operator and negative contraction in spoken british english · is no clear distinction between...
TRANSCRIPT
1
Operator and Negative Contraction in Spoken British English: A Change in Progress
1
José Ramón Varela Pérez
University of Santiago de Compostela
1. Introduction
There is a growing interest among corpus linguists in changes of morphology and
syntax in standard varieties of English. So-called ‘comparative corpus linguistics'
(Leech et al. 2009) is now seen as an extension of a method that has been used in
historical (socio-)linguistics for some time, and which involves plotting current changes
in English grammar based on comparisons of well-matched corpora of English. Apart
from the growing number of papers on this new corpus-based approach (Mair and
Hundt 1995, 1997; Hundt and Mair 1999; Leech 2003, 2004; Smith 2003; Leech and
Smith 2006, Smith and Leech, this volume, among others), we have now at our disposal
three monographs that take a corpus-based approach to changes in progress in
contemporary English (Bauer 1994; Mair 2006; Leech et al. 2009).
Most of this new research enterprise has made use of the Brown family of
corpora of written English, i.e. Brown, FROWN, LOB, FLOB and others representing
different regional varieties of the language. However, it is widely acknowledged that
face-to-face spoken interaction is the source of most grammatical changes, and it is in
this medium that changes spread faster. Only recently have scholars started to make use
of comparable corpora to study on-going grammatical changes in spoken English. For
example, Leech (2003, 2004) has supplemented the study of contemporary written
English with the use of two ‘mini-corpora' for studying language change in recent
spoken British English. These subcorpora are based on the recently released Diachronic
Corpus of Present-Day Spoken English (DCPSE), a corpus compiled by Bas Aarts and
colleagues at University College London that is based on the London-Lund Corpus
(LLC) and the British Component of the International Corpus of English (ICE-GB).
The present paper shows the possibilities of using parts of LLC and ICE-GB in a
trend study that combines a real time and an apparent time approach to one aspect of the
grammar of negation in present-day English, viz. the variability between operator and
negative contraction. This is a feature that has been disregarded by previous studies of
grammatical change in contemporary standard varieties of English. Standard varieties of
English show variation in the way present tense forms of BE2 (is, are), HAVE (have,
has, had), will, and would realize verbal negation by means of the negative particle
NOT: (1) Full or uncontracted forms; (2) Negative contraction; (3) Operator
contraction:
(1) She is not at home and they are not either.
(2) She isn’t at home right now and they aren’t either.
(3) She’s not at home right now and they’re not either.
1 The research reported here was supported by the Autonomus Government of Galicia (INCITE
grant 08PXIB204033PR) and Generalitat de Catalunya (Grup de Recerca Consolidat UVAL – Grant
2009SGR650). These grants are gratefully acknowledged. I also thank Geoffrey Leech, Bas Aarts,
Ignacio Palacios Martínez, Sean Wallis, and Paloma Núñez Pertejo for their comments on this paper. All
remaining errors are mine. 2 I follow the common practice of using capital letters to indicate a form representing a lemma (HAVE)
and lower case italicized letters to represent individual word forms (have, has, had).
2
Negative contraction is realized by attaching the suffix -n’t to the operator in such a way
that the resulting form is a negative operator (isn’t, aren’t). With operator contraction
the operator becomes cliticized to a preceding host while not remains uncontracted. This
paper uses the methodology and aims of comparative corpus linguistics to address the
variation between these two contraction strategies in a diachronic corpus of
conversational British English comprising two samples taken from the London-Lund
Corpus (LLC) (dating from the late 1950s to the early 1970s) and from the British
Component of the International Corpus of English (ICE-GB). The paper is organized as
follows: Section 2 reviews some of the existing synchronic literature on this type of
variation in British English; Section 3 provides a diachronic perspective on the
grammaticalization of not and the emergence of negative and operator contraction in the
history of English; Section 4 introduces the main objectives of the present paper, the
data collection method, and the delimitation of the variable context; Sections 5 and 6
present and discuss the main findings. Finally, Section 7 offers the main conclusions.
2. Previous research
Some of the scholars who have undertaken a comparative grammatical study of
different regional varieties of British English have not mentioned variation between
negative and operator contraction (e.g. Trudgill and Hannah 1994) or have made casual
statements based on personal opinion rather than on empirical data. Occasionally, there
is no clear distinction between present tense BE and the other operators which allow
variation between operator and negative contraction (HAVE, will, and would). Trudgill
(1978: 13) ignores the status of present tense BE when he states that speakers in the
south of England tend to use contracted forms such as I won’t do it and I haven’t done
it, while “the further north one goes”, the more likely it will be to hear instances of
operator contraction such as I’ll not do it or I’ve not done it. He further hypothesizes
that ‘northern’ forms might be on the increase in the south “in certain, particularly
formal usages.” However, as he himself recognizes, asking informants how they use
these forms in different social and regional contexts “does not work.” Quirk et al.
(1985: 123) state that “generally speaking, the variant with contracted negation is more
common than the variant with contracted verb and full negation.” Similarly, Selkirk
(1981: 114), while recognizing that negative contraction with HAVE (e.g. I haven’t left)
is preferable to I’ve not left, speculates that with BE a sentence like She’s not leaving
“varies freely” with She isn’t leaving. Hughes, Trudgill and Watt (2005: 18) point out
that southern English speakers “quite frequently use the you’re not, we’re not, they’re
not forms alongside the more typically southern-type forms with aren’t.” Yet it is still
not clear what "quite frequently" means in this context. Finally, Swan (2005: 142)
remarks rather vaguely that “the two negative forms of BE (e.g. she isn’t and she’s not)
are both common in BrE.”
Recent empirical research has made a clear distinction between the behaviour of
BE and that of the other operators (HAVE, will, and would) across different dialects of
English. In an elicitation test carried out in a U.S. college, students reported that
operator contraction’s not and ’re not was more frequent in their speech than isn’t and
aren’t after personal pronouns and proper nouns (Greenbaum 1977: 98). In an early
corpus-based study of different registers in the LLC corpus, Hiller (1987) shows that
operator contraction with are and is was the dominant pattern (cf. Castillo González
2007, for similar results). Forms like aren’t (6%) and isn’t (21%) are found to be much
3
less common. Biber et al. (1999: 1132) also find a preference for operator contraction in
nearly all the registers investigated, although such a tendency is more marked in
conversation (70% vs. 10%) than in fiction (45% vs. 25%) and news (20% vs. 10%), no
doubt because in these two registers there is also competition with full forms.
Gasparrini (2001) finds that in the spoken registers of the British National
Corpus (BNC) such as sportscasts, club meetings, and call-in/chat programmes the
distribution of operator contraction was the opposite of the one predicted by Trudgill
(1978): South UK (75%) > Midland UK (74%) > North (67%). Likewise, in a study of
the demographically-sampled component of the BNC, Anderwald (2002: 75-76) points
out that the use of operator contraction for BE with subject pronouns is very high
overall, with a mean frequency of about 92% of all cases. Such a high frequency was
consistently found in all dialect areas, so in both the north and the south of Great Britain
operator contraction with BE is clearly the dominant strategy. One of the explanations
that she offers for the mismatch between her findings and what the literature says is that
since the publication of Trudgill’s comments, the contraction patterns have changed
radically, so this “can only be investigated by comparisons with historical data”
(Anderwald 2002: 85).
Tagliamonte and Smith (2002) conducted a comparative cross-variety approach
to the variation between operator contraction and negative contraction. Their analysis of
two southern and six northern varieties in Britain shows that, despite claims in the
literature that operator contraction is dominant in the northern communities, there are no
clear north-south, east-west, or Scots vs. English divisions as regards this variable,
which was thus found to be a poor diagnosis for distinguishing varieties of BrE on
geographic grounds (Tagliamonte and Smith 2002: 276). With regard to is and are,
Tagliamonte and Smith (2002: 270) find that most northern varieties have categorical or
near-categorical operator contraction. However, two northern communities (Maryport
and York) show a lower presence of operator contraction: Maryport (is 36%; are 93%),
York (is 56%; are 58%), while in the two southern varieties operator contraction was
also variable: Henfield, in West Sussex (is 84%; are 79%) and Tiverton, in Devon (38%
is; 57% are).
Finally, Yaeger-Dror, Hall-Lew, and Deckert’s (2002: 102-104) analysis of is +
NOT and are + NOT in nineteenth and early twentieth century British literary dialogue
offers mixed results. The Brontë sisters, born in York and writing mostly about northern
characters, were much more likely to use operator contraction than southern authors like
Hardy, Trollope or Virginia Wolf. However, two London novelists, Dickens and
Collins, used operator contraction in their novels to a very high degree, and a similar
situation was found among London teenagers in the Bergen Corpus of London Teenage
Language (COLT), where operator contraction was present in 86% of the cases
(Yaeger-Dror, Hall-Lew, and Deckert 2002: 96).
As regards the other operators (HAVE, will, would), Hiller (1987: 535-539) and
Castillo González (2007: 273-277) note that in the LLC corpus they prefer negation
with -n’t. Likewise, Anderwald (2002: 78-79)3 finds a very high frequency of negative
contraction across the different areas of the UK: 94.64%. Again there was no clear
south-north divide as postulated by Trudgill. In no region was operator contraction for
verbs other than BE the dominant strategy. The only areas where the ratio of operator
contraction was higher than 10% were South Midlands (19.7%), Central Midlands
(12.6%), and Scotland (10.9%). Gasparrini (2001) notes a similar preponderance of
3 Actually, Anderwald (2002: 75) also included CAN with the other operators on account of the fact that
CAN has “the two forms can’t and cannot.” Obviously, these are not instances of the variation between
operator and negative contraction so they should not have been included in her analysis of the data.
