Search This Blog

Thursday, 1 August 2019

CB 7

Page 1
§ 5.4 Pragmatic presupposition
41
issue. If she in fact never tried to secure her son’s release, [ii] is strictly speaking true, but
it would normally be a very inefficient or misleading way of conveying that information.
A simpler, more direct and more explicit way of doing so would be to say She never tried
to secure her son’s release. The fact that I didn’t say this but said [ii] instead will lead you
to infer that the negation applies to the stopping, so that [ii] implicates that she is still
trying. Similarly with questioning. If I didn’t know, and wanted to find out, whether she
formerly tried to secure her son’s release, I would be expected to ask Did she try to secure
her son’s release? If I ask [iii] instead, the natural inference will be that I am trying to find
out about the present state of affairs.
The kind of reasoning just described is similar in kind to that invoked in discussing
conversational implicatures, reflecting the fact that both phenomena are pragmatic.17
Like conversational implicature, presupposition applies in the first instance to utterances,
but we can apply it derivatively to sentences with the same ‘normally’ qualification as
before:
[12]
X normally presupposesY ≡in saying X the speaker, in the absence of indications
to the contrary, takes the truth of Y for granted, i.e. presents it as something that
is not at issue.
Again, then, we allow that in special circumstances a presupposition may be cancelled.
Consider, for example, the following exchange:
[13] A: Have you stopped using bold face for emphasis?
B: No I haven’t (stopped using bold face for emphasis); I’ve always used small caps.
A’s question presupposes that B formerly used bold face for emphasis. But suppose it
turns out that A was mistaken in believing this. B answers the question with a negative,
and since this reflects the form of the question it too would normally presuppose that B
formerly used bold face for emphasis. But in the context given here that presupposition
is cancelled.
The presupposition associated with the verb stop coincides with an entailment when
X is positive and declarative, as in [11i], but with a conversational implicature when X
is negative or interrogative, as in [11ii–iii]. You cannot stop doing something that you
have never done before, so [11i] cannot be true unless [11iv] is true. This gives the latter
the status of an entailment. But it is not an entailment of the negative [11ii], as evident
from the example in [13]. Nevertheless, if I say [11ii] I will normally be taken to have
implicitly committed myself to [11iv], and the latter therefore counts as a conversational
implicature. Likewise with the interrogative [11iii], which does not have entailments.
This represents the most usual pattern for presuppositions. For the most part they are
entailed if X is positive and asserted to be true, and otherwise they are conversationally
implicated. But this is not a necessary feature of presuppositions: we will see that they
do not always follow this pattern.
17An alternative view is that presupposition is a logical or semantic concept. On one version of this account, a
presupposition is a proposition that must be true if the presupposing proposition (or the sentence expressing
it) is to be either true or false. In the case of [11], for example, in a context where [iv] was false, where she had
never tried to secure her son’s release, [i–ii] would be neither true nor false: they would simply lack a truth
value (or would take a third truth value distinct from both truth and falsity). We do not adopt that concept of
presupposition here, and take the view that if a proposition is not true, then it is false.
2
Syntactic overview
Rodney Huddleston
1 Sentence and clause 44
2 Canonical and non-canonical clauses 46
3 The verb 50
4 The clause: complements 52
5 Nouns and noun phrases 54
6 Adjectives and adverbs 57
7 Prepositions and preposition phrases 58
8 The clause: adjuncts 59
9 Negation 59
10 Clause type and illocutionary force 61
11 Content clauses and reported speech 62
12 Relative constructions and unbounded dependencies 63
13 Comparative constructions 64
14 Non-finite and verbless clauses 64
15 Coordination and supplementation 66
16 Information packaging 67
17 You leave and anaphora 68
43
44
Given the length and nature of this book, there will be relatively few readers who begin at
the beginning and work their way through the chapters in order to the end. We envisage,
rather, that readers will typically be reading individual chapters, or parts thereof, without
having read all that precedes, and the main purpose of this syntactic overview is to enable
the separate chapters to be read in the context of the grammar as a whole.
We begin by clarifying the relation between sentence and clause, and then intro-
duce the distinction between canonical and non-canonical clauses, which plays an im-
portant role in the organisation of the grammar. The following sections then survey
very briefly the fifteen chapters that deal with syntax (as opposed to morphology or
punctuation), noting especially features of our analysis that depart from traditional
grammar.
