Vous êtes sur la page 1sur 133

Tasks, talk and teaching:

Task-based language learning and the

negotiation of meaning in oral interaction

Michael Courtney
RESEARCH REPORTS

General Editor: Gregory James

VOLUME ONE

Tasks, talk and teaching:

Task-based language learning and the

negotiation of meaning in oral interaction

Michael Courtney

LANGUAGE CENTRE
The Hong Kong University of Science and Technology
This report is a shortened, edited version of the authors thesis, Input, output and interaction.
Oral task design, performance and evaluation, for which he was awarded the degree of
PhD at the City University of Hong Kong, 1998.

Language Centre
The Hong Kong University of Science and Technology
Copyright July 2001. All rights reserved.
The extract from Happy Talk (from South Pacific), by Richard Rodgers and Oscar Hammerstein II,
Williamson Music International, is reproduced with permission.
ISBN 962-7607-14-2

Postal Address: Language Centre,


The Hong Kong University of Science and Technology,
Clear Water Bay,
Kowloon, Hong Kong SAR, CHINA

Telephone: (852) 2358 7880

Facsimile: (852) 2335 0249


Contents

Editorial Foreword ix

Acknowledgements xi

Synopsis xiii

Chapter One: Introduction 1

Research background 1
The research design 4
The quantitative research questions 4

Four definitional problems 4


The problem of task description 4
The problem of negotiated meaning 5
The problem of strategic motivation 6
The problem of task contextualisation 8

Chapter Two: Oral tasks and language learning 9

Pedagogic assumptions 9
Main objections to be addressed 11

Oral tasks and research 13


Oral tasks and acquisition 14
Creative construction 17
An integrated model automatisation and restructuring 18

Input, interaction and output 20


Evidence for the role of interaction in acquisition 22
The input hypothesis 23
The interaction hypothesis 25
The output hypothesis 26
Noticing and modification of output 27
Task performance and deviance 28

Summary 30

Chapter Three: Oral task description 31

Oral task classification 31


Information-gap tasks 32
Simulation tasks 33
Role-plays and scenarios 33
Opinion-gap (discussion) tasks 34

The development of the task typology 34


Task topic 38
Task difficulty and complexity 39
Task authenticity, relatability, memorability and salience 41

Chapter Four: A quantitative comparison of oral task performance 45

Research design 45
Research hypotheses 45

Research definitions 47
Utterances, turns, moves and repairs 47
The information-related turn 49
Turns off task 49
Repairs and perception of task difficulty 49

Task performance coding 50


Coding for negotiated meaning 50

Research variables 53
Output on task 53
Learner perception of difficulty 53
Learner management of interaction 54
Moderator and control variables 54
Triangulation secondary data 54

Research procedure 54
Subject control variables 54
Standardised task implementation controls 55
Transcription procedure 56

Data analysis 56
General procedures and assumptions 56
Analysis of variance and hypothesis testing 57
Analysis of negotiation of meaning 58
Analysis of topic management 58
Sampling 59
Coding reliability 59

Results 60
Hypotheses in relation to learner output 60
Hypotheses related to learner perception of difficulty 61
Hypotheses associated with learner co-operative management of interaction 61
Hypotheses associated with task complexity 61
Task-related output 62
Words per turn and turn lengths 63
Negotiation of meaning 63
Performance differences in relation to task complexity 63
Performance differences in relation to task type 64
Discussion of results 64

vi
Chapter Five: The problem of negotiated meaning 67

Introduction 67
Strategies, moves and speech acts 69
Proficiency and strategy use 71
Evidence for negotiated meaning 73
Interaction and task type 74
Types of objection to the concept of negotiated meaning 75
Does negotiation of meaning exist? 75
Does negotiation of meaning always occur? 76
Is negotiation of meaning useful? 77
Negotiated meaning and motivation 78

Strategic competence 79
Strategic competence and the context of performance 80
Strategy marking in discourse 80

Chapter Six: Research and pedagogic conclusions 83

Oral task types and the negotiation of meaning 83


Functions of negotiated meaning 85
Strategies and negotiated meaning 86
Oral task types and topics 87

Pedagogic recommendations 88
Task-based syllabuses 91
Oral tasks in the Hong Kong secondary school curriculum 94
Oral tasks in the Hong Kong tertiary curriculum 97
Specific-purpose oral tasks 97
Focused oral tasks 98

Oral task evaluation 99


Fluency and accuracy 102

Conclusion 105

References 107

Appendix: Bibliography of published sources of oral interactive tasks 119

vii
viii
Editorial Foreword

The principal canons of Greco-Roman rhetoric heuresis, taxis, lexis, mneme and hypocrisis have
stood the test of time as a taxonomy of the process of communication. The formal study of classical
rhetoric no longer enjoys the prominence it had for centuries in educational curricula, but its precepts
remain deep in the European psyche. The current focus on communication in the range of
applications of this term, including the classical notions of decorum and kairos is a much closer
continuation of the classical norm than is generally appreciated.

people seek the same things from social interaction as public speaking. The orator must gain and hold
the attention of the audience. He does this by attending to both form and content of discourse. Social
communication is an exchange of short speeches. Even the ordinary oral discourse of mundane social
relationships contains the potential for inartistic rhetoric. (Phillips 1991:6874)

In the present study, Michael Courtney offers a principled analysis of the effects of a range of tasks
designed to elicit oral production in English as a second language, and finds a significant cor-relation
between task difficulty and the quality of language produced. With data from a series of experiments
carried out in the Language Centre at UST, his investigation constitutes an important contribution to
the under-researched areas of how particular features of language (e.g. accuracy, fluency,
organisation) alter under different task conditions, the way tasks are developed and classi-fied, how
tasks may influence quantitative assessment scores, and how language output may be influenced by
different levels of task difficulty (cf. Porter 1993). Indeed, it has been shown how only small changes
in the characteristics or conditions of tasks can influence scores for oral per-formance (Wigglesworth
1994, 1997, 2000), and that different factors influence different types of tasks to differing degrees
Henry (1998) has suggested, for example, that variability in a persons oral performance may even
be affected by his/her interlocutors language ability, but the findings are inconclusive.

This issue has crucial implications for the general Hong Kong educational context, where oral
proficiency constitutes a significant component of public language examinations at secondary level,
and is included in many tertiary language assessment schemes. The import of Courtneys findings is
that candidates grades for oral performance may be less a reflection of their real proficiency, more
a function of the tasks they are called upon to accomplish for the purposes of evaluation. Whilst it is
likely that some features of language, e.g. pronunciation, are more generalisable across different
tasks than others, performance in, for example, the use of vocabulary, may vary greatly depending on
the task (James 1989; Huhta 1996). Great faith is put in grades, but the generalisability of universal
oral grades awarded on the basis of performance on one task, to other tasks and contexts is, by
Courtneys study, called into doubt.

Courtney concludes that the design of tasks as assessment instruments should include a precise
specification of the task and the roles of the interlocutors. What is needed is a calibration of tasks
according to difficulty and expected outcome, which have been extensively trialled on a range of
learners with a range of interlocutors.

Insofar as formal assessment is concerned, this argues strongly for a portfolio containing a
progressive record of a persons language skills. One initiative in this direction, that launched by the
Council of Europe for the 2001 European Year of Languages, is the European Language Portfolio,
a personal document that can be presented to teachers, the authorities and employers in a transparent
and transnational manner so as to promote educational and occupational mobility
(press.coe.int/cp/2001/50a(2001).htm), to support and give recognition to language learning and
intercultural experience at all levels (culture.coe.int/lang). The portfolio will comprise a Language

ix
Passport, indicating the languages the holder has studied, the level attained in each language, the
holders intercultural experiences and, where appropriate, qualifications achieved; a Language
Bio-graphy consisting of a detailed description of the holders oral and written skills, as well as
practical language experience gained through visits to other countries; and a file containing personal
in-formation deemed useful by the holder, such as a (multimedia) compendium illustrating skills in
the languages recorded in the Language Passport and Language Biography, and documentary
evidence of classes attended and qualifications gained.

Such an instrument has been proposed for Hong Kong, initially at tertiary level, but at the time of
writing, the report of the inter-institutional Feasibility Study has not been published. Should the
Study recommend the introduction of a pan-city portfolio approach to tertiary language assessment,
the need will be to develop banks of tasks, for both oral and written performance, graduated
according to locally agreed benchmark scales not necessarily the European six-band range
(Schneider & North 2000). The design, construction and analysis of such tasks will, as Courtneys
investigation demonstrates, be no easy matter, and will require a considerable commitment of
resources. But, as a compact city, with a population of highly qualified language professionals, both
researchers and practitioners, and access to global communication, Hong Kong has the opportunity
to make a unique, and lasting, contribution in the realm of frameworks of language assessment.
Courtneys research, with its emphasis on oral communication, is a starting point. Much more needs
to be, and can be, done. But there must be a collaborative will to achieve the common objectives.

