Annual Review of Applied Linguistics (1998) 18, 20-42. Printed in the USA.

Copyright ® 1998 Cambridge University Press 0267-1905/98 $9.50

THEORETICAL PERSPECTIVES ON SPEAKING*

Martin Bygate

INTRODUCTION

Second language (SL) speaking has not been recognized as a research area
as readily as have reading, writing, or listening. For instance, the 25th anniver-
sary volume of TESOL Quarterly had papers on reading, writing, and listening, but
apart from a paper on pronunciation, not a single one on speaking. This may be
due to assumptions such as the following: 1) SLA studies in any case focus on oral
language production (e.g., the morpheme acquisition studies in the 1970's, the
input-interaction studies of the 1980's for which oral interaction was only a way of
acquiring new target language items, and the task-based learning studies of the
1990's); 2) language pedagogy starts with oral skills and hence special attention is
only needed to reading and writing skills; or 3) grammars are still best based on
written corpora (an assumption noted by Carter and McCarthy 1995). While some
of the neglect may be more apparent than real (Crookes [1991] remarks that much
oral language research is distributed across a number of distinct areas of applied
linguistics), nonetheless, speaking is not fully accepted as a distinct area.

This review aims to cover a broad area, and inevitably will fail to do full
justice to substantial and significant research, both within the areas that it surveys,
and beyond the limits that it sets itself. The review approaches SL speaking in
terms of issues of process and of product: 1) what processes do learners use to
manage oral performance and how do the processes affect their performance? and
2) how far do SL learners approximate to target patterns of speech? Four main
questions structure this paper. First, is it the case that speech is different from
writing? Second, is it the case that oral discourse patterns differ from one language
to another? Third, are non-native speaker oral skills significantly different from
those of native speakers? Fourth, how does second language speaking develop? In
order to situate these questions, this article refers to early landmark papers as
appropriate, particularly in areas where little recent research has been published.

20
 THEORETICAL PERSPECTIVES ON SPEAKING 21

THE ORAL LANGUAGE PRODUCT

On the assumption that language processing operates in the service of

targeted language output, and that language output is patterned, it is important first
to establish the norms of oral language output. The specific characteristics of
speech compared to writing have been studied over three decades (see Akinasso
1982 for a review). While recognizing that oral and written modes do not involve
categorical differences, Hatch (1992) identifies three parameters which can serve to
differentiate spoken and written discourse production: planning, contextualization,
and formality. It is probable that speech is more commonly unplanned, contex-
tualized, and informal than is writing. In addition, speech is more commonly
reciprocal than is writing. These parameters of discourse production and the
notion of reciprocity lead to the possibility of identifying a number of features that
distinguish oral from written language. One can further claim that spoken
language is highlighted by certain features and that while speech does not
necessarily use these features, they are more prototypical of speech than of writing.
These features include at least the following: clausal and phrasal organization
(rather than sentence organization); left dislocation; features of involvement versus
features of detachment; lower levels of lexico-grammatical formality; and nextness
(or parataxis), parallelism (and presumably repetition), repair, and conjunction
(rather than subordination). In addition, it is reasonable to claim that spoken
language is characterized in typical ways:

1. by certain types of discourse structures or moves (e.g., reciprocal

openings and closings, interactive negotiation of meaning, conversation
structures, question-answer sequences, pre-sequences, and exchange
clusters [Hoey 1993]);
2. by certain characteristic types of speech acts (e.g., apologies,
compliments, greetings, interruptions);
3. by certain grammatical features (such as first and second personal
pronouns, deictics, question forms, contractions—see Eggins and Slade
1997:67-115 and Carter and McCarthy 1995 for identification of additional
grammatical features of speech);
4. by typical features of the speech stream (such as segmental and
suprasegmental features, pauses, and hesitations).

Speech, then, is constituted by constellations of these features occurring with

typical patterns of frequency. The patterns of frequency vary of course according
to different genres and subgenres, but together can be conceived of as forming
'families' of related constellations. If it is the case that speech does pattern across
these different dimensions in statistically regular (though not of course invariable)
ways, then it is reasonable to assume that through experiencing speech, speakers
will come to form mental representations of these patterns (see Levelt 1989, and
below).
22 MARTIN BYGATE

Various studies have demonstrated aspects of patterning of speech.

Empirical studies demonstrate differences between the patterns of oral and written
language (e.g., Carter and McCarthy 1995, Chafe 1985, Schleppegrell 1996,
Tannen 1982, Thompson 1994). Biber (1988) and Biber, Conrad and Reppen
(1994) illustrate how specific types of speech (such as conversation) can be situated
along several dimensions, distinctively from other types of discourse. Tannen
(1984) shows how speakers from different speech communities within the same
language may differ in their use of overlaps, expressions of involvement, and
approaches to story-telling. In a similar way, her subsequent publications (e.g.,
1990; see also Holmes 1995) provide evidence of differences between genders in
aspects of their discourse patterning and in the use of speech acts (such as
politeness strategies, e.g., Holmes 1995). The patterns used in particular speech
events have consequences and may contribute to communication problems, for
instance, between doctor and patient (e.g., Coupland, Robinson and Coupland
1994, Wodak 1996), between speakers of one region and another (e.g., Hatch
1992), or in creating and maintaining power and role relationships in the family
(e.g., Ochs and Taylor 1992). In other words, in private and public roles,
discourse structures are used to construct or maintain identities and relationships,
whether individual or group, often not to mutual satisfaction.

If problems can arise between native speakers, similar problems are at

least as likely to be the case where there are differences in discourse norms
between cultures. Gumperz's (1982) classic volume on discourse strategies in
interaction demonstrates how differences in the structures and cues in discourse can
give rise to misunderstandings regarding such factors as interpersonal attitude and
personality; these findings are similar to the kind found between native speakers
identified by Tannen (1984). Ranney (1992) also illustrates ways in which the
discourse expectations of NNS's can be at odds with those of the target
community, enabling predictions of cross-cultural misunderstandings in potentially
crucial contexts (e.g., doctor-patient interactions). Cross-cultural and cross-
linguistic instances of this are reported and discussed in studies of cross-cultural
communication in the work place (Roberts, Davies and Jupp 1992) and in
academic contexts (Bardovi-Harlig and Hartford 1990). Cross-cultural studies
have extended into the area of interlanguage pragmatics, demonstrating differences
between cultures in the selection and realization of speech acts (see particularly
Kasper and Dahl 1991 for a detailed review). Clearly then, speakers use
knowledge of discourse structures and patterns in the selection and realization of
speech acts in negotiating communication; hence, in contexts in which SL students
must interact, knowledge of the target-language patterns may be needed.

