Clause Type and Illocutionary Force PDF

10
Clause type and

illocutionary force
Rodney Huddleston
1 Type as a grammatical system of the clause 853

2 Distinctive grammatical properties of the major clause types 855
3 Some semantic and pragmatic preliminaries 858
3.1 Illocutionary force 858
3.2 Indirect speech acts 861
4 Kinds of question 865
4.1 Question as a semantic and as a pragmatic category 865
4.2 Summary classification of questions 867
4.3 Polar questions 868
4.4 Alternative questions 868
4.5 Variable questions 872
4.6 Direction questions 876
4.7 Biased questions 879
4.7.1 Kinds and degrees of bias 880
4.7.2 Declarative questions 881
4.7.3 Negative interrogative questions 883
4.7.4 Positively- and negatively-oriented polarity-sensitive items (some vs any, etc.) 884
4.8 Echo questions 886
4.8.1 The contrast between echo and ordinary questions 887
4.8.2 The grammatical form of variable echo questions 888
4.8.3 The form of polar echo questions 889
4.8.4 Repetition vs clarification echoes 890
5 Interrogative tags and parentheticals 891
5.1 The formation of interrogative tags 892
5.2 The use and interpretation of tags 894
5.3 Parentheticals 895
6 The presuppositions of information questions 897
6.1 Q–A presuppositions of alternative questions 897
6.2 Q–A presuppositions of polar questions 898
6.3 Q–A presuppositions of variable questions 899
6.4 Rejection of Q–A presuppositions 900
6.5 Cancellation of Q–A presuppositions 900
6.6 Secondary presuppositions 901
7 Interrogative words and phrases 902
7.1 Which 902
7.2 Whose 904
851
Downloaded from https:/www.cambridge.org/core. Boston University Theology Library, on 04 Jun 2017 at 01:28:02, subject to the Cambridge Core terms of use,
available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.011
852
7.3 Who and whom 904

7.4 When 905
7.5 Where 905
7.6 Why 906
7.7 How 907
7.8 What 909
7.9 Upward percolation of the interrogative feature: interrogative phrases 910
7.10 Upward percolation of the question variable: the questioned element 912
7.11 Open interrogatives as an unbounded dependency construction 914
7.12 Ambiguities concerning the role of an interrogative phrase in complex clauses 914
7.13 Modification of interrogative words 916
7.14 Complex-intransitive interrogatives: S–P–PC vs PC–P–S 917
8 Exclamatives and exclamations 918
8.1 The syntax of exclamatives 918
8.1.1 Exclamative how 919
8.1.2 Exclamative what 919
8.1.3 Position of the subject 920
8.1.4 Verbless exclamatives 921
8.2 Meaning and use of exclamative main clauses 922
8.3 Non-exclamative exclamations 923
9 Imperatives and directives 924
9.1 Subtypes of imperative clauses 924
9.2 Ordinary imperatives 925
9.2.1 Omissibility of the subject 925
9.2.2 Subject vs vocative in imperatives 927
9.2.3 Imperatives with auxiliary do 928
9.2.4 Imperatives as directives 929
9.2.5 Agentivity in imperatives 931
9.3 1st person inclusive let-imperatives 934
9.3.1 Grammatical properties 934
9.3.2 Use of 1st person inclusive imperatives 936
9.4 Open let-imperatives 936
9.5 Imperatives interpreted as conditionals 937
9.6 Non-imperative directives 939
9.6.1 Interrogatives as directives 939
9.6.2 Declaratives as directives 941
9.6.3 Non-finite and verbless directives 942
9.7 Imperatives with interrogative tags 942
9.8 No subordinate imperative construction 943
10 Minor clause types 944
853
1 Type as a grammatical system of the clause
The five major categories

Clause type is the grammatical system whose five major terms are illustrated in:
[1] i You are generous. [declarative]
ii Are you generous? [closed interrogative]
iii How generous are you? [open interrogative]
iv How generous you are! [exclamative]
v Be generous. [imperative]
Characteristic use and general definitions
Each of the categories is associated with a characteristic use as follows:
[2] clause type characteristic use
i declarative statement
ii closed interrogative closed question
iii open interrogative open question
iv exclamative exclamatory statement
v imperative directive
A closed question is one with a closed set of answers: for example, the answers to [1ii] are
just “Yes” and “No”. By contrast, [1iii] has any number of possible answers, and is therefore
an open question; similarly with Who attended the meeting?, and so on. In [2iv] we have
used ‘exclamatory statement’ rather than the more familiar ‘exclamation’, because an
exclamatory meaning can be added to any of the use categories, but the special syntactic
construction shown in [1iv] is associated just with a particular kind of statement. For
example, the exclamatory command Get the hell out of here or the exclamatory question
What on earth are you doing? belong syntactically with [1v] and [1iii] respectively, not
with [1iv]. ‘Directive’ in [2v] is a cover term for requests, commands, instructions, and
the like; traditional grammars tend to use the term ‘command’, but this is far too narrow
and specific for our purposes if understood in its everyday sense.
The correlation shown in [2] provides the basis for general definitions of the clause
type categories:
[3] Imperative clause is a grammatically distinct class of clause whose members are
characteristically used to issue directives.
And similarly for the others.
854 Chapter 10 Clause type and illocutionary force
Complex relation between form and meaning

We have spoken of ‘characteristic’ use because, as is so often the case, the correlation
between major categories of grammatical form and categories of meaning or use is by
no means one-to-one. Compare, for example:
[4] clause type use
i Passengers are requested to remain seated. declarative directive
ii Would you mind opening the door for me. closed interrogative directive
iii Sleep well. imperative wish
Examples [i] and [ii] illustrate conventional ways of expressing a polite request, a kind of
directive, but syntactically they belong to the same structural class as You are generous and
Are you generous? respectively. And though [iii] belongs syntactically with Be generous it
is not used to tell or ask somebody to do something, but to express a wish. Numerous
further examples of this and other kinds will emerge during the course of the chapter. It is
essential therefore to maintain a sharp conceptual distinction between the grammatical
clause types and the categories of meaning or use – between declarative and statement,
imperative and directive, and so on. The situation is closely parallel to that which obtains
in the area of tense (form) and time (meaning), but whereas it is standard practice for
grammars to distinguish terminologically between tense and time, many use ‘question’
both for form (our ‘interrogative’) and for meaning. Again we emphasise, therefore, that
interrogatives aren’t always used as questions, and not all questions have the syntactic
form of interrogatives.
Clause type in subordinate clauses

Four of the clause type categories apply to subordinate clauses as well as to main clauses:
[5] main subordinate
i It’s a bargain. She says that it’s a bargain. [declarative]
ii Is it a bargain? I wonder if it’s a bargain. [closed interrogative]
iii Which one is a bargain? I know which one is a bargain [open interrogative]
iv What a bargain it is! I realise what a bargain it is. [exclamative]
Imperatives, however, are normally restricted to main clauses.1 Our main focus in this
chapter will be on main clauses, with most of the material on subordinate clause types
deferred till Ch. 11.
The type system applies to the clause, not the sentence

What we refer to here as ‘clause type’ is more often called ‘sentence type’, but given
the way in which we have defined sentence and clause in this grammar (Ch. 2, §1), it
is evident that it is to the clause, not the sentence, that the system applies. In the first
place, the subordinate clauses of [5] are not sentences but are nevertheless classified for
type. And secondly, in sentences with the form of a clause-coordination the coordinated
1
For arguments against treating infinitivals like I told them to be quiet as subordinate imperatives, see §9.8.
Examples like It’s time we were going home, because don’t forget we have to be up early in the morning are of
somewhat marginal grammaticality, and the internal structure here remains like that of a main clause.
§2 Distinctive grammatical properties of the major clause types 855
clauses do not have to be of the same type:

[6] i Come around six, or is that too early for you? [imperative + closed interrog]
ii You can come too, but please bring your lunch. [declarative + imperative]
iii What a fine player she is, and she’s still only ten! [exclamative + declarative]
In [i] the first clause is imperative and the second closed interrogative: the sentence as a
whole cannot be assigned to any of the type categories. And similarly with [ii–iii].
Clause fragments
We include within the category of clause various kinds of verbless construction, such as
open interrogative What about the others? or exclamative What a disaster!
The clause types are mutually exclusive

Clause type is a grammatical system in the sense that no clause can belong to more than
one of the categories: they are mutually exclusive. There can be ambiguity: How many
problems remain (considered in abstraction from punctuation and prosody) can be an
open interrogative (“What is the number of problems that remain?”) or an exclamative
(“What a lot of problems remain!”), but any particular instance of it will be one or the
other, not simultaneously both.
Echo question not a clause type
Echo questions are illustrated in:
[7] stimulus echo question
i a. Give it to Angela. b. Give it to who?
ii a. Did you use a macro? b. Did I use a what?
Suppose you say [ia] and I don’t quite catch the name: I might respond with [ib] to ask
you to repeat it, and similarly in [ii]. The [b] examples are known as echo questions:
they echo the stimulus, what has just been said, with a view to questioning some aspect
of it. We examine this construction in §4.8, but we mention it in the present context to
develop what has just been said about the concept of system. It is evident from [7] that
the echo construction is not mutually exclusive with the clause type categories. In [i]
the echo feature is superimposed on an imperative and in [ii] on a closed interrogative:
from a syntactic point of view, echo question is therefore a different kind of category
from the clause types, not a sixth term on the same dimension. With respect to clause
type, then, [ia] and [ib] are both imperatives, [iia] and [iib] both closed interrogatives;
in each pair [b] differs from [a] by virtue of being an echo question. This reinforces the
need to distinguish carefully between the clause types and the categories of meaning/use:
[ib], for example, is a question, but it is not syntactically interrogative.
2 Distinctive grammatical properties of the major clause types
Declarative is the default clause type: a clause is declarative if it lacks the special properties
that define the other types. In this section, therefore, we outline the distinctive properties
of the other four major types with respect to main clauses.
Closed interrogatives
[1] Closed interrogatives have subject–auxiliary inversion triggered by the clause
type, and hence are always tensed.
[2] declarative closed interrogative
i a. It is true. b. Is it true?
ii a. They saw her. b. Did they see her?
Subject–auxiliary inversion is a do-support construction (Ch. 3, §2.1), so empty do is
required if there would not otherwise be an auxiliary verb, as in [ii]. Subject–auxiliary
inversion is not limited to closed interrogatives but other main clauses in which it occurs,
with one exception, all have the inversion triggered by the placement of a non-subject
element in initial position. In declarative None of them did he consider satisfactory, for
example, the inversion is triggered by the initial negative. The one exception (in main
clauses) is the optative may construction of May you be forgiven! 2
Open interrogatives
[3] i Open interrogatives contain an interrogative phrase based on one of the inter-
rogative words who, whom, whose, which, what, when, where, how, etc.
ii A non-subject interrogative phrase is usually fronted, and this triggers subject–
auxiliary inversion.
iii Open interrogatives are usually tensed, but can also be infinitival.
iv Open interrogatives can be reduced to just the interrogative phrase.
[4] i Who broke the window? [interrogative phrase as subject]
ii Which one did he choose? [non-subject interrogative phrase with inversion]
iii So you told him what, exactly? [non-fronted interrogative phrase]
iv Why make such a fuss? [infinitival]
v Which one? [reduction to interrogative phrase]
In [4i] the interrogative phrase is subject and occupies the same position as the subject of
a declarative (Kim broke the window). In [4ii] the interrogative phrase is a non-subject in
prenuclear position: we say that the interrogative phrase has been fronted. As usual, the
process terminology is not to be interpreted literally: it is merely a shorthand way of saying
that the interrogative phrase occupies front position rather than the post-verbal position
of the corresponding element in a syntactically more basic clause (cf. He chose this one).
In main clauses fronting of the interrogative phrase always triggers inversion, whereas in
subordinate clauses it normally doesn’t. (The qualification ‘normally’ is needed because
inversion is possible in subordinate clauses under restrictive conditions discusssed in
Ch. 11, §5.3.2: %She asked how could she help us.) Example [iii] shows that fronting of a
non-subject interrogative phrase is not obligatory: the interrogative phrase here remains
in situ, i.e. it occupies the same place as the corresponding element in a declarative
clause. Example [iv] is a bare infinitival; to-infinitivals are also possible, as in How to
explain his attitude? Reduced clauses like [v] are naturally heavily dependent on context
for their interpretation.
2
Untriggered non-interrogative inversion occurs also in subordinate clauses functioning as conditional
adjunct: Had I known earlier, I’d have done something about it (see Ch. 11, §4.7).
§2 Distinctive grammatical properties of the major clause types 857
Exclamatives
[5] i Exclamatives contain an initial exclamative phrase, based on one or other of the
two exclamative words what and how.
ii They may be reduced to just a predicative exclamative phrase; otherwise they are
always tensed.
iii They usually have subject + predicator order, but subject postposing and
subject–auxiliary inversion are also possible.
[6] i What a disaster it was!
ii How great would be their embarrassment if the error were detected!
iii How happy would he be if he could see her once more!
iv What a disaster!
Examples [6i–iii] have fronting of a non-subject exclamative phrase; in [ii] this is accom-
panied by postposing of the subject their embarrassment and in [iii] by subject–auxiliary
inversion. In [iv] the clause is reduced to the exclamative phrase, understood predicatively
(“What a disaster it was!”).
Imperatives
[7] i Imperatives are normally restricted to main clauses.
ii A 2nd person subject is omissible.
iii The verb is in the plain form.
iv In verbal negation, emphatic polarity, and code, supportive do is required even
in combination with be.
v Verbal negatives with you as subject usually have the order don’t + you.
The examples in [8] show how these properties distinguish imperatives from
declaratives:
[8] declarative imperative
i a. You look after yourself. b. (You) look after yourself.
ii a. You are very tactful. b. Be very tactful.
iii a. Everybody stands up. b. Everybody stand up.
iv a. You aren’t late. b. Don’t be late.
v a. You don’t worry about it. b. Don’t you worry about it.
You look after yourself is ambiguous between declarative and imperative (it could be
used as a statement about your behaviour or as a directive), but Look after yourself is
unambiguously imperative (having only the directive interpretation). Examples [8ii–iii]
have a difference in verb-form. Present tense are in [iia] contrasts with plain form be: be
is the only verb lexeme that does not have syncretism between the plain form and one
of the present tense forms. The difference in [iii] is of greater generality: here the plain
form of the imperative contrasts with the 3rd person singular present tense form of
the declarative, whatever the lexeme involved. In [8iv] do isn’t needed in the declarative
because be is an auxiliary verb, but it is nevertheless required in the imperative (cf. ∗Be
not late). Finally, in [8v] we have a difference in the position of the subject. Example
[va] has the default S–P order; it is a statement, one that might well be followed by
the question tag do you? In imperative [vb], a directive, the subject follows don’t.
Closed and open interrogatives: subclasses of a larger class

or distinct primary classes?
The terms ‘closed interrogative’ and ‘open interrogative’ suggest that they are subclasses
of ‘interrogative’. Yet what they have in common is much more a matter of meaning
than of syntax: they both characteristically express questions. From a syntactic point of
view, they are in fact strikingly different. The most important property of open interrog-
atives is the presence of an interrogative phrase, based on the special set of interrogative
words – they can in fact be reduced to just an interrogative phrase, as seen in [4v]. The
distinctive property of closed interrogatives (in main clauses) is subject–auxiliary inversion.
Inversion is found in the open interrogative as well as the closed, but only as a secondary
feature, triggered by the fronting of a non-subject interrogative phrase. As a result, closed
Did she win the race? and Who won the race? share no syntactic feature distinguishing them
from declarative She won the race. Note, moreover, that inversion is also found as a sec-
ondary feature in a variety of other constructions too, such as negatives like Not once did she
smile.
It is for these reasons that we have treated closed and open interrogatives as each on a
par with declarative, exclamative, and imperative within the syntactic system of clause type. It
would not be helpful, however, to coin new terms for them without a shared component – and
we will make use of ‘interrogative’ as a cover term, generally when the focus is on the relation
with question rather than on the syntactic structure.
3 Some semantic and pragmatic preliminaries
Before looking systematically at the relation between the clause types and their meaning
or use, we need to clarify some of the concepts we will be using in talking of meaning in
this area.
3.1 Illocutionary force

Statement and directive are in the first instance pragmatic categories: we are concerned
with the way the speaker is using the clause when uttering it in a particular context. A
more specific term for this aspect of pragmatic meaning is illocutionary force. If, for
example, I utter the clause Tom has arrived with the intention of thereby committing
myself to the truth of the proposition “Tom has arrived”, I have uttered it with the
illocutionary force of a statement – or, to put it slightly differently, I have performed the
illocutionary act of making a statement. If I say Sit down with the intention of telling
you to sit down, my utterance has the illocutionary force of a directive – or, again, I have
performed the illocutionary act of issuing a directive. And so on. Question is commonly
used in the same kind of way: asking a question is a kind of illocutionary act contrasting
with making a statement or issuing a directive.
More specific kinds of illocutionary force

Statement, directive, and question are very general categories of illocutionary force, but
there are in addition innumerable more specific illocutionary categories. Some of these
can be regarded as simply special cases of the more general categories. For example, Bring
§ 3.1 Illocutionary force 859
the water to the boil might be said with the force of a command, a request, advice, an
instruction (e.g. in a recipe), all of which can be subsumed under the broader category
of directive, for they all count as attempts to get you to do something.
In other cases, the specific illocutionary force is different in kind from the three general
ones:
[1] I promise to return the key tomorrow.
The natural use of this is to make a promise, and a promise is different in kind from
a statement. In making a statement I commit myself to the truth of some proposition,
whereas in making a promise I commit myself to doing something – in the case of [1],
to returning the key tomorrow.
Primary and secondary force

Strictly speaking, a natural utterance of [1] would be both a statement and a promise,
though the promise is of course more important, more salient than the statement. We will
speak of the promise force as primary and the statement force as secondary. Making
the statement can be regarded as simply the means of making the promise. I make a
promise by stating that I do, and the statement is true simply by virtue of my uttering
the clause with the intention of making a promise. The greater salience of the promise
over the statement is reflected in the way the utterance of [1] would most naturally be
reported, in comparison with the way in which an utterance with primary statement
force would be:
[2] i a. I returned the key yesterday. [statement]
b. You said you returned the key yesterday. [report of statement]
ii a. I promise to return the key tomorrow. (=[1]) [promise]
b. You promised to return the key tomorrow. [report of promise]
You said you promised to return the key tomorrow, although possible, is much less natural
than [iib] precisely because it reports [iia] as a statement, whereas [iib] reports it as a
promise.
Notice, then, that whereas the clause types are mutually exclusive, the illocutionary
categories are not: it is possible for an utterance to belong simultaneously to more
than one such category. This applies, indeed, also to the general categories: I order
you to leave would naturally be both a directive (the primary force) and a statement
(secondary).
The performative use of verbs

The two illocutionary forces of [1] are expressed by quite different linguistic devices. The
statement force derives from the declarative clause type, whereas the promise force derives
from the presence of the verb promise itself. Promise belongs to the class of illocutionary
verbs, verbs that denote illocutionary acts, and in [1] it is used performatively, i.e. to
effect the performance of the illocutionary act it denotes. The performative use of promise
in [1] may be contrasted with its non-performative use in examples like [2iib]. The latter
is not a promise but simply a report of a promise, hence just a statement. The first device,
declarative clause type, is a matter of grammar, and there is accordingly only a very small
number of possible contrasts; the second device, the performative use of promise, is
primarily a lexical matter, which allows for a large number of contrasts. In [3] we give a
small sample of verbs that can be used performatively:

[3] admit advise apologise ask beg
bet claim command commend concede
congratulate entreat estimate name order
postulate promise repudiate resign suggest
swear thank urge warn welcome
There are also expressions consisting of verb + dependent: declare . . . open (I declare the
meeting open), give one’s word, and so on. For convenience we will refer to such utterances
as [1]/[2iia] as performatives.3
Perlocutionary effect
Illocutionary force contrasts with perlocutionary effect, the effect the utterance has on
you, the addressee. If I say Tom has arrived with the illocutionary force of a statement, the
default perlocutionary effect is that you will accept it as true. But of course statements do not
invariably have this effect: you may know or believe me to be mistaken. Similarly, if I say Sit
down with the illocutionary force of a directive, the default perlocutionary effect will be that
you comply by sitting down; but again this is not the only possible result. Typically, then, an
illocutionary force is associated with a particular perlocutionary effect which the speaker is
aiming to achieve, but failure to achieve this effect does not normally deprive the utterance
of its illocutionary force: a statement is still a statement even if it is not accepted as true, a
directive is still a directive even if it is not complied with, and so on.
Verbs which denote illocutionary acts can normally be used performatively, like promise
in [1]. Those – such as persuade, convince, annoy, intimidate, impress – which denote perlocu-
tionary acts cannot similarly be used to perform those acts. I can warn you (an illocutionary
act) that the car is unroadworthy by saying [4i], but I cannot persuade you (a perlocutionary
act) that it is unroadworthy by saying [4ii]:
[4] i I warn you that the car is unroadworthy. [performative]
ii I persuade you that the car is unroadworthy. [non-performative]
(The second is indeed pragmatically unlikely: it needs some such continuation as and yet you
buy it nevertheless!) Similarly, there are differences in the way I might ask you to clarify your
illocutionary and perlocutionary intentions. Compare

[5] i a. Is that a threat or a promise?
[illocutionary]
b. Are you asking me or telling me?

ii a. Is that intended to intimidate me?
[perlocutionary]
b. Are you trying to annoy me or to amuse me?
When questioning the perlocutionary intention, one typically needs to include some such
verb as intend or try, but this is generally not necessary when questioning the illocutionary
intention. Saying I’ll be back at six with the intention of making a promise is sufficient for
the utterance to be a promise, but telling a joke with the intention of amusing the addressee
is not sufficient to achieve that goal.
3
Clauses like I promise to return the key and I order you to leave are ambiguous, having also less salient
interpretations in which they are statements about my habitual behaviour (“I habitually promise to return
the key / order you to leave”): in this interpretation they are not performatives since they do not themselves
constitute a promise or order. The tomorrow in [2iia] makes the habitual reading even less salient, but it is
still possible in principle.
§ 3.2 Indirect speech acts 861
Propositional and non-propositional components of meaning

The propositional content of a sentence is that part of its meaning that determines
what propositions it can be used to express; clause type, however, contributes to non-
propositional meaning (see Ch. 1, §5.2). Consider the relation between a declarative
and its closed interrogative counterpart, as in:
[6] declarative closed interrogative
i a. Kim is in Paris. b. Is Kim in Paris?
ii a. Pat saw them. b. Did Pat see them?
In each pair, [a] and [b] are partly alike and partly different in both form and meaning. In
form, the closed interrogative contains the same elements as the declarative (sometimes
with the addition of do, as in [ii]) but with a different order of subject and verb. As for
meaning, what they have in common is that they have the same propositional content:
both express the proposition “Kim is in Paris” or “Pat saw them”. They differ in the
non-propositional component, more specifically, in their illocutionary meaning. In a
normal use of [a] the proposition is asserted, whereas in [b] it is questioned. In both
[i] and [ii] the illocutionary force is separate from the propositional content of the
utterance: when I use [ia] to make a statement I do not express the proposition that I
am making a statement – I simply make it by uttering a declarative with the appropriate
intention. Similarly, when I ask a question by means of [ib] I do not say that I am asking
a question. Note, then, that the term ‘express’ is neutral as to illocutionary force. And
the terms ‘true’ and ‘false’ can be applied to propositions or to statements, but not of
course to questions.
A unique feature of performatives like [1], I promise to return the key tomorrow, is that
here the (primary) illocutionary force is identified in the propositional content of the
utterance. Thus [1] itself, for example, expresses the proposition “I promise to return
the key tomorrow”. As a result of this feature of performatives, the primary illocutionary
force of the utterance is more explicit and precisely specified in such utterances than it
normally is elsewhere.
Other non-propositional markers of illocutionary force
Clause type and the performative use of illocutionary verbs are not the only linguistic
devices for indicating illocutionary force. Intonation plays an important role too, as we
shall see, and there are also particular words, such as please, which serve this purpose. But
these further devices are like clause type in that they do not contribute to the propositional
content of the utterance. When I say, for example, I’d like a cup of tea, please, the please
serves to indicate that I am making a request, that I am asking for a cup of tea, but I do
not express the proposition that I am doing so. Please is quite irrelevant to the truth or
falsity of the utterance, and hence does not express any part of its propositional content.
3.2 Indirect speech acts

Illocutionary force is very often conveyed indirectly rather than directly. Consider:
[7] Do you know what time it is?
A likely context for this (not the only possible one of course) is where I don’t know
the time, want to know the time, and believe you may well be able to tell me. In this
context it would indirectly convey “What time is it?” This is why it would be thoroughly
unco-operative in such a context for you to respond merely with Yes. Yes would answer
the question that is actually asked, but not the one that I in fact want to have answered.
Another plausible context for [7] is where it is addressed to a child (by a parent, say)
when it is known to be past the child’s bedtime: here my intention may well be to convey
a directive to go to bed.
In either contextualisation, I perform two illocutionary acts simultaneously, one
directly (a question as to whether you know what time it is), and one indirectly (a
question as to what time it is, or a directive to go to bed). We will follow the established
practice of referring to indirect illocutionary acts as indirect speech acts (with the
understanding that the term covers writing as well as speech). Commonly, the direct act
is obviously less important than the indirect one – as when the interest of the question
whether you know the time is simply that if you do you will be able to answer the question
that I really do want an answer to. There is an analogy here with performatives, and we
will again, where appropriate, apply the terms ‘primary’ and ‘secondary’ to the different
acts or forces. Thus just as in the performative I promise to return the key tomorrow the
promise is primary and the statement secondary, so in the first contextualisation of [7]
the question about the time is primary and that about your knowledge secondary. The
difference is that in the performative case the primary act is direct, whereas in such cases
as [7] it is not.
Definition of indirect speech act

An indirect speech act is one where (a) the propositional content actually expressed
differs from that which the speaker intends to convey with some illocutionary force, or
(b) where the illocutionary force is different from that normally conveyed by the clause
type concerned.
Most cases are covered by condition (a). The propositional content expressed in [7],
for example, is “You know what time it is”, whereas the propositional content that I
intend to convey is “What time is it?” (with question force) or “You go to bed” (with
directive force). Case (b) is illustrated by [4iii] of §1, Sleep well. At the direct level it is a
directive (the force characteristically associated with imperative clause type), but since
sleeping well is not something that we normally regard as being under our control it will
generally have the indirect force of a wish.4
Degrees of indirectness
There are varying degrees of indirectness, depending on how different the two propo-
sitional contents are. The first suggested contextualisation of [7], for example, is less
indirect than the second because the propositional content of the conveyed “What time
4
The sense of ‘indirect’ introduced in this section is quite different from the one it has in traditional grammar
in such expressions as ‘indirect question.’ A traditional indirect question, such as the underlined clause in
She asked who had done it, is in our terminology a subordinate interrogative, whereas an indirect question in
the speech-act sense might be I’d be interested to hear your view, when used to convey “What is your view?”
As it happens, both senses are applicable in [7], but to different parts of it. The subordinate clause what
time it is is an indirect question in the traditional sense (it is a subordinate interrogative clause), whereas the
whole utterance is an indirect question in the speech act sense (in the use where it conveys “What time is it?”).
We will use ‘indirect’ solely in the speech-act sense.
§ 3.2 Indirect speech acts 863
is it?” is included as part of that which is actually expressed, whereas “You go to bed” is
not. Intuitively (for we are not suggesting that the degree of indirectness can be precisely
calculated), the following are less indirect again, though they still qualify as indirect
speech acts:
[8] i I should like to order two copies of the Penguin edition of Plato’s ‘Republic’.
ii May I remind you that you agreed to pay for the drinks?
In the context of a letter to a bookshop, the writer of [i] will be taken to have performed
the illocutionary act of ordering the goods, but the act is performed indirectly because the
propositional content expressed is “I should like to order . . .”, not “I (hereby) order . . .”.
The inference from “I should like to order” to “I order” is a very easy one to make in this
context, for the wish to order can be fulfilled instantaneously simply by writing the letter
(and perhaps enclosing payment). Nevertheless, “I should like to order” and “I order”
are obviously not propositionally equivalent, and it is easy to imagine other contexts
where the inference would not go through – e.g. in a conversation where the speaker
adds: but in my present financial plight I can’t afford to do so. Example [ii] conveys “I
remind you that you agreed to pay for the drinks”, but again that is not the same as the
propositional content actually expressed. The question concerning permission is here
vacuous since merely mentioning that you agreed to pay for the drinks itself reminds
you of that fact, but this does not alter the fact that there is a difference between the
propositional content expressed and that which I wish to convey.
The pervasive nature of indirect speech acts

Indirect speech acts are an immensely pervasive phenomenon. Some kinds of illocu-
tionary act are more often performed indirectly than directly, either in general or in a
certain range of contexts.
Requests
Take first the case of requests, not in general but in a context where speaker and
addressee are social equals yet not closely intimate. Here a request is much less likely
to be made directly than indirectly. Instead of the direct Please open the window, for
example, I am likely to use one of the indirect directives in [9] or something along
similar lines:
[9] i Can/Could you (please) open the window.
ii Will/Would you (please) open the window.
iii Would you be good enough to open the window (please).
iv Would you mind opening the window (please)?
v Would you like to open the window (please)?
vi I wonder if I might trouble you to open the window?
(There is some variation in the punctuation of interrogatives such as [iv], with a full
stop reflecting the primary directive force, a question mark the secondary question
force.)
Job applications
As a second example, consider the more specialised illocutionary act of applying for a job,
an act normally performed in writing. Some of the formulations used for this purpose
are illustrated in [10], where those in [i] are direct, the others indirect:
[10] i a. I hereby apply for the position of Lecturer in Philosophy advertised in ‘The
Australian’ of 30 November.
b. I apply for the position . . .
c. This is an application for . . .
ii a. I would/should like to apply . . .
b. I wish to apply / make application . . .
c. I am writing to apply . . .
d. I would/should like to be considered for . . .
e. I would/should be grateful if you would consider me for . . .
f. Please consider this letter as my formal application for . . .
g. I beg/wish to offer myself as a candidate for . . .
h. The purpose of this letter is to express my interest in securing . . .
i. I am very glad to have this opportunity to apply . . .
Again, only a small minority of applications are performed directly, and there are innu-
merable variations on the indirect formulations exemplified in [ii].
Non-propositional markers of indirect force