Happy Talk, keep talkin happy talk,


Talk about tings youd like to do.
You gotta have a dream, if you dont have a dream,
How you gonna have a dream come true?
If you dont talk happy, an you never have a dream,
Den youll never have a dream come true!
Is good idea? You like?
(Bloody Mary, Happy Talk, in South Pacific 1949)

References

Huhta, A. 1996. Variability vs generalizability of oral proficiency across different test tasks. Language Testing Research
Colloquium, Tampere. [Abstract online.] Available at: www.surrey.ac.uk/ELI/ilta/wip2/html.
James, G. C. A. 1989. Considerations in the design of an oral test in English for Academic Purposes. PhD, University of
Exeter.
Phillips, G. M. 1991. Communication incompetencies. A theory of training oral performance behavior. Carbondale IL:
Southern Illinois University Press.
Porter, D. 1993. Performance conditions: Stability of effects across cultural groups and task types in the assessment of
spoken language. Language Testing Research Colloquium, Cambridge & Arnhem. [Abstract online.] Available at:
www.surrey.ac.uk/ELI/ilta/ltrc93.html.
Rodgers, R. & Hammerstein, O., II. 1949. South Pacific. New York NY: Williamson Music.
Schneider, G. & North, B. 2000. Fremdsprachen knnen was heisst das? Chur/Zrich: Rgger. [Report of a Swiss
National Science Foundation project for the Common European Framework and the European Language Portfolio.]
Wigglesworth G. 1994. The investigation of rater and task variability using multi-faceted measurement. Report for the
National Centre for English Language Teaching and Research, Macquarie University.
______ 1997. Task variation in oral interaction tests: Increasing the reality. Prospect 12, 1, 3549.
______ 2000. Developing, using and evaluating oral communication tasks for the language classroom. Paper presented
at the Applied Linguistics Association of Korea Summer International Conference, Korea University, Seoul, June.

x
Acknowledgements

I should like to acknowledge the very valuable assistance of the following colleagues in this
investigation:

In the Language Centre of the Hong Kong University of Science and Technology (UST), several
colleagues have been especially helpful throughout the course of this investigation. I would like to
thank particularly the Director, Professor Gregory James, and Ada Fan Shui-fun for their
pain-staking and detailed editing of my initial draft for this report. Without their professionalism and
attention to detail, this record of the investigation would not have seen the light of day. During the
research itself, Richard Pemberton, Sarah Carmichael and Elsie Christopher were particularly
help-ful with commentary and advice as the research progressed. I am also indebted to Norma
Pemberton for assistance with task transcriptions and rating checks. I have to thank Dr Man Yu
Wong of the UST Department of Mathematics for essential guidance in relation to the complexities
of multi-variate analysis. At the City University of Hong Kong, Professor John Flowerdew and Dr
Alastair Allen were also generous in their very valuable and detailed commentary throughout the
research.

Colleagues who worked with me on the original Hong Kong Task Oriented Curriculum (TOC)
proposals, particularly Dr John Clark and Dr Keith Johnson, provided useful initial discussion
opportunities on the fundamental concept of task and task-based curricula in the early stages of my
study. At the Hong Kong Examinations Authority (HKEA), John Fullilove and Graham Kennedy
provided me with a very valuable opportunity to work on the design and assessment of oral tasks for
the HKEA Advanced Level Use of English oral examinations.

Form 6 and 7 English Department staff at selected Hong Kong secondary schools also provided
opportunities for me to trial these tasks and collect valuable feedback on classroom and syllabus
constraints.

Finally, I would like to acknowledge important debts to all the UST students who took part in the
intact grouping research and, of course, to my family, whose forbearance during the long process of
this investigation ensured that the research, as recorded here, was successfully completed.

Michael Courtney
Hong Kong
June 2001

xi
xii
Synopsis

Undergraduates in Hong Kong universities are taught in an English-medium environment but come
from an educational background where classroom opportunities to practise speaking in English are
limited. Stimulating and motivating oral materials are an essential tool to help students to practise in
the classroom in meaningful ways. Interactive peer-group oral tasks not only provide a practical basis
for oral development within a language syllabus, but they can also promote the essential skill of
critical thinking. A theoretical justification for their use can be derived from second language
acquisition research but the best justification is from learners themselves.

This report explores the key theoretical issues surrounding the use of such pedagogy. The main issue
is so-called negotiated meaning, specifically in relation to such classroom tasks. However,
exploration of this theoretical area inevitably takes us into a general discussion of the role of
interaction in discourse analysis particularly in relation to developments in the fairly recent field of
conversation analysis. Previous studies have made strong claims in relation to the potential of
interactive communication tasks for language acquisition. However, researchers have based their
claims on small samples and limited descriptions of task design and task performance. The present
study used a much larger intact sample and a more detailed and inclusive task taxonomy.

The first part of the investigation establishes a method for classifying the design of commonly used
oral interactive tasks (the input). A method is developed to quantify the performance (the output) of
such tasks statistically. Finally, the central performance activity of negotiated meaning is discussed
in some detail in relation to task participant interaction and the theoretical question of language
acquisition.

The investigation concludes that statistically significant links between input and output were evident,
thereby offering a principled basis for the choice and use of oral interactive tasks for both pedagogic
and evaluation purposes.

xiii
xiv
Chapter One: Introduction

Research background

There are two main approaches currently employed in the teaching of oral skills in the language
classroom (Richards 1990: 76). One approach is characterised by the conversation class, where
primarily social interaction skills are practised, and the second is through the use of a variety of
communicative or interactive oral tasks (hereafter referred to as oral tasks) which provide learners
with an opportunity for oral practice with other learners. This approach is potentially useful in
teaching situations such as those encountered in Hong Kong, where large classes, and lack of oral
English practice opportunities outside the classroom, create particular problems for learners.

Oral tasks are deliberately designed to involve learners in a co-operative process of information
sharing and negotiation in relation to task procedure and task goals and their value for language
development has been widely advocated (see Lynch & Anderson 1992; Pica et al. 1990). This study
is about the design of oral tasks and the relationship of task design input to learner performance on
task the output of the task. Through a comprehensive quantitative investigation, it attempts to show
that it is possible to establish both a detailed descriptive framework of oral task design and a detailed
description of oral task performances resulting from the use of specific task types. Such descriptions
are essential in order accurately to assess the pedagogic value of oral tasks in relation to language
development requirements, and to assess the value of different oral task types for the purposes of oral
evaluation. As Dornyei & Thurell (1994: 40) have stated, one of the main problems with task-based
oral work in the classroom is that teachers can only make a random, intuition-based selection of
general communicative activities in the uncertain expectation that certain task types may produce
certain types of performances. Is it, in fact, possible for language teachers and language testers to
design or select specific oral tasks for specific output purposes? The fact that oral examination task
designers already assume a reliable correlation between task type and task output makes research in
this area all the more important.

Oral tasks in communicative language curricula are not only discrete activities within a wider
pedagogical framework but also an embedded part of that framework itself. Their suitability and
effectiveness in language enhancement programmes in Hong Kong is a matter of conjecture,
particularly since there has been a growing interest in the notion of task-based approaches to
language teaching (see Clark et al. 1991) and there is a growing body of international research in
relation to oral task usage in particular. However, there are very few detailed studies of oral task
design and usage in the Hong Kong language teaching context, and none that classifies and use a
range of different oral task types.

This study uses the dyadic oral task as a basis for task performance investigation, where dyad refers
to two learners of English, although some of the oral task types investigated can be performed by
groups of three or four learners. In a dyadic oral task, typically, one learner attempts to work with
another to follow a set of task instructions. An example of a typical published dyadic discussion task
outline is given in Figure 1 overleaf:

1
You are a member of a committee in a university department which is responsible for making decisions about
departmental applications for travel grants. Each committee member will be given information about four
applicants in the form of a brief reference for each applicant, written by their tutor or supervisor. Read the
references carefully and select what you think are the relevant facts. When you have sufficient information on
the applicants, the next step is to decide which of them should be awarded the travel grant.

Figure 1: A dyadic discussion task (adapted from Lynch & Anderson 1992).

This study uses established techniques in discourse analysis to investigate oral tasks, and owes much
in a very general sense to the work of British philosophers such as Austin, Searle and Grice who were
influential in the promotion of the study of language as social action through speech-act theory,
conversational logic and pragmatics the study of meaning in context. The field of con-versational
analysis has also been greatly influenced by the functional approach to language of Halliday, who
views language basically as a social phenomenon with three basic functions ideational,
interpersonal, and textual. Hallidays discourse model emphasises the social functions of language
and the thematic and informational structure of speech and writing. Halliday also relates grammar at
the clause and sentence level to situational constraints, referred to as field (purpose of
communication), tenor (relationships among participants) and mode (channels of communication).