Patterning exists at a number of levels, from macro-structures of

discourse, through lexico-grammatical patterns, to the distribution of intonation
patterns, pauses, and hesitations in the stream of speech. The identification of
such genuine discourse norms is a necessary background to the development of
language pedagogy.
 THEORETICAL PERSPECTIVES ON SPEAKING 23

There is, however, a view of discourse patterns which gives them a

dynamic interpretation. Gumperz (1982), while clearly identifying what might be
termed canonical discourse patterns in the talk of people from different cultures,
nonetheless sees discourse as strategically negotiated. Speakers exercise choices,
and interaction therefore provides opportunities for joint construction of communi-
cation. This jointly constructed communication will be returned to in subsequent
sections. For now it is important to note that frequency data—remembering that
frequency data is always subject to variation—provides an important frame of
reference for language learning and language use; but such patterning situates
rather than determines speech, involving choices as well as knowledge of patterns.
This patterning is an important basis for understanding processes of speech
production, which is the focus of the next section.

PROCESSES OF ORAL LANGUAGE PRODUCTION

In order to provide a frame of reference for discussing SL oral production,

this section first reviews the study of native-speaker (NS) production processes. It
first draws on the literature to present a framework for understanding oral language
production and then discusses issues of discourse routines, lexical processing,
prefabricated chunks, and pausing.

In terms of the underlying processes of production, speaking is broadly

seen as an activity integrating distinct aspects of interpersonal and psycho-motor
control. Drawing on a cognitive tradition going back to Miller, Galanter and
Pribram (1960), Levelt (1989) argues that messages are 'planned,' and, for this,
have to be 'conceptualized,' then 'formulated,' and finally 'articulated.' Faerch
and Kasper (1983), Garman (1990:376-377), and Garrett (1980) share a broadly
similar view. The speed of processing, which is largely constrained by limitations
in the capacity and availability of working memory (Gathercole and Baddeley
1993), crucially has two requirements: chunking of language units into
economically bundled sequences of words and phonemes (Schmidt 1992), and
automation in the implementation of the process. The bundling of phonemes and
words into optimally sized chunks depends on the process of 'restructuring'
(McLaughlin 1990, Schmidt 1992), whereby familiar information is reorganized in
memory into more manageable units. Automation may then be needed for four
key processes: 1) selecting appropriate conceptualizations, 2) accessing and
sequencing appropriate formulations, 3) relating formulations to articulations, and
4) producing the articulations (Levelt 1978). It is likely that automation takes the
form of a degree of routinization at the level of conceptualization (where messages
are selected in the light of contextual factors and goals). Processing also involves
monitoring (Faerch and Kasper 1983, Levelt 1989) which can take place both
'covertly' during production but before articulation, and 'overtly' following
articulation (Laver 1970). This processing takes place within contextual
constraints (see below). Hargie (1986) and Hargie and Marshall (1986) illustrate
24 MARTIN BYGATE

how this psycholinguistic account can be integrated into an interpersonal

framework.

In order to demonstrate some of the issues contained within Levelt's

framework, we now consider four crucial subtopics: discourse routines, lexical
processing, prefabricated chunks, and pausing. Discourse routines implicate some
degree of storage of accessible information sequences at the conceptual level. A
study by Snow (1987) suggests that second language production of definitions and
explanations correlates significantly with speakers' ability to perform the same task
in the first language. Findings of this kind would suggest that familiarity with
conceptual demands of specific discourse tasks can have an impact on
performance.

With respect to lexical processing, there are a number of research

questions concerning lexical access under conditions of speech (e.g., de Bot and
Schreuder 1993, Schreuder and Flores d'Arcais 1992). In terms of understanding
capacities for lexical production in speech, one issue involves the impact of
background knowledge on lexical access. A second issue concerns the extent to
which the semantic networking of a speaker's lexical system can be promoted by
different types of production tasks. A third issue concerns the relationship between
lexical access and the size of a speaker's lexicon. A fourth issue (though again a
relatively neglected one) concerns the relationship between lexical access and
morpho-syntactic co-text. Levelt's (1989) model places lexis at the center of
language processing. Whether this is an appropriate view or not, the question still
remains as to how the reciprocal relationship between preceding or following
morpho-syntactic elements of an utterance and lexical access are integrated (see
Bock 1987). It would be useful to understand how speakers engage the central task
of interweaving the two classic parameters of language selection: paradigmatic
alternatives in relation to the speaker's intended meaning, and syntagmatic
possibilities in relation to the speaker's emerging co-text. Few papers however
address issues of lexical processing in relation to speech production.

Associated with the discussion of lexical processing, a third issue attracting

increasing interest is the processing of prefabricated chunks. The discussion of the
role of prefabricated chunks in oral language processing (Pawley and Syder 1983)
raises various interesting questions: First, what kinds of sequence are internalized
and stored as collocational units? Second, how do these sequences evolve (in
particular how are they grammaticized) during acquisition? Third, to what extent
is the appropriate selection of chunks a function of the mode and register of the
discourse? Most NS accounts of prefabricated chunks (e.g., Bates, et al. 1988,
Locke 1993, Ochs and Schieffelin 1983, Peters 1983) are concerned with child
language, combining developmental factors with issues of speech processing,
though Wray (1992) is a notable exception to this. As with lexis, by analyzing
cases of speech processing of prefabricated chunks (using both normal and
pathological data)—that is, by taking an overt category as the focus of obser-
 THEORETICAL PERSPECTIVES ON SPEAKING 25

vation—it is possible to derive a sense of the underlying processes which affect the
surface distribution of these chunks. Thus, prefabricated chunks provide a further
source of data on the processes of conceptualization, formulation, and articulation.

A fourth issue of interest concerns the patterns of pausing in speech, an

area Griffiths (1991) describes as having been seriously neglected. Once again, the
distribution of pauses can be taken as a clue to some of the underlying processes
involved in speech production. Evidence (e.g., Beattie 1980) suggests that, when
holding constant both task and context, native speakers use characteristic pausing
patterns. Beattie argues that while pausing is undoubtedly multi-functional, it is
possible to identify distinct distribution patterns of pausing, suggesting that some
distinct planning functions can be identified, notably 'proximal' and 'distal'
planning (associated with formulation and conceptualization respectively).
Consistent with Beattie's view, Pawley and Syder (1983) argue that the nature of a
given communication task may significantly affect the type of pausing according to
the pressure on precision of formulation that is involved. As with the issues of
lexical selection and prefabricated chunks, patterns of pausing relate to the three
speech processes underlying planning: conceptualization, formulation, and
articulation.