We have explained the concept of indirect speech act primarily by reference to proposi-
tional content; non-propositional components may, however, relate to the illocutionary
act which is conveyed indirectly rather than the one which is directly expressed. This is
illustrated by the please of [9], Can you please open the window, etc., or of the earlier,
I’d like a cup of tea, please. The latter is an indirect speech act in that the propositional
content expressed (“I’d like a cup of tea”) differs from that conveyed with directive force
(“you give me a cup of tea”), but the please serves to signal this indirect directive force
(marking it, more specifically, as a request): it does not relate to the direct statement.
Similarly in [9] please works at the indirect level of request, not the direct level of ques-
tion. The distinction between direct and indirect is therefore not to be identified with
that between explicit and inexplicit: the please explicitly marks the above examples as
requests, but they are still indirect because of the discrepancy between the propositional
content expressed and that implied.
Prosody and punctuation commonly serve as markers of indirect force:
[11] i Could you turn your radio down a little.
ii Isn’t she fantastic!
At the direct level these are questions, but I am unlikely to use them with question as
the primary force: I would generally be indirectly conveying “Turn your radio down a
little” (directive) and “How fantastic she is!” (exclamatory statement). In this use they
would typically have falling intonation, rather than the rising intonation that is the most
characteristic prosodic accompaniment of closed questions, and they are very often not
punctuated with a question mark.
Such markers of the indirect force have the effect of increasing the difference in salience
between the indirect speech act and the direct one, pushing the latter further into the
background. In such an example as Boy, am I ever hungry! the combination of falling
intonation with the non-propositional elements boy and ever causes the direct question
force to be completely overshadowed by the indirect exclamatory statement force.
§4 Kinds of question 865
Idiomatic forms of indirect speech acts

As will be evident from the examples given, certain forms of expression are idiomatically
or conventionally used in the performance of indirect speech acts. A clear distinction
can be seen in such a pair as:
[12] a. Can you turn the light on. b. Are you able to turn the light on?
The construction with can is a much more frequent and natural way of making an
indirect request than that with be able. It is certainly possible to make indirect requests
with be able, but the degree of backgrounding of the direct inquiry force is significantly
less than it is with can. With can, the inquiry force is commonly vacuous, in that the
answer is self-evidently Yes, but this is not normal for be able. Thus in a context where
it is obvious that you can turn the light on, [12a] is appropriate, or idiomatic, whereas
[b] is not. Version [b] needs a context where there is genuine doubt as to your ability
to turn the light on (e.g. one where you are carrying some shopping). This difference
is reflected in the fact that the request marker please, which backgrounds the question
force, would be very much more naturally inserted before turn in [a] than in [b]: Can
you please turn the light on, but hardly Are you able to please turn the light on.
A similar, probably sharper, distinction is seen in
[13] a. Have a good match. b. Win the match.
Like the earlier Sleep well, [a] is likely to be used as an indirect wish, whereas it is hardly
possible to convey a wish by means of [b]. The range of imperatives conventionally used
as wishes is very limited: Sleep well, Get well soon, Have a good . . . , Enjoy . . . , but not
many more.
To say that [12a] is an idiomatic way of making a request is not to say that it, or just the
initial part, can you, is an idiom. An idiom (such as kick the bucket with the sense “die”) is
an expression whose meaning is not systematically derivable from its parts, but the request
meaning conveyed by [12a] is derivable from the meanings of can, you, and the remainder of
the clause together with the inference from an inquiry about your ability to do something to
a request that you do it.
4 Kinds of question
4.1 Question as a semantic and as a pragmatic category

Semantic questions and their answers
The term ‘question’ is commonly used at both the semantic and pragmatic levels. At
the semantic level, a question is distinguished by the fact that it defines a set of logically
possible answers:
[1] question answers
i a. Have you seen it? b. I have seen it. I haven’t seen it.
ii a. Who broke it? b. I broke it. Kim broke it. The priest broke it.
One of her children broke it . . .
Instead of saying I have seen it in answer to [ia], I might say Yes or I have or Yes, I have
or Yes, I’ve seen it, and so on. Although these are different in form they are equivalent,
and we will regard them as (expressing) the same answer. Similarly for the negatives:
I have not seen it; I have not; I haven’t; No; No, I haven’t. These all count as the same
answer. It is in this sense of the term that we can say that [ia] defines a closed set of
just two possible answers. Questions like [iia], on the other hand, define in principle an
open set of answers: there are indefinitely many others besides those given in [iib]. It
was this distinction that provided the basis for general definitions of closed and open
interrogatives.
The distinction between answer and response

Answer is to be distinguished from response, which is a purely pragmatic concept. If
you ask question [1ia], Have you seen it?, I could give any of the following as response,
or of course indefinitely many others:
[2] i No. I have.
ii I’m not sure. I can’t remember. Possibly. Does it matter?
iii I’ve already told you that I have. It’s on your desk. I saw it yesterday.
The responses in [i] are answers, but the others are not. In [ii] I avoid giving an answer –
whether on the grounds of insufficient knowledge or for some other reason. The re-
sponses in [iii] implicate or entail the answer Yes, but they are not logically equivalent to
Yes: they are not themselves answers. With It’s on your desk I interpret your question as
indirectly asking “Where is it?”, and answer that – Have you seen my pen?, for example,
is a conventional way of indirectly conveying “Where’s my pen?” The final response in
[iii], I saw it yesterday, is not an answer because it contains extra information not called
for in the question.
It is clear, then, that for a wide range of reasons one very often responds to a question
in some other way than by giving an answer. And such a response will sometimes contain
less information than an answer would, and sometimes more.
‘The answer’ and ‘the right answer’

A semantic question defines a set of answers, but commonly one speaks of the answer
to a question. Unless there are special features of the context indicating otherwise, the
expression the answer is understood to mean “the right answer”. Usually the right answer
is the one that is true, but we will see below (§4.6) that there is a kind of question where
‘right’ cannot have this interpretation.
Pragmatic questions
Inquiry
The pragmatic concept of question is an illocutionary category. Prototypically, a ques-
tion in this sense is an inquiry. To make a (genuine) inquiry is to ask a question to
which one does not know the answer with the aim of obtaining the answer from the
addressee. An inquiry can be thought of as effectively a kind of a directive – a direc-
tive (usually a request) to the addressee to supply the answer. The directive force is
indirect, however, since the propositional content of the implied directive (“Tell me
the answer to the question . . .”) is not the same as that which is actually expressed. As
with the indirect directives discussed in §3.2, the request force can be signalled explic-
itly in the non-propositional component by the marker please, as in What time is it,
please?
§ 4.2 Summary classification of questions 867
Not all questions are inquiries

The category of question is much broader than that of inquiry. Consider, for example:
[3] i A: Ed’s coming round tonight. B: Is he? I didn’t know he was still in London.
ii What will become of her, I wonder?
iii What were the names of Henry VIII’s six wives?
iv How can this problem be overcome? I suggest that the first step is . . .
Example [i] illustrates the case where a question is used to indicate surprised or interested
acknowledgement of new information. B’s Is he? is not an inquiry: it doesn’t seek to find
out the answer, for A has just provided it, and B is not challenging what A has said. In
[ii] I am wondering, not inquiring – probably not asking for an answer (much less ‘the’
answer). Question [iii] might be used in a quiz or exam: in this case it’s not an inquiry
since presumably I already know the answer, my aim being to test whether you do. And
[iv] is intended as an expository question. Instead of asking you for the answer, I am
directing your attention to a question whose answer I’m about to give you. Other cases
of questions that are not inquiries include indirect speech acts like Could you turn your
radio down a little or Isn’t she fantastic! ([11] of §3), where the question force is secondary
and very much backgrounded.
In comparison with a statement, a question on its own is informationally incomplete:
it needs the answer to complete it. In an utterance with question as its primary force, I
draw attention to this need for a completing answer. What we are calling an inquiry is
then the special, but most common, case where I ask you to provide this answer.
4.2 Summary classification of questions

Questions can be classified in numerous different ways. In the following sections we will
examine distinctions on the four dimensions shown in [4] where the first distinguishes
three kinds of question, the others two each:
[4] i polar alternative variable
Is it breathing? Is it alive or dead? Why isn’t it moving?
ii information direction
What time is it? Shall I put some music on?
iii neutral biased
Have you read it? Haven’t you read it yet?
iv ordinary (non-echo) echo
What’s he going to do? He’s going to what?
Dimension [i] is based on the way the question defines the set of answers. Polar and
alternative make up the class we have called closed question, and both are characteristi-
cally expressed by closed interrogatives. Variable questions are open questions, and are
expressed by open interrogatives. We examine polar, alternative, and variable questions
in turn in the next three sections.
For the other three dimensions the category in the left column of [4] can be regarded
as the default, and §§4.6–8 will therefore focus respectively on direction questions (where
the answers have the force of directives, not statements), biased questions (where the
speaker is biased in favour of one answer over another) and echo questions (which seek
repetition or clarification of what has just been said).
In addition, we take the view that while intonation may mark a question it does
not mark interrogative clause type, and hence with respect to the syntactic form of
(non-echo) polar questions we distinguish:
[5] interrogative question declarative question
Are you ready? You’re ready?
4.3 Polar questions

Answers to polar questions
A polar question has as answers a pair of polar opposites, positive and negative. The
answers to Is it ready? are It is ready and It is not ready (or equivalently Yes and No, or Yes,
it is and No, it’s not, and so on). The propositional content of one answer is expressed in
the question itself, and that of the other is obtained by reversing the polarity.5
Usually it is the positive that is expressed in the question, but it can also be the negative,
as in the biased question Isn’t it ready?
Choice between Yes and No determined by answer
Yes and no are used in positive and negative answers respectively: the choice between
them is determined by the polarity of the answer, with the polarity of the question
being irrelevant. Thus Yes, it is and No, it’s not are answers to both Is it ready? and Isn’t it
ready? Yes on its own, however, is relatively unlikely to be used as a response to a negative
question.
The form of polar questions

Polar questions prototypically have the form of a closed interrogative clause, as in Is it
breathing? in [4i]. They do not always have this form, however; other possibilities are
shown in:
[6] i Your aim that evening, then, was to go to the discotheque?
ii So you went to the party but your brother stayed at home?
iii Another cup of tea?
Example [i] is what we are calling a declarative question: it has declarative not interrog-
ative syntax, with the question meaning normally signalled by rising intonation or the
punctuation; see §4.7.2 for further discussion. Example [ii] has the form of a coordina-
tion of declarative clauses: there are two clauses, but it is a single question. Finally, [iii]
is a clause fragment.
4.4 Alternative questions

Answers to alternative questions
Alternative questions have as answers a set of alternatives given in the question itself.
For example, the answers to Is it right or wrong? are It’s right and It’s wrong, which are
derivable directly from the question. This example contains two alternatives, but there
may be more: e.g. three in Would you like to meet in the morning, the afternoon, or the
evening?
5
Other terms for polar question include ‘yes/no question’, ‘general question’, ‘total question’, ‘nexus-question’.
§ 4.4 Alternative questions 869
The propositional content of an alternative question is, or is logically equivalent to, a dis-
junction of propositions, disjunction being the relation expressed by or (see Ch. 15, §2.2.1).
Each of these propositions gives the content of one of the answers. The propositional content
of Is it alive or dead?, for example, is “It is alive or dead”, which is logically equivalent to “It
is alive or it is dead”.
The form of alternative questions

The essential feature of alternative questions is the coordinator or which relates the
alternatives. The or-coordination is normally prosodically marked by a rise on the first
coordinate and a fall on the final one, as indicated in [7], where we put after a word
or phrase that is uttered with rising pitch and after a word or phrase that is uttered
with falling pitch:
[7] i Is it a boyor a girl ? [closed interrogative]
ii Is it genuineor is it a hoax? [coordination of closed interrogatives]
iii You’re staying here, or coming with us? [declarative]
iv Teaor coffee? [clause fragment]
With multiple coordination the intermediate coordinates take rising intonation, like the
first: Would you like orange juice, lemonade , or coke?
Alternative questions usually have closed interrogative syntax. In [7i] the or-coordi-
nation is within the clause, whereas in [ii] it is between clauses. In the latter case, then, we
have two interrogative clauses but a single question. It is also possible for an alternative
question to have the form of a declarative, as in [iii], or of a clause fragment, as in [iv].
Prosodically marked declaratives, however, are much less readily used for alternative
questions than for polar ones; this is no doubt because questions with declarative form
are biased, and alternative questions tend to be neutral.6
Or in alternative and polar questions

The coordinator or is an essential component of an alternative question, but it may also
occur incidentally in a polar question: Will I be able to get some tea or coffee at the bus
station? Here the answers are Yes, you will and No, you won’t: I’m not asking which drink
is available but whether or not I’ll be able to get one or other of the drinks.
In writing there will often be ambiguity between an alternative question and a polar
question that happens to contain an or-coordination, but the two cases are distinguished
in speech by the intonation. An alternative question, we have noted, has a rise on the first
coordinate and a fall on the last: a polar question will not distinguish the coordinates in
this way but will normally have a rising pitch on the last:
[8] i Are you free on Tuesday or Wednesday? [alternative]
ii Are you free on Tuesday or Wednesday? [polar]
The answers to the alternative question [i] are I am free on Tuesday and I am free on
Wednesday. The polar question [ii] is equivalent to Are you free on Tuesday or Wednesday,
or not?, its answers being Yes, I am free on Tuesday or Wednesday and No, I’m not free
on Tuesday or Wednesday. It should again be borne in mind, however, that while the
answers to the alternative and polar questions are sharply distinct, the responses may
6
A special case is where the alternatives are identical: Is it hot, or is it hot? This serves as an indirect emphatic
statement, “It is remarkably hot”.
be less so. A co-operative addressee might respond to the polar question with Yes, I’m
free on Tuesday, giving more specific information than is actually asked for.
The ambiguity which is found in the written form Are you free on Tuesday or Wednesday? is
seen also when the coordination is between clauses:
[9] Have you moved or are you about to move?
The answers to the alternative question are I have moved and I am about to move, and those
to the polar question are Yes, I have moved or I am about to move (unlikely as a response: you
would generally give more specific information) and No, I have not moved nor am I about to
move. The example is taken from a bank statement, where the continuation – If so, please call
us on the number below – makes clear that the polar interpretation is intended, but it is equally
easy to imagine contexts where the alternative one applies. Again, the ambiguity would be
resolved in speech by the intonation.
One grammatical difference between the two kinds of question is that or cannot be
paired with either when it is the marker of an alternative question. Are you free on either
Tuesday or Wednesday?, for example, is unambiguously polar. This explains the anomaly
of examples like Would you prefer to watch with the light either on or off? or (to an
expectant mother) Are you hoping for either a boy or a girl? – the either forces a polar,
yes/no, interpretation which conflicts with normal assumptions that there are no other
possibilities than those expressed.
No alternative interrogative clause type corresponding to alternative question

Although [8i] and [ii] are, semantically, different kinds of question, they do not belong,
grammatically, to different clause types. This is one reason why we have adopted different
terminologies for subclassification at the two levels. Semantically we distinguish three
kinds of question on the basis of the way they define the set of answers: polar, alternative,
and variable. But grammatically there are just two subtypes of interrogative: closed and
open.
The reason we do not treat the or of alternative questions as a clause type marker is that
the coordination in which it figures may be between clauses, as in [7ii]. This is a clause-
coordination, not a clause, so the issue of what clause type it belongs to doesn’t arise. And as
for the component clauses, is it genuine and is it a hoax, they have the same syntactic form as
clauses expressing polar questions. Note that with embedded questions the clause subordi-
nator whether can appear in both coordinates: I don’t know whether it’s genuine or whether it’s
a hoax. From a grammatical point of view, therefore, alternative questions are distinguished
from polar questions not by the system of clause type but by a special use of coordination.
Polar-alternative questions
A special type of alternative question has the alternatives consisting of a positive and its
negative counterpart. Questions of this kind are logically equivalent to polar questions,
and we refer to them as polar-alternative questions:
[10] i a. Are you ready or are you not ready?
b. Are you ready or aren’t you ready?
c. Are you ready or aren’t you? [polar-alternative]
d. Are you ready or not?
e. Are you, or are you not, ready?
ii Are you ready? [polar]
§ 4.4 Alternative questions 871
As shown in [i], the second coordinate can be reduced by the omission of repeated
material, and its position relative to the first can be varied, as in [ie]. With embedded
polar-alternative questions there is also the possibility of having or not adjacent to the
subordinator whether: They want to know whether or not you’re ready.
The questions in [10] are logically equivalent in that they define the same set of answers.
They do so, however, in different ways. Polar [ii] expresses a single proposition and the
answers are provided by this and its polar opposite, whereas the polar-alternatives in [i]
express two propositions, each of which provides an answer. The distinction between polar,
alternative, and variable questions is based on the way they define the set of answers, and
in accordance with the definitions given above, therefore, [i] and [ii] belong to different
categories despite their logical equivalence. The term ‘polar-alternative’ is to be understood
as denoting a subclass of alternative questions.
Apart from the issue of how the answers are derived, there are two other respects in which
[10i] behave like alternative questions rather than polar ones. These involve the subordinate
constructions illustrated in:
[11] i a. I wonder/doubt whether it is alive. [polar]
b. I wonder/∗doubt whether it is alive or dead. [alternative]
c. I wonder/∗doubt whether it is alive or not. [polar-alternative]
ii a. ∗I’m marrying her whether you like her. [polar]
b. I’m marrying her whether you like her or hate her. [alternative]
c. I’m marrying her whether you like her or not. [polar-alternative]
While verbs like wonder license interrogative complements expressing all three kinds of
question, doubt accepts only the polar type: the polar-alternative is excluded just as other
alternative questions are (see Ch. 11, §5.3.3). Conversely, the ungoverned exhaustive condi-
tional construction [ii] excludes the polar type, while allowing polar-alternatives as well as
other alternative questions (see Ch. 11, §5.3.6).
Pragmatic differences between polar and polar-alternative questions

Although polar questions are logically equivalent to their polar-alternative counterparts,
there are considerable pragmatic differences between them. The polar version is simpler
and much more frequent: it can be regarded as the default version. We draw attention
here to a selection of contexts favouring one rather than the other of the two.
(a) Polar-alternative emphasises choice
The explicit expression of the negative often has an emphatic effect. One reason for my
emphasising the choice might be that you have failed to give a satisfactory response,
i.e. an answer, to a previous polar question. In such a context, the polar-alternative
question is likely to have an impatient, hectoring, or petulant tone, conveying “Make
up your mind”, “Give me an answer”, or the like. The different versions of the polar-
alternative, as illustrated in [10i], vary in the extent to which they convey such emotive
meaning. In general, the less elliptical the form, the greater the emotive meaning is likely
to be: compare Are you going or not? with the more insistent Are you going or aren’t you?
Typically, however, the version where the second coordinate is interpolated within the
first, as in [10ie], is the most hectoring.
(b) Polar-alternative emphasises the exhaustiveness of the two alternatives
[12] a. Was it good? b. Was it good or not?
The answers to both polar [a] and polar-alternative [b] are simply It was good and It
wasn’t good, but (no doubt because of the tendency for the negative It wasn’t good to
be interpreted as “It was bad” – see Ch. 9, §5) [a] will often receive such non-answer
responses as It was okay, It wasn’t too bad, and the like. Such responses locate ‘it’ in
the middle ground between good and bad. The polar-alternative can serve, then, to
insist on a simple, unequivocal choice between ‘good’ and ‘not good’ – and again the
emotive meaning of impatience or the like will be more evident in the less elliptical
versions.
(c) Polar version preferred when answers are of unequal status for the speaker
[13] i Have you any idea how much these things cost?
ii Will they agree to the proposal, do you think, or not ?
iii Is it the sixteenth today?
I might use [i] rhetorically, to convey that I believe that you haven’t any idea of the
cost: in this use (which is not the only one, of course) it would be a biased question,
one where I am predisposed to one answer over another. In this case it would be very
unnatural to add or not, for this would take away the rhetorical effect. Negative polar
questions – e.g. Don’t you like it? – are always biased, and will never be pragmatically
equivalent to polar-alternatives. The polar-alternative, by expressing both positive and
negative propositions, tends to assign them equal status.
In a case like [13ii] the effect is to give you full freedom to choose between them:
it avoids any appearance of according greater likelihood to one answer. Especially in
combination with the parenthetical do you think, the effect of the polar-alternative may
then be to suggest a certain diffidence or deference to the addressee.
Bias is not the only factor that can make the answers of unequal status. Consider
[13iii], for example. I might say this when my concern is to find out what date it is,
and in that case a “yes” answer gives me the desired information but a “no” does not
(so that a co-operative response would go further: No, it’s the fifteenth, say). For this
reason, the answers are of unequal value, and in such a context the polar-alternative
version would be very unlikely. There are numerous other ways in which the an-
swers might be of unequal status. To give just one, note that we say Are you awake?,
not (normally) Are you awake or not? The latter suggests that positive and negative
answers are on a par, but they are not. If you are awake you can answer Yes, but
if you’re not you can’t answer No, so only one of the answers is a possible (true)
response.
4.5 Variable questions

Answers to variable questions
Variable questions have a propositional content consisting of an open proposition, i.e.
a proposition containing a variable (Ch. 1, §5.1). The answers express closed proposi-
tions derived by substituting a particular value for the variable. If we use the symbol
‘x ’ for the variable, we can represent the propositional content of What did they give
her? as “They gave her x”, and the answers have different values for the variable x :
They gave her some books; They gave her the key; They gave her everything she asked for ;
and so on.
Prototypically, there is no logical limit to the number of different possible values, so
that the set of answers will be open-ended. It need not be so, however: a limit to the
§ 4.5 Variable questions 873
possible values may be incorporated into the question, as in Which of the two proposals
suits you better? 7
The form of variable questions

These questions have the form of an open interrogative clause. They are marked by a
phrase containing an interrogative word – whose role is to express the variable. Details
of the interrogative words and their properties are given in §7. The question may consist
solely of the interrogative element, forming a clausal fragment: Who?; What about the
others?; and so on.
Fronting of the interrogative phrase
When the interrogative phrase has a function other than that of subject of the inter-
rogative clause, it is normally fronted to prenuclear position, and in main clauses this
triggers subject–auxiliary inversion. Fronting of a non-subject interrogative phrase is
not obligatory in main clauses: it can remain in situ, i.e. in the default position of cor-
responding non-interrogative phrases, following the verb. Open interrogatives with a
post-verbal interrogative phrase generally occur, however, only in contexts of sustained
questioning, such as quizzes and interrogations by legal counsel, police, and so on.
Compare, then:
[14] fronted: inversion in situ: no inversion
i a. Where are those senses located? b. And those senses are located where?
ii a. What were the results of that b. And the results of that examination were
examination? what?
The [b] versions are attested examples used in court during the cross-examination of
a medical witness. As such, they are quite distinct from echo questions: the aim was to
elicit new information, not a repetition or clarification of what had just been said.
Infinitivals
Open interrogatives may have infinitival form, with or without to; in either case, no
subject is permitted.
The to-infinitival construction
Two non-embedded cases of this are to be found:
[15] i What to do in the event of fire [titular]
ii How to persuade her to forgive him? [main clause]
Type [i] is a non-sentential construction: infinitivals of this kind are used as titles of
books, articles, etc., or headings for lists, notices, and the like. They have the same
function as an NP: compare How to get rich quick and Five easy ways to get rich quick.
In [ii] the interrogative is a main clause, forming a sentence – note the difference in
punctuation between [ii] and [i]. By virtue of forming a sentence, it will normally have
illocutionary force: it’s a matter of asking, or at least wondering. This type is somewhat
rare and literary; one case of it is in interior monologue, where one is pondering over a
question. The meaning here is essentially “How could he persuade her . . . ?”
7
Other terms to be found in the literature as equivalent to our ‘variable question’ include ‘x-question’,
‘wh-question’, ‘specific question’, ‘partial question’, and ‘information question’.
The bare infinitival construction

This is found only in main clauses, almost invariably with why:
[16] a. Why be so soft with them? b. Why not accept his offer?
These convey that I don’t think there is any valid reason, and this leads to an interpretation
as an indirect directive: “I suggest that you not be so soft with them / that you accept his
offer”. With negatives, a finite construction can be used in the same way: Why don’t you
accept his offer? A negative finite, however, does not have to be interpreted in this way.
Thus Why don’t you let him drive your car? can be interpreted literally as asking for your
reasons or indirectly as suggesting you should let him, whereas Why not let him drive
your car? has only the second interpretation.8
Single-variable vs multi-variable questions

All the variable questions given so far have contained a single variable, but it is possible
for there to be more than one. Compare:
[17] i Who said that? [single-variable Q]

ii Who said what?
iii Who said what to whom? [multi-variable Q]
We are representing the propositional content of [i] as “x said that”, with ‘x ’ as the
variable; the content of [ii] can similarly be represented as “x said y”, with two variables,
and that of [iii] as “x said y to z”, with three. We therefore distinguish [i] and [ii/iii] as
respectively single-variable and multi-variable questions. As there is a straightforward
match with the grammar, we can use the same terms to distinguish the corresponding
grammatical categories: clause [i] is a single-variable open interrogative, while clauses
[ii] and [iii] are multi-variable ones.
Coordination of interrogative phrases

It is possible for two or more interrogative phrases to be coordinated:
[18] i How many sheets and how many towels do we need to take?
ii When and where did you see her?
These two examples differ in that in [i] the two coordinate phrases are functionally
alike (with the coordination as a whole understood as object of take), whereas in [ii]
they are functionally distinct (when being an adjunct of temporal location, where one
of spatial location). In answers to [i], phrases expressing the values of the variables will
be coordinated, as in We need to take six sheets and a dozen towels, but in answers to [ii]
they generally will not, as in I saw her last Saturday at the Planetarium.
Restriction on fronting of interrogative phrase

In multi-variable questions where the variables are not coordinate, no more than one
interrogative phrase can be fronted:
[19] i They did what to whom?
ii What did they do to whom?
iii ∗What to whom did they do?
8
Bare infinitivals with how are occasionally attested, but they are of questionable acceptability: ?How leave the
matter rest? The interpretation is similar to those with why, suggesting that there is no way in which one could
reasonably leave the matter rest.
§ 4.5 Variable questions 875
Answers to multi-variable questions

Multi-variable questions may have either singulary answers or multiple answers:
[20] i Who beat who?
ii Kim beat Max. [singulary answer]
iii Kim beat Max and Pat beat Bob. [multiple answer]
A singulary answer simply provides a single value for each variable. A multiple answer
provides sets of values – pairs if there are two variables, triples if there are three, and so on.
Thus for “x beat y ” in question [20i], the singulary answer [ii] provides the value “Kim”
for “x” and “Max” for “y”, whereas the multiple answer [iii] provides the pairs {“Kim”,
“Max”} and {“Pat”, “Bob”} for the variable pair {“x ”, “y”}. For example, [ii] might be
the answer when the question is used to inquire about the final match in some sporting
competition, [iii] when it is used to inquire about the semi-finals.
A two-variable question with multiple answers can be used as an inquiry in a range
of contexts differing with respect to what information I already have, as opposed to that
which I am seeking to obtain:
[21] i I don’t know the values of either variable.
ii I know the values of one variable but not the other.
iii I know the values for both variables, but not how they are paired.
Answer [20iii] would be used under condition [21i] when I simply know that two
matches took place but don’t know who the players were, or I know who played but
not who won or lost. A context where [21ii] applies would be one where I know who has
won through to the final but want to know who they beat in the semi-finals. Context
[21iii] would obtain if I knew who was playing in the semi-finals and ask Who beat
who? to find out the results. Or [21iii] might similarly apply to the answer to such a
question as Who’s going to teach which courses this semester? A likely context for this
is the planning of a teaching-programme when we know who the teaching staff are
and what the courses are: it is then just a matter of matching teachers with courses.
Multiple answer vs multiple response

Single-variable questions (and multi-variable ones where the variables are coordinate, as
in [18]) have only singulary answers. Again, however, we need to invoke the distinction
between answer and response, for sometimes a single-variable question can receive a multiple
response:
[22] i Where did she buy these books? [single-variable Q]

ii She bought them at Heffer’s.
[singulary answers]
iii She bought them at Heffer’s and Dillon’s.
iv She bought this one at Heffer’s and that one at Dillon’s. [multiple response]
With [ii] we have a straightforward singulary answer: all the books were bought at one
place. Answer [iii] is still singulary, but as the books were bought at more than one place
a coordinate phrase is used to give the value of the variable. This situation then implic-
itly raises the multi-variable question of which books were bought at which shop, and [iv]
gives the answer to this question, not to [i]: it gives more information than is needed to
answer the question to which it was a response. As we have noted, a co-operative participant
commonly provides more information than is directly asked for. The difference in direct-
ness can be brought out by contrasting the anomaly of [23i] with the naturalness of [ii]:
[23] i #I know that she bought these books at Heffer’s and Dillon’s, but I don’t know where she
bought these books.
ii I know that she bought these books at Heffer’s and Dillon’s, but I don’t know where she
bought which.
The single-variable where she bought these books in [i] is the embedded counterpart of [22i].
Example [23i] is anomalous because it is self-contradictory. Thus to know [22iii] is to know
the answer to [22i], even in a context where [22iv] is true: information about the pairing
of books with shops is not part of the answer. The multi-variable where she bought which in
[23ii] is the embedded counterpart of Where did she buy which?, and the naturalness of the
example shows that to know [22iii] is not sufficient to know the answer to this latter question:
in the multi-variable case, information about the pairing is part of the answer.
Juxtaposition of variable and polar or alternative questions