1 Sentence and clause
Syntax is concerned with the way words combine to form sentences. The sentence is the
largest unit of syntax, while the word is the smallest. The structure of composite words
is also a matter of grammar (of morphology rather than syntax), but the study of the
relations between sentences within a larger text or discourse falls outside the domain of
grammar. Such relations are different in kind from those that obtain within a sentence,
and are outside the scope of this book.
We take sentences, like words, to be units which occur sequentially in texts, but are
not in general contained one within another. Compare:
[1] i Jill seems quite friendly.
ii I think Jill seems quite friendly.
iii Jill seems quite friendly, but her husband is extremely shy.
Jill seems quite friendly is a sentence in [i], but not in [ii–iii], where it is merely part
of a sentence – just as in all three examples friend is part of a word, but not itself a
word.
In all three examples Jill seems quite friendly is a clause. This is the term we apply to
a syntactic construction consisting (in the central cases) of a subject and a predicate. In
[1ii] one clause is contained, or embedded, within a larger one, for we likewise have a
subject–predicate relation between I and think Jill seems quite friendly. In [iii] we have
one clause coordinated with another rather than embedded within it: her husband is
subject, is extremely shy predicate and but is the marker of the coordination relation. We
will say, then, that in [i–ii] the sentence has the form of a clause, while in [iii] it has the
§ 1 Sentence and clause
45
form of a coordination of clauses (or a ‘clause-coordination’).1 Within this framework,
the clause is a more basic unit than the sentence.
To say that sentence [1i] has the form of a clause is not to say that it consists of
a clause, as the term ‘consists of’ is used in constituent structure analysis of the type
introduced in Ch. 1, §4.2. There is no basis for postulating any singulary branching here,
with the clause functioning as head of the sentence. This is why our tree diagram for the
example A bird hit the car had the topmost unit labelled ‘clause’, not ‘sentence’. ‘Sentence’
is not a syntactic category term comparable to ‘clause’, ‘noun phrase’, ‘verb phrase’, etc.,
and does not figure in our constituent structure representations.
Most work in formal grammar makes the opposite choice and uses sentence
(abbreviated S) rather than clause in constituent structure representations. There are
two reasons why we do not follow this practice. In the first place, it creates problems for
the treatment of coordination. In [1iii], for example, not only the whole coordination but
also the two clauses (Jill seems quite friendly and but her husband is extremely shy) would
be assigned to the category sentence. The coordination, however, is quite different in its
structure from that of the clauses: the latter are subject–predicate constructions, while
the coordination clearly is not. Most importantly, assigning the whole coordination to
the same category as its coordinate parts does not work in those cases where there is
coordination of different categories, as in:
[2]
You must find out [the cost and whether you can pay by credit card].
Here the first coordinate, the cost, is an NP while the second is, on the analysis under
consideration, a sentence, but the whole cannot belong to either of these categories.
We argue, therefore, that coordinative constructions need to be assigned to different
categories than their coordinate parts. Thus we will say, for example, that Jill seems quite
friendly is a clause, while [1iii] is a clause-coordination, Jill and her husband an NP-
coordination, and the bracketed part of [2] an NP/clause-coordination (a coordination
of an NP and a clause).
The second reason why we prefer not to use ‘sentence’ as the term for the syntactic
category that appears in constituent structure representations is that it involves an un-
necessary conflict with the ordinary, non-technical sense of the term (as reflected, for
example, in dictionary definitions). Consider:
[3]
a. The knife I used was extremely sharp.
b. I’m keen for it to be sold.
The underlined sequences are not sentences in the familiar sense of the term that we
adopted above, according to which sentences are units of a certain kind which occur
in succession in a text. The underlined expressions nevertheless contain a subject (I, it)
and a predicate (used and to be sold), and hence belong in the same syntactic category
as expressions like Jill seems quite friendly. If we call this category ‘sentence’ rather than
‘clause’, the term ‘sentence’ will have two quite different senses.