Planning processes in speaking imply choice as well as automation. The

identification of three areas of planning processes (conceptualization, formulation,
and articulation) is consistent with a functional (Halliday 1985, Bloor and Bloor
1995) and pragmatic (Hatch 1992, Yule 1996) view of language which recognizes
the possibility of a speaker's exercising choice both among meanings to commu-
nicate and among a range of more or less conventional formulations for expression
of the selected meaning. Yet while choice is possible, a degree of social
convention and individual routinization is also recognized for each area of
planning. Individual routines develop because our perceptions, understandings,
processing capacities, memories, and experiences of language are inevitably an
idiosyncratic mixture. Social conventions arise for a number of reasons: 1)
because concepts, while mentally stored by individuals, also reflect socialization
outcomes (Bolinger 1975, Bruner 1983, Pawley and Syder 1983); 2) because an
individual's store of formulations will reflect the conventions of socio-economic,
professional, gender, and generation groupings (Giles, Coupland and Coupland
1991); and 3) because an individual's articulations usually approximate to a greater
or lesser extent to those of social groupings.

Hence speech production involves the following components and concepts:

distinct areas of planning, a range of choices in each planning area, limited
processing capacity due to limitations of working memory, a consequent need for
routinization or automation in each area of production, the monitoring of speech
during and following production, and the managing of communication under a
range of external pressures. It remains unclear how the different areas of planning
interrelate, whether as separate modules as in Levelt's 1989 model, as hierarchical
26 MARTIN BYGATE

areas in which planning in one area is correlated hierarchically to planning in

another, or as loosely networked decision-making paths for different elements of
the lexico-gramatical store. It is clear, however, that separate types of mistakes
can be made for each type of planning; that is, mistakes may occur in message
selection, formulation, or articulation (Garman 1990, Garrett 1980), lending
support to the belief that different operations are involved at each level and that the
three types of planning activity (including conceptual planning of the overarching
message) are normally engaged simultaneously (Levelt 1989), albeit with activity
in articulation slightly delayed in relation to the others.

It could be argued from the above that at all levels of planning, two major
dimensions are constantly in play and in need of attention by speakers, selection
and combination. Plans clearly have to be selected that are judged appropriate—
whether at the level of conceptualization, formulation, or articulation—bringing
together meaning and form. But equally, plans have to be combined, both with
preceding and with following plans. This involves adaptation of current plans to
what has gone before. It also entails adjusting current plans to what may follow.
As with selection, combination is needed at each level of planning. Finally, these
processes need to be monitored for meaning under pressures of time, with possible
effects on accuracy and recall (Hildyard and Hidi 1985). We will return to these
issues below (see sections on second language production processes and oral
second language development). For now we can conclude that fluent native
speakers integrate a number of processes in which selection and combination are
equally important. Not only is it incorrect to stress the focus on meaning without
reference to form, but equally it is partial to exclude the dimension of combination
from the processes of selection. In the next section, we relate these issues to the
study of SL production processes.

SPEAKING IN A SECOND LANGUAGE

In this section, we consider studies of processes involved in oral second

language production in terms, firstly, of internal psycholinguisdc processes, and
secondly, of the conditions of speech on speech output.

1. Second language production processes

The few studies of oral SL production processes available fall broadly in

line with the account of oral language production outlined above. A central
question is whether SL speakers undertake similar kinds of processing decisions as
NS speakers. It is broadly assumed that they do. For instance they have to fix on
appropriate message content, including appropriate component concepts within
messages (such as the appropriate entities, attributes, and processes needed to
constitute elements of utterances). They have to formulate the message content,
articulate it, and monitor the production process so as to make adjustments as and
when necessary according to what they judge appropriate to interlocutors' com-
 THEORETICAL PERSPECTIVES ON SPEAKING 27

prehension and acceptance. To this extent, non-native speakers are presumed to

process speech like native speakers. However, non-native speakers may also
engage processes that are distinct from those of native speakers. This unique
processing may occur in at least four ways: in terms of lexical access, in terms of
pausing, in terms of compensatory communication strategies, and in terms of the
balance between selection and combination processes.

First, terms of lexical access, de Bot (1992) discusses whether each of the
three processes of conceptualization, formulation, and articulation are language
specific, or whether it is possible that resources at any of the levels are shared
across languages. He concludes that the conceptualizer is partly language specific
and partly language independent, and that formulators are language specific, but
draw on a single non-language-specific lexicon. Poulisse and Bongaerts (1994)
find that the use of the LI in SL production supports this view. SL conceptualizing
and formulating seem in some ways dien to be clearly different from LI
processing.

Second, terms of pausing in the SL, as mentioned above, Lennon (1990)

argues that it is a complex of several factors and that there is no clear correlation
between any one specific factor and proficiency. While the phenomenon of
pausing clearly combines with a number of formal features of fluency, such as
location, duration, and length of intervening uninterrupted 'runs' (Lennon 1990),
and while it is also multi-functional (Griffiths 1991), the relevance of pausing for
understanding oral SL production is reflected in three further types of evidence: 1)
pauses may relate to a speaker's current processing capacity—and hence represent
one aspect of a speaker's proficiency; 2) pauses may vary according to task
demand, including the processing of conceptualization, formulation, or
articulation; and 3) pauses may vary according to task conditions. One area of
neglect concerns syntactic placement of pauses. By extrapolating from Beattie's
(1980) account of NS pausing in terms of 'proximal' and 'distal' planning (see
above), it is likely that pause distributions, particularly with relation to immediate
planning, vary in terms of proficiency, yielding different distributions of pausing
from those of native speakers. This is an area for further investigation and one
mat could shed additional light on patterns of SL speech processing.

Third, terms of strategy use, it has been widely assumed that NNSs differ
from NSs in terms of the nature and use of strategies for dealing with
communication problems. This assumption follows partly from the nature and
number of problems that NNSs have to deal with, and partly from the limited
number of resources available. Here data seem to reflect level of proficiency
differences (Bialystok 1983) with less proficient speakers using more LI-based
strategies. Bongaerts and Poulisse (1989) found strong similarities in LI and L2
uses of strategies when these are defined in terms of conceptual (holistic or
analytic) strategies rather than in terms of compensatory strategies. (Some con-
ceptual strategies could also be termed 'formulative' in that they are sometimes
28 MARTIN BYGATE

devoted to generating words capable of filling lexical gaps.) Speakers would tend
to use holistic strategies in terms of concept selection, whether in LI or L2. This
finding would be consistent with Bialystok's research, which focused more on use
of formulative strategies. In other words, NNS communication strategies would
appear to differ from NS strategies in terms of frequency and formulation type
rather than in terms of mental processes. (For a recent review of work in
communication strategies, see Dornyei and Scott 1997.)

Finally, a fourth possible difference between second and first language

speech processing concerns the use of the processes of selection and combination
in relation to the use of formulaic chunks. Following work by Wong Fillmore
(1979) and others, both Nattinger and DeCarrico (1992) and Weinert (1995) argue
that, just as in LI acquisition (Peters 1983), word sequences are commonly first
acquired and used in SL speech as whole unanalyzed expressions, to be possibly
re-analyzed subsequently.