It is not uncommon for a variable question to be followed by a polar or alternative
question which pragmatically supersedes it:
[24] i What’s her name? Is it Anne?
ii What’s her name? Is it Anne or Anna?
In [i] the polar question suggests an answer to the variable one. If the answer is positive,
then answering the second also provides the answer to the first – though the typical form,
Yes, shows that it is given in response to the polar question. If the answer is negative,
then another answer to the variable question is needed – No, it’s Anna (as we saw above,
this kind of response is often given when the variable question is merely implicit). In
[ii] the alternative question reformulates the variable one, narrowing down the range
of possible answers; this time, whatever the answer to the second question is, it will
simultaneously be an answer to the first. (To respond with Neither: it’s Amy is not to
answer the alternative question but to reject its presupposition: see §6.1 below.)
In either the polar or the alternative case, the second question will often be elliptically reduced:
[25] i What’s her name? Anne?
ii What’s her name? Anne or Anna?
Such reduction will be particularly likely when the full form would involve more lexical
repetition than is the case with [24]: Who do you think I am? (Do you think I am) Father
Christmas?; What are you going to give him? (Are you going to give him) a book or just money?
We have punctuated the questions in [24] and [25] as separate sentences, but they could
also be integrated into a single written sentence: e.g., for [24], What’s her name – Anne? With
an alternative question, but not a polar, they can be separated simply by a comma: What’s her
name, Anne or Anna? Similarly, in speech there are different degrees of prosodic integration
between the two questions. This is one of the places where it is difficult to draw a sharp
distinction between a succession of two grammatical sentences on the one hand and a single
sentence on the other.
4.6 Direction questions

Distinguished from information questions by the illocutionary force
of the answers
The great majority of questions are information questions: when used as inquiries they
seek to elicit information. The characteristic illocutionary force of their answers is that of a
§ 4.6 Direction questions 877
statement. There is also, however, a kind of question whose answers characteristically have
the force of directives. They seek not information but direction, and we accordingly call
them direction questions.9 The distinction applies to all three of the polar, alternative,
and variable categories of question. Compare:
[26] information question direction question
i a. Did he open the window? b. %Shall I open the window? [polar]
ii a. Did he do it then or later? b. %Shall I do it now or later? [alternative]
iii a. When did he come back? b. %When shall we come back? [variable]
The answers to the information questions are He opened the window /He didn’t open the
window; He did it then / He did it later; He came back at six (or whenever): these would
all have the force of statements. The answers to the direction questions are Open the
window / Don’t open the window; Do it now / Do it later; Come back at six (etc.): and
these would have the force of directives. With the polar questions, Yes and No could of
course be used in either case, but they would still have the statement force in response to
[a] and directive force in response to [b]. We annotate the [b] examples with ‘% ’ because
some varieties use will in place of shall here.
Distinction between direction and information questions only

weakly grammaticalised
As far as unembedded questions are concerned, direction questions are not sharply
distinguished in grammatical form from information questions. Matters are complicated
by the fact that a response giving an answer to a question may, like utterances generally,
have more than one illocutionary force. Consider, for example:
[27] i a. Do you promise not to tell him? b. I promise not to tell him.
ii a. Would you advise me to accept? b. I would advise you to accept.
iii a. Have I got to eat it all? b. You have got to eat it all.
iv a. %Shall I tell the police? b. You shall tell the police.
For convenience, we give the answers in unreduced form, rather than as Yes, I do, etc.
Answer [ib] has the illocutionary verb promise used performatively, so that it is a promise
as well as a statement. Similarly [iib] both makes a statement and gives advice: it differs
from [ib] in that the advice force is indirect rather than direct, because we have I would
advise, not I advise. Answer [iiib] is literally a statement but, in the context of answering
question [iiia], stating that you have to eat it all amounts to telling you to do so. Here
then we have an answer which is both statement and directive. However, from the point
of view of its form, and more specifically of the relation between its form and that of its
answer, there is nothing special about [iiia]. What makes [iva] different from the other
questions is that shall here has a sense that is specialised to direction questions: [ivb] is
not a possible answer precisely because the meaning of shall is not the same.
The semantic development of shall has led to a situation where it has a use in 1st
person interrogatives that specifically marks direction questions. It can also be used with
a futurity sense, as in the information question Shall I ever need it again? As a result,
there is potential for ambiguity:
[28] Shall I get my money back? [direction or information question]
9
Direction questions are also known as ‘deliberative questions’.
As a direction question, this is concerned with the choice between future actions by
the speaker: I’m asking you to tell me to retrieve it or not to do so. As an information
question, it is concerned with predictions as to what will happen: will the money be
returned or not?10
The case with should is less clear-cut. Should I get my money back? is ambiguous between a
deontic reading (“Is getting my money back the right thing for me to do?”) and an epistemic
one (“Is it probable that I’ll get it back?”). But the ambiguity is also found in You should get
your money back. So the development of should has certainly not been entirely parallel to
that of shall. Nevertheless, we do see something partly analogous. While Yes, you should is a
perfectly natural (deontic) response to Should I tell the police?, it would be odd to respond in
this way to Should I open the window?, said in a context where it is a matter of my possibly
opening the window there and then. In this use should behaves like shall – and could not be
replaced by ought . . .to, as it could in Should I tell the police?
Embedded direction questions have infinitival form

We have seen that with unembedded questions the distinction between direction and
information questions is not matched by any sharp grammatical distinction; with em-
bedded ones, however, the distinction is clearly marked. For here there is a construction,
the infinitival interrogative, that is used exclusively for direction questions. Compare the
following, as used, say, in the frame I asked :
[29] information question direction question
i a. whether she told him b. whether to tell him [polar]
ii a. whether he left then or later b. whether to leave then or later [alternative]
iii a. how she got home b. how to get home [variable]
Because they are embedded, these are questions only in the semantic sense, not in
the pragmatic sense, as they do not themselves have illocutionary force. But we can
relate them to unembedded questions by considering them in the suggested frame. I
asked whether she told him reports my asking the information question “Did she tell
him?”, while I asked whether to tell him reports my asking the direction question “Shall
I tell him?”11 Note that subordinate interrogatives with shall do not express direction
questions. I must ask him whether I shall get my money back, for example, lacks the
ambiguity of [28], having only the information question interpretation.
Right answers to direction questions

It is because of the existence of direction questions that we cannot identify a ‘right’
answer with a true one (cf. §4.1). Right answers to information questions are true, but
the categories true and false are not applicable to directives, and hence to the answers to
direction questions.
10
In neither interpretation is You shall get your money back an answer; with a 2nd or 3rd person subject shall
indicates commitment on the part of the speaker, rather than obligation on the part of the subject-referent
(cf. Ch. 3, §9.6.1). The positive answers to the two readings of [28] are therefore Get your money back and You
will get your money back.
11
We are concerned here with indirect reported speech, where one reports the content of what was said rather
than the actual words, but for present purposes we can consider the idealised case where there is the closest
possible match. The embedded infinitival construction does not always correspond to an unembedded shall
question. How shall I turn the machine on? is hardly idiomatic whereas I don’t know how to turn the machine
on is; nevertheless, the answers to the infinitival question involve directions for turning the machine on.
§ 4.7 Biased questions 879
If I use a direction question to ask you to tell me what to do, then the issue of whether
the answer given is right or wrong is trivial. Consider, for example:
[30] A: Shall I call a taxi for you? B: No, thanks. I’ll enjoy the walk.
A’s question is an indirect offer, and it is up to B to accept or reject. The issue of whether
the directive answer is right or wrong is comparable to that of whether a statement like
I promise to help you, when used to make a promise, is true or false. This latter issue
is trivial because the statement is made true simply by virtue of its being uttered with
the relevant intention, and similarly B’s directive answer to the direction question in
[30] will be right simply by virtue of B’s deciding to deliver that directive rather than
another.
There are also cases, however, where the issue of what the right answer is to a direction
question can arise in a non-trivial way. For example, such questions can be used to ask
for advice:
[31] A: Shall I take a taxi? B: No, you’d be better off walking.
Here it is easy to imagine circumstances under which B could be said to have given bad
advice, given the wrong answer – e.g., if the distance were too great for A to be able to
walk it comfortably in the time available. Similarly when one puts a direction question
to oneself, in wondering: what the right answer is in this case is of course a crucial issue.
Determining what is the right answer to direction questions used to seek advice or in
wondering involves a judgement as to what course of action is in the best interests of
the one uttering the question (or, in shall we questions, of the group containing that
person).
4.7 Biased questions

The distinction between neutral questions and biased questions
A biased question is one where the speaker is predisposed to accept one particular an-
swer as the right one. A neutral question lacks such bias towards one answer rather
than another: it is the default category on this dimension. The distinction between
neutral and biased questions applies primarily to polar questions. Compare, for
example:
[32] i Did you get any annuity, superannuation, or other pension? [neutral]
ii Doesn’t she like it? [biased]
Example [i] is taken from an income tax form: it is addressed individually to all those
filling in the form, and for any individual there is no expectation on the part of the
‘speaker’ (the Income Tax Commissioner) that the answer will be positive rather than
negative, or vice versa. A plausible context for [ii], though not the only one, is that her
behaviour or her remarks suggest that she doesn’t like it: I ask the question to confirm
whether this is so. In such a context the question is biased towards the negative answer
She doesn’t like it.12
12
Alternative terms for ‘neutral’ and ‘biased’ are ‘open’ and ‘conducive’ respectively.
4.7.1 Kinds and degrees of bias

(a) Epistemic bias
There are different kinds of bias. It may be simply a matter of the speaker thinking,
expecting, or knowing that one answer is the right one. We will refer to this kind of bias
as epistemic, a term whose primary application is in the closely related field of modality.
In the contextualisation of [32ii] suggested above, for example, the bias towards the
negative answer will be an epistemic one.
(b) Deontic bias

Alternatively, it might be a matter of the speaker judging that one answer ought to be
the right one. Again we will take over a term from the field of modality, and refer to this
kind of bias as deontic. It is seen in the natural interpretations of:
[33] i You’re surely not going to let them get away with outrageous behaviour like that, are
you?
ii Aren’t you ashamed of yourselves?
In [i] there is a deontic bias towards a negative answer: I convey a judgement that you
ought not to let them get away with their outrageous behaviour. Example [ii] shows a
deontic bias towards a positive answer: you ought to be ashamed of yourselves. But at
the same time, [ii] has an epistemic bias towards a negative answer: it appears from your
behaviour that you are not ashamed of yourselves.
(c) Desiderative bias

A third kind of bias, not greatly different from the deontic, is that where the speaker
wants one answer to be the right one – desiderative bias, as we shall call it. For example,
when I indirectly request something by means of such a question as
[34] Can I have some more ice-cream?
there will be a desiderative bias towards a positive answer: I want a Yes answer. The
negative epistemic bias of [32ii] could also be accompanied by a positive desiderative
bias. This could be the case in a context where it refers to something I am responsi-
ble for (a painting, say, that I have painted or chosen): I want her to like it but think
she doesn’t.
Different degrees of bias

We also find considerable differences in the degree or strength of the bias. Take, for
example, the case illustrated in [13iii], where I ask, wanting to know the date:
[35] Is it the sixteenth today?
It is likely that there will be some positive bias here (primarily epistemic), for if I had no
idea whether it was the sixteenth or not, it would generally be more natural to use the
variable question What date is it today? But my confidence that I am right about the date
can vary greatly. Similarly in [32ii] the negative bias could be a matter of a mere flimsy
suspicion or a strongly supported conviction.
The limiting case is where the bias is complete: I am in no doubt at all as to what is
the right answer.
§ 4.7.2 Declarative questions 881
[36] i A: May I speak to Ms Jones? B: I’m afraid she’s no longer here. Didn’t you
know that she went overseas yesterday?
ii A: I wasn’t able to get a ticket. B: Weren’t you? I’m sorry to hear that.
In [i], assuming that B does not consider the possibility that A intended to deceive,
B will be in no doubt that A didn’t know that Ms Jones went overseas yesterday, i.e.
that the answer to the question is negative. Nor is there any doubt in [ii]: A has just
given the answer, and B accepts it. Where the bias is complete, as here, the question
cannot have the force of an inquiry: the intention is not to elicit information. In [i]
the question serves to inform A that Ms Jones went overseas yesterday: it is an indirect
statement. In [ii] the question serves to acknowledge the information A has just supplied
(cf. [3i] above).
Complete bias can also be found with variable questions, as in Who’s a clever girl? In a context
where the addressee or the speaker is a girl who has just done something clever, there will be
no doubt as to what is the value for the variable in “x is a clever girl”, and the question will
therefore indirectly convey “What a clever girl you are!” or “What a clever girl I am!”.
The encoding of bias

The inference that a question is biased towards a particular answer may be based simply
on the context, together with assumptions about the speaker’s intentions. This is likely
to be the case in [35], for example. In other cases, bias may be reflected in the prosodic
properties of the question. For example,
[37] Have you any idea how much these knives cost?
could be used as an indirect way of inquiring about the cost of the knives or as a rebuke to
someone considered to be misusing a certain knife. In the first case there is some positive
desiderative bias (I am no doubt hoping for a positive answer to the direct question, for
otherwise you will not be able to answer the indirect question about the cost, the one I
am primarily interested in), but it could be epistemically quite neutral (I have no reason
to think that one answer rather than the other is actually the right one). In the rebuke
use, on the other hand, there will be a strong negative epistemic bias: the suggestion is
that the maltreatment of the knife indicates lack of awareness of its value.
Our focus in what follows will be on the grammatical marking of bias, on cases where
the bias is reflected in the grammatical structure of the question. Declarative questions
and negative interrogative questions are always quite strongly biased; note that neither
of these would occur in the kind of context attested for the neutral question [32i]. In
addition, a weaker bias can be conveyed by the use of positively- and negatively-oriented
polarity-sensitive items, such as some and any. We take these three cases in turn; see also
§5.2 for bias in tag questions.
4.7.2 Declarative questions

Positive declarative questions have an epistemic bias towards a positive answer, negative
ones towards a negative answer:
[38] a. They’ve finished? b. They haven’t finished?
The expected answer is here the statement with the same propositional content as
the question – i.e. They’ve finished and They haven’t finished respectively. In asking a
declarative question I am typically seeking confirmation of a proposition that I am in-

clined, with varying degrees of strength, to believe. There may be deontic or desiderative
bias as well as epistemic, but this is not inherent to the construction as such.
Lexical reinforcement of bias

The bias may be reinforced by lexical markers indicating confidence in the truth of the
proposition expressed:
[39] i They no doubt misunderstood her intentions?
ii You’re surely not going to agree?
iii And the manager has been informed, of course?
iv There isn’t any chance of her changing her mind, I take it?
These confidence markers are outside the propositional content of the question, outside
the scope of the question. The positive answer to [39i], for example, is not They no
doubt misunderstood her intentions, but simply They misunderstood her intentions. Such
markers would not naturally occur in interrogative questions, where comparable items
are epistemically much weaker:
[40] i Did they perhaps misunderstand her intentions?
ii Isn’t there any chance of her changing her mind, I wonder?
Responses to declarative questions
The bias of declarative questions is reflected in the fact that they can naturally receive
confirmatory responses like That’s right, Exactly, Quite so, which would be out of place
with a neutral question. Yes can even occur here with a following negative, which is not
normally possible: Yes, there’s no chance at all is a plausible response to [39iv], but not to
interrogative Is there any chance of her changing her mind? or even Isn’t there any chance
of her changing her mind?
Declarative questions as indirect speech acts

The illocutionary force of declarative questions is not that characteristically associated with
the clause type, and hence they are indirect speech acts in the sense of §3.2. At the direct
level they are statements, but the intonation overrides this to yield an indirect question. The
indirectness is reflected in the fact that pragmatic inferences may be involved in determining
the scope of the question, which need not be the whole of the propositional content. Consider,
for example:
[41] i I take it there isn’t any chance of her changing her mind?
ii I hope you’re not proposing to leave it like that?
iii I don’t suppose I could borrow your car for a couple of hours?
Example [41i] differs grammatically and semantically from [39iv], but pragmatically there
is little difference. In [39iv] I take it is a parenthetical (§5.3), but in [41i] it occurs as part
of a complex clause construction: there isn’t any chance of her changing her mind is this time
a subordinate clause, the grammatical complement of take, and its content is semantically
integrated into that of the complex clause. Nevertheless, I take it is again outside the scope of
the question: the expected answer is not You take it there isn’t any chance . . . , but simply There
isn’t any chance . . . At the direct level, [41i] is a statement with the propositional content “I
take it there isn’t any chance of her changing her mind”, but indirectly it is a question
with the content “There isn’t any chance of her changing her mind”, or perhaps “It is true
§ 4.7.3 Negative interrogative questions 883
that there isn’t any chance of her changing her mind”. Example [ii] is just like [i], but [iii]
involves two additional factors. One is that the negative is associated with the complement of
suppose (cf. the discussion of increased specificity of negation in Ch. 9, §5), giving “Couldn’t
I borrow your car for a couple of hours?”; and the other factor is that the latter question in
turn indirectly conveys a request to borrow the car.
4.7.3 Negative interrogative questions

Questions with negative interrogative form are always strongly biased. They typically
allow a range of interpretations, and the epistemic bias can be towards either the negative
or the positive answer. Consider:
[42] Didn’t I tell you Kim would be coming?
One context for this is where it has become apparent that I have, or probably have,
omitted to tell you that Kim would be coming. Here the bias is towards the negative
answer (I didn’t tell you). But I could equally use [42] in a context where I remember
quite well having said that Kim would be coming: my prediction was not accepted at the
time but has now been shown by Kim’s presence to have been correct and I am asking
you to admit that I was right. Here, then, the bias is towards the positive answer (that I
did tell you).
Similarly with such examples as:
[43] i Wasn’t I right?
ii Isn’t it all as simple as she predicted?
iii Aren’t they spending Christmas with their uncle?
iv Isn’t it raining?
For [i] possible interpretations are: “It appears that I wasn’t right – is that so?” (negative
epistemic bias) and “It is now evident I was right – admit it” (positive). For [ii]: “It is
looking as though it is not as simple as she predicted, isn’t it?” (negative) and “It has
turned out just as simple as she said it would, hasn’t it?” (positive). For [iii]: “It seems
I was wrong in thinking they are spending Christmas with their uncle” (negative) and
“Remember they are spending Christmas with their uncle” (positive). A context for this
latter, positively biased, interpretation could be one where you have suggested inviting
them over for Christmas and I point out that they won’t be able to come because they will
be away at their uncle’s – another case where a question is used not to obtain information
but to indirectly impart it. For [iv], a negatively biased interpretation is “I thought it was
raining but there is now evidence suggesting it is not” (for example, you may be showing
signs of going out without protection against rain), and a positively biased one is “Let
me remind you that it’s raining” (e.g. in response to Why aren’t you going out? – “Surely
the fact that it is raining is reason enough!”).
Negative interrogative questions typically suggest some element of contrast. We will
consider this feature first in cases where the bias is negative, and then in those where it
is positive.
Negative interrogative questions with negative bias

The negative epistemic bias commonly contrasts with a positive deontic bias. This was
illustrated in [33ii] above, Aren’t you ashamed of yourselves? The most salient interpreta-
tion here carries an implied contrast between the state of affairs which apparently obtains
(negative) and my judgement of what should be the case (positive). When such a contrast
reflects adversely on you, the question will be an indirect reproach or rebuke, as in [33ii].
This is also the natural interpretation of [44i], and a quite likely one for [44ii–iii]:
[44] i Can’t you think of a more positive response?
ii Didn’t you turn the oven off?
iii Don’t you know where it goes?
In this interpretation [44i] conveys “It appears you can’t think of a more positive response,
but you ought to be able to”. Similarly for [ii] we may have “You apparently didn’t turn
the oven off, but you ought to have done”, and for [iii], “You have been told where it goes
but have apparently forgotten”. On the other hand, I may myself accept responsibility
for the contrast between what is and what should be, and then the question may be
accompanied by an apology. An alternative, apologetic, implicature of [iii], therefore, is
“I should have told you where it goes but apparently didn’t do so”.
Such a contrast between what is and what should be (whether a matter for reproach or
apology) is not, however, the only kind of contrast that may be suggested by the negative
interrogative construction. The negative bias reading of [43iv], for example, is unlikely
to suggest that it ought to be raining. There is still a contrast, however: between what
now appears to be the case (negative: it’s not raining) and what I previously thought to
be the case (positive: it was raining).
Negative interrogative questions with positive bias

Where the epistemic bias is positive, there is commonly an implicit contrast between
my belief in some proposition and previous unwillingness on the part of you or others
to accept it. This is the context suggested for the positive bias interpretations of [42]
and [43].
One case where such a contrast is not in evidence, or at least not obviously in evidence,
is that where the question is used as indirectly equivalent to an exclamatory statement:
[45] Aren’t they lovely! Haven’t they made a good job of it! Doesn’t he talk fast!
Haven’t I been a fool! Didn’t it rain!
The interpretations can be given in the form of exclamatives: “How lovely they are!”;
“What a good job they made of it!”; “How fast he talks!”; “What a fool I have been”;
“How it rained!” The indirect statement force is reflected in the falling intonation that
marks this use. Such exclamatory questions typically involve gradable expressions – very
often adjectives (predicative lovely or attributive good ) or adverbs (fast), but also certain
nouns (fool ) or verbs (rain).
4.7.4 Positively- and negatively-oriented polarity-sensitive

items (some vs any, etc.)
The distinction between contrasting pairs of items such as unstressed some vs any, or
already vs yet, is discussed in detail in Ch. 9, §4.3; here we focus on the relation between
this distinction and bias in questions.
Positive interrogative questions

Here, the selection of a positively-oriented item rather than its negatively-oriented coun-
terpart confers some degree of positive bias. Compare:
§ 4.7.4 Polarity-sensitive items 885
[46] negatively-oriented item positively-oriented item

i a. Is anything wrong? b. Is something wrong?
ii a. Has anybody told Ed about it? b. Has somebody told Ed about it?
iii a. Have they gone yet? b. Have they gone already?
In each pair [b] suggests that I am rather more inclined towards a positive answer than
does [a]. The difference is particularly evident in [iii]: already is a stronger marker
of positive bias than some and its compounds. The bias is potentially either epistemic
or deontic/desiderative. The positively-oriented somebody in [iib], for example, could
reflect some positive evidence that somebody has told him about it (e.g. he may have said
something which suggests that he knows about it) or it could reflect the need for someone
to tell him (e.g. it has been recognised that his behaviour is causing inconvenience and
it is necessary that someone inform him of this).
Positively-oriented items with desiderative bias often appear in questions used as
indirect speech acts of various kinds:
[47] i Could you please do something about that noise.
ii Would you like some coffee?
Example [i] is an indirect request, with the direct question force very much back-
grounded. In the natural interpretation, I assume you can do something about the
noise (so that there is complete positive bias) and ask you to do so. Negatively-oriented
anything would be out of place here. Example [ii] would typically be used as an offer, a
more hospitable one than the form with any: the positive bias signalled by some suggests
that I am favourably inclined towards an acceptance of the offer, whereas any suggests
indifference. In a more effusive offer such as Would you care for some of this delicious cof-
fee? substitution of any for some would be unlikely, because of the inconsistency between
the indifference of any and the enthusiasm of the rest.
Positive declarative questions

These always have a strong positive bias, as we observed in §4.7.2. For this reason they
do not take negatively-oriented items:
[48] There’s something/∗anything else you need? ∗
You have ever been to Paris?
Negative interrogatives
Negatively-oriented items give these a negative bias:
[49] i Haven’t they seen anybody about it yet?
ii Wasn’t I right about anything else?
Question [i] expects the negative answer They haven’t seen anybody about it yet, and
analogously for [ii]. Example [ii] differs strikingly from [43i], Wasn’t I right?: it allows
only the negative bias interpretation, whereas [43i] allows both positive and negative.
Positively-oriented items occur in negative interrogatives with either bias, though the
positive case will often be more salient:
[50] i Didn’t you like some of it?
ii Haven’t you forgotten something?
iii Shouldn’t someone do something about it?
iv Weren’t some of them marvellous!
Question [i] can be interpreted with positive bias, conveying “It wasn’t all bad: there
was some of it you liked, wasn’t there?”, or (less likely) with negative bias: “It apparently
wasn’t a complete success: there was some of it you didn’t like, did you?” The others
generally have positive bias. Example [ii] would typically convey “You have apparently
forgotten something”, and would often be used as a reminder (e.g. to a child to say
please when asking for something). Similarly, [iii] is likely to be seeking agreement to
the proposition that something should be done about it. Finally, [iv] illustrates the use
of positively-oriented items in indirect exclamatory statements.
4.8 Echo questions

The prototypical use of the echo question is to question whether one has correctly heard
what the previous speaker said – heard the stimulus, as we call it. My doubt as to whether
I heard the stimulus correctly may arise because it was not perceptually clear (I may have
had difficulty making it out above some background noise) or because its content is
surprising or remarkable in such a way that I want to verify whether you did in fact say,
or mean to say, what I apparently heard.
Polar, alternative, and variable echo questions

Echo questions are predominantly of either the polar or the variable kind, but alternative
echoes are also possible:
[51] stimulus echo question
i A: She’s leaving on Saturday. B: She’s leaving on Saturday? [polar]
ii A: He gave it to Anne. B: He gave it to Anne or Anna? [alternative]
iii A: He’s proposing to resign. B: He’s proposing to what? [variable]
The polar echo question prototypically repeats the stimulus but with a rising into-
nation imposed on it; like ordinary polar questions, it has Yes and No as answers. The
alternative echo question substitutes an or-coordination for part of the stimulus, and
each answer will include just one of the coordinates; it generally has the rise + fall in-
tonation pattern of ordinary alternative questions. The variable question, which – like
the polar – prototypically also has rising intonation, substitutes an echo question word
expressing a variable for part of the stimulus, and the answers involve replacing the
variable by its possible values.
Modification of stimulus
The stimulus is often modified by reduction – by omitting parts or replacing them by
shorter expressions such as pro-forms. An echo response to Kim is going to try and
persuade him to buy a microwave, for example, could take one of many forms, including:
[52] i To try and persuade him to buy a microwave /one /a what?
ii To buy a microwave /one /a what?
iii Kim/Who is?
And since the stimulus will normally be produced by a different speaker, there will be a
change in deictic pronouns: I like it will be echoed as You like it?, and so on.
§ 4.8.1 The contrast between echo and ordinary questions 887
4.8.1 The contrast between echo and ordinary questions

Echo questions as indirect speech acts
The propositional content of echo questions is not the same as that which is actually
expressed in the utterance, and such questions therefore belong to the class of indirect
speech acts. The propositional content of the echo question in [51i] is not “She’s leaving
on Saturday”, the content actually expressed, but something like “You said she’s leaving
on Saturday”. (We say ‘something like’ for it could be “You’re telling me . . .”, “You’re
suggesting . . .”, and so on: precisely because it is implicit rather than directly expressed
there is some imprecision as to what it is.) And, correspondingly, the answer Yes is
equivalent not to She’s leaving on Saturday, but to I said she’s leaving on Saturday. The
difference between these two interpretations may not seem very important, but is of
much greater significance in a case like:
[53] A: Is he going to resign? B: Is he going to resign?
The propositional content of B’s echo question is clearly not “He’s going to resign”, but
“You said he’s going to resign”, and a Yes answer in this context is equivalent to I asked
whether he is going to resign, not to He’s going to resign. The latter would be an answer to
A’s question, not B’s – to an ordinary question, not an echo question. The same applies
to the other kinds of question. For example, the propositional content of B’s variable
question in [51iii] is approximately “You said that he’s going to x”.
This indirectness is a crucial property distinguishing echo questions from ordinary questions.
They are not distinguished simply by the fact that an echo question questions what has just
been said, for it is perfectly possible to use an ordinary question to do that:
[54] previous utterance ordinary question
i A: She’s leaving on Saturday. B: Did you say she’s leaving on Saturday?
ii A: He’s proposing to resign. B: What did you say he’s proposing to do?
Note that B’s utterances here may be said with the same prosodic signals of incredulity or
the like that commonly accompany echo questions: these prosodic features are likewise not
what is crucial for the echo question. There is nothing special about B’s questions in [54]:
they are just ordinary questions whose subject matter happens to be the content of a previous
utterance.
The following also belong to the category of ordinary questions, not echoes:
[55] i A: She’s leaving on Saturday. B: Is she? [sc. leaving on Saturday]
ii A: He’s proposing to resign. B: What’s he proposing to do?
B’s questions here are not echoes in the technical sense we are giving to that term in that they
do have their face value. That is, they are genuinely questions as to whether she is leaving
on Saturday and as to what he is proposing to do: assuming that what A said was true, the
answers are She’s leaving on Saturday and He’s proposing to resign. Since A has already given
the answers, B may be construed as challenging what A said or asking for repetition, but
this is not a reason for identifying them with the echoes in [51]. (As we have seen, the polar
question here could also be used simply to acknowledge, with surprise, what A has said.)
The difference between these examples and the echoes is reflected in the difference in form:
in [55] B’s questions have the characteristic grammatical form of ordinary questions, but in
[51] they do not.
4.8.2 The grammatical form of variable echo questions

