Grammar and the Spoken Language1

RONALD CARTER and MICHAEL McCARTHY

University of Nottingham

In this paper, we argue that language teaching which aims to foster speaking
skills and natural spoken interaction should be based upon the grammar of

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

spoken language, and not on grammars which mainly reflect written norms
Using evidence from a mini-corpus of conversational English, we look at how
four grammatical features which occur with noticeable frequency in the corpus
are dealt with in currently popular pedagogical grammars Our investigation
shows that treatment of the selected features varies from adequate to patchy to
complete absence from the grammars surveyed We conclude that research in
discourse analysis does offer some helpful insights into the usage of these
features, but that, in the absence of easy access to discourse analysis work and
given the mixed treatment in grammar books, teachers and learners will often
be thrown back on their own resources However, we argue that even verysmall
amounts of real spoken data can yield significant evidence which can be used
imaginatively within inductive and language awareness approaches in and out
of the classroom to increase awareness and knowledge of the grammar of
conversation

1 INTRODUCTION
In the last couple of decades, the emphasis of English language teaching has
shifted from a general notion of proficiency or competence towards skills-based
approaches, and the speaking skill has been foregrounded in a way that was not
so some twenty-five to thirty years ago In parallel, the communicative approach
has provided language teachers with innovative techniques and materials for
fostering the spoken skills At the same time, language courses and works of
reference are increasingly advertising themselves as offering 'real English1 and
'real-life communication' to the learner 2 Equally significant and relevant is the
debate over model(s) of English selected for teaching the speaking skills,
whether it be southern, middle-class British English, American English, one of
the 'new English1 varieties, or else a more neutral, culturally unattached variety
of the language for intercultural communication that may not involve the use of
English as a native or first language (see, for example, Bex (1993) on English in
Europe) But it may be argued, as we shall argue in this paper, that the models of
grammar which underpin most of the laudable attempts at representing and
activating the use of the spoken language are still rooted in descriptions of the
grammar of written English and have failed to take on board some interesting
features of the grammar of informal, interactive talk Just as it would be
questionable to base a writing skills course on grammatical statements based
only on informal spoken data, in our opinion it is equally the case that spoken

Applied Linguistics, Vol 16, No 2 O Oxford University Press 1995

142 GRAMMAR AND THE SPOKEN LANGUAGE

language instruction based solely or mainly on written language description is

an unsound methodological foundation upon which to build The problematic
base from which we launch our argument includes ideological confusion too so
often, in our experience, attempts to discuss the grammar of speech in teacher-
training seminars and similar contexts are clouded by prejudgements that many
of the grammatical features observable in everyday, unplanned conversation
are simply "wrong1, and are corruptions of and lapses from the standards
enshrined in the scholarly grammars (this may be particularly so in the
perception of the public at large, see Pinker 1994)

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

What we shall argue in this paper is that wntten-based grammars exclude
features thai occur widely in the conversation of native speakers of English,
across speakers of different ages, sexes, dialect groups, and social classes, and
with a frequency and distribution that simply cannot be dismissed as aberration
If our speakers are 'wrong', then most of us spend a lot of our time being 'wrong'
Such a view of grammar is absurd and needs to be put aside from the outset In
this paper, we shall not make particular reference to course books, but suffice it
to say that the available evidence points to a lack of fit between what conversa-
tional data tell us and what textbook dialogues suggest, and not just for Bntish
English (e g see Pica (1983) on American English, O'Connor di Vito (1991) on
French data and textbooks, and Glisan and Drescher (1993) on a similar lack of
fit between real data and Spanish textbooks)
Nowadays, computational analysis of large amounts of natural language data
is relatively easy, and pioneering projects such as COBUILD at the University
of Birmingham have revolutionized the way we look at naturally-occurring
language And yet even the COBUILD Grammar (Collins COBUILD 1990,
hereafter COB), along with other recent grammars, seems to underplay or omit
recurring features of everyday spoken English In some cases, a lack of
descriptive coverage may be due to a bias towards broadcast talk, where
language is usually more measured, self-conscious, often pre-planned and more
formal, and more likely to emulate written standards because of traditional
perceptions of institutions such as the British Broadcasting Corporation (BBC)
(see the views of Summers (1994), representing rival publishers Longman, on
the question of bias towards broadcasting in relation to COBUILD in the EFL
Gazette article 'Corpuses', August 1994 18) In the case of the COBUILD
project, great stndes have been made in incorporating more and more spoken
data of vanous kinds into its huge corpus, and others have not been slow to
follow its lead m building major corpora which are either devoted to, or include,
spoken material The Bntish National Corpus, now under development,
promises scholars access to millions of words of spoken data (see Crowdy 1993
for details) and its published products, it is to be hoped, will have the
opportunity to reflect the nature of'real' spoken language in a way that has not
hitherto been possible In the meantime, though, language practitioners do not
have to sit back and wait for the results of such vast investigations for, with
relatively small but carefully targeted corpora, much can be learnt about the
spoken language, and small corpora can. in themselves, be directly exploited as
 RONALD CARTER AND MICHAEL McCARTHY 143

a valuable resource in teaching The fact that lexis is essentially a huge number of
items whose occurrence, except for the most common words, is relatively
infrequent means that convincing corpora for major lexicographical work need
to be vast, perhaps tens of millions of words Grammar, on the other hand,
consists of a small number of items and patterns frequently repeated, and thus
much smaller corpora can yield regularly patterned data for grammatical
analysis and exploitation in grammar teaching Put another way, even in a very
large corpus of millions of words, many idiomatic lexical expressions will only
occur once or a couple of times (e g Arnaud and Savignon (forthcoming) cite