In line with this view, Skehan (1996) suggests that the conditions of
second language learning can lead learners to vary far more than first language
learners in the way they select and combine language, notably in terms of the size
of chunk that is selected and the extent to which they work with chunks or with
combinatorial potential. Memory-oriented learners (Skehan 1986) and commu-
nicatively-oriented learners (Skehan 1996, Willing 1987, Wong-Fillmore 1979;
1982) may be more inclined to communicate through unanalyzed or unmanipulated
chunks of language. In contrast, learners who are more rule-oriented (Skehan
1986, Wong-Fillmore 1982), more analytic (Willing 1987), or more authority-
oriented (Willing 1987, Wong-Filmore 1982) might prefer to concentrate on
exploring the combinatorial possibilities of words rather than on building up a
repertoire of collocational phrases. Compared with first language learners, who
from the start find their own fluent balance between the processes of selection and
combination, second language learners are likely to manage the two processes less
seamlessly. While it is tempting to suggest that one style augurs better than the
other for language development, the truth is more likely that both styles have
strengths and that a 'flexible' style combining both dimensions may be the best.
While this position clearly anticipates the discussion below on the processes of oral
second language development, we can recognize the possibility of 'unbalanced'
processing as a fifth 'systemic' difference between LI and L2 speech processes.

A processing perspective, then, suggests that the speech production of

NNSs in some ways resembles, and in other ways is distinct from, that of NSs. It
also suggests that patterns of oral SL development may be sui generis, that is,
distinct from NS patterns of development and distinct from patterns of development
in SL listening or written skills. These factors will be drawn on below to
contribute to a perspective on the learning, teaching, and testing of speaking.
 THEORETICAL PERSPECTIVES ON SPEAKING 29

2. Conditions on speech production

Because of its particular relevance to second language education, the issue

of conditions on speech production has been left until this point. It is, nonetheless,
an important topic because it helps us to understand ways in which performance
can be related to its context. A person's level of skill is conceptualized as a
capacity for performance. This, however, has to function in—and can be
influenced by—contexts of operation. In terms of LI performance, Levelt (1978)
identifies the following significant contextual factors: demand, arousal, and
feedback. Each of these factors can interact positively and negatively with a
person's skill: positively in the sense that each can stimulate higher levels of
performance, and negatively in the sense that each may cause lower levels of
performance.

According to Levelt (1978), 'demand' refers to the amount of processing

required by a given task. An appropriate level of demand elicits an optimum level
of skilled performance. Level of demand affects the level of response, with
increasing demands tending to increase the amount of planning required. A
demand that is too high or too low can lead to malfunction or under-achievement
respectively. 'Arousal' refers to the speaker's emotional and cognitive response to
a task and can be a function of the importance attached by the individual to
successful performance. Like demand, arousal can either be enabling or else be
too high or too low. Hence being asked to do a simple task can be made difficult
by giving great importance to the performance being successful. Finally, feedback
can influence performance simply in terms of the quantity and quality of feedback
being relayed to the performer. Once again, either an excess or an absence of
feedback can have a negative impact on performance.

Turning to SL performance, the impact of conditions on speech

performance, particularly in terms of task and task conditions, is, of course, a
central concern in language teaching and testing. Skehan (1996) identifies the
following three major factors and relevant sub-components which incorporate two
of Levelt's categories:

1. Code complexity
2. Cognitive complexity
•Cognitive processing
•Cognitive familarity
3. Communicative stress
•Time pressure
•Modality
•Scale
•Stakes
•Control
30 MARTIN BYGATE

(Note that 'modality' concerns the distinction between speech and writing. For
another model of language processing, see Bachman and Palmer 1996.) These
factors have not been widely researched. However, there are studies that have
considered cognitive familiarity, both in terms of topic familiarity and in terms of
scale and control.

In terms of topic familiarity, Selinker and Douglas (1985) first presented

NNS data supporting the view that interlanguage talk is influenced by familiarity of
'discourse domain'—a combination of topic, interpersonal context, and discourse
type. On the basis of their data, speech on familiar topics and in congenial
contexts is likely to be more accurate and fluent than speech in other contexts.
Whether this result is due to speech being more automatic and less monitored in
familiar domains, while requiring more (potentially developmental) work in less
familiar domains, is open to discussion. The apparently positive impact of topic on
speech production is also demonstrated by Zuengler and Bent (1991). If more
competent production is a result of speaking on familiar topics, this would be
important for testers in terms of sampling a range of speech, as well as for
pedagogy, although once again, the developmental impact of familiar topics needs
fuller investigation.

Ellis (1987) finds evidence that planning and rehearsal of a piece of

discourse improves accuracy in regular past tense forms, while Crookes (1989)
reports a planning effect on complexity and fluency, rather than accuracy. Foster
and Skehan (1996) report significant results with respect to the effect of unfocused
planning on fluency, accuracy, and complexity. These studies, while being
consistent with the work by Selinker and Douglas, and Zuengler and Bent,
introduce the point that control may be extended into fluent performance through
planning and rehearsal. This set of findings suggests that the vernacular dimension
of speech, rather than being relatively less open to developmental influences, may
in fact be influenced by circumstances of performance in more subtle ways than
those outlined in the Labovian approach (Tarone 1987): All vernacular speech is
not the same, some perhaps being more form-sensitive than others. This
perspective may be important for both language pedagogy and testing. There have
been few other recent studies on this topic, and those that exist do not consider the
impact of content familiarity on discourse performance.

Scale concerns principally the number of interactants/interlocutors.

Crookes (1986) and Pica, et al. (1993) both indicate ways in which the number of
interactants can affect the discourse. Shohamy's (1994) study of testing formats,
however, suggests that scale may be less important than interpersonal relations.
This result indicates the impact of the control factor, a point further supported by
Young and Milanevic (1992) and Young (1995). Control, in particular, affects the
interaction patterns, especially in terms of initiations and topic changes. While
these studies concern NNS-NS interaction, a study by Yule and MacDonald (1990)
demonstrates a similar strong effect of proficiency and interactive role on NNS-
 THEORETICAL PERSPECTIVES ON SPEAKING 31

NNS interaction. In other words, it may be that control in terms of speech roles,
power relations, and proficiency relations all have a similar impact on speech
patterns, and hence on speech processing.

To summarize this section, NNS talk is subject to types of variation similar

to that of NS talk. However, it can be differently influenced by planning, rehear-
sal, or familiarity (which can be construed as a form of unconscious planning),
apparently aiding production in terms of complexity, accuracy, and fluency. We
can note that the processes of selection and combination in the performance of
speech plans are likely to be affected by the conditions of performance, whether in
the LI or in the L2. A demanding task accompanied by high arousal and no
feedback will test both a native and non-native speaker's use of language, just as an
undemanding task, with low arousal and meager feedback, is unlikely to stimulate
a high level of performance.

