University of Pennsylvania Working

Papers in Linguistics
Volume 5
Article 6
Issue 2 Current Work in Linguistics

1-1-1998

The Conflict Between Future Tense and Modality:

The Case of Will in English
Anoop Sarkar

This paper is posted at ScholarlyCommons. http://repository.upenn.edu/pwpl/vol5/iss2/6

For more information, please contact libraryrepository@pobox.upenn.edu.
The Conflict Between Future Tense and Modality: The Case of Will in
English

This working paper is available in University of Pennsylvania Working Papers in Linguistics: http://repository.upenn.edu/pwpl/vol5/
iss2/6
 The Conflict Between Future Tense and Modality:
The Case of Will in English

Anoop Sarkar
There have been differing views in the literature on what the semantics of will
should constitute. Some consider will to be homonymous between a modal and
a periphrastic future tense, while some deny that it is a future tense, indicating
that its futurity is derived from its modality. This paper reviews the evidence
for both views and draws a conclusion based on an empirical comparison.

1. Introduction

The debate about a semantics for will can be summarized as follows:

Can the use of will in sentences like He will speak tomorrow be part
of the morphological tense-system of English, i.e. is will speak the
periphrastic future tense of speak just as speaks is the present tense
and spoke is the past tense. Or should sentences like He will have left
already (epistemic will) be taken as evidence that will is part of the
modal system, parallel to He must/can/may/ speak. Is will part of
the tense system or the modal system or is it simply homonymous.

In this paper we will review the evidence for both sides and try to motivate the
need for a unified semantics for will. We ask the following two questions and
then evaluate various extant analyses of will to see if they can account for the
data.
Can the future be empirically shown to be different from the past?
Is the future distinguishable from modality?

2. The Problem

Before embarking on a search for a semantics for will, we must motivate the
desire to show that the different senses of will should in fact be unified. As
Kratzer (1977) points out, nobody would claim that a semantic description of
the word will should try to capture whatever is common to the meaning of the
two instances of the word will in (1).
(1) I will read your will at your death-bed.

U. Penn Working Papers in Linguistics, Volume 5.2, 1998

92 ANOOP SARKAR

This is an instance where the two wills are taken to be occurrences of two
distinct words which just happen to look the same. Now consider the sen-
tences in (2) with the interpretations given in parentheses with each sentence.
The different interpretations of will in these sentences could be attributed to
different wills: will , will , will .
(2) a. It will rain for hours in Stockport. (generic statement about a place)
(Haegeman 1983)
b. John will have left already. (epistemic)
c. John will leave right now. (directive) (Hornstein 1990)
However, by analogy to the argument presented in Kratzer (1977) for
modals such as must, we can shift the meaning of will in (2) to the single
temporal meaning of will in (3) by changing the context (given in emphasis).
A theory that maintains ambiguous wills must also have an additional neutral
sense of will for the sentences in (2). However, this neutral meaning of will
(embedded in an appropriate theory) is what we need to unify the semantic
description of will and to account for the sentences in both (2) and (3). This is
why seeking a unified semantics for will is a worthwhile goal.
(3) a. It will rain for hours in Stockport tomorrow night due to a low pres-
sure system moving into the area. (future)
b. John will have left by eight o’clock tomorrow night. (future perfect)
c. John will leave tomorrow. (future) (Hornstein 1990)

3. Tense and Modality

Before we look at the particulars of will we must clarify some notational is-
sues. In the literature, the term tense or tense system are sometimes used to
denote completely different things.

1. In one sense, the term tense is used to indicate the tense morphology of
a language which refers (although not exclusively) to the temporal repre-
sentation. Usually this notion is used to denote the grammatical category
realized by the inflectional element I(NFL). We shall refer to this sense
as tense or tense morphology in this paper.
However this meaning is not the only one adopted in the literature. For instance,
in Hornstein (1990), tenses, modals, the perfect auxiliary and temporal modifiers are
assigned the same temporal structures: Reichenbachian tense diagrams. He also as-
sumes there is a one to one mapping between the tense morphology and his temporal
structures.
 FUTURE TENSE AND MODALITY 93

2. In the other sense tense system is taken to mean the mechanisms of tem-
poral interpretation common to all natural languages, for instance, a Re-
ichenbachian tense diagram (Reichenbach 1947) or a Priorian past tense
operator (Prior 1967). In this paper this sense is referred to as temporal
interpretation.
These two definitions correspond to the grammatical (syntactic) tense dis-
tinction as opposed to the notional divisions of time represented in a natural
language (Jespersen 1924:255). The distinction is important because an ap-
proach that argues that will is not formally a tense morpheme does not preclude
the notion of future temporal interpretation, however a theory could also deny
future temporal interpretation in natural language altogether. In this paper, we
will attempt to choose between these two theories that have been proposed
to account for future time reference shown schematically in Figures 1 and 2
(where, S represents the speech or utterance time, bold lines represent events
and E represents a future event distinguished by the utterance).

world 1

world 2
S

E
world 3

time

Figure 1: Possible worlds model

The two models are:

futurity as a corollary of modality in a Kripke model of possible worlds
(actually entire histories of possible worlds) as pictured in Figure 1. This
theory is enriched by a notion of context dependence (Kratzer 1977, 1991).
In this model, will is purely modal with the future being epiphenomenal.
the modal-temporal model, where both future time reference and modality
coexist as pictured in Figure 2. An argument for the notion of future
temporal interpretation also has ramifications for ontological commitment
in a theory of temporal knowledge representation, see Steedman (1996).
In this model, will is ambiguous between receiving a modal and a future
temporal interpretation.
94 ANOOP SARKAR

world 3
world 1

world 2 world 4

time E

Figure 2: The modal-temporal model

Most of the discussion in the literature could be characterized as taking

one of these two options in order to adequately describe the semantics of will.
Since several efforts at categorizing the nature of will are reviewed and
compared here, and since many of these efforts do not share a common the-
oretical background, we have tried to remain theory-neutral throughout the
following discussion.