We turn now to the form of echo questions, beginning with the variable kind. The
variable echo question is grammatically marked by an echo-question word – what in
the above examples. As there is here a fairly straightforward relation between grammar
and meaning we can use the term ‘variable echo’ at both levels, speaking of a variable
echo construction at the level of grammar and of a variable echo question at the level of
meaning. And just as there is a difference in meaning between a variable echo question
and a variable ordinary question, so there are differences in form.
Grammatical differences between variable echoes and open interrogatives

(a) Differences between what/who as echo-question words and as interrogative words
One difference between the two constructions is that these words, especially what, have a
wider range of uses in variable echo clauses than in open interrogatives. For example, in
[51iii], He’s proposing to what?, the echo word what is a verb and combines with to to form
a verb phrase, but interrogative what is not a verb and hence we need do in [55]. Note also:
[56] i A: He was enthusing about the film. B: He was whatting about the film?
ii A: They gave it to Angela Cooke. B: They gave it to Angela who?
Example [i] shows that echo what, unlike interrogative what, can inflect. And [ii] shows
that echo who likewise differs from interrogative who: the latter must always be initial in
the interrogative phrase.
(b) Position of echo-question and interrogative words,
and subject–auxiliary inversion
In open interrogative clauses a non-subject interrogative element is normally fronted
and triggers subject–auxiliary inversion; in variable echo clauses a non-subject question
element always remains in situ and therefore never triggers inversion. Compare, for
example, echo He’s proposing to what? ([51iii]) with ∗What is he proposing to?, or They
gave it to Angela who? ([56ii]) with ∗Angela who did they give it to? The absence of fronting
and inversion serves of course to make the echo maximally like the stimulus that it echoes.
However, there are places where we find overlap between echoes and ordinary ques-
tions:
[57] i Who made a mistake? [subject]
ii And the purpose of that was what? [post-verbal non-subject]
Example [i] is the case where the interrogative or echo element is subject, and hence will be
in pre-verbal position in either case. It could be either an ordinary question with answers
like The Secretary-General made a mistake or an echo question with the latter as a possible
stimulus and answers like I said, ‘The Secretary-General made a mistake’. Question [ii] has
non-subject what after the verb: this is a possible if rare position in open interrogatives,
as well as the only possible one in echoes, so [ii] could be an ordinary question with such
answers as The purpose of that was to test the PH level or an echo of a stimulus like the
latter. The two interpretations would normally be distinguished prosodically: the echo
construction will generally have a fall + sharp rise on the question word.
(c) Relation with clause type
The open interrogative is a clause type, and as such it is mutually exclusive with the
other clause types. The variable echo construction, by contrast, is not a clause type. It
§ 4.8.3 The form of polar echo questions 889
is a construction on a quite independent dimension, and can combine with any of the
clause types:
[58] stimulus variable echo question
i A: She’s a genius. B: She’s a what? [declarative]
ii A: Did Kim complain? B: Did who complain? [closed interrogative]
iii A: What did he do last week? B: What did he do when? [open interrogative]
iv A: What a fuss Ed made! B: What a fuss who made? [exclamative]
v A: Give the key to Angela. B: Give what to Angela? [imperative]
In each of these, B’s echo question belongs to the same clause type as the corresponding
stimulus, being derived from it by substituting a question element for some element of
the stimulus.
The independence of the variable echo construction from clause type is explicable in
terms of both form and meaning. As far as its form is concerned, it is marked simply by
the presence of a question word, and there is no reason why this should not co-occur
with any of the clause type markers. Notice in particular that the variable echo does
not determine any features of order, which means that the order of elements in the
clause is able to be determined by clause type (or other properties). As far as meaning is
concerned, the variable echo question is indirect; at the direct level it is merely a partial
repetition, citation, of the stimulus and hence there is no reason why the stimulus should
not have an illocutionary force of the kind characteristically associated with any of the
clause types.
Multi-variable echo questions

Like an ordinary question, an echo question can contain more than one variable. All
the examples so far have been single variable echoes, but multi-variable echoes can be
formed simply by substituting a question word for two or more elements in the stimulus:
[59] stimulus single variable echo multi-variable echo
i A: Kim’s a genius. B: Kim’s a what? B: Who’s a what?
ii A: Give the key to Pat. B: Give what to Pat? B: Give what to who?
Each question word will normally bear the main stress in its own intonation group.
Because the differences between interrogative words and echo-question words are relatively
slight, there is scope for a great deal of potential ambiguity as to whether a word that does not
trigger subject–auxiliary inversion is an interrogative word or an echo-question word. Who’s
a what?, for example, is given in [59i] as a two-variable echo of a declarative, but it could also
be a single-variable echo of an open interrogative, used in response, say, to the stimulus Who’s
a genius? In the two-variable interpretation, who and what are both echo-question words,
but in the single-variable reading only what is, who being an interrogative word. Similarly,
Who saw what? can be a non-echo multi-variable open interrogative, a single-variable echo
of a single-variable open interrogative (with a stimulus such as Who saw the weasel?) or a
multi-variable echo of a declarative (with a stimulus like The butler saw the weasel).
4.8.3 The form of polar echo questions

Unlike the variable echo question, the polar echo is not, for the most part, expressed by
any special grammatical construction: rather, the echo is signalled prosodically, by the
rising intonation. A polar echo question can have the syntactic form belonging to any
of the clause types. Polar counterparts of the variable echoes given in [58], for example,
are as follows:
[60] stimulus polar echo question
i A: She’s a genius. B: She’s a genius? [declarative]
ii A: Did Kim complain? B: Did Kim complain? [closed interrogative]
iii A: What did he tell her? B: What did he tell her? [open interrogative]
iv A: What a fuss Ed made! B: What a fuss Ed made? [exclamative]
v A: Give the key to Angela. B: Give the key to Angela? [imperative]
The echo repeats the stimulus, with rising intonation signalling a request for repetition,
or justification.
We have seen that rising intonation can also combine with declarative clause type to yield an
ordinary polar (or alternative) question, and there is accordingly again potential ambiguity
in examples like She’s a genius? It can be an ordinary (direct) question whose answers are
She’s a genius and She’s not a genius; or it can be, as in [60i], a polar echo (indirect) question
whose answers are I said she’s a genius and I didn’t say she’s a genius. In both cases the question
is biased – towards an answer which expresses the same propositional content as is expressed
in the direct question or implied in the indirect one. For example, in:
[61] a. She gave it to him? b. She didn’t give it to him?
the expected answers for the ordinary question interpretations are, for [a], She gave it to him
and, for [b], She didn’t give it to him, whereas those for the echo question interpretations are
respectively I said, ‘She gave it to him’ and I said, ‘She didn’t give it to him’.
The bare predication construction in polar echoes

[62] stimulus polar echo (bare predication)
i A: Kim has resigned. B: Kim resign?
ii A: She’s a genius. B: Her a genius?
One case where a polar echo does have a special syntactic form is the bare predication
construction. The clause consists of a subject together with a non-finite VP (as in [i]),
or simply a predicative complement (as in [ii]). The echo belongs to relatively informal
style, and a personal pronoun subject will thus tend to take accusative case.
4.8.4 Repetition vs clarification echoes

The echoes considered so far have all been what we will call repetition echo questions, as
their answers include a repetition of the stimulus. They are distinguishable from a more
peripheral kind of echo used to seek clarification of some element in the stimulus – the
clarification echo question. The distinction may be illustrated by means of a potentially
ambiguous example, such as:
[63] i A: I’ve finally solved the problem
of the missing cents. B: You’ve finally solved what? [repetition]
ii A: I’ve finally solved it. B: You’ve finally solved what? [clarification]
In [i] the echo what substitutes for the problem of the missing cents, which B has not
properly perceived or understood. The answer would be I said, ‘I’ve finally solved the
problem of the missing cents’: this is the repetition echo interpretation. In [ii], by contrast,
§5 Interrogative tags and parentheticals 891
B has heard perfectly well what A has said but does not know what A intended to refer to
by the pronoun it. B seeks therefore not a repetition of the stimulus, but a reformulation
that expresses A’s intended meaning more successfully. Similarly, B’s echo in [56ii], They
gave it to Angela who?, was given above as a repetition echo, aiming to obtain repetition
of Cooke, but it could also be used as a clarification echo to A’s stimulus They gave it to
Angela: A mistakenly assumes that the name Angela on its own is sufficient to pick out
the intended referent, and B asks for a fuller referring expression.
One obvious use of the clarification echo is when the stimulus is an incomplete
utterance like A’s I need to buy a new er, er, . . . ; here B’s echo response You need to buy a
new what? aims to elicit the word or expression that A was trying to find.
Clarification echo questions can again be of the polar, alternative, or variable kind.
In response to the stimulus Give the key to Angela, for example, we might have:
[64] i Give her the front-door key? [polar echo]
ii Give her the front-door key or the back-door one? [alternative echo]
iii Give her which key? [variable echo]
There are intonational differences between clarification and repetition echoes, espe-
cially in the variable kind. For example, You’ve finally solved what? as a repetition echo
(in [63i]) will have a fall followed by steep rise on what, but as a clarification echo (in
[63ii]) it will have falling intonation.
The implicit propositional content for the repetition echo begins, roughly, “You
said . . .”, whereas that for the clarification echo might be given as “You meant . . .”.
And indeed the distinction between the two kinds of echo question can be made explicit
by means of parentheticals using one or other of these verbs:
[65] i Give the key to Angela, did you say? [repetition echo]
ii Give her the front-door key, do you mean? [clarification echo]
In spite of these differences, the clarification echo belongs grammatically with the repetition
echo in that it too exhibits the grammatical properties that distinguish the variable echo
construction from the open interrogative. This is evident from the above examples. In You
need to buy a new what?, the echo-question word what is a common noun with a determiner
and adjectival modifier as dependents: interrogative what, by contrast, is a pronoun, unable
to take such dependents. And in [64iii] the echo question element which key occurs in an
imperative clause, whereas interrogative which key is a clause type marker, and can occur only
in open interrogatives.
5 Interrogative tags and parentheticals
In this section we look at the form and interpretation of questions like:

[1] i He’s rather aggressive, isn’t he? [interrogative tag]
ii He’s rather aggressive, don’t you think? [interrogative parenthetical]
An interrogative clause is here added as a supplement to another clause, changing the
illocutionary force of the utterance. The clause to which the interrogative is attached we
refer to as the anchor, e.g. he’s rather aggressive in [1]. In the default case the anchor is
declarative, but it can belong to any of the five major clause types: closed interrogative
in Is it genuine, do you think?, imperative in Be quiet, will you!, and so on (those with
imperative anchors are discussed in §9.7).
No great significance attaches to the terminological distinction between tag and paren-
thetical, but it is useful to have a separate (and well-established) term for the construction
shown in [1i], which we examine first.
5.1 The formation of interrogative tags

Reversed polarity and constant polarity tags
A tag is a short interrogative clause which may be negative or positive:
[2] i Your friends made a good job of it, didn’t they? [negative tag]
ii They haven’t finished it, have they? [positive tag]
In [i] a negative tag attaches to a positive anchor and in [ii] a positive tag attaches to a
negative anchor: we refer to these as reversed polarity tags. It is also possible to have
constant polarity tags, where the tag has the same polarity as the anchor. Thus in [i] we
could have did they? instead of didn’t they?, and in [ii] haven’t they? instead of have they?
The four possibilities are shown in:
[3] positive anchor negative anchor
i a. He is ill, isn’t he? b. He isn’t ill, is he? [reversed polarity tag]
ii a. He is ill, is he? b. %He isn’t ill, isn’t he? [constant polarity tag]
As far as the meaning is concerned, the important issue is not whether the tag is positive
or negative, but whether it has reversed or constant polarity. Reversed polarity tags are
much the more frequent, and constant polarity tags occur predominantly with posi-
tive anchors: many speakers reject examples like [iib]. A more formal variant of the
inflectional negative isn’t he? is the analytic negative is he not?
Formation of tags by reduction of full interrogatives

In the examples so far, the tag might be regarded as a reduced version of a full closed
interrogative clause corresponding to the anchor. The tag in [2i], for example, might be
derived from its anchor in three steps, as follows:
[4] i your friends made a good job of it [anchor]
ii your friends didn’t make a good job of it [step i: reverse polarity]
iii didn’t your friends make a good job of it? [step ii: form interrogative]
iv didn’t they? [step iii: reduce]
For constant polarity tags, step i would of course be skipped. Although this procedure
works in the great majority of cases, it runs into difficulties with anchors like those in:
[5] i Few of them liked it, did they?
ii It’s hardly fair, is it?
The only way we could apply step i to the anchor few of them liked it would be to change
this into few of them didn’t like it, which would lead via steps ii and iii to the tag didn’t
they? But in fact the reversed polarity tag for this anchor is did they?, as in [5i]. The reason
is that few of them liked it is in fact negative (see Ch. 9, §3.3), so the reversed polarity tag
must be positive. It is a negative, however, with no positive counterpart, so step i can’t
be applied. The same applies with [ii].
§ 5.1 The formation of interrogative tags 893
Direct formation of the tag

It follows that we need rules to account for the tags directly, rather than by reduction of
some full interrogative clause. The form of the tag is as in [6], and it can be derived by
procedure [7]:
[6] Auxiliary as predicator + personal pronoun as subject (+ not)
[7] i Subject: if anchor subject is a personal pronoun, repeat it; otherwise take the
anchor subject as antecedent and select the appropriate personal pronoun.
ii Auxiliary lexeme: if anchor predicator is an auxiliary, select the same lexeme,
otherwise select do.
iii Auxiliary tense: same as anchor tense.
iv Auxiliary person–number properties (if any): determined by agreement with
subject.
v Polarity: opposite to that of anchor for reversed polarity tags, the same for con-
stant polarity tags.
vi Negation: if tag is negative, choose between the less formal synthetic negation
(with negative form of auxiliary) and the more formal analytic negation (neutral
auxiliary, with final not)
Step [i] will often require pragmatic information: the tag for the anchor the boss has
arrived can be either hasn’t he? or hasn’t she?, depending on the sex of the boss. Step [ii]
reflects the normal rules for closed interrogative formation: the closed interrogative is a
do-support construction. Step [iv] selects person–number properties by reference to the
tag subject rather than the anchor predicator to cater for cases like Everybody has read
it, haven’t they?, where the anaphoric personal pronoun for singular everybody is plural
they. Steps [v] and [vi] handle the polarity of the tag and of the auxiliary in accordance
with the account given above.
Tags based on subordinate clauses

Special provision must be made for the construction illustrated in:
[8] a. I think it’s legal, isn’t it? b. I don’t think it’s legal, is it?
In [a] the first constituent structure boundary is between I think it’s legal and isn’t it?: the tag is
appended to the main clause I think it’s legal, which is in this sense the anchor. But the form of
the tag is based on the subordinate clause it’s legal, so from this point of view it is the latter that
is treated as anchor. This conflict reflects the mismatch between the grammatical structure of
I think it’s legal and its communicative meaning. Grammatically, it’s legal is subordinate to the
think clause, but communicatively it is the subordinate clause that is primary: I think simply
expresses some modal qualification. The anchor is comparable to It’s probably legal, where
the modal qualification is expressed in a grammatically subordinate way, by an adverbial
adjunct, or It’s legal, I think, where the qualification is parenthetical: for both of these the
procedure [7] would give isn’t it? as tag quite straightforwardly.
Example [8b] is similar but has the added complication of negative polarity. Is it? is a
reversed polarity tag, but the negative which it reverses is in the think clause. Again the form
of the tag reflects the communicative meaning rather than the grammatical structure – by the
process we call specificity increase (Ch. 9, §5) the negative is interpreted as applying to the
complement of think (“I think it isn’t legal”), so the tag is positive, just as it is in It’s probably
not legal, is it? or It isn’t legal, I think, is it?
Other expressions allowing the tag to be based on a subordinate clause complement

include: I believe/suppose/guess/reckon; it seems/appears; it follows /this means; and so on
(cf. It seems we made a mistake, didn’t we?; It follows that we won’t have to pay any more,
will we?).
Minor departures from the main pattern of tag formation

We find a number of departures from the forms predicted by [7] (but with a good deal of
idiolectal and dialectal variation), all indicating that meaning rather than exact syntactic
form is what is important in tag selection:
[9] i The non-prototypical auxiliary ought is sometimes replaced by the synonymous
should: You ought to have told them the whole truth, shouldn’t you?
ii The rules predict mayn’t it? as the informal reversed polarity tag for It may rain, but
most speakers do not have the form %mayn’t; there is no clearly established way of
filling the gap: possibilities include mightn’t it?, won’t it?, the more formal may it not?,
or a structurally independent interrogative such as parenthetical don’t you think?, isn’t
that so?, etc.
iii Do may be found as a variant of have in the tag to an anchor with have got: He’s got
problems, doesn’t he? (which may be regarded as a blend of He’s got problems, hasn’t
he? and He has problems, doesn’t he?).
iv Be + 3rd person pronoun can occur as tag to a verbless anchor: Lovely day, isn’t it?;
Beautiful ship, isn’t she?
5.2 The use and interpretation of tags

Reversed polarity tags
The illocutionary force of an utterance with the form anchor + tag depends on the
prosody. The two principal patterns both have falling tone on the anchor; the tag itself
is either rising or, more frequently, falling:
[10] positive anchor negative anchor
i a. He was here, wasn’t he? b. He wasn’t here, was he? [rising tag]
ii a. He was here, wasn’t he? b. He wasn’t here, was he? [falling tag]
(a) The rising tag
This expresses doubt or asks for verification: the question is biased towards an answer
that confirms the anchor. A special case, involving a negative anchor, is prosodically
distinguished by a somewhat wider pitch movement and the lack of any rhythmic break
between anchor and tag. Here there is no such bias towards an answer with the same
polarity as the anchor:
[11] It isn’t raining again, is it? It isn’t my turn already, is it?
If anything, there is a bias towards a positive answer, but in addition the construction
has an emotive component of meaning – a suggestion of being afraid that the positive
answer is the true one.
(b) The falling tag
The version with falling intonation on the tag does not express doubt: the question merely
seeks acknowledgement that the anchor is true. Thus it can be used in a context where
the anchor is obviously true: Good gracious, you’re up early this morning, aren’t you?,
§ 5.3 Parentheticals 895
uttered at 4 a.m., say. There may be, as perhaps in this example, an implicit invitation
to provide an explanation (Yes, I’ve got a train to catch). Or I may want you to admit
something you didn’t previously accept (I was right all along, wasn’t I?). Or again I
might be asking for your agreement to some minor uncontroversial proposition (It’s
a lovely day again, isn’t it?). Thus an exclamative anchor will normally take a falling
tag because I can hardly ask you to confirm my exclamation: What a mess I’ve made of
things, haven’t I? With an exclamative the truth of the proposition is not at issue (see
§8.2), so that such an anchor is inconsistent with the expression of doubt. The falling
tag may therefore have the character of a rhetorical question, where an answer-response
is unnecessary.
Constant polarity tags

The characteristic intonation for constant polarity tags is slightly but not steeply rising.
They do not, however, express doubt: the content of the anchor is typically something I
am repeating or inferring from what you have just said or from what was said earlier. For
many speakers they occur only in the positive. One use, commonly accompanied by so
or a comparable item such as oh, I see, etc., carries an emotive meaning of disapproval,
reproach, belligerence, or the like:
[12] i So you have forgotten your homework again, have you?
%
ii So you haven’t done your homework, haven’t you?
These suggest a context where you have just revealed that you have forgotten your
homework or failed to do it. Because the anchor proposition is implicitly attributed to
you, this use lends itself to sarcasm, as when I say to someone who has performed badly:
So you’re the one who was going to come back laden with prizes, are you?
Such belligerence is not, however, a necessary feature of the constant polarity tag
construction. A second use is where I accept what you say, indicating some surprise or
at least acknowledging that the information is news to me:
[13] A: Jones is coming over next semester.
B: Jones is coming, is he? In that case we can ask him to give some seminars.
In both uses, the anchor proposition derives from the addressee, rather than representing
a prior belief of the speaker. Exclamatives, which do not readily occur in such contexts,
normally allow only reversed polarity tags.
5.3 Parentheticals
By parentheticals we mean expressions which can be appended parenthetically to an
anchor clause but which also have a non-parenthetical use in which they take a declarative
content clause as complement – expressions like I think, don’t you think?, and so on.
Compare:
[14] non-parenthetical use parenthetical use
i a. I think it is quite safe. b. It is quite safe, I think.
ii a. Don’t you think it is safe? b. It is safe, don’t you think?
iii a. Would you say it is safe? b. Is it safe, would you say?
iv a. When did she say it’ll be safe? b. When will it be safe, did she say?
In the parenthetical use they can, in general, interrupt the anchor instead of following
it, as in, for example, It is, I think, quite safe or Has there, would you say, been any serious
attempt at compromise?
In the parenthetical construction the anchor is syntactically a main clause, whereas the
corresponding clause in the non-parenthetical construction is subordinate. And while
the verbs think and say have complements in [a], in [b] they do not: in this construction
verbs which (with the meanings they have here) normally require a complement occur
without one. The syntactic structure signals that the anchor is the communicatively
most important part of the message, with the parenthetical supplement correspondingly
backgrounded.
Declarative anchor + declarative parenthetical: It is quite safe, I think

Many of the parentheticals here serve, like I think, to weaken the speaker’s commitment
to the truth of the anchor proposition: I believe, I reckon, I guess, I suppose, I suggest,
it seems, it appears, etc. Others are somewhat stronger: I’m sure, I have no doubt. And
some have concessive force: it is true, I admit.
In the non-parenthetical construction the matrix verb and its subject may also be
backgrounded, but this is not signalled syntactically, being rather a matter of pragmatics,
dependent on context and the content of the subordinate clause. I believe that there is a
God, for example, will be taken as a statement about my beliefs, with the informational
status of believe matching its syntactic status as verb of the main clause. This is not the
natural way of taking I believe that nominations close on Tuesday, however. Here, the
subordinate clause will generally be pragmatically foregrounded and the matrix I believe
reduced in status to a modal qualifier, making it like the parenthetical in Nominations
close on Tuesday, I believe. Note that while Do you? is a pragmatically perfectly natural
response to the God example, it would be rather odd as a response to the nominations one,
indicating that I believe had been taken as foreground material. With the parenthetical
construction this response would be anomalous.
Declarative anchor + interrogative parenthetical: It is quite safe,

don’t you think?
The effect of the parenthetical is much like that of a tag (isn’t it?): I seek your confirmation
that the anchor proposition is true. Other such parentheticals include wouldn’t you say?,
don’t you reckon?, and so on. One might also say am I right?, don’t you think so?, and the
like, but these would have the status of independent questions rather than parentheticals.
Interrogative anchor + interrogative parenthetical: Is it safe, would you say?

From a grammatical point of view we have here a sequence of two interrogative clauses,
but in normal use I would be asking one question, not two. The parenthetical does not
serve to ask a second question but to clarify the way the question expressed in the anchor
clause is to be interpreted and answered: it is not part of the propositional content that
is questioned. The parenthetical makes clear that I am asking for an expression of your
opinion – and, indeed, a more formal variant might be In your opinion, is it safe?, where
the adjunct in your opinion likewise does not contribute to the propositional content of
the question.
The non-parenthetical Would you say it is safe? might well be interpreted in the same
way, but again this is a pragmatic matter, not signalled in the syntactic structure. Note,
§6 The presuppositions of information questions 897
for example, that Did you say it’s safe? can be used as an indirect speech act to ask whether
it is safe, i.e. essentially equivalent to Is it safe, did you say?, but it can also be used at face
value as a question about what you said.
Other interrogative parentheticals that are appended to interrogative anchors include
would/do you reckon/guess/believe/argue? We also find similar expressions with 3rd person
subjects: does she think?, did she say?, etc. In general, the same items are used with closed
and open interrogatives. An exception is do you know?, which appears as a parenthetical
only with the closed type: Are they valuable, do you know?
Interrogative parentheticals with echo-question anchors

Parentheticals can also be appended to echo questions:
[15] i He’s going to what, did you say?
ii Did I help him, do you mean?
Again the parenthetical does not express a second question, but clarifies the force of the
whole – [i] is a repetition echo, [ii] a clarification echo.
Note that with these there is no comparable non-parenthetical construction. Did you
say that he’s going to what?, for example, is an (unlikely) echo of a question (with a
stimulus like Did I say that he’s going to emigrate?), whereas [15i] is an echo of a statement
(with a stimulus like He’s going to emigrate).
6 The presuppositions of information questions
Pragmatic Q–A presuppositions

In general, when I ask a question I presuppose that it has a right answer. We are concerned
here with pragmatic presupposition: to presuppose something in this sense is to take
it for granted, as not at issue, to present it as uncontroversial background. In the case of
information questions, to presuppose that a question has a right answer is to presuppose
that one of the possible answers is true.
There are several different kinds of pragmatic presupposition, and we will accordingly
refer more specifically to the kind we are concerned with here as a question–answer
presupposition (‘Q–A presupposition’). The most straightforward case is perhaps that
of the alternative question: we will look at this first, and then examine how the concept
of Q–A presupposition applies to polar and variable questions.
6.1 Q–A presuppositions of alternative questions

The possible answers to [1i] are given in [ii], and its Q–A presupposition in [iii]:
[1] i Is he leaving on Monday or Tuesday? [alternative Q]
ii “He is leaving on Monday”; “He is leaving on Tuesday” [answers]
iii “He is leaving on Monday or Tuesday” [presupposition]
To presuppose that [i] has a right answer is to presuppose the disjunction (or-coordi-
nation) of the possible answers, i.e. “Either he is leaving on Monday or he is leaving
on Tuesday”, which is logically equivalent to [iii]. The presupposition of an alternative
question can thus be derived simply by taking the corresponding statement.
The or in alternative questions and their presuppositions is interpreted exclusively – i.e. the
question presents a set of alternatives with the presupposition that one, but only one, is true.
Just as [1i] does not countenance the possibility that he will not leave on Monday or Tuesday,
so it does not countenance the possibility that he will leave on both days. This relation of
mutual exclusiveness between the alternatives is perfectly consistent with the use of such
expressions as or both:
[2] i Would you like cheese, fruit, or both? [alternative question]
ii “You would like cheese”; “You would like fruit”;
“You would like both fruit and cheese” [answers]
iii “You would like cheese or fruit or both” [presupposition]
The or both simply adds a third alternative, mutually exclusive with the first two, for “cheese”
and “fruit” in the first two answers are interpreted as “just cheese” and “just fruit”.
6.2 Q–A presuppositions of polar questions

The presuppositions of polar questions are arrived at in a similar way:
[3] i Has the clock stopped? [polar question]
ii “The clock has stopped”; “The clock hasn’t stopped” [answers]
iii “(Either) the clock has stopped or it hasn’t” [presupposition]
To presuppose that [i] has a true answer is again to presuppose the disjunction of the
answers listed in [ii], i.e. to presuppose [iii]. The Q–A presupposition of a polar question
is therefore the disjunction of the corresponding statement and its polar opposite.
This applies equally when the question is negative:
[4] i Didn’t she see them? [polar question]
ii “She didn’t see them”; “She did see them” [answers]
iii “(Either) she didn’t see them or she did” [presupposition]
We have seen that negative questions are biased, but bias is perfectly consistent with
presupposition. The presupposition is that one of the answers is true; the bias goes
beyond this and favours one rather than the other as the true one.
The presuppositions in [3] and [4] are necessarily true: they are logical truths, tautolo-
gies. This will always be the case with polar questions: the disjunction of any proposition
and its polar opposite is necessarily true. Polar questions clearly differ in this respect
from alternative ones: [1], for example, is obviously not a logical truth. It should not
be thought, however, that this property of polar presuppositions makes them vacuous:
from a pragmatic point of view they can be of considerable significance. This may be
illustrated by such an example as:
[5] i Are you telling the truth? [polar question]
ii “(Either) you are telling the truth or you are not” [presupposition]
The presupposition is what I take for granted, and the significance here is that the
presupposition is so weak – it doesn’t go beyond a mere tautology. In asking [5i] I take
for granted the truth of [ii], but in so doing conspicuously fail to take for granted the
truth of the proposition “You are telling the truth”. The question entertains the possibility
that you are not telling the truth, thereby casting doubt on your veracity. At the direct
level [i] is clearly a question, not an accusation, but is not uncommon for such questions
to be interpreted as indirect accusations.
§ 6.3 Q–A presuppositions of variable questions 899
6.3 Q–A presuppositions of variable questions

The presuppositions of variable questions cannot be derived in quite so simple a way as
with alternative and polar questions. This is because the set of answers is prototypically
open-ended: we shall not therefore derive the presupposition by taking the disjunction
of the possible answers. Instead we derive it by substituting an appropriate indefinite
phrase for the variable in the open proposition expressed in the question:
[6] i Who wrote the editorial? [variable question]
ii “Person x wrote the editorial” [open proposition]
iii “Someone wrote the editorial” [presupposition]
We have seen that the answers to a variable question assign values to the variable(s) in the
open proposition. To presuppose that [6i] has a right answer, therefore, is to presuppose
that there is a true proposition in which a value is assigned to the variable in [ii] – and
to presuppose this is to presuppose [iii]. Note that the latter does not constitute an
answer to the question, and to give it as a response would be clearly unco-operative.
This is precisely because it does not provide any information beyond that which the
speaker presents as already established. Thus [iii] no more constitutes an answer to [i]
in set [6] than it does in sets [2–4]. With alternative and polar questions an answer
must be more specific than the presupposition by selecting one of the coordinates;
with variable questions it must be more specific by genuinely providing a value for the
variable.
It will be noted that in [6ii] the variable x is attached to ‘person’ instead of standing
on its own, as in the formulations which, for simplicity of presentation, we have been
using hitherto. ‘Person’ is needed to show that the variable ranges only over the set of
persons: #The lawnmower wrote the editorial, for example, is not a possible answer to
[6i]. ‘Person’, of course, is part of the meaning of who, and is encapsulated also in that
of someone, which appears in the presupposition. For when, a comparable schema is
given in:
[7] i When did they move to Edinburgh? [variable question]
ii “They moved to Edinburgh at time x” [open proposition]
iii “They moved to Edinburgh at some time” [presupposition]
A detailed account of the presuppositions of variable questions must therefore draw on
a description of the individual interrogative words: we take this up in §7 below.
Q–A presuppositions in relation to the set consisting of all the answers