Finally, in considering the learning of oral SL skills, learners need to

integrate the following abilities:

• planning and implementation at the levels of conceptualization,

formulation, and articulation;
• the development of fluency in terms of automation of plans at all levels;
• attention to accuracy at all levels;
• attention to complexity;
• the development of the ability to self-monitor.

Skehan (1996) suggests that fluency, accuracy, and complexity are qualities that
compete for processing capacity, a point we return to in the final section.

ORAL SECOND LANGUAGE DEVELOPMENT

So far, this paper has considered the nature of speech processing in

general, the nature of speech as 'product' or target norm, and the nature of second
language speech processing in particular. This section now considers the develop-
ment of second language speech processing, viewed principally from two angles:
from a sociolinguistic perspective, viewing interactive speech as a source of new
language input to the learner; and from a cognitive perspective, emphasizing
theories of skill development.

1. Sociolinguistic accounts

Sociolinguistic approaches, rather than considering the development of

skilled abilities by an individual, tend to place the emphasis on the contribution of
group interaction to development. Four developmental approaches can be dis-
tinguished here. One approach has emphasized the role of language as a marker of
group affiliation. If a learner's desire for group affiliation is significant, if
32 MARTIN BYGATE

language plays an important role in group affiliation, and if the learner perceives
the connection between the adoption of language patterns and group affiliation, the
speaker will attempt to approximate to the language patterns preferred by a target
group. The speaker would seek to imitate, to pick up phrases used by the group,
to interact in the manner of the group. This 'accommodation' can operate at the
level of conceptualization, as well as more obviously at the levels of formulation
and articulation. Zuengler (1991) reviews the contribution of communication
accommodation theory to second language acquisition studies. Tarone and Swain
(1995) provide data to suggest that at secondary school levels, a student's sense of
group affiliation can play a key role in second language acquisition, and Ross and
Berwick (1992) apply the concept to an analysis of oral SL examination interviews
(though not to developmental issues).

A second approach studies the discourse patterns between NNS and NS in

terms of the strategies used to support communication. This type of study is
developed by Bremer, Roberts, Vasseur, Simonot and Broeder (1996) in their
recent account of the way understandings are achieved in intercultural encounters
between migrant workers and members of the host community. Bremer, et al. 's
project adopts an ethnographic approach to the study of different strategies used by
speakers in negotiating understanding. This line of research assumes that the
strategic use of discourse moves and interpretative schemata can play a significant
role in promoting language development: Out of discourse processes can emerge
communicative competence. This 'thick' approach to the analysis of data—using
extended extracts, actual and simulated interactions, role plays, and feedback
sessions—enables the dynamics of the interaction to emerge, permitting reasonable
inferences about the impact of different strategies on language development. Tyler
(1992; 1994), among others (e.g., Madden and Myers 1994), uses a similar
methodology in her study of the oral discourse of international teaching assistants
(ITAs).

A third approach is adopted by diose exploring the development of

pragmatic competence (see Kasper 1996 for a recent review). Bardovi-Harlig and
Hartford (1990; 1993) illustrate this approach in their studies of the pragmatic
development of NNS students within English-medium higher education, tracing the
development of students' abilities to deal with specific pragmatic problems.

The fourth approach, closely related to the third, involves concentrating on

the role of negotiated NNS output in language development. In line with the view
of output activities as an important site for learners to develop by experimenting
with language, Swain (1985) outlines a case for seeing comprehensible 'pushed'
output as an important contributor to the development of language proficiency, a
view also developed in Bygate (1988). Oral language production becomes an
occasion for learners to establish patterns in the selection and combination of
language items. In more recent work, Swain and Lapkin (1995; 1997) argue that
through collaborative co-construction, NNS's are able to help each other
 THEORETICAL PERSPECTIVES ON SPEAKING 33

conceptualize and formulate meanings, thus developing their interlanguage

interactively. This view sees interaction as enabling significant acquisitional
processes to engage.

Peirce (1995) and Angelil-Carter (1997) develop a distinct but related

point in arguing that the personal investment of the speaker in role relationships, in
the use of the target language in general, and in specific discourses can either
hinder or facilitate language acquisition. Although this shift in direction raises
fresh methodological questions, it is a development that many will endorse (e.g.,
McNamara 1996:85-87).

Sociolinguistic approaches then are more concerned with the impact of

social conditions on language development, making specific predictions of
language development difficult to make. The study of the collaborative production
of output, however, provides a basis for interrelating interactive processes with
psycholinguistic processes.

2. Psycholinguistic processes of oral SL development

What then of the patterns of oral SL development? While oral SL

development, like most human development, is socially situated, it is clear that
language constitutes the interface between social and cognitive processes.
Language that is learned must be both socially motivated and cognitively managed.
The approach adopted in this paper follows Hulstijn's (1990) and, to some extent,
Johnson's (1996) argument that acquisition can proceed from analyzed knowledge
to unanalyzed knowledge, and from controlled processing to automatic processing,
as well as the reverse. While language use seems to involve the combination of
analyzed and unanalyzed judgements, a central issue concerns how the two
dimensions develop, the extent to which they can complement each other, and the
extent to which either one can be stimulated. Mitchell and Martin (1997) point to a
conflict of interpretation between researchers and teachers on the role of formulaic
chunks in their data. While the researchers see the contextualized use of formulaic
chunks as providing the learners with a site for progressive grounded analysis of
the chunks during communication, teachers seem to view the use of unanalyzed
chunks as best proceeding in parallel with the teaching of grammar rules.

As discussed earlier, a speaker's content knowledge appears capable of

influencing oral SL performance. However, there have been relatively few studies
that focus on the effect of content familiarity on oral development over time.
By gate (1996) reports a case study in which topic familiarity is deliberately
encouraged through task repetition. The repeated version of the performance
demonstrates more native-like formulation, greater complexity, and more self-
correction (compared with more hesitation for lexical accessing on the first per-
formance). The implication is that an increase in content familiarity can lead
speakers from an initial concern with the conceptual content of their message, and
34 MARTIN BYGATE

with finding a passable formulation for it, to a point where they can shift attention
towards a greater concern with the quality of a chosen form, notably in terms of
lexical access and self-correction. This view may complement and give a
developmental dimension to Zuengler and Bent's (1991) and Selinker and
Douglas's (1985) findings.