4
haven’t over ’ve not in other spoken texts in the BNC, while Biber et al. (1999: 1131-
1132) have confirmed Anderwald’s results. In the Longman Spoken and Written English
Corpus (LSWEC), negative contraction is overwhelmingly the preferred contraction
strategy with HAVE, will and would, while operator contraction is extremely rare in all
registers. In conversation, for example, operator contraction is used in only about 5% of
all cases. Finally, Tagliamonte and Smith (2002: 268-269) found very little or no use of
operator contraction with HAVE and would. The distribution of WILL, by contrast, was
different. There were three northern varieties where ’ll not was very frequent:
Cullybackey (91%), Cumnock (88%), and Wheatley Hill (70%). However, Buckie, a
Scots variety, had no cases of operator contraction with WILL.
Summing up, recent corpus-based accounts have demonstrated two things: (a)
There is little evidence to support the view that operator contraction increases “the
further north one goes.” (Trudgill 1978: 13); (b) HAVE, will, and would show an
overwhelming preference for negative contraction (haven’t, hasn’t, hadn’t, won’t,
wouldn’t) in negative declarative sentences in spoken and written varieties of English.
Present tense BE, on the other hand, has a special status in being more commonly
cliticized to a previous host in negative sentences, so that these two groups of operators
should have a different treatment in any study of this type of variation.
3. Diachronic background
The appearance of negative forms for auxiliaries (with the suffix -n’t) and evidence for
a system of clitic forms for BE, HAVE, will and would were two new formal properties
which characterized auxiliary operators and set them apart from the class of full verbs at
the beginning of the Early Modern English period (Warner 1993: 206-209).
Phonological reduction appeared simultaneously for these operators as an integral
part of grammaticalization, i.e. the process through which a series of premodal verbs
became auxiliaries (McElhinny 1993b: 375). The first written evidence of cliticized
operators appeared at the end of the 16th
century (Jespersen 1917: 117; McElhinny 1993b:
369). Early attestations consisted of instances of the cliticization of auxiliary verbs onto
pronominal hosts (McElhinny 1993b: 372). Comedies from the 1560s and the 1570s such
as Gummer Gurtons Needle and Roister Doister contain a handful of monosyllabic forms
such as Ile and Ill. By the end of the sixteenth century, the OED and concordances from
Shakespeare’s plays show that forms like the following were common: I’m, thou’rt, we’ll,
she’ld, he’s (‘he is’ and ‘he has’), I’ve, etc. Warner (1993: 208) suggests an even earlier
date for cliticization in speech, as evidenced by ME written spellings like ichulle (‘ich
wulle’) and ichot (‘ich wot’).
During the last stages of the grammaticalization of not in ME, the originally
strong negative particle seems to have weakened and become attached to an auxiliary
verb in the same way as ne had done before, as predicted by Jespersen's cycle.
Traditionally, it was assumed that contracted -n’t first came into use in speech around
1600 (Jespersen 1917: 117). However, recent research has dated the weakening of not in
speech earlier, to the 16
th century right after the loss of ne, as shown by transition forms
like wynnot, wilnot, shallnot, didnot, etc, as well as by occasional instances of negative
questions with pre-subject not (Risannen 1994, 1999). In any case, the first orthographic
instances of forms like of won’t, can’t, and don’t appeared in plays in the second decade
of the 17th
century, thus somewhat later than the development of operator contraction
(Brainerd 1989: 182-183).
5
It is difficult to know exactly which type of contraction predominated in
previous stages of the language. For one thing, there has always been a delay between
the weakening of operators and not in speech and their representation in writing.
Moreover, there have been no quantitative studies of the use of contractions in Early
Modern English, the period when contractions first became established in writing. For
late Modern British English, López-Couso (2007) offers some data on the use of
operator and negative contractions with BE and HAVE in ARCHER (1700-1899).
While instances of operator contraction for BE in negative sentences are found in the
first decade of the eighteenth century, occurrences of the negative operator an’t (‘am
not’; ‘is not’) are first attested in the corpus fifty years later. Regarding the diachronic
competition between the two variants, she points out that operator contraction
predominates over negative contraction in the eighteenth and the nineteenth centuries,
although the difference levels out somewhat during the second half of the nineteenth
century: operator contraction (60%) vs. negative contraction (40% of all cases) (López-
Couso 2007: 311). Regarding HAVE, although enclisis of this operator to the subject is
clearly disfavoured in ARCHER, there is a slight increase of this variant throughout the
whole period, so at the end of the nineteenth century forms such as ’ve not/’s not/’d not
represent over 17 percent of the cases with negative HAVE.
The review of the synchronic and the diachronic literature makes it clear that
corpus-based studies of present-day varieties of English show that variation between
operator and negative contraction no longer appears to provide a critical diagnostic for
situating dialects geographically in Britain (Anderwald 2002: 78-81; Tagliamonte and
Smith 2002: 276). All the varieties appear to be following the same pattern: present
tense BE has higher rates of operator contraction, while negative contraction is the
dominant strategy for HAVE, will, and would. The divergence of these results from
what is stated in the dialectal literature suggests that there has been some change in
contraction patterns in standard British English.
This suggestion is confirmed when we compare published findings from
ARCHER (1850-1899), LLC (1960s-1970s), and BNC-demographic (1990s). Tables 1
and 2 below show a comparison between the existing historical record in late Modern
British English (López-Couso 2007) and the literature for present-day British English.
Table 1. Operator and negative contraction with is and are in ARCHER (1700-1900)
(López-Couso 2007: 307), LLC (Hiller 1987: 540), and the demographically sampled
component of the BNC (Anderwald 2002: 78).
ARCHER
(1700-
1749)
ARCHER
(1750-
1799)
ARCHER
(1800-
1849)
ARCHER
(1850-
1899)
LLC
(1960s-
1970s)
BNC
Demog
(1990s)
’s not /
’re not
3
(100%)
8
(61.5%)
19
(82.6%)
33
(60%)
288
(83.2%)
7,202
(91.9%)
isn’t /
aren’t
−
5
(38.5%)
4
(17.4%)
22
(40%)
58
(16.8%)
636
(8.1%)
Total 3 13 23 55 346 7838
6
Table 2. Operator and negative contraction with HAVE in ARCHER (1700-1900)
(López-Couso 2007: 312), LLC (Hiller 1987: 538), and the demographically sampled
component of the BNC (Anderwald 2002: 78)4.
ARCHER
(1700-
1749)
ARCHER
(1750-
1799)
ARCHER
(1800-
1849)
ARCHER
(1850-
1899)
LLC
(1960s-
1970s)
BNC
Demog
(1990s)
’ve not /
’s not /
‘d not
−
− 1
(6.4%)
6
(17.1%)
25
(9.3%)
577
(5.4%)
haven’t /
hasn’t /
hadn’t
2
(100%)
1
(100%)
14
(93.3%)
29
(82.9%)
228
(90.7%)
11,807
(94.6%)
Total 2 1 15 35 253 12384
Bearing in mind that we are comparing speech-based written texts (ARCHER) with
transcriptions of conversation (LLC, BNC) and that the evidence offered by ARCHER
is too scanty for conclusive results, we find in Tables 1 and 2 diachronic evidence for an
increasing relative frequency of operator contraction with present-tense BE5 in more
recent times and, at the same time, the reverse process with HAVE, that is, a rise in the
relative frequency of negative contraction. These findings thus show that it would be
interesting to investigate whether there are current developments in the use of operator
and negative contraction in spoken British English and to determine the impact of
language-internal and external determinants on this type of variation.
4. Aims, data and methodology
4.1. Aims
In view of the existing research on this topic, my paper will address questions such as
the following:
(a) Is operator contraction, which seems to be the majority pattern, becoming more
frequent than negative contraction with BE? Is there any evidence of change in
progress among educated (mainly southern) speakers of British English?6 What
is the role of speaker age in the distribution of the two forms?
(b) What is the influence of language internal factors on the choice between ’s not
and isn’t? In spoken English the following constraints have been said to favour
one construction over the other: (i) the type of subject; (ii) the status of is/are
either as a copula or as an auxiliary; (iii) the presence of ellipsis of the
4 Data from Anderwald (2002: 76) include HAVE, will and would.
5 There is some evidence from Castillo González’s (2007: 319) study of the LOB and FLOB corpora that
in written British English there has been a similar rise in the frequency of operator contraction to the
detriment of negative contraction. Yet her figures are not easy to interpret because of the use in her
analysis of the frequency of each variant per thousand words instead of percentages of occurrence of each
variant. 6 The LLC and ICE-GB spoken data include educated mainly southern speakers of British English. Other
spoken corpora such as the demographic component of the BNC are more differentiated by class and
region.
7
predication. The present paper will explore these and other language internal
factors and will go beyond merely descriptive studies of grammatical variation
and offer possible functional, cognitive and frequency-related explanations that
account for the choice of pattern.
(c) Are there any discourse or pragmatic factors that favour either variant?
(d) Which contraction strategy is favoured in face-to-face conversation (vs. other
registers of spoken language)?
(e) Why is operator contraction favoured with BE whereas negative contraction
predominates with HAVE, will and would?
4.2. Data
In this paper, I will use two comparable samples of spoken British English taken from
two corpora originally compiled with different sampling and design characteristics over
periods of different length: the LLC corpus was collected over the period 1958-1977
(Svartvik 1990), and the ICE-GB corpus was collected over the 1990-1992 period
(Nelson,Willis, and Aarts 2002).7 The resulting selection (see Table 3 below) contains
(mainly conversational) spoken British English from the 1960s-mid 1970s and the early
1990s, thus covering a period between one and a half and two decades.