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

Rosamund Moon's evidence of only one occurrence of the expression "it's all in
a day's work' in the 18-million-word Oxford Hector Pilot Corpus of British
English) However, the inclusion of such instinctively useful expressions in
dictionaries of idioms would pass largely without controversy On the other
hand, even in a very small (by current standards) corpus of tens of thousands of
words of conversation, one can easily find grammatical patterns not described
in existing grammars occurring five, ten, or more times across different speakers
and in different contexts Furthermore, huge corpora often throw up vastly
unmanageable numbers of occurrences of common grammatical features (e g
the articles), resulting in compilers of dictionaries and grammars having to work
with a sub-sample anyway And it should never be forgotten that even the very
big corpora currently in use are themselves only small samples of the spoken
and wntten output of a speech community over the course of many years Each
average individual processes a phenomenal amount of language every day If we
were only exposed to three hours of spoken language per day (many people
experience this much and more from radio and television alone, not to mention
their exposure to live conversation and to written language), the Nottingham
corpus (see below) average figure of 100 hours for one million words of
conversation would suggest that an individual might well process the 12 million
"spoken sample' words claimed by the Cobuild 'Bank of English' (see Fox,
quoted in 'Corpuses', EFL Gazette August 1994 18) in just over a year of their
adult life (many might achieve this in an even shorter time) Naturally, small
corpora cannot be the basis of sweeping generalizations about the spoken
language, but they can be used as the basis of further research or day-to-day
observations which can confirm or refute their limited evidence Above all,
small corpora can be collected by teachers and learners and used as a class
resource where existing published resources fail to supply data or descriptions
of features of interest in the spoken language
This paper bases its arguments on data taken from a mini-corpus of conversa-
tional English which forms part of a larger spoken corpus currently under
development at the University of Nottingham For the present purposes, we
have taken 35 extracts of conversation averaging four minutes each,3 giving us a
total of approximately two-and-a-half hours of transcribed talk and 30,000
words In this mini-corpus, 45 different speakers are represented, 22 males and
23 females, ranging in age from 11 years to 78 years The speakers come from
across a broad social spectrum covenng unskilled working-class on one side to
144 GRAMMAR AND THE SPOKEN LANGUAGE

professional people on the other They come from widely distributed areas of
the British Isles The 35 extracts are classified by genre (see McCarthy and
Carter 1994 24-38) We wish to concentrate principally on features of casual
conversation as the core genre for any conversational analysis relevant to the
teaching of general spoken English, with a secondary interest in narrative, and
so the mmi-corpus is weighted accordingly, with other genres represented by
smaller amounts of talk For more reliable cross-generic comparisons, statistical
normalization procedures such as that outlined by Biber (1988 75-8) can be
brought into play, though in the first instance we are interested in gross

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

frequency across all genres The classification and amounts of data are as
follows
A Casual conversation (bi- and multi-party informal talk with frequent turn-
changes, no pre-set topics, equally distnbuted or shifting conversational roles),
14 extracts, approximately 16,000 words, 35 of the total speakers represented
B Narratives (stretches of talk amenable to narrative-structure analysis
according to accepted models of narrative, for example Labov 1972) 11
extracts, approximately 6,000 words, 15 of the total speakers represented
C Service encounters (talk between server and served parties in shops,
restaurants, etc, in the negotiation of goods and services), 5 extracts,
approximately 4,000 words, 9 of the total speakers represented
D Language-in-action (talk which accompanies some real-world task, for
example, talk while preparing food, moving furniture, etc), 5 extracts,
approximately 4,000 words, 17 of the total speakers represented

The genre-classification is important, since the grammatical features we shall

look at are not equally distnbuted across the genres, and we shall argue for a
more genre-sensitive description of the spoken language as being the most
useful resource for teachers and learners of English (see Biber, Conrad, and
Reppen (1994) for a similar argument, though they refer to differences across
registers, see also Rings (1992)) Our choice of the non-casual genres is
influenced by our long expenence as English language teachers and the kinds of
genres most often found in textbook dialogues It is, unsurpnsingly, not always a
simple, straightforward matter to assign extracts to genres For example,
narratives often occur embedded in longer casual stretches, and defining what a
minimal narrative may consist of is also occasionally problematic, but overall
labelling is generally feasible and has a pedagogical usefulness that transcends
any unresolved local difficulties
Our purpose in analysing the data was to see how often features of grammar
occurred which, intuitively in the first instance, could be labelled as unlikely to
occur in formal, written text, but which, none the less, did not seem 'wrong' or lll-
formed, but which occurred repeatedly in the mini-corpus We have been able
to use large numbers of our own students and other seminar participants over
the past year as informants to support our intuitions as to which features of the
 RONALD CARTER AND MICHAEL McCARTHY 145

texts corresponded to those criteria When at least five occurrences of any such
features were recorded in the data, and provided the features in question
occurred across different speakers independently of sex, age, dialect group, and
social class, and across different extracts, searches were made in current
descriptive and pedagogical reference grammars to see what coverage, if any,
the features received4 Our interest was in forms and structures which, although
unlikely to occur in the formal written mode, seem to pass as perfectly
acceptable and grammatical in the informal spoken mode, and on which any