4. The Data

As mentioned before, will is not uniquely used to refer to future time. will is
also commonly used as a modal with reference to present or past time.
futurity
(4) a. Tomorrow morning I will wake up in this first-class hotel suite.
b. He will go to London tomorrow. (Boyd and Thorne 1969)
c. I’ll be 21 next week. (Haegeman 1983)
d. Between 6 and 7 I’ll be having my bath. (duration) (Haegeman
1983)
e. Well, I’ll ring you tonight sometime. (volition) (Palmer 1986)
f. I will do it. (volition) (Haegeman 1983)
g. The queen will now hand the trophy to the captain. (immediate fu-
ture) (Haegeman 1983)

epistemic modality
(5) a. That will be the milkman.
b. Tell him Professor Cressage is involved – he will know Professor
Cressage. (Palmer 1979)
 FUTURE TENSE AND MODALITY 95

c. In the 1920s Wilkinson Sword introduced the stroppable razor and

then the ‘Empire’ range which many people will remember. (Palmer
1979)
d. He will have read it yesterday. (Huddleston 1995)

dynamic modality
(6) a. John will get angry over nothing.
b. John will work one day and loaf the next.
c. Ed will lie in bed all day, reading trashy novels. (Huddleston 1995)

capability/generic
(7) a. Nitric acid will dissolve zinc. (Boyd and Thorne 1969)
b. Oil will float on water. (Haegeman 1983)
c. Accidents will happen. (Elvis Costello)
d. The French will be on holiday today. (Palmer 1979)
e. In 20 years, cockroaches will prey on humans.
f. According to predictions, typhoons will arise in this part of the Pa-
cific.
directives
(8) a. You will do as I say, at once.
b. Will you please stop that racket?
c. You will report back for duty on Friday morning. (Huddleston 1995)
Of course, these examples do not exhaustively cover the various modal-
ities that will can participate in . Both Palmer (1979) and Haegeman (1983)
attempt to give a more exhaustive list of contexts in which will can be used.
For the purposes of this paper, we will simply try to distinguish the modal uses
of will from its temporal use to refer to future time.

5. Various Analyses of Will

There have been several first approximations towards giving a unified seman-
tics for will (and such morphemes). Comrie (1989) refutes the conclusions
As the examples show, modal uses of will like other modals are sensitive to context.
Kratzer (1977, 1991), formalizes these contextual dependencies. We will return to this
subject later when we discuss the temporal contribution of modals.
We’ll make a simplifying assumption and not look at shall in this paper. With
respect to our study, shall does not differ from will in any significant way. shall has a
sense of obligation in contrast with will, however it shows a similar ambiguity between
a future and a modal present time interpretation (see Palmer 1979:62).
96 ANOOP SARKAR

drawn by such theories and we give a brief summarization of the arguments

below:
1. Since the future is inherently less certain, future time reference is dif-
ferent from present or past time reference, and is thus inevitably modal
(Jespersen 1931, Lyons 1977, Yavaş 1982, Palmer 1979, 1986). However,
statements about next week or tomorrow in examples like (9a) have more
in common with past time reference in (9b) than the modality in (9c).
Comrie (1989) calls this the “conceptual non-argument”.

(9) a. #It will rain tomorrow but it won’t rain at all.

b. #It had rained yesterday but it didn’t rain at all.
c. It must rain tomorrow but it might not rain at all.

It should be noted that this leaves open the possibility that a particular
language might subsume future time reference under modality, thus mak-
ing overt the lower degree of certainty usually associated with statements
about the future. This is the case in Burmese (Comrie 1985:50–51) where
declarative sentences take either realis or irrealis particles. The irrealis
particle subsumes possibility in the non-future, but also all future-time
reference.
2. In terms of their syntactic distribution, the tokens will and shall are auxil-
iaries that distribute exactly with the modal auxiliaries. Hence they should
have identical semantics (Palmer 1979, Coates 1983, Perkins 1983, Quirk
et al. 1985). However this view is not tenable. Consider the determiner
a which distributes identically with the, many, etc. Non-specific uses of
a however are not considered to be quantifiers like other determiners in
some semantic theories (e.g. those in the DRT vein). Comrie (1989) also
argues against this view and calls it a “formal non-argument” and shows
that if considered cross-linguistically such a premise is neither a necessary
nor a sufficient condition on a description of the data.
3. Since most natural languages seem to have future tense morphology am-
biguous with respect to future temporal interpretation and modality this
is an indication of a language universal. First of all, as Comrie (1989)
points out, according to Haiman’s (1980) grammar of Hua, verb inflec-
tions in Hua are solely used for future time reference. They are not used
to express modal values with present or past time reference. Even if all
languages had ambiguous future tense morphology, this does not rule out
true ambiguity as a result of the morphological resources available to a
language.
 FUTURE TENSE AND MODALITY 97

With the inconclusive approaches out of the way, the remaining sections
lay out some evidence that will help us choose an analysis for will.

6. Future Past

In this section we look at empirical evidence that attempts to show that future
time reference is different from past time reference.

6.1. Present/Future Ambiguity

Zagona (1989) points out that while ambiguity between present or future in-
terpretations of an event is always possible, such a “shifting” between past and
present is not. The following examples are taken from Zagona (1989).
As the comparison in (10) and (11) shows, unlike morphologically present
sentences, morphologically past sentences cannot be construed as contempo-
raneous with the utterance time (the now in (11) crucially has to refer to utter-
ance time for ungrammaticality) or to some future time.
(10) a. *John sang now/tomorrow.
b. John is singing now/tomorrow.
(11) a. *John was singing now/tomorrow.
b. John sings now/tomorrow.
Also, non-past tense morphology does not admit a past adverbial as in (12).
But, by contrast, non-past tense morphology can take future interpretation as
in (13). Zagona (1989) also cites a similar ambiguity between past and non-
past morphology in Spanish.
(12) a. *Plácido sings yesterday.
b. *Plácido is singing yesterday.
c. *Plácido will be singing yesterday.
(13) a. Juan sings tomorrow.
b. Juan is singing tomorrow.
Thus, past and non-past morphology do not behave alike when it comes to
temporal modification.

6.2. The Perfect

In a Reichenbachian system, the past perfect is associated with the tense dia-
gram E-R-S, where data such as (14) is explained by the fact that neither the
E point nor the R point can be associated with the S point (Hornstein 1990).
98 ANOOP SARKAR

(14) a. John had eaten the cake yesterday. (Hornstein 1990)

b. *John had left right now/at this very moment.
c. *John had left tomorrow.
The future perfect is given an analogous tense diagram: S-E-R. However,
we can give the future perfect an interpretation where the E point can precede
the S point (without invoking any modal behavior) as in (15). In order to an-
alyze these cases, Hornstein (1990) gives the structure (S-R) o (E-R) to
the future perfect and along with the future in past, (R-S) o (R-E), it is
the only tense diagram that does not compose to give a well-formed Reichen-
bachian tense diagram like the other tense diagrams do .
(15) John will have finished his manuscript by tomorrow. (see 7.4)