Table 3. Data files used in this study
Time Register / Files No. Words
LLC-mini-sp
(1960s-mid 1970s)
Private conversation: S.1.1-11; S.2.1-9; S.3.1-6;
S.4.1-7; S.5.8-11
Broadcast discussions and interviews: S.5.1-7;
S.1; S.3; S.6.5-7
Total number of words:
180,000
60,000
240,000 ICE-GB-mini-sp
(early 1990s)
Private conversation: S1A-001 – S1A-090
Broadcast discussions and interviews: S1B-021 –
SB1-050
Total number of words:
180,000
60,000
240,000
Private conversation and broadcast discussions and interviews are two registers of
spoken English that differ in terms of topic, type of addressee (private conversation vs.
conversation) and attention to speech (colloquial vs. more formal speech). All the
speakers represented in the two conversational samples are of British origin, mainly
from the south-east of England, and educated through the medium of English to at least
the end of second-level schooling. In addition, the samples are stratified by speaker age
(18-25, 26-45, and 46+) and sex.
Diachronic comparison between LLC and ICE must necessarily offer tentative
results “in the expectation that further corpus-based or other empirical research will
confirm and refine the findings” (Leech and Smith 2006: 187). In particular, it is
7 The recently released Diachronic Corpus of Present-Day Spoken English (DCPSE) includes about
800,000 words of spoken English from comparable categories in the LLC and ICE-GB (400,000 words
from each corpus) (Leech et al. 2009).
8
important to discuss some of the limitations of these two corpora for a comparative
sociolinguistic study:
(a) The fact that LLC was collected over a period of over 20 years makes it difficult
to compare findings obtained from this spoken corpus with those from the
Lancaster-Oslo/Bergen (LOB), a written corpus compiled with 1961 materials.8
(b) LLC and ICE-GB are two corpora that were originally compiled with somewhat
different sampling methods and corpus design features. For example, the text
samples differ in length: those of the LLC corpus are approximately 5,000
words in length, whereas those of the ICE-GB corpus are about 2,000 words in
length (Leech et al. 2009: 44).
(c) Compared to LLC, ICE-GB includes a larger amount of material from younger
speakers and women.
(d) While ICE-GB includes detailed information on speaker region in the UK, it is
unfortunately not possible to determine the region assigned to the speakers in
LLC, so this important factor for the choice between operator and negative
contraction cannot be taken into consideration in this study.
(e) Whereas most face-to-face conversations in LLC are surreptitious (i.e. one or
more of the participants did not know that their conversation was being
recorded), ICE-GB includes only non-surreptitious recordings.
(f) The individuals represented in the two samples are speakers of the standard
variety. Most of the recordings took place in the environment of University
College in London, so they include speech by teachers, students, administrators,
etc. This makes the two samples rather cohesive but not good representations of
conversational English spoken by people from different social, educational and
regional backgrounds.
Despite all these caveats, a comparative study of the two corpora, though provisional,
“would be preferable to a survey of recent grammatical change which took no account
of the spoken language” (Leech 2004: 64).
4.3. Methodology
An accurate definition of the variable context is crucial to a quantitative analysis of
grammatical variation. It is thus necessary to define clearly the envelope of variation,
i.e. all and only those contexts in which a given instance of variation could be possible.
Unlike other studies dealing with the same topic (Hiller 1987; Biber et al. 1999;
Tagliamonte and Smith 2002; Yaeger-Dror,Hall-Lew, and Deckert 2002; López-Couso
2007), which only consider variation between operator and negative contraction, I
decided to include instances of non-contraction in the analysis, the rationale being that
speakers do not first decide to contract and then choose which contraction to apply, but
they form an utterance and in the process may choose any of the three alternatives.
Am not does not have a variant with negative contraction in standard English
(e.g. *I amn’t watching TV), so this combination was omitted from further analysis.
Also absent from the corpora were instances of ain’t. Have, had, will, and would are
8 Leech’s (2004: 64) solution to this problem was to construct a sample from the SEU/LLC corpus that
rejected any material not contemporaneous with LOB, “a constraint we interpreted rather liberally to
exclude any material outside the time frame 1959-1965.” The resulting subcorpora (termed SEU-mini-sp
and ICE-GB-mini-sp) were rather small in size: 80,000 words each.
9
normally cliticized to a previous pronoun; combinations like ?All the men’ve not gone
are not institutionalized in written transcriptions of speech and, moreover, they may be
pronounced in the same way as All the men have gone, with a weak vowel, i.e. /ǝv/. Are
can only be attached to a previous personal pronoun (we, you, and they); preceding
nouns and pronouns such as these, there, which, etc., block the cliticization of are for
phonological reasons (e.g. ?These’re not the things that I need) (Hiller 1987: 539). In
the case of ’s (= has), no instances of operator contraction with non-pronominal hosts
were attested in negative sentences in the corpora, so they were also omitted.
Other cases in which alternation between the three variants was not possible
were left out of the analysis. The majority of knock-out contexts identified in the
samples involve instances in which it is impossible for the operator to be attached to the
subject due to phonological or syntactic reasons. Operator contraction with is is blocked
in the following cases:
(a) In tag questions and interrogatives with inverted subject-operator order (e.g. *’s
not he at home?).
(b) With a subject host ending in a sibilant consonant (LLC-mini-sp: n = 33; ICE-
GB-mini-sp: n = 8), as in (4) and (5)
(4) But this isn’t definite (ICE-GB:S1A-082 #63:1:A)
(5) He went to this awful down-market university which isn’t either Oxford
or Cambridge (ICE-GB:S1B-029 # 166-167:1:D)
(c) In sentences with no available subject host, as in (6) (LLC-mini-sp: n = 5; ICE-
GB-mini-sp: n = 1). This restriction also goes for the other operators:
(6) b it looks Tudor#
a but obviously isn’t# (LLC 4.2.398-399)
Finally, false starts and unintelligible utterances were also discarded from further
analysis:
(7) an Xray isn’t no## it’s an investigation# (LLC 2.9.1142-1143)
5. Results
5.1. HAVE, will, and would
Table 4 shows the overall distribution of full forms (have not, has not, had not, will not,
would not), negative contraction (’ve not; ’s not; ’d not; ’ll not; ’d not) and operator
contraction (haven’t; hasn’t; hadn’t; won’t; wouldn’t) with pronominal hosts in LLC-
mini-sp and ICE-GB-mini-sp.
10
Table 4. Overall distribution of full forms, negative contraction and operator contraction
with HAVE, will, and would across time
Full form Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-mini-sp 22 4.5% 436 89.3% 30 6.2% 488
ICE-GB-mini-sp 16 4.2% 354 93.4% 9 2.4% 379
Chi-square = 7.176; 1 df; p < 0.01
Full forms are uncommon because of the spoken nature of the two samples and their
frequency has remained stable over time. The ratio of operator contraction is very low
in both corpora, although in LLC-mini-sp (6.2%) it is more than twice as high as in
ICE-GB-mini-sp (2.4%). Despite the fact that the use of negative contraction with these
operators is at a very advanced stage in both samples, there is still some room for
change. In ICE-GB-mini-sp, the proportion of negative contraction represents 93.4% of
negative constructions and has risen by 4.2%. The difference between the two samples
is statistically significant. Here are some samples of operator contraction from both
corpora.
(8) I've not heard that we're having any trouble with it (ICE-GB:S1A-024
#12:1:A)
(9) here’s one fairly well he’s not been playing very well recently# (LLC
4.6a 32 286)
(10) Chief Faulkner turned it into a freehold# and we annexed it# that was a
great error# if we’d not annexed it# the lease would have just run out#
(LLC 6.7.994)
(11) when they are used they will be totally anonymous they’ll not be merely
anonymous they will have all the names changed (LLC 2.2a.244)
Although tokens such as those in (8)-(11) above were too rare to permit further
qualitative analysis, Table 5 reports the relative frequency of operator contraction for
each of the operators out of the number of forms.
Table 5. Frequency of operator contraction for HAVE (’ve not / ’s not / ’d not), will (’ll
not), and would (’d not)
’ve not; ’s not;
’d not
’ll not ’d not Total
LLC-mini-sp 28/307
(9.1%)
2/66
(3%)
0/115
(0%)
30/488
(6.1%)
ICE-GB-mini-sp 9/199
(4.5%)
0/64
(0%)
0/116
(0%)
9/379
(2.4%)
Compared to the rest of the HAVE forms, the relative frequency of ’ve not, ’s not and ’d
not in LLC-mini-sp is twice as high as that we find in ICE-GB-mini-sp (9.1% vs. 4.5%).
In addition, ’ll not is found twice in LLC.mini-sp but it is not attested in ICE-GB-mini-
sp. Finally, although forms like I’d not like to eat so much are theoretically possible
(Quirk et al. 1985: 777), would not never contracts to ’d not in either of the samples.
11
In both absolute and relative terms, ’ve not is the most frequent instance of
operator contraction attested in both corpora. 22 instances out of 31 in LOB and 5 out of
9 in FLOB involve this form. Since virtually the majority of the tokens of ’ve have
either I or you as subject hosts, we can say that variation between operator contraction
and negative contraction in negative sentences with HAVE, will, and would in present-
day spoken British English is to some extent restricted to collocations of ’ve and a
preceding first or second-person personal pronoun.
There are no instances of operator contraction either for main verb HAVE (?I’ve
not a lot of money) or for the semi-modals HAVE to and HAVE got to (?I’ve not to go
now). Regarding possessive HAVE got, the auxiliary is cliticized to a previous host only
once in each corpus. All other instances of HAVE got with verbal negation show
negative contraction.