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

observant language learner or teacher might well expect to find information in
available published resources We did not count aborted structures or
anacolutha (changing from one structure to another in mid-stream, before the
first construction is complete) Some work on spoken grammatical features to
date (eg Brown 1980, Lewandowska-Tomaszcyk 1990) has concentrated on
starkly obvious anomalies such as anacolutha and pronoun-copying in relative
clauses (e g That's the record that I lent it to her weeks ago') These features are
undoubtedly of interest, since they are by no means isolated phenomena, but
there are also other, less immediately striking features which do not seem so
openly to violate traditional norms and yet which still slip through the net of
grammatical descnption or are relegated to positions of minor importance in
the descnptive and pedagogic apparatus, as we shall attempt to show5
In the following section, we shall take just four grammatical features, and
report findings which we believe to be relevant for dealing with grammar in the
spoken mode in the language teaching and learning context These features do
not by any means exhaust our list of areas offering great potential for new and
revised description and pedagogy, and more findings will be published as our
project develops

2 EVIDENCE FROM A SELECTION OF GRAMMA TICAL FEATURES

2 1 Ellipsis
Ellipsis, the omission of elements otherwise considered required in a structure,
occurs widely in the mini-corpus Here we shall concentrate on just one kind of
ellipsis, what Quirk et al (1985 895ff) (hereafter QUIRK) refer to as
situational ellipsis Situational ellipsis differs from textual and structural ellipsis
in that the unrealized items of the conventional account of structure are
retnevable from the immediate situation Textual ellipsis is characterized by
retnevability from the text itself (either anaphoncally or cataphoncally), while
structural ellipsis occurs when a purely structural element is omitted (e g 'I'm
surprised [that] no-one told you ') Situational ellipsis is particularly apparent in
casual data It is also notably present in language-in-action data, where not only
the participants but the objects and entities and processes talked about are
typically prominent in the immediate environment There is also situational
ellipsis in our service-encounter examples It is notably absent from the
narrative data, where the participants and processes of the story are usually
separated in time and place from the moment of telling
146 GRAMMAR AND THE SPOKEN LANGUAGE

There are over 80 places in the mini-corpus where one or more items of
structure which would be expected in the formal written m o d e d o not appear,
but whose referents are retrievable from the immediate situation Of these
places, 65 are ellipses where subject p r o n o u n s are retrievable from the
contextual environment In 41 of these 65 cases, a copular or auxiliary verb is
also absent where it might be expected in written text Extract (1), from a
language-m-action segment, gives the typical flavour of these types of ellipses
(1) [members of a family in their kitchen cooking nee in preparation for a party)

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

1 (01) Foreign body in there
2 (02) It's the raisins
3 (01) Oh is it oh it's nee with raisins is n
4 (02) L No no no it's not supposed to be
5 [laughs| erm
6 (01) There must be a raisin for u being in there
7 (03) D'you want a biscuit
8 (01) Erm
9 (03) Biscuit
10 (01) Eryeah
11 |9secs]
12 (03) All right
13 (01) Yeah
14 [10 sees)
15 (03) Didn't know you used boiling water
16 (02) Pardon
17 (03) Didn't know you used boiling water
18 (02) Don't have to but its erm they reckon it s erm
19 quicker

Here we may observe ellipsis of the subject and verb phrase and the indefinite
article in line 1, possibly either 'there's a o r you've got a foreign body in there",
though these two interpretations by no means exhaust the possibilities ( e g / can
see a is an equally possible candidate for the understood part of the message) In
line 9, biscuit might minimally expect to attract a, while in lines 15,17, and 18,
the pronouns / and you are not realized These types of pronoun and/or
pronoun + operator absences are well desenbed in existing grammars (eg see
QUIRK ibid), although the tendency is to explain them as elements of
informality QUIRK simply states that sttuational ellipsis is 'restricted to
familiar (generally spoken) English' (ibid 896) While it may be true that such
ellipses do not occur in highly formal contexts, it is also true that the wholly
informal and, by any account, 'familiar' narratives in our mini-corpus do not
have them either, and so the formality/informality or familiarity distinction is
anything but the whole story We would argue that genre and context are the two
key factors that mediate beyond the choice of formality/familiarity The
narrative genre, with its spatio-temporal displacement, no matter how informal
or familiar, cannot easily retrieve its elements from the immediate context and
thus spells out explicitly the participants and verbal operations which may be
 RONALD CARTER AND MICHAEL McCARTHY 147

assumed to be retrievable from the environment in other, more situation-

dependent forms of talk, such as tanguage-in-action, informal service encoun-
ters, and casual talk, where participants are face-to-face Extract (2), from a
service encounter, illustrates such retnevabihty
(2) [At a dry-cleaner's (02) is leaving a pair of trousers for cleaning|
(01) Wednesday at four be okay
(02) Er yeah that's fine just check the pockets a minute
Initial will or would from (Ol)'s turn and /'// from (02)'s second clause are not

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

realized Our conclusion is that a proper description of a feature such as
situational ellipsis, and any descnption claiming pedagogical usefulness, should
be able to state those environments in which the types of ellipsis described tend
to and tend not to occur, as well as stating the structural restnctions on what
elements in the clause may and may not be elhpted
Even on the purely structural questions of what is permissible or not in
situational ellipsis, existing grammars fail to take into account some interesting
features of correlation between grammar and lexis In the mini-corpus, it is
noticeable that, on many occasions, items are elhpted from what are often
termed lexical phrases (Nattmger and DeCarnco 1992), institutionalized
expressions (Lewis 1993 94), or fixed expressions (identified by the normal
criteria of lack of syntactic or lexical productivity, see McCarthy 1992) QUIRK
(899) does permit this for the definite and indefinite article with fixed and
idiomatic expressions, but our data show ellipsis of other items too in the
environment of fixed expressions Nattinger and DeCarnco (ibid) argue that
such expressions are fundamental in the construction of text, and, indeed, it is
their very fixedness and cultural commonality which makes them good
candidates for ellipsis, the unspoken elements of the message can always be
assumed to be known Some examples follow, with relevant fixed expressions
manifesting ellipsis in italic