1Traditional grammar classifies the sentences in [1] as respectively simple, complex, and compound, but this
scheme conflates two separate dimensions: the presence or absence of embedding, and the presence or absence
of coordination. Note that in I think Jill seems quite friendly, but her husband is extremely shy there is both
embedding and coordination. We can distinguish [i–ii] from [iii] as non-compound (or clausal) vs compound;
[i–ii] could then be distinguished as simple vs complex clauses but no great significance attaches to this latter
distinction, and we shall not make further use of these terms.
Chapter 2 Syntactic overview
46
2 Canonical and non-canonical clauses
There is a vast range of possible clause constructions, and if we tried to make descriptive
statements covering them all at once, just about everything we said would have to be
heavily qualified to allow for numerous exceptions. We can provide a simpler, more
orderly description if in the first instance we confine our attention to a set of basic, or
canonical, constructions, and then describe the rest derivatively, i.e. in terms of how
they differ from the canonical constructions.
The contrast between canonical and non-canonical clauses is illustrated in the fol-
lowing examples:
[1]
canonical
non-canonical
i a. Kim referred to the report.
b. Kim did not refer to the report.
ii a. She was still working.
b. Was she still working?
iii a. Pat solved the problem.
b. The problem was solved by Pat.
iv a. Liz was ill.
b. He said that Liz was ill.
v a. He has forgotten the appointment.
b. Either he has overslept or he has
forgotten the appointment.
■ Dimensions of contrast between canonical and non-canonical constructions
The examples in [1] illustrate five major dimensions of contrast between canonical and
non-canonical clauses. In each case the canonical clause is syntactically more basic or
elementary than the non-canonical one.
The examples in [1i] differ in polarity, with [a] positive and [b] negative. In this
example, the negative differs from the positive not just by virtue of the negative marker
not but also by the addition of the semantically empty auxiliary do.
The contrast in [1ii] is one of clause type, with [a] declarative and [b] interrogative.
The syntactic difference in this particular pair concerns the relative order of subject and
predicator: in [a] the subject occupies its basic or default position before the predicator,
while in [b] the order is inverted. In the pair She finished the work and Did she finish the
work? the interrogative differs from the declarative both in the order of elements and
in the addition of the auxiliary do. All canonical clauses are declarative; non-canonical
clauses on this dimension also include exclamatives (What a shambles it was!) and im-
peratives (Sit down).
In [1iii], canonical [a] is active while [b] is passive. These clauses differ strikingly in
their syntactic form, but their meanings are very similar: there is a sense in which they
represent different ways of saying the same thing. More precisely, they have the same
propositional content, but differ in the way the information is presented – or ‘packaged’.
The passive is one of a number of non-canonical constructions on this dimension. Others
include preposing (e.g. Most of them we rejected, contrasting with canonical We rejected
most of them), the existential construction (e.g. There were several doctors on board,
contrasting with Several doctors were on board), and the it-cleft (e.g. It was Pat who spoke
first, contrasting with Pat spoke first).
The underlined clause in [1ivb] is subordinate, whereas [a] is a main clause. In this
example, the non-canonical clause is distinguished simply by the presence of the sub-
ordinator that, but many kinds of subordinate clause differ from main clauses more
§ 2 Canonical and non-canonical clauses
47
radically, as for example in This tool is very easy to use, where the subordinate clause
consists of just the VP subordinator to together with the predicator, with both subject
and object left unexpressed. The clause in which a subordinate clause is embedded is
called the matrix clause – in [ivb], for example, subordinate that Liz was ill is embedded
within the matrix clauseHe said that Liz was ill. Subordination is recursive, i.e. repeatable,
so that one matrix clause may be embedded within a larger one, as in I think he said that
Liz was ill.
Finally, the underlined clause in [1vb] is coordinate, in contrast to non-coordinate
[a]; it is marked as such by the coordinator or. A greater departure from canonical
structure is seen in Jill works in Paris, and her husband in Bonn, where the predicator
works is missing.
It is of course possible for non-canonical constructions to combine, as in:
[2]
I can’t understand why I have not been questioned by the police.
The underlined clause here is negative, interrogative, passive, and subordinate. But these
are independent properties, and we can describe the structure in terms of its difference
from canonical clause structure on four separate dimensions.
