reading. In (19a), on the reading where the event of crying is
interpreted as occurring at the same time as the event of saying, the
waiting can not be interpreted as occurring at the Speech time: (19a)
cannot mean, John said that (as he was speaking), the passenger who is now
waiting for flight #307 was crying. Instead, the waiting is interpreted as
taking place at the time of saying: (19a) must mean, John said that (as he
was speaking), the passenger who was waiting at that time for flight #307
was crying. The relevant tense structure is provided in (19b), where the
tense structure of the gerund links to the tense structure of the embedded
clause, which is in turn linked to the matrix clause.


(19)	a.	John said that the passenger waiting for flight #307 was crying.
	b.	E , R _ S
		 |
		S , R , E
		 |
		R , E

Since the Speech time is linked to the matrix tense structure in a SOT
construction, as in (19b), it does not receive the default utterance time
value, and hence a gerundive relative within a SOT clause cannot be
interpreted with respect to the utterance time by linking to the Speech
time. Therefore, a gerund within a relative of the SOT clause cannot
receive an utterance time reading. This explanation shows that a gerundive
relative is interpreted with respect to the Speech time on its utterance
time reading; when the Speech time is reoriented to the Event time of the
main clause, the gerund is interpreted with respect to this Event time.
	This analysis of gerundive relatives within SOT clauses is supported by
their contrasting behavior in another type of temporally dependent clause,
temporal adjuncts. In contrast to SOT clauses, a gerundive relative within
a temporal adjunct clause is ambiguous between an Event time and a Speech
time reading; (20) may be interpreted with the waiting occurring at the
time of complaining or at the utterance time.

(20)	John saw Mary before the passenger waiting for flight #307 complained
to the flight 	attendant.

	The contrasting behavior of temporal adjunct and SOT constructions is due
to the different time linking involved in these constructions. As proposed
in Hornstein (1990), temporal adjunct clause constructions involve linking
of the adjunct Speech and Reference times to the matrix Speech and
Reference times. Thus the derived tense structure of (20) is as in (21).

(21)		E , R _ S		Main clause tense
	        |       |
		E , R _ S		Adjunct clause tense
	            |
	            R , E		Gerund tense

	Since the Speech time of the adjunct is reoriented to the Speech time of
the matrix clause, it still receives a default utterance time
interpretation (SOT clauses, on the other hand, reorient the Speech time to
the Event time of the matrix clause). Therefore a gerundive relative within
a temporal adjunct clause can link to the adjunct Speech time and receive
an utterance time reading.
	To summarize this section, although the Speech time is typically
interpreted as the deictic utterance time, when it is reoriented to another
value, as in historical present and SOT constructions, gerundive relatives
are interpreted with respect to this reoriented time, showing that the
utterance time reading results from linking to the Speech time.

4.0  Reduced Clause Structure of Gerundive Relatives
In this section, I examine the clause structure of gerundive relatives. As
has been discussed in the literature on relative clauses, gerundive
relatives seem to have a reduced clause structure when compared to full
relatives (see Williams 1975, Stowell 1982). Given the present analysis of
tense whereby the Speech time is associated with TP, the Reference time
with AspP, and the Event time with VP, since gerundive clauses consist of
only a Reference and an Event time, I propose that they are structurally
AspP.
	
4.1  Missing CP Projection
Gerundive relatives, in contrast to full relatives, do not appear to have a
CP projection. This is shown by the fact that they do not permit overt
complementizers, although full relatives do, as illustrated in the contrast
between the full relative in (22a) and the gerundive relative in (22b).

(22)	a. The passengers that were waiting for flight #307 complained to the
flight attendant.
	b. *The passengers that waiting for flight #307 complained to the flight
attendant.

	In addition, as shown in (23a) and (23b), gerundive relatives contrast
with full relatives in that they do not permit fronted temporal PPs,
although they do permit temporal PPs in final position, as shown in (23c).

(23)	a. The passengers who at 6:00 were waiting for flight #307 complained
to the flight attendant.
	b. *The passengers at 6:00 waiting for flight #307 complained to the
flight attendant.
	c. The passengers waiting for flight #307 at 6:00 complained to the flight
attendant.

Fronted temporal PPs are located in a position above the subject, as is
shown by the example in (24).

(24)	At 6:00, the passengers were waiting for flight #307

The unacceptability of gerundive relatives with overt complementizers and
fronted PPs is predicted given the claim that gerundive relative clauses
are AspPs.

4.2  Missing Inflectional Projections 		
Gerundive relatives also lack certain inflectional projections, as shown by
the fact that they do not occur with sentential adverbs or modal verbs. The
data in (25a) and (25) show that the sentential adverb probably is
permitted with full relatives but not with gerundive relatives (examples
from Williams 1975:251).