Another, but equivalent, approach to an explanation of the Q–A presupposition of a
variable question is to see it as following from every one of the answers. Thus instead of
[6], for example, we might have:
[8] i Who wrote the editorial? [variable question]
ii “Ian wrote the editorial”; “I wrote the editorial”;
“One of the directors wrote the editorial”; . . . [answers]
iii “Someone wrote the editorial” [presupposition]
If Ian wrote the editorial, it follows that someone wrote it, and similarly for all other
answers.
It is important to note, however, that the presupposition may follow not from the answer
alone, but from the answer together with the premise that it is an answer. Consider [9i] and
the sample of responses given in [ii]:
[9] i a. What Soviet president won the Nobel Peace Prize? [question]
b. “Some Soviet president won the Nobel Peace Prize” [presupposition]
ii a. Mikhail Gorbachov won the Nobel Peace Prize.
b. Leonid Brezhnev won the Nobel Peace Prize.
c. Willy Brandt won the Nobel Peace Prize.
The right answer is [iia], but this does not by itself entail [ib]: we need the additional
premise “Mikhail Gorbachov is/was a Soviet president”. Though false, [iib] is still an answer.
By contrast, [iic] is true, but not a true answer, because the additional premise – “Willy
Brandt is/was a Soviet president” – is false. Presuppositions thus play a significant role in the
interpretation of responses to variable questions. If you interpret a response as an answer you
add to its content whatever additional premise is needed for the presupposition to be entailed.
No such additional premises are needed with alternative and polar questions. Here the pre-
supposition is a disjunction of the answers, and since any proposition “p” entails “p or q”, any
one answer will always entail the disjunction of the set of answers. In [1], for example, if “He is
leaving on Monday” is true, then obviously “He is leaving on Monday or Tuesday” is also true.
6.4 Rejection of Q–A presuppositions

The Q–A presuppositions of alternative and variable questions are generally not logical
truths, and they will not always be accepted by the addressee:
[10] i A: Will Kim or Pat chair the meeting? B: Neither. I’m chairing it this time.
ii A: Who helped her? B: Nobody – she did it herself.
B’s responses here are not answers. They contradict the presuppositions of the questions:
B rejects these presuppositions and hence cannot answer. Notice that although nobody
has the grammatical form of an NP, it does not have reference and hence does not supply
a value for the variable in the question. It indicates, rather, that there is no value of x
such that “Person x helped her” is true.
6.5 Cancellation of Q–A presuppositions

We have said that Q–A presuppositions are pragmatic in nature: they are not invariant,
immutable, as they would have to be to qualify as a semantic phenomenon. In particular,
there are cases where I can use a variable question without taking it for granted that a
value can be supplied for the variable. We will refer to this as presupposition cancella-
tion: the question is used without the presupposition that characteristically accompanies
a question.
Some examples are given in:
[11] i Who cares?
ii What do I care? / What does it matter?
iii How do you know Jill didn’t do it herself?
iv How should I know?
v Why should he do a thing like that?
Question [i] is a conventional way of indirectly asserting “Nobody cares”. Similarly [ii]
typically conveys “I don’t care at all / It doesn’t matter at all”. A difference between them
§ 6.6 Secondary presuppositions 901
is that in [i] there is nothing in the grammatical form of the sentence to indicate that
the expected presupposition is cancelled, whereas in [ii] there is. It would be possible in
principle to use [i] in a presupposition-preserving way, and this is reflected in the fact
that it can be used as complement to know: She knows who cares and who doesn’t care.
This is not so with [ii] – cf. ∗She knows what I care / what it matters. They involve semi-
fixed expressions, with what here meaning “how much”. Questions like [iii], beginning
with how do you know, are often used to challenge what the addressee has just said. I am
suggesting that you have not considered the possibility that Jill did it: I am therefore not
taking it for granted that you know she didn’t.
Question [11iv] is a conventional way of conveying “There’s no reason why I should (be
expected to) know”. Modal should quite often occurs with presupposition cancellation,
but is not an unequivocal marker of it. Question [11v], for example, could be used either
with or without the Q–A presupposition. Suppose you say I’ve just discovered that Max has
been tampering with the computer; I might then respond with [11v] in two different ways:
[12] i Yes, I can’t understand it. Why should he do a thing like that?
ii Oh, surely not! Why should he do a thing like that?
The question in [i] presupposes that Max ‘did a thing like that’ (i.e. tampered with the
computer), whereas in [ii] the presupposition is cancelled: it conveys that Max probably
didn’t do it. (Note that even in [i] should serves an emotive role, indicating puzzlement
or surprise, rather than being part of the propositional content: the presupposition is
“He did a thing like that for some reason”, not “He should do a thing like that for some
reason”.) Would is possible instead of should in all these examples.
Negatively-oriented polarity-sensitive items generally serve to cancel the presuppo-
sitions of variable questions:
[13] i When will you ever learn not to trust them?
ii Where could you find anything better?
There is no presupposition here that you will at some time learn not to trust them or
that you could find something better somewhere. On the contrary, these questions tend
to convey the speaker’s belief that you won’t, that you couldn’t.
The presupposition is also cancelled in the bare infinitival construction illustrated in
[16] of §4. Why invite them both? suggests that there is in fact no reason to do so. Likewise
in the negative finite Why don’t you invite them both?, when it’s a matter of some future
situation – the presupposition is preserved in examples like Why don’t you like it?
6.6 Secondary presuppositions

In addition to the Q–A presupposition that a question normally has simply by virtue of
being a question, there may be further presuppositions that follow from it – we will refer
to these as secondary presuppositions:
[14] i Has he stopped smoking? [polar question]
ii “(Either) he has stopped smoking or he hasn’t” [Q–A presupposition]
iii “He formerly smoked” [secondary presupposition]
Here [ii] presupposes [iii]. If he has stopped smoking, then necessarily he formerly
smoked and precisely because you can’t stop doing something that you have never
done I also wouldn’t normally say He hasn’t stopped smoking unless he had previously
done so (cf. Ch. 1, §5.4). If [i] presupposes [ii] and [ii] presupposes [iii], it follows that
[i] presupposes [iii]. When I ask [i], the issue is normally just whether he smokes now: I
take it for granted that he formerly did. We call this a secondary presupposition because
it doesn’t derive directly from the question form itself, but depends crucially on the
properties of the verb stop.
A second example is seen in:
[15] i Did he break it intentionally? [polar question]
ii “(Either) he broke it intentionally or he didn’t” [Q–A presupposition]
iii “He broke it” [secondary presupposition]
If he didn’t break it, the issue of whether he broke it intentionally doesn’t arise, so in
asking [i] I normally take [iii] for granted. But again [iii] is only a secondary presuppo-
sition of the question because it derives from properties of the verb + manner adverb
construction.
7 Interrogative words and phrases
This section is concerned with the grammatical and semantic properties of the following
interrogative words, and phrases based on them:
[1] how what when where which
who whom whose why
The interrogative subordinators whether and if are discussed in Ch. 11, §5.2: we confine
our attention here, therefore, to the open interrogative words, which are used in both
main and subordinate clauses. Who and whom are inflectional forms of the lexeme who,
and will here be treated together, the difference being a matter of case (Ch. 5, §16.2.3). The
issue of the grammatical number of interrogative phrases, as reflected in subject–verb
agreement, is discussed in Ch. 5, §18.
The dual role of interrogative words

Open interrogatives are marked as such by the presence of an open interrogative word,
but these words all have some other role in the syntactic structure. Compare:
[2] i a. Who has taken my umbrella? b. What mistakes did I make?
ii a. Someone has taken my umbrella. b. I made some mistakes.
As well as marking the clause type, who in [ia] is head of the subject NP, like someone in
[iia], and what in [ib] is determiner in the object NP, like some in [iib]. We refer to these
two roles respectively as their interrogative role and their core role. In their interrogative
role they serve as a variable that is replaced by a value in the set of answers; the core role
differs from one to another, and hence has to be described individually.
7.1 Which
Which differs from all the other interrogative words in having a property we shall call
‘selective’. It implies that the value which an answer substitutes for the question variable
is to be selected from some definite set:
§ 7.1 Which 903
Which + partitive phrase

The set may be specified by a partitive phrase within the interrogative phrase itself:
[3] i Which (one) of the chapters did you write?
ii Which of quiche, pizza, and lasagna would you prefer?
A possible answer to [i] is I wrote Ch. 3, where Ch. 3 refers to a member of the set referred
to by the chapters in the question. The partitive phrase usually consists of of + definite
NP, as in this example. An alternative to a definite NP is a coordination defining the
set by listing its members, as in [ii]. Note, however, that we can’t have an indefinite NP:
which of us / them / the boys / these books / your shares, but not ∗which of boys / some
books / any shares. Instead of of we sometimes find out of: Which out of the three cheapest
ones do you think we should take? A cardinal numeral commonly precedes the partitive:
which one/two of the chapters.
Which without an overt partitive phrase

The set from which selection is made need not be specified in a partitive phrase, but may
be given elsewhere or just be contextually implicit:
[4] i Which would you prefer, quiche or pizza?
ii It comes in three colours, red, blue, and green. Which would you prefer?
iii Which chapter(s)did you write?
In [i] the set is defined by the coordination which is added at the end, effectively convert-
ing the variable question into an alternative one. In [ii] the set is given in the preceding
text, while in [iii] it is contextually given: we are talking about some book, and the
chapters in that book constitute an identifiable set.
Which vs what
What cannot occur just before a partitive: ∗What of the chapters did you write? It is also
normally excluded before a cardinal numeral + partitive: Which/∗What one of them is
defective? Elsewhere, what is not grammatically excluded, though there will be a strong
pragmatic preference for which in cases like [4ii], where the interrogative word constitutes
the whole interrogative phrase and the set is given in the preceding text – what could
substitute much more readily for which in [i], where the set is defined later.
When the interrogative word is determiner to a noun head, both which and what are
possible:
[5] i Which/What approach to the problem would you recommend?
ii Which/What king of England had six wives?
In [i] the contrast is quite sharp. The set of approaches to a problem is not inherently
clearly defined. Normally, therefore, one would present it as an identifiable set by using
which only in a context where a number of possible approaches have been mentioned:
the question implies a choice from such a contextually defined set. What does not have
the selective feature and is the form one would use when there is no identified set to
choose from. What is not inconsistent with there being a contextually defined set, as
evident from the fact that it can be used in [ii], since the kings of England do form
an identifiable set. In such cases the distinction between which and what is effectively
neutralised: which encodes the fact that the choice is from an identifiable set while what
doesn’t, but as that is part of background knowledge it doesn’t matter from a pragmatic
point of view whether it is encoded or not. Compare, similarly, Which/What gear are we
in?, as said of a car with five gears.
7.2 Whose
Interrogative whose is genitive and (unlike relative whose) personal, so that presupposi-
tions to whose questions contain someone:
[6] question presupposition
i a. Whose bicycle did she take? b. “She took someone’s bicycle”
ii a. Whose is that? b. “That is someone’s / belongs to someone”
iii a. Whose do you prefer? b. “You prefer someone’s”
In [i] whose is determiner to a noun head, while [ii] is the predicative use, with answers
like It’s mine. In [iii] whose is a fused determiner-head, with the interpretation recoverable
from the context – e.g. Kim and Pat don’t need their bicycles today: whose would you prefer
to borrow? This is a relatively infrequent construction: one would be more likely to use
which. Whose can be used when the variable ranges over a contextually identifiable set,
but it is hardly possible with a partitive of phrase: ∗Whose of the two of them would you
prefer?
7.3 Who and whom

Nominative who and accusative whom occur only in head function, contrasting with
whose as non-genitive vs genitive, and with what as personal vs non-personal:
[7] i a. Who is that? [non-genitive]
b. Whose is that? (=[6ii]) [genitive]
ii a. Who have you got as tutor this year? [personal]
b. What have you got as set text this year? [non-personal]
Who vs what in predicative complement function
[8] i A: Who is Lesley? B: She’s their solicitor. [specifying be]
ii A: What is Lesley? B: She’s a solicitor. [ascriptive be]
In [i] we have the specifying be construction: the question asks about Lesley’s identity,
and the answer identifies her as their solicitor. Here their solicitor is used referentially.
In [ii] be is ascriptive, and the NP a solicitor in the answer is non-referential (cf. Ch. 5,
§8.3). The contextually most neutral interpretation of such what questions about people
is that the variable ranges over occupations, as in this example. The context, however,
may specify other kinds of property: the question What are you?, for example, could be
a question about your political affiliation, your religion, or your martial art grade, when
following such statements as I’m Labour, I’m Catholic, I’m a Blue Belt – and so on.
Who vs which
[9] i There are two contestants left, Kim and Pat. Which/Who do you think will win?
ii Who/Which is Lesley?
The contrast between who and which is similar to that between what and which. Unlike
which, who does not encode that selection is to be made from an identifiable set, but
§ 7.4 When 905
it can substitute for which in cases where the set is defined in context, as in [i]. Again,
we could add of them with which but not who. (A partitive phrase with of or out of is
not in general excluded with who, however: Who (out) of all the conductors you have
worked with was the most inspiring?) Both who and which are found in the specifying
be construction, [ii]. With who the more likely interpretation is that who has the iden-
tifier role, Lesley the identified, as in [8i]: I don’t know who Lesley is and am wanting
to find out. With which, by contrast, the salient interpretation has which as identified
and Lesley as identifier: I do know who Lesley is and am wanting to find out which
of a certain set of persons (e.g. on a group photograph, or on a stage) is identifiable
as her.
7.4 When
When is used to question time, and is used with a range of functions:
[10] i When is she leaving? [adjunct (temporal location)]
ii When is the concert? [complement]
iii When would be a good time to meet? [subject of specifying be]
iv When would the best time be for her lecture? [complement of specifying be]
v Since when have you been in charge? [complement of preposition]
Question [i] presupposes “She is leaving at some time”, but it is more general than
What time / At what time is she leaving?, which normally refers to clock time, so that
while in August is not an answer to the latter it is an answer to [i]. Similarly for [ii].
In the specifying be construction when could be replaced by what (or what time), and
the answer would generally be given without at: five o’clock rather than at five o’clock
(though the afternoon and in the afternoon are equally possible). One case of specifying
be where what could not replace when is the it-cleft construction: When/∗What was it
you saw her?
Since when, as in [10v], is often used sarcastically, with cancellation of the presupposi-
tion. So as well as asking how long you have been in charge, it might be used to suggest
that you are behaving as though you were in charge when in fact you are not.
7.5 Where
Where questions spatial location or goal and occurs in the same range of functions as
when:
[11] i Where are we going to have lunch? [adjunct (spatial location)]
ii Where are you? Where are you going? [complement]
iii Where would be a good place to meet? [subject of specifying be]
iv Where would the best place be for her lecture? [complement of specifying be]
v Where have you come from? [complement of preposition]
Example [i] presupposes “We are going to have lunch (at) some place”. As complement,
where can question location (Where are you?: “at what place?”) or goal (Where are you
going?: “to what place?”). In the latter case, to can be used, giving a structure like [v]:
Where are you going to? In [iii–iv] we could again substitute what. In the prepositional
construction [v] the preposition is normally to or from (though at is found in idiomatic
Where are we at?). The preposition is usually stranded, as in [v]: fronting of the prepo-
sition (From where have you come?) is confined to markedly formal style.13
7.6 Why
This has a narrower range of use than when and where, being restricted to adjunct
function or the complement of the it-cleft construction; note, for example, that other
specifying be constructions allow only what:
[12] i Why is she going home? [adjunct (cause)]
ii Why is it that we keep getting the wrong results? [it-cleft]
iii What/∗Why was the reason for her sudden departure? [specifying be]
Why questions cause (reason or purpose): answers to [i] include Because she’s hungry,
To get some food, and so on. In adjunct function (but not in [ii]) why is replaceable by
the idiomatic, and relatively informal, what . . . for: Why did you do that? or What did
you do that for?
Presupposition cancellation
Question [12i] presupposes “She is going home for some reason”, but why is often found
with cancellation of the presupposition. Compare:
[13] i a. Why is Max so naughty? b. Why am I naughty?
ii a. Why don’t you go to the beach? b. Why not go to the beach?
iii a. Why don’t you be more tolerant? b. Why not be more tolerant?
In its salient interpretation [ia] presupposes that Max is naughty and asks for the cause,
or explanation. Question [ib] can be used in a similar way, but I’m more likely to use
it to ask for evidence or justification for the claim that I’m naughty: the presupposition
is cancelled in this use, whether I’m naughty being the main issue, not something that
is taken for granted. (A similar interpretation is possible for [ia], but the so makes it less
likely.) Why questions with presupposition cancellation often contain modal should, as
noted in §6.6. Example [iia] allows a face value interpretation where the presupposition
is retained: “Why is it that you don’t go to the beach?”, but it can also be used when I
don’t think there is any valid reason and the cancellation of the presupposition results in
an indirect directive: “You should go to the beach”. The other three examples have only
this directive interpretation: see §4.5.
Direction questions and to-infinitivals

Unlike the other interrogative words, why is not normally possible in direction questions.
Thus %Why shall I get my money back? can only be an information question, with answers
like You will get your money back because the manufacturer has accepted responsibility for
the defect. In to-infinitivals, why is just possible in the titular use: Why to vote yes in the
referendum.
13
Whereabouts can be used instead of where to indicate that only an approximate answer is envisaged. However,
the noun whereabouts (derived from the preposition by shift of stress from the last syllable to the first) is not
an interrogative word: His whereabouts aren’t known is declarative and has no embedded interrogative clause
within it. Interrogative whence (“from where”) and whither (“to where”) are archaic. So too is wherein, which
is the only one of the compounds of interrogative where + preposition that one is likely to encounter; its
archaic status is seen in the fact that it does not require do-support but allows inversion of subject + lexical
verb, as in Wherein lies its appeal?
§ 7.7 How 907
Why in response to directives and questions

Why also differs from the other interrogative words in examples like:
[14] i A: Get your money back. B: When?/Why?
ii A: Did you see her? B: When?/Why?
B’s When? in [i] asks for a more specific directive, whereas Why? asks for a reason for
complying with the directive. Features such as time and place count as features of the
action itself (getting your money back tomorrow is a different action from getting it
back next week), but reason does not (getting your money back counts as the same
action, whether you do it for this reason or for that). Similarly in [ii] When? asks for
further specification of the propositional content of A’s question: it is a clarificatory echo
question, elliptical for Did I see her when? (cf. §4.8.4). Why?, by contrast, asks for A’s
reason for asking the question – and could accompany an answer to the question, as in
Yes, why?
7.7 How
Interrogative how has a considerable range of uses which we will consider in turn.
(a) Adjectival predicative complement

[15] i a. How are you ( feeling)? b. How was the concert? [PCs ]
ii a. How did you find the seminar? b. How do you like your coffee? [PCo ]
Adjectival how is used predicatively only, not attributively (we can’t say ∗How films do you
like?, but need a periphrastic expression such as what kind of ). In [i] the PC is subjective,
and in [ii] objective. It is difficult to state the presuppositions precisely, because there is
no word standing in the same relation to how as someone does to who or somewhere to
where, for somehow can only be an adverb, not an adjective. Nor is there any comparable
NP or PP standing in the same relation to how as some time does to when or for some
reason to why.
The salient interpretation of [15ia] is as a question about your present state of health,
with such answers as I am well/ill/better, I have a headache, and so on: the presupposition
is thus roughly “You are in some state of health”. Question [ib] would typically be
used to ask for your subjective evaluation of the concert, with such answers as It was
good / fair / disappointing /a disaster. An example like How does the house look now?,
which has a more specific verb than be, seems to allow a wider range of values for the
variable, though evaluative ones will still be the most likely; the presupposition is that
the house has some property relating to its appearance.
Example [15iia] again asks for evaluation (I found it enlightening /a waste of time).
The complex-transitive construction with like takes a very limited range of PCo s – black,
white, sweet, hot, with cream, etc., and the presupposition of [15iib] will be that you like
your coffee with some property from this range.
Questioning of a predicative (like that of a predicator: see §7.10) is much more re-
stricted than questioning an argument or circumstance. How occurs with relatively few
copulative and complex-transitive verbs, so that [15] may be contrasted with examples
like ∗How did she become?, ∗How did you think the concert?, etc. With strike it is only in the
interrogative that the PC can have the form of an adjective rather than a PP: compare
How does it strike you? and It strikes me as quite interesting.
(b) Adverbial degree modifier

[16] i [How old ]is your father? [modifier of adjective]
ii [How many]children have they got? [modifier of degree determinative]
iii [How seriously]are they taking his threat? [modifier of adverb]
iv How did you like the concert? [modifier of verb]
How modifies adjectives, degree determinatives, adverbs, and verbs to question degree,
extent, quantity. Only a very small number of verbs (like, enjoy, please, etc.) take how in
its degree sense: with other verbs, degree is normally questioned with how much, as in
How much do you care about him?
Whatever its category, the word modified by how is always gradable. The presuppo-
sition of the question then depends on the nature of the scale involved. Consider:
[17] a. How deep is the water? b. How shallow is the water?
While [b] presupposes that the water is shallow, [a] does not presuppose that it is deep –
only that it has a value on the scale of depth, a value which may or may not fall within
the range denoted by the unmodified adjective deep.
(c) Adjunct in clause structure, questioning means

[18] i A: How did you get in? B: By climbing through the kitchen window.
ii A: How is she going to pay for it? B: By cheque.
iii A: How can I remove it? B: With a razor-blade.
The presupposition is “You got in by some means / in some way”, and so on. This
represents the most usual sense of adverbial how when it modifies the verb.
(d) Adjunct in clause structure, questioning manner

[19] i A: How did she speak? B: With a strong French accent.
ii A: How does he drive? B: Rather recklessly.
The presupposition this time is “She spoke in some manner/way”, etc. Very often such
questions ask for evaluative judgements, as in [ii]. How can also occur as complement
with those verbs requiring manner specification: How do they treat you? This again invites
an evaluative answer, such as Well or Badly.
(e) Adjunct in clause structure, asking for evidence:

[20] i How does he know she is going to resign?
ii How can you be so sure that it was an accident?
This use is found with know and a few similar expressions, and serves to challenge what
has been said or implied. There is no presupposition “He knows (in some way) that she
is going to resign”: I am calling into question the contention that she is going to resign.
(f) Questioning reason in the it-cleft construction, and in the idiom how come
[21] i How is it you didn’t tell me before?
ii How come the fridge is switched off?
The use of how shown in [i] is restricted to the it-cleft construction: the closest non-
cleft counterpart has why rather than how: Why didn’t you tell me before? Like the latter,
[i] presupposes “You didn’t tell me before (for some reason)”.
§ 7.8 What 909
The idiom how come, a common alternant to why in informal speech, derives from
a construction where come is a verb taking a clausal subject (cf. How does it come to
be that the fridge is switched off ). But it is best regarded as a compound interrogative
word functioning as adjunct in a simple clause. Note, for example, that it is impossible
to insert that after come, a strong indication that the following NP (the fridge) has
been reanalysed as subject of a main clause. Moreover, the lack of any inflection on
come works against its construal as a verb with a 3rd person singular subject. How
come is nevertheless very exceptional as an interrogative expression in that it doesn’t
trigger subject–auxiliary inversion. It normally occurs in main clauses, but is not entirely
excluded from subordinate interrogatives: That’s how come they stay No. 1.
(g) In the idiom how about

[22] i How about another drink?
ii How about helping me with the washing-up?
iii How about we leave the others until next week?
iv I think it’s excellent; how about you?
How about belongs to informal style – with how’s about more informal still. The primary
use is to put forward a suggestion. This may have the indirect force of an offer (as in
a possible contextualisation of [i]), a directive (as in a likely interpretation of [ii]), and
so on. The complement of about can be an NP, a gerund-participial clause, or a tensed
declarative clause, as in [i–iii]. Besides its use in suggestions, the idiom can be used to
introduce a change to a related topic – e.g. from the speaker’s opinion to the addressee’s
in [iv]. What about is a possible variant except in [iii], where the complement is a tensed
clause.
(h) In certain non-question idioms

[23] i How do you do.
ii How dare you speak to me like that!
The boundary between direct and indirect speech acts is by no means sharply drawn, but it is
arguable that these are not questions indirectly conveying a greeting and a rebuke respectively,
but simply a greeting and a rebuke at the direct level. How do you do may be contrasted with
How are you?, which is used as a greeting, but is clearly also a question. I’m very well is an
answer to the latter, but I do very well is archaic (in the sense “I am in good health”), so that
[i] can no longer be regarded as defining a set of possible answers. Similarly it is difficult to
accept that any declarative clause could properly be regarded as expressing an answer to [ii]
or other rebukes beginning with how dare.
7.8 What
What occurs as determinative or pronoun, with respectively some or something as the
non-interrogative counterpart, and, in general, presuppositions of what questions can
be derived by substituting these:
[24] i a. What class is she in? b. “She is in some class”
ii a. What did the doctor say? b. “The doctor said something”
Determinative what contrasts with which and whose, while pronoun what contrasts
with these two and also with who/whom, when, and where: the meaning of what has been
discussed above in relation to these contrasts.
What in verbless constructions

[25] i A: Tom. B: What? A: Can you come here a moment?
ii A: I’ve just discovered something. B: What?
iii A: Kim’s just got a new job: B: So what?
In [i] what? serves to acknowledge a call; yes? would be a rather more polite alternative.
In [ii] it is elliptical: we understand “What have you just discovered?”; this elliptical use
is found with all the interrogative words. (What? can also be a repetition echo question,
and [25ii] is thus ambiguous between the ordinary question interpretation just given
and the echo question interpretation “What did you say?”) The idiom so what? is a
conventional way of expressing, none too politely, a lack of interest in what has just been
said: the meaning is “What is the relevance/significance of that?”.
What about? and what if?