(3)
(01) We did quite well out of it actually
(02) Great
(03) Mm saved a fortune [ellipted we\
(4)
(01) How many pillows do you have, one two
(02) I have one Graeme won't have any
(01) Yeah right okay good job you said [elhpted it's/u was a\
(5)
(01) It's lovely
(02) Good winter wine that
(01) A terrific one
(02) Put hairs on your chest that one [elhpted it'll]
Further, casual observation in the field suggests this feature is quite common,
and future research will devote itself to investigating just how widespread it is as
148 GRAMMAR AND THE SPOKEN LANGUAGE

the corpus develops Once again, we would argue that any description of ellipsis
in spoken language is incomplete which does not investigate fully and render an
account of this essentially culturally embedded feature of lexico-grammatical
form
Ellipsis is a good starting point for our examination of grammatical features,
for it is a feature which is described in existing published resources, but often
with inadequate attention to precisely those features that leap out of natural
spoken data Swan (1980) (hereafter SWAN6) gives very good coverage of most
of the types of situational ellipsis we find in our data but makes no reference to

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

fixed expressions or to the uneven distribution over genres, prefernng formality
as the determining factor and permitting elhpsis of words 'when the meaning can
be understood without them' (SWAN section 196, a similar quote may be
found in the new edition, section 183) Alexander (1988) (hereafter ALEX) has
no separate section devoted to ellipsis, textual ellipsis is referred to with
examples as and when they arise On the kinds of everyday situational ellipsis we
have referred to, ALEX is singularly unmformative COB (399-400)
concentrates on the elliptical features of questions, short replies, and agreement
sequences, and makes no reference to much of what we have observed in our
data Downing and Locke's grammar (1992 242) (hereafter DOWL) has only
an extremely short and cursory subsection on situational ellipsis, but at least
recognizes the 'institutionalized1 nature of ellipsis in many everyday expres-
sions What we have, therefore, with regard to situational ellipsis in these major
popular reference resources is an uneven and patchy treatment, even though the
feature is extremely common in conversational data in the genres we have
referred to
Often in such cases it is worthwhile turning to discourse-analytical research
papers and relevant books, and we shall argue for other features below that this
is sometimes the best course of action for teachers and materials writers in
pursuit of more accurate description In the case of ellipsis in conversation, the
two most notable recent papers are Thomas (1987) and Ricento (1987)
Thomas covers auxiliary contrasting (e g "Has she arrived7' - "She should have
by now'), and Ricento concentrates on clausal ellipsis which produces the kind
of verbless short replies exemplified in COB No paper or book we know of
covers adequately the kinds of features we have found to be frequent and
significant in our data

2 2 Left dislocation and topical information

Left dislocation is one of the names commonly used for the phenomenon where
items semantically co-referential with the subject or object of the clause are
positioned before the subject An example from our data is
(6)
(01) Well Sharon, where I'm living, afriendof mine, she's got her railcard, and
Here, a friend of mine is 'copied' in the subject pronoun she There are seven
clear examples in the mini-corpus (uttered by seven different speakers in seven
 RONALD CARTER AND MICHAEL MCCARTHY 149

different extracts) where the pronoun subject of the clause is co-referential with
an initial noun phrase uttered before the main clause gets under way, as in
example (6), and again here in example (7)

(7)
(01) The one chap in Covenl Garden who I bought the fountain pen off he was saying
that he'd
where he copies the whole of the preceding long noun phrase Examples (6)
and (7) fit in with Geluykens' (1992) model for left dislocation, where a friend of

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

mine in (6) would be termed the 'referent', she would be termed the 'gap", and 's
got her railcard the 'proposition' But other comparable phenomena, labelled
'quasi-left dislocation' by Geluykens (ibid 131) are also apparent in our data,
and ment a closer look For example, the initial noun phrase may be only
indirectly related to the subject, which need not be a pronoun copy, as in (8)

(8)
(01) Thisfriendof mine, her son was in hospital and he'd had a serious accident
Or there may be discord of person and number between the front-placed noun
phrase and the main-clause subject, as in (9), but which does not seem to
hamper pragmatic decoding by the hstener(s)
(9)
(01) Thai couple thai we know in Portsmouth, I don't hearof/ier for months then
Alternatively, the front-placed item may be grammatically indeterminate and its
relationship with the main-clause subject only topically/pragmatically coherent
For instance, in (10) below, is the italicized portion a non-finite clause, or a
noun-phrase (a colloquial version of 'your saying')9 The precise grammatical
status seems irresolvable, and yet its pragmatic link with the subject ('one of
dad's many stones of how he escaped death during his long life') is apparently
clear to the participants and unproblematic
(10)
(01) You saying about that chap with the newspaper, that, one of dad's many stones of
how he escaped death [laughs] during his long life was
The point about these canonical examples of left dislocation and the less-clear
but related examples is that they have in common the utilization of an available
'slot' before the core constituents of the clause (Subject, Verb, Object/
Complement, Adjunct, in whatever order they occur) are realized Indeed, it
would seem to be a misnomer and a misleading metaphor to talk of dislocation,
for it suggests that something has been pushed out of place to a somewhat
aberrant position This metaphor may be an unfortunate legacy of a Chomskyan
view of syntax 7 We would argue that left-placed or fronted items of this kind are
perfectly normal in conversational language, and are quite within their 'right
place' The phenomenon occurs especially in the narrative genre, where ten of
our total of fourteen examples of this type of feature occur It is apparent that
150 GRAMMAR AND THE SPOKEN LANGUAGE

speakers use the available slot to flag a variety of items of information that will
be helpful to the listener in identifying participants, in linking current topics to
already mentioned ones, in reactivating old topics, and generally anchoring the
discourse, offering what Quirk et al (1985) call 'a convenience to hearer'
(QUIRK 1417) This is a quintessential example of'grammar as choice', where
the speaker chooses to fill an available slot for textual and interpersonal
motives The grammatical indeterminacy of what may fill the available slot,
which we shall term the topic because of its proclivity to carry topic-prominent
items, actually means that it is quite easy for language learners to manipulate—