(25)	a. The person who was probably playing the music you heard used to be
my roommate.
	b. *The person probably playing the music you heard used to be my roommate.

	When probably occurs between the subject and VP, it is associated with an
inflectional projection, as seen by the fact that it cannot be associated
with VP in VP fronting or pseudoclefting, as shown in (26a) and (26b).
However, it can occur with the inflectional projection in these
constructions, as in (26b) and (26c). (see Travis 1988 for discussion of
the syntax of sentential adverbs).

(26)	a. *John said that Mary will probably play the music and probably play
the music she will.
	b. *Bill decided not to go to the party. What he did was probably sit around and listen to jazz.
	c. John said that Mary will probably play the music and play the music she
probably will.
	d. Bill decided not to go to the party. What he probably did was sit
around and listen to jazz.

	The data discussed here is explained if the inflectional projection that
probably is associated with (which is plausibly Modality Phrase), is absent
in gerundive relatives, since gerundive relatives are composed of only an
AspP.
	As shown by the contrast between (27a) and (27b), gerundive relatives,
unlike full relatives, do not occur with modal verbs, showing that
gerundive relatives lack the relevant inflectional projection. (As in the
examples in (26), this projection is plausibly Modality Phrase).

(27)	a. The passengers who should/could/may/might be waiting for the flight
spoke to the flight attendant
	b. *The passengers should/could/may/might waiting for the flight spoke to
the flight attendant
		
	In this section, we have seen that gerundive relatives consist of a
reduced clause structure; in contrast to full relatives, they do not permit
complementizers, fronted PPs, sentential adverbs, or auxiliary verbs. This
reduced structure is explained by the current analysis of gerundive
relatives; since they include only a Reference and Event time, gerundive
relatives are syntactically AspPs.

5. Syntax of Subject Gerundive Relatives
5.1  Proposal
Recall that gerundive relatives in subject position are temporally
interpreted with respect to the Speech time or the Event time of the matrix
clause. These readings are summarized for (28) in (29a) and (29b). Given
that the Event and Speech time readings are the result of the linking of
different times in tense structure, the syntactic issue is how these
readings are represented in the sentence structure.

(28)	A passenger waiting for flight #307 complained to the flight attendant.

(29)	a. Event time reading - event of gerund (waiting) is interpreted as
occurring at time	of matrix Event (complaining)
	b. Speech time reading - event of gerund (waiting) is interpreted as
occurring at time	of Speech

	I propose that the Event time reading requires the gerundive tense to be
in a syntactically local relation with the Event time reading of the main
clause, and the Speech time reading requires the gerundive tense to be
local with the Speech time of the main clause. This locality requirement is
formulated in (30a), the Condition on Time Linking, which states that in
order for times to link in tense structure, they must be located in the
same checking domain at LF. The definition of checking domain is
illustrated in (30b), where the checking domain of V contains DP1 and H
(Chomsky 1995: 178).

(30)	a.	Condition on Time Linking - In order for time a to link to time b,
a and b must be within the same checking domain at LF.
	b.	     VP
                     /\
		   VP  PP  		   Checking
		   /\ 		    Domain
                DP1  V'
	        /\
	      V  DP2
	     /\
	    H V

Event and Speech time readings of gerundive relatives thus correlate with
the position of the subject at LF. Assuming the VP Internal Subject
Hypothesis, the subject is generated within VP (I assume that this position
is Spec, VP), and moves from this position to Spec, IP (see Zagona 1982,
Kitagawa 1986, Speas 1986, Koopman and Sportiche 1988, 1991). Gerundive
relatives interpreted as temporally dependent on the Event time involve
interpretation of the subject within VP at LF, since this is the position
of the Event time. Gerundive relatives interpreted as temporally dependent
on the Speech time involve interpretation of the subject within TP at LF,
since this is the position of the Speech time.,


(31)	Temporal interpretation		LF interpretation site of subject
	Event time reading			Spec, VP
	Speech time reading			Spec, IP

5.2  Structural Evidence
Initial evidence for this analysis of gerundive relatives is provided in
(32a), with a gerundive relative in subject position of a matrix clause,
and (32b), with a gerundive relative in subject position of an embedded
clause. The gerund in (32a) and (32b) can be interpreted with respect to
the Speech time or the Event time of its immediate clause only. The gerund
in the matrix clause in (32a) can be interpreted with the waiting occurring
at the time of the matrix event of saying and not the embedded event of
complaining. Likewise, the gerund in the embedded clause in (32b) can be
interpreted with the waiting occurring at the time of the embedded event of
complaining, and not the matrix event of saying. This is predicted; since
the subject with the gerundive relative will be within the checking domain
of a time of its own clause, according to the Condition on Time Linking, it
can only be interpreted with respect to one of these times.