There is also a wealth of relevant research from sociological studies in the ethnomethodological and
symbolic interactionist traditions. In both of these approaches to the analysis of society, the
exam-ination of language is seen to be essentially the same as the examination of social organisation.
A major strand in symbolic interactionist work has been concerned with the development of the self,
particularly through language exchange in conversation, where negotiation of the exchange is seen
as central to the moves of participants (Goffman 1976). Goffmans work has been seminal in many
respects, but he has been criticised for his lack of methodology and reference to a corpus of data
(Atkinson 1996) and his characteristic stylistics and use of metaphor and irony. His use of a theatrical
metaphor to explain social interaction has been criticised for being a purely textual repre-sentation
of the real world (ibid.: 108). Work in America by Gumperz and Hymes has helped relate
conversational analysis to the behaviour of participants in conversations with respect to recurring
patterns of speech over a wide range of natural data. The work of Goffman, Sudnow, Sacks,
Schegloff and Jefferson has also been important in the study of conversational rules, turn-taking and
other features of spoken interaction. These ethnomethodologists see the exchange of talk as the
central activity in the analysis of social organisation. Such conversational analysis allows us to
examine conversation to see how people make themselves understood in turn-taking procedures.
Sinclair & Coulthards (1975) work on conversational analysis, particularly spoken interaction in
school classrooms, based on a rank scale of units of discourse from larger stretches of talk
(transac-tions) to individual acts of speech (classes of act) has been influential in promoting
descriptive re-search into patterns of talk in pedagogic situations. The SinclairCoulthard discourse
model views the exchange as the minimal unit of interaction and attempts to establish units and rules
for defining oral interaction through participant moves on the basis of this speech act theory (ibid.:
257).

The field of pragmatics has attempted to link language-related behaviour to the problem of how
listeners uncover speakers intentions. This should not be seen as a purely receptive, but as a reflexive,
process. Collaborative theorists stress the active search for common ground (the grounding process)
which participants in a conversation are searching for. This grounding process involves the gradual
establishment of a mutual set of beliefs about the meaning of the interaction (Wilkes-Gibbs 1997).

2
Moving from description to hypothesis about the relationship between classroom interaction and
language acquisition, the claim that the use of oral tasks constitutes effective classroom pedagogy is
supported by research into the interactive processes stimulated by such tasks (cf. Hatch 1978b; Swain
1985; White 1987;Yule & Macdonald 1990; Brown 1994). This type of research has largely centred
on the processes whereby interlocutors reach and agree on mutually acceptable understandings in the
context of task-based solutions, a process termed the negotiation of meaning, which has been seen
as central to task-based interaction (Swain 1985: 248) in much the same way that social
interactionists have viewed the exchange as central to all human organisation. Such negotiation
between learners is hypothesised as taking place during the course of their interaction when either
participant in the task signals with questions or comments that the others message has not been
successfully conveyed. This may then set off a response aimed at modifying the message and
attempting to make it more comprehensible. This process has been seen as conducive to language
acquisition itself (see Hatch 1978b; Long 1981) although the earlier strong position that negotiated
meaning was a sufficient condition for acquisition has largely been abandoned for the weaker
position that it is necessary but not sufficient (see Yule & Tarone 1991).
The output from such tasks is arguably closer to a genuine conversational form than other types of
classroom language output, such as teacherstudent questioning, since there is a high degree of
unpredictability and output contingency between turns as task participants attempt to understand task
input to make it more accessible or more comprehensible.

An important question about task design itself follows from these considerations, one that centres on
the issue of the role of task input, which learners can then make comprehensible or accessible during
the task performance, and its relation to eventual language acquisition. What types of classroom oral
tasks and patterns of interaction provide learners with the greatest amount of comprehensible input?
(Nunan 1991: 50). For example, Long (1989) stated that jigsaw tasks are more effective oral
communication tasks in this respect than other types of information exchange tasks because they
appear to stimulate more negotiation of meaning during task interaction. However, there are
conflicting claims in relation to the effectiveness of different task types, claims which ultimately
depend not only on a reliable description of significant aspects of the process of participant
interaction on task, but also on what is meant by effective or successful in the context of task
performance.

A basic description of oral tasks should attempt to include task input, task interaction and task output,
and the relationship between these concepts, where input initially refers to the task instructions,
although we might also want to include the subsequent output from task participants as another form
of input in itself. This approach should not be taken to be adherence to a mechanistic data-processing
model of task performance, but should be seen as a basic organisational device to facilitate initial
description of the three basic stages of task performance. Output in this context refers to the
turn-based structuring of a task-based dialogue in response to task input, which is commonly verbal,
textual, audio-visual or graphical. In order to consider fully the relationship between task type and
task performance, it is also necessary to give the performers themselves a central position in the
analysis, since their motivation for the task and their perception of the task completion process within
the changing social context of the performance can be hypothesised to have a central influence on
task outcomes.

3
The research design

My original study was in two parts, in what has been termed a bi-methodological research design
(Lazaraton 1995). Part One, that reported here, is an experimental design, consisting of a quantitative
study of task performance in relation to the development of a typology of task types for an English for
Academic Purposes (EAP) teaching domain in Hong Kong. Following an overview of four main
descriptive problems deriving from a consideration of the concepts of task input, output and
interaction, a review of the literature relevant to the quantitative investigation is presented. The
establishment of the research task typology is a central concern for the quantitative investigation, and
is based on a consideration of previous task typologies, published oral task examples and the views
of a sample of regional language teachers.

Part Two, which is not included in this Report, a qualitative investigation of task performance in
relation to selected case studies of undergraduates in a tertiary domain in Hong Kong, was
undertaken as a perspective on the whole question of oral task interaction and the use of specific
task-completion strategies by task participants. It comprised a qualitative investigation of ten case
studies, each of two subjects who performed the range of research task types.

The quantitative research questions

The following research questions concerning the relationship between task type and task
per-formance were investigated in the quantitative study:

1. For oral task performances in a specific research domain, are there significant differences
between learner output and task type and complexity for the range of tasks in the typol-ogy?
2. Is there any manifest evidence from oral task performances to suggest that learners regard
some tasks as more difficult than others?
3. How do learners attempt to cope with a perception of task difficulty during task
per-formance?

Four definitional problems

The study investigates these two sets of questions and four related definitional problems. An
understanding of the issues surrounding these definitional problems is fundamental for an
under-standing of discussion and investigation in relation to questions concerning the pedagogic
effects of oral tasks. These issues are primarily concerned with problems of adequate description and
with theories of causation in relation to task-based interaction and language acquisition.

The problem of task description

The word task is now used so widely in language teaching materials that there is a danger of it being
used as a label for every language teaching activity. Indeed, we may already have reached this
position. A major review of research into teacher planning and decision-making (Shavelson & Stern
1981) concluded that the instructional activity generally referred to as a task was, in fact, perceived
by teachers to be the basic unit of all lesson planning. In the Hong Kong context, Clark et al. (1991)
define task as the smallest organised wholes or unities of learning. However, the general currency of
the term neither derives from, nor has resulted in, a widely accepted definition.

4
The starting point for any investigation into the effects of task types must therefore be the
establish-ment of an oral task typology which classifies tasks currently in use, and justifies their
distinction as discrete types. The typology must also supply an adequate descriptive base, the lack of
which has resulted in superficial descriptions of tasks in many previous studies (Donmoyer 1990:
181). This description has usually been in terms of single features of task design such as convergent
or divergent, categories which refer only to a particular task goal requirement to reach a consensus
or not. Both the construction of the research task typology for this study investigation and the
decision concerning the amount of data to collect in relation to the typology were influenced by
Duffs (1993: 5986) recommendations which relate to previous research in this area:

Researchers need to gather data on a range of tasks before assuming that they have a representative or
adequate sample of interlanguage performance. Research on the effects of task types on interlanguage
development has attracted criticism specifically in terms of the types of tasks employed for elicitation
purposes and the relatively small number of performances observed. Critics have called for
validation of the findings with a wider range of different types of tasks [my italics].

The problem of negotiated meaning

Once a method for oral task classification has been established, it then becomes possible to examine
learner interaction in relation to specific oral task types. The central focus in many studies has been
the concept of negotiated meaning, a type of interaction that has been viewed as a keystone of the
communicative curriculum. Hymes (1972) broadened Chomskys (1965) original conception of
competence as the mental comprehension of a language system, to include an important
sociolin-guistic perspective which paved the way for conceptions of strategic competence, or the
ability to use language effectively and appropriately in a social context. Breen & Candlin (1980)
defined communicative competence in performance terms as the interpretation, expression and
negotiation of intended meanings between speakers. Given this central conceptual position, links
between negotiated meaning and second language acquisition (SLA) have been suggested.