6.3. Sequence of Tense

Consider sentence (16): this sentence has two distinct readings, a “shifted
reading” (Enç 1987) in which John hears at a past time that Mary was pregnant
at a time previous to that. It also has a “simultaneous reading” (the so-called
sequence of tense reading) in which John hears at a past time that Mary is preg-
nant at the time of hearing. These two readings occur in languages like English
which is a strict sequence of tense language (Steedman 1996). See Enç (1987),
Abusch (1988), Hornstein (1990), Ogihara (1995) for further discussion about
this phenomenon.
(16) John heard that Mary was pregnant.
However, simultaneous readings are only available with stative complements
as in (17) but not with eventive complements as in (18). The examples are
from Enç (1987).
(17) a. You knew that I was upset about the results.
b. I heard that Sally was in London.
(18) a. John heard that Mary failed the test.
b. The gardener said that the roses died.
c. Sally thought that John drank the beer.
Also, the present perfect combines happily with stage level predicates but the fu-
ture perfect does not:
(i) #He will have been available. (future perfect reading)
(ii) He has been available.
Further discussion on the future perfect is given in 7.4.
 FUTURE TENSE AND MODALITY 99

In the literature, the issue of whether sentences with will undergo se-
quence of tense phenomena has been debated. (Hornstein 1990) argues that
will does participate in sequence of tense, i.e. in (19) the complement clause
is predicted to be co-temporal with the main clause.
(19) Mary will say that she will be tired. (Enç 1996)
However, as Enç (1996) points out, Hornstein’s (1990) theory also must
predict that the sequence of tense rule is optional and hence in a sentence such
as (20) there must exist a reading in which the S point of the embedded clause
is left free, and by default is identified at the utterance time. On this reading,
the time of thinking and the time of being pregnant are both claimed to be after
the utterance time, but they are not ordered with respect to each other. Among
the non-ordered readings, the reading where the pregnancy precedes John’s
thinking does not seem to be available.
(20) John will think that Mary will be pregnant. (Hornstein 1990)

6.4. Aspect

We will consider four tests used by Vendler: (a) compatibility with adverbials
like for 15 minutes, (b) in 15 minutes, (c) the entailment arising from the pro-
gressive and (d) compatibility with the perfect (see Steedman 1996:6). We will
only consider a category of events like walking, climbing and writing which
Vendler called activities (as opposed to events which are achievements like
arriving, reaching the top or fishing or events that are accomplishments like
writing a sonnet or flying to Paris). Activities are extended in time and when
the tense morphology is past, they can combine with for-adverbials but not
with in-adverbials, that the progressive does carry a factive entailment, and
that they are odd with the perfect. The following examples are from Steed-
man (1996:6).
(21) a. Keats wrote for 15 minutes.
b. #Keats wrote in 15 minutes.
c. Keats is writing. ( Keats will have written)
d. #Keats has written.
Now consider activities with will instead of the past tense morphology.
(22) a. Keats will write for 15 minutes.
b. Keats will write in 15 minutes.
c. Keats will be writing. ( Keats will have written)
d. Keats will have written.
100 ANOOP SARKAR

The examples in (22) show that there are some aspectual difference in future
time reference and past time reference.

6.5. Conclusion

There is empirical evidence to believe that the future can be distinguished in

terms of temporal interpretation from the past.

7. The Future and Modality

In this section we look at the various arguments presented in the literature that
attempts to show that the will used for futurity can be empirically shown to be
different from the will of modality.

7.1. Passivisation

In Wekker (1976) and Davidson-Nielsen (1988) the effect of passivisation on

will-sentences is taken to be a formal device that shows the distinction between
will as a future tense auxiliary and the volitional or a modal will. The argument
is as follows: the sentences in (23) (from Haegeman 1983) are synonymous in
their future reading, however in their volitional reading, (23a) means that John
is the unwilling party, while in (23b) it is Mary who is unwilling.
(23) a. John won’t meet Mary.
b. Mary won’t be met by John.
This distinction in meaning was surprising in the context of the relation
between active and passive sentences in earlier transformational grammar. Of
course, the fact that future will is voice-neutral and volitional will isn’t cannot
be a litmus test for tense and modality because:
epistemic will is voice-neutral as shown in (24).
(24) a. John will have finished the job yesterday. (Haegeman 1983)
b. =The job will have been finished by John yesterday.
and the voice-neutral future will in (25) patterns like the past tense mor-
phology and the modal auxiliary may.
(25) a. The rain delayed/may delay/will delay the start. (Huddleston 1995)
b. =The start was delayed/may be delayed/will be delayed by the rain.
 FUTURE TENSE AND MODALITY 101

7.2. Conditionals

Wekker (1976), Davidson-Nielsen (1988) and Declerck and Depraetere (1995)

distinguish the future tense auxiliary from the modal auxiliary by reference to
the non-appearance of the future tense will in the antecedents of conditionals.
This is simply not true, as evidenced by the sentences in (26).
(26) a. And I will greatly appreciate it if you will not tell your husband.
(Brown corpus cn19)
b. “And I am not sure that I have any cash – any money , that is – but
if you will wait just a minute I will write you out a check if I can
find my checkbook. Won’t you step into the living room , where it’s
cozier”? (Brown corpus ck22)
c. “I’ll have a drink , then , if you’ll have one with me”.
“If you will promise to make it weak”. (Brown corpus ck22)

7.3. Adverbial Modifiers and Free Choice Any

Hornstein (1990) argues that the future tense will in English can be easily
distinguished from the modal will by some simple empirical tests.
The first involves the modification of the present-tense adverb such as
now. The claim is in (30) (from Hornstein 1990:33) the modal will, but not the
future tense will as in (29a), is modifiable by a present-tense modifier.
(27) a. That will be Max at the door now.
Accepting that this is true, let us take a case of a future tense will such as
(29a). Compare it with the sentence in (29b). It seems that analogous to the
argument given for will, must too is ambiguous between a sense compatible
only with future-time modifiers like tomorrow and a sense compatible with
a present-tense modifier like now. In fact, all the modals in (30a) seem to
participate in this ambiguity. But Hornstein (1990) goes on to make crucial
distinctions between modals like must which is assumed not to be ambiguous
and will which is (see 7.4). Hence, sentences in (30) do not solve the problem
of deciding whether will is ambiguous.
(28) a. George will leave now.
b. Suzie will go to sleep now.
(29) a. Tomorrow, John will leave for Paris in a week.
b. Tomorrow, John must leave for Paris in a week.
(30) a. John could/should/might/may/can/must go to school now/tomorrow/
*yesterday.
102 ANOOP SARKAR

b. Go to school now/tomorrow/ *yesterday.