(32)	a. The passenger waiting for flight #307 said that Mary will complain
to the flight attendant.
	b. Mary said that the passenger waiting for flight #307 will complain to
the flight attendant.

	To summarize this section, I have argued that the Event time reading of
gerundive relatives in subject position correlates with VP-internal
position of the subject at LF, and that the Speech time reading of
gerundive relatives correlates with Spec, IP position of the subject.
	
6.0  Structural Evidence
In this section, I present evidence for the proposed analysis of the syntax
of gerundive relatives from constructions involving coordination,
existential there, the scope of quantificational and cardinality
adverbials, extraposition, and ECM constructions.
	
6.1  Coordination
Full relatives which are coordinated in subject position may receive
independent temporal interpretations. This is illustrated in (33), where
the first relative is interpreted as past with respect to the Speech time
(the entering takes place before the Speech time), and the second relative
is interpreted as past with respect to the time of the event of going to
the conference (the finishing takes place before the time of going to the
conference, but may be after the Speech time).

(33)	Three students who entered the department in 1993 and who finish their
work will go to the conference next month.

	In contrast to full relatives, when gerundive relatives are coordinated in
subject position, they must be evaluated with respect to the same time. In
(34), the events of waiting and suffering may be both interpreted as
occurring at the Event time, with the reading, Three passengers got off the
boat at the time that they were waiting for the next boat and suffering
from seasickness. The events of waiting and suffering may also be
interpreted with respect to the Speech time, with the meaning, Three
passengers who are now waiting for the next ship and who are now suffering
from seasickness got off the boat. Although it is pragmatically plausible,
it is not possible to interpret the waiting as occurring at the time of
Speech and the suffering as occurring at the time of getting off the boat,
with the meaning, Three passengers who are now waiting for the next ship
got off the boat at the time that they were suffering from seasickness.

(34)	Three passengers waiting for the next ship and suffering from
seasickness got off the	boat.
	
	Since distinct temporal interpretations for coordinated relatives are
possible with full relatives, it does not seem to be due to a semantic
problem that coordinated gerundive relatives cannot be interpreted with
respect to different times. This effect is predicted on the current
analysis of gerundive relatives; since the subject including the conjoined
relatives must be interpreted either in Spec, IP, or in Spec, VP, both
gerunds must be interpreted either with respect to the Speech time or with
respect to the Event time.

6.2  Existential Constructions
Support for the claim that subject gerundive relatives are interpreted
within VP when they relate to the Event time and within IP when they relate
to the Speech time comes from existential constructions. Note that (35a) is
ambiguous; it can have an Event time reading, where the waiting is
interpreted with respect to the time of storming into the room, or it can
have a Speech time reading, where the waiting is interpreted with respect
to the utterance time. However, the existential construction version of
(35a) in (35b) does not permit the Speech time reading; the waiting here is
necessarily interpreted with respect to the event of storming into the
room.

(35)	a. Three passengers waiting for the flight stormed into the room.
	b. There stormed into the room three passengers waiting for the flight.

	I assume, following den Dikken (1995) and Groat (1995), that existential
constructions involve interpretation of the associate of the expletive
within VP. Evidence for the VP-internal position of the associate discussed
by den Dikken (1995) comes from contrasts in reciprocal licensing. In
(36a), the subject some applicants can bind each other in the PP, but
binding is not permitted in the existential version of (36a) in (36b).
Assuming that the reciprocal each other must be c-commanded at LF by its
antecedent, this contrast shows that whereas the subject of (36a)
c-commands the reciprocal, the associate in (36b) does not c-command the
reciprocal at LF. This is explained if the associate is interpreted within
VP.

(36)	a. Some applicants seem to each other to be eligible for the job.
	b. *There seem to each other to be some applicants eligible for the job.
	
	Given that the associate of the expletive is interpreted within VP at LF,
the present analysis correctly predicts that only the Event time reading is
permitted in the existential construction, since when the subject is
located within VP, it is in the checking domain of the Event time.

6.3  Scope of Quantificational Adverbs
Certain quantificational adverbs show a scope ambiguity with respect to the
subject. This is illustrated in (37), which may be interpreted with the
adverb taking scope over the subject, with the meaning It is usually the
case that there are some three passengers or other such that they get
stranded here, or may be interpreted with the subject taking wide scope,
with the meaning there are three particular passengers such that they
usually get stranded here (see Lewis 1975, Kamp 1981, Heim 1982, de Swart
1993 for discussion of quantificational adverbs).