Canale & Swain (1980) hypothesised a typology of competencies to include:

(i) Grammatical competence the mastery of a perception of the language system


(ii) Socio-linguistic competence the mastery of appropriateness
(iii) Discourse competence the mastery of decoding techniques
(iv) Strategic competence the mastery of strategies to communicate effectively and to cope
with communication breakdown.

Here, both discourse competence and strategic competence are conceptualised as facilitating the
process of negotiating task-based meaning. In a later paper, Swain (1985: 248) argued that the
process of negotiating meaning is a necessary precursor to grammatical acquisition:

... it paves the way for future exchanges, where, because the message is understood, the learner is free
to pay attention to form.

Negotiation of meaning between interlocutors has been seen as a central activity in oral task
performance, intimately connected with questions of input and output. However, the concept of
negotiated meaning is by no means clear in relation to the assumed sociolinguistic processes it
encompasses. The term is often taken to mean the linguistic processes whereby two interlocutors

5
appear to co-operate to achieve a comprehension of, and consensus about, questions of task-related
meaning (Breen & Candlin 1980; Long 1983b; Ellis 1984).

Most of the work done on how learners interact has examined the way in which learners appear to use
certain strategies in this process of negotiation particularly confirmation checks, clarification
requests, repetition and restatement to build a common definition of the task and its goals.
Participant motivation is assumed to be co-operative and task-related during this process. However,
there appear to be three main types of objection in relation to the concept of negotiated meaning:

1. The first type of objection acknowledges that negotiated meaning is a possibility in


learnerlearner exchanges but questions whether it always occurs (see Pica 1988).
2. The second type of objection questions its value to the learner if it does occur (see Prabhu
1987).
3. The third type of objection questions the existence of the concept at all (see Aston 1986).

The problem of strategic motivation

In order to present learners with effective learning opportunities, we need to know what types of oral
task learners in a particular domain find motivating and why. Once learners are engaged on a task,
we also need to know something about their motivation for the choice and use of particular
task-related communication strategies what I shall term their strategic motivation in order to
verify or otherwise our assumptions from surface discourse features about the function or functions
such strategies actually perform. In short, to use a sociological distinction, the manifest function
attributed to a discourse feature by a participant (and perhaps the researcher) may not be the latent
function, intended or otherwise, that actually resulted.

Bachman (1990) viewed the ability to use strategies in communication (strategic competence) as an
important part of all communicative language use and conceptualised it as not just the ability to deal
with perceived problems in communication, but as a vital link between competence and
per-formance for all speakers. In a recent reformulation, Bachman & Palmer (1996) present a model
which gives strategic competence a central position, linking topical knowledge, language
knowl-edge and personal characteristics with language tasks and language task settings. Bachman
& Palmer divide general language competence into organisational and pragmatic competence
thereby expanding the role of strategic competence. This competence is seen as central to all
communi-cation and mediates between language competence, background knowledge and context of
situation. It is therefore not just a compensatory skill, employed when problems are perceived. It is
implicated in all performances by all speakers, not just language learners. Strategic competence is in
general terms seen to include:

(i) goal setting formulating communicative intentions


(ii) assessment examining underlying competencies to see if they are adequate or not for
upcoming communicative encounters
(iii) planning organising resources for communication and translating them into action.

This reformulation of the role of strategic competence constitutes a major advance. The earlier
Canale & Swain model was essentially compensatory and therefore had a limited relationship with
performance. However, the Bachman & Palmer model is still not an effective working model of
language use, and is too generalised to account for performance in specific domains. There is little
treatment of underlying processing factors, and without this no theory of transfer from competence to

6
performance is possible (see Skehan 1998).

Canale & Swain (1980) included the concept separately within their framework of communicative
competence and conceptualised it as consisting of a store of verbal and non-verbal communication
strategies that may be used by the learner to compensate when communication breakdowns occur,
due either to performance variables or inadequate communicative competence. As such, it can be
thought of as a reserve of communication strategies (communicative strategies) which can be
consciously employed to repair such breakdowns either by direct attempts to solve the perceived
problem or by the use of avoidance strategies. The term repair was used by Chaudron (1977) in a
comprehensive account of corrective actions taken by students in second language classrooms and
covered a wide range of actions. Repair in my investigation is treated as the generic term, and
attempted correction of errors, for example, regarded as one type of repair. This conception does not
deny a possible wider role for the term, but facilitates a research focus on particular strategies which
appear to deal directly with problem sources in oral task-related discourse.

However, there is little general agreement in relation to the classification of communication


strategies and the concept of communication breakdown and its associated repair. There are currently
two broad approaches to the question of what constitutes a communication strategy, both
conceptually difficult (Kasper & Kellerman 1997a). The present study generally follows the first
approach, which assumes that strategy use is conscious on the part of the learner and that it is
observable in the discourse outcome. These observed forms are divided into two categories
reduction and achievement strategies. Reduction strategies are those which involve the learner in
avoiding, changing or perhaps totally abandoning an exchange when he or she perceives a
communication problem. Achievement or compensatory strategies are those which the learner uses
to try to solve or get round perceived problems.

The second broad approach to the question of communication strategies focuses on the internal,
cognitive processes used by learners and as such tends to focus on linguistic competence rather than
performance. Existing models of communicative competence fail to link competence to
per-formance participant knowledge of a language to the motivation for actual CS employment
adequately since processing arguments would have to account for both participant analysis and
participant control with respect to interlanguage, which also appears to be highly context-specific.

The notion of consciousness is important in this context. What does it mean to say that strategy use
is conscious on the part of the learner? Schmidt (1990, 1994) has argued that the critical notion of
unconscious is inadequately described in previous accounts of language acquisition. He takes the
strong position that noticing is the necessary and sufficient condition for converting input to intake,
and we might therefore conclude that a degree of noticing in relation to personal strategy use might
link such attention with modification of strategies to increase their effectiveness a process we might
also want to link to eventual acquisition in relation to such facets of strategic competence.

The concept of communication breakdown itself has been criticised for being overly mechanistic and
one that ignores the possible phatic value of avoidance strategies (Smith 1970). Learners may simply
elect to ignore the problem, rather than reach a point where their strategic competence is revealed as
inadequate, as suggested in the Bachman & Palmer (1996) model of strategic com-petence through
the feature of assessment. It may therefore be the case that strategic moves by task participants to
overcome perceived communication problems in the performance of the task may have multiple
functions, and also multiple realisations in the task transcript (see Chun et al. 1982).

7
The problem of strategic motivation is therefore not only a definitional problem, but also a problem
of knowing what would constitute evidence for strategy employment and how we might access
confirmation of that evidence in the social context of task performance.

The problem of task contextualisation

The last definitional problem to be addressed here is the important question of the social context of
task performance. Since oral tasks are not performed in social vacuums, we can hypothesise that the
performance setting itself is an important influence on outcomes. Despite this recognition of the
importance of social context, conversation analysis has been criticised for its emphasis on the
mechanics of the exchange with the implication that the sequential position of an utterance is more
important in the determination of meaning. For example, a perception by task participants of unequal
rights and duties in relation to task performance (one participant may be seen by others to be more
of an authority on the issue under discussion) is likely to be an important factor in the resulting
contingency pattern of turns.

This contingent orderliness (Fairclough 1995: 28) in conversation is hypothesised to derive from
task participants conforming to an implicitly understood framework of discoursal rights and
obli-gations, involving turn taking, topic control, rights to question and obligations to answer (see
Sinclair & Coulthard 1975; Goffman 1976).

8
Chapter Two: Oral tasks and language learning

Pedagogic assumptions

There is a sense in which all oral communication involves an information gap of some kind and the
act of communication itself can be basically seen as the attempt to bridge a perceived information gap
(Johnson 1979). A central implicit feature therefore, of all oral tasks, is the information gap task
participants have to share new information in the process of task performance. This feature may be
made explicit in so-called information gap tasks, or it may remain implicit in other oral task types
such as those designed primarily to promote discussion. Are there other basic features which all oral
tasks should possess? Taylor (1983), drawing on Johnson & Morrows (1981) work, listed the
information gap as one of five desirable pedagogic features of effective oral tasks:

1. Information gap: One of the fundamental features of any oral task should be that it requires
learners to share information not previously possessed.

2. Extended discourse: Since sentences, whether spoken or written occur not in isolation but
in the context of a discourse, oral tasks should be designed to give learners experience in
processing language above the sentence level.

3. Uncertainty: Genuine communication is unrehearsed and task design should reflect this.

4. Goal-orientation: Genuine communication involves the perception of goals by


interlocu-tors, so task design should include goals.