(12) because you’ve not got a good road into London then# (LLC 11.1b.726)
(13) cos a actually thinking about it I’ve not got a uhm record player or
anything (ICE-GB:S1A-042 #135:1:C)
In order to check whether the distribution of the receding forms depends on speaker age
differences, I adopted an apparent-time approach (cf. Labov 1972) in Table 6, dividing
the subjects into a younger subject group (age ≤ 25) and an older one (age ≥ 25)
Table 6. Distribution of full forms, negative contraction and operator contraction with
HAVE, will, and would according to age
Age Full form Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-mini-sp ≤ 25
> 25
1
21
0.9%
5.5%
103
333
95.4%
87.6%
4
26
3.7%
6.9%
108
380
ICE-GB-mini-sp ≤ 2
> 25
2
14
1.8%
5.6%
106
223
97.3%
89.9%
1
8
0.9%
3.2%
109
248
Due to the relatively low number of instances of full forms and operator contraction, no
statistically significant differences can be reported between the age groups. In any case,
in both samples younger speakers choose negative contraction to a higher extent than
older ones. In fact, in the more recent corpus speakers under 25 hardly ever use operator
contraction. We seem to be facing the last stages of a change that has been going on for
quite some time and has reached the final stages of the slow-quick-slow S-curve. These
findings seem to be in line with Greenbaum’s (1977: 97-98) results of a test that
measured American students’ acceptance of forms such as ’ve not and haven’t. In the
test, the haven’t form was greatly preferred to’ve not.
5.2. Are
Actual variation between negative contraction and operator contraction with are is only
possible when this operator is preceded by a subject pronoun host, basically we, you,
and they (Hiller 1987: 539). Table 7 shows that the distribution of variants is quite one-
sided, with operator contraction representing the majority of the cases.
12
Table 7. Overall distribution of full forms (are not), negative contraction (aren’t) and
operator contraction (’re not) with we, you, and they as subject hosts across time
Full form
Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-mini-sp 6 4.9% 4 3.3% 112 91.8% 122
ICE-GB-mini-sp 8 6.5% 4 3.3% 110 90.2% 122
Not statistically significant
The -n’t type of negation is very rare in both samples, and operator contraction is used
in the large majority of cases in which are can be cliticized to a preceding personal
pronoun host. The use of ’re not is so advanced in both corpora that we cannot claim
that there is ongoing change with this form of BE. The low distribution of aren’t
compared to other negative operators has also been pointed out by Kjellmer (1998: 181)
for LOB (5%) and by Hiller (1987: 539) for the texts from LLC that he analysed (6%).
If the spread of grammatical change is normally characterized by a slow-quick-slow
rate, it can be said that the use of ’re not has reached the slow end of change. As a
result, examples like the following were rare in both corpora:
(14) and I think this is the way they (sic) England copes# is the fact that we
never# we [nev] we aren’t a militaristic nation# (LLC 2.3.557-560)
(15) I'm only sorry that we aren’t actually having a holiday in Provence
(ICE-GB:S1A-011 #64:1:A)
5.3. Is
5.3.1. Overall distribution across time
Compared to are, negation with the third person singular is shows rather more
interesting results. Table 8 and Figure 1 show the absolute figures and the percentages
of both variants in the two corpora
Table 8. Overall distribution of full forms (is not), negative contraction (isn’t) and
operator contraction (’s not) across time
Full form Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-mini-sp 35 10.2% 90 26.2% 218 63.6% 343
ICE-GB-mini-sp 28 7.8% 60 16.8% 270 75.4% 358
Chi-square =12.003; 2 df; p < 0.01
13
Figure 1. Overall distribution of full forms (is not), negative contraction (isn’t), and
operator contraction (‘s not) across time
These results are in accordance with observations by Hiller (1987), Anderwald (2002),
and Tagliamonte/Smith (2002), among others, that 3rd
person singular present tense
forms of BE show a strong preference for cliticization in negative sentences. ICE-GB-
mini-sp reveals that in a period of 15-20 years there has been a statistically significant
11.8% increase in the use of operator contraction and a corresponding decrease in the
frequency of negative contraction with is, just the opposite tendency that was found for
HAVE, will, and would. The high frequency of operator contraction in ICE-GB-mini-sp
further testifies to the fact that the replacement of negative contraction with operator
contraction has now reached an advanced stage in those contexts in which variation
between the two constructions is possible.
One more research question is to determine whether change is taking place in both
registers at the same time. In Table 9, statistically significant differences can be
observed if the two samples of private conversation are compared. There is a 12.5%
increase of ’s not in the more recent sample. Regarding the two samples of radio
broadcasts and interviews (Table 10), we also detect an increase in the percentage of
operator contraction (39.3% in LLC-mini-broadcast vs. 50.7% in ICE-GB-mini-
broadcast). In this case, though, the findings do not yield statistically significant results,
probably due to the low overall number of tokens in these two samples.
Table 9. Face-to-face conversation. Distribution of full forms (is not), negative
contraction (isn’t) and operator contraction (’s not) across time
Full form
Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-mini-conv 23 8.2% 65 23% 194 68.8% 282
ICE-GB-mini-conv 12 4.2% 42 14.5% 235 81.3% 289
Chi-square =12.236; 2 df; p < 0.001
LLC-mini-sp ICE-GB-mini-sp
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100% ’s not isn’t is not
14
Table 10. Broadcast discussions and interviews. Distribution of full forms (is not),
negative contraction (isn’t) and operator contraction (’s not) across time
Full form
Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-mini-broadc 12 19.7% 25 41% 24 39.3% 61
ICE-GB-mini-broadc 16 23.2% 18 26.1% 35 50.7% 69
Not statistically significant
Face-to-face conversation is considered to be the locus of most grammatical changes.
Tables 9 and 10 confirm that the tendency for operator contraction to be the preferred
variant is more advanced in private face-to-face conversation than in more formal
registers of spoken English, which still rely on is not and isn’t. Moreover, since both
registers point in the same direction, towards an increasing ratio of operator contraction
with’s not, we can be confident that there is on-going grammatical change in this area.
5.3.2. Register variation
Despite the fact that both operator and negative contraction “are found in modern
colloquial English” (Kjellmer 1998: 178), not much is known yet about their
distribution in different spoken registers. Quirk et al. (1985: 123) claim that ’s is typical
of informal and is of formal contexts, but their comments refer to the variation between
’s and full is in positive sentences. Hiller (1987: 540-541) claims that the distribution of
’s not and isn’t in conversation, radio discussions and speeches in LLC cannot be
ascribed to influences of register, but his figures for speeches are too low to yield
relevant results. Finally, a comparison between BNC context-governed registers
(Gasparrini 2001) and Anderwald’s (2002) results from the BNC demographic
component shows that, although isn’t is the less frequent pattern in both subcorpora, it
is better represented in the more formal conversations (sportscasts, club meetings, and
call-in/chat programmes).
Tables 11 and 12 show the distribution of the three variants in each corpus
according to the two registers of private conversation and broadcasts. Register does
seem to favour the selection of variants.
Table 11. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to register in the LLC sample
Full form Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-mini-conv 23 8.2% 65 23% 194 68.8% 282
LLC-mini-broadc 12 19.7% 25 41% 24 39.3% 61
Chi-square =19.509; 2 df; p < 0.001
15
Table 12. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to register in the ICE-GB sample
Full form Negative
contraction
Operator
contraction
Total
n % n % n %
ICE-GB-mini-conv 12 4.2% 42 14.5% 235 81.3% 289
ICE-GB-mini-broadc 16 23.2% 18 26.1% 35 50.7% 69
Chi-square =37.155; 2 df; p < 0.001
The distribution of the three variants is clearly dependent on register, which confirms
the evidence we found by comparing the results reported in Gasparrini (2001) and
Anderwald (2002). In both corpora, the ratio of both full forms (is not) and negative
contraction (isn’t) is higher in broadcasts, the more formal register. By contrast, ’s not
predominates in face-to-face conversational interaction whereas its presence in the more
formal broadcasts is less common. In each case the divergence between the two
registers is statistically significant. There is thus a gradation in the use of operator
contraction that is time and register-dependent: in earlier more formal recordings there
is a balance between ’s not and isn’t, while in more recent informal recordings ’s not is
widely preferred in the majority of cases: LLC-mini-broadcast (39.3%) < ICE-mini-
broadcast (50.7%%) < LLC-mini-conversation (68.8%) < ICE-mini-conversation
(81.3%). These results further confirm the claim that conversation is leading the
diffusion of grammatical change.
By comparing the distribution of ’s not in conversation and in broadcasts in each
corpus (Tables 11 and 12), it is clear that the difference between one and the other
register is very much the same: LLC (29.5%) vs. ICE-GB (30.6%). This means that the
relative frequency of operator contraction has risen in both registers to a similar extent.
5.3.3. Type of subject
It has been long known that pronoun subjects usually favour the occurrence of ’s while
non-pronominal subjects correlate with full is (Black 1977: 174; Quirk et al. 1985: 122-
123; López-Couso 2007: 317, among others). A similar constraint was found by Labov
(1972: 106) in his study of copula deletion and contraction in African-American
Vernacular English (AAVE). While deletion and contraction were favoured by a
preceding pronoun, these two processes were disfavoured by noun phrase subjects, and
this finding has been replicated in subsequent analyses of AAVE and Euro-American
varieties (McElhinny 1993a: 387-388).
Hiller (1987: 539-543) compares’s not/isn’t and ’re not/aren’t (thus disregarding
full forms) in the LLC spoken corpus. Operator contraction is the predominant form
when the subject of the sentence is a personal pronoun:’re not (83%) > ’s not (79%).
After a full NP or a proper name, however, we get the opposite result and negative
contraction predominates: isn’t (79%). These results are confirmed by Biber et al.
(1999: 1132), who find that in conversation operator contraction is the most frequent
option with 2nd
person pronouns (85%), 1st person plural pronouns (80%) and other
pronouns (70%). By contrast, negative contraction predominates with non-pronominal
subjects (55%).