More importantly, viewing oral SL development within the controlled

context of a selected task (rather than attempting to trace one aspect of language
development such as the use of a particular morpheme, syntactic structure, or
discourse function) permits a fuller profiling of development within the particular
skill domain. Thus Tonkyn (1996) is able to use longitudinal data sets to point to
evidence of trade-offs between accuracy, fluency, and complexity, a picture that
Foster and Skehan (1996) are also able to relate to the effects of task performance.
Fluency, accuracy, and complexity stand as qualities of performance, reflecting the
ways in which the underlying processes of conceptualization, formulation, and
articulation can be managed. Here there are interesting avenues of research
opening up to enable a fuller picture of the ways in which NNS's can become—and
be helped to become—fluent target-like users of a second language.

CONCLUSION

Research in SL speaking sets three major challenges. The first challenge

is understanding the patterning of oral language development as an integrated
system, including the ways in which fluency, accuracy, and complexity can
interact developmentally. The second involves understanding the impact of the
conditions of speaking on development, including the ways in which learners
situate themselves and the target language in relation to their broader context.
There is, however, a third challenge, which has not been discussed in the context
of this paper, but which is nonetheless central to a full understanding of the
construct, and that involves the ways in which written skills and listening can have
an impact on speaking, and the ways in which speaking can have an impact on the
development of interlanguage. It is inevitable that the real time processing of
listening activities, the exposure to language via reading and listening, and the
attention to form-meaning relations in all skills can wash forward to help the
development of speaking. At the same time, exploration of meaning-form relations
through speaking activities can be expected to contribute in some ways uniquely to
interlanguage development.

Nonetheless, as Bachman and Palmer point out (1996), understanding

language development cannot be free from performance factors. Ultimately we are
always assessing language development within the context of communication and
under a specific range of conditions. Speech is a uniquely self-defining activity.
For this reason, it is becoming increasingly timely for research to assess the
development of SL speaking skills in their own right.
 THEORETICAL PERSPECTIVES ON SPEAKING 35

NOTES

* I would like to thank Virginia Samuda for comments on drafts of this paper.
Any shortcomings are my own.

ANNOTATED REFERENCES

Bardovi-Harlig, K. and B. S. Hartford. 1993. Learning the rules of academic talk:

A longitudinal study of pragmatic development. Studies in Second
Language Acquisition. 15.279-304.

This paper reports on the pragmatic development of NNS students in

academic advising sessions. It demonstrates students' abilities to converge
without instruction towards target norms for managing the interactive
session in terms of managing their own discourse role. The picture given
by the paper is, however, more reassuring than the reader might intuitively
feel is justified, and further research is needed to study pragmatic
development at the workplace.

Bongaerts, T. and N. Poulisse. 1989. Communication Strategies in LI and L2:

Same or different. Applied Linguistics. 10.253-268.

This study succeeds in disentangling conceptual and formulation strategies

of SL speakers and in relating the former to their NS performance. The
study, however, leaves for further exploration issues involving level of
proficiency, learner style, and the relationship between conceptual
strategies and formulation strategies.

By gate, M. 1996. Effects of task repetition: Appraising the developing language

of learners. In D. Willis and J. Willis (eds.) Language teaching
alternatives: Methodological innovation in English language teaching.
Oxford: Heinemann. 136-146.

This study attempts to demonstrate that on-task focus of attention can shift
from conceptualization to formulation processes in speech (integrating
attention to form and meaning within the same task), and that this
integration can be exploited to promote acquisition. Weaknesses in the
operationalization of measures, and the fact that this is a case study,
indicate a need for further work.

Ellis, N. C. 1996. Sequencing in SLA: Phonological memory, chunking, and

points of order. Studies in Second Language Acquisition. 18.91-126.
36 MARTIN BYGATE

This study provides an exposition of current psychology theory on the

relationship between memory for specific language encounters and rule
learning. It deserves specific interpretation and application to the various
kinds of learning implied by an integrated model of oral SL production.

Hoey, M. 1993. The case for the exchange complex. In M. Hoey (ed.) Data,
description, discourse. London: HarperCollins. 115-138.

This paper proposes a structure of oral interaction analogous to the role of

the paragraph in written discourse. The argument supports the need for an
improved understanding of oral discourse structure beyond the exchange,
but poses problems of operationalization, particularly in terms of
reliability.

Ochs, E. and C. Taylor. 1992. Family narrative as political activity. Discourse

and Society. 3.301-340.

This article presents a very interesting contextualized study of native-

speaker interaction within the family, demonstrating the possibility of
systematic gathering of large data bases of spontaneous speech. The
methodology, however, leaves questions of subject validation unanswered.

Roberts, C , E. Davies and T. Jupp. 1992. Language and discrimination. A study

of communication in multi-ethnic workplaces. London: Longman.

This is a remarkable book, dealing with issues of methodology and the

substantive analysis of contextualized spontaneous talk. The sensitivity of
the analysis and the clarity of the methodology make this a reference book
which would be valuable for carrying out a wide range of research projects
in oral SL discourse.

Schmidt, R. W. 1992. Psychological mechanisms underlying second language

fluency. Studies in Second Language Acquistion. 14.357-385.

This article provides a theoretical account of SL fluency which, unusually,

provides valuable links to developmental processes.

Swain, M. and S. Lapkin. 1995. Problems in output and the cognitive processes
they generate: A step towards second language learning. Applied
Linguistics. 16.371-391.

This paper illustrates how second language interaction between pupils may
spur development. The approach is intuitively appealing, but the lack of
longitudinal data suggest that this is a type of study to be developed
further.
 THEORETICAL PERSPECTIVES ON SPEAKING 37

Tonkyn, A. 1996. The oral language development of instructed second language

learners. In H. Coleman and L. Cameron (eds.) Change and language.
Clevedon, Avon: British Association for Applied Linguistics/Multilingual
Matters. 116-130. [British Studies in Applied Linguistics 10.]

This is a rare longitudinal study of oral SL development; it provides a

demonstration of the ways in which a speaker's oral development may
vary in the extent to which it involves fluency, accuracy, or complexity
over a given period. The paper, though, is based on the productions of
only eight subjects on limited tasks, and the patterns deserve fuller study.

UNANNOTATED BIBLIOGRAPHY

Akinasso, F. N. 1982. On the differences between spoken and written language.