5. Real-time processing: Oral tasks should require that learners deal with language
spon-taneously since genuine communication involves the ability to attend to many
linguistic and extra-linguistic factors at once.

If two learners are given an oral task which ostensibly embodies these global features, what is it that
is assumed to happen that makes the exercise valuable in terms of language learning? Why might
such a task be useful in the process of second language acquisition? These are fundamental questions
for oral task design and implementation. Even if our answers to these questions are only in terms of
the increased opportunities such tasks offer for oral practice, we still need some idea of how these
opportunities might be useful to language development since, as Allwright & Bailey (1991: 150)
point out:

It seems clear that making learners actively participate as much as possible cannot be universally
right, if only because not all learners learn best in the same way.

The following summary review of previous studies isolates the major areas of pedagogic assumption
and research interest, beginning with the most common rationale, that of increased opportunity for
oral practice. These points are not in any order of priority.

1. Oral tasks offer more opportunity for learners actually to produce the target language in practice
situations. This provides an alternative to the normal lockstep situation of large language classes
(Hoetker & Ahlbrand 1969; Fanselow 1977). Flanders (1970) found that in a typical lockstep
lesson, the teacher talked for as much as two thirds of the lesson. Long & Porter (1985: 208)
estimated that in an EFL class of 30 students in a public secondary school classroom, each

9
student achieved an average of only one hour per year of actual oral production. Another study
by Long (1983a) revealed that lessons involving learner discussion in pairs promoted a greater
quantity of speech and a greater variety of speech acts and social uses of language than did
teacher-led class discussions.

2. In modifying, linguistically and conceptually, the basic information and rules supplied for the
tasks, learners are increasing the comprehensibility of this task-related input and thereby,
according to the strongest claims (Long 1983c; Pica & Doughty 1985a; Swain 1985;
Car-michael 1996) providing a necessary basis for further language acquisition in relation to the
forms manipulated.

3. Despite possible objections with regard to the use of uncorrected interlanguage forms, the
learner output which results is a possible source of input for other learners which has potential
for eventual language acquisition (Long & Porter 1985; Porter 1986; Slimani 1987).

4. In negotiating the meaning of the basic information and rules for the task, learners are engaged
in interaction which is a more potent source of input for eventual language acquisition than that
from interactions between teacher and learner. It has even been claimed that this input is a richer
source than that resulting from native speakers (Long & Porter 1985) since it is more accessible
to the learner.

5. In order to develop the communicative competence of students, it has been argued (van Lier 1988)
that it is crucial to allow learners to monitor and repair their own performances rather than doing it
for them. This follows the processes of naturally occurring conversation more closely and oral tasks
should attempt to mirror these processes.

6. Learners engaged on such tasks will develop and practise performance skills, such as co-operating
in task completion, negotiating a consensus, evaluating information etc., which form a general core
of genre-related target language communicative skills.

7. When oral tasks are embedded in an ESP programme, learners will be engaged in developing and
practising cognitive and metacognitive target language strategies and skills relevant to a target
discourse community when they attempt to solve the problems posed by such tasks.

8. The resulting communication strategies employed by learners to complete oral tasks are in
themselves teachable devices to improve their communicative competence (particularly with regard
to the strategic competence component of communicative competence).

9. Such tasks offer an environment of reduced communicative stress for learners to practise and
manipulate target forms, when compared to the exposure of native speakernon-native speaker
(NSNNS) interactions (Brown & Yule 1983a; Long & Porter 1985). Long & Porter (op. cit.: 212)
suggested:

Freedom from the requirement of accuracy at all costs and entry into the richer and more
accommodating set of relationships provided by small-group interaction promote a positive affective
climate. This in turn allows for the development of the kind of personalized, creative talk for which most
aural-oral classes are trying to prepare learners.

10. Peer group interaction itself is easiest to achieve when interactants are orientated towards
symmetry (van Lier 1996: 193). Symmetry here refers to the idealised state of an equal distribution

10
of rights and duties in talk (ibid.: 175). In this sense, oral tasks performed by peers of similar social
status and, importantly, similar language proficiency, are more appropriate (and effective)
pedagogical vehicles for language development than interaction between partners (such as teacher
and student) who might perceive their status to be unequal. The contention is that learning which
might result from task interaction does not occur effectively when the learner is in a subordinate role
(Knowles 1975). The hypothesis here is that when such an imbalance in status is perceived,
contingency, or the dependence of utterances on previous utterances, is adversely affected. On the
basis of this argument, oral tasks are thought to represent more effective peda-gogy because of the
opportunity for learner interaction at a similar level of social power.

11. Oral tasks improve the quality of learner production because learners have the following
oppor-tunities which are unique to small group settings:

(i) Group face to face communication is a natural setting for conversation.


(ii) Group work encourages longer turns (rather than the characteristic short turns of class
responses) and more attempts at increased cohesion and coherence.
(iii) Group participants can play roles which require attempts at a greater range of language
functions, conversational management techniques and topic nomination.
(iv) Group situations are the only classroom situations where genuinely communicative practice
can take place (Long & Porter 1985) since they are the only situations where spontaneous
language use is encouraged and the focus is on meaning as well as form.

12. Individual differences in learner background, capabilities and needs are ignored by traditional
lockstep teaching environments. Group work offers opportunities to take many of these individual
aspects into account. Small-group tasks allow students to work simultaneously on material suited to
their needs and capabilities, in a way that offers opportunities for each learner to contribute
individually.

13. Oral tasks are more motivating than other more formal methods of classroom instruction. Learners
often report that they feel less inhibited to speak and make mistakes in a small-group setting than in
a teacher-led class (Littlejohn 1982).

14. The quality of language exposure between peers is high because accessibility and
comprehen-sibility of task input is improved. This leads to a concomitant improvement in output
quality.

15. Peer group interaction integrates the cognitive and social aspects of language. Both Barnes (1976)
and Cazden (1988) based their support for peer group interaction on the premise that oral and
written language use is an act which is rooted in a social context, a perspective which followed the
important shift in discourse analysis in the 1950s and 1960s from mainly structural concerns to a
consideration of the social context of speech.

Main objections to be addressed

There have been many counter-arguments in relation to both research and pedagogy, which either
point to assumed misconceptions or assumed inadequacies in the arguments summarised above.
These objections are outlined below.

1. Van Lier (1996) objected to the research term input for two main reasons. Firstly, many input

11
studies have neglected the active role of the learner in terms of strategy choice in relation to task
input. Secondly, there is a danger that the quality of the exposure is often overshadowed by
concerns about the quantity of the exposure, particularly the quantity of specific items thought to
have a triggering effect on acquisition. The input-output metaphor is therefore seen as
dangerously mechanistic, although van Lier conceded that when the analysis is linked to
interaction, it is a step in the right direction (ibid.: 50). Other critics have pointed to the
conceptual difficulties in establishing discrete definitions of input if it is seen as everything that
can be attended to by the learner (Schmidt 1994).

2. Critics (e.g. Aston 1986) have pointed out that evidence of negotiation of meaning during the
performance of oral communication tasks may indicate activity which is predominantly phatic
rather than task-based, or interactional rather than transactional, to use Brown & Yules (1983a)
distinction. For example, apparent usage of a strategy such as a comprehension check, where an
interlocutor attempts to check understanding of the content of a previous turn, may indicate the
desire to relinquish the present turn (for personal reasons) rather than a co-operative attempt to
establish more comprehensible input. These critics acknowledge the potential value of such
exchanges, but object to the apparent dismissal or avoidance of phatic concerns.

3. A related objection concerns the possibility of multiple functions in relation to observed strategy
usage, and the difficulty for an observer to establish the motivation for such strategy usage. For
example, hesitation and pausing, which might be taken as an indication of a problem source, is a
natural, although largely unexplained, phenomenon in native speaker speech, and may in fact
indicate the use of planning strategies by a more competent speaker (see Fulcher 1996).

4. There is a pedagogic concern in relation to the possible swapping, mutual endorsement and
fossilisation of interlanguage errors in oral task performance, where corrective intervention by a
teacher is not always possible. This was why, in the Bangalore project (Prabhu 1987), such
inter-language practice was discouraged. It is assumed that the resulting limited interlanguage
exchange is one of deviant forms with the danger that the models presented will be subject to
fossilisation.

5. Further, there is a concern that oral tasks are only performance tasks they do nothing for a learners
competence (Prabhu 1987). This is associated with a general concern that the assumed linkage
between task performance and language acquisition is inferential only. It is argued that many widely
used information-gap tasks would not be suitable as tools to develop a learners competence since
they lack specificity in terms of the linguistic focus of instruction. Doyle (1983) hypothesised that
task participants learn from a task whatever they are led to do in that task, and that the connection
between grammatical structure and task performance requirements has been either nonexistent or
loose.