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

and experienced teachers will here recognize a common feature of learner
discourse manifested in examples such as 'My father, he has two brothers and
one sister*, which are .sometimes lamented as examples of aberrant structure
The problem may well be better pul down to lack of pragmatic motivation for
using the structure rather than to any inherent ungrammatically
QUIRK handles the topic slot under a wider phenomenon termed reinforce-
ment, which includes what we discuss below in section 2 3, and describes it as a
•feature of colloquial style used for 'purposes of emphasis, focus or thematic
arrangement (QUIRK 1416) COB seems to ignore the phenomenon
altogether, and no heading anywhere would seem to point the student in the
direction of this frequent feature of 'real' English (our claim to its frequency and
normality is here doubly reinforced by Geluykens' (ibid) data from a much
larger corpus) SWAN (1980 edition) offers no guidance in connection with the
topic slot except for one isolated example sentence with a topic-slot object and a
copied pronoun in the main clause, under the heading emphasis (section
201 3 a) The new edition of SWAN improves on this with several examples of
what Swan calls "detached fronted subjects and objects' (section 217 3), but with
no guidance as to the function of fronting other than stating that it is common
We can find no examples of the topic-slot phenomenon in ALEX and only the
briefest of mentions in DOWL, which, using Halhdayan terminology, refers to
'preposed themes' (234) Once again, the picture is inadequate and terribly
patchy We would argue that ouPdata, which points to genre-correlations,
suggests that more detailed descriptions of this phenomenon are needed, and, in
the case of narrative, which is where the majority of instances occur in our mini-
corpus, the topic-slot needs to be related to basic narrative functions such as
abstracts and orientations, to use Labov's (1972) terms

2 3 Reinforcement the tail slot

Just as there seems to be an available slot at the front of the clause, which we
have called the topic, so too, when all the core clause constituents have been
exhausted, is there a final available space which speakers often choose to fill
with different types of information Tags occupy this slot A typical example is
the reinforcement tag (e g 'You're stupid, you are *), which is well-covered in the
sources we have surveyed But also recurrent in the mini-corpus are amphfica-
tory noun-phrases, the reverse, as it were, of the topic-slot noun phrase and a
subsequent copying pronoun Some examples follow
 RONALD CART ER AND MICHAEL McCARTHY 151

<01) Its lovely

{02) Good winter wine that
(12)
(01) It's very nice that road up through Skipton to the Dales [(02) Yeah] I can't
remember the names of the places
(13)
(01) And he's quite a comic the fellow, you know

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

(02) Llshe Lyeah
(14) [(01) is the host, (02) a dinner guest)
(01) Look get started you know putting all the bits and pieces on
(02) L Bits and pieces on
(01) Cos otherwise they tend to go cold don t they pasta
Eleven such nominal post-clause items occur in the mini-corpus, uttered by
eight different speakers across seven different extracts The speakers come from
Wales, Northern England, and Southern England, suggesting it is not confined
to one dialect Example (14) demonstrates that the nominal items are not
mutually exclusive with tags It will also be noted that examples (11) to (14) are
all in contexts where evaluative statements are being made So too are the other
seven examples not quoted here This fits in with Aijmer's (1989) study of tails,
as she calls the phenomenon, and as we shall do hereafter Tails, in Aijmer's data
and in ours, tend to occur with phatic, interpersonal functions, usually in
contexts of attitudes and evaluations
Tails are therefore an important part of what may be called interpersonal
grammar, that is to say speaker choices which signal the relationships between
participants and position the speaker in terms of his/her stance or attitude Tails
may therefore be compared with topics, and a model for the clause in
conversation may be posited which offers the structural potential shown in
Table 1
The topic and the tail are optional, but when utilized, carry important inter-
personal functions, as our and others' data show With this model, elements
commonly found in conversational data need not be scattered around the
descriptive apparatus or relegated to minor or marginal sections of the
grammar, but achieve their proper place as items of 'real' language
Tails do get treatment in conventional grammars QUIRK (1417) calls them
amphficatory tags SWAN (section 524 3, section 472 3) gives them scant
attention and puts them under "subject tags', which is a sub-heading of
'reinforcement tags" In ALEX (260), they receive equally short shrift and are
called 'reinforcement tags' We have been unable to find any treatment of them
in COB, and if there is any, it is well hidden DOWL (234) brings topics and tails
together under the headings 'preposed and postposed themes' and gives them
only a passing glance, offering no information other than that, as a form of
repetition, they 'allow speakers to process their information as they go along' It
is not at all clear what such a cryptic statement might mean
152 GRAMMAR AND THE SPOKEN LANGUAGE

Table 1 Structural potential of the clause in conversation

Sequence pre-clause clause post-clause

Function TOPIC CORE TAIL
Constituents S-tp/O-tp/RI S/V/O/C/A TAG/S-tl/O-tl
Forms NP/NFC NP/VP/ADVP NP/VP