16
Table 13. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to the type of subject host9
Type of subject Full form
Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-
mini-sp
Personal prons.
Other prons.
Full NPs
7
10
18
3%
13.9%
43.9%
50
25
15
21.6%
34.7%
36.6%
175
37
8
75.4%
51.4%
19.5%
232
72
41
Chi-square = 83.559; 4 df; p < 0.001
ICE-
GB-
mini-sp
Personal prons.
Other prons.
Full NPs
10
9
9
4.3%
9.6%
29%
31
19
10
13.3%
20.2%
32.3%
192
66
12
82.4%
70.2%
38.7%
233
94
31
Chi-square = 36.068; 4 df; p < 0.001
In Table 13 we see a clear distinction between pronominal and non-pronominal hosts.
Frequently recurring sequences of adjacent forms are more likely to be affected by
phonological reduction, which explains why the enclitic form ’s is more commonly
attached to personal pronouns and other pronouns such as that, who or what, while the
less frequent full noun phrases promote isn’t and the full form is not. We also notice
that the rise in frequency of operator contraction has affected personal pronouns and the
rest of the hosts to a different degree. The change has advanced faster with the least
frequent combinations, i.e. with other pronouns and full noun phrases.
5.3.4. Ellipsis of the predication
Operators function in a range of reduced constructions where the main verb and its
complements are omitted by ellipsis and the sentence is understood to repeat the content
of an earlier sentence. This type of operator function has been referred to as ‘code’ or
‘stranding’ (Quirk et al. 1985: 126; Huddleston and Pullum 2002: 99). The only scholar
who has shown that the effect of eliding the predication after is + NOT and are + NOT
is important in the choice of variant is Hiller (1987: 543-544). He finds that operator
contraction is more frequent in 'mid-position' in the LLC corpus. However, when there
is ellipsis of the elements following the operator negative contraction becomes the more
common alternative (73% of the cases). Confirming this, in the present study instances
like (16) and (17) were found to be more common than (18).
(16) no, it isn’t. (LLC 2.1.940)
(17) well there isn’t actually (LLC 3.3.119)
(18) I suppose it's not. (LLC 3.1.1280)
9 McElhinny (1993a: 386) finds that preceding vowels overwhelmingly favour and preceding consonants
disfavour contraction of auxiliaries. This difference in phonological constraints is “what one might
expect, given the universal tendencies towards CV syllable structure.” Such a factor, however, did not
play any role in the choice of variant in my data. As a matter of fact, pronouns ending in a consonant such
as it and that highly promote operator contraction.
17
In (16) and (17), isn’t carries the nuclear focus of the utterance. In (18), it is the full
negator not that carries nuclear focus. In the corpora, occurrences of the operator in end-
position in negative sentences usually represent: (a) an answer to a previous yes/no
question; (b) an explicit denial of a previous statement; (c) a statement that agrees with
a previous negative suggestion. Table 14 shows the effect of eliding the predication in
negative sentences with is.
Table 14. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to ellipsis of the predication
Ellipsis of the
predication
Full form Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-
mini-sp
Ellipsis
No ellipsis
0
35
0%
10.9%
18
72
81.8%
22.4%
4
214
18.2%
66.7%
22
321
Chi-square = 37.689; 2 df; p < 0.001
ICE-GB-
mini-sp
Ellipsis
No ellipsis
2
26
5.6%
8.1%
13
47
36.1%
14.6%
21
249
58.3%
77.3%
36
322
Chi-square = 10.755; 2 df; p < 0.01
These findings are in accordance with Hiller (1987: 543). In both corpora, the
distribution of ’s not is higher in cases of non-ellipsis than with ellipsis of the
predication although the difference between the two contexts has decreased
spectacularly over time. In LLC-mini-sp there is a wide divergence in the distribution of
the two variants in the two contexts. Isn’t predominates with ellipsis (81.8%), whereas
’s not is more frequent in mid-position in the sentence (66.7%). A radical change has
occurred in ICE-GB-mini-sp, where we notice that the frequency of ’s not in contexts of
ellipsis has risen spectacularly, in such a way that it is now the majority pattern
(58.3%). The frequency of ’s not, though, is still lower in this context than in those
cases in which the whole predication is present, where we notice a 10.6% rise in its
distribution. Once again we see that the increase in the frequency of ’s not is much more
noticeable in a linguistic context where it was not so well represented in LLC-mini-sp.
One important factor not considered in previous research is the informational
status of the operator in cases of ellipsis: discourse-old ellipsis and discourse-new
ellipsis (Huddleston and Pullum 2002: 99). In (19) below the stranded operator (the is of
isn’t) is discourse-old in the sense that it, as well as the missing material (poison),
occurs in the preceding context. In (20), on the other hand, the stranded operator is new
to the discourse.
(19) I mean it cos we more or less know what’s poison and what isn’t (LLC
2.10.1437) [is = discourse-old] + neg [discourse-new]
(20) Well no because you’re still doing it in four four time and it isn’t (ICE-
GB:S1A-026-338-340) [is = discourse-new] + neg [discourse-new]
18
Table 15. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to the informational status of the operator in cases of
ellipsis of the predication
Information
status of the
operator
Full form Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-
mini-sp
Discourse-old
Discourse-new
0
0
0%
0%
13
5
76.5%
100%
4
0
23.5%
0%
17
5
ICE-GB-
mini-sp
Discourse-old
Discourse-new
1
1
3.4%
14.3%
7
6
24.1%
85.7%
21
0
72.4%
0%
29
7
Table 15 reveals two things. First, although in LLC-mini-sp isn’t is the preferred option
when is is discourse-old (76.5%), this preference is reversed in ICE-GB-mini-sp, where
s not has gained a lot of ground and isn’t only represents 24.1% of discourse-old cases.
Secondly, Table 15 shows that isn’t, with full is, is deeply entrenched in those contexts
of ellipsis of the predication in which the operator encodes discourse-new information.
All instances of ’s not, i.e. cases in which the operator is phonologically reduced,
express discourse-old information.
5.3.5. Denials and correctives
The potential difference in meaning between ’s not and isn’t has not been much
discussed in the literature. Biber et al. (1999: 167) claim that, since not is unreduced in
cases of operator contraction, this variant “may be felt as the more emphatic alternative
in clause-final position, and perhaps more generally.” More specifically, they claim that
unreduced not frequently occurs “where there is a marked contrast with the immediate
context.” Hiller (1987: 545-546) argues otherwise. He offers quantitative data from the
prosodically annotated version of the LLC corpus to show that isn’t is slightly more
common when one of the variants carries its own tonal nucleus. More importantly, he
further notes that isn’t “bears much more stress and emphasis” than ’s not, claiming
that“[t]his is particularly true whenever two clauses are contrasted in which two nouns
are opposed” (Hiller 1987: 545):
(21) this this is the truth# it isn’t a story (LLC 11.3f.13)
(22) it’s an investigation# it isn’t a sigh (LLC 2.9.1144-5)
Unfortunately, neither Hiller (1987) nor Biber et al. (1999) define in a clear way what
they mean by “emphasis” nor do they support their claim with empirical data. In a much
earlier comment on the difference in meaning between he’s not a fool and he isn’t a
fool, Sweet (1900: 126) has suggested that in the former sentence
the not is detached from the verb, and is thus at liberty to modify the following
noun.” As a result, the sentence is equivalent to ‘I assert that he is the opposite
of a fool’. On the other hand, in he isn’t a fool the negation is attached to the
operator and “must necessarily logically modify the whole sentence (…) so that
the sentence is equivalent to ‘I deny that he is a fool’.
19
Jespersen (1917: 126) links this difference to that between ‘special’ (i.e. local,
constituent) negation and ‘nexal’ (i.e. sentence, clausal) negation. Yet both sentences
pass Klima’s (1964) syntactic tests for sentence negation (he’s not a fool, is he?; he
isn’t a fool, is he?; he’s not a fool, and neither is his sister; he isn’t a fool, and neither is
his sister, etc.). In fact, in another part of his monograph Jespersen does not attach much
importance to the difference in meaning between the two forms of negation:
Similarly, it seems to be of no importance whether we look upon one notion only
or the whole nexus as being negative in she is not happy = ‘she is (positive) not-
happy’ or ‘she is not (negative nexus) happy’ (Jespersen 1917: 43).
In the process of conducting this analysis, it was found that the discourse function of
negation exerted a strong effect on the choice of construction. Thus, ’s not and isn’t may
appear in three discourse environments: (a) explicit denials; (b) implicit denials; and (c)
corrective sentences. Explicit denials, as in (23)-(24), deny a proposition which has
been explicitly asserted, while implicit denials, as in (25)-(26) deny something that
might merely be expected or which can be contextually inferred but has not been
asserted by anybody (Tottie 1991: 21). Corrective sentences, as in (27)-(28), take the
form not/-n’t X but Y, in which case the focus of the negative is the constituent before
but, which need not be made explicit in the sentence:
(23) A: I mean it sounds like there's uh technical <,> things like release
work and it sounds like some kind of therapy that you're doing
B: Yeah <,,> Uhm <,> I think the answer to that is no it's not
therapy uhm <,> and I'm certainly I 'm not trained as a <,> as a
therapist (ICE-GB:S1A-004 #90:1:B)
(24) l: and it’s quite silly to try and blame ourselves # for something
that’s quite beyond our control
f: Bill Mallalieu very shortly #
m: it isn’t beyond our control# (LLC 1.54.929-932)
(25) (coughs) well Billy`s always said that there isn`t enough water in this
country to make hydroelectric possible (LLC 2.8b.97)
(26) A: Have you seen it before Caroline
C: I have never seen it still I really <unclear-words> really it's not
my sort of film <,,> (ICE-GB:S1A-049 #282:1:B)
(27) … I’ve suddenly realized # it’s he’s not there as a dramatist now # he’s
there as a satirist (LLC 1.4.290)
(28) That’s not the guy you were talking to That was someone else (ICE-
GB:S1A-058 #13:1:B)
20
Table 16. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to discourse function: correctives, implicit denials and
explicit denials
Type of subject Full form
Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-
mini-sp
Correctives
Implicit denials
Explicit denials
4
28
3
16.7%
10.6%
5.4%
3
61
26
12.5%
23.1%
47.3%
17
175
26
70.8%
66.3%
47.3%
24
264
55
Chi-square = 17.032; 4 df; p < 0.01
ICE-
GB-
mini-sp
Correctives
Implicit denials
Explicit denials
1
24
3
4%
9.1%
4.2%
1
40
19
4%
15.2%
26.4%
23
199
50
92%
75.7%
69.4%
25
263
72
Chi-square cannot be calculated
As predicted by Tottie (1991), Table 16 shows that implicit denials are much more
numerous than explicit denials, even in spoken language. In both LLC-mini-sp and ICE-
GB-mini-sp, corrective contexts are extremely favourable to ’s not (70.8% and 92%,
respectively), followed by implicit denials (66.3% and 75.7%, respectively). The least
favourable context for ’s not in both corpora is that of explicit denials: 47.3% of the
cases in LLC-mini-sp, while in ICE-GB-mini-sp the frequency of ’s not in this context
has risen to 69.4%.