6. Some (e.g. Swales 1990) have suggested that since oral tasks theoretically have a multiplicity of
possible outcomes, then the notion that particular types of tasks will consistently produce parti-cular
types of outcome is an essentially false one that stems from a common expectation that teachers
themselves should be in control (Murphy 1993). However, it is not clear whether objec-tions in this
respect relate to empirical observations that tasks actually do have multiple outcomes, or whether
the objection is essentially a value judgement that we should do nothing to encourage predictability
through educational engineering (Newmark 1971: 11).
7. Another related objection is in terms of the variability of the interlanguage produced although this
objection still relates to the central concern of control of outcomes:

12
It is hard to see how we can manipulate the input to the speech production mechanisms ... We can, for
example, ask our subjects to describe pictures or the actions depicted in little movies and thus gain
some control over what our subjects will talk about. But we cannot manipulate other critical variables,
like what syntactic form our subjects sentences will take and what words they will choose to put in
their sentences. (Matthei & Roeper 1983: 162)

The objections above can be countered, but they do warrant serious consideration. They are substantial
objections to very different aspects of oral task pedagogy and research. Many of the objections are
fundamentally definitional. There is also an evident accuracyfluency tension between those
ad-vocating attention to grammatical features, and those favouring the development of more affective
learning environments leading to improved learner strategic competence.

The definition of task type is crucial here, and the danger previously referred to of treating all such tasks
equally for the purposes of discussion is evident. For example, using a very basic categorisation, we
should expect a difference in outcome between a task which is designed to facilitate practice of specific
structural features, and one that is primarily designed to stimulate interaction and negotiation between
task participants. However, we might also expect both of these task types to show a degree of
predictability in their respective outcomes in certain pedagogic domains. One function of task-based
research might be to establish the degree of predictability expected from different task types.

Oral tasks and research

Applied linguistics research into oral language development in relation to groupwork has been in several
distinct traditions, and many of these approaches have been narrow and exclusive (see Yule & Tarone
1991). This is possibly for reasons associated with the design complexities of more holistic
investigations (Courtney 1996a) or perhaps as a result of deterministic beliefs in linkages between
certain types of communicative activity, for example, negotiated meaning, and second language
acquisition. There has also been a considerable amount of group-work research deriving from
psychological and managerial concerns with group decision-making processes. Some of this work is
relevant to the issues of task performance here, particularly McGraths (1984) performance typology
based on the processes of generating ideas, choosing a correct answer, negotiating between interests and
executing decisions (see also Thompson 1967; Telfer 1993). In terms of a focus on oral language
processes, there are two main research perspectives which have attempted to supply answers to the
question of how group work might be linked to language development and acquisition:

1. Investigating what learners do on task. The input/interaction/output approach, by far the most
extensive in terms of published research, focuses on the concept of negotiated meaning and its
possible links to second language acquisition.

2. Investigating why learners do what they do on task. The language socialisation view, which
focuses on learner motivation and the learners developing sense of, and definition of, the task
and its linguistic requirements in a social context.

It seems reasonable to assume that these two approaches are not mutually exclusive and that research
designs investigating oral task performance should also include what Yule & Tarone (1991: 162) have
termed the other side of the page the perspective of the task participants themselves.

Free production oral tasks which are open-ended in terms of task goals, have traditionally been the

13
primary source of data in longitudinal studies of grammatical development (Duff 1993: 60). However,
there is still controversy concerning the fundamental notion that conversation is the most natural source
of language for research purposes (Atkinson & Heritage 1984; van Lier 1989). Claims (e.g. by Tarone
1988) that conversation offers the most vernacular, systematic style of speech, largely because of the
relative lack of monitoring involved, have justified studies such as McNeills (1990), which employed a
spontaneous conversation task for error analysis purposes.

An example of a more structured approach to research task design in a longitudinal study is provided by
Duff (1993) in her study of the interlanguage development of JDB, a 28-year-old Cambodian stu-dent.
A possible problem with this type of research, however, is that even if we conceive of a task type which
elicits the kind of spontaneous conversation referred to above, and we attempt to hold that type
constant for the duration of longitudinal research, not only will the relationship between the
researcher and the researched change but, more importantly, the relationship between the sub-ject
and the task will also change, even if task content is renewed. The nature of the research itself then
changes to a qualitative assessment of this changing relationship.

Oral tasks and acquisition

The main theoretical models which have been used to link the use of oral tasks with language
acquisition are referred to in this study as the input hypothesis, the interaction hypothesis and the
output hypothesis. All of these theoretical models (sometimes termed interactionist) seek to
explain how the performance of such tasks might be conducive to the eventual aim of second
language acquisition itself, but each model gives prominence to one particular aspect of the process.
There is an important initial question of adequacy of explanation in this context, which is related to
the search for a general inductive theory of language acquisition. Can we expect task-based
interac-tions implicitly to embody all the processes of language acquisition (assuming these are
finitely definable) or should we view task-based interactions as one helpful part of this process? In
order to consider this question, we need to examine these positions in more detail.

Table 1 relates these hypothesis to specific studies and specific task types. The Table is based on an
original smaller compilation by Ellis (1990: 96), which I have extensively revised and expanded here.
This is not intended to be exhaustive, but to identify research studies which are particularly relevant
to hypotheses concerning the possible links between particular types of task activity or design and
second language acquisition. These hypotheses are outlined in this study as the input, interaction,
output, discourse and collaborative hypotheses and are given in Column 3 of the Table.

Table 1: Research linking oral tasks to acquisition.

Associated task Relevant Associated Activities or Studies (in


types acquisition hypothesis features linked chronological
theory to acquisition order)
structural tasks descriptive frequency frequency of quantitative
occurrence of input correlational
language drills reception based features
Larsen-Freeman
rote exercises behaviourism comprehension of (1976a, 1976b)
input Lightbown (1983)
creative Long & Sato
construction accessing input (1984)
Lightbown (1985)

14
cognitive skill Wells (1985)
learning Bailey (1989)
Pienemann (1989)
Loschky &
Bley-Vroman
(1990)
problem-solving explanatory input accessing input indirect evidence
tasks reception-based
comprehension of Cross (1977)
information-gap creative input Scarcella & Higa
tasks construction (1981)
quantity of input Ellis (1982)
structural tasks cognitive skill quality of input Fillmore (1985)
learning (i+1) Gass & Madden
comprehension of (1985)
input Kelch (1985)
Krashen (1985)
simplified input Long (1981)
Chaudron (1985)
two-way Schachter (1986)
interaction Pica et al. (1987)
Long (1992)
Fotos (1994)
Information-gap explanatory interaction accessing input indirect evidence
tasks reception-based
and interaction Wagner-Gough &
discussion tasks production-based Hatch (1975)
negotiation of Seliger (1977)
simulations creative meaning Snow &
construction Hoefnagel-Hohle
scenarios negative input (1978)
cognitive skill Schwartz (1980)
learning co-operation Johnson (1981)
Long (1981)
unco-operation Scarcella & Higa
(1981)
strategy use Fillmore (1982)
van Lier (1982)
motivation Bialystok (1983)
Long (1983a)
Tong-Fredericks
(1984)
Ellis (1985)
Fillmore (1985)
Gass & Varonis
(1985)
Long & Porter
(1985 )
Pica & Doughty
(1985a, 1985b)
Wells (1985)
Crookes (1986)
Demetras et al.
(1986)
Duff (1986)
Frch & Kasper

15
(1986)
Pica & Long (1986)
Porter (1986)
Rulon & McCreary
(1986)
Ellis & Rathbone
(1987)
Beebe (1988a)
Bohannon &
Stanowicz (1988)
Bygate (1988)
Long (1988)
Samuda & Rounds
(1988)
Pica et al. (1989)
Brown (1991)
Crookes & Schmidt
(1991)
Luk (1992)
Bly (1993)
Ng (1995)
Carmichael (1996)
Pica et al. (1996)
van Lier (1996)

problem-solving explanatory output production indirect evidence


tasks opportunity for
production-based resource Swain (1985)
information-gap employment Schmidt & Frota
tasks creative (1986)
construction negotiation of Pica (1988)
convergent & meaning Anderson et al.
divergent cognitive skill (1993)
discussions learning demand for Green (1995)
accuracy, Swain (1995)
simulations precision and
appropriateness
interaction
negotiation of
meaning
structural tasks explanatory discourse comprehensible indirect evidence
input
role-plays production-based Swain (1985)
genre-based Pavesi (1986)
scenarios creative participation
construction
debates
cognitive skill production
discussion tasks learning opportunities

exposure to formal
planned language
use

information gap explanatory collaborative production indirect evidence

16
discourse opportunities
convergent production-based Wagner-Gough &
discussions scaffolding interaction Hatch (1975)
creative Long & Sato
construction co-operation (1984)
Ellis (1985)
cognitive skill Sato (1986)
learning Hawkins (1988)

divergent explanatory topicalisation learner topic indirect evidence


discussions nomination
reception-based Hatch (1978a)
role-plays learner topic Long (1983c)
production-based control Ellis (1984)
scenarios Wells (1985)
creative production Slimani (1987)
simulations construction Christopher & Ho
learner initiative (1996)
cognitive skill
learning

The interaction hypothesis stresses the importance of the process of interaction itself, whereas the
input and output hypotheses stress the importance of the resources offered to participants by task
in-put and task output respectively. In so doing, these hypotheses ultimately rest on more general
theories as to how languages are acquired, attempting to explain how the learners experiences and
reactions to natural language stimuli are conducive to language learning and acquisition. They do this
by recourse to general theories of language learning which give prominence to the learners active
role the constructivist position or to post-behavioural models which rest on hypotheses
concerning learner progress towards automatic oral responses. Integrated models which involve
conceptions of both automatic and constructed responses are possibly more fruitful compromises,
and many of the cognitive theories such as those outlined by McLaughlin (1987) appear to embody
components of the two assumed basic acquisitional processes of automatisation and creative
con-struction.

