Conditional Clauses: External and Internal Syntax

LILIANE HAEGEMAN

Abstract: The paper focuses on the difference between event-conditionals and premise-
conditionals. An event-conditional contributes to event structure: it modifies the
main clause event; a premise-conditional structures the discourse: it makes manifest a
proposition that is the privileged context for the processing of the associated clause. The
two types of conditional clauses will be shown to differ both in terms of their ‘external
syntax’ and in terms of their ‘internal syntax’. The peripheral structure of event condi-
tionals will be shown to lack the functional head Force, which encodes illocutionary
force. Event conditionals are merged inside the IP of the matrix clause. Premise-
conditionals contain the head Force and they are merged outside the associated CP.

1. Aim and Scope of the Paper

This paper will relate some of the interpretive properties of conditional sub-clauses
to their syntax. The paper focuses on the difference between the indicative
conditional clauses in (1).

(1) a. If it rains we will all get terribly wet and miserable.

b. If [as you say] it is going to rain this afternoon, why don’t we just stay
at home and watch a video?

This research was undertaken in view of the colloquium on conditionals organised at the
University of London Philosophy Programme of the School of Advanced Study on 8th March
2002. I thank the organisers of the colloquium for inviting me and thus sparking off my renewed
thinking on this issue. Thanks to Renaat Declerck for generously making available the manuscript
of Declerck and Reed (2001). Thanks are also due to the following for generous help with data
and for discussion of specific points: Enoch Aboh, David Adger, John Collins, Siobhan Cottell,
Dorothy Edgington, Marcus Giaquinto, Sam Guttenplan, Ruth Huart, Barry Lee, David
Lightfoot, Joan Maling, Jim McCloskey, Philip Miller, Bernadette Plunkett, Genoveva Puskas,
Luigi Rizzi, Andrew Simpson, Neil Smith, Ellen Thompson, Deirdre Wilson.
Versions of the paper were presented at the English linguistics meetings at Lille III, at the
Tournesol conference on Tense and Point of View (Paris 15.12.2001), at SOAS, at the
Linguistics department of the University of Reading. Thanks to the various audiences for their
help. Needless to say, none of those mentioned above can be held responsible for any remaining
errors.

Address for correspondence: UFR Angellier Université Charles de Gaulle- Lille III BP 149,
59653 Villeneuve d’Asq Cedex, France.
E-mail: Haegeman@univ-lille3.fr

Mind & Language, Vol. 18 No. 4 September 2003, pp. 317–339.

# Blackwell Publishing Ltd. 2003, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
318 L. Haegeman

The conditional clause in (1a) modifies the main clause event: it expresses an event
which will lead to the main clause event; the conditional clause in (1b) expresses a
premise leading to the question raised in the matrix clause. I refer to the conditionals
illustrated in (1a), which relate to event structure as ‘event-conditionals’, and to those
illustrated in (1b), which relate to discourse structure, as ‘premise-conditionals’.
Against Declerck and Reed (2001, pp. 37–8), who consider the difference between
(1a) and (1b) to be purely interpretive, I argue that the conditional clauses in (1)
differ both in terms of their ‘external syntax’ and in terms of their ‘internal syntax’.
The structural differences between conditional clauses extend to other adverbial
clauses, such as those introduced by while, because, when, since, as, so that etc. The
distinction between different types of adverbial clauses has been signalled by, among
others, Rutherford (19701), Hooper and Thompson (1973), and more recently
Verstraete (2002). Elaborating my own earlier work on this issue (Haegeman,
1984a,b,c, 1991), I will propose that the conditional clauses in (1a) are more deeply
embedded than those in (1b). The event-conditional is inserted inside the matrix
domain, while the premise-conditional is attached outside the matrix domain.
Given this analysis, the label ‘matrix clause’ is actually a misnomer when referring
to the clause associated with a premise-conditional, and I will replace it by ‘associated
clause’. The different level of insertion/attachment or—to use minimalist
terminology (Chomsky, 1995; Epstein, 1999)—the different timing of the merger
of the conditional clause with the associated clause determines its semantic/
pragmatic contribution. While the event-conditional in (1a) is integrated in the
speech act of the matrix clause, the premise-conditional in (1b) has independent
illocutionary force, as also shown by the fact that such conditional clauses often
have an echoic reading. I will link the concept of illocutionary force to syntactic
structure.
The paper assumes a broadly generative framework. Section 2 briefly introduces
the two types of conditional clauses. Section 3 discusses scope effects distinguishing
the conditional clauses. Section 4 concerns the external syntax of conditionals.
Section 5 shows that the distinctions between the two conditional clause-types
generalise to other adverbial clauses. Section 6 proposes that in addition to
a distinction in terms of external syntax, the two types of conditional clauses
(and more generally the two kinds of adverbial clauses) also differ with respect to
their internal syntax. Assuming that the CP layer is a hierarchy of discrete
functional projections (the so-called ‘Split CP’), I propose that the CP domain
of the event-conditionals has less functional structure than that of peripheral
conditional clauses. Section 7 is more speculative and relates the internal make
up of the CP-domain to the anchoring of the utterance to the speaker. Section 8
summarises the paper.

1
Rutherford (1970) distinguishes restrictive vs. non-restrictive adverbial clauses. Hooper and
Thompson distinguish presupposed vs. non-presupposed adverbial clauses (1973, p. 492)
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 319

2. Event-conditionals vs. Premise-conditionals

2.1 Starting Point

It is intuitively clear that the attested conditional clauses in (2) and in (3) receive
distinct interpretations.2 (2) illustrates event-conditionals; (3) illustrates premise-
conditionals. In (2) the conditional sub-clause structures the event: it expresses a
cause leading to the effect expressed in the matrix clause. In (3) the conditional
sub-clause structures the discourse: it makes manifest a context for the question
raised in the associated clause. The premise-conditional is usually echoic (see also
Declerck and Reed (2001, p. 83)).