Negative contraction (isn’t) is better represented in explicit denials than in
implicit denials and correctives in both corpora. LLC-mini-sp shows a balance between
negative contraction and operator contraction, both of which are equally common
(47.3% each). In ICE-GB-mini-sp, the ratio of isn’t in explicit denials has decreased to
25.4%, a percentage that is still higher than that for correctives and implicit denials.
In any case, the highest increase of ’s not in ICE-GB-mini-sp is detected
precisely in the two contexts where it shows the lowest absolute frequency, i.e. in
correctives and explicit denials.
5.3.6. Function of the operator
Few scholars have addressed the impact of this factor in negative sentences. Kjellmer's
(1998: 172) results are a bit misleading because his figures for the auxiliary function
combine instances of BE and HAVE, and these two operators show different behaviour
with respect to contraction. Moreover, he collapses under the heading “uncontracted”
both full forms and operator contraction. Hiller (1987: 545) and Tagliamonte (2002:
276) show that in spoken British English the use of is/are as an auxiliary preceding the
progressive promotes the use of operator contraction, although differences were found
to be non-significant. Similarly, Yaeger-Dror, Hall-Lew, and Deckert (2002: 99) do not
find any significant correlation between type of contraction and the distinction between
a copula and an auxiliary. Table 17 illustrates the distribution of variants according to
the function of is.
21
Table 17. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to the function of the operator
Function of
the operator
Full form Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-
mini-sp
Copula
Auxiliary
31
4
7.4%
13.3%
84
6
26.8%
20%
198
20
63.3%
66.7%
313
30
Not statistically significant
ICE-GB-
mini-sp
Copula
Auxiliary
23
5
7.4%
10.9%
53
7
17%
15.2%
236
34
75.6%
73.9%
312
46
Not statistically significant
The results show that the effect of the function of is on the choice of variant is
negligible (see Tagliamonte and Smith 2002: 276, for similar results). We cannot say
that auxiliary is turns out to be a much stronger constraint favouring the frequency of
operator contraction. Copula and auxiliary instances cliticize onto a previous host to a
similar degree.
5.3.7. Speaker age
The influence of age patterns on the choice of the two variants in standard varieties of
English has not been taken into account in previous research. In order to combine a real-
time approach and an apparent-time approach to change, speakers were divided into
three groups: a younger subject group (18-25 years of age), an in-between group (26-45)
and an older group (+46). The assumption was that the language of the younger subjects
should be indicative of a newer language state than the language of the older subjects.
Table 18 and Figure 2 illustrate the relative frequency of isn’t and ’s not according to
the three age groups in both corpora.
Table 18. Distribution of full forms (is not), negative contraction (isn’t) and operator
contraction (’s not) according to speaker age
Speaker age Full form
Negative
contraction
Operator
contraction
Total
n % n % n %
LLC-
mini-sp
18-25
26-45
+46
1
25
8
1.8%
14%
7.6%
12
43
34
21.1%
24.2%
32.4%
44
110
63
77.2%
61.8%
60%
57
178
105
Chi-square = 11.727; 4 df; p < 0.05
ICE-
GB-
mini-sp
18-25
26-45
+46
5
14
9
3.9%
11.3%
9.7%
13
24
21
10.1%
19.4%
22.6%
111
86
63
86%
69.4%
67.7%
129
124
93
Chi-square = 13.776; 4 df; p < 0.01
22
Figure 2. Distribution of operator contraction (’s not) according to speaker age across
time
In both corpora, the proportion of ’s not and isn’t are distributed in the same way across
the three age groups. There is a steady rise in the use of ’s not as the speaker age
decreases, with a peak in the youngest age group and the lowest frequency in the oldest
group, a pattern which is suggestive of age grading. At the same time, there is an
increasing ratio of negative contraction with isn’t as speaker age increases. Assuming
that the language of the older subjects is indicative of an older language state and given
the fact that for each of the age groups the percentage for ’s not is a bit higher in ICE-
GB-mini-sp than in LLC-mini-sp, we can confirm the existence of change in real time.
5.3.8. Interaction of language-internal factors
After reviewing the way a series of factors, considered individually, affect the choice of
variants differently, the aim of this section is to consider how the different language-
internal factors interrelate. When a speaker chooses a certain contraction strategy, the
choice is influenced by a combination of coexisting influences, each of which has a
distinct relative force compared to the others.
18-25 26-45 46+
LLC-mini-sp
ICE-GB-mini-sp
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
23
0
10
20
30
40
50
60
70
80
90
100
-n/-e/-d -n/-e/+d -n/+e/+d +n/-e/-d -n/+e/-d +n/-e/+d +n/+e/-d +n/+e/+d
%
's not isn't is not
Figure 3. LLC-mini-spoken. Interaction of language-internal factors influencing the
choice of variant (n = full NP; e = ellipsis of the predication; d = explicit denial).
0
10
20
30
40
50
60
70
80
90
100
-n/-e/-d -n/-e/+d -n/+e/+d +n/-e/-d -n/+e/-d +n/-e/+d +n/+e/-d +n/+e/+d
%
's not isn't is not
Figure 4. ICE-GB-mini-spoken. Interaction of language-internal factors influencing the
choice of variant (n = full NP; e = ellipsis of the predication; d = explicit denial)
We can observe in the two corpora a parallel hierarchy in the distribution of the
statistically-significant language-internal factors influencing the choice of variant. In
both corpora, the rate of ’s not is highest when the subject is not a full NP (-n), when
there is no ellipsis of the predication (-e) and when the sentence does not express an
explicit denial (-d). The second highest rate for ’s not occurs when the subject is not a
full NP (-n), when there is no ellipsis (-e) and the negation expresses an explicit denial
(+e). As a result of these combinations, we can deduce that the most important factor in
the choice of ’s not is the type of subject, followed by ellipsis of the predication and
finally explicit denial.
24
Although the hierarchy of factors involved in the preference for ’s not is similar
in both corpora, the strength of each factor varies from one corpus to the other. In LLC-
mini-sp, there are only two combinations where the rate of ’s not is higher than that of
isn’t and both involve the use of a pronoun subject and non-ellipsis of the predication.
In ICE-GB-mini, ’s not is the preferred choice in five combinations. In the latter corpus,
the surviving stronghold for isn’t is represented by three combinations: a full NP plus
both ellipsis and explicit denial, a full NP plus ellipsis and implicit denial, and a full NP
plus non-ellipsis and explicit denial.
6. Discussion
An important question posed by the results reported above is the existence of a
mismatch between present tense BE, which prefers operator contraction, and HAVE,
will, and would, which overwhelmingly take negative contraction. After all, other
contractible operators such as DO, can, could, shall, etc., only take -n’t. Anderwald
(2002: 91-92; 2004: 60-62) considers that the entire paradigm of present tense BE
behaves irregularly and that this phenomenon of a “localized hierarchy reversal” is well
attested in linguistic typology. BE is very different from other auxiliaries in that it
carries the least semantic content. With the cliticization of is and are the amount of
coding material thus matches the semantic content to be coded.
Like other auxiliaries, copulas cannot normally function as predicates on their
own (Pustet 2003: 6). These features are reflected in the fact that in some languages and
under varying circumstances the copula can be omitted without affecting the meaning of
the sentence (Hengeveld 1992: 32; Stassen 1997: 65). For this reason, it is not
surprising that is and are tend to be contracted (She's not, we're not) rather than the
negative, which carries a greater semantic load.
There is further evidence in support of Anderwald's link between the empty
meaning of is and are and their tendency to appear in reduced form. In many languages,
copulas are sensitive to grammatical categories (Ferguson 1971: 147). Thus, copulas
tend to be dropped in the least marked (i.e. most frequent) contexts: (a) in the present
tense, which is also the least marked form in terms of cognitive complexity; (b) with a
pronoun subject; and (c) in informal style (Pustet 2003: 187). These are precisely the
contexts in which’s and ’re are the majority forms in the corpora. Even in languages
that do require a copula, such as English, the copula has a merely supportive role; the
copula "enables a non-verbal predicate to act as a main predicate" (Hengeveld 1992:
32). For example, the non-verbal predicate imposes the selection restrictions (*This
table is ill vs. Sheila is ill) and also determines the number of arguments (*This book is
identical vs. This book is identical to that one) (Hengevald 1992: 29). Thus, the copula
simply functions as a carrier for tense, mood, person, number, etc. In fact, in English is
and are may be left out in cases in which both the subject and the operator can be
deduced from the context.