Hornstein (1990:202n38) also claims that free choice (FC) any can be
used to detect modal uses of will from the temporal future sense. The contrast
is given in (31) (from Hornstein 1990:202).
(31) a. Leave this instant on any available flight.
b. John will leave this very instant on any available flight.
c. *John left yesterday on any available flight.
d. You will leave tomorrow on any flight. (directive)
e. ??I simply believe that you will leave tomorrow on any flight.
The argument is that (31e) is odd because of a lack of directive force
which is seen in (31d). Consider (32a) which has will in a complement clause
where it is impossible to get imperatives in English. The use of FC any seems
to be grammatical. Also the emphasized segment of the discourse in (32b)
seems to me difficult to reconcile with an imperative use of will, but it clearly
has a FC any under its scope.
(32) a. John told/assured me that Mary will catch any available flight to-
morrow in order to reach the meeting on time.
b. “Information is hereby given that Mr. Timothy Palmer of Newbury-
port, Mass. has agreed to take charge of the concerns of the Paten-
tees of the Chain Bridge, in the states of Massachusetts, New Hamp-
shire, Vermont, Rhode Island, and Connecticut, so far as relates to
the sale of Patent rights and the construction of Chain Bridges
“Mr. Palmer will attend to any applications relating to bridges and
if desired will view the proposed site , and lay out and superintend
the work, or recommend a suitable person to execute it.
John Templeman. “Approved, Timothy Palmer”.
(Brown corpus ce18)

7.4. The Future Perfect

Hornstein (1990:38) cites the future perfect as evidence to support the view
that the future tense acts quite differently from the modal will. The assump-
tion is that in English the two senses of will have the Reichenbachian tense
diagrams given in (33).
(33) a. future perfect will have := S-E-R
b. modal + have will have := E,R-S
 FUTURE TENSE AND MODALITY 103

The tense diagram for (33b) is that of the simple past. This tense diagram
is shared with all other modals such as must, etc. but crucially Hornstein
(1990) gives evidence to show that other modal verbs with have do not get
the tense diagram for the future perfect in (33a). The reason is the contrast
between the examples of modal verbs with have in (34) and the sentences with
will have in (35) (taken from Hornstein 1990:39).
(34) a. John must have eaten at 3 pm.
b. John should have eaten at 3 pm.
c. At 3 pm, John must have eaten.
d. At 3 pm, John should have eaten.
(35) a. John will have left the office at 3 pm.
b. At 3 pm, John will have left the office.
The sentences in (35) can be interpreted with John’s leaving occuring ei-
ther at or before 3 pm (i.e. either modifying the E or the R point, Hornstein
(1990) shows that sentence initial time adverbials tend to modify the R point
easier than the E point). The sentences in (34) show no such ambiguity. This is
striking evidence since this tries to show that will have has a R point which can
be modified (which is distinct from the E point), and existence of a temporal R
point is strong evidence that we are dealing with a tense interpretation of will
rather than a modal one.
To verify this evidence, we should test whether the particular choice of
modal has anything to do with the judgments seen in (34). Consider the ex-
amples in (36) and (37). They seem to allow modification of the R point more
readily than the examples in (34).
(36) a. John might have eaten his lunch at 3 pm.
b. John may have eaten lunch at 3 pm.
c. At 3 pm, John might have eaten his lunch.
d. At 3 pm, John may have eaten his lunch.
(37) a. The train must have left by now.
b. The train may have left by now.
c. The train will have left by now.
Also, as seen in the sentences in (38) (if my judgments are correct) the
modification of the R and the E point in will have sentences is not as robust as
in the cases with will.
(38) a. Now we’ll be broke at the end of the month.
b. ??At the end of the month, we will have been broke on the 15th.
c. Tomorrow, John will leave for Paris in week. (Hornstein 1990)
104 ANOOP SARKAR

d. ??Day after tomorrow, John will have left for Paris at 3 pm tomor-
row.
If Hornstein’s (1990) story about modal have and will have sentences is
not the right one, what could account for the data presented in this section.
Consider the simplest answer: the underlying tense in all the modal have sen-
tences including the will have sentences is the present perfect E-S,R. The
range of adverbs that the present perfect can take in English is limited. As
Hornstein (1990) points out it is odd to say “John has left yesterday”.
If this simplistic analysis can be justified, then all modals always have
present tense which combines with the perfect auxiliary have to give us the
present perfect. The tense diagram of the present perfect allows us to explain
the objection in Comrie (1985:71) (see Hornstein 1990:200n15) that (39) has
an interpretation where E is prior to S. In order to analyze these cases, Horn-
stein (1990) gives the structure (S-R) o (E-R) to the future perfect and
along with the future in past, (R-S) o (R-E), it is the only tense diagram
that does not compose to give a well-formed Reichenbachian tense diagram
like the other tense diagrams do.
(39) John will have finished his manuscript by tomorrow. (Hornstein 1990)
Such an analysis also predicts why the sentences in (37) allows the mod-
ification of the R point by now. The interaction of modality with the present
perfect (see 7.4) can be used to explain cases like (40) where the R point in a
Reichenbachian tense diagram is not associated with the E point.
(40) a. If you remember how we were and how we lived, then we will have
lived again.
b. By 1965, several or all of these systems will have been fully tested
and their reliability established. (Brown corpus ch21)

7.5. Conclusion

This section reviewed the evidence presented in the literature to show that
future tense and modality in English can be shown to be different. We saw
that there seems to be no good argument against a unified treatment of will as
a modal . There is also evidence that diachronically will and shall have modal
This section did not consider other less principled arguments put forward (citing
evidence from negation, replacement by shall in indirect speech, alternation with I
think and probably and combination with a simple present tag, like does s/he?) to show
that the future will differs distributionally from the volitional or the modal will. See
Haegeman (1983), Huddleston (1995) for counter-examples to these arguments.
 FUTURE TENSE AND MODALITY 105

origins in English (Jespersen 1931, Comrie 1989, Haegeman 1983:39–40).