Language and Speech. 25.97-125.
Angelil-Carter, S. 1997. Second language acquisition of spoken and written
English: Acquiring the skeptron. TESOL Quarterly. 31.263-287.
Bachman, L. and A. S. Palmer. 1996. Language testing in practice. Oxford:
Oxford University Press.
Bardovi-Harlig, K. and B. S. Hartford. 1990. Congruence in native and normative
conversations: Status balance in the academic advising session. Language
Learning. 40.467-501.
Bates, E., et al. 1988. From first words to grammar: Individual differences and
dissociable mechanisms. Cambridge: Cambridge University Press.
Beattie, G. W. 1980. The role of language production processes in the organization
of behaviour in face-to-face interaction. In B. Butterworth (ed.) Language
production. Volume 1. London: Academic Press. 69-107.
Bialystok, E. 1983. Some factors in the selection and implementation of
communication strategies. In C. Faerch and G. Kasper (eds.) Strategies in
interlanguage communication. London: Longman. 100-118.
Biber, D. 1988. Variation across speech and writing. Cambridge: Cambridge
University Press.
, S. Conrad and R. Reppen. 1994. Corpus-based approaches to issues in
applied linguistics. Applied Linguistics. 15.169-189.
Bloor, T. and M. Bloor. 1995. The functional analysis of English. London: E.
Arnold.
Bock, J. K. 1987. Coordinating words and syntax in speech plans. In A. Ellis (ed.)
Progress in the psychology of language. Volume 3. London: L. Erlbaum.
337-387.
Bolinger, D. 1975. Meaning and memory. Forum Linguisticum. 1.2-14.
de Bot, K. 1992. A bilingual production model. Applied Linguistics. 13.1-24.
38 MARTIN BYGATE

de Bot, K. and R. Schreuder. 1993. Word production and the bilingual lexicon. In
R. Schreuder and B. Weltens (eds.) The bilingual lexicon. Amsterdam: J.
Benjamins. 191-214.
Bremer, K., C. Roberts, M. Vassear, M. Simonot and P. Broeder. 1996.
Achieving understanding: Discourse in intercultural encounters. London:
Longman.
Brown, G. and G. Yule. 1983. Teaching the spoken language. Cambridge:
Cambridge University Press.
Bruner, J. S. 1983. Child's talk. Cambridge: Cambridge University Press.
Bygate, M. 1988. Units of interaction and acquisition in small group work. Applied
Linguistics. 9.59-82.
Carter, R. and M. McCarthy. 1995. Grammar and spoken language. Applied
Linguistics. 16.141-158.
1997. Exploring spoken English. Cambridge: Cambridge
University Press.
Chafe, W. L. 1985. Linguistic differences produced by differences between
speaking and writing. In D. R. Olson, N. Torrance and A. Hildyard (eds.)
Literacy, language and learning: The nature and consequences of reading
and writing. Cambridge: Cambridge University Press. 105-123.
Coulmas, F. (ed.) 1981. Conversational routines explorations in standardized
communication situations and prepatterned speech. The Hague: Mouton.
Coupland, J., J. D. Robinson and N. Coupland. 1994. Frame negotiation in
doctor-elderly patient consultations. Discourse and Society. 5.89-124.
Crookes, G. 1986. Task classification: A cross-disciplinary review. Honolulu, HI:
Center for Second Language Classroom Research. Social Science Research
Institute, University of Hawaii at Manoa. [Technical Report No. 4.]
1989. Planning and interlanguage variation. Studies in Second
Language Acquisition. 11.367-383
1991. Second language speech production research. Studies in Second
Language Acquisition. 13.113-132
Dornyei, Z. and M. L. Scott. 1997. Communication strategies in a second
language: Definitions and taxonomies. Language Learning. 47.173-210.
Eggins, S. and D. Slade. 1997. Analysing casual conversation. London: Cassell.
Ellis, R. 1987. Interlanguage variability in narrative discourse: Style shifting in the
use of the past tense. Studies in Second Language Acquisition. 9.1-20.
Faerch, C. and G. Kasper. 1983. Plans and strategies in foreign language
communication. In C. Faerch and G. Kasper (eds.) Strategies in
interlanguage communication. London: Longman. 20-60.
Foster, P. and P. Skehan. 1996. The influence of planning and task-type on second
language performance. Studies in Second Language Acquisition.
18.299-324.
Garman, M. 1990. Psycholinguistics. Cambridge: Cambridge University Press.
Garrett, M. F. 1980. Levels of processing in sentence production. In B.
Butterworth (ed.) Language production. Volume 1. London: Academic
Press. 155-176.
 THEORETICAL PERSPECTIVES ON SPEAKING 39

Gathercole, S. E. and A. D. Baddeley. 1993. Working memory and language.

Hove: L. Erlbaum.
Giles, H., N. Coupland and J. Coupland. 1991. Accommodation theory:
Communication, context, and consequence. In H. Giles, N. Coupland and
J. Coupland (eds.) Contexts of communication. Cambridge: Cambridge
University Press, and Paris: Editions de la Maison des Sciences de
1'Homme. 1-68.
Griffiths, R. 1991. Pausological research in an L2 context: A rationale, and review
of selected studies. Applied Linguistics. 12.345-364.
Gumperz, J. 1982. Discourse strategies. Cambridge: Cambridge University Press.
Halliday, M. A. K. 1985. An introduction to functional grammar. London: E.
Arnold.
Hargie, O. 1986. Communication as skilled behaviour. In O. Hargie (ed.) A
handbook of communication skills. Beckenham: Croom Helm. 7-21.
and P. Marshall. 1986. Interpersonal communication: A theoretical
framework. In O. Hargie (ed.) A handbook of communication skills.
Beckenham: Croom Helm. 22-56.
Hatch, E. 1992. Discourse and language education. Cambridge: Cambridge
University Press.
Hildyard, A. and S. Hidi. 1985. Oral-written differences in the production and
recall of narratives. In D. R. Olson, N. Torrance and A. Hildyard (eds.)
Literacy, language and learning: The nature and consequences of reading
and writing. Cambridge: Cambridge University Press. 285-306.
Holmes, J. 1995. Women, men and politeness. London: Longman.
Hulstijn, J. 1990. A comparison between the information-processing and the
Analysis/Control approaches to language learning. Applied Linguistics.
11.30-45.
Johnson, K. 1996. Language teaching and skill learning. Oxford: Blackwell.
Kasper, G. (ed.) 1996. The development ofpragmatic competence. [Special issue
of Studies in Second Language Acquisition. 18.2.]
and M. Dahl. 1991. Research methods in interlanguage pragmatics. Studies
in Second Language Acquisition. 13.215-247.
Laver, J. D. M. 1970. The production of speech. In J. Lyons (ed.) New horizons
in linguistics. Harmondsworth: Penguin. 53-75.
Lennon, P. 1990. Investigating fluency in EFL: A quantitative approach.
Language Learning. 40.387-417.
Levelt, W. J. M. 1978. Skill theory and language teaching. Studies in Second
Language Acquisition. 1.53-70.
1989. Speaking: From intention to articulation. Cambridge, MA:
MIT Press.
Locke, J. 1993. The child's path to spoken language. Cambridge, MA: Harvard
University Press.
Madden, C. and C. Myers (eds.) 1994. Discourse and performance of
international teaching assistants. Alexandria.VA: TESOL.
McLaughlin, B. 1990. Restructuring. Applied Linguistics. 11.129-158.
40 MARTIN BYGATE

McNamara, T. 1996. Measuring language performance. London: Longman.