(2) If your back-supporting muscles tire, you will be at increased risk of

a.
lower-back pain. (Independent on Sunday, Sports, 14.10.1, p. 29, col. 3)
b. [President Bush and Mr Blair] will be taking even more [risks] if, and when,
a land war starts. (Independent on Sunday, Comment, 14.10.1, p. 25, col. 2)
(3) a. If we are so short of teachers (‘Jobs crisis grows as new term looms’,
August 30), why don’t we send our children to Germany to be
educated? (Letters to the editor, Eddie Catlin, Norwich, Guardian,
31.8.1, p. 9, col. 5)
b. If, as Bush and Blair maintain, they aim to leave Afghanistan better than it
was when they found it, then the west is committed to defend it against
all oppressors, whoever they might be. (Guardian, 9.10.1, p. 7, col. 3)

2.2 Typology
There is a considerable literature on the typology of conditional clauses, with the
labels differing from author to author. In (4) I provide just a few of the labels that
have been proposed for the distinction. The list is not exhaustive, and I refer to
Dancygier (1987), and also Declerck and Reed (2001) for a survey.

(4) Two kinds of if-clauses:

(1a),(2) ‘event-conditional’ (1b) (3) ‘premise-

conditional’
Event structure Discourse structure
Dancygier and Mioduscewska Consequential Non-consequential
(1984)
Van der Auwera (1986) Speech act about Conditional speech act
conditional
Sweetser (1984) Content conditional Epistemic conditional
Haegeman(1984a) Occurrence conditional Utterance conditional

2
I will often use attested examples from journalistic prose. This is because it is sometimes
difficult to provide natural-sounding examples without a context. With premise-conditionals,
I will provide extra material of the context when this seems particularly relevant.
# Blackwell Publishing Ltd. 2003
320 L. Haegeman

Declerck (1991) Situation conditional Truth conditional

Haegeman (1984a,b,c, 1991) Central Peripheral

I have proposed (Haegeman (1984a,b,c; 1991)) that the interpretive differences

between the two conditional types are reflected in their syntax, disagreeing with
Declerck and Reed, who say:

a subordinate clause is a syntactically dependent clause. Such questions as the

scope of negation, focusing, modality, etc. in the head clause are immaterial to
this, as they pertain, not to syntactic, but to semantic integration (Declerck
and Reed 2001, pp. 37–8)

Though if clauses are ‘subordinate’ (in that they are not normally used as isolated
root clauses3), my claim will be that the degree of embedding distinguishes the
conditional clauses. I will show that the distribution and interpretation of the
conditional clauses is also reflected in their internal structure. (For more details
on the internal syntax, see Haegeman (2001a, 2001b).)

3. Scope Effects

This section shows how operators in a matrix/associated clause can extend their
scope to the event-conditional, though not to the premise-conditional. Such scope
effects suggest that the event-conditional is more closely integrated with (and
hence dependent on) the matrix/associated clause than the premise-conditional,
which remains peripheral and hence more independent.

3.1 Time and Tense

Consider the event-conditional (2a), repeated here as (5). The conditional sub-
clause contains a present tense verb (tire) but it clearly refers to a future situation.
The future reading of the sub-clause is determined by the future time reference in
the matrix clause:

(5) (¼2a) If your back-supporting muscles tire (future), you will be at

increased risk of lower-back pain.

The tense in the event-conditional is said to be ‘temporally subordinated’ to that of

the matrix clause. This phenomenon sometimes has been referred to as ‘will

3
Except in cases like (i) and (ii):
(i) If you will come this way.
(ii) A: I will write the report.
B: If you like.
I assume that these are to be seen as elliptical, the associated clause being implied.
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 321

deletion’ (see Jespersen, 1940; Palmer, 1965, 1974; Zandvoort, 1975; McCawley,
1971; Leech, 1971; Close, 1980; Wekker, 1976, 1977; also Declerck, 1991 and
Declerck and Reed, 2001, for further references).
The premise-conditional clause is not temporally subordinated; a present tense
will have the temporal reading of a root clause present tense. This is illustrated in
(3) above. As shown by (6), in premise-conditionals future time is expressed
independently. The choice of tense in the premise-conditional is determined by
the factors that determine the choice of tense in an independent clause.4

(6) If I’m no longer going to be arrested for possessing cannabis for my own
consumption (Cannabis laws eased in drugs policy shake-up, October
24), shouldn’t I be able to grow my own? (Jason Cundy, Letter to the
editor, Guardian, 25.11.1, p. 9, col. 8)

That an event-conditional is temporally dependent on the associated clause in a

way that a premise-conditional is not is also suggested by the fact that in the
former, though not in the latter, if may often be expanded to if and when, as shown
by attested (2b). A similar expansion is not possible for premise-conditionals, as
shown by (7), based on (3a):

(7) (¼3a) *If and when we are so short of teachers (‘Jobs crisis grows as new
term looms’, August 30), why don’t we send our children to
Germany to be educated?

3.2. Adverbial Adjuncts

Event-conditionals may, and premise-conditionals may not, be within the scope of
adverbial adjuncts in the associated clause. This is illustrated in (8) and (9). (8a)
means that there is sometimes a causal relation between pressure and John’s
working best, (8b) says that it is often the case that John knows who is calling.

(8) a John sometimes works best if there is a lot of pressure.

b John will often know the caller if the phone rings.

Note that I do not wish to imply that the event-conditional must be in the scope of
a matrix clause adverbial. In (8c) the adverb regularly does obviously not quantify
the event of moving:

(8) c If you move to Paris, you will regularly be invited to give a talk at
their research group.

4
Present tense can be used for future time reference, but only when the future event is
considered as a present fact, for instance when the event is scheduled
# Blackwell Publishing Ltd. 2003
322 L. Haegeman

In (9a), the past tense in the premise-conditional is independent of the associated

clause and the main clause adjunct never does not bear on the conditional clause. In
(9b), the adverbial adjuncts probably and never do not scope over the premise-
conditional.

(9) a If it was already four o’clock when he left, John will never make it.
b If John lives in Rome, he probably never uses his bike.

In the light of the discussion of (8c), this is not a bi-directional relation. If an

adverbial in the matrix clause does not scope over the conditional, this does not mean
that the conditional must be a premise-conditional. The claim is only that when a
conditional is in the scope of a matrix clause adverbial it will be an event-conditional.

3.3 Affective Operators

Event-conditionals are within the scope of matrix clause negation and premise-
conditionals are not, as illustrated by the contrast between (10a) and (10b).

(10) a John won’t finish on time if there’s a lot of pressure on him.

b John won’t finish on time, if there’s (already) such a lot of pressure
on him now.

Scope effects are also observed with the other affective operators (Klima, 1964;
Ladusaw, 1980). For the event-conditional in (11a), the yes/no question bears on
the causal relation between John’s taking exercise and his fitness. In (11b), the
wh-constituent how targets the event-conditional.