(37)	Three passengers usually get stranded here.

	This scope ambiguity correlates with Event or Speech time readings of
gerundive relatives in subject position. When the gerund receives an Event
time reading, the subject is interpreted as within the scope of the adverb.
On this reading of (38), the meaning is, There are usually some three
passengers or other such that they get stranded here when they are waiting
for flight #307. It is not possible for the subject to take wide scope,
meaning, There are three particular passengers such that they usually get
stranded here when they are waiting for flight #307. However, when the
gerund receives a Speech time reading, the subject is interpreted as
outside the scope of the adverb; here, the meaning of (38) is, There are
three particular passengers who are waiting for flight #307 such that they
usually get stranded here. The reading with wide scope for the adverb is
not possible: There are usually some three passengers or other who are now
waiting for flight #307 such that they get stranded here.

(38)	Three passengers waiting for flight #307 usually get stranded here.

	Given that the quantificational adverb is located between the IP subject
position and VP, since on the present analysis the Event time reading is
linked to VP-internal interpretation of the subject, we correctly predict
that on this interpretation the subject is within the scope of a
quantificational adverb. Conversely, given that the Speech time reading is
linked to IP interpretation of the subject, on this interpretation, the
subject is outside the scope of the quantificational adverb.

6.4  Scope of Cardinality Adverbials
A similar effect to the one observed with quantificational adverbs is seen
with adverbials that specify the cardinality of an event. The example in
(39) illustrates the scope ambiguity of subjects and cardinality
adverbials; the adverb may take scope over the subject, with the meaning
there are five events of three passengers complaining, or the subject may
take scope over the adverb, with the meaning there are three passengers
such that they complained five times, with up to fifteen complaining
events.

(39)	Three passengers complained to the flight attendant five times.

	As expected from the previous discussion of quantificational adverbs, the
narrow scope reading of the subject correlates with the Event time
interpretation, and the wide scope reading correlates with the Speech time
interpretation. In (40), on the reading where the waiting takes place at
the time of complaining, the meaning is: There are five events of three
passengers complaining to the flight attendant when they were waiting for
their flight(s). It is not possible to interpret this sentence as meaning,
There are three passengers such that they complained to the flight
attendant five times when they were waiting for their flight(s). On the
interpretation where the waiting takes place at the time of Speech, the
reading is There are three passengers who are now waiting for their
flight(s) such that they complained to the flight attendant five times. The
following interpretation is not possible: There are five events of three
passengers who are now waiting for their flight(s) complaining to the
flight attendant.

(40)	Three passengers waiting for the flight complained to the flight
attendant five times.

	These facts have the same explanation as the data involving
quantificational adverbs; assuming that the cardinality adverbial is
located between the IP subject position and VP, since the Event time
reading is linked to VP-internal subject interpretation, on this reading,
the subject falls within the scope of the cardinality adverbial. Since the
Speech time reading is linked to IP subject interpretation, on this
reading, the subject takes scope over the cardinality adverbial.

6.5  Extraposition
Extraposition constructions provide further evidence for the proposal
outlined here. Note that (41a), with a gerundive relative and an adjunct PP
within the subject, is ambiguous; it permits either an Event time reading
for the gerundive relative, where the waiting takes place at the time of
entering the room, or a Speech time reading, where the waiting takes place
at the time of Speech. However, (41b), the version of (41a) with
extraposition of the PP, permits only the Event time reading (although it
is slightly marginal); it can only mean that the waiting takes place at the
time of entering the room.

(41)	a. A passenger from California waiting for the announcement entered
the room.
	b. ?A passenger waiting for the announcement entered the room from California.

	Following Culicover and Rochemont (1990), I assume that elements
extraposed from the subject are adjoined to VP. Evidence for this claim
comes from VP adverbial constructions. Constituents extraposed from the
subject can occur before VP adverbials, such as as quickly as possible in
(42) (see Travis 1988 for discussion of the VP-adjunction site of manner
adverbials). Given that extraposed constituents appear before VP-adjoined
material, they must also be adjoined to VP.

(42)	Some women came in (who were) from Chicago as quickly as possible.
	
	I assume that in order for a modifier to modify an element, it must be
adjoined to this element or to the phrase which contains this element.
Therefore, assuming that extraposed elements are VP-adjoined, a subject
that has been extraposed from may be interpreted in Spec, VP; the
extraposed PP is adjoined to the phrase (VP) which contains the subject,
and hence modification is permitted. However, interpreting a subject that
has been extraposed from in Spec, IP is barred, since the extraposed PP is
not adjoined to the subject in Spec, IP or to the phrase (IP) that contains
the subject, and hence modification is not licensed.
	The absence of the Speech time reading with extraposition from a subject
with a gerundive relative is thus explained on the present account; since
extraposition allows only the VP-internal interpretation of the subject,
only the Event time reading is permitted with this construction.