8. The Modality of Prediction

Having shown in 7 that the futurity of will is not incompatible with a modal
semantic interpretation let us look at some implementations of this idea.
Verbs that express desires or demands evaluate their complements in a
future time relative to their own event time. According to Boyd and Thorne
(1969), Palmer (1979), Haegeman (1983) and Enç (1996), the futurity of will
is a consequence of its interpretation as a modal of prediction and hence will
can unambiguously be taken to be a modal.
According to this view the futurity in (41c) and (41d) is a consequence
of the modality of will analogous to the futurity in (41a) and (41b) being a
consequence of the lexical semantics of the verbs expect and want.
(41) a. I expect to win the race. (Enç 1996)
b. “He wants me to go with him tomorrow”, she told Kate.
(Brown corpus ck15)
c. I will win the race.
d. “That critter will be back tomorrow”, predicted George Rust, “and
he’ll bring fifty of his kind back with him. Blue Throat won’t stand
for this. He’ll shoot up the town”. (Brown corpus cn26)
However this cannot be an adequate theory if will is the only modal that
displays futurity due to the modality of prediction. Enç (1996) gives the ex-
amples in (42) to show that other forms of modality show the same effect of
futurity. (42a) and (42b) are examples of deontic modality and (42c) is an
imperative .
(42) a. You must do fifty push-ups. (Enç 1996)
b. Sally may go to the party if she finishes her work. (Enç 1996)
c. Do fifty push-ups. (Enç 1996)
The time of doing push-ups in (42a) and (42c) and the time of going to
the party in (42b) are required to be after the utterance time. However, the fu-
turity is not conclusively the result of the intensional expressions in the above

No specific proposal is made (see Palmer 1979:11).

Deontic modality denotes the feasibility and permissibility of the core proposition,
and ability and obligation of the agent (see Palmer 1986).
This example is relevant under the assumption that imperatives have an intensional
expression associated with their semantics.
106 ANOOP SARKAR

examples. For instance, the sentences in (43) are identical to the sentences in
(42) except that the intensional expressions are replaced with some appropri-
ate context to elicit a reading in which the time of doing push-ups and the time
of going to the party are after the utterance time (analogous to (42)). The sen-
tence in (45) is the real-life counterpart of the constructed examples in (43).
As (43b) shows, the prediction of will is not at the utterance time. Rather, the
prediction holds at a time specified by the when clause. The examples in (43)
and (45) give evidence against Enç’s (1996) implicit assumption that English
does not have any present tense interpretation that cannot be collapsed to the
utterance time, and a sentence like “John must leave.” does not project tense
(Enç 1996:354).
(43) a. You do fifty push-ups and I’ll give you your money.
b. When you do fifty push-ups, I’ll give you your money.
c. Sally goes to the party only if she finishes her work.
(44) a. You’ll do fifty push-ups and I’ll give you your money.
b. When you’ll do fifty push-ups, I’ll give you your money.
c. Sally will go to the party only if she will finish her work.
(45) But come the next session of Congress, State can expect only that its
summer guest will bite its hand when it goes to the Capitol asking
money for diplomatic entertaining expenses abroad or for living ex-
penses for its diplomats. (Brown corpus cf46)
The futurity in the present tense sentences (43) and (45) could be claimed
to be derived from an implicit will present in those sentences (as explicitly
shown in (44)). We will show in 9 that this cannot be true.
Note that all we are arguing for here is that while will can be treated as a
modal of prediction, it does not always have the utterance time as its reference
point.
The examples in (43) also show why the analysis in Boyd and Thorne
(1969) leads to difficulties. Boyd and Thorne (1969) attempt to give an analy-
sis of modals using the notion of a speech act. They see the difference between
the sentences in (46) as reflected in their illocutionary force as characterized
by the sentences in (47).
(46) a. He goes to London tomorrow.
b. He will go to London tomorrow.
(47) a. I state He goes to London tomorrow.
b. I predict He goes to London tomorrow.
(48) a. He will go to London now/tomorrow.
b. I predict He goes to London now/tomorrow.
 FUTURE TENSE AND MODALITY 107

c. I direct He goes to London now/tomorrow.

However, this approach does not explain the temporal interactions of the
various illocutionary uses of will. For example, in (48) the illocutionary forces
of a prediction and a directive can both be interpreted at utterance or future
time (see (49c) for a more convincing example of the temporal interpretation
of (48b) as the utterance time). Palmer (1979) also points out other problems
with this approach.
Modals like must and may also have interpretations where no shift to fu-
ture time takes place. This is also true of will. It is important to note that
examples (49a) and (49b) although predicated of the present time are cases
of deontic modality, while (49c) although predicated of the present time is,
however, a case of prediction.
(49) a. Sally must be in her office now. (Enç 1996).
b. Go home, your mother may be worried.
c. Go home, your mother will be worried.
d. Sally will be in her office now.
Until the precise nature of the temporal contribution of modals is resolved
the analysis of the futurity of will being derived from the modality of predic-
tion cannot be conclusive. We will return to the notion of the temporal nature
of modality in 11.
Haegeman’s (1983) analysis of will is one in which it is neither exclusively
a modal nor tense morphology, rather it is given a more general meaning from
which both its interaction with tense and modality is predicted based on con-
textual and epistemic facts about the discourse. The particular lexical meaning
given to will is as follows (from Haegeman 1983:162):

1. non-factuality, i.e. time-based objective uncertainty

2. actuality, speaker-based subjective certainty
3. event-time orientation

Haegeman (1983) tries to provide will with one basic contribution towards
a discourse model, with a wide range of contextually and situationally defined
specifications which express under which conditions in a discourse which par-
ticular sense of will is appropriate. However, the crucial point of future time
reference is resolved via ascribing to the basic meaning of will the ability to
shift the event time (see (3)). While the basic meaning attributed to will is
that of a modal of prediction (“a modal of conditionality indicating maximal
likelihood”), the analysis presupposes an intention towards future-time speci-
fication in order to distinguish the future from the modal.
108 ANOOP SARKAR

Enç (1996) gives a reasonable answer within possible world semantics

to the temporal contribution of modals. The assumption is that possibility
and necessity is over world-time pairs. You could equivalently assume that
quantification is over entire histories of possible worlds (as in Figure 1). The
interpretation given to modals that refer to future time is given in (50).
(50) MODAL[S] is true at iff in every world accessible to there
is an interval such that and S is true at . (Enç 1996)
Here, will is considered one of the modals that can be interpreted by (50),
and the futurity is derived due to the fact that the worlds are ordered as histo-
ries. Crucially, there is a particular sense of will that has an interpretation as
in (50). The will that is used, say, in John will leave. However, there is also
the will that is used in Sally will be in her office now which will get an inter-
pretation that is given by (51) (as far as we can tell, the epistemic will with
now does not have to be evaluated with respect to the current world). Notice
that this interpretation is still consistent with considering will as a modal of
prediction.
(51) MODAL[S] is true at iff in every world accessible to there
is an interval such that and S is true at .
Considering the interpretation available to the epistemic will in John will
have left yesterday, yet another interpretation has to be made available, namely
the one in (52) (assuming that the R temporal point for the perfect can be
handled appropriately in some compositional way).
(52) MODAL[S] is true at iff in every world accessible to there
is an interval such that and S is true at .
Hence, such a view has to derive all the possible temporal interpretations
of the modal will via ambiguous models.
We will come back to the notion of will as a modality of prediction in 11.