■ Counterparts
In the examples of [1] we presented the non-canonical clauses side by side with their
canonical counterparts, i.e. canonical clauses differing from them simply as positive
rather than negative, declarative rather than interrogative, and so on. Where a clause
combines two non-canonical features, its counterpart with respect to each feature will
be non-canonical by virtue of retaining the other. Thus It wasn’t written by Sue has as
its active counterpart Sue didn’t write it (non-canonical by virtue of being negative)
and as its positive counterpart It was written by Sue (non-canonical by virtue of being
passive).
Itmustbeemphasised,however,thatnotallnon-canonicalclauseshavegrammatically
well-formed counterparts. Compare, for example:
[3] i a. I can’t stay any longer.
b.
*
I can stay any longer.
ii a. Have they finished yet?
b.
*
They have finished yet.
iii a. Kim was said to be the culprit.
b.
*
Said Kim to be the culprit.
iv a. There was an accident.
b.
*
An accident was.
v a. If it hadn’t been for you,
b.
*
It had been for you.
I couldn’t have managed.
Example [ia] has no counterpart differing from it as positive vs negative, and similarly
there is no declarative counterpart to interrogative [iia]. There is no active counterpart
to the passive [iiia], partly because say + infinitival (with this sense) is restricted to
the passive construction, partly – and more generally – because there is no element
corresponding to the subject of an active clause. Existential [iva] differs from the one
cited above (There were several doctors on board) in that again there is no non-existential
counterpart. And finally [va] contains a subordinate clause with no main clause coun-
terpart. It had been for you is of course grammatical in the interpretation where it refers
to something identifiable in the context (cf. The parcel had been for you), but that is not
how it is interpreted in [va].
Chapter 2 Syntactic overview
48
■ Syntactic processes
We follow the practice of much traditional and modern grammar in commonly describ-
ing non-canonical structures in terms of syntactic processes. We talk, for example, of
subject–auxiliary inversion, of passivisation and relativisation, or preposing and post-
posing, and so on. It should be made clear, however, that such process terminology
is merely a convenient descriptive device. When we say, for example, that Is she still
working? involves subject–auxiliary inversion, we are not suggesting that a speaker ac-
tually starts with the declarative She is still working and then reverses the order of the
first two elements. Apart from the inherent implausibility of such an interpretation of
process terminology, it cannot be reconciled with the point illustrated in [3], namely
that in many cases a non-canonical clause has no grammatically well-formed canonical
counterpart.2 It is always possible to translate the process description into an equivalent
one couched in purely static terms. In the present example, we are merely saying that
the order of the auxiliary and the subject is the opposite of that found in canonical
clauses.
■ Extension of the apparatus for the representation of syntactic structure
The kind of syntactic analysis and representation we introduced in Ch. 1, §4.2, works
well for canonical constructions, but needs some extension to cater for certain kinds of
non-canonical construction. Compare, for example:
[4]
a. Liz bought a watch.
b. I wonder what Liz bought.
While [a] is a canonical clause, the underlined clause in [b] is non-canonical in two re-
spects: it is interrogative and subordinate. It is the interrogative feature that distin-
guishes it from the canonical [a], inasmuch as what is understood as object of bought
although its position relative to the verb differs from that of the object a watch in
canonical [a]. (Clause [a] is not the declarative counterpart of what Liz bought because
it contains the NP a watch, but it illustrates a comparable declarative structure.) The
representations we propose are as in [5].
[5]
a.
b.
Clause
Subject:
NP
Predicate:
VP
Preacher:
V
Object:
NP
Head:
N
Prenucleus:
NP i
Nucleus:
Clause
Subject:
NP
Predicate:
VP
The:
D
Head:
N
Head:
N
Object:
GAP i
bought
what
Liz
watch
a
bought
Liz
Head:
N
Clause
Preacher
V
––
Structure [a] needs no commentary at this stage: it is of the type introduced in Ch. 1. In
[b] what precedes the subject in what we call the prenucleus position: it is followed by
2Note also that what we present in this book is an informal descriptive grammar, not a formal generative
one: we are not deriving the ‘surface structure’ of sentences from abstract ‘underlying structures’. Thus
our process terminology is not to be interpreted as referring to operations performed as part of any such
derivation.
§ 2 Canonical and non-canonical clauses
49
the nucleus, which is realised by a clause with the familiar subject–predicate structure.