(11) a Will John get any fitter if he takes more exercise? [? (P ! Q)]
b How can John get any fitter?
If he takes (¼future) more exercise in the future.

The premise-conditional in (12a) is outside the scope of the polar question operator.
This is reflected typographically: a clause -final question mark is somehow inappro-
priate. In (12b) B’s response cannot count as an appropriate reply to A’s question.

(12) a Is John getting any fitter (?), if [as you say] he is taking so much
exercise?
A: How can John get any fitter?
B: # If he is already taking so much exercise now.

The focus marker only in (13a) can be interpreted as focusing on the event-
conditional. The premise-conditional is not in the scope of only in (13b).

(13) a John will only finish the book if there is a lot of PRESSURE on him.
(‘only if’)
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 323

b John will only finish the BOOK, if there is already such a lot of
pressure on him. (i.e. ‘he won’t finish anything else’)

Clefting confirms the contrast: event-conditionals may be clefted, premise-

conditionals may not.

(14) a It is (only) if he takes more exercise that John will get fitter.
b *It is only if there is already such a lot of pressure on him now, that
John will finish the book.

3.4 Quantifiers and Bound Pronouns

Binding effects also distinguish the two kinds of conditional clauses. In (15a), the
pronoun he may either refer to some specific salient entity in the discourse, or it
may be a variable bound by the negative quantifier no one. On the other hand, in
(15b) the pronouns in the premise-conditionals have independent reference, they
are not bound by the quantifier no one:

(15) a No one will answer the phone if he thinks it’s his supervisor.
b Why does no one answer the phone, if he probably thinks it’s his
supervisor?

A pronoun in an event-conditional can be in the scope of a quantifier in the matrix

clause, a pronoun in a premise-conditional cannot. I assume that bound pronouns
must be c-commanded by their binder (Safir, 1984; Haik, 1984; May, 1986;
Hornstein, 1984 etc). A c-commands B if A is a sister of a constituent dominating
B. In (16) Q is a sister of YP, which may contain a pronoun bound by Q.5 In our
examples, YP dominates the conditional clause containing the bound pronoun.
XP

Qi YP

ð16Þ
conditional

pronouni

5
For various definitions of c-command, including a minimalist approach see (Epstein, 1999).
# Blackwell Publishing Ltd. 2003
324 L. Haegeman

From the fact that pronouns in premise-conditionals cannot be bound by the

subject in the associated clause, I conclude that such conditional clauses are not
c-commanded by the subject NP of the associated clause.

3.5 Parasitic Gaps

In (17), the relative pronoun who is construed as the object both of know and of
love. This phenomenon, in which one relative pronoun who relates to two empty
positions or ‘gaps’ (Ø), is referred to as the ‘parasitic gap phenomenon’ (see
Nissenbaum, 2000 and the references he cites).

(17) He is a man who if you know [Ø] you will love [Ø] (Jespersen, 1940,
p. 202)

It is assumed that in a parasitic gap construction both gaps are c-commanded by the
relative pronoun (here who) and by the antecedent (here a man). Nissenbaum
(2000, p. 64) proposes that an adjunct clause containing a PG forms a kind of
complex predicate with the VP of the main clause.6 Let us assume that the
formation of this complex predicate is only possible if the relevant VP c-commands
the adjunct with which it composes and which contains the parasitic gap. If, as I
will argue below, event-conditionals are, and premise-conditionals are not, in the
c-command domain of the matrix predicate, we correctly predict that parasitic gaps
will be licit in event-conditionals,7 and illicit in premise-conditionals.

(18) a. . . .a paper which if he finally sends [Ø] to me I shall read [Ø] with
interest.
b. ?*. . .a paper which I will receive [Ø] today, if (,as you say,) he
probably sent [Ø] to me yesterday.

4. External Syntax

I have argued (Haegeman, 1984a,b,c, 1991) that event-conditionals and premise-

conditionals are distinguished in terms of their ‘external’ syntax. Event-
conditionals are structurally integrated in the domain of the matrix clause, while
premise-conditionals are peripheral to the associated clause. The proposed struc-
tural difference has a prosodic reflex. The event-conditional is integrated in the
prosodic contour of the matrix, the premise-conditional is set off from the asso-
ciated clause both intonationally and by a pause.

6
In fact, he proposes that the complex predicate is formed by the adjunct and vP. ‘vP’ is a
projection dominating the lexical VP whose specifier hosts the subject argument. I use VP
here as a cover term for vP and VP.
7
In general, parasitic gaps are more easily accepted in non finite adverbial clauses than in finite
ones, though.
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 325

This section provides additional evidence for the structures proposed (sec-
tion 4.1.), formalises them in terms of the generative syntax framework (section
4.2.) and briefly comes back to the interpretation of premise-conditionals
(section 4.3.).

4.1 VP Anaphora
In the co-ordinated structure in (19), the VP of the second conjunct is ellipted.
The fronted event-conditional can be construed as being part of the substi-
tuted constituent, with so will Mary interpreted as (19b) or (19c). Observe crucially
that within the substituted VP, the referent of the possessive need not be
strictly identical to that of the preceding VP (his), but the possessive may also be
interpreted as relating to Mary (19c). This would be a case of so called ‘sloppy
identity’.

(19) a If his paper is accepted, John will go to the conference and so will
Mary.
b Mary will go to the conference if John’s paper is accepted.
c Mary will go to the conference if her paper is accepted.

In (20a), containing a premise-conditional, VP substitution is possible, but inter-

pretation (20b) with sloppy identity reading of the possessive is not available.

(20) a If his children aren’t in the garden, John will already have left
home, and so will Mary.
b If her children aren’t in the garden, Mary will already have left
home.

I conclude that VP substitution can affect event-conditionals but not premise-

conditionals. This contrast is expected if a premise-conditional is really ‘outside’
the associated clause, while the event-conditional is an integral part of it. The next
section formalises this idea.

4.2 Embedding, C-command and Scope

Following standard assumptions, I assume that the clause consists of three inter-
locking layers: (i) the thematic domain, headed by the predicate, say the verb,
which assigns its thematic roles, (ii) the functional domain headed by inflectional
head(s), summarised as IP, in which functional categories such as time, mood,
aspect are licensed, and (iii) finally the periphery of the clause, centred around the
subordinating conjunction, and labelled C(omplementiser)P.