6.6  ECM Constructions
ECM constructions lend support to the claim that the LF position of a
subject with a gerundive relative correlates with its temporal
interpretation. In contrast to the subject of an embedded finite clause,
the subject of an ECM construction permits a gerund to be interpreted with
respect to the matrix Event time. (See section 4.1.3.2 for discussion of
gerundive relatives in the subject of temporally independent versus
dependent embedded clauses.) In (43a), the waiting event of the embedded
finite clause may not be interpreted as occurring at the matrix time of
saying, but in (43b), the waiting event may be interpreted as occurring at
the matrix time of expecting.

(43)	a. John said that the passenger waiting for flight #307 will complain
to the flight attendant. 	
	b. John expected the passenger waiting for flight #307 to complain to the
flight attendant.

	I assume, following Lasnik and Saito's (1991) reformulation of Postal's
(1974) analysis, that ECM subjects raise to Spec, AgrOP of the matrix
clause to check case at LF. Evidence for this claim comes from reciprocal
constructions. In (44a), the defendants cannot bind each other in the
adjunct clause, showing that the subject of an embedded finite clause does
not c-command into a matrix adjunct. However, in (44b), the defendants can
bind each other, showing that the subject of an ECM clause c-commands into
a matrix adjunct. This is predicted if an ECM subject raises at LF to Spec,
AgrOP of the matrix clause.

(44)	a. ?*The DA proved that the defendants were guilty during each other's
trials.
	b. ?The DA proved the defendants to be guilty during each other's trials.

	Assuming Lasnik and Saito's analysis, the fact that ECM subjects can be
interpreted with respect to the matrix Event time is explained by the
proposal presented here; in matrix Spec, AgrOP, the subject is in the
checking domain of the Event time of the matrix clause, and hence can be
interpreted relative to this time.
	To summarize this section, I have presented data involving coordination,
existential there constructions, the scope of quantificational adverbs and
cardinality adverbials, extraposition, and ECM constructions as evidence
for the proposal that a subject with an Event time reading is interpreted
within VP, whereas a subject with a Speech time reading is interpreted
within IP.

7.0  Reconstruction Effects and Gerundive Relatives
It has been noted in the literature that relative clauses contrast with
complement clauses in that they do not show binding-theoretic
reconstruction effects with WH-movement (see van Riemsdijk and Williams
1975, Freidin 1986, Lebeaux 1988 for discussion). In (45a), with the
complement clause that John was asleep, the WH-phrase which claim that John
was asleep behaves as if it is located in object position; he cannot be
coreferent with John. However, in (45b), with the relative clause that John
made, which claim that John made behaves as if it is outside the c-command
domain of the subject; he can corefer with John.

(45)	a. *Which claim that Johni was asleep was hei willing to discuss?
	b. Which claim that Johni made was hei willing to discuss?

	Gerundive relatives, unlike full relatives, do not seem to circumvent
reconstruction effects. This is illustrated in the contrast between (46a),
with a full relative, and (46b), with a gerundive relative; in (46a),
Kennedy and he can corefer, but in (46b), coreference is not permitted. The
gerundive relative, unlike the full relative, behaves as if it
reconstructs.

(46)	a. Which student who was reading Kennedy'si book did hei really like?
	b. *Which student reading Kennedy'si book did hei really like?

	The analysis of gerundive relatives presented here, in combination with
Lebeaux's (1988) proposal for the anti-reconstruction effects of relative
clauses, makes possible an explanation of this contrast between full and
gerundive relatives. Lebeaux (1988) argues that the relative clause of
(45b), unlike the complement clause of (45a), is not present before
WH-movement has taken place, but is adjoined to the WH-phrase by
generalized transformation after the WH-phrase moves to Spec, CP. Since the
relative clause is never in object position, it cannot reconstruct to
object position.
	I claim that although full relatives can be adjoined by generalized
transformation after WH-movement has taken place, this option is not
available for gerundive relatives because they are temporally dependent on
the main clause. If a gerundive relative were to adjoin to the WH-phrase
after the WH-phrase moves to Spec, CP, it would not be within the checking
domain of a time of the main clause, since the times are located in the
inflectional projections and VP. Hence the gerund would not receive a
temporal interpretation. Therefore, gerundive relatives must be present
before movement takes place in order to be interpreted relative to a matrix
time, and thus they show reconstruction effects.