[26] i A: The car’s in fine shape now. B: What about the tyres?
ii A: I’ve invited Peter. B: And what about Paul?
iii A: You know that knife I found? B: Yes, what about it?
iv What about a game of squash?
v What if we can’t get back in time?
What about is often used to introduce a new but related topic, as in [i–ii]; in [i] it serves to
challenge what has been said, the implicature being that the judgement fails, or may have
failed, to take account of the tyres. In [iii] the topic marked by about is not new, but old
information: I’m asking you to say something about the topic that you introduced. What
about can also be used to make suggestions, as in [iv]; how about could be substituted
for it here, and also in [i–ii]. The interpretation of [v] is along the lines of “What will
happen / will you do / shall I do if . . . ?”
7.9 Upward percolation of the interrogative feature: interrogative phrases

The markers of the open interrogative clause type are in the first instance words. We
must also recognise a category of interrogative phrase, however, for it is whole phrases,
not single words, that are affected by the fronting rule:
[27] a. She took which car? b. Which car did she take?
In [a] which car remains in situ, while in [b] it is fronted, but we can’t have fronting
of which alone: ∗Which did she take car? We will say, therefore, that which is an interrog-
ative word and which car an interrogative phrase. And since which car is an interrogative
phrase by virtue of containing the interrogative word which, we can speak metaphorically
of the interrogative feature as percolating upwards from the word which to the phrase
which car.
This is a useful way of speaking because such upward percolation may involve more
than one step, as we see from such examples as:
[28] a. What size shoes do you take? b. How big a hole did it make?
§ 7.9 Upward percolation of interrogative feature 911
[29] a. NPINTERROG b. NPINTERROG
Det: Head: Mod: Head:

NPINTERROG N AdjPINTERROG NP
Det: Head: Mod: Head: Det: Head:

DINTERROG N AdvINTERROG Adj D N
what size shoes how big a hole

Both involve two steps. In [a] the interrogative feature first percolates up from the word
what to the NP what size, and then from this to the higher NP what size shoes. Similarly
in [b]: from how to the AdjP how big, and thence to the NP how big a hole. There may,
therefore, be one interrogative phrase within another, and when fronting applies it is the
topmost one that is fronted: compare ∗What size do you take shoes?; ∗How big did it make
a hole?
The question then arises: what determines how far upwards the interrogative feature
percolates? How far up the tree do we have to go before we come to the maximal inter-
rogative phrase, i.e. the one that is fronted? The examples in [28] share two properties:
[30] i The maximal interrogative phrase is the highest phrase beginning with the in-
terrogative word.
ii The maximal interrogative phrase is an element of clause structure.
In [28a], for example, what size shoes is higher in the tree than the other phrase beginning
with what, namely what size; and what size shoes is a clause element, namely object,
whereas what size is not (it is determiner in NP structure).
Usually, both these properties obtain – but not always:
[31] i In which drawer do you keep the bank statements? [fronted preposition]
ii Which drawer do you keep the bank statements in? [stranded preposition]
In [i] the fronted interrogative phrase is in which drawer; again the upward percolation
involves two steps, from which to the NP which drawer, and then from this to the PP
in which drawer. But although property [30ii] applies (in which drawer is a locative
complement), property [30i] does not, for the maximal interrogative phrase begins
with the preposition. In [31ii], by contrast, the upward percolation stops at the NP which
drawer, which begins with the interrogative word but is not an element of clause structure,
being complement of the preposition in. Here, then, the preposition is stranded in situ
while its complement is fronted. In general, informal style conforms to [30i], formal
style to [ii].
We should also note that a PP containing an interrogative phrase as complement will
not always be an element of clause structure – it may be the complement in NP or AdjP
structure, as in:
[32] i a. Which country was she the president of ? [informal]
b. Of which country was she the president? [formal]
c. ∗The president of which country was she?
ii a. What subjects are you interested in? [informal]
b. In what subjects are you interested? [formal]
c. ∗Interested in what subjects are you?
Informal style continues to follow principle [30i], but the formal style conforms to neither
[30i] nor [30ii]: it represents a compromise between the two principles. If the highest
phrase beginning with the interrogative word is complement of a preposition, upwards
percolation proceeds towards an element of clause structure as far as is consistent with
the fronted interrogative phrase being a PP. This rather complex formulation is needed
to exclude examples like [32ic/iic], where the inadmissibly fronted interrogative phrase
is not a PP, while nevertheless allowing such structures as:
[33] To the daughter of which famous statesman was he engaged?
Here the fronted interrogative phrase is not the PP containing which famous statesman
as complement, but the next higher PP. Upward percolation of the interrogative feature
here involves four steps: from which to the NP which famous statesman, then to the
of PP, then to the NP the daughter of which famous statesman, and finally to the fronted
PP itself. The contrast between the fronted and stranded preposition constructions
applies much more widely than in interrogatives, and is discussed in more detail in
Ch. 7, §4.1.
7.10 Upward percolation of the question variable: the questioned element

The upward percolation of the grammatical feature interrogative has an analogue at the
semantic/pragmatic level, as illustrated in:
[34] A: Which team do you support? B: United.
From a grammatical point of view, as we have seen, the fronting rule applies to the
whole NP which team, not just to the interrogative word which. Similarly, from the
point of view of the meaning, the answer substitutes United for the whole NP which
team. We will therefore use the same metaphor, and speak of the question variable as
percolating upwards from the interrogative word to the phrase containing it, and we will
refer to which team as the questioned element, the one for which the answer provides a
replacement, giving a value for the variable.
Although there is this similarity between form and meaning, quite extensive differ-
ences are to be found in the way the upward percolation works at the two levels. One
small difference applies, in fact, to example [34], for a very limited subset of answers
to A’s question involve replacement of the interrogative determinative alone: this team,
your team, etc. We will consider the major differences between syntactic and semantic
upward percolation under three headings: determinative whose, PPs, and predications.
(a) Determiner whose

[35] A: Whose father is on duty today? B. Kim’s.
The answer, without ellipsis, is Kim’s father is on duty today, where it is just whose that is
replaced. The syntactic interrogative phrase is whose father, but the semantic questioned
element is just whose. Note that Mr Roberts is not a possible answer, even in a context
where it is known that Mr Roberts is Kim’s father. With determiner whose, upward
percolation is obligatory at the syntactic level, but barred at the semantic level.
(b) PPs
[36] A: To whom are you referring? / Who are you referring to?
B: (I’m referring to) your mother.
§ 7.10 Upward percolation of question variable 913
The answer retains the preposition, making a substitution just for its NP complement.
This is the usual case: in general there is no semantic percolation into a PP – and hence
it is in the informal preposition stranding construction that we find the closer match
between form and meaning.
There are nevertheless some exceptions, cases where the answer involves replacement
of the whole PP:
[37] i a. A: What are you closing the window for?
B: (I’m closing the window) because I’m cold / to cut out the noise.
b. A: What’s the new boss like?
B: (The new boss is) quite pleasant.
ii a. A: Under what conditions would you take on the job?
B: (I’d take on the job) if they gave me adequate support staff.
b. A: In what way can I help you?
B: (You can help me) by minding the children for a couple of hours.
In [i] we have the idioms what . . . for and what . . . like: the answers must provide a
replacement for the whole PP. The same applies with certain very general PPs, such as
under what circumstances/conditions and in what way in [ii], but this time the syntax
matches, having obligatory percolation into the PP.
(c) Predication
[38] i A: What did you do? B: I called the police.
ii A: What happened? B: The car rolled into the ditch.
The syntactic interrogative phrase is what, but the semantic questioned element is in [i]
the VP, the predicate, and in [ii] the whole clause. There is no verb among the interrogative
words, and hence to question the predication it is necessary to use a very general verb
together with the pronoun what; syntactically there is no percolation beyond the what,
but semantically there is. Do is used to question the predicate, happen the whole clause.
Just as the answer in [34] implicates “United is a team”, so the answer in [38i] implicates
“For me to call the police is for me to do something”. This is why I was older than my sister,
I was seen by one of the guards, etc., are not possible answers: the implicit propositions
“For me to be older than my sister / seen by one of the guards is for me to do something”
are not true. An answer to such a do question must denote a situation which is dynamic
rather than static, and one in which the subject-referent has an agentive role (cf. Ch. 17,
§7.6).14 Similarly, the answer in [38ii] implicates “For the car to roll into the ditch is
for something to happen”. Happen also denotes a dynamic situation, but it need not be
agentive; I was older than my sister is thus again not a possible answer to [38ii], but I was
seen by one of the guards is.
Such predication questions may include specification of various circumstances of the
situation denoted by the answer: What did you do in the morning?; What happened to
make you change your mind? They may also include specification of an argument:
[39] i A: What did you do to/with my hat? B: I dropped it in the mud / put it away.
ii A: What happened to your father? B: He was taken away for questioning.
14
An exception is the idiom ‘What be X doing Y? ’ (where Y is a locative or a gerund-participial): What are your
gloves doing on my desk?; What are you doing sleeping in my bed? These convey “Why are your gloves on my
desk?” and “How come you are sleeping in my bed?” – with the suggestion that the gloves shouldn’t be there,
that you shouldn’t be sleeping in my bed.
With do, the preposition to indicates that the complement NP is associated with an
affected, patient role, as in B’s I dropped it in the mud. Another answer here might be I
sat on it, where ‘it’ does not inherently have an affected, patient role, but is interpreted
as having such a role because for I sat on it to be an answer to What did you do to my
hat? implicates “To sit on your hat is to affect it”. With involves disposing of something,
putting it somewhere: I put it away / gave it to Kim. With happen, to indicates a less
specific participant role: it will often be an affected patient, as in B’s answer in [ii], but
it can also be an agent (He escaped through the bathroom window). Become can be used
with an of phrase in a related way (A: What became of his sister? B: She went to China).
7.11 Open interrogatives as an unbounded dependency construction

The open interrogative is what is known as an unbounded dependency construction,
as illustrated in the following examples:
[40] i Whati did he [buy i ]?
ii Whati did she [say [he bought i ]]?
iii Whati do you [think [she said [he bought i ]]]?
iv Whati do you [think [she said [he wanted [to buy i ]]]]?
In all of these what occupies initial (prenuclear) position in the open interrogative clause
but its core role is that of object of the verb buy. The buy clause is embedded within the
interrogative clause and it will be evident from these examples that there is no limit as
to how deeply embedded it may be. The paired brackets indicate clause boundaries, and
by adding verbs that take clausal complements we can increase the number of clause
boundaries between what and buy. The dependency relation between what and buy is
thus unbounded: there is no grammatical limit on how many clause boundaries may
separate them.
There are a number of constructions of this kind, including exclamatives and relatives.
We discuss the general properties of this kind of construction in Ch. 12, §7, and here we
will merely note that we show the relationship between what and buy by co-indexing
what with a gap ( ) located in the position of direct object to buy (cf. Ch. 2, §2). This
gap pre-empts the filling of the direct object function by any other NP: ∗What did he buy
some meat?, etc.
7.12 Ambiguities concerning the role of an interrogative phrase

in complex clauses
Two kinds of ambiguity can arise when an open interrogative clause has another clause
embedded as complement within it. One concerns the core role of a fronted interrog-
ative phrase, the other the interrogative role of an interrogative phrase that remains in
situ.
Core role ambiguities resulting from fronting

Examples such as the following are ambiguous in that the interrogative phrase may have
its core role in either of two clauses:
§ 7.12 Ambiguities: interrogative phrases in complex clauses 915
[41] i a. When did they decide to leave?

b. Wheni did they [decide [to leave] i ]? [gap in decide clause]
c. Wheni did they [decide [to leave i ]]? [gap in leave clause]
ii a. Why do you think he lied?
b. Whyi do you [think [he lied ] i ]? [gap in think clause]
c. Whyi do you [think [he lied i ]]? [gap in lie clause]
In [ia] when may question the time of deciding (“When did they make the decision
to leave?”) or the time of leaving (“What was the time of leaving they decided on?”).
Similarly in [iia] why can be reason adjunct in the think clause (“What makes you think
he lied?”) or in the lie clause (“In your opinion, why did he lie?”).
Such ambiguities most often involve adjunct interrogative phrases, as in these exam-
ples. Under quite restricted conditions, however, ambiguity may arise involving a more
central core function:
[42] i Who do you expect to play?
ii Whoi do you [expect i [to play]]? [gap as object of expect]
iii Whoi do you [expect [to play i ]]? [gap as object of play]
The question here is ambiguous according as it concerns your expectation as to who will
play or who you will play: answers for the two interpretations might be respectively I
expect Jones to play and I expect to play Jones. For reasons given in Ch. 14, §2, we analyse
I expect Jones to play as having Jones as object of expect, whereas I expect to play Jones has
Jones as object of play: on this account, therefore, the ambiguity in [i] results from the
fact that who can be syntactically object of expect or of play.
The ambiguity in [42] is clearly dependent on the interrogative phrase being fronted,
since the core positions are different. This is typically not so with the adjunct examples.
The ambiguity of [41ia], for example, is found also in They decided to leave when?, just
as it is in declarative They decided to leave yesterday.
Interrogative role ambiguities in multi-variable constructions

The ambiguities illustrated above have to do with the core role of an interrogative phrase but
it is also possible for there to be ambiguity concerning its interrogative role. Such ambiguity
can arise when one open interrogative clause is embedded within another and at least one of
them is multi-variable:
[43] i Who knows which universities offer the best courses in which subjects?
ii Kim knows which universities offer the best courses in medicine and Pat knows which
universities offer the best courses in law.
iii Kim knows which universities offer the best courses in which subjects.
The core functions of the three interrogative phrases in [i] are clear: who belongs in the know
clause as subject, while which universities and which subjects belong in the offer clause as
subject of the clause and complement of in respectively. The issue is: which of the two clauses
is the multi-variable one?
Who clearly expresses a variable with respect to the know clause: we will say that who
(or, more precisely, the interrogative component of who) has the know clause as its scope.
Similarly for which universities in the offer clause: it has the offer clause as its scope. The
issue can thus be put more specifically as: what is the scope of the interrogative component
of which subjects? Is the variable it expresses associated with the know clause (making the
main clause the multi-variable one) or with the offer clause (making the subordinate clause
the multi-variable one)? The ambiguity can be brought out by using variables just for the
propositional content of the know clause:
[44] i “x knows which universities offer the best courses in y”
ii “x knows which universities offer the best courses in which subjects”
Thus [i] corresponds to the reading where the know clause is multi-variable: this is the one
for which [43ii] is an answer, since it supplies paired values {“Kim”, “medicine”} and {“Pat”,
“law”} for the pair of variables {“x ”, “y ”}. Different people are assumed to be informed about
different fields, and the answer supplies a pairing of people with fields. By contrast, [44ii]
matches the reading where the know question is a single-variable one: this is the one for which
[43iii] is an answer, since it supplies a value for the variable “x ”. Here the presupposition is that
someone is informed about the whole range of fields, and the answer identifies such a person.
From a formal point of view, the ambiguity is attributable to the fact that only one
interrogative phrase can undergo fronting. In single-variable constructions the fronting of
the interrogative phrase shows clearly which clause its interrogative role is associated with:
[45] i She will say what i she saw i. [see clause interrogative]
ii What i will she say she saw i? [will clause interrogative]
In both [i] and [ii] what is at the front of the clause over which it has scope. In multi-variable
constructions only one phrase can have its scope marked in this way: the scope of the other(s)
is not overtly signalled. Hence the scope ambiguity of [43i].
7.13 Modification of interrogative words

Interrogative phrases may contain a limited range of modifiers, as illustrated in:
[46] i a. [What ever] did you do that for? b. [Why ever] would he do that?
ii a. [What the hell ] is she trying to do? b. [Who on earth] can that be?
iii a. [Who else] will be there? b. [What exactly] do you mean?
The items in [i–ii] express surprise or bafflement, and hence suggest that the speaker does
not know the answer to the question. They tend to emphasise the open-endedness of the
set of possible values for the questioned variable; as a result they are hardly compatible
with which, for this involves selection from an identifiable set. They do not contribute
to the propositional meaning, and we will refer to them as emotive modifiers. Variants
of hell in [iia] are heck, blazes, deuce, dickens, fuck, etc.
These modifiers are constrained to occur immediately after an interrogative word
in head, not dependent, function: ∗[What on earth reason] could they have for rejecting
it?; ∗[How ever much]did that cost? The interrogative phrase must be in initial posi-
tion: contrast ∗And after that you went [where the hell]? Except in combination with why,
ever is found written as part of the interrogative word: Whatever did you do that for? 15
Interrogatives are among those words that accept else as postmodifier, as in [46iiia].
Else is semantically like other (“who other than those mentioned”); it licenses a than
complement (What else than this?). The adverbs exactly and precisely may follow the
head, as in [46iiib], or precede: Just exactly who do you think you are?
15
The more conservative usage manuals say that ever should always be written separately in this interrogative
construction, but that is in conflict with actual usage. This ever is semantically distinct from the ever that
occurs – always as part of a compound – in fused relatives ([Whoever said that] was mistaken) and the
exhaustive conditional construction (I won’t sell [however much you offer]).
§ 7.14 Complex-intransitive interrogatives 917
7.14 Complex-intransitive interrogatives: S–P–PC vs PC–P–S

A main clause of the form ‘NP1 be NP2 ’, where NP1 is interrogative, may have one of two
structures: S–P–PC or PC–P–S. Compare:
[47] S P PC PC PS
a. Who is editor of the magazine? b. What time is it?
Example [a] represents the default order of clause elements, while [b] arises through
fronting of the interrogative phrase and subject–auxiliary inversion. Answers might be,
respectively, Kim is and It’s five o’clock, with the interrogative phrase replaced by the
subject Kim or predicative complement five o’clock. The structures can be distinguished
by manipulating them in any of the three ways shown in:
[48] interrogative S interrogative PC
i a. Who will be editor of the magazine? b. What time will it be?
ii a. Who did he say was editor of the magazine? b. What time did he say it was?
iii a. Ask who is editor of the magazine. b. Ask what time it is.
In [i] we have added modal will, so that the positions of editor of the magazine in [a] and
it in [b] are now differentiated: it occurs after auxiliary will, indicating that it is subject,
while editor of the magazine follows be, showing that subject–auxiliary inversion has not
applied. In [ii] the interrogative phrase has its core role in a subordinate clause: in [a] it
is linked to a gap in subject function (Whoi did he say [ i was editor of the magazine]?),
and in [b] to a gap in predicative complement function (What timei did he say [it was
i ]?). In [iii] the interrogative clauses are subordinate, so subject–auxiliary inversion is
inapplicable: as a result it is clear that the subject of is is who in [a] but it in [b].
Example [47a] is itself unambiguous, as the post-verbal NP lacks the determiner that
it would need if it were in subject function. Other features that may clearly signal the
structure are verb agreement and case:
[49] interrogative S interrogative PC
a. Which is me? b. Which am I?
Here is agrees with the subject which in [a], while am agrees with the subject I in [b].
Accusative me can only be PC; nominative I can occur as subject or (in formal style)
as PC and hence does not serve to distinguish the structures. Which is him? is therefore
S–P–PC, while Which is he? is ambiguous.
Ascriptive and specifying be

Where be is interpreted ascriptively, the PC–P–S structure is somewhat more likely, but it will
be pragmatically clear from the content which NP is subject:
[50] i Who is a friend of John’s? [interrogative S (S–P–PC)]
ii What is your uncle? [interrogative PC (PC–P–S)]
These may be glossed roughly as “Who belongs in John’s circle of friends?” and “What does
your uncle do?”
Where be is used in the specifying sense, either structure will generally be possible, and it
will usually make no significant difference whether it is taken as one or the other:
[51] What was the cause of the delay? [PC–P–S or S–P–PC]
The tests of [48] allow either analysis: compare What do you think was the cause of the delay?
or What do you think the cause of the delay was? ; We asked what was the cause of the delay or
We asked what the cause of the delay was ; and so on.
8 Exclamatives and exclamations
8.1 The syntax of exclamatives

Exclamatives, we have said, are marked by one or other of the exclamative words how
and what. These enter into the structure of an exclamative phrase, which is fronted when
it is not subject:
[1] exclamative subject exclamative non-subject
i a. How much remains to be done! b. How she hated it!
ii a. What strange people inhabit these parts! b. What a disaster it was!
Where the exclamative phrase is subject, the order is the same as in matching declaratives:
compare Much remains to be done and Strange people inhabit these parts. Where the
exclamative phrase is non-subject, the fronting is obligatory: there is no exclamative
analogue of open interrogatives like And the results of the examination were what?
Like the open interrogative (cf. §7.11), the exclamative is an unbounded dependency
construction, so that the initial exclamative phrase can be linked with a gap in a subor-
dinate clause embedded within the exclamative clause, as in:
[2] i How impossibly politei she expected them [to be i ]!
ii What a waste of timei they thought [it was likely [to be i ]]!
As evident from the above examples, the exclamative feature percolates upwards in
the same way as the interrogative feature; in [1iia], for example, it goes from what to the
NP what strange people, while in [2i] it goes from how to the AdvP how impossibly, and
thence to the AdjP how impossibly polite. It may percolate up into a PP or stop at the NP
complement, leaving the preposition stranded:
[3] i With what unedifying haste he accepted the offer! [fronting of preposition]
ii What unsavoury people he associates with! [stranding of preposition]
Ambiguity between exclamative and open interrogative
How and what can be either exclamative or interrogative, and in abstraction from the
prosody/punctuation such examples as [1ia/iia] are ambiguous between an exclamative
reading (“A remarkably large amount remains to be done”; “Remarkably strange people
inhabit these parts”) and an open interrogative reading (“What is the amount that
remains to be done?”; “Who are the strange people that inhabit these parts?”). Where
the clause is not ambiguous, as in the other examples given above, this is due to the
distributional differences between exclamative and interrogative how and what or to
differences in the order of elements in the clause. We will consider the exclamative words
in the next two subsections, and then take up the issue of order.
Whereas there are a fair number of interrogative words, the exclamative class has
only two members, how and what: clauses like Who remains to be seen? or Which strange
people inhabit these parts? are unambiguously interrogative. Like interrogative words,
exclamative how and what have a dual role: on one dimension they are markers of
exclamative clause type, but they also have what we are calling a core role. And in
their core role they show some differences from their interrogative counterparts in both
grammar and meaning.
§ 8.1.1 Exclamative how 919
8.1.1 Exclamative how

We will consider two uses of how, both adverbial; the first is closely parallel to interro-
gative how, the second quite different.
(a) Modifying an adjective, degree determinative, or adverb

How commonly functions as a degree modifier. Compare the following exclamative and
interrogative examples:
[4] exclamative open interrogative
i a. How tall they are! b. How tall are they?
ii a. How much time we wasted! b. How much time did we waste?
iii a. How quickly it grows! b. How quickly does it grow?
iv a. How very tactful he is! b. ∗How very tactful is he?
In both constructions we are concerned with degree: with exclamative how the degree
is remarkably great, with interrogative how it is to be indicated in the answer. Two
differences are to be noted. In the first place, as illustrated in [iv], exclamative how
can modify another degree modifier such as very, absolutely, remarkably, etc., whereas
this is not possible with interrogative how. Secondly, while [ia] says that they are tall
(remarkably tall), [ib] does not presuppose that they are tall, only that they have some
degree, small or large, on the scale of tallness.
(b) Modifying a verb

[5] i a. How they deceived her! b. How did they deceive her?
ii a. How I hated it! b. #How did I hate it?
Here we find a sharp difference: exclamative how is again concerned with degree, whereas
interrogative is rarely used in that way (cf. §7.7 above). Exclamative [ia] means “They
deceived her greatly / to a remarkable degree or extent”, while interrogative [ib] means
“By what means / In what way did they deceive her?”
8.1.2 Exclamative what

What occurs in NPs with a following noun head: there is no exclamative counterpart to
the interrogative what of What did they bring with them? In NPs with a noun head, we
find the following patterns:
i a. What a game it was! b. What game was it? [count singular]
ii a. What games he played! b. What games did he play? [plural]
iii a. What music he played! b. What music did he play? [non-count]
With count singulars, the NPs are overtly distinct, as shown in [i]: the exclamative
has what a, the interrogative just what. In other cases, i.e. with plurals and non-count
singulars, as in [ii–iii], there is no overt distinction in the form of the NP itself. We
take interrogative what to be a determinative functioning as determiner; exclamative
what , by contrast, is an adjective functioning as external or internal modifier (see Ch. 5,
§§7.1.3, 12).
Interrogative what is concerned with identity: answers to questions [6iia/iiia] will

identify the games/music (e.g. He played cricket and tennis / Schubert’s octet). Excla-
mative what, by contrast, is concerned with quality and degree: [iia/iiia] indicate that
the games/music he played were remarkable for their quality (though whether the
speaker approves or disapproves is not indicated: the quality may be remarkably good
or remarkably bad). However, if the head noun is gradable, the difference is like that
described above for how in [4i–iii]: compare What a size it was! (its size was remarkable)
and What size was it?
Differences between exclamative what and how

Style
How is associated with a somewhat formal style, especially in main clauses: How well she
plays!, for example, is appreciably more formal than What a fine player she is!
Distribution
What occurs only as modifier in NP structure. How modifies adjectives, degree determi-
natives, adverbs, and verbs; it does not function immediately in NP structure, but phrases
containing it can occur as external modifier in count singular NPs. (The distributional
difference is similar to that between such and so: see §8.3.) Compare, then, the following
exclamative phrases, which might occur in such a frame as ‘ we have on our hands!’
[7] i a. what a difficult problem b. how difficult a problem [count singular]
ii a. what difficult problems b. ∗how difficult problems [plural]
iii a. what difficult work b. ∗how difficult work [non-count]
Although what is not directly in construction with the adjective difficult, it nevertheless
indicates a remarkable degree of the property expressed by it, so that with count singulars
the meaning of [a] is effectively the same as that of [b]. For plurals and non-count
singulars, only the what construction is available.
8.1.3 Position of the subject

The normal position for the subject in exclamatives is before the predicator. A major
grammatical difference between main clause exclamatives and open interrogatives is
thus that whereas fronting of an interrogative phrase is obligatorily accompanied by
subject–auxiliary inversion, fronting of an exclamative phrase is not:
a. What a mistake they made! b. What mistake did they make?
Subject–auxiliary inversion in exclamatives
However, subject–auxiliary inversion is available as an option in exclamatives, though it
is relatively infrequent and characteristic of fairly literary style:
[9] i How much clearer does it seem now that you have explained it yourself!
ii How rarely does one see such chivalry nowadays!
iii What a row would there have been if they had known you were here!
The effect of course is to make the structure more like an interrogative, so that from a
grammatical point of view there will often be ambiguity as to clause type. In abstraction
from the prosody/punctuation, for example, [9i] can be either exclamative (“How much
clearer it now seems”) or interrogative (“To what extent does it now seem clearer?”).
§ 8.1.4 Verbless exclamatives 921
Example [iii] is unambiguously exclamative because what is here in a count singular

NP, where the constructions are distinct. Inversion is not possible with how when it is
modifying the verb, as in [5ia] (How they deceived her!): the inverted How did they deceive
her? is unambiguously interrogative.
When the clause is grammatically ambiguous, it is not always as clear as one might expect
which is the contextually appropriate reading. Consider, for example:
[10] How often have I told you not to put your feet on the sofa!
The most salient interpretation of this is as a rebuke and directive (“Don’t put your feet
on the sofa”), but in spite of the exclamatory prosody/punctuation this corresponds to the
interrogative reading. In this interpretation the force at the direct level is that of question, but
it is a rhetorical question which indirectly conveys “I have told you very many times not to put
your feet on the sofa”; this is close to the meaning of an exclamative, but not quite the same.
An unco-operative addressee could respond with, say, forty-five times, thereby answering the
question at the direct level. Note, moreover, the clear style difference between [10] and [9ii]:
the latter suggests the rather formal style that is characteristic of exclamative how, whereas
[10] does not.
Subject postposing
It is also possible for the subject to be postposed when the exclamative phrase is an
adjectival predicative:
[11] How great would have been her disappointment if she had known what they had
actually thought!
8.1.4 Verbless exclamatives

An exclamative clause often consists of just the exclamative phrase (or of this plus a
coordinator or the like), normally an NP or an AdjP:
[12] What nonsense! What an insensitive way to behave! What a strange thing for
him to say! How fantastic! How incredibly unlucky!
The infinitival clauses here are relatives. The exclamative phrase is interpreted as a pred-
icative complement, with be and the subject understood: “What nonsense that is!”. It is
also possible to omit just be, with an overt subject in final position: What a terrible thing,
that ‘wailing wall’ in Berlin!
Another frequent verbless construction consists of an exclamative phrase followed by
a declarative content clause, infinitival, or (less frequently) a gerund-participial:
[13] How strange that nobody noticed the error! What a coincidence that they were on
the same bus! How kind of you to let me know! What a nuisance having to do
them all again!
These can be regarded as involving the omission of it + be: compare How strange it
is/was that nobody noticed the error! The subordinate clause thus functions as extraposed
subject – and the infrequency of the gerund-participial reflects the fact that such clauses
do not readily undergo extraposition.16
16
An idiomatic verbless exclamative is and how!, added (in informal style) to what has just been said (by the
same or another speaker) as an exclamatory intensifier: She can certainly play the piano! – And how!
8.2 Meaning and use of exclamative main clauses

Exclamative utterances normally have the force of exclamatory statements. Consider
again our initial example:
[14] How much remains to be done! (=[1ia])
At the most general level of classification we include this in the statement family because of
its resemblance to the straightforward statement Much remains to be done. The meaning
of the latter is contained within that of the exclamative.
Nevertheless, exclamative utterances are by no means prototypical statements. The
exclamatory component gives them a strongly subjective quality, so that they are not
presented as statements of fact. Rather, they express the speaker’s strong emotional
reaction or attitude to some situation. The exclamative indicates that the situation obtains
(e.g. [14] indicates that much remains to be done), and this is the statement component of
the meaning – but this component is backgrounded relative to the emotive exclamatory
component. Several more specific properties follow from this general account.
In the first place, because the statement component is backgrounded it is presented as
uncontroversial, not at issue. Thus one normally doesn’t envisage disagreement, dispute.
This is reflected in the use of interrogative tags:
[15] i What a disaster it was, wasn’t it! [reversed polarity tag]
ii ?What a disaster it was, was it! [constant polarity tag]
Example [i] (with falling intonation on the tag) is possible because such a tag can be
used to seek agreement – but note that it is agreement with the subjective attitude
(that the situation is remarkable), not just the statement component. However, [ii]
would not normally be used because the constant polarity tag in this case would be
seeking acknowledgement of the statement component, which is inconsistent with its
background status.
Secondly, exclamatives do not give answers to (non-echo) questions: How I enjoyed
it! is not an answer, and a very unlikely response, to Did you enjoy it? In part, this
relates to what we have said about the backgrounded status of the statement component:
information giving the answer to an explicit question will normally be foregrounded.
But a more important reason is the semantic one that exclamative how and what express
variables rather than constants. The construction indicates that the value of the variable
is remarkable, but does not explicitly specify what it is. This is why exclamative utterances
are not naturally assessed as true or false. Thus one would hardly express agreement by
saying Yes, that’s true : more normal would be something like Yes, indeed, or – where the
exclamative phrase is non-subject – a reduced interrogative, such as Yes, didn’t she! in
response to How she hated it!
The resemblance between exclamatives and open interrogatives

The property of expressing a variable is of course common to both exclamative and
interrogative uses of how and what, and it is a very widespread phenomenon in the
world’s languages that exclamative clauses bear strong formal resemblances to open
interrogatives. (Because subject–auxiliary inversion generally does not apply in English
exclamatives, their resemblance is greater to subordinate interrogatives, and this too
is illustrative of the most common pattern.) We have seen that closed interrogatives
§ 8.3 Non-exclamative exclamations 923
like Wasn’t it a disaster! or Did she hate it! are also commonly used as exclamatory
statements, but with these the force is indirect: at the direct level, they are questions.
In the exclamative construction, however, the exclamatory statement force has been
grammaticalised: in spite of the resemblance, exclamatives are grammatically distinct
from open interrogatives, as we have noted above, and they do not belong in the semantic
category of question.
Exclamatives have a narrower range of uses than the other major clause types. And
there are no cases where they are conventionally used with some indirect force, as various
kinds of declarative and interrogative are used as indirect directives, or imperatives as
components of conditional statements, and so on.
8.3 Non-exclamative exclamations

Not all exclamations take the form of exclamative clauses. The concept of exclamation is,
moreover, a somewhat nebulous one, and it is not possible to present a well-defined set
of grammatical constructions that express exclamatory meaning; very often, of course,
it is signalled prosodically rather than, or as well as, by the lexicogrammatical form. We
give here a sample of structures that are characteristically associated with such meaning.
(a) Closed interrogatives

As we have noted, closed interrogatives such as negative Isn’t it cold! or positive Is it
cold! can be used as rhetorical questions indirectly conveying exclamatory statements:
the implicit meaning is close to that of the positive exclamative How cold it is!
(b) So and such