(29) A: That is a reason to want to get married. Not the most romantic reason
in the world but… (ICE-GB:S1A-050 #198:1:A)
The correlation between the type of subject and contraction is easy to establish.
Historically, the first instances of the cliticization of auxiliaries occurred with
pronominal hosts (McElhinny 1993a: 372; López-Couso 2007: 317). Frequent repetition
makes processing easier and more automated through phonological reduction (Zipf
1935). Conventionalized contractions in English are reduced due to their high frequency
25
(Bybee and Thompson 1997: 65-66). Krug (1998) has shown that the contraction of
auxiliary verbs varies with the string frequency, i.e. the joint frequency of the subject
and the auxiliary (see also Biber et al. 1999: 166-167; Krug 2003: 23-33). Given that
pronominal subjects are much more frequent than lexical subjects and that ’s is among
the 10 most frequent words in the BNC (Leech, Rayson, and Wilson 2001), it does not
come as a surprise that operator contraction is more restricted with non-pronominal
subjects.
There are additional reasons why non-pronominal subjects tend to take isn’t or is
not, the two variants with the full form of the operator, instead of ’s not. According to
the so-called 'complexity principle' (Rohdenburg 1996: 151), more explicit grammatical
options will be favoured in cognitively more complex environments. Using data from
past and present-day varieties of English, Bailey, Maynor, and Cuckor-Avila (1989:
291) show that 'heavy' noun phrases are more likely to be followed by an inflected verb
than personal pronouns. Furthermore, according to Labov (1972: 85), in AAVE the
single most important constraint on deletion and contraction of the copula is "whether or
not the subject is a pronoun or some other noun phrase (…). In every case, the
percentages of deleted and contracted forms are greater when a pronoun precedes."
Regarding the variation between’s not and isn’t/is not, it is clear that the more complex
(and difficult to process) the subject becomes, the longer it takes to recognize the
relationship between the subject and the following operator. Whereas the more explicit
full form of is (isn’t/is not) unambiguously stands for a tensed verb, the contracted
operator (’s) is more ambiguous. For example, as a clitic, ’s may function as one of the
markers of perfect aspect (He’s finished his meal).
Two of the contexts in which isn’t seems more firmly entrenched are represented
by cases of ellipsis of the predication and explicit denials. On the other hand, ’s not is
especially common in mid-position, in implicit denials and corrective clauses. Two
iconic principles seem to account for the correlation between meaning and form in these
two variants. First of all, according to the so-called “quantity principle” (Zipf 1935;
Givón 1991: 87-89), a larger chunk of form will be given to: (a) a larger chunk of
information; (b) less predictable information; (c) more important information. In the
second place, the “proximity principle” predicts that entities that are closer together
functionally, conceptually, or cognitively will be placed closer together at the code
level, i.e. temporally or spatially (Givón 1991: 89; see also Haiman 1985: 106 and
Bybee 1985: 12).
The data have also shown that isn’t, with a full copula and sentence stress on the
operator, is surviving longer in contexts in which the copula makes a clear contribution
to the meaning of the clause, i.e. in explicit denials and contexts of ellipsis of the
predication. In the latter context, the real predicate is omitted and is becomes a predicate
of its own, so that it is more likely to appear in full form. As we saw above, this is
particularly the case when the operator encodes hearer-new information. When the
operator expresses discourse-old information, it may be reduced phonologically ('s not).
In explicit denials, the negative particle is often attached to the element to which
it is more relevant conceptually, i.e. the operator, which carries sentence stress. In (30),
for example, the form of contraction (isn’t) evinces that the negator goes with is, i.e. the
speaker expresses a disagreement with a previous assertion that there is “something
that’s quite beyond our control.” The resulting sentence (“it isn’t beyond our control”)
can be paraphrased as “(It is not the case / I deny) that it is beyond our control.”
(30) l: and it’s quite silly to try and blame ourselves # for something
that’s quite beyond our control
26
f: Bill Mallalieu very shortly #
m: it isn’t beyond our control# (LLC 1.54.929-932)
In implicit denials like (31), the speaker expresses a positive assessment of the fact that
the film under discussion is different from the ones that he likes, i.e. “it’s not my sort of
film”, so in such cases the full negator tends to be placed iconically next to the predicate
“my sort of film.”
(31) A: Have you seen it before Caroline?
C: I have never seen it still I really <unclear-words> really it’s not
my sort of film (ICE-GB:S1A-049 #282:1:B)
Data from explicit and implicit denials confirm Sweet's (1900: 126) suggestion (see
above) that’s not is much more common in assertions of negative facts, where the
negation is attached to the predicate, while isn’t is typical of explicit denials, where
negation is attached to the operator. The data also give some credence to Hiller's
observation that’s not is especially frequent non-contrastively and non-emphatically in
understatements. In such cases, not is attached to a following adjective and the speaker's
communicative strategy is not meant to be negative but rather positive: "From a
pragmatic point of view, "'not bad' or 'not well off' e.g. are just another way of saying
'good' or 'badly off'" (Hiller 1987: 545).
In corrective sentences of the type not/-n’t X but Y, there is a similar tendency
for the full negator (not) to be associated with the following predicate X, which is the
focus of negation. In (32) below, the placement of full not mirrors its close association
with the following noun phrase (“the guy you were talking to”). As a result, the copula
becomes associated with the second term of the correction, i.e. “someone else”. The
sentence may thus be paraphrased as “That (guy) is someone else, not the guy you were
talking to”:
(32) That's not the guy you were talking to. That was someone else (ICE-
GB:S1A-058 #13:1:B)
Also note that a sentence like (32) is similar to other corrective sentences in which the
full negator is placed immediately after the predicate, as in She saw not the play but the
opera. In such a sentence, the scope of the negative word not includes the direct object
but not the verb saw. The sentence can thus be paraphrased as “It was not the play but
the opera that she saw” (Quirk et al. 1985: 122)
Further support for the observation that one of the strongholds for isn’t is
represented by explicit denials while ’s not is much more advanced in contrastive
sentences comes from Yaeger-Dror’s (1997: 2-3) observation that negative contraction
is conditioned by interactional type. According to the so-called Cognitive Prominence
Principle, the negative word remains uncontracted when it expresses semantically focal
information, as is the case with full not in correctives. On the other hand, there is a
Social Agreement Principle that predicts contraction of the negative when it is face-
threatening to focus on disagreement, as is the case with explicit denials. Speakers are
thus more likely to use isn’t when they minimize the extent to which they disagree with
coparticipants.
27
7. Conclusions
Corpus-based data on real and apparent time drawn from two matching samples of
spoken educated British English has confirmed that there is an increasing polarization
between BE, which favours operator contraction (’s not and ’re not rather than isn’t and
aren’t), and the other contractible forms HAVE, will and would, which favour negative
contraction (haven’t rather than ’ve not, etc.). The near-categorical use of negative
contraction with HAVE, will, and would is in line with that of other auxiliaries such as
can, could, DO, etc., which are increasingly taking -n’t in registers that are prone to the
influences of speech. Operator contraction percentages for HAVE, will, and would are
so low that it is very difficult to determine what factors influence the choice of this
variant apart from the strength of collocations. The majority of the instances of operator
contraction in the two corpora are represented by frequent ready-made collocations like
I’ve not and you’ve not. These are combinations that are also extremely common in
positive sentences, which makes them strong in memory and easy to access as wholes,
so they resist the adoption of the more productive pattern with -n’t (cf. Bybee and
Thompson 1997: 67).
There is also evidence of an on-going change that involves the increasing
productivity of operator contraction with is in those contexts in which there is variation
between is not, ’s not and isn’t. The trend is consistent across the two registers
considered in this study. Thus, in both conversation and broadcast discussions we
observe that operator contraction (i.e.’s not) is much more common than negative
contraction (i.e. isn’t). Although the frequency of ’s not has risen across the board in
ICEBG-mini-sp, the increase has become sharper in those linguistic contexts in which
isn’t used to better represented. As a result, ’s not is now the preferred choice with non-
personal subject hosts, in cases of ellipsis of the predication and in explicit denials.
Finally, it is the youngest generation that seems to favour the use of ’s not in both
corpus samples, while the oldest generation shows a significantly greater tendency to
cling on to the use of the fast-waning isn’t.
These results are provisional in the expectation that further corpus-based or other
empirical research will either confirm or refine the findings (Leech and Smith 2006:
187). More synchronic and diachronic data are needed to understand the effect of
variables such as region, social class and education on these two types of variation.
References
Anderwald, Lieslotte. 2002. Negation in Non-Standard British English: Gaps,
Regularizations and Asymmetries. London and New York: Routledge.
Anderwald, Lieslotte. 2004. ‘Local Markedness as a Heuristic Tool in Dialectology:
The Case of amn’t.’ In: Bernd Kortman (ed.), Dialectology Meets Typology:
Dialect Grammar from a Cross-linguistic Perspective. Berlin and New York:
Mouton de Gruyter, pp. 47-67.
Bailey, Guy, Natalie Maynor and Patricia Cuckor-Avila. 1989. ‘Variation in Subject-
Verb Concord in Early Modern English.’ Language Variation and Change 1,
285-300.
Bauer, Laurie. 1994. Watching English Change: An Introduction to the Study of Linguistic
Change in Standard Englishes in the Twentieth Century. London: Longman.
Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward Finegan.
1999. Longman Grammar of Spoken and Written English. Harlow: Pearson.
28
Black, Michael. 1977. ‘An Investigation into Factors Influencing the Choice between
the Syllabic and Contracted Form of is.’ In: Wolf-Dietrich Bald and Robert Ilson
(eds.). Studies in English Usage: The Resources of a Present-Day English for
Linguistic Analysis. Frankfurt and Bern: Peter Lang, pp. 171-183.