9. The Futurate

In many cases (cf. 8), the reference to future time in sentences containing will
is compared to what is termed as the futurate construction (Smith 1983), e.g.
(53).
(53) Tomorrow, the Yankees play the Red Sox. (Vetter 1973)
An argument can be made (as in Lakoff 1971) that this evidence is not
relevant since one can hypothesize an implicit will in sentences like (53) hence
 FUTURE TENSE AND MODALITY 109

explaining why they refer to a future time. However, as (Vetter 1973) points
out using the examples in (54), the futurate behaves differently from sentences
with an explicit will in many contexts.
(54) a. Tomorrow, the Yankees will play well.
b. #Tomorrow, the Yankees play well.
c. Tomorrow, the astronauts will splash down safely.
d. #Tomorrow, the astronauts splash down safely.
e. The Yankees will have played the Red Sox next Thursday.
f. #The Yankees have/had played the Red Sox next Thursday.
This tells us two things:
An analysis of the futurate construction cannot be collapsed to that of will.
Reference to future time is not uniform. The futurity of will is somehow
distinct from the futurity displayed in the futurate construction.
Some consider the simple present tense in sentences like (54) to be the
“true” future tense in English (Steedman 1996), while others like Vetter (1973),
Huddleston (1977), Smith (1983) consider the futurate to be formally in the
present and making a modal statement (a schedule or plan) about the current
time .
Steedman (1996) (citing Isard and Longuet-Higgins 1973) points out that
the past tense demands that its past reference point (the R point) be explicitly
established, either by a modifier, such as a when clause, or by preceding dis-
course. Thus (55a) is inappropriate as the first utterance of a discourse (except
under cases where a temporal reference is accommodated), while (55b) is fine.
(55) a. #Chapman breathed a sigh of relief. (Steedman 1996)
b. When Nixon was elected, Chapman breathed a sigh of relief.
The futurate too, is anaphoric, like the past with the same need for an “an-
chored” reference point. Hence, (56a) is inappropriate when discourse-initial,
whereas the suitably anchored (56b) is fine. Steedman (1996) gives this as
evidence that the present tense morphology (or the pure future tense interpre-
tation) is a true tense since it behaves analogous to the past tense. Given the
evidence in (56a) and (56b) an explanation of the present tense morphology as
being under-specified with respect to temporal interpretation might be a better
alternative. This can also help us explain cases of the “dramatic present” (also
Also Binnick (1971, 1972), Palmer (1979), Haegeman (1989) give evidence to
support the view that other “periphrastic futures” in English such as be to, be about to,
be going to are also modal in character.
110 ANOOP SARKAR

called the historic present) in (57) where the present tense is bound in some
discourse by a past tense (sometimes immediate past) interpretation.
(56) a. #Harry moves to Philadelphia. (Steedman 1996)
b. Next Tuesday, Harry moves to Philadelphia.
(57) But Voltaire perseveres. He goes to the chief himself. “At what uni-
versity did you study”? He asks. He refuses to believe that the bandit
chief never attended a higher institution. “To have become so corrupt”,
he says, “surely you must have studied many arts and sciences”. (dra-
matic present) (Brown corpus ck08)
We shall see in 11 that will too shows anaphoricity but of a somewhat
more complicated nature.

10. The Evidence from Would

Certain syntax-semantic reasons have been given to show that would is syn-
chronically related to will, i.e. would = woll + PAST
Recall the discussion of sequence of tense from 6.3. Now consider the
view that would = woll + PAST. Then embedded clauses in past tense with
would in the matrix clause should also show sequence of tense effects. Both
Abusch (1988) and Ogihara (1995) argue that this is true. The argument is
that sequence of tense in English ensures that for a clause to be co-temporal
with its complement clause which has past tense morphology, the clause itself
has to carry past tense morphology. For instance, in (58a) John’s talk with his
mother is co-temporal with their meal together and in (58b) the time of saying
is the same as the time the predicate “Stalin and Molotov being less reliable
defenders of Russia” holds.
(58) a. John decided a week ago that in ten days at breakfast he would say
to his mother that they were having their last meal together.
(Abusch 1988)
b. The White Russians and the Ukrainians would say that Stalin and
Molotov were far less reliable defenders of Russia than Curzon and
Clemenceau. (Brown corpus cj36)
Replacing would with will in (59) only gives the shifted reading (see
Ogihara 1995).
PAST indicates past tense morphology. Note that this conclusion is not as obvious
as it seems. Hornstein (1990) for instance, does not give would such a representation.
The judgment is from Ogihara (1995). Some speakers seem to get the co-temporal
 FUTURE TENSE AND MODALITY 111

(59) a. John decided a week ago that in ten days at breakfast he will say to
his mother that they were having their last meal together. (Ogihara
1995)
b. The White Russians and the Ukrainians will say that Stalin and
Molotov were far less reliable defenders of Russia than Curzon and
Clemenceau.
Also, the co-temporal reading vanishes when the complement clause is an
event as in (60) where only the shifted reading is available. Thus (given that
the above judgments are valid) clauses with would pattern with other clauses
with a morphological past tense with respect to sequence of tense phenomena.
(60) a. Few months’ later, John would hear that Mary failed the test.
b. Few months’ later, the gardener would say that the roses died.
c. Sally would think that John drank the beer.
There are, however, cases where would in an embedded clause can be
modified by future adverbials and are not co-temporal with the higher clause
as in (61a). But ordinary past tense morphology also shows this behavior as
shown in (61b).
(61) a. Phoebe didn’t realize that the Yankees would play the Red Sox the
next day (so she agreed to fly to Mauritius with Henry Kissinger).
b. Phoebe didn’t realize that the Yankees played the Red Sox the next
day. (Vetter 1973)
It is important to note that we have not made any commitment towards a
particular interpretation for the PAST tense morphology. However, a unified
interpretation might be given for cases where would can take on a modal in-
terpretation as in (62) where a counterfactual reading is obtained. This can be
summarily explained by referring to the theories which have linked the mor-
phological past tense to the modality of counterfactuals (Isard and Longuet-
Higgins 1973, Isard 1974, Iatridou 1996).
(62) Mary would have finished the book.
Also, Huddleston (1995) gives the following examples (63) and (64) to show
the temporal contrast of would with will. These examples can be explained
easily if would has a morphological past tense but are problematic for theories
reading in (59a), but get only the shifted reading in (59b). Perhaps the higher verb
decided is to blame, although this would be unexpected given that sequence of tense is
considered to be local in nature (Hornstein 1990, Enç 1987).
Also see the discussion in Palmer (1986:200–215).
112 ANOOP SARKAR

that treat would as being systematically ambiguous (as in Hornstein 1990).