# Blackwell Publishing Ltd. 2003

326 L. Haegeman

I think CP

C′

C IP

ð21Þ NP I′

I VP

that John will buy the book

(21) is standardly assumed, but it is generally agreed that it is a simplification, and

that each layer (VP, IP, CP) in fact decomposes into a sequence of layered
projections. For discussion of the articulation of VP, see Larson (1988, 1990) and a
great deal of later work; for the internal make-up of IP, see Pollock (1989) and Cinque
(1999), among others. In sections 6 and 7, I return to the articulation of CP.
I will assume, with among others Nissenbaum (2000, p. 23) and Thompson
(1994), that event-conditionals are adjoined to a projection of the matrix V.8
An event-conditional is a ‘central’ adverbial clause; it is part of the complex
matrix predicate. Event conditional clauses remain in the scope of matrix operators
and can be affected by VP ellipsis (see Thompson, 1994 for more precise discussion).
Premise-conditionals are peripheral. They do not have a local relation with V or
with I, hence they do not modify the predicate of the associated clause, nor are
they within the scope of operators in the associated clause. I assume that they are
adjoined to the associated CP.
Derivationally, adverbial clauses differ in the timing of the merger with the
associated clause. Central adverbial clauses (e.g. event-conditionals) are inserted
in (or merged with) the matrix clause early on in the derivation; they are merged
before IP is completed. Peripheral clauses (e.g. premise-conditionals) are merged
after the associated CP has been completed. The structure in (22b) is close to that
of co-ordination.9

8
Or to a higher projection between VP and the surface position of the subject.
9
In future research I hope to refine the structural account to distinguish peripheral premise-
conditionals from co-ordinated clauses.
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 327

a Central adverbial/‘event conditional’

I think CP

C′

C IP

ð22Þ
NP I′

VP

I VP Conditional clause

that John will buy the book if he finds it

b ‘Peripheral’ adverbial/premise-conditional

CP
ð22Þ
CP1 CP2
Peripheral conditional

The next section shows that in addition to the difference in external syntax,
conditional clauses also differ in terms of their internal make-up. See also
Haegeman (2001a). I will first briefly turn to the interpretation of peripheral
conditionals.

4.3 Interpretation of Peripheral Conditional Clauses

In my earlier work, I have related the interpretive differences between the two
types of conditional clauses to their external syntax. Essentially, what I proposed
there, and what I maintain here, is that the event-conditional is integrated in the
associated matrix clause. Its role is to modify the event-structure. The premise-
conditional is peripheral to the associated clause. Its role is to structure the
discourse. The merger of the premise-conditional, CP2, with the associated clause,
CP1, encodes a processing constraint on CP1 (cf. Blakemore, 1987; Sperber and
Wilson, 1986), in the sense that CP2 makes manifest the context from which CP1
# Blackwell Publishing Ltd. 2003
328 L. Haegeman

can be inferred, or in which the question raised by CP1 arises, or in which the
question raised in CP1 has to be answered. The fact that the particular proposition
CP2 and no other is made manifest as a processing context means that the speaker
thinks that this is the most relevant context for to the conclusion formulated or the
question raised in the associated CP (see also Declerck and Reed, 2001). The use of
if to introduce the contextual premise makes it manifest that the speaker does not
necessarily endorse the premise. He is entertaining it ‘for the sake of the argument’,
to draw the inference in the associated clause. In the italicised passage in (23),
below, the speaker explicitly distances himself from the conditional:

(23) Such is the respect felt everywhere for Media Monkey, your Media
Guardian diarist, I am reluctant to point out that its story (November 5)
about Rupert Murdoch rejecting candidates to be press columnist at
the Times belongs with the rubbish at the bottom of its cage. If
Mr Murdoch knows that Brian MacArthur is considering retirement
(and I somewhat doubt it), he has played the same role in his replacement
as in every other decision of mine to appoint or remove columnists
in the almost 10 years that I have been editor—absolutely none.
(Guardian, 6.11.1, p. 9, col. 5, letter to the editor, Peter Stothard,
The Times).

Further insightful discussion of the interpretation of premise conditional clauses is

provided by Dancygier (1987, 1998).
Declerck and Reed (2001, p. 131) propose that in what I call premise-
conditionals ‘the speaker makes two independent predictions: there are, as it were, two
illocutionary speech acts’. Obviously, if there are two speech acts, the interlocutor
may actually respond to either of them as discussed by Dancygier (1987, 1998, pp.
126–8). In section 7 below I return to the availability of illocutionary force in the
premise-conditionals, but I will first show that the distinction between the two
kinds of conditional clauses generalises to other adverbial clauses. Adverbial clauses
are either merged early on in the derivation, at which point they are semantically
integrated in the matrix proposition and structure the event, or they are merged
late in the derivation, at which point they make manifest the context for the
associated proposition and structure the discourse.

5. Adverbial Clauses: Central vs. Peripheral

This section provides additional examples of the distinction between two types of
adverbial clauses. The discussion is not exhaustive. I do not survey all possible
adverbial clauses nor will I apply all the diagnostics to each case. I simply show
briefly that the issue at stake is not confined to conditionals. (For a recent non-
generative view see Verstraete (2002)).
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 329

The conjunction while has two interpretations: it introduces a temporal adverbial

clause indicating that the event of the main-clause is co-temporal with that of the
sub-clause, or it makes manifest an assertion that is to be interpreted in parallel with
the assertion expressed in the associated clause. In the temporal interpretation, while
structures the event, in the contrastive interpretation, while is near equivalent to
whereas and structures the discourse.

(24) a John always works best while his children are asleep.
b John studied at Oxford, while Bill has a Cambridge Ph.D.

Like the event-conditional, a temporal while clause is fully integrated in the matrix
clause. Like the premise-conditional, a non-temporal while clause is structurally
peripheral. From now on, I refer to adverbial clauses that are integrated into
the matrix clause as ‘central’ adverbial clauses and to those that are not as
‘peripheral’.
As before, VP substitution distinguishes the two types of while clauses. A
temporal while clause can be construed as part of a substituted VP: in (25a) so
does James may be interpreted as ‘James works most efficiently while his wife is in
bed’, with a sloppy identity reading of the possessive. No such interpretation is
possible with non-temporal while. In (25b) so does James is construed as ‘James also
has a high powered job’. There is no implication that James is married or that
James’s wife is also unemployed.