S-tp and O-tp ~ subject or object in the topic slot S-il and O-tl ™ subject or object in
the tail slot, RI — related item (as in example (8) above), and TAG — any of the types of

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

English sentence-tags (see Bennett 1989) S/V/O/C/A represent the core clause
constituents (subject, verb object complement, and adjunct), which may be re-ordered
for a variety of thematic and focusing purposes (eg object-fronting, see Hietaranta
1984) NP = noun phrase NFC = non-finite clause, as in example (10) above, VP =•
verb phrase and ADVP = adverbial phrase

2 4 Indirect speech
Indirect speech is an area which, on the face of it, seems thoroughly covered in
the grammars we have surveyed Hundreds of examples abound of the type 'X
said that Y\ where the reporting verb (typically say or tell) is in the simple past
tense and the sequence of tense rules apply to the reported clause And yet in
our mini-corpus, eleven examples of indirect speech (uttered by eight different
speakers in eight different extracts) have the reporting verb in past continuous
Some examples follow
(15)
(01) 1 mean 1 was saying to mum earlier that I'm actually thinking not for the money
but for the sort of fun of it really trying to get a bar or a waitressing job I was saying to
you wasn't 11(02) Yeah) in the summer well over Christmas or Easter
(16)
(01) Tony was saying they should have the heating on by about Wednesday
(17)
(01) Where were, yeah because I was saying to Ken that you wouldn t be in a pub at
twelve o clock in Corby would you you would have to be in somebody's house
(18)
(01) Yes Maureen and Derek were telling me you have to get a taxi
Example (18) above, it will be noted, has tell in past continuous In addition to
the eleven examples in the mini-corpus, we have collected a further eleven
instances of indirect reports with past continuous say or tell by eleven different
speakers on separate occasions during casual listening over the last three
months, thus revealing the value of a mini-corpus as a stimulus for further tar-
geted observation in the field None of the illustrative sentences in any of the
grammar books surveyed contain a single example of this feature All examples
are with say and tell (and other reporting verbs) in simple past tense The same is
true of research articles on indirect speech Most papers are concerned with the
problem of backshift in the sequencing of tenses (e g present becomes past)
 RONALD CARTER AND MICHAEL McCARTHY 153

when direct speech becomes indirect, for example, Coulmas (1985), Comne
(1986), Goodell (1987), Huddleston (1989), Harman (1990) A good survey of
debates among linguists concerning direct and indirect speech may be found in
Baynham (1991) Yet despite the shift Baynham notes towards looking at
speech reporting within discourse pragmatics, even those investigations using
real data fail to pick up on the use of past continuous in indirect speech
reporting verbs, for example, Philips (1985), Tannen (1986), Wald (1987)
Notable too is Yule # a/'s (1992) paper, which sets out to take to task previous
work on speech reporting and which shows with real spoken and written data a

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

wide range of ways of reporting speech, but which, none the less, fails to note the
past continuous phenomenon for indirect speech Perhaps even more
remarkable is the absence of any reference to past continuous with say and tell
in the recent COBUILD Reporting Guide (Thompson 1994), which is corpus-
based with hundreds of real examples of indirect speech, and which is
thoroughly detailed in every other respect
The language teacher and the interested and observant learner are thus
thrown back on their own intuition in understanding examples (15) to (18) We
note that say and, in example (18), tell may be used in this way, that the report
may be of what the speaker him/herself said or what another said, and that past
simple could have been used, but with a distinctly different force the past simple
seems to give more authority to the actual words uttered, while the past
continuous seems more to report the event of the uttenng (exploiting perhaps
the basic foregrounding and backgrounding functions of past simple and past
continuous, respectively) Once again, though, what is interesting is that no
examples of indirect speech in the narrative extracts have past continuous We
checked ten other narrative extracts not in the mini-corpus and none of them
have any occurrences either All our examples occur in the casual genre, except
one, which crops up in the middle of a language-in-action sequence All the
indirect speech reports in our narratives are introduced by reporting verbs in
past simple or so-called historical present (see Johnstone 1987 on this) This
may hint at a truly reporting function for past simple reporting verbs, where
what someone said and what words they used (although creatively recon-
structed) are considered important On the other hand, past continuous say and
tell show a tendency to emphasize message content rather than form, and to
report or summarize whole conversational episodes rather than individual
utterances At the moment, our conclusions must be tentative until much more
data can be analysed As a feature of spoken English grammar, it seems, past
continuous reporting verbs have slipped through the grammarians' net We
would argue that this is precisely because (a) most notions of reported speech
are taken from wntten data (including literary text), and (b) that where spoken
data have been consulted, there has been an over-emphasis on oral narrative,
which is simply unlikely to yield such examples if the present corpus is anything
to go by, and an over-emphasis on broadcast talk, where indirect speech is likely
to be of a truly quotational type, where form and content are both important
(e g news reporting)
154 GRAMMAR AND THE SPOKEN LANGUAGE

3 CONCLUSIONS IN I HE PEDAGOGICAL CONTEXT

The last points in section 2 4 concerning indirect speech underscore some
central pedagogical considerations with regard to spoken English grammar
Principally, we would argue that descriptions that rest on the written mode or on
restricted genres and registers of spoken language are likely to omit many
common features of everyday informal grammar and usage, or at best relegate
them to marginal places in the description, regardless of corpus size But other
considerations arise too, which we summanze as follows