Brainerd, Barron. 1989. ‘The Contractions of not: A Historical Note.’ Journal of
English Lingusitics 22, 176-196.
Bybee, Joan. 1985. ‘Diagrammatic Iconicity in Stem-Inflection Relations.’ In: John
Haiman (ed.). Iconicity in Syntax. Amsterdam and Philadelphia: John
Benjamins, pp. 11-48.
Bybee, Joan and Sandra Thompson. 1997. ‘Three Frequency Effects in Syntax.’
Berkeley Linguistics Society 23, 65-75.
Castillo González, Pilar. 2007. Uncontracted Negatives and Negative Constructions in
Contemporary English: A Corpus-Based Study. Santiago de Compostela:
Universidade de Santiago de Compostela [CD-ROM].
Ferguson, Charles. 1971. ‘Absence of Copula and the Notion of Simplicity: A Study of
Normal Speech, Baby Talk, Foreigner Talk and Pidgins.’ In: Del Hymes (ed.).
Pidginization and Creolization of Languages. Cambridge: Cambridge University
Press, pp. 141-150.
Gasparrini, Désirée. 2001. It isn’t, it is not or it’s not? Regional Differences in
Contraction in Spoken British English. Master’s thesis. University of Zürich.
Givón, Talmy. 1991. ‘Isomorphism in the Grammatical Code.’ Studies in Language 15,
85-114.
Greenbaum, Sidney. 1977. ‘Judgments of Syntactic Acceptability and Frequency.’
Studia Linguistica 31, 83-105.
Haiman, John. 1985. Natural Syntax. Cambridge: Cambridge University Press.
Hengeveld, Kees. 1992. Non-Verbal Predication. Berlin and New York: Mouton de
Gruyter.
Hiller, Ulrich. 1987. ‘She isn’t studying vs. She’s not studying.’ Die Neueren Sprachen
86, 531-553.
Huddleston, Rodney and Geoffrey K. Pullum. 2002. The Cambridge Grammar of the
English Language. Cambridge: Cambridge University Press.
Hughes, Arthur, Peter Trudgill, and Dominic Watt. 42005 [1979]. English Accents and
Dialects: An Introduction to Social and Regional Varieties of English in the
British Isles. London: Edward Arnold.
Hundt, Marianne, and Christian Mair. 1999. ‘‘Agile’ and ‘Uptight’ Genres: The Corpus-
Based Approach to Language Change in Progress.’ International Journal of
Corpus Linguistics 4, 221-242.
Jespersen, Otto. 1917. Negation in English and in Other Languages. Copenhagen: Ejnar
Munksgaard.
Kjellmer, Göran. 1998. ‘On Contraction in Modern English.’ Studia Neophilologica 69,
155-186.
Klima, Edward S. 1964. ‘Negation in English.’ In: Jerry A. Fodor and Jerrold J. Katz
(eds.). The Structure of Language. Englewood Hills: Prentice Hall, pp. 246-323.
Krug, Manfred. 1998. ‘String Frequency: A Cognitive Motivating Factor in
Coalescence, Language Processing, and Linguistic Change.’ Journal of English
Linguistics 26, 286-320.
Krug, Manfred. 2003. "Frequency as a Determinant in Grammatical Variation and
Change." In: Günther Rohdenbrug and Britta Mondorf (eds.). Determinants of
Grammatical Variation in English. Berlin and New York: Mouton de Gruyter,
pp. 7-68.
29
Labov, William. 1972. Language in the Inner City. Philadelphia: University of
Pennsylvania Press.
Leech, Geoffrey. 2003. ‘Modality on the Move: The English Modal Auxiliaries 1961-
1992.’ In: Roberta Facchinetti, Manfred Krug and Frank Palmer (eds.). Modality
in Contemporary English. Berlin: Mouton de Gruyter, pp. 223-240.
Leech, Geoffrey. 2004. ‘Recent Grammatical Change in English: Data, Description,
Theory.‘ In: Karin Aijmer and Bengt Altenberg (eds.). Advances in Corpus
Linguistics: Papers from the 23rd
International Conference on English Language
Research on Computerized Corpora (ICAME 23) Göteborg 22-26 May 2002.
Amsterdam: Rodopi, pp. 61-81.
Leech, Geoffrey and Nicholas Smith. 2006. ‘Recent Grammatical Change in Written
English 1961-1992: Some Preliminary Findings of a Comparison of American
with British English.’ In: Antoinette Renouf and Andrew Kehoe (eds.). The
Changing Face of Corpus Linguistics. Amsterdam and New York. Rodopi, pp.
185-204.
Leech, Geoffrey, Paul Rayson and Andrew Wilson. 2001. Word Frequencies in Written
and Spoken English. Harlow: Longman.
Leech, Geoffrey, Marianne Hundt, Christian Mair, and Nicholas Smith. 2009. Change
in Contemporary English. A Grammatical Study. Cambridge: Cambridge
University Press.
López-Couso, María José. 2007. ‘Auxiliary and Negative Cliticisation in Late Modern
English.’ In: Javier Pérez Guerra et al. (eds.) ‘Of Varying Language and
Opposing Creed.’ New Insights into Late Modern English. Bern: Peter Lang, pp.
301-323.
Mair, Christian. 2006. Twentieth-Century English. History, Variation, and
Standardization. Cambridge: Cambridge University Press.
Mair, Christian and Marianne Hundt. 1995. ‘Why is the Progressive Becoming More
Frequent in English? A Corpus-based Investigation of Language Change in
Progress.’ In: Wolfgang Riehle (ed.). Anglistentag 1994 Graz: Proceedings.
Tübingen: Niemeyer, pp. 247-254.
Mair, Christian and Marianne Hundt. 1997. ‘The Corpus-Based Approach to Language
Change in Progress.’ In: Uwe Böker and Hans Sauer (eds.). Anglistentag 1996
Dresden. Trier: Wissenschaflicher Verlag, pp. 71-82.
McElhinny, Bonnie. 1993a. ‘Copula and Auxiliary Contraction in the Speech of White
Americans.’ American Speech 68, 371-399.
McElhinny, Bonnie. 1993b. ‘The Interaction of Phonology, Syntax and Semantics in
Language Change: The History of Modal Contraction in English.’ Chicago
Linguistics Society 28, 367-381.
Nelson, Gerald, Sean Willis, and Bas Aarts. 2002. Exploring Natural Language:
Working with the British Component of the International Corpus of English.
Amsterdam and Philadelphia: John Benjamins.
Pustet, Regina. 2003. Copulas. Universals in the Categorization of the Lexicon. New
York: Oxford University Press.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik 1985. A
Comprehensive Grammar of the English Language. London: Longman
Rissanen, Matti. 1994. “The Position of Not in Early Modern English Questions.” In:
Dieter Kastovsky (ed.). Studies in Early Modern English. Berlin: Mouton de
Gruyter, pp. 339-348.
Rissanen, Matti. 1999. ‘Isn’t It? or Is It Not? On the Order of Postverbal Subject and
Negative Particle in the History of English.’ In: Ingrid Tieken-Boon van Ostade,
30
Gunnel Tottie and Wim van der Wurff (eds.). Negation in the History of English.
Berlin: Mouton de Gruyter, pp. 189-205.
Rohdenburg, Günter. 1996. ‘Cognitive Complexity and Increased Grammatical
Expliciteness in English.’ Cognitive Linguistics 7, 149-182.
Selkirk, Elizabeth O. 1981. The Phrase Phonology of English and French.
Bloomington, IN: Indiana University Linguistics Club.
Smith, Nicholas. 2003. ‘Changes in the Modals and Semi-modals of Strong Obligation
and Epistemic Necessity in Recent British English.’ In: Pam Peters, Peter
Collins, and Adam Smith (eds.). New Frontiers of Corpus Research.
Amsterdam: Rodopi, pp. 317-330.
Stassen, Leo. 1997. Intransitive Predication. Oxford: Clarendon Press.
Svartvik, Jan (ed.). 1990. The London-Lund Corpus of Spoken English: Description and
Research. (Lund Studies in English, Volume 82). Lund: C. W. K. Gleerup.
Swan, Michael. 32005 [1980]. Practical English Usage. Oxford: Oxford University
Press.
Sweet, Henry. 1900. A New English Grammar Logical and Historical. Part I:
Introduction, Phonology and Accidence. Oxford: Clarendon Press.
Tagliamonte, Sali, and Jennifer Smith. 2002. ‘‘Either it isn’t or it’s not’ NEG/AUX
Contraction in British Dialects.’ English World-Wide 23, 251-281.
Tottie, Gunnel. 1991. Negation in English Speech and Writing. San Diego: Academic
Press.
Trudgill, Peter (ed.). 1978. Sociolinguistic Patterns in British English. London: Edward
Arnold.
Trudgill, Peter. 1984. Language in the British Isles. Cambridge: Cambridge University
Press.
Trudgill, Peter and Jean Hannah. 31994 [1982]. International English. A Guide to
Varieties of Standard English. London: Arnold.
Warner, Anthony R. 1993. English Auxiliaries. Cambridge: Cambridge University
Press.
Yaeger-Dror, Malcah. 1997. ‘Contraction of Negatives as Evidence of Variance in
Register-specific Interactive Rules.’ Language Variation and Change 9, 1-36.
Yaeger-Dror, Malcah, Lew Hall-Lew, and Sharon Deckert. 2002. ‘It’s not or isn’t it?
Using Large Corpora to Determine the Influences on Contraction Strategies.’
Language Variation and Change 14, 79-118.
Zipf, George. 1935. The Psychobiology of Language: An Introduction to Dynamic
Philology. Cambridge, Mass.: MIT Press.