8.0  Extraposition and Gerundive Relatives
Williams (1975) notes that gerundive relatives, unlike full relatives, do
not undergo extraposition, as shown in the contrast between (47a) and
(47b). ((47b) is unacceptable on the relative clause reading of wearing a
fedora.)

(47)	a. A man said hello to me who was wearing a fedora.
	b. *A man said hello to me wearing a fedora.

This contrast is predicted by the present analysis of gerundive relatives.
We have seen that gerundive relatives are interpreted with respect to the
Speech or Event time of the matrix clause by being located within the
checking domain of IP or VP at LF. An extraposed relative, since it is
adjoined, is not within the checking domain of any time. Hence an
extraposed gerundive relative cannot be temporally interpreted.
	If an extraposed gerundive relative were allowed to reconstruct, it should
be able to be interpreted with respect to the time that the subject is
within the checking domain of. However, it seems that extraposition does
not permit reconstruction, as shown in the binding data in (48), from
Guéron (1980:650). Coreference is possible between Mary and her in (48a),
but not in (48b), with extraposition, showing that reconstruction in order
to avoid a Binding Condition C violation is not possible with
extraposition.

(48)	a. A picture of Maryi was sent to heri
	b. *A picture was sent to heri of Maryi

	The same effect is seen with extraposed relative clauses as with PPs, as
seen in the contrast between (49a) and (49b).

(49)	a. A picture that Rembrandti painted was sent to himi
	b. *A picture was sent to himi that Rembrandti painted

9.0  Conclusion
In this paper, I have shown that the structure of tense proposed here is
supported by the interpretation site of subjects at LF. Gerundive relatives
in subject position are temporally dependent on the matrix Event or Speech
time, and since the Event time is represented in VP, assuming that temporal
dependence requires times to be in the same checking domain, a subject
gerundive relative with an Event time reading is interpreted within VP.
Since the Speech time is associated with IP, when the gerundive relative
receives a Speech time reading, the subject is interpreted in Spec, IP.
	Constructions involving coordination, existential there, scope of
quantificational and cardinality adverbials, extraposition, and ECM
constructions show that when the subject is located in Spec, VP, only the
Event time reading is available, whereas when the subject is located in
Spec, IP, only the Speech time reading is available.
	This analysis also explains why, unlike full relatives, gerundive
relatives do not circumvent reconstruction effects with WH-movement. Full
relatives avoid reconstruction effects by adjoining directly to the
WH-element in Spec, CP, but this adjunction is not permitted with gerundive
relatives, since in Spec, CP, they will not be in a local relation with
either the Speech or Event time, and hence cannot receive a temporal
interpretation.
 	The fact that gerundive relatives, in contrast to full relatives, do not
undergo extraposition is also accounted for. An extraposed relative is not
with the checking domain of a matrix time and hence a gerundive relative in
this position cannot be temporally interpreted.



References	



Abusch, D. (1988). "Sequence of Tense, Intensionality and Scope".
Proceedings of WCCFL 7:1-14.

Beghelli, F. and T. Stowell. (forthcoming). The Direction of Quantifier
Movement.

Chomsky, N. (1995). The Minimalist Program. Cambridge, Massachusetts: MIT Press.

Diesing, M. (1989). The Syntactic Roots of Semantic Partition. Doctoral
dissertation, University of Massachusetts at Amherst.

Diesing, M. (1992). Indefinites. Cambridge, Massachusetts: MIT Press.

den Dikken, M. (1995). "Binding, Expletives, and Levels". Linguistic
Inquiry 26:347-353.

Dowty, D. (1982). "Tenses, Time Adverbs, and Compositional Semantic
Theory". Linguistics and Philosophy 5:23-55.

Freidin, R. (1986). "Fundamental Issues in the Theory of Binding", in
Studies in the Acquisition of Anaphora, volume 1, B. Lust, ed., 151-188.
Boston: D. Reidel.

Giorgi, A. and F. Pianesi. (1991). "Toward a Syntax of Temporal
Representations". Probus 3:1-27.

Groat, E. (1995). "English Expletives: A Minimalist Account". Linguistic
Inquiry 26:354-365.

Guéron, J. (1980). "On the Syntax and Semantics of PP Extraposition".
Linguistic Inquiry 11:637-676.

Guéron, J. and R. May (1984). "Extraposition and Logical Form". Linguistic
Inquiry 15:1-31.

Heim, I. (1982). The Semantics of Definite and Indefinite Noun Phrases.
Doctoral dissertation, University of Massachusetts at Amherst.

Hornstein, N. (1977). "Towards a Theory of Tense". Linguistic Inquiry 8:521-557.

Hornstein, N. (1981). "The Study of Meaning in Natural Language: Three
Approaches to Tense", in Explanation in Linguistics, N. Hornstein and D.W.
Lightfoot, eds., 116-151. London and New York: Longman.