Cognitive skill learning theories view language learning as a conscious information-processing


activity by the learner to overcome perceived shortfalls in knowledge. Learning is usually said to
have taken place when the learner has internalised these techniques and is then able to carry out
linguistic operations automatically as a result of practice. However, there is very little research
evidence to support the main rationale of many types of skill-based performance tasks that of the
transfer of internalised genre-related skills from the instructional setting to the target language
environment (Twelker 1971; Crookes & Gass 1993). Such evidence that does exist is inferential in
relation to pre-test, post-test research designs (Courtney 1994).

Creative construction

Negotiation of meaning is a hypothetical process which is ultimately related to natural learning or


creative construction models of language acquisition, since these implicity accord a central role to
communicative interaction (see Hatch 1978a; Halliday 1989). According to creative construction
theory (and all naturalistic theories of language learning), learners develop their own interlanguage
systems using powers of observation and generalisation of rules which they consciously construct

17
and reconstruct in order to comprehend and use the language they experience. Creative construction
theory originated from studies which investigated how children acquire their mother tongue. In
response to the perceived limitations of behavioural theories, evidence was sought that learning was
not just a process of habit formation, but that it must also consist of a cognitive dimension in-volving
the learner developing personal rules for creating and understanding new utterances.

Significantly for this study, under this theory, errors take on a positive acquisitional value. This
means that it actually becomes necessary to produce errors in order to develop an adequate internal
grammar and progress to fluency with target forms one learns from ones mistakes. It is this
implication which is at the heart of the division between those stressing the importance of accuracy
in task-based language work, and those stressing the importance of fluency. This is not to say that
tasks should be designed to promote errors, but that communicative language practice is best
achieved in an instructional environment which encourages learners to experiment with a
devel-oping interlanguage, even though the resulting output may not be completely structurally
accurate.

An integrated model automatisation and restructuring

Task-related theories of acquisition often attempt to synthesise elements of behavioural and


cogni-tive psychology (cf. Littlewood 1992) to present a model of how automatic and creative
responses might act together to achieve oral task completion. What exactly is automatic behaviour in
this con-text? According to McLaughlin (1987: 134) automisation involves:

a learned response that has been built up through the consistent mapping of the same input to the same
pattern of activation over many trials.

Automatisation in relation to task performance might therefore occur through repeated task practice
using the same type of input in the same way, and this concept would seem to take us close to the
commonsense belief that practice makes perfect (see Seliger 1977). Andersons (1985) review of
the literature does indicate the positive potential of such practice on vocational task performance.
However, as second language acquisition studies have shown, simply practising a given language
structure does not invariably yield a gradual approximation to perfect performance (Lightbown 1983;
Pienemann 1989; Loschky & Bley-Vroman 1990). On the other hand, where the learner is
consciously engaged in restructuring, then this less mechanical, more intuitive process may be more
beneficial for eventual language acquisition through conscious cognitive engagement in the learning
process. This concept is more closely related to creative construction theory since it has a basis in
learner hypothesis formation and testing (McLaughlin 1987).

Skehan (1998) also sees a role for automaticity, in that the language user does not rely on a
rule-hypothesis system, but on the ability to access memorised chunks of language remembered to be
appropriate in specific situations. This method is much less demanding of a speakers processing
ability and is the basis of fluency in the target language.

As well as part-skill practice in which individual language sub-skills are isolated for individual
rehearsal, learners need whole-task practice which requires them to integrate a variety of
meta-cognitive and discourse level plans in order to communicate a range of meanings. The real
value of whole-task practice is in enabling learners to formulate new knowledge about the target
language, but the process is a complex one and the type of practice in this context is a social event
involving a high degree of personal involvement on the part of the learner. It consists of a particular

18
kind of interaction which is negotiated by the participants in accordance with the social and personal
factors that prevail in a given teaching context (Ellis 1992: 116). Another important consideration
is that if we define controlled practice as input both in the sense of task instruction input and
performance input between learners, then we have to accept the ever-present possibility that what is
learned may not be what is intended by task designers. Another important consideration is that the
benefits to learners from oral task performance may not be immediate but may well appear later in
the course, or even after the course is over. It is for these reasons that we have to be cautious in
relation to the question of the effectiveness of task-based pedagogy or any language teaching
pedagogy for that matter.

The type of practice and the type of task become important at this point in the discussion. Controlled
practice is designed to automatise rather than to sensitise, though both outcomes are necessary. A
problem-solving task, for example, might be potentially more useful for language learning than an
exercise designed to practise a given syntactic structure, since the former task requires a mental
engagement in the process of formulating possible solutions of creating new knowledge rather
than merely repeating a target structure.

A basic distinction should be made here between declarative and procedural knowledge in the
context of task performance (Anderson 1985). Declarative knowledge consists of such information
as the definitions of task-input words, facts and rules. Procedural knowledge is knowledge about how
the task could be completed, or how it actually was completed. Anderson identifies three stages in the
learning process, which synthesise declarative and procedural knowledge in an integrated model of
acquisition which is applicable to the question of acquisition in relation to the process of negotiated
meaning:

1. In the cognitive stage the task participant makes use of conscious activity. The knowledge
acquired is typically declarative and the learner is able to introspect in relation to it.

2. In the associative stage, the task participant detects errors in his or her original declarative
knowledge and makes efforts (negotiation of meaning) to correct these.

3. In the final, autonomous stage the performance of the task participant becomes more or less
automatic and errors disappear. Working memory is relied on less, and task per-formance
becomes subconscious and procedural.

The value of learnerlearner interaction on task becomes most evident during the associative stage
(2), where the learner not only gets opportunities to try out hypotheses about specific features of a
developing interlanguage, but also, according to interaction theories of acquisition, receives
feed-back on performance from other learners, which some theorists contend is of comparable value
to native-speaker attempts at modification of learner interlanguage. For example, Pica et al. (1996)
found that learners performing a specific task type provided effective on task feedback to each other,
largely through the use of simple segmentation strategies (repetitions of abstracted items from
pre-vious turns). Significantly, this type of feedback was more frequent between learners than
between learners and native speakers. These utterances contained considerable amounts of
morphosyntax and might therefore also have served as a source of input to learners.

The notion of oral task type is central to attempts to integrate task-related skill learning and natural
learning in a methodological framework which links different abilities and relates them to the goal
of communicative ability. Indeed, it also seems central to Bialystoks model (Bialystok & Sharwood

19
Smith 1985) of a cognitive theory of L2 learning. On this model, language proficiency is again
described with reference to the two dimensions of creative construction (the analysed factor) and
automaticity. The analysed factor concerns the extent to which the language learner is aware of the
structure of his or her linguistic knowledge. This means that it can be used by the learner, and appears
to be context-specific in the sense that it might relate to specific types of target tasks for example,
seminar discussion. It follows that learners who do not have this analytic ability will be restricted in
terms of their language competence.

The central issue then becomes the role of oral task practice in achieving not just automaticity in this
process, but also the role of task in promoting and facilitating an analytical comprehension of the task
input in order to justify its redefinition as new, acquired knowledge about the target language.