(25) a While his wife is in bed John works most efficiently and so does
James.
b While his wife is unemployed, John has a high-powered job and so
does James.

Temporal while clauses are temporally subordinated. This is shown most clearly by
examples in which a present tense refers to a future time (26a). Non-temporal while
clauses are temporally independent of the associated clause (26b).

(26) a He will finish the book while he is on holiday.

b John’s son will go to Oxford, while John’s daughter will go
to UCL.

Temporal while clauses can be the focus of a question (27a) or focus (28a) operator
in the matrix clause, non-temporal while clauses cannot (27/8b):

(27) a A: When will John finish the book?

B: While he is on holiday.
b A: When will John’s son go to Oxford?
B: *While his daughter will go to UCL.

# Blackwell Publishing Ltd. 2003

330 L. Haegeman

(28) a It’s only while (time) you’re alive that human selfishness, or what-
ever, is held against you (Independent on Sunday, Review 14.10.1,
p. 9, col. 1)
b *It is only while (¼whereas) John’s daughter will go to UCL that
his son will go to Oxford.

Temporal while clauses may contain a parasitic gap licensed by a gap in the matrix
clause (see 3.5.), non-temporal while clauses cannot contain a parasitic gap licensed
by a gap in the associated clause.

(29) a This is the document which John managed to memorise -Ø while

he was copying Ø. (Haegeman, 1991, p. 235)
b *This is a research topic which John first taught -Ø in Cambridge,
while his son will be working on Ø- in Oxford.

A further contrast between temporal and non-temporal while clauses is that the
latter, though not the former, are compatible with question tags. In (30a) only a tag
corresponding to the matrix verb is licit, while in (30b) the tag corresponds to the
while clause itself (Hooper and Thompson, 1973, p. 468, Haegeman, 1991).

(30) a John had to be careful with his money while his daughter was a
student, didn’t he/*wasn’t she?
b Bill took a degree at Oxford, while his daughter is studying at
Cambridge, isn’t she/*didn’t he?

The contrast between a central type and a peripheral reading is instantiated by

other conjunctions: since and as may be either temporal or they may introduce a
premise; an adverbial clause introduced by so that is central when expressing
a purpose and peripheral when expressing a result etc. See Rutherford (1970),
Haegeman (1991) and Verstraete (2002) for discussion.

6. The Internal Syntax of Conditional clauses (and Adverbial Clauses)

6.1 The Basic Idea

So far, I have shown that adverbial clauses can be distinguished by their external
syntax, specifically in terms of the timing of the merger with the associated clause.
A closer examination of the adverbial clauses also reveals differences in their
internal syntax. This point is discussed in this and the next sections. My claim is
that the CP domain of the central adverbial clause (see 4.2. (21) for the notion CP)
is truncated (in the sense of Rizzi, 1994, 1995): central adverbial clauses, such as
event-conditionals, contain a structurally deficient CP, while peripheral adverbial
clauses, such as premise-conditionals, contain a complete CP, similar to that of
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 331

independent root clauses. I will provide the empirical arguments for this proposal
and I will sketch an account of the difference. For reasons of space, my presenta-
tion does not go into all the details and ramifications of the proposal. For further
discussion, I refer to Haegeman (2001a, b).
That structural reduction may be relevant for the classification of adverbial
clauses is suggested, though not explored in detail, by Hooper and Thompson
(1973). While these authors do not attempt to provide a structural explanation for
the reduced compatibility of so called M(ain) C(lause) P(henomena) with
embedded clauses, they do point out that such MCP are unavailable in what
they call ‘reduced clauses’, i.e. infinitives, gerunds, and present subjunctives
(1973, pp. 484–5).

6.2 The Split CP (Rizzi, 1997)

Section 4.2. presents the clausal periphery as a unitary projection CP, with a head,
C, a complement (the clause, IP) which it introduces, and a specifier. Clearly, a
proposal according to which CP is structurally deficient necessitates postulating a
more richly articulated CP, thus allowing for a ‘deficient’ i.e. ‘partial’ realisation.
That the unitary CP is a simplification has indeed been observed before (Culicover,
1991; Koizumi, 1995; Nakajima, 1996; Rizzi, 1997 etc). Assuming that the
conjunction that occupies the head C, the examples in (31) show that between
C and the subject there must be additional positions. In (31a) a fronted adjunct
intervenes, in (31b) a fronted topic intervenes, in (31c) a focalised argument
intervenes, and in (31d) a fronted negative constituent intervenes, triggering
inversion of the auxiliary:

(31) a I believe [CP that next week [IP they will introduce changes in the
law]].
b I think [CP that these three exams, [IP you should definitely pass this
year]].
c I believe [CP that JOHN [IP they won’t invite]].
d Lee said [CP that at no time would [IP she agree to visit Robin]].
(Culicover, 1991 p. 4 his (7a))

Based on comparative evidence Rizzi (1997) proposes that the interface between
the clause and its context be seen not as the projection of a single head ‘C’, but
rather as a layered structure articulated around a sequence of hierarchically organ-
ised functional heads, schematically presented in (32). The head Force hosts the
subordinating conjunction that. The specifier of the Topic projection (TopP) hosts
a topicalised constituent. To accommodate multiple topics, it is assumed that TopP
is recursive (hence *). The Focus projection (FocP) is unique, its specifier hosts the
focalised constituent. Finally, the head Finiteness encodes the [
FINITENESS] feature
of the selected IP. In the main body of his paper (specifically in his section 9), Rizzi
(1997) proposes that fronted adjuncts (31a) are similar to fronted topicalised
# Blackwell Publishing Ltd. 2003
332 L. Haegeman

arguments (31b) and move to TopP,10 but in footnotes (26, 30, 32) he envisages as
an alternative that they are left adjoined to FinP.11

(32) The split CP (Rizzi, 1997)

Force >Top*>Foc>Top*>Fin

6.3 Embedded Topicalisation/Focalisation

It has been observed in the literature that argument fronting (topicalisation or
focalisation) is not generally available in embedded domains (cf. Hooper and Thompson,
1973). Concerning data such as (33) Maki et al. say ‘embedded topicalisation is
consistently impossible in an adjunct clause’ (Maki, Kaiser and Ochi, 1999, pp. 3–4).