These words closely match how and what in their grammatical distribution, except that
they are not exclamative and hence the phrase containing them is not obligatorily fronted.
Compare, then, the following with [1] above:
[16] i a. So much remains to be done! b. She hated it so!
ii a. Such strange people inhabit these parts! b. It was such a disaster!
So resembles how in that it can modify an attributive adjective only in count singular
NPs with following a, while such occurs in the three main kinds of NP, like what. The
following therefore match [7] above, but would not be fronted, so that they might occur
in the frame ‘We have on our hands! ’:
[17] i a. such a difficult problem b. so difficult a problem
ii a. such difficult problems b. ∗so difficult problems
iii a. such difficult work b. ∗so difficult work
So and such are not markers of a distinct clause type since they can occur in any
of the major clause types – closed interrogative (Have you ever seen such chaos?), open
interrogative (Why do you torment me so?), imperative (Don’t be such a wet blanket!),
as well as the above declaratives. There is nevertheless a slight difference of meaning
in that these non-declarative examples are implicitly comparative (“Have you ever seen
such chaos as this?”; “Why do you torment me as you are doing?”; “Don’t be such a wet
blanket as you are being”), whereas the examples in [16] are not. The non-comparative
use is more clearly exclamatory, and does not readily occur in non-declaratives. It is not
altogether excluded, however: they can occur in interrogatives used with the indirect
force of an exclamatory statement, as in Haven’t they such charming manners!
(c) Extraposable NPs

NPs with the form the + . . . N + integrated relative clause / of phrase can stand on their
own as exclamations:
[18] i The money he spends on clothes!
ii The cost of these clothes!
iii The way he treats his wife!
We call these ‘extraposable NPs’ because they can appear in extraposed subject po-
sition with predicates such as amazing: It’s amazing the money he spends on clothes /
the cost of these clothes; It’s a scandal the way he treats his wife. The verbless examples in
[18] are understood in much the same way as these with some generalised exclamatory
predicate understood; the attitude implied is usually one of disapproval. An alternative
to the + N + relative is the fused relative construction: What some people will do to save
a few dollars! (compare: The things some people will do . . . !).
(d) Imprecative retorts

[19] A: I’ll invite them round for dinner. B: Like hell you will!
The structure consists of an expletive + personal pronoun subject + auxiliary with
anaphoric ellipsis of the complement.
9 Imperatives and directives
9.1 Subtypes of imperative clauses

The grammatical properties which together define the class of imperative clauses in
English were summarised in §2. The most important points are:
[1] i The subject is an optional rather than obligatory element.
ii The verb is in the plain form.
iii Supportive do is used in relevant constructions with be, not just lexical verbs.
Ordinary imperatives vs let-imperatives
The main syntactic division within the class is between ordinary imperatives (the default
subclass) and let-imperatives:
[2] ordinary imperative LET-imperative
i a. Open the window. b. Let’s open the window.
ii a. Please let us borrow your car. b. Let’s borrow Kim’s car.
Let-imperatives are marked by a special use of let distinct from the normal use with the
sense “allow”. Let in the “allow” sense is found in all clause types (cf. declarative He let
us borrow his car), including ordinary imperatives like [iia]. The let of [ib] and [iib] has
been bleached of this meaning and serves as a marker of this special type of imperative
construction.
§ 9.2 Ordinary imperatives 925
Subtypes of let-imperative: 1st person inclusive vs open

The [b] examples in [2] are, more specifically, 1st person inclusive let-imperatives.
These contain an accusative form of we which can be, and usually is, contracted to
’s, and whose reference normally includes the addressee(s) as well as the speaker. In
[2ib/iib], for example, I’m proposing that you and I (or one of us) open the window and
borrow Kim’s car. (The us of [2iia], by contrast, is exclusive: it refers to me together with
one or more other persons, excluding you.) The special use of let is also found in what
we call open let-imperatives:
[3] i If that is what the premier intends, let him say so.
ii Let that be a lesson to you.
iii Since I/we/you did most of the work, let me/us/you receive the credit.
These usually have 3rd person reference, as with him and that in [i–ii], but in principle
the full range of person–number combinations is permitted, as illustrated in [iii].
The 1st person inclusive let-imperative is certainly grammatically distinguishable
from ordinary imperatives (by the potential contraction of us to ’s), but the status of
the open subtype is much more problematic: it could be argued that it simply involves
a semantically special use of let and is not grammatically distinct from the ordinary
imperative construction. It will nevertheless be convenient to treat them separately,
after first examining clear cases of ordinary imperatives and then 1st person inclusive
let-imperatives. (For simplicity we will generally, in discussing ordinary imperatives,
leave the ‘ordinary’ understood.)
9.2 Ordinary imperatives

9.2.1 Omissibility of the subject
The prototypical imperative has no subject – and this of course immediately distinguishes
it from most other main clause types. Normally, such clauses are interpreted as though
they had you as subject: in Tell her the truth, for example, it is a matter of you, the
addressee(s), telling her the truth. This is reflected in the use of reflexive pronouns in
such examples as:
[4] a. Get yourself /∗you a new hat. b. Try to leave yourselves/∗you plenty of time.
The choice of the reflexive over the non-reflexive here matches that found in clauses
with you as subject, such as You never get yourself/ ∗you a new hat, You always try to leave
yourselves / ∗you plenty of time, and so on.
Imperatives with overt subject

Imperatives do not always have the subject missing, however. You itself can appear as
subject, or we may have a 3rd person NP:
[5] i You be wicket-keeper and I’ll bowl. [2nd person subject]

ii Somebody get me a screwdriver.
[3rd person subject]
iii All those in the front row take one step forward.
With a 2nd person subject, be is the only verb where the plain form and the present tense
are not syncretised; with other verbs, therefore, there is potential ambiguity between
imperative and declarative, as in You give the first lecture. As an imperative this would
be some kind of directive for you to give the first lecture, and as a declarative it would
be a statement about what you do (e.g. as part of some scheduled lecture programme).
With a 3rd person subject, imperative and declarative will in the singular always have
overtly distinct verb-forms, but in the plural again only with be. Thus [5ii] contrasts
with declarative Somebody gives me the screwdriver, whereas [iii] is ambiguous between
imperative (with directive force) and declarative (with statement force).17
3rd person subjects
The range of possible subjects is more limited in imperatives than in declaratives, though
it is questionable how far this is a matter of grammatical rule. The subject must normally
have personal denotation, and dummy or clausal subjects are thus categorically ruled out:
∗
There be no more talking ; ∗That he’s over 60 don’t be forgotten. The most likely 3rd person
subjects are the compound determinatives (someone, nobody, everybody, etc., alone or with
dependents – Everybody over here stay still; Anybody with a faulty disk please let me know),
other fused determiner-head constructions with of you as complement (Some/One of you
give me a hand with this trunk; Those of you who’ve finished please put up your hands), and
bare plurals (Passengers on flight QF2 please proceed to Gate 6; Gentlemen lift the seat). Definite
NPs with the are less likely, but possible (The boy by the door please turn on the light), and the
same applies to proper names (Kim move upstage a little). These 3rd person definites occur
somewhat more readily in coordinative constructions (You and Kim play on the other court;
You give the first four lectures and the others do the rest). Personal pronouns other than you
are very unlikely though they probably cannot be categorically ruled out, especially in the
coordinative constructions just illustrated.
2nd person subjects
Given that you can be omitted, why is it sometimes retained? One factor is the need
to mark contrast. In [5i], for example, you contrasts with I; compare, similarly, You do
the washing-up tonight please: Kim did it last night. Where there is no such contrast, the
addition of you has an emotive effect:
[6] i (Just) you watch where you put your feet.
ii You mind your own business.
iii You sit down and have a nice cup of tea; everything is going to be all right.
iv You go back and tell him you need more time.
Very often it contributes to a somewhat impatient, irritated, aggressive, or hectoring
effect, as in a natural use of [i–ii]. But [iii–iv] show that it can also have very much the
opposite effect of soothing reassurance, encouragement, support. Whether the effect is
of the first or the second kind will of course depend on the tone of voice, the content,
and the context. What the two cases have in common is perhaps that expression of you
emphasises the speaker’s authority. In the aggressive case, the you emphasises that I am
telling you, not asking you, to do something. In the reassuring case, I assume the position
of one who is assured, one who knows best what to do.
17
It should be borne in mind that declaratives can be used as indirect directives: directive force is thus no
guarantee that an ambiguous clause is construed as imperative. A common way of giving street directions, for
example, is illustrated in You take the first road on the right after the church, which is normally intended (and
pronounced) as a declarative, even though it is (indirectly) a directive. Note that the contextual conditions
for an imperative with overt subject would not normally apply to such cases – see the final paragraph of this
section.
§ 9.2.2 Subject vs vocative in imperatives 927
9.2.2 Subject vs vocative in imperatives

In declarative clauses we find a very sharp grammatical distinction between the subject
and a vocative (i.e. the underlined element in We need to talk this over, my boy or Kim,
I’m just slipping out to the shops). One obvious factor is that the subject is obligatory
whereas the vocative is optional. It follows that if only one of them is present it can only
be the subject. In Someone in the back row is not in tune, for example, someone in the back
row must be subject: there is no possibility of it being vocative.
In imperatives the two functions are less sharply distinct because they are both
optional, so that it is possible to have either element without the other:
[7] i Nobody move. [subject]
ii Kim, dear, just come and see what I’ve found. [vocative]
iii Someone in the back row(,) please turn on the fan. [subject or vocative]
Nobody in [i] is unambiguously subject because a vocative can’t be negative, and in [ii]
dear marks the NP as unambiguously vocative, but in abstraction from punctuation and
prosody someone in the back row in [iii] could be either.
Apart from the grammatical factor, there is a pragmatic reason why the distinction is
less sharply drawn in imperatives than in declaratives. In declaratives subject and vocative
are referentially quite independent. There is no intrinsic connection between the subject-
referent and the addressee(s): the subject may of course refer to the addressee(s), but
it then does so only coincidentally. But in imperatives the subject is always referentially
tied to the addressee(s). This is perfectly consistent with the NP being grammatically 3rd
person: somebody and all those in the front row in [5] are also interpreted as “somebody
among you”, “all those of you in the front row”. It follows that with a few exceptions like
those in [7i–ii], expressions which have the potential to function as imperative subject
can also appear in vocative function, and vice versa.
Single NP in imperatives: subject or vocative?
NP in initial position
Where the NP is initial, vocative and subject will be distinguished prosodically: the vocative
is set apart intonationally, whereas the subject is intonationally linked with the predicator;
in writing, the vocative, but not the subject, is set off by a comma. The two functions can of
course combine, with vocative + subject effectively the only possible order:
[8] i You at the back(,) please make less noise. [vocative or subject]
ii Kim, you be umpire please. [vocative + subject]
One grammatical difference is that a subject, but not a vocative, can serve as antecedent for
a pronoun. Compare:
[9] i Somebody at the front(,) write your name on the board. [vocative or subject]
ii Somebody at the front write their name on the board. [subject only]
In [i] your is deictic, not anaphoric, and hence has no bearing on the status of the initial NP.
In [ii], however, their is anaphoric to the initial NP, which requires that it be read as subject.
NP in final position
Here the distinction between subject and vocative is somewhat less determinate:
[10] i Turn the fan on please, somebody.
ii Stand up all those who wish to leave.
Final position is in other clause types a very common one for vocatives, and must surely
be allowed in imperatives too. We will thus take somebody in [i] to be a vocative, just like
that in interrogative What time is it, somebody? More problematic is [ii], where the final
NP is prosodically integrated into the clause. With declaratives a final vocative need not be
prosodically set apart as clearly as an initial one, but it still cannot carry the focal stress as
the NP in [ii] can. Moreover it would be possible to have a clear vocative before the verb, as
in Now, children, stand up all those who wish to leave. On the other hand, declarative subjects
are allowed in final position only under very restrictive conditions, and there is certainly no
declarative counterpart of [ii] with final subject: ∗Stood up all those who wished to leave. The
evidence from anaphora is not entirely conclusive. Put their hands up all those who wish to
leave is less natural than the form with deictic your, but it is clearly more acceptable than, say,
[9ii] would be if read with the initial NP in vocative function – and in any case its relative
unnaturalness could be due to the fact that there are constraints on the use of anaphoric pro-
forms when the antecedent follows rather than precedes. On balance, the evidence seems to
favour the subject rather than the vocative analysis, but the matter is very far from clear-cut.
9.2.3 Imperatives with auxiliary do

Verbal negation
Like clauses of other types, negative imperatives may have the negation associated with
the verb (verbal negation) or incorporated within some other element:
[11] i Don’t say anything that could compromise you. [verbal negation]
ii Say nothing that could compromise you. [non-verbal negation]
In verbal negation, imperatives differ from other constructions in requiring the dummy
auxiliary do unconditionally, not just when there is no other auxiliary present (see
Ch. 9, §2.2). Compare, for example:
[12] i You weren’t sitting in that chair when your father returned. [declarative]
ii Don’t be sitting in that chair when your father returns. [imperative]
Do in combination with a subject
Imperatives with verbal negation may contain an overt subject; the subject then either
precedes or (more often) follows don’t:
[13] subject + DON’T DON’T + subject
i a. You don’t be so cheeky. b. Don’t you be so cheeky.
ii a. Those with a bus to catch don’t b. ? Don’t those with a bus to catch
hesitate to leave. hesitate to leave.
With you the subject-first order is strongly disfavoured, whereas with other, especially
longer, subjects, the subject-first order tends to be preferred and examples like [iib]
are somewhat marginal. Contrastiveness will tend to favour the subject-second order:
The girls can board now, but don’t the boys move until I say so. The choice may also be
determined by the scope of the negative:
[14] i One of you don’t forget to turn off the light. [narrow scope negation]
ii Don’t one of you forget to sign the register. [wide scope negation]
One is outside the scope of the negative in [i], inside in [ii]: the meanings can be
contrasted as “One of you remember to turn off the light” and “All of you remember to
sign the register”.
§ 9.2.4 Imperatives as directives 929
Don’t and do not

Analytic do not occurs as a somewhat more formal variant of inflectional don’t, except
that it is of somewhat doubtful acceptability in constructions like [13i] which have you
as overt subject. The imperative differs from the interrogative in that the subject cannot
come between do and not. Compare:
[15] i Don’t you tell her!/? [imperative or interrogative]
ii Do you not tell her? [interrogative only]
Emphatic imperatives
Supportive do is also used in imperatives to emphasise the positive polarity and again it
occurs unconditionally, not just in the absence of an auxiliary, as in non-imperatives:
[16] a. Do hurry up. b. Do be careful.
Here too it can either follow or precede an overt subject (though this is unlikely to be
just you):
[17] i Those with a bus to catch do please feel free to leave.
ii Do at least some of you make a commitment to contribute.
9.2.4 Imperatives as directives

Whereas declarative clauses are prototypically concerned with the truth of propositions,
imperatives are prototypically concerned with carrying out some future action. Imper-
atives are characteristically used as directives, and directives do not have truth values.
The issue that arises with a directive is not whether it is true or false, but whether it is
(subsequently) complied with. A directive expresses a proposition representing a po-
tential situation: realising or actualising that situation constitutes compliance with the
directive.
The terms ‘directive’ and ‘compliance’ are to be understood in a broader sense than
they have as non-technical terms. There is no everyday word whose normal sense is gen-
eral enough to embrace the quite wide range of (direct) uses of imperatives: we therefore
extend the sense of ‘directive’ so that it covers not just orders, requests, instructions,
and the like but also advice or merely giving permission. Similarly, ‘compliance’ cov-
ers obeying orders, acceding to requests, following advice, or simply doing what one is
given permission to do. What is common to the various more specific kinds of directive
is that they all ‘promote’ compliance – with varying degrees of strength, of course. At
the stronger end of the spectrum, compliance is required, whereas at the weaker end it
is merely accepted: the range of the imperative is therefore comparable to that of the
deontic modals must, should, may/can together.
We list and illustrate below a sample of directive categories – with the proviso that
since illocutionary force depends on the interplay of a whole variety of factors many of
the examples could also be used in other ways than those suggested.
(a) Orders, commands, demands

[18] i a. By the left, quick march! b. Get out of my way!
ii a. Release all detainees! b. Do as you’re told!
iii a. Keep off the grass. b. Don’t move!
With orders, commands, and demands, compliance is required: failure to comply is

not countenanced – or is liable to provoke sanctions. For a command, I generally need
institutionalised authority to tell you to do something. A demand doesn’t have this kind
of backing, but I nevertheless forcefully insist on compliance. ‘Order’ is the most general
term for a strong directive; it can cover commands, some demands, and directives issued
by some legal authority to the public at large (. . . By order of the Council ). A prohibition
is an order not to do something.
(b) Requests, pleas, entreaties
[19] i a. Please help me tidy up. b. Kindly lower your voices.
ii a. Open the door, will you? b. Give me one more chance, I beg you.
Here I give you the option of not complying: I am asking, not telling – though very often
it will be assumed that you will do as you are asked. It is of course possible for me to
ask even when I have the authority to tell: I simply do not present myself as invoking
the authority to require compliance. The examples in [19] illustrate various ways in
which the ‘asking’ force is commonly signalled: by means of please or (less frequently)
kindly, by an interrogative tag like will you?, or by a performative parenthetical like
I beg you. There are other devices too, such as just in Just hold the hammer for me a
moment.
The distinction between telling and asking is scalar rather than categorical, and many
directives could be reported with either verb. For example, if I say to my spouse Don’t
forget to buy some milk on the way home this could lie somewhere in the middle ground,
construable as indeterminate between telling and asking. Categories (a) and (b) to-
gether have been called ‘wilful’ directives: it is, with varying strength, my will that you
comply.
(c) Advice, recommendations, warnings

[20] i a. Keep your options open. b. Don’t put all your eggs in one basket.
ii a. Wait until the price is right. b. Don’t let yourself become too complacent.
iii a. Mind the step. b. Try your uncle, perhaps.
These are a kind of non-wilful directive: compliance is not something I will, not for my
benefit, but rather something I present as being in your interest. It is then up to you
whether you comply or not. Suggestions belong in the same family, though here I am
merely putting forward a possible course of action for you to consider: there is not the
same accountability as there is with advice, in that I am not expected to be able to justify
the action as being the best thing for you to do. Some kinds of warning (e.g. Look out!)
may be very peripheral to this category in that immediate compliance can be more or
less a reflex action.
(d) Instructions and expository directives

[21] i Insert a cassette as illustrated with its labelled side facing you.
ii Dilute 1ml to 20ml with water, and gargle for 30 seconds.
iii Blend lemon juice, orange rind, and cornflour and add to the cottage cheese.
iv Take the first road on the right after the post office.
v Compare these figures with those shown in Table 1 above.
vi Take, for example, the case of my uncle.
§ 9.2.5 Agentivity in imperatives 931
These too are non-wilful: compliance is primarily in your interest rather than mine, but
is presented as necessary for the achievement of the relevant goal – using some appliance,
cooking some dish, finding your way somewhere, etc.
Examples [21v–vi] are what we will call expository directives. Directives of this kind
are used in various kinds of expository discourse, especially written, to engage the active
participation of the addressee (there are many examples in the text of this book). They
have it in common with instructions like [i–iv] that compliance will serve the purpose
in hand: in this case, following the speaker’s exposition.
(e) Invitations
[22] i a. Come over and see my etchings. b. Bring your family too if you like.
ii a. Have some more soup. b. Feel free to call in at any time.
These have some similarity with advice in that you can choose whether or not to comply
(accept) and doing so is intended to be primarily for your benefit – but it is a matter of
what you’d like rather than what is calculated to be in your best interest. Invitations may
lie at the boundary between the wilful and non-wilful categories, since compliance may
be something I’d like too. Where this is not so, they tend to merge with offers, where the
speaker has an initiating and enabling role.
(f) Permission
[23] i a. Yes, go ahead. b. Take as many as you’d like.
ii a. [Knock at the door]Come in. b. Yes, borrow it by all means.
The action is something you want to do, but I have the authority to permit or prohibit
it. Giving permission promotes compliance in the rather weak sense of not exercising
power to stop it or, to put it more positively, removing a potential obstacle.
(g) Acceptance
[24] i Well, tell her if you want to – it’s all the same to me.
ii OK, buy it if you insist – it’s your money, after all.
iii Take it or leave it – it’s my final offer.
This is the weakest kind of directive. Compliance is not something I positively want,
but I haven’t the authority or power to prevent it; I thus merely express acceptance,
perhaps with defiance, perhaps with indifference. As [iii] shows, the acceptance use is
not sharply distinct from that where the imperative is more or less equivalent to an
exhaustive conditional. Compare, similarly:
[25] i Say what you like, it won’t make any difference.
ii Double your offer: I still won’t sell.
It is arguable that the imperatives here have lost all directive force, and that such ex-
amples are instances of indirect speech acts, with direct directive + statement indirectly
conveying a concessive statement (“Whatever you say, it won’t make any difference”;
“Even if you double your offer, I still won’t sell”). For other indirect uses of imperatives,
see [29] and [39–41] below.
9.2.5 Agentivity in imperatives

Prototypically, compliance with a directive with the form of an ordinary imperative is a
matter of future action by the addressee(s).
Preference for dynamic VPs

Compliance is a matter of doing something. This is why an imperative is likely to have
a dynamic rather than a stative VP: Apply for Australian citizenship is more natural
than Be Australian and Help yourself to some more coffee more so than Want some more
coffee. Progressive be (like be generally) and perfect have are stative, and hence relatively
infrequent in imperatives. This is especially so with the perfect, where essentially equiv-
alent non-stative constructions of less or at least no greater structural complexity are
available – compare:
[26] i Have finished it before I return. [perfect: stative have]
ii Have it finished before I return. [causative: dynamic have]
iii Finish it before I return. [dynamic finish]
Example [ii] uses the dynamic causative have rather than the perfect and is much more
natural than [i]. This alternative will of course be available only in a limited set of
cases (for example, the non-finite complement of have must be transitive), but simply
dropping the perfect component, as in [iii], is a quite generally available option. The loss
of perfect meaning will not normally be significant, because if you comply with [iii] you
will necessarily comply with [i] as well.
Agentivity conferred by imperative construction itself

It must be emphasised, however, that there is no grammatical rule excluding inherently
stative VPs from imperatives. The imperative construction can itself affect the interpre-
tation of the VP, assigning an agentive role to the subject when it would not have (or not
necessarily have) such a role in a corresponding declarative. Compare:
[27] i a. Kim is patient. b. Be patient.
ii a. Kim saw what time it was. b. See what time it is.
Declarative [ia] describes a state, and Kim has a non-agentive role, whereas in [ib] the
imperative leads us to assign an agentive role to the understood subject-referent: we
interpret it as a directive to exercise self-control, to refrain from acting impatiently.
Similarly in [ii]: in declarative [a] the salient interpretation describes a happening, with
Kim in the non-agentive role of perceiver, while [b] assigns an agentive role: “Find
out / Go and look what time it is”. Where the normal sense of a predicate is stative,
acceptability of the imperative will thus depend on how readily it lends itself to such an
agentive reinterpretation. For the earlier Want some more coffee, for example, we need
the interpretation “Get yourself into the state where you want some more coffee”, and
this is a pragmatically rather unlikely directive.
Passive imperatives
Because the agentive role is associated with subject function, passive imperatives are
relatively infrequent. This reflects the fact that in declaratives whose predicate assigns
an agentive role to one of the arguments the argument concerned is aligned with the
subject of the active, not the passive. Compare active Kim attacked him and passive He
was attacked by Kim, where only the former has an agentive subject. Thus Attack him
makes a perfectly natural imperative, but Be attacked by Kim does not. Passive imperatives
§ 9.2.5 Agentivity in imperatives 933
are not ungrammatical, however, for the imperative construction can itself, as we have
just seen, confer agentivity on a subject that is not assigned an agentive role by the
predicate:
[28] i Be warned! (“Heed this warning”)
ii Don’t be intimidated. (“Don’t allow yourself to be intimidated”)
iii Get checked out by your own doctor. (“Get your own doctor to check you out”)
Positive passives with be are not often found with directive force: [i] has something of
the character of a fixed phrase. But negatives lend themselves more readily to such an
agentive interpretation, as in [ii] or, say, Don’t be seen (“Avoid being seen”). Get also
facilitates an agentive interpretation, as in [iii], where be would be unidiomatic. It is
more usual, however, to have a reflexive object here (Get yourself checked out by your
own doctor), but this involves a different construction, one where it is not the imperative
matrix clause itself that is passive.
Non-agentive imperatives and indirect speech acts

In certain cases the unnaturalness of an agentive interpretation is associated with the use
of the imperative as an indirect speech act:
[29] i Win $60,000 for an extra $1.10.
ii Sleep well. Get well soon. Have a good week-end. Enjoy your holiday.
Example [i] illustrates a form commonly used in advertising. While it suggests that
winning is subject to your control, that is not in fact so (we may assume): what is
subject to your control is just paying the extra $1.10, but whether doing so results in
your winning $60,000 is a matter of chance. Your role relative to the predicate actually
expressed in the verb win is therefore non-agentive, and the imperative indirectly conveys
another directive where the role is agentive – something like “Spend an extra $1.10
(on your Gold Lotto ticket or whatever) in order to give yourself a chance of winning
$60,000”.
Imperatives like [29ii] are normally interpreted as wishes. Sleeping well, recovering
from an illness, and so on, are situations we do not normally think of as being under
our control, and this inhibits a direct interpretation as directives. This use of imperatives
as indirect wishes is highly conventionalised, but it is limited to a very narrow range of
situations, hardly going beyond the types exemplified here. Note that these illustrate the
way indirect speech acts may involve the backgrounding of the direct force without it
being totally lost. Although enjoying oneself is generally not thought of as something
you can choose to do, your role is not wholly passive, and enjoy can certainly be used
in imperatives with full directive force (as in Come on, join in and enjoy yourself ): there
seems to be a residue of this directive force in the wishes.
Other non-agentive imperatives are found in constructions with a conditional inter-
pretation: see §9.5.18
18
Controlled compliance is incompatible with past time reference, and imperatives like Please don’t have eaten
it are highly exceptional and again not interpreted as genuine directives. This example might be used
in addressing someone not actually present, expressing the hope that they have not eaten something
(e.g. because I know it to have been contaminated and am concerned for their safety).
9.3 1st person inclusive let-imperatives

9.3.1 Grammatical properties
Dialect differences
This construction involves a specialised use of let, but there are dialect differences in
the extent to which it has diverged from the ordinary verb let meaning “allow”. We will
therefore distinguish two varieties, Dialects A and B, on the basis of their rejection or
acceptance of examples like [30ii]:
[30] Dialect A Dialect B
√ √
i Let’s go for a walk.
√
ii %
Let’s you and I/me make it ourselves. ∗
Dialect A
This is the more conservative dialect, and here 1st person inclusive let-imperatives are
analysable as containing the catenative verb let together with an NP object and (except
in ellipsis) a bare infinitival clause as second complement. The object is us, actually or
potentially contracted to ’s, with an interpretation that includes the addressee(s) in the
reference along with the speaker: Let’s go for a walk proposes that you and I go for a walk,
not that I go with some third person.19
The differences between this construction and an ordinary imperative may be illus-
trated with reference to the following pair:
[31] 1st inclusive LET-imperative ordinary imperative
a. Let us / Let’s go with her. b. Let her go with you.
(a) Contraction of us
In 1st person inclusives us can be contracted to ’s, whereas in ordinary imperatives, as
indeed in all non-imperatives, it can’t: Don’t make us/∗’s look ridiculous; He won’t let
us/∗’s join in. As it stands, Let us go with her is ambiguous between a 1st person inclusive
(“I propose that we go with her”) and an ordinary imperative (“Allow us to go with
her”), but if we reduce the us to ’s, only the first possibility remains. Contraction reduces
the pronoun to the status of a clitic (i.e. let’s is phonologically like a single word); this
is the usual form, with uncontracted us found only in relatively formal style (cf. Let us
pray, as said in a church service).
(b) No subject allowed for let
The 1st person inclusive doesn’t allow let to have a subject, whereas the ordinary imper-
ative normally does. Compare ∗You let’s go with her with the ordinary imperative You let
her go with you.
(c) Interrogative tags
Ordinary imperatives normally allow the addition of an interrogative tag with you as
subject; comparable tags with 1st person inclusives have we. Compare:
[32] a. Let’s go with her, shall we? b. Let her go with you, will you?
19
As with the 1st person plural pronoun generally (see Ch. 17, §2.2.2), there are also peripheral uses where us
refers just to the addressee(s), as when a parent says to a young child, Now, let’s just eat up these carrots, or
just to the speaker, as in the very informal Let’s have a look.
§ 9.3.1 Grammatical properties 935
(d) Scope of negation

Ordinary imperatives with catenative let as verb allow negation of either the imperative
clause itself or of the infinitival complement, with a corresponding difference in the
semantic scope of the negative. Both constructions are possible with 1st person inclusives,
but without the semantic scope difference:
[33] i a. Don’t let’s go with her. b. Don’t let her go with you.
ii a. Let’s not go with her. b. Let her not go with you.
There is a very clear difference in meaning between [ib] and [iib]. In [ib] let is inside
the scope of negation, so it is a matter of your not allowing something (her going with
you); in [iib] the let is outside the scope of negation, so it is a matter of your allowing
something (her not going with you). There is no comparable semantic difference between
[ia] and [iia]. Both would normally be used to propose/suggest that we not go with her:
the difference is simply stylistic, with [ia] a little more informal than [iia].
(e) Let not omissible in ellipsis
The verb let cannot be omitted in ellipsis in 1st person inclusives:
[34] i A: Let her go with you. B: Yes, do. / No, don’t. [ordinary]
ii A: Let’s go with her. B: ∗Yes, do. / ∗No, don’t. [1st inclusive]
In [i] A addresses the ordinary imperative to some person C and another person B gives
an elliptical response understood as “Yes, do let her go with you” or “No, don’t let her
go with you”. But these are not coherent responses to a 1st person inclusive imperative;
what we would get in [ii] (from the person to whom the directive was addressed) might
be, for the positive, Yes, let’s, with ellipsis only of go with her.
These properties of the 1st person inclusive indicate that let has here lost its proposi-
tional meaning. It does not contribute to the propositional content, does not help specify
what action would constitute compliance with the directive. It serves, rather, as a marker
of illocutionary meaning. In Dialect A, however, there is no compelling reason to sug-
gest that there has been a reanalysis of the syntactic structure. The data are compatible
with an analysis where let is still a catenative verb: it is semantically bleached and partly
fossilised in its syntax.
Dialect B
This is the dialect that allows, in informal style, examples like [30ii], Let’s you and I/me
make it ourselves. This would appear to be widely enough used to qualify as acceptable
informal style in Standard English.
Syntactically, this construction indicates that the specialisation of let has been taken a signif-
icant step further. The ’s is not here replaceable by us; for this reason (and also because of
the prosody) it is not plausible to treat the NP you and I/me as being in apposition to ’s. It
seems clear, rather, that let and ’s have fused syntactically as well as phonologically, and are
no longer analysable as verb + object: they form a single word which functions as marker
of the 1st person inclusive imperative construction. The NP you and I/me will be interpreted
not as object of let but as subject of the following verb.20
20
Some speakers of Dialect B have a negative construction that provides even stronger evidence of reanalysis:
%Let’s don’t bother. This is much less common than the construction with an NP after let’s, and cannot be
regarded as acceptable in Standard English. Its syntactic interest is that it shows conclusively that let is no
longer construed as a verb: a subjectless don’t could not appear in the complement of a catenative verb.
9.3.2 Use of 1st person inclusive imperatives