Hornstein, N. (1990). As Time Goes By. Cambridge, Massachusetts: MIT Press.

Hornstein, N. (1994). "An Argument for Minimalism". Linguistic Inquiry
25:455-480.

Hornstein, N. (1995). Logical Form: From GB to Minimalism. Oxford: Blackwell.

Huang, C.-T. J. (1993). "Reconstruction and the Structure of VP: Some
Theoretical Consequences". Linguistic Inquiry 24:103-138.

Hudson, R.A. (1973). "Tense and Time Reference in Reduced Relative
Clauses". Linguistic Inquiry 4:251-256.

Kamp, H. (1981). "A Theory of Truth and Semantic Representation", in Formal
Methods in The Study of Language, J. Groenendijk, T. Janssen and M.
Stokhof, eds., 1-41. Amsterdam: Mathematisch Centrum.

Kitagawa, Y. (1986). "Subjects in English and Japanese", Doctoral
dissertation, University of Massachusetts.

Koopman, H. and D. Sportiche. (1988). "Subjects". Ms., University of
California at Los Angeles.

Koopman, H. and D. Sportiche (1991). "The Position of Subjects". Lingua
85:211-259.

Ladusaw, W. (1977). "Some Problems with Tense in PTQ", in Texas Linguistic
Forum 6, University of Texas, Austin.

Lasnik, H. and Saito, M. (1991). "The Subject of Infinitives". Paper
presented at Chicago Linguistic Society, Ms., University of Connecticut at
Storrs.

Lebeaux, D. (1988). Language Acquisition and the Form of the Grammar.
Doctoral dissertation, University of Massachusetts at Amherst.

Lewis, D. (1975). "Adverbs of Quantification", in Formal Semantics of
Natural Language, E. Keenan, ed., 3-15. Cambridge: Cambridge University
Press.

Musan, R. (1995). On the Temporal Interpretation of Noun Phrases, Doctoral
dissertation, MIT.

Nunes, J. (1994). "The Discourse Representation of Tense Sequencing in
Narratives". MIT Working Papers in Linguistics.

Postal, P. (1974). On Raising: One Rule of English Grammar and It's
Theoretical Implications. Cambridge, Massachusetts: MIT Press.

Raposo, E. and J. Uriagereka. (1995). "Two Types of Small Clauses (Toward a
Syntax of Theme/Rheme Relations)", in Small Clauses, A. Cardinaletti and M.
T. Guasti, eds., 179-206. San Diego, California: Academic Press.

Reichenbach, H. (1947). Elements of Symbolic Logic. New York: The Macmillan
Company.

van Riemsdijk, H. and E. Williams. (1981). "NP-Structure". The Linguistic
Review 1:171-217.

Speas, M. (1986). "Adjunction and Projections in Syntax". Doctoral
dissertation, MIT.

Stowell, T. (1982). "The Tense of Infinitives". Linguistic Inquiry 13:561-570.

Stowell, T. (1993). "The Syntax of Tense". Ms., UCLA.

de Swart, H. (1993). Adverbs of Quantification: A Generalized Quantifier
Approach. New York: Garland Publishing, Inc.

Travis, L. (1988). "The Syntax of Adverbs". Paper presented at NELS 19,
Ithaca, New York.

Williams, E. (1975). "Small Clauses in English", in Syntax and Semantics,
volume 4, J. Kimball, ed., 249-273. New York: Academic Press.

Zagona, K. (1982). Government and Proper Government of Verbal Projections.
Doctoral dissertation, University of Washington at Seattle.

Zagona, K. (1988). Verb Phrase Syntax. Dordrecht: Kluwer.





References	



Abusch, D. (1988). "Sequence of Tense, Intensionality and Scope".
Proceedings of WCCFL 7:1-14.

Beghelli, F. and T. Stowell. (forthcoming). The Direction of Quantifier
Movement.

Chomsky, N. (1995). The Minimalist Program. Cambridge, Massachusetts: MIT Press.

Diesing, M. (1989). The Syntactic Roots of Semantic Partition. Doctoral
dissertation, University of Massachusetts at Amherst.

Diesing, M. (1992). Indefinites. Cambridge, Massachusetts: MIT Press.

den Dikken, M. (1995). "Binding, Expletives, and Levels". Linguistic
Inquiry 26:347-353.

Dowty, D. (1982). "Tenses, Time Adverbs, and Compositional Semantic
Theory". Linguistics and Philosophy 5:23-55.

Freidin, R. (1986). "Fundamental Issues in the Theory of Binding", in
Studies in the Acquisition of Anaphora, volume 1, B. Lust, ed., 151-188.
Boston: D. Reidel.