Input, interaction and output

The input, output and interaction hypotheses attempt to explain not just aspects of task-related
performance, but also language acquisition itself. The term acquisition is generally used in two
distinct ways in the literature. Its main use is to refer to the learners internalisation of new language
forms. The second use of the term is to describe the learners increase in control over forms that have
already been internalised, and in this sense seems closely related to the component of language
competence which is referred to as strategic competence. The concept of attention to task (Schmidt
1990) is important here, and stems from the work of Stevick (1976), where successful
communication was hypothesised to be dependent on attentiveness, which in itself, is dependent on
the level of involvement of task participants. This seems close to Allwright & Baileys (1991: 23)
notion of receptivity, where receptivity is taken as a positive attitude to the learning process and
the task in particular. Both attentiveness and receptivity highlight the crucial importance of learner
motivation in any discussion of task-based learning since, clearly, in order for task input to have a
potential for grammar restructuring in a developing interlanguage, the input must first be noticed
(Gass 1988). This is not so much stating the obvious as underlining the point that the process of
negotiating the meaning of input is only valuable because task participants focus their attention on
the initially unsuccessful parts of their task-related discourse. In this sense, negotiation of meaning,
rather than automatisation, is viewed as the interactive process which is the main trigger for further
language acquisition (Crookes & Gass 1993).

Under the integrated model, the three basic conditions of on task learning are, firstly, exposure to the
language; secondly, interaction with other task participants for construction and automatisation to
take place; and thirdly, a specific need for task participants to communicate. But just what constitutes
input and output in oral task interaction is a complex and important question. We have seen that in
theory, input can be defined as everything that is potentially available to the learner to attend to, but
what is output? According to Selinker (1972), output is a distinct interlanguage (IL) system which
results from a learners attempted production of the target language in relation to generalised rules of
accuracy and appropriateness. Of relevance to the present investigation is Selinkers contention that
predictions of behavioural events in a theory of second language learning should be primarily
concerned with the linguistic shapes of the utterances produced in interlanguages. Selinker (ibid.: 35)
stated that the psychologically relevant data of second language learning will be the surface
structure of IL sentences. However, as we have seen, it is the highly inferential and superficial
nature of the data which makes it particularly problematic.

Even if we assume that interlanguage is a viable research resource, exactly what constitutes input and
output in relation to oral task performance is a definitional question which has no easy answers.

20
Output from one task participant can function as input for another. Krashen (1985) initially focused
attention exclusively on input with his notion of comprehensible input which, he claimed, was a
sufficient condition for language acquisition to occur. The notion of comprehensible input refers to
input to the learner either from task instructions or from interaction with others about the task
instructions, which the learner can then operate on to attempt task completion. Comprehension of
task input can be optimised by task participants through the process of negotiation of meaning.

Swain (1985) shifted the focus from input to the related question of task output, and in so doing,
usefully broadened the debate. Output, viewed as the linguistic performance based on the negotiation
of task input, was seen as the more influential factor, particularly since Swains own students
appeared to fail to achieve grammatical competence despite a learning environment apparently rich
in input made comprehensible. Schachter (1984) appears to have supported this view but had
suggested the use of the term negative input for responses which indicate that learner attempts at
communication have not been totally understood by one or more of the task participants. According
to Schachter, negative input is a source of feedback to learners that a communication problem has
been perceived. Schachter suggested that this feedback might consist of any expression made to the
interlocutor that indicated that another interlocutor was having difficulties with the exchange. The
following are examples of such feedback which can be taken as an indication of such problem
sources, although not an indication as to the type or origin of the problem source:

1. Overt problem admission.


2. Repetition (segmentation) and clarification requests.
3. Procedural repair requests.
4. Content repair requests.
5. Functional reduction (avoidance) strategies.

Carroll (1995) argued that such verbal feedback was irrelevant since it was an impoverished
stimulus to subsequent modification by the speaker. This is because the types violate Grices (1975)
Maxim of Relevance and are perceived by the speaker to be irrelevant to the ongoing purpose of the
talk at that point. To the extent that the learner cannot decide and interpret the stimulus for
interlanguage change, Carroll claims that it is difficult to assign a role to listener feedback in changes
to the speakers grammar (op. cit.: 77).

It can be seen that some definitions of task input might be redefined by other researchers as task
output, depending on whether the resulting performance is in response to another input stimulus, or
is viewed as having the potential to be such a stimulus in itself. These essentially behaviouristic and
mechanistic metaphors of input and output have been necessarily challenged (see van Lier 1996).
The notions of input and output as reified and finite quantities are seen as misleading rather than false.
Van Lier claims that a preoccupation with this particular mechanistic metaphor has led inputoutput
theorists to neglect the essential factor of individual learner motivation on task. If input and output
are seen not as mutually exclusive quantities, but as interrelated potentials of task-based interaction,
then an important question in relation to oral task performance concerns the nature of this
interrelationship. For example, can task participant output also function as task input for acquisition
purposes? How can we distinguish input and output in the metalanguage of task- based language
acquisition? To move further from this mechanistic inputoutput model, it seems crucial to attempt
to consider not only learner motivation but also learner perceptions of the process of oral task
performance. With these considerations in mind, the concepts of task input, interaction and output
can still be used by the researcher as organising devices to refer to the language and procedures
which learners attend to on task, the mutual, negotiative process of that attention, and the new

21
understanding of this language and procedure which results from this process.
Evidence for the role of interaction in acquisition

An important consideration at this stage in the discussion is whether task-based input, interaction and
output would, in themselves, provide a sufficient basis for language acquisition. Carroll (1995: 74)
doubted that this would be the case:

... linguistic stimuli alone, that is, forms and meanings arising from the analysis and interpretation of
speech heard in communicative interactions, do not appear to be an adequate basis for the learning of
complex and subtle properties of language.

An integrated theory of language acquisition must have some mechanism for explaining how the
learners task-based cognitive constructions are confirmed as appropriate and correct, and,
conversely, how errors are disconfirmed. All the evidence reviewed so far that task-based
interaction contributes to second language acquisition is inferential. There is no direct evidence for
any link between interaction and acquisition, and although this could be regarded as potentially
difficult for theories which regard learner interaction on task as a crucial element in the process of
language acquisition, the absence of any established link appears to be no more notable than the
absence of any proven causal link between acquisition and other likely influencing factors such as the
social context of performance.

[A]fter 40 years ... there is still no general inductive theory of language acquisition which will explain
how the abstract categories and constraints of grammar could be induced from linguistic stimuli.
(Carroll, op. cit.: 74)

We can assume, since all tasks are performed in a social context, that, despite the great difficulties
associated with adequate delineation, this must be regarded as an important variable. The same
difficulties are involved in setting up suitable longitudinal studies and there are also important
questions to be considered with regard to the inherent limitations of positivistic research approaches
in the social sciences. Ellis (1990: 109) conclusion that There have been no L2 studies
investigating whether interactional adjustments promote acquisition is partially true. There have,
however, been many studies which have inferred an important linkage between interaction and
acquisition, particularly during the 1980s. Pica & Long (1986) concluded that since there was less
interaction in classroom settings, there was less eventual language acquisition. They concluded that
the teacher-dominated classroom was a poorer environment for negotiation of meaning and therefore
for language acquisition. Unfortunately, many of these studies made claims which were stronger than
the evidence warranted, but their apparent failure to establish stronger connections between
interaction and acquisition has to be viewed in relation to the continuing debate as to what should
properly constitute proof in the physical and social sciences.

It is important to remember, that any theory of language acquisition on which other research
hypotheses may rest is itself dependent on hypotheses which cannot logically be proved to be true
and therefore can only represent deductive conclusions which have to be viewed in terms of
probability and therefore operational adequacy (see Medawar 1984).

Recently, interest in the role of interaction in acquisition has shifted from deterministic quests to the
examination of qualitative issues such as the effects of different forms of feedback, particularly
recasts where the speaker receives implicit correction by the listener saying the equivalent of So
what you are saying is ..., which the first speaker can interpret as a request for clarification. This
type of research has shown that such implicit feedback not only maintains a focus on meaningful

22
conversation (it doesnt violate the Relevance maxim) but also results in substantial incorporation
and subsequent retention of the recast forms in learner speech (Carroll & Swain 1993).

The input hypothesis

One of the more influential hypotheses perhaps initially, the most influential one linking oral
performance to acquisition, was Krashens (1985) input hypothesis. It is essentially a reception
hypothesis, since assumptions concerning its role in acquisition concentrate more on the input itself
than any active decoding role on the part of the learner. The input hypothesis asserts that learner
interlanguage develops as a result of learners comprehending input that contains linguistic features
one step beyond their current levels of language competence this is the well known i + 1 formula.
There are many corollaries to this position, but two in particular put its validity in some doubt. Firstly,
if it is true, speaking must actually result from acquisition and is not a cause of acquisition (or at least
not accounted for as a possible cause). Secondly, if comprehensible input increases linguistic
competence, the comprehensible input hypothesis alone could not justify the use of oral tasks to
promote language acquisition since it assumes that input works on the linguistic competence base,
not strategic competence.

Some researchers, such as Long (1981), have adopted strong positions on the assumed linkage of task
input to acquisition:

1. Access