Like ordinary imperatives, 1st person inclusives are normally used as directives – but
with a narrower range of subcategories than was illustrated for ordinary imperatives in
§9.2.4. Compliance normally involves joint action by speaker and addressee(s), alone or
with one or more others. I commit myself to the action and seek your agreement.21 For
this reason, a verbal response is normally expected, indicating agreement or refusal:
[35] A: Let’s go for a walk.
B: Okay, just let me put some shoes on. / Not just now: I must finish this letter.
The force is thus of a proposal for joint action, which the addressee can accept or reject.
The speaker’s attitude towards compliance can range from strongly wanting it (Come on,
let’s get going: the bus leaves in five minutes) to merely accepting it (Okay, let’s invite Kim
as well, if that’s what you want). With expository directives, the 1st person inclusive tends
to suggest less inequality between speaker and addressee than the ordinary imperative,
and where us is contracted it is less formal:
[36] i Consider now the effect of increasing the velocity. [ordinary imperative]
ii Let’s consider now the effect of increasing the velocity. [1st incl let-imperative]
This is one use where no verbal response is expected: agreement is taken for granted.
9.4 Open let-imperatives

This construction – if it is indeed a syntactically distinct construction – is illustrated in:
[37] i If he has any evidence to support his allegation, let him produce it.
ii Let anyone who thinks they can do better stand for office at the next election.
iii If this is what the premier really intends, let him not / don’t let him pretend otherwise.
These differ (in their salient interpretation) from prototypical ordinary imperatives with
let in that they are not understood as directives to the addressee(s) to allow or permit
something. They are roughly paraphrasable with deontic should: “he should produce it”;
“anyone who thinks they can do better should stand for office”; “he shouldn’t pretend
otherwise”. They can be used where the speaker has no specific addressee(s) in mind,
e.g. in newspaper editorials, and the one(s) on whom the obligation is laid need not be
among the audience. They are therefore somewhat peripheral members of the speech act
category of directives. Nevertheless, they have it in common with more central directives
that they define some future action and call for it to be carried out. As in 1st person
inclusives, the let does not contribute to defining that future action, but serves as an
illocutionary marker.
For this reason, it is again not possible to insert you as subject, or to have an inter-
rogative tag such as will you? – indeed, no comparable tag at all is possible with these.
And as with 1st person inclusives there is no semantic scope contrast with negatives. Let
. . . not is a great deal more likely than don’t let, no doubt because such imperatives are
characteristically used in relatively formal style, but don’t let is not excluded, as in [37iii].
21
A special case is where the action is in fact to be carried out by just one (typically the speaker). For example,
I might say Let’s open the window with the aim of securing your agreement to my opening it. But note that
in this scenario I could still report the action subsequently by saying We opened the window.
§ 9.5 Imperatives interpreted as conditionals 937
There is, however, no positive grammatical property that sets such clauses apart as a
distinct construction (as contraction of us does the 1st person inclusives). An alternative
analysis, therefore, would be to group them grammatically with ordinary imperatives,
treating the difference as a matter of meaning and use rather than form.
One advantage of this is that it avoids the problems raised by the very fuzzy boundary that
would separate them. Consider the range illustrated in:
[38] i Let the prisoners be brought in.
ii This proposal was first made, let it be noted, by the Liberal Party.
iii Let ‘u’, ‘v’, ‘w’ be the velocity components along the ‘x’, ‘y’, ‘z’ axes of a molecule moving
with velocity ‘q’.
iv Now, let me see, what’s the best way of tackling the problem?
Example [i] differs from Bring the prisoners in in that it is not so specifically addressed to those
who are to bring the prisoners in. Nevertheless the audience is more directly involved than
in cases like [37] – and of course a reformulation with should would here be much too weak
to capture the meaning. Let clearly doesn’t mean “allow”, but it might be argued that it has
a causative sense, and hence does contribute to the propositional content, with compliance
being a matter of causing the prisoners to be brought in – compare the ordinary imperative
Have the prisoners brought in.22
Let it be noted in [38ii] can be roughly glossed as “it should be noted”, but it can be
regarded as an expository directive to the addressee(s); it is comparable to an ordinary
imperative with note: Note that this proposal was first made . . . Example [38iii] illustrates
an expository device in scientific discourse, where the speaker assigns values to arbitrary
symbols, by fiat, as it were, and invites the addressee to accept these decisions. Because the
structure is conventionalised, it would not be possible to insert you as subject or add a will
you? tag, but otherwise the meaning is consistent with let having its basic “allow” sense. Note,
moreover, that such let-imperatives can be coordinated with ordinary ones in the expository
use: To keep things simple, let I be an open interval and assume that all functions mentioned have
domain I.
In [38iv] let me see is a conventional way of giving oneself time to think. As such, it
doesn’t permit manipulation of the usual kind (adding you, a tag, etc.), but this is no reason
for denying that let here has the “allow” sense and contributes to the propositional content
rather than being an illocutionary marker.
9.5 Imperatives interpreted as conditionals

When an imperative is the first element in a clause-coordination, it is commonly inter-
preted as a conditional:
[39] i Ask him about his business deals and he quickly changes the subject.
ii Do that again and you’ll regret it.
iii Persuade her to agree and I’ll be forever in your debt.
iv Don’t make him the centre of attention and he gets in a huff.
Thus we understand “If you ask him about his business deals he quickly changes the
subject”, and so on. The examples illustrate the prototypical case, where the second clause
is declarative and overtly linked to the imperative by and. The conditional interpretation
22
Let has a causative sense not restricted to imperatives in the idiom let . . . know : I’ll let you know means “I’ll
tell you”, i.e. “I’ll cause you to know”.
derives from the implicature of consequence that is commonly conveyed by and – com-
pare I’ll offer him a 10% discount and he’s bound to take it. The first clause is usually
positive, but it is just possible for it to be negative, as in [iv]; the form of the neg-
ative shows clearly that it is indeed the imperative construction that we are dealing
with here.
Condition is not part of the meaning of the imperative, and the present examples can
be regarded as involving indirect speech acts. The direct force of the imperative, that of
directive, is lost or backgrounded in varying degrees.23 In the salient interpretation of
[39i] I am not directing you (even in the broad sense we have given to that term) to ask
him about his business deals. No illocutionary force attaches to the imperative clause
itself: the coordination has a force as a whole – that of a conditional statement. Example
[ii] is similar except that a further step follows: the whole indirectly conveys “If you do
that again you’ll regret it”, and this in turn conveys “Don’t do that again”, the opposite
of what would be directly conveyed by the imperative clause standing on its own. The
indirect negative directive results from the undesirability of the consequence expressed
in you’ll regret it. In [iii] the consequence is desirable, so that the conditional “If you
persuade her to agree I’ll be forever in your debt” conveys “Persuade her to agree”, the
meaning of the imperative itself. In such cases the distinction between direct and indirect
is blurred: it is hardly possible to distinguish [iii] from, say, Come over around seven and
then we’ll be able to avoid the rush hour traffic, which surely directly conveys a directive
to come around seven.
Relaxation of constraints applying to imperatives in their directive use

The use of imperatives to convey conditions is a highly conventionalised one, with the result
that certain constraints on form and propositional meaning that normally apply to impera-
tives in their directive use are here relaxed:
[40] i Do that ever again and I’ll brain you. [negatively-oriented ever]
ii Feel slightly off-colour and he thinks you’re dying. [absence of agentivity]
iii Buy myself the slightest luxury and I’m branded a spendthrift. [1st sg]
iv Express any misgivings and he accused you of disloyalty. [past time reference]
Negatively-oriented items like ever are normally excluded from positive imperatives, but they
occur readily in conditionals (cf. if you ever do that again), and the conditional meaning in [i]
thus sanctions the ever. Imperatives used with directive force (other than those with open let)
associate an agentive role with the subject, but conditionals of course do not, so that there is
no suggestion in [ii] that feeling off-colour is under your control. The reflexive myself in [iii]
shows that the understood subject is 1st person singular, a normal possibility for conditionals,
but not imperatives with directive force. And in [iv] the tense of accused shows that we are
concerned with past time (“if you expressed any misgivings”), whereas imperatives normally
involve future situations. Note that any in [iv] is a further case of a negatively-oriented item
like ever in [i]. Examples like [iii] and [iv] are comparatively rare, but they are possible; they
provide further illustration of the way an indirect use can have repercussions on the formal
and semantic properties of constructions.
23
This distinguishes the and construction from the one with or : Hurry up or we’ll be late. This conveys “If you
don’t hurry up we’ll be late”, but the full directive force of the imperative is retained. For further discussion
of the and and or constructions, see Ch. 15, §2.2.3–4.
§ 9.6 Non-imperative directives 939
Clause type of second coordinate

The second clause can belong to other clause types than declarative; what is important
is not the form but what the clause conveys:
[41] i Invite one without the other and what a row there’ll be. [exclamative]
ii Tell the truth and who’ll believe you / what’ll they do? [open interrogative]
iii Act in haste and repent at leisure. [imperative]
Exclamatives, we have seen, have essentially the same force as declaratives and hence [i]
needs no explanation. The interrogatives in [ii] would not be used as inquiries. The most
likely interpretation of who’ll believe you? is as a rhetorical question conveying “No one
will believe you”; with what’ll they do? the answer might again be contextually obvious or
else be given immediately by the one who put the question (e.g. They’ll say you’re being
disloyal to your friends). In [iii] the second imperative indirectly conveys approximately
“You’ll regret it (for a long time)”.
Let-imperatives
Our conditional examples have all been of ordinary imperatives. Open let-imperatives are
also possible: Let anyone question what he says and he flies into a rage. But 1st person inclusives
are not used in this way: in normal use they always retain their directive force. For example,
Let’s put up the price and they’ll cancel the order cannot be used like [39ii] to convey the
opposite of what is expressed in the imperative (“If we put up the price they’ll cancel the
order, so let’s not put up the price”).
9.6 Non-imperative directives

9.6.1 Interrogatives as directives
Directives are very often conveyed indirectly by means of interrogatives. This is espe-
cially so with requests – particularly when speaker and addressee are not intimates. The
imperative structure, we have noted, can be used for a wide range of directives, including
orders: to make a request by means of an imperative may therefore run the risk of ap-
pearing too brusque or peremptory, even if illocutionary modifiers like please and kindly
are added. In many circumstances indirect directives with interrogative form are con-
sidered more polite. This is not to suggest that there is any simple correlation between
interrogative form and politeness with directives. In the first place, prosody plays an
important and in part independent role: Can you move your car? will typically be more
polite than Move your car, but this can be overridden by an impatient, emphatic tone of
voice. Secondly it depends on the content of the interrogative: Can you move your car
lends itself much more readily to use as a polite request than, say, Must you park your car
across my driveway?
We find a great variety of interrogative directives, but four of the most important
semantic categories concern: (a) your ability to do something; (b) your desire or willing-
ness to do something; (c) the deontic necessity for you to do something; (d) the reason
for you to do something:
[42] i Can you open the door. (“Open the door”) [ability]
ii Would you like to / Will you sign here? (“Sign here”) [desire/willingness]
iii Must you talk so loud? (“Don’t talk so loud”) [deontic necessity]
iv Why don’t you bring your radio? (“Bring your radio”) [reason]
(As noted in §3.2, there is some variation with respect to punctuation, with a full stop
often preferred to a question mark in the ability or desire/willingness cases.)
(a) Ability questions

These lend themselves to indirect directive use since a likely reason for me to be interested
in your ability to do something is that I want you to do it. Typical openers are:
[43] can you, could you, is it possible ( for you), will/would it be possible (for you), are
you able, will/would you be able
The versions with can are most frequent, and are especially likely in contexts where the
answer is obviously “yes”, so that the direct inquiry force is effectively lost; this will usually
apply with such everyday examples as Can you pass the salt, etc. The forms with preterites
(could/would, with the preterite indicating tentativeness) are regarded as more polite.
All the above could be negated: can’t you, couldn’t you, etc. Negative questions are
always biased, and in the present case the negation adds some emotive component of
meaning, perhaps impatience (Can’t you talk a little louder?, suggesting you ought to be
able to) or persuasiveness (Couldn’t you stay a little bit longer?). Inflectional negatives
always have the ability predicate within their scope; analytic ones with can are potentially
ambiguous as to scope. Can you not stand by the door, for example, can have the not in
the can clause or in the stand clause. In the first case it is a negative question conveying
the positive directive “Stand by the door”, and in the second it is a positive question
conveying the negative directive “Don’t stand by the door”.
(b) Desire/willingness questions

These likewise have a natural connection with directives: if you want or are willing to do
something you are likely to comply with a request to do it. Typical formulae are:
[44] i will/would you, would you like to / care to / be so kind as to [+ infinitival]
ii do/would you mind [+ gerund-participial]
Again the tentative preterite would adds to the effect of politeness. (Note that want occurs
in this use with do but not would, and is appreciably less polite: Do you want to clear
the table so that we can have lunch?) Inflectional negatives are possible for those taking
infinitival complements, especially the first three (won’t/wouldn’t you, wouldn’t you like
to); they do not occur with mind, for the salient bias of the question would be positive,
suggesting that you do/would mind. The analytic negatives will/would you not have the
scope ambiguity illustrated above for can. Compare:
[45] i Will you not [take a seat]? [not in will clause]
ii Will you [not put your feet on the sofa]. [not in put clause]
The pragmatically salient interpretations have primary verb negation in [i], “Take a seat”,
and secondary negation in [ii], “Don’t put your feet on the sofa”.
Closely related to desire/willingness is prospective intentional future, as in Are/Aren’t
you going to tidy your room? These express doubt as to whether you intend to do what
you should do, and for this reason are quite well down the politeness scale.
(c) Deontic necessity questions

Deontic necessity is usually questioned by must, as in [42iii] or have (Do you have to talk
so loud?), though need, necessary, etc., are also possible (cf. Need you / Is it necessary to talk
so loud?). The situation denoted in the complement clause (your talking so loud) is one
§ 9.6.2 Declaratives as directives 941
that I regard as undesirable, and the question has, in context, a clear negative bias: I don’t
think there is any necessity for you to talk so loud. Hence the implied directive – whose
content is this time the opposite of that expressed in the complement clause. Suggesting
that you are unnecessarily doing what I don’t want does not of course make for a polite
directive.
(d) Reason questions

These are usually expressed by means of why and have been discussed in §4.5. Here again
the content of the implicit directive is the opposite of that expressed in the question,
whether the latter is negative, as in [42iv], or positive, as in Why accept less? (“Don’t
accept less”).
3rd person
The above examples all have you as subject, but just as imperatives can have 3rd person
subjects, so can interrogatives with directive force: Will everyone remember to sign the
register. It is also possible to have 3rd person subjects in examples like those given in [42]:
Can he come a little earlier tonight?; Would he like to return my wrench?; Must they talk
so loud?; Why doesn’t she bring her radio? These aren’t equivalent to imperatives because
they are not addressed to the person(s) concerned, but can still have indirect directive
force, suggesting that you should convey the directive to whoever is to comply.
1st person inclusive

Interrogatives used with essentially the same force as 1st person inclusive imperatives
generally begin with shall we or why :
[46] i Shall we go for a swim?
ii Why don’t we eat out tonight? Why waste our time on it?
At the direct level, [i] is a direction question whose answers are expressed by 1st person
inclusive imperatives: in context the question is biased towards the positive answer, “Let’s
go for a swim”, which it therefore indirectly conveys.
9.6.2 Declaratives as directives

Declaratives can be used with either direct or indirect directive force. The direct cases
involve the performative construction (§3.1):
[47] i I order/beg you to return her letters.
ii The riding of bicycles on the walkway is strictly prohibited.
Among the many indirect cases, mention may be made of those involving (a) the speaker’s
wants or needs; (b) the addressee’s future actions; (c) deontic necessity:
[48] i I want / need / would like someone to hold the ladder. [speaker’s wants/needs]
ii You are going to / will apologise. [addressee’s future actions]
iii You must / have to come in now. [deontic necessity]
The indirectness in these cases does not contribute to politeness. Case [48ii] in partic-
ular is strongly wilful and coercive: in telling you what you are going to do I conspicuously
leave you no choice. Politeness can be achieved, however, by combining declarative and
interrogative in a doubly indirect directive:
[49] I wonder whether you would mind moving your car a little.
The statement actually expressed conveys an indirect question (“Would you mind moving
your car a little?”), and this in turn is interpreted as an indirect directive. This device
generally involves the more polite of the interrogative constructions (note, for example,
that would is not omissible from the whether clause), and the extra indirectness increases
the impression of politeness.
9.6.3 Non-finite and verbless directives

Directives of this form are commonly used in written notices, where there is a need for
brevity:
[50] i Smoking prohibited. No visitors allowed beyond this point. [non-finite]
ii No smoking. No entry. Slow. [verbless]
Non-finites are typically abbreviated passive performatives (cf. Smoking is prohibited ).
The verbless construction is commonly used in speech to indicate what one is ordering or
asking for: Two black coffees; The hammer, please; Two adults, please (“I request admission
for two adults”); Single to Manchester (as in booking transport); and so on.
9.7 Imperatives with interrogative tags

Ordinary and 1st person inclusive let-imperatives may be anchor to an interrogative tag:
[51] i Help yourself, will you / won’t you? [positive ordinary imperative]
ii Don’t tell anyone, will you? [negative ordinary imperative]
iii Let’s (not) go with them, shall we? [1st person inclusive let-imperative]
The tags attached to imperatives cannot be derived by grammatical rules of the kind
we suggested for tags attached to declaratives (§5.1). They can be regarded as ellipti-
cal versions of the full interrogatives Will you help yourself ?, Won’t you help yourself?,
Will you not tell anyone?, Shall we (not) go with them?, but the reason truncated ver-
sions of these can be attached to the imperative anchors is that they are interroga-
tives of the types that are commonly used as indirect directives, as described in §9.6.1.
The indirect force of the interrogative thus matches the direct force of the imperative
anchor.
This is why we can have either a positive or a negative tag with the positive imperative
in [51i], but only a positive tag with the negative imperative in [ii]. Will you help yourself ?
and Won’t you help yourself ? can both be used to convey “Help yourself ”, and hence
both will you? and won’t you? can be attached to Help yourself. “Don’t tell anyone”,
however, can be indirectly conveyed by Will you not tell anyone, but not normally by
?
Won’t you not tell anyone, and hence only will you? is an appropriate tag for Don’t tell
anyone.
Nevertheless, the construction is conventionalised in that the tags correspond to only
a subset of the interrogatives that can be used with indirect directive force. For example,
Can/Could/Will/Would you not touch it could all convey “Don’t touch it”, but the tag
for a negative ordinary imperative is virtually restricted to will you? With positives the
range is considerably greater: the most frequent are the above will you?, won’t you?, but
would you?, can you?, can’t you?, could you? and, especially in AmE, why don’t you? are
also common (and wouldn’t you?, couldn’t you? are possible):
§ 9.8 No subordinate imperative construction 943
[52] i Just give me a hand with these boxes, would you?

ii Let me have your reply by the end of the week, can you /could you?
iii Watch where you’re putting your feet, can’t you?
Can and could tend to retain some of their direct inquiry force. Can’t you? typically
conveys some impatience. Won’t you? and would you? are generally the most polite.
One context favouring won’t you? is where the directive is a reminder or otherwise
not unexpected: Be there at six, as we agreed, won’t you? The normal tag for 1st person
inclusives is shall we?, but why don’t we? is also a possibility.
9.8 No subordinate imperative construction

Imperatives normally occur as main clauses: there is no grammatically distinct con-
struction that can properly be regarded as the subordinate counterpart of a main clause
imperative, as whether she liked it and why she liked it are the subordinate counterparts
of the closed and open interrogative main clauses Did she like it? and Why did she like it?
respectively.
Reporting of directives
Imperatives are generally used as directives, and directive speech acts can of course be re-
ported. But they are reported by means of constructions where the subordinate clauses are
syntactically and semantically very different from imperative clauses. Compare, for example:
[53] i Leave her alone. [imperative]
ii Max ordered/told/asked/advised me to leave her alone. [infinitival]
iii Max asked that I leave her alone. [mandative subjunctive]
All three constructions contain the plain form of the verb, but the imperative differs from
the other two in taking auxiliary do for verbal negation and emphatic polarity:
[54] i Don’t be late. [imperative]
ii ∗He told me to do not be late. [infinitival]
iii ∗He asked that I do not be late. [mandative subjunctive]
In other clause types subordination does not exclude do in this way (compare main Why
didn’t they like it? and subordinate He asked why they didn’t like it), so the data in [54] sug-
gests we are dealing with different constructions, not main and subordinate versions of a
single construction. Compelling evidence for this view comes from the fact that both in-
finitivals and mandatives allow a much wider range of subject–predicate combinations than
we find with imperatives. There are, for example, no main clause imperatives matching the
subordinate clauses in:
[55] i a. The house was shown to be in need of repair.
b. She was the first one to realise its significance.
c. It’s unusual for it to rain so much in August.
d. We can’t afford for there to be more disruption.
ii a. He suggested that the meeting be postponed.
b. It is essential that there be no more disruption.
c. It’s important that she get all necessary assistance.
#
Be in need of repair and #Realise its significance are pragmatically anomalous, while ∗It rain
so much in August, ∗There be more disruption, and so on are ungrammatical.
The infinitival construction has a vastly greater range of use than the imperative, and
examples like those in [55i] bear no significant relation to imperatives at all. The meaning
is compatible with its use to report directives, but the interpretation of [53ii] as reports of
directives depends crucially on the lexical verb in the matrix clause – order, tell, etc. The range
of the mandative construction is more limited, but still considerably broader than that of the
imperative, as evident from the lack of imperative counterparts to the subordinate clauses in
[55ii]. The differences in form and meaning between either infinitivals or mandatives on the
one hand and imperatives on the other cannot be explained in terms of subordination: we
must recognise three syntactically quite distinct constructions.
10 Minor clause types
In this final section of the chapter we review summarily a number of main clause con-
structions that do not belong to any of the major clause types discussed so far. See also
the elliptical constructions discussed in Ch. 17, §7.8.
Optatives
[1] i Long live the Emperor. God save the Queen! God help you if you’re not
ready on time! Far be it from me to complain. So be it.
ii May all your troubles be quickly resolved! Long may she reign over us!
iii Would that he were still alive! Would to God I’d never set eyes on him!
These three constructions express wishes. The examples in [i] are subjunctives. Though
the subjunctive construction is fully productive in subordinate clauses, in main clauses
it is found only in a narrow range of fixed expressions or formulaic frames. In some the
subject occupies its basic position, while in others it is postposed to the end of the clause
or to the right of be. Construction [ii], which belongs to somewhat formal style, has may
in pre-subject position, meaning approximately “I hope/pray”. There is some semantic
resemblance between this specialised use of may and that of let in open let-imperatives,
but syntactically the NP following may is clearly subject (witness the nominative form
she). The construction has the same internal form as a closed interrogative, but has no
uninverted counterpart. Construction [iii] is archaic; syntactically it consists of would
as predicator with a finite clause complement (and optionally the PP to God as another
complement), but is of course exceptional in that the understood subject (I ) is not
expressed. The subordinate clause is a modal preterite, with the same interpretation as
in the regular construction with I wish (Ch. 11, §7.2).
Clauses with the subordinate form

[2] i That it should have come to this!
ii To think that he was once the most powerful man in the land!
The meaning is close to that of the exclamatives How amazing it is that it should have
come to this / to think that he was once the most powerful man in the land, with subordinate
clauses in extraposed subject function. The subordinate form of the clauses in [2] suggests
that, when they stand on their own, as here, they are fragments, containing a subordinate
clause but with the matrix frame omitted. Other infinitival constructions that are not
overtly embedded include the Oh to be in England pattern (which also belongs in the
§ 10 Minor clause types 945
optative category: “I wish I were in England”) and the negative Not to worry (“Don’t /
Let’s not worry”), which is rarely found with verbs other than worry.
Conditional fragments
[3] i If only you’d told me earlier!
ii Well, if it isn’t my old friend Malcolm Duce!
iii If you’d like to move your head a little.
iv Supposing something happens to part us, June?
Various kinds of conditional adjunct can be used on their own, with the apodosis left
unexpressed. Construction [i], with if only + modal preterite indicating counterfactu-
ality, is used to express regret: “How unfortunate you didn’t tell me earlier (because if
you had done, things would have been better)” – see Ch. 8, §14.2.1. Construction [ii]
involves a fixed frame of the form if it/that isn’t X; it is used to express surprise at seeing X
(so [ii] itself conveys “It is my old friend Malcolm Duce”). Construction [iii] is a further
type of indirect directive: “Please move your head a little” (as said by doctor to patient,
for example); the missing apodosis is understood along the lines of “that would be help-
ful”. Example [iv] is understood as a question: “What if . . . ?”; the same construction
can be used with directive force, as a suggestion or invitation: Supposing we meet at six.
This fragment construction represents the most common use of conditional supposing;
suppose can be used with the same meaning, but syntactically that gives an imperative
clause rather than a fragment.
Verbless directives
[4] Out of my way! On your feet! This way! Everybody outside! All aboard!
Head up! Shoulders back! Careful! Off with his shoes! On with the show!
Fragmentary structures like these are commonly used for a peremptory type of directive,
where immediate compliance is required. In many a verb could be supplied (Get out of
my way!; Come this way!; Everybody move outside; Put your shoulders back!; Be careful!),
but the off /on + with construction can’t be expanded in this way. Military commands
often take this form: Eyes right!; At ease! ; etc. Other types of verbless directive have the
form of NPs: No talking! or Two coffees, please, as used in restaurants, shops, etc.
Parallel structures
[5] The sooner, the better. More haste, less speed. Out of sight, out of mind.
No work, no pay. Once bitten, twice shy. Like father, like son.
There are numerous lexicalised expressions of this kind, fixed phrases, proverbs, and the
like. They consist of a juxtaposition of two expressions of like form. The first two can be
seen as elliptical versions of the correlative comparative construction (Ch. 13, §4.6) – cf.
The sooner you decide, the better it will be. Some have a conditional interpretation (e.g. “If
you do no work, you get no pay”). Some, such as the last, bear no clear resemblance to
any productive syntactic construction. Similar parallelism is seen in pairs of imperative
clauses like Spare the rod, and spoil the child, which also have conditional interpretations
like those of the non-lexicalised examples discussed in §9.5.

Clause Type and Illocutionary Force PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Clause Type and Illocutionary Force PDF

Uploaded by

Copyright:

Available Formats

10

Clause type and

1 Type as a grammatical system of the clause 853

7.3 Who and whom 904

1 Type as a grammatical system of the clause

The five major categories

Complex relation between form and meaning

Clause type in subordinate clauses

The type system applies to the clause, not the sentence

clauses do not have to be of the same type:

The clause types are mutually exclusive

2 Distinctive grammatical properties of the major clause types

Closed and open interrogatives: subclasses of a larger class

3 Some semantic and pragmatic preliminaries

3.1 Illocutionary force

More specific kinds of illocutionary force

Primary and secondary force

The performative use of verbs

small sample of verbs that can be used performatively:

Propositional and non-propositional components of meaning

3.2 Indirect speech acts

Definition of indirect speech act

The pervasive nature of indirect speech acts

Non-propositional markers of indirect force

Idiomatic forms of indirect speech acts

4.1 Question as a semantic and as a pragmatic category

The distinction between answer and response

‘The answer’ and ‘the right answer’

Not all questions are inquiries

4.2 Summary classification of questions

4.3 Polar questions

The form of polar questions

4.4 Alternative questions

The form of alternative questions

Or in alternative and polar questions

No alternative interrogative clause type corresponding to alternative question

Pragmatic differences between polar and polar-alternative questions

4.5 Variable questions

The form of variable questions

The bare infinitival construction

Single-variable vs multi-variable questions

Coordination of interrogative phrases

Restriction on fronting of interrogative phrase

Answers to multi-variable questions

Multiple answer vs multiple response

Juxtaposition of variable and polar or alternative questions

4.6 Direction questions

Distinction between direction and information questions only

Embedded direction questions have infinitival form

Right answers to direction questions

4.7 Biased questions

4.7.1 Kinds and degrees of bias

(b) Deontic bias

(c) Desiderative bias

Different degrees of bias

The encoding of bias

4.7.2 Declarative questions

declarative question I am typically seeking confirmation of a proposition that I am in-

Lexical reinforcement of bias

Declarative questions as indirect speech acts

4.7.3 Negative interrogative questions

Negative interrogative questions with negative bias