Within this nuclear clause there is no overt object present. But the prenuclear what
is understood as object, and this excludes the possibility of inserting a (direct) object
after bought:
*
I wonder what Liz bought a watch. We represent this by having the object
realised by a gap, an abstract element that is co-indexed with what (i.e. annotated with
the same subscript index, here ‘i ’): this device indicates that while what is in prenuclear
position, it also functions in a secondary or derivative sense as object of bought.
Note that it would not be satisfactory to replace the ‘prenucleus’ label by ‘object’,
and then simply dispense with the object element on the right of bought. Functions,
we have said, are relational concepts and ‘object’ is a relation between an NP and a VP
construction. Directly labelling what as object would not show that it is object of the VP
headed by bought. This can be seen more easily by considering such an example as [6],
where the bracketed clause has the structure shown in [7]:
[6]
I can’t remember [what Max said Liz bought ].
[7]
Clause
Prenucleus:
NP i
Nucleus:
Clause
Subject:
NP
Predicate:
VP
Comp:
Clause
Predicate:
VP
Subject:
NP
Head:
N
Head:
N
Preacher:
V
Head:
N
Preacher:
V
said
Liz
bought
Object:
GAP i
what
Max
––
What is in prenuclear position in the clause whose ultimate head is the verb said, but it
is understood as object of bought, not said. Simply labelling what as object would not
bring this out, whereas the co-indexed gap device does serve to relate what to the bought
VP whose object it is.
We make use of the same device to handle subject–auxiliary inversion. Compare
the structures in [8] for canonical He is ill and interrogative Is he ill? The nucleus in
[b] is identical to structure [a] except for the gap, and this accounts for the fact that
the functional relations between he, is, and ill are the same in the two clauses: he is
subject, ill is predicative complement, and is in [b] is shown to be predicator by virtue
of its link to the gap element that fills the predicator position directly. Main clause
interrogatives like What had Liz bought? will thus have one prenucleus + nucleus con-
struction (had + Liz bought) functioning as nucleus within another (what + had Liz
bought).
Chapter 2 Syntactic overview
50
[8]
a.
b.
Clause
Subject:
NP
Subject:
NP
Predicate:
VP
Predicate:
VP
Prenucleus:
V i
Nucleus:
Clause
Head:
N
Head:
N
Preacher:
V
Preacher:
GAP i
PredComp:
AdjP
PredComp
AdjP
ill
ill
is
is
he
he
Clause
––
■ Organisation of the grammar
The distinction between canonical and non-canonical clauses plays a major role in the
organisation of the present grammar. The early chapters deal predominantly with the
structure of canonical clauses, and with units smaller than the clause: phrases headed
by nouns, adjectives, adverbs, and prepositions. Chs. 9–16 then focus on non-canonical
constructions; subordination requires more extensive treatment than the other dimen-
sions mentioned above, and is covered in Chs. 11–14. The final chapter devoted to syntax
(Ch. 17) deals with deixis and anaphora, phenomena which cut across the primary part-
of-speech distinction between nouns, verbs, etc. There follow two chapters dealing with
the major branches of morphology, and we end with a short account of punctuation.
3 The verb
The head of a clause (the predicate) is realised by a VP, and the head of a VP (the
predicator) is realised by a verb. The verb thus functions as the ultimate head of a
clause, and is the syntactically most important element within it: properties of the verb
determine what other kinds of element are required or permitted.3
■ Inflection
Most verbs have six inflectional forms, illustrated here for the lexeme take:
[1]
preterite
I took her to school.
primary forms
3rd sg present tense
He takes her to school.
plain present tense
They take her to school.
plain form
I need to take her to school.
secondary forms
gerund-participle
We are taking her to school.
past participle
They have taken her to school.
Auxiliary verbs also have negative forms (She isn’t here, I can’t help it, etc.), while the
verb be has two preterite forms (was and were) and three present tense forms (am, is,
are). The were of I wish she were here we take to be an irrealis mood form, a relic of
an older system now found only with the verb be with a 1st or 3rd person singular
subject.
3 Since the verb is the ultimate head, we can identify clauses by the verb. In [6] of §2, for example, we can refer
to the most deeply embedded clause as the buy clause.
Posted by at