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

1 There is no common, agreed metalanguage for talking about the features of
spoken grammar that we have focused on in this article Our surveys of current
grammars reveal a bewildering and often user-unfriendly variety of terms used
to descnbe the features illustrated This makes it difficult to find and compare
information, even where it exists
2 Research-minded teachers and learners can get a lot of information on
spoken grammar from articles and books within the area of discourse analysis,
but these sources are sometimes obscure and difficult of access
3 Real spoken data is hard to come by for many teachers and learners, and the
concocted dialogues of course books often suffer from grammatical artificiality
and do not feature the kinds of grammatical items we have looked at Without
either improved descriptions of the spoken grammar or materials which reflect
it, grammar teaching within the communicative approach will always be out of
kilter with the typical kinds of informal interaction teachers hope to foster and
which learners frequently hope to be able to enter into in a natural way
4 Even very small corpora, if properly targeted and approached with an open
mind and careful observation, can yield recurring patterns of grammar that are
not fully (or not at all in some cases) described in conventional grammars Size is
no guarantee that the corpus will yield up its secrets if it is badly targeted or not
subject to detailed searching (see Owen (1993) for a relevant critique of corpus-
based grammar) But even small corpora may be difficult to collect in many
situations, and publishers should be encouraged to publish user-friendly
corpora and more real spoken language extracts as resources for teachers and
learners
5 Corpora should be carefully chosen to reflect the model of English the
learners may want or need to acquire There is little point in agonizing over
interactive features of informal spoken British English grammar if such features
simply do not occur in the target vanety This agument goes alongside our view
that there is equally little point in basing grammar teaching exclusively on
written models if the goal is to encourage speaking skills
6 In the present situation, the best course of action would seem to be to expose
learners to natural spoken data wherever possible and to help them to become
observers of the grammar of talk in its natural contexts and in different genres.
 RONALD CARTER AND MICHAEL McCARTHY ] 55
8
as advocated by Riggenbach (1990) Ellis has claimed that learners need to
learn how to observe language through tasks which foster comprehension as
much as production, and an over-emphasis on production in grammar may not
necessarily be the best approach in relation to the features we have described
7 The kinds of features we have looked at may not be entirely suitable for the
traditional 'three Ps' (Presentation-Practice-Production) mode of teaching,
and approaches based on observation, awareness, and induction may prove to
be a more satisfactory way of dealing with the interpersonal subtleties of choice

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

in matters such as ellipsis and topicahzing Our mnemonic would be the "three Is*
(Illustration-Interaction-Induction) where illustration stands for looking at
real data—which may be the only option since the grammar books and current
materials so often fall short, interaction stands for discussion, sharing of
opinions and observations, and induction stands for making one's own, or the
learning group's, rule for a particular feature, a rule which will be refined and
honed as more and more data is encountered On this last point, the patchy,
confusing, and often inadequate treatment of the grammar of spoken language
in published resources may turn out to be a cue for imaginative discovery and
problem-solving work in the grammar class One only needs an initial cunosity,
some real data, and the feeling that there is a lot to be discovered to get started
(Revised version received August 1994)

This paper was reviewed and accepted by the previous editorial team, before the
appointment of Michael McCarthy as joint editor of Applied Linguistics

NOTES
1
This article is a version of a paper delivered by the authors at the Second MATSDA
Conference at the University of Luton, UK in January 1994, and at TESOL, Baltimore,
USA, in March 1994 Further versions were presented at TESOL Greece, Athens
March 1994, IATEFL Argentina, Buenos Aires, July 1994, and BRAZTESOL, Riberao
Preto, Brazil, July 1994 The authors are grateful for the many comments offered by
participants at those conferences which have helped to shape and refine the present
article
2
'Helping learners with real English' is the slogan used to promote the Collins
COBUILD reference books and materials "Real-life communication' has been used
recently in publicity literature lo promote the Look Ahead multi-level English course,
published by Longman of Harlow, UK
3
There is a difficulty in controlling length of extract in that narratives and service
encounters are closed episodes of unpredictable length, and many of these extracts are
shorter than four minutes, while the open-endedness of casual conversation means that
episodes can be considerably longer before participants negotiate some sort of break or
closure
4
The figure of five occurrences is based on a projected average of once per 5,000
words of conversation, 5,000 words representing approximately 24-25 minutes of talk
Some canonical features (eg adding-clauses with which and w/t-clefts) occur with this
156 GRAMMAR AND THE SPOKEN LANGUAGE

frequency in the mini-corpus and so the figure is claimed to have sufficient reliability for
our present purpose
5
While we concentrate here on what we call spoken grammatical features, we are
aware that some of the features we have chosen to highlight {e g situational ellipsis) may
also occur in certain types of written text, for instance friendly and informal letters,
postcards, diaries, e-mail messages, etc, as well as written advertising and literary texts
which purposely evoke conversational style (see McCarthy 1993 for examples) The
existence of such texts underlines the difficulty of creating a sharp boundary between
spoken' and 'written1 grammars and the need for comprehensive descriptions covering

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

the full range of genres and contextual types, which traditional grammars have usually
failed to provide
A
A new edition of this work is in press and is due in late 1994 We are very grateful to
Michael Swan for letting us read proofs of the new edition Where two section references
are given they refer to the 1980 edition and the new edition in that order
7
We are grateful to Robin Fawcett of the University of Wales, Cardiff College, for
offering us this view of the provenance of the dislocation metaphor
* Rod Ellis, MATSDA conference talk. University of Luton, UK, January 1 1994