[33] *Before this book, Mary read, John had already read it.
(Maki, Kaiser and Ochi (1999, p. 3 (2e))

The ban on argument fronting in adverbial clauses to account for the ungram-
maticality of (33) is, however, stated too generally by Maki et al. As the contrasts
between (34) and (35) show, central adverbials disallow argument fronting, but
peripheral adverbials are more easily compatible with it.

(34) a *If these final exams you don’t pass, you won’t get the degree.
b *Mary listened to the radio while the dinner she was preparing.

(35) a If his SYNTACTIC analysis we can’t criticise, there is a lot to be said

against the SEMANTICS of the paper.
b While YOUR book they are using in two courses, MINE they haven’t
even ordered for the library.

Pursuing the idea of structural deficiency, I propose, as a first approximation, that

central adverbial clauses such as event-conditionals lack TopP and FocP, while the
CP layer of peripheral adverbial clauses such as premise-conditionals is complete. If
TopP and FocP are lacking in event-conditionals, we predict that argument
fronting is ungrammatical.

(36) a Event-conditional clause (central adverbial clause)

Force >Fin

10
In fact he distinguishes between adjunction to TopP for fronted adjuncts and movement to
SpecTopP for fronted arguments.
11
Rizzi (2001) postulates a more articulated CP-structure, which includes Int, a position for
the yes/no operator, and, more central to our discussion, a lower adjunct position, Mod(ifier),
to the left of a lower topic position. In this paper I will adopt the 1997 analysis according to
which adjuncts move to Fin. The argumentation is compatible with the assumption that
locally moved adjuncts target Mod. See Haegeman (2001b).
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 333

b Premise-conditional s (peripheral adverbial clause)

Force >Top*>Foc>Top*>Fin
c Root clause
Force >Top*>Foc>Top*>Fin

6.4 Adjunct Fronting in Conditionals

As shown by (37), adjuncts can be fronted in event-conditionals.

(37) a If with all these precautions you don’t succeed, you will have to try
again next week.

If the adverbial clause in (37a) is structurally deficient in that it lacks TopP, then
fronted adjuncts cannot be claimed to always move to TopP (one option envisaged
by Rizzi (1997)). Let us assume that fronted adjuncts can be associated with FinP.
Since peripheral conditionals are not structurally deficient, they will also be able to
contain fronted adjuncts:

(37) b If last week you were too tired to do anything after your class, why
do you think that next week will be any better?

At first sight, adjunct fronting does not distinguish the two kinds of conditional/
adverbial clauses, so they would not strictly speaking bear on the issue. However,
adjunct fronting does become relevant once we take into account the fronting of
adjuncts originating in a lower clause. In (38a) the fronted PPs in the coming academic
year and across the country are construed with the lower clause and they are fronted
into the higher domain. As shown by the (38b), long construal of a fronted adjunct
is also available in an embedded clause:12

(38) a. In the coming academic year, across the country, the government says
50% of all students will be exempt from paying a share of their
tuition fees. (Guardian, 26.6.1, p. 4, col. 1)
b She says that without her mother she doesn’t know what she would
have done. (Observer, 22.4.1, Review, p. 2, col. 2)

12
There are various discourse-related conditions on this kind of fronting, and I do not claim to
understand them. See Postal and Ross (1971), Bouma et al. (2001), Pollard and Sag (1987),
Hukari and Levine (1995) for more data. For a discussion of constraints on long fronting see,
for instance, Kempson (1975, p. 127), who shows that long adjunct extraction from factive
complements leads to ungrammaticality: (i)–(iii) are her judgements.
(i) *Tomorrow I regret that Mary is leaving.
(ii) *Tomorrow I resent it that Mary is leaving.
(iii) *Tomorrow I am surprised that Mary is leaving.
# Blackwell Publishing Ltd. 2003
334 L. Haegeman

Observe now that in the event-conditional (39), the fronted PPs in the next three
years and across the country can only be construed with says, implying that long
movement cannot have taken place.13

(39) The protests will only end if in the next three years, across the country, the
government says 50% of all students will be exempt from paying a share
of their tuition fees. (no construal with will be exempt).

We conclude that adjunct fronting is not a homogeneous phenomenon. Short

moved adjuncts target FinP. I propose that long moved adjuncts target TopP or
FocP. If event-conditionals lack TopP and FocP, my hypothesis in this paper, then
the unavailability of long construal of the adjunct in (39) follows.
If peripheral adverbial clauses have a complete CP structure, and, specifically, if
their periphery does contain TopP and FocP, we correctly predict that they will be
able to display long extraction of adjuncts. (40) illustrates a long fronted adjunct in
a contrastive while clause:

(40) Major improvements in funding can be expected, while in the next three
years, across the country, the government says 50% of all students will be
exempt from paying a share of their tuition fees.

Table 1 summarises the data discussed so far and relates the (im)possibility of
fronting in adverbial clauses to the structure of their CP.

7. Speculations on Subordinating Conjunctions and Illocutionary Force

Hooper and Thompson (1973) suggested that MCP only occur in ‘asserted
clauses’.14 Let us assume that MCP occur in clauses with illocutionary force. As
one piece of confirmation, recall that the premise-conditionals, which are compa-

Table 1 Fronting and CP-structure

Clause type Root Central adverbial Peripheral adverbial

CP structure Complete No FocP, no TopP Complete

Local adjunct fronting Yes Yes Yes
Argument fronting Yes No Yes
Long adjunct fronting Yes No Yes

13
Thanks to David Adger, Jim McCloskey, Ruth Huart, Joan Maling,, Philip Miller,
Bernadette Plunkett and Neil Smith for judgements.
14
Green (1976) and Andersson (1975) show that the term ‘assertion’ is perhaps too narrow.
I do not want to go into this in detail here.
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 335

tible both with argument fronting and long adjunct fronting, often have an echoic
interpretation (cf. Declerck and Reed, 2001, p. 131, for discussion, and see the
examples in (3) and (6)). This comes down to saying that they echo someone else’s
speech act. Event-conditionals lack this echoic reading. The availability of tags in
while clauses (see section 5) also confirms that peripheral adverbial clauses have
illocutionary potential. The attested because clause in (41), which embeds a (rhet-
orical) question, offers further support for assuming that peripheral adverbial clauses
have illocutionary force.15

(41) She was thinking of how Philip had buttoned up her fur coat on the
platform at Paddington, saying that she mustn’t catch cold because
what would they do then. (Ellis, Alice Thomas. The other side of the
fire, Penguin 1985, 1986, p. 93)