Miller, G. A., E. Galanter and K. H. Pribram. 1960. Plans and the structure of
behaviour. New York: Holt, Rinehart and Winston.
Mitchell, R. and C. Martin. 1997. Rote learning, creativity and 'understanding' in
classroom foreign language teaching. Language Teaching Research.
1.1.1-27.
Nattinger, J. and J. DeCarrico. 1992. Lexical phrases and language teaching.
Oxford: Oxford University Press.
Ochs, E. and B. B. Schieffelin. 1983. Acquiring conversational competence.
London: Routledge & Kegan Paul.
Pawley, A. and F. H. Syder. 1983. Two puzzles for linguistic theory: Nativelike
selection and nativelike fluency. In J. C. Richards and R. W. Schmidt
(eds.) Language and communication. London: Longman. 191-227.
Peters, A. 1983. Units of language acquisition. Cambridge: Cambridge University
Press.
Peirce, B. N. 1995. Social identity, investment and language learning. TESOL
Quarterly. 29.9-31.
Pica, T., R. Kanagy and J. Falodun. 1993. Choosing and using communicative
tasks for second language instruction. In G. Crookes and S. Gass (eds.)
Tasks and language learning: Integrating theory and practice. Clevedon,
Avon: Multilingual Matters. 9-34.
Poulisse, N. and T. Bongaerts. 1994. First language use in second language
production. Applied Linguistics. 15.36-57.
Ranney, S. 1992. Learning a new script: An exploration of sociolinguistic
competence. Applied Linguistics. 13.25-50.
Ross, S. and R. Berwick. 1992. The discourse of accommodation in oral
proficiency interviews. Studies in Second Language Acquisition.
14.159-176.
Schleppegrell, M. J. 1996. Conjunction in spoken English and ESL writing.
Applied Linguistics. 17.271 -285.
Schreuder, R. and G. B. Flores D'Arcais. 1992. Psycholinguistic issues in the
lexical representation of meaning. In W. Marslen-Wilson (ed.) Lexical
representations and process. Cambridge, MA: MIT Press. 409-436.
and B. Weltens (eds.) 1993. The bilingual lexicon. Amsterdam: John
Benjamins.
Selinker, L. and D. Douglas. 1985. Wrestling with 'context' in interlanguage
theory. Applied Linguistics. 6.190-204.
Shohamy, E. 1994. The validity of direct versus semi-direct oral tests. Language
Testing. 11.99-124.
Skehan, P. 1986. Cluster analysis and the identification of learner types. In V.
Cook (ed.) Experimental approaches to second langage acquisition.
Oxford: Pergamon.
1996. A framework for the implementation of task-based instruction.
Applied Linguistics. 17.38-62.
 THEORETICAL PERSPECTIVES ON SPEAKING 41

Snow, C. 1987. Beyond conversation: Second language learners' acquisition of

description and explanation. In J. Lantolf and A. Labarca (eds.) Research
in second language learning: Focus on the classroom. Norwood, NJ:
Ablex. 3-16.
Stenstrom, A-B. 1994. An introduction to spoken interaction. London: Longman.
Swain, M. 1985. Communicative competence: Some roles of comprehensible input
and comprehensible output in its development. In S. M. Gass and C. G.
Madden (eds.) Input in second language acquisition. Rowley, MA:
Newbury House. 235-253.
and S. Lapkin. 1997. Task-based collaborative dialogue: Two
adolescent French immersion students interacting and learning. Paper
presented at Annual Meeting of the American Association for Applied
Linguistics. Orlando, Florida, March 1997.
Tannen, D. 1982. Oral and literate strategies in spoken and written narratives.
Language. 58.1-21.
1984. Conversational style: Analyzing talk among friends. Norwood,
NJ: Ablex.
1989. Talking voices. Cambridge: Cambridge University Press.
1990. You just don't understand. Women and men in conversation.
New York: William Morrow & Co Inc.
Tarone, E. 1987. Variation in interlanguage. London: Arnold.
and M. Swain. 1995. A sociolinguistic perspective on second-language
use in immersion classrooms. Modern Language Journal. 79.166-178.
Thompson, S. 1994. Aspects of cohesion in monologue. Applied Linguistics.
15.58-75.
Towell, R., R. Hawkins and N. Bazergui. 1996. The development of fluency in
advanced learners of French. Applied Linguistics. 17.84-119.
Tsui, A. B. 1994. English conversation. Oxford: Oxford University Press.
Tyler, A. 1992. Discourse structure and the perception of incoherence in
international teaching assistants' spoken English. TESOL Quarterly.
26.713-730
1994. The role of syntactic structure in discourse structure: Signaling
logical and prominence relations. Applied Linguistics. 15.243-262.
Weinert, R. 1995. The role of formulaic language in second language acquisition:
A review. Applied Linguistics. 16.180-205.
Willing, K. 1987. Learning styles and adult migrant education. Adelaide: National
Curriculum Resource Centre.
Wodak, R. 1996. Disorders of discourse. London: Longman.
Wong-Fillmore, L. 1979. Individual differences in second language acquisition. In
C. J. Fillmore, D. Kempler and W. Wang (eds.) Individual differences in
language ability and language behaviour. New York: Academic Press.
1982. Instructional language as linguistic input: Second
language learning in classrooms. In L. Wilkinson (ed.) Communicating in
the classroom. New York: Academic Press.
42 MARTIN BYGATE

Wray, A. 1992. The focusing hypothesis: The theory of left-language re-examined.

Amsterdam: John Benjamins.
Young, R. 1995. Conversational styles in language proficiency interviews.
Language Learning. 45.3-42.
and M. Milanovic. 1992. Discourse variation in oral proficiency
interviews. Studies in Second Language Acquisition. 14.403-424.
Yule, G. 1996. Pragmatics. Oxford: Oxford University Press.
and D. Macdonald. 1990. Resolving referential conflicts in L2
interaction: The effect of proficiency and interactive role. Language
Learning. 40.539-556.
and E. Tarone. 1991. The other side of the page: Integrating the study of
communication strategies and negotiated input. In R. Phillipson, et al.
(eds.) Foreign/second language pedagogy research. Clevedon, Avon:
Multilingual Matters.
Zuengler, J. 1991. Accommodation in native-nonnative interactions: Going beyond
the 'what' to the 'why' in second-language research. In H. Giles, J.
Coupland and N. Coupland (eds.) Contexts of accommodation.
Cambridge: Cambridge University Press and Paris: Maison des Sciences
de l'Homme. 223-244
and B. Bent. 1991. Relative knowledge of content domain: An
influence on native-non-native conversations. Applied Linguistics.
12.397-415.