(63) a. I have no money on me but he won’t lend me any. (volitional)
b. I had no money on me but he wouldn’t lend me any. (volitional)
c. In a few months’ time their love will change to hate.
d. Only a few months’ later their love would change to hate.
(past reference time)
(64) a. It will rain before we get home.
b. You said it would/*will rain before we got home.
Hence, there seems to be good evidence to show that would = woll +
PAST tense morphology.

11. Thesis

In previous sections we have laid out a fairly comprehensive survey of the

various viewpoints and related empirical facts cited in the literature on what
semantics should be assigned to will. We also explored some additional facts
such as the futurate and the use of would which were related to the use of will.
In this section, we see if we have enough evidence to answer the question that
was posed at the beginning of the paper: Is will part of the tense system or the
modal system or is it simply homonymous?
First, let us encapsulate the conclusions from the previous sections:

Future temporal interpretation in sentences with will have distinct effects

on temporal modifiers, aspectual markers and sequence of tense when
compared with past temporal interpretations.
There was no convincing argument against the treatment of will as a
modal.
While will can be suitably treated as a modal of prediction, the prediction
does not always have the utterance time as the reference point.
would can be thought of as a modality of prediction plus PAST tense
morphology.

After reviewing several arguments presented for and against the two sides
of this question, we are led to the conclusion that the question is ill-posed since
neither alternative alone could account for the empirical facts. However, there
are some additional facts considered here which might shed some new light
on the problem.
 FUTURE TENSE AND MODALITY 113

We have seen in 9 that the PRES tense morphology can take various
temporal interpretations,
The arguments presented in 7.4 can be used to justify will have as having
the present perfect as a temporal interpretation,
The PRES tense morphology can also be bound by a generic operator,
giving us the future generics in (7).

This leads us to the conclusion that a semantics for will can be always
decomposed into a composition of the semantics given to a modal contribution
and the temporal contribution of the PRES tense. It is important to note that
both of these components can contribute to the temporal interpretation of will.
Hence, the semantics for will is not exclusively modal as argued by one camp,
neither is it ambiguous between a tense and a modal as argued by the other
camp, but rather each instance of will seems to be simultaneously a modal and
a tense morpheme. That is, will equals the modality of prediction plus PRES
tense morphology.
This is implicit in the various analyses in 10. Let us take would to be
woll + PAST and analogously will to be woll + PRES (borrowing the notation
from Abusch 1988).
But given that both the modal and the temporal parts of will can be anaphoric
in nature, it would mean that will can be anaphoric in two dimensions simulta-
neously. Is this property empirically observed? The remainder of this section
will attempt to show that such a property for will has been attested.
As Binnick (1972) points out sentences with will that are out of context
seem elliptical. Sentences such as the ones in (65) require some context to be
felicitous.
(65) a. The rock’ll fall. (Binnick 1972)
b. In fact, she’ll die.
(66) is not elliptical in the same way.
(66) The rock’ll fall if you pull the wedge out from under it. (Binnick 1972)

There are two kinds of anaphoric reference being considered here. The first is
exemplified by the sentence in (67).
(67) Don’t pull the wedge out from under that boulder, you nitwit! The
rock’ll fall. (Binnick 1972)
The distinction between futurity in will and that in the futurate is now explicit.
Whereas in the futurate only PRES tense morphology exists, in will there is both the
modality of prediction and the existence of PRES tense morphology.
114 ANOOP SARKAR

The will in (67) is making reference to an argument (in the sense of Stone
1994), where in this case will has an epistemic meaning. It is important to note
that in (67) will is not necessarily temporally bound to a future time.
In conditionals, this also explains why will can shift into the future twice
as in (68). The fact that Tommy cries or Eric drops out of school is a conse-
quence of the antecedent in both the conditionals in (68) and hence in those
worlds where Sally wins the race, Tommy’s crying is predicted to hold at those
worlds.

(68) a. If Sally wins the race, Tommy will cry. (Enç 1996)
b. If Eric fails the test, he must drop out of school.

The other kind of anaphoric reference is when the PRES tense morphol-
ogy in will gets its temporal interpretation from discourse (see Partee (1973),
Muskens (1995, 1996) for other forms of temporal anaphora in discourse)
shown in the examples in (69) via a co-indexation between the PRES tense
in will and the location in discourse where the temporal interpretation is lo-
cated.

(69) a. When the sun sets , we’ll be frozen. (Binnick 1972)

b. Someday Americans will be able to visit Albania.
c. Now Salomé will do her world-famous Dance of the Seven Ostrich
Feathers.

Any semantic analysis of will must account for these two cases of anaphoric
reference.
This analysis also extends to account for would as pointed out in Stone
(1996), when we take would to be woll + PAST tense morphology which was
argued for in 10.

(70) a. Only a few months later their love would change to hate.
b. My neighbours would kill me. (Stone 1996)

As sentences such as (63d) repeated as (70a) show, the temporal reference

point for would is strongly anaphoric. This patterns with the anaphoric modal
uses of would (Stone 1996) in (70b) uttered while looking at a high-end stereo
in an electronics store. The anaphoric (or accomodated) context is one in
which the speaker bought the stereo and played it at its natural volume. Anal-
ogous to this, the PAST tense morphology of would in (63d) makes anaphoric
reference to a past temporal point where two people are in love, with respect
to which love changes to hate in the future (a few months’ later).
 FUTURE TENSE AND MODALITY 115

12. Conclusion

In this paper, we began with the following question: Is will part of the tense
system or the modal system or is it simply homonymous? After reviewing
several arguments presented for and against the two sides of this question,
we were lead to the conclusion that the question was ill-posed since neither
alternative alone could account for the empirical facts. Any semantics for will
must account for a simultaneous contribution from the modal as well as the
tense system. Note that this is distinct from stating that will is ambiguous
between a modal and a future tense. In comparison to existing analyses, by
increasing the dependence on contextual information a much simpler account
for the semantics of will can be afforded.