REFERENCES
Aijmer, K. 1989 Themes and tails The discourse function of dislocated elements' Nor-
dic Journal of Linguistics 12/2 137-54
Alexander, L. G 1988 Longman English Grammar Harlow Longman
Arnaud, P and S Savignon forthcoming 'Rare words, complex lexical units, and the
advanced learner' in J Coady and T Huckin (eds ) forthcoming Second Language
Vocabulary Acquisition A Rationale for Pedagogy Cambridge Cambridge University
Press
Baynham, M 1991 'Speech reporting as a discourse strategy Some issues of acquisition
and use Australian Review of Applied Linguistics 14/2 87-114
Bennett, W. 1989 'The structure of English lags' Word 40/3 315-33
Bex, A 1993 'Standards of English in Europe" Multiltngua 12/3 249-64
Biber, D 1988 Variation across Speech and Writing Cambridge Cambridge University
Press
Biber, D , S Conrad, and R. Reppen 1994 Corpus-based approaches to issues in
applied linguistics 'Applied Linguistics 15/2 169-89
Brown, E. K 1980 "Grammatical incoherence" in H Dechert and M Raupach (eds)
1980 Temporal Variables in Speech Studies in Honour of Frieda Goldman-Eisler
The Hague Mouton
Collins COBU1LD 1990 COBU/LD English Grammar London Collins
Conine, B. 1986 Tense in indirect speech Folia Linguistica XX/3-4 265-96
Coulmas, F. 1985 "Nobody dies in Shangn-Ia Direct and indirect speech across
languages' Georgetown University Round Table on Languages and Linguistics 140-
53
Crowdy, S 1993 "Spoken corpus design " Literary and Linguistic Computing 8/4 259-
65
Downing, A. and P Locke 1992 A University Course in English Grammar London
Prentice HaJl
Geluykens, R. 1992 From Discourse Process to Grammatical Construction On Left-
dislocation in English Amsterdam John Benjamins
 RONALD CARTER AND MICHAEL McCARTHY 157

Glisan, E. and V. Drescher. 1993 Textbook grammar Does it reflect native speaker
speech9' The Modem Language Journal 77/1 23-33
Goodell, E. W. 1987 "Integrating theory with practice An alternative approach to
reported speech in English' TESOL Quarterly 21/2 305-25
Harman, I. 1990 Teaching indirect speech Deixis points the way' ELT Journal 44/3
230-8
Hieteranta, P 1984 'A functional note on topicalisation' English Studies 1 48-51
Huddleston, R. 1989 The treatment of tense in indirect speech' Folia Lingutstica
XXlll/3-4 335-40
so 1 said11 Verb tense alternation and narrative

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012

Johnstone, B. 1987 ""He says
depictions of authority in American English' Linguistics 25/1 33-52
Labov, W. 1972 Language in the Inner City Oxford Basil Blackwell
Lcwandowska-Tomaszcyk, B 1990 'Conversational data and lexicographic practice' in
T MagayandJ Zigany (eds) 1990 BudaLEX '88 Proceedings Budapest Akademiai
Kiado
Lewis, M 1993 The Lexical Approach The State ofELT and a Way Forward LTP
Hove, UK
McCarthy, M. J. 1992 'English idioms in use' Revista Canana deEstudios Ingleses 25
55-65
McCarthy, M. J. 1993 "Spoken discourse markers in written text' in J M Sinclair, M
HoeyandG Fox (eds) 1993 Techniques of Description London Routledge
McCarthy, M.J and R A. Carter. 1994 Language as Discourse Perspectives for
Language Teaching London Longman
Nattinger, J. R and J, S DeCarrico 1992 Lexical Phrases and Language Teaching
Oxford Oxford University Press
O'Connor di Vito, N. 1991 'Incorporating native-speaker norms in second language
materials 'Applied Linguistics 12/4 383-96
Owen, C 1993 "Corpus-based grammar and the Heineken effect Lexico-grammatical
description for language learners' Applied Linguistics 14/2 167-87
Philips, S U. 1985 'Reported speech as evidence m an American trial' Georgetown
University Round Table on Languages and Linguistics 154-70
Pica, T 1983 'The article in American English What the textbooks don't tell us" in N
WolfsonandE Judd(eds) 1983 Soaoltnguistics and Language Acquisition Rowley
Newbury House
Pinker, S. 1994 'Grammar Puss' New Republic 210/5 19-25
Quirk, R., S. Greenbaum, G. Leech, and J Svartvik. 1985 A Comprehensive Grammar
of the English Language London Longman
Ricento, T 1987 'Clausal ellipsis in multi-party conversation' Journal of Pragmatics 11
751-75
Riggenbach, H 1990 'Discourse analysis and spoken language instruction' Annual
Review of Applied Linguistics 1 1 152-63
Rings, L 1992 'Authentic spoken text;, as examples of language variation Grammatical,
situational and cultural teaching models' IRAL XXX/1 21-33
Summers, D. 1994 'Corpuses' EFL Gazette August 1994 18
Swan, M . I 980 Practical English Usage Oxford Oxford University Press New edition
published 1995
Tannen, D 1986 introducing constructed dialogue in Greek and American con-
versational and literary narrative'in F Coulmas(ed) 1986 Direct and Indirect Speech
Berlin Mouton de Gruyter
158 GRAMMAR AND THE SPOKEN LANGUAGE
Thomas, A 1987 T h e use and interpretation of verbally determinate verb group ellipsis
inEnglish 1RAL XXV/1 1-14
Thompson, C 1994 Reporting Collins-Cobuild English Guides No 5 London
HarperCollins
Wald, B 1987 "Spanish-English grammatical contact in Los Angeles The grammar of
reported speech in the East Los Angeles English contact vernacular' Linguistics 25
53-80
Yule, G , T Mathis, and M Hopkins 1992 "On reporting what was said ' ELT Journal
46/3 245-51

Downloaded from http://applij.oxfordjournals.org/ at McGill University Libraries on September 25, 2012