In the discussion of the internal syntax of event-conditionals, and of central

adverbial clauses in general, I proposed that they have the reduced functional
hierarchy (36a), while premise-conditionals display the complete array of func-
tional projections (36b), also available in a root clause (36c). Obviously, given my
current discussion of the interpretation of these adverbial clauses, labelling the
highest node in (36a) ‘Force’ is unfortunate. If the adverbial clause-types are
distinguished by the absence vs. presence of illocutionary force, then ideally this
difference should be reflected in their structural make-up. Clearly, we cannot just
eliminate the highest slot Force from (36a) since that very slot hosts the conjunc-
tion. A second problem with (36) is that it fails to distinguish root clauses from
embedded clauses, though the latter, and not the former, are systematically intro-
duced by a subordinating conjunction.
These problems can be solved if, as a first approximation and following Bhatt
and Yoon (1992) (see also Bennis, 2000), we distinguish the head that encodes
‘illocutionary force’ from the head that serves simply to subordinate a clause, to
‘make it available for (categorial) selection independently of its force’ (Rizzi, 1997,
note 6). Central adverbials would have the structure in (42a); peripheral adverbials
that in (42b), root clauses that in (42c):

(42) a Central adverbial: Sub >Fin

b Peripheral adverbial: Sub >Force>Top*>Focus>Top*>Fin
c Root clause: Force >Top*>Focus>Top*>Fin

Force encodes ‘illocutionary force’, the fact that the speaker takes on the propos-
ition as part of a speech act (assertion, prediction, question, etc). I assume that
topicalisation and focalisation depend on the presence of Force (see Bayer (2001)

15
In addition, if wh-fronting targets the focus position (Rizzi, 1997) then the data also confirm
that peripheral adverbials may contain FocP
(i) because [FocP what [Foc would [IP they do then.]]]
# Blackwell Publishing Ltd. 2003
336 L. Haegeman

for a similar proposal concerning German). For reasons of space I cannot develop
in more detail the syntactic implications of the analysis proposed here. See
Haegeman (2001a, b) for discussion of the syntax.
Data from Japanese confirm that central adverbial clauses such as event-
conditionals have a more restricted structure than peripheral adverbial clauses as
represented by premise-conditionals. Maki (pc)16 signals that in Japanese event-
conditionals are introduced by the conjunction ba and premise-conditionals by
naraba. While ba is incompatible with wa-topicalisation, naraba is compatible with
wa-topicalisation. Following my proposal it would be tempting to take ba to
correspond to the head Sub and nara to Force. Obviously further research is
required to bear this out.

(43) a *Mosi sono yoona zassi-wa, (anata-ga) yome-ba,

if that like magazine-top (you-nom) read (conditional)-if
(anata-wa) yasai-ga sukini narimasu.
(you-top) vegetable-nom like become
‘If these magazines, you read, you will come to like vegetables’
b Mosi sono yoona zassi –wa (anata-ga) sukide-nai (conclusive)-naraba,
if that like magazine-top (you-nom) like-not-if
naze (anata-wa) (sorera-o) kai-tuzukerunodesu ka?
why you-top them-acc buy-continue, Q
‘If such magazines, you don’t like, why do you keep buying them?’

8. Conclusion

The paper analyses the semantic and syntactic differences between event-
conditionals and premise-conditionals. An event-conditional structures the event:
it modifies the matrix predicate. A premise-conditional structures the discourse: it
makes manifest a context for the proposition in the associated clause. I have argued
that the conditional clause types differ both in their ‘external syntax’ and in their
‘internal syntax’. The contrast displayed by conditionals generalises to other sub-
ordinating conjunctions such as while, because, when, since, as, so that etc.
Event-conditionals are central adverbial clauses; premise-conditionals are per-
ipheral adverbial clauses. A central adverbial clause is merged with the IP-domain
of the associated matrix clause, while the premise-conditional is merged with the
associated CP. The different level of insertion or the different timing of the merger
determines the semantic/pragmatic differences.
I propose that central adverbial clauses are structurally deficient in that their
CP domain lacks the functional projection ‘Force’, which encodes illocutionary
force, and they also lack TopP and FocusP, both of which are dependent on
Force. While the central adverbial clause is integrated in the speech act of the

16
I warmly thank Hideki Maki for generously and spontaneously providing these data.
# Blackwell Publishing Ltd. 2003
 Conditional Clauses: External and Internal Syntax 337

matrix clause, the peripheral adverbial clause has independent illocutionary force.
Its CP domain is complete. Central adverbial clauses disallow MCP such as
topicalisation and focalisation. Peripheral adverbial clauses allow topicalisation
and focalisation.

Université Charles de Gaulle- Lille III

UMR 8258 Silex du CNRS
Haegeman@commat;univ-lille3.fr

References

Andersson, L.G. 1975: Form and Function of Subordinate Clauses. Ph.D. Diss. Department
of linguistics, University of Goteborg.
Bayer, J. 2001: Asymmetry in emphatic topicalisation. Ms. University of Utrecht.
Bennis, H. 2000: On the interpretation of functional categories. In H. Bennis,
M. Everaert, and E. Reuland (eds), Interface Strategies. KNAW publications.
Bhatt, R. and Yoon, S. 1992: On the composition of Comp and parameters of V-2.
WCCFL, 10, 41–52.
Blakemore, D. 1987: Semantic Constraints on Relevance. Oxford: Blackwell.
Bouma, G., Malouf, R. and Sag, I. 2001: Satisfying constraints on extraction and
adjunction. Natural Language and Linguistic Theory, 19, 1–65.
Chomsky, N. 1995: The Minimalist Program. Cambridge: MIT Press.
Cinque, G. 1999: Adverbs and Functional Heads: A Cross-linguistic Perspective. Oxford:
Oxford University Press.
Close, R.A. 1980: Will in if- clauses. In S. Greenbaum, G. Leech and J. Svartvik (eds),
Studies in English Linguistics. London: Longman.
Culicover, P. 1991: Topicalisation, inversion and complementisers in English. In
D. Delfitto et al. (eds), OTS Working Papers. Going Romance and Beyond. University
of Utrecht.
Dancygier, B. 1987: Conditions and relevance. Ms. Institute of English Studies, University
of Warsaw.
Dancygier, B. 1998: Conditionals and prediction. Cambridge: Cambridge University
Press.
Dancygier, B. and Mioduszewska, E. 1984: Semanto-pragmatic classification of condi-
tionals. Studia Anglica Posnaniensia, xvii, 121–134.
Declerck, R. 1991: Tense in English. London and New York: Routledge.
Declerck, R. and Reed, S. 2001: Conditionals: a Comprehensive Empirical Analysis.
Amsterdam: Mouton de Gruyter.
Epstein, S. 1999: Un-principled syntax: the derivation of syntactic relations. In
S. Epstein and N. Hornstein (eds), Working Minimalism. Cambridge: MIT Press.
317–345.
Green, G. 1976: Main clause phenomena in subordinate clauses. Language, 52, 387–97.
# Blackwell Publishing Ltd. 2003
338 L. Haegeman