Giorgi, A. and F. Pianesi. (1991). "Toward a Syntax of Temporal
Representations". Probus 3:1-27.

Groat, E. (1995). "English Expletives: A Minimalist Account". Linguistic
Inquiry 26:354-365.

Guéron, J. (1980). "On the Syntax and Semantics of PP Extraposition".
Linguistic Inquiry 11:637-676.

Guéron, J. and R. May (1984). "Extraposition and Logical Form". Linguistic
Inquiry 15:1-31.

Heim, I. (1982). The Semantics of Definite and Indefinite Noun Phrases.
Doctoral dissertation, University of Massachusetts at Amherst.

Hornstein, N. (1977). "Towards a Theory of Tense". Linguistic Inquiry 8:521-557.

Hornstein, N. (1981). "The Study of Meaning in Natural Language: Three
Approaches to Tense", in Explanation in Linguistics, N. Hornstein and D.W.
Lightfoot, eds., 116-151. London and New York: Longman.

Hornstein, N. (1990). As Time Goes By. Cambridge, Massachusetts: MIT Press.

Hornstein, N. (1994). "An Argument for Minimalism". Linguistic Inquiry
25:455-480.

Hornstein, N. (1995). Logical Form: From GB to Minimalism. Oxford: Blackwell.

Huang, C.-T. J. (1993). "Reconstruction and the Structure of VP: Some
Theoretical Consequences". Linguistic Inquiry 24:103-138.

Hudson, R.A. (1973). "Tense and Time Reference in Reduced Relative
Clauses". Linguistic Inquiry 4:251-256.

Kamp, H. (1981). "A Theory of Truth and Semantic Representation", in Formal
Methods in The Study of Language, J. Groenendijk, T. Janssen and M.
Stokhof, eds., 1-41. Amsterdam: Mathematisch Centrum.

Kitagawa, Y. (1986). "Subjects in English and Japanese", Doctoral
dissertation, University of Massachusetts.

Koopman, H. and D. Sportiche. (1988). "Subjects". Ms., University of
California at Los Angeles.

Koopman, H. and D. Sportiche (1991). "The Position of Subjects". Lingua
85:211-259.

Ladusaw, W. (1977). "Some Problems with Tense in PTQ", in Texas Linguistic
Forum 6, University of Texas, Austin.

Lasnik, H. and Saito, M. (1991). "The Subject of Infinitives". Paper
presented at Chicago Linguistic Society, Ms., University of Connecticut at
Storrs.

Lebeaux, D. (1988). Language Acquisition and the Form of the Grammar.
Doctoral dissertation, University of Massachusetts at Amherst.

Lewis, D. (1975). "Adverbs of Quantification", in Formal Semantics of
Natural Language, E. Keenan, ed., 3-15. Cambridge: Cambridge University
Press.

Musan, R. (1995). On the Temporal Interpretation of Noun Phrases, Doctoral
dissertation, MIT.

Nunes, J. (1994). "The Discourse Representation of Tense Sequencing in
Narratives". MIT Working Papers in Linguistics.

Postal, P. (1974). On Raising: One Rule of English Grammar and It's
Theoretical Implications. Cambridge, Massachusetts: MIT Press.

Raposo, E. and J. Uriagereka. (1995). "Two Types of Small Clauses (Toward a
Syntax of Theme/Rheme Relations)", in Small Clauses, A. Cardinaletti and M.
T. Guasti, eds., 179-206. San Diego, California: Academic Press.

Reichenbach, H. (1947). Elements of Symbolic Logic. New York: The Macmillan
Company.

van Riemsdijk, H. and E. Williams. (1981). "NP-Structure". The Linguistic
Review 1:171-217.

Speas, M. (1986). "Adjunction and Projections in Syntax". Doctoral
dissertation, MIT.

Stowell, T. (1982). "The Tense of Infinitives". Linguistic Inquiry 13:561-570.

Stowell, T. (1993). "The Syntax of Tense". Ms., UCLA.

de Swart, H. (1993). Adverbs of Quantification: A Generalized Quantifier
Approach. New York: Garland Publishing, Inc.

Travis, L. (1988). "The Syntax of Adverbs". Paper presented at NELS 19,
Ithaca, New York.

Williams, E. (1975). "Small Clauses in English", in Syntax and Semantics,
volume 4, J. Kimball, ed., 249-273. New York: Academic Press.

Zagona, K. (1982). Government and Proper Government of Verbal Projections.
Doctoral dissertation, University of Washington at Seattle.

Zagona, K. (1988). Verb Phrase Syntax. Dordrecht: Kluwer.