References

Abusch, D. (1988) “Sequence of Tense, Intensionality and Scope,” in Proceedings of

WCCFL 7, Stanford, 1–14.
Binnick, R. I. (1971) “Will and Be Going To,” in Papers from the Seventh Regional
Meeting, Chicago Linguistic Society, Chicago, Chicago Ling. Soc.
Binnick, R. I. (1972) “Will and Be Going To II,” in P. M. Peranteau, J. N. Levi, N.
Judith, and G. C. Phares, eds., Papers from the Eighth Regional Meeting, Chicago
Linguistic Society, Chicago, Chicago Ling. Soc.
Boyd, J., and J. P. Thorne (1969) “The Semantics of Modal Verbs,” Journal of Lin-
guistics 5, 1–192.
Coates, J. (1983) The Semantics of Modal Auxiliaries, Croom Helm, London.
Comrie, B. (1985) Tense, Cambridge University Press.
Comrie, B. (1989) “On Identifying Future Tenses,” in W. Abraham and T. Janssen,
eds., Tempus – Aspekt – Modus: die lexicalischen und grammatischen Formen in
den germanischen Sprachen, Niemayer, 51–63.
Davidson-Nielsen, N. (1988) “Has English a future?: remarks on tense,” in I. Lindbad
and M. Ljung, eds., Proceedings from the Third Nordic Conference for English
Studies, volume 1, Stockholm, Almquist and Wiksell, 53–59.
Declerck, R., and I. Depraetere (1995) “The Double System of Tense Forms Referring
to Future Time in English,” Journal of Semantics 12:3, 269–310.
Enç, M. (1987) “Anchoring Conditions for Tense,” Linguistic Inquiry 18, 633–657.
Enç, M. (1996) “Tense and Modality,” in S. Lappin, ed., The Handbook of Contem-
porary Semantic Theory, Blackwell Reference, Oxford, UK ; Cambridge, Mass.,
345–358.
116 ANOOP SARKAR

Haegeman, L. M. (1983) The semantics of will in present-day British English : a uni-

fied account, Verhandelingen van de Koninklijke Academie voor Wetenschappen,
Letteren en Schone Kunsten van Belgie, Klasse der Letteren. Jahrg. 45, nr. 103,
Paleis der Academien, Brussel.
Haegeman, L. M. (1989) “Be Going to and Will: A Pragmatic Account,” Journal of
Linguistics 25, 291–317.
Haiman, J. (1980) Hua: a Papuan language of the Eastern Highlands of New Guinea,
J. Benjamins, Amsterdam.
Hornstein, N. (1990) As Time Goes By: Tense and Universal Grammar, MIT Press,
Cambridge, Mass.
Huddleston, R. (1969) “Some observations on tense and deixis in English,” Language
45, 777–806.
Huddleston, R. (1977) “The futurate construction,” Linguistic Inquiry 8, 730–736.
Huddleston, R. (1995) “The Case against a Future Tense in English,” Studies in Lan-
guage 9:2, 399–446.
Iatridou, S. (1996) “Fakes?” presented at GLOW ’96.
Isard, S. (1974) “What would you have done if ,” Theoretical Linguistics 1, 233–
255.
Isard, S., and C. Longuet-Higgins (1973) “Modal Tic-Tac-Toe,” in R. J. Bogdan and I.
Niiniluoto, eds., Logic, Language, and Probability, D. Reidel, Dordrecht-Holland,
189–195.
Jespersen, O. (1924) Philosophy of Grammar, W. W. Norton (publ. 1965).
Jespersen, O. (1931) A Modern English Grammar on Historical Principles. Part IV:
Syntax. Third Volume: Time and Tense, (Various Editions).
Kratzer, A. (1977) “What ‘must’ and ‘can’ must and can mean,” Linguistics and
Philosophy 1, 337–355.
Kratzer, A. (1991) “Modality,” in A. von Stechow and D. Wunderlich, eds., Semantics:
An international handbook of contemporary research, de Gruyter, Berlin, 639–550.
Lakoff, G. (1971) “Presupposition and Relative Well-formedness,” in D. Steinberg and
L. A. Jakobovits, eds., Semantics, Cambridge University Press, Cambridge.
Lyons, J. (1977) Semantics I, Cambridge University Press, Cambridge.
Muskens, R. (1995) “Tense and the logic of change,” in Lexical Knowledge in the
Organization of Language, volume 114 of Current Issues in Linguistic Theory,
John Benjamins, Philadelphia, 147–184.
Muskens, R. (1996) “Combining Montague semantics and discourse representation
theory,” Linguistics and Philosophy 19:2, 143–186.
Ogihara, T. (1995) “The Semantics of Tense in Embedded Clauses,” Linguistic Inquiry
26:4, 663–679.
Palmer, F. R. (1979) Modality and the English modals, Longman, London.
 FUTURE TENSE AND MODALITY 117

Palmer, F. R. (1986) Mood and Modality, Cambridge University Press.

Partee, B. H. (1973) “Some structural analogies between tenses and pronouns in En-
glish,” Journal of Philosophy 70, 601–609.
Perkins, M. R. (1983) Modal Expressions in English, Frances Pinter, London.
Prior, A. (1967) Past, Present and Future, Oxford University Press.
Quirk, R., S. Greenbaum, G. Leech, and J. Svartik (1985) A Comprehensive Grammar
of the English Language, Longman, London.
Reichenbach, H. (1947) Elements of Symbolic Logic, Reprinted in 1966 by Free Press.
Smith, C. S. (1978) “The Syntax and Interpretation of Temporal Expressions in En-
glish,” Linguistics and Philosophy 2, 43–100.
Smith, C. S. (1983) “The Temporal Reference of the English Futurate,” in L. Tas-
mowski and D. Willems, eds., Problems in Syntax, Studies in Language. Series
No. 2, Ghent, 273–288.
Steedman, M. (1996) “Temporality,” in J. van Benthem and A. ter Meulen, eds.,
Handbook of Logic and Language, Elsevier North Holland.
Stone, M. (1994) “The Reference Argument of Epistemic must,” in Proceedings of the
International Workshop on Computational Semantics, 181–190.
Stone, M. (1996) “The Anaphoric Parallel Between Modality and Tense,” ms. Univer-
sity of Pennsylvania.
Vetter, D. C. (1973) “Someone Solves This Problem Tomorrow,” Linguistic Inquiry 4,
104–108.
Wekker, H. (1976) The expression of future time in contemporary British English,
North-Holland, Amsterdam.
Yavaş, F. (1982) “Future Reference in Turkish,” Linguistics 20, 411–429.
Zagona, K. (1989) “On the Non-Identity of Morphological Tense and Temporal In-
terpretation,” in C. Kirschner and J. DeCesaris, eds., Sel. Papers from 17th Ling.
Symposium on Romance Langs. (XVII LSRL), Rutgers Univ., 27-29 Mar. 1987,
Studies in Romance Linguistics, Benjamins, Amsterdam.

Dept. of Computer and Information Science

University of Pennsylvania
200 South 33rd Street,
Philadelphia, PA 19104–6389 USA
anoop@linc.cis.upenn.edu