Haegeman, L. 1984a: Parasitic gaps and adverbial clauses. Journal of Linguistics, 20, 229–232.
Haegeman, L. 1984b: Pragmatic conditionals in English. Folia Linguistica, 18, 485–502.
Haegeman, L. 1984c: Remarks on adverbial clauses and definite anaphora. Linguistic
Inquiry, 15.
Haegeman, L. 1991: Parenthetical adverbials: the radical orphanage approach. In
S. Chiba, A. Shuki, A. Ogawa, Y. Fuiwara, N. Yamada, O. Koma and T. Yagi (eds),
Aspects of Modern English Linguistics: Papers presented to Masatomo Ukaji on his 60th
Birthday. Tokyo: Kaitakushi, 232–254.
Haegeman, L. 2001a: Anchoring to speaker and the structure of CP. Ms. Université
Charles de Gaulle, Lille III.
Haegeman, L. 2001b: Speculations on adverbial fronting and the left periphery. Ms.
Université Charles de Gaulle, Lille III.
Haegeman, L. and Wekker, H. 1984: The syntax and interpretation of futurate
adverbials in English. Journal of Linguistics, 20, 45–55.
Haik, I. 1984: Indirect binding. Linguistic Inquiry, 15, 185–224.
Hooper, J. and Thompson, S. 1973: On the applicability of root transformations.
Linguistic Inquiry, 4, 465–97.
Hornstein, N. 1984: Logic as Grammar. Cambridge: MIT Press.
Hukari, T. and Levine, R. 1995: Adjunct extraction. Journal of Linguistics, 31, 195–226.
Jespersen, O. 1940: A Modern English Grammar on Historical Principles. Volume 5
London: Allen and Unwin.
Kempson, R. 1975: Presupposition and the Delimitation of Semantics. Cambridge:
Cambridge University Press.
Klima, E.S. 1964: Negation in English. In J. Fodor and J. Katz. (eds), The Structure of
Language. N. Jersey: Prentice Hall, 246–232.
Koizumi, M. 1995: Phrase Structure in Minimalist Syntax. Ph.D. Diss. MIT.
Ladusaw, W. 1980: On the notion affective in the analysis of negative polarity items.
Journal of Linguistic Research, 1, 1–16.
Larson, R. 1988: On the double object construction. Linguistic Inquiry, 19, 335–392.
Larson, R. 1990: Double objects revisited: reply to Jackendoff. Linguistic Inquiry, 21,
589–632.
Leech, G. 1971: Meaning and the English Verb. London: Longmans.
Maki, H., Kaiser, L. and Ochi, M. 1999: Embedded topicalisation in English and
Japanese. Lingua, 109, 1–14.
McCawley, J. 1971: Tense and time reference in English. In C. J. Fillmore and
D. T. Langendoen (eds), Studies in Linguistic Semantics. New York: Holt, Rinehart
and Winston. 96–113.
Nakajima, H. 1996: Complementizer selection. The Linguistic Review, 13, 143–164.
Nissenbaum, J. 2000: Investigations of Covert Phrase Movement. Ph.D. Diss. MIT.
Palmer, F. 1965: A Linguistic Study of the English Verb. London: Longman.
Palmer, F. 1974: The English Verb. London: Longman.
Pollard, C. and Sag, I. 1987: Information Based Syntax and Semantics, CSLI Publications,
Stanford.

# Blackwell Publishing Ltd. 2003

 Conditional Clauses: External and Internal Syntax 339

Pollock, J-Y. 1989: Verb movement, UG and the structure of IP. Linguistic Inquiry, 20,
365–425.
Postal, P. and Ross, J.R. 1971: A problem of adverb preposing. Linguistic Inquiry,
1, 145–6.
Rizzi, L. 1994. Early null subjects and root null subjects. In T. Hoekstra and
B. Schwartz (eds), Language Acquisition Studies in Generative Grammar. John
Amsterdam: Benjamins, 151–177.
Rizzi, L. 1995. Some notes on linguistic theory and language development: the case of
root infinitives. Language Acquisition, 3, 371–393.
Rizzi, L. 1997: The fine structure of the left periphery. Haegeman, L (ed.), Elements of
Grammar. Dordrecht: Kluwer, 289–330.
Rizzi, L. 2001: Locality and Left Periphery. Ms. University of Siena.
Rutherford, W. 1970: Some observations concerning subordinate clauses in English.
Language, 46, 97–115.
Safir, K. 1984: Multiple variable binding. Linguistic Inquiry, 15, 603–38.
Sperber, D. and Wilson, D. 1986: Relevance. Oxford: Blackwell.
Sweetser, E.E. 1984: Semantic Structure and Semantic Change: a Cognitive Linguistic Study
of Modality, Perception, Speech Acts and Logical Relations. Ph. D. Diss. Berkely
University of California.
Thompson, E. 1994: The syntax and semantics of temporal adjunct clauses. Studies in
the Linguistic Sciences, 24.
Van der Auwera, J. 1986: Conditionals and speech acts. In E. Traugott, A. ter Meulen,
J Snitzer Reilly and C.A. Ferguson (eds), On Conditionals. Cambridge: Cambridge
University Press, 197–214.
Verstraete, J.-C. 2002: Interpersonal Grammar and Clause Combining. Ph. D. Diss.
University of Louvain, Belgium.
Wekker, H.C. 1976: The Expression of Future Time in Contemporary British English.
Amsterdam: North Holland.
Wekker, H.C. 1977: Future reference in adverbial clauses. Interlanguage Studies Bulletin,
2, 64–77.
Zandvoort, R.W. 1975: A Handbook of English Grammar. London: Longman.

# Blackwell Publishing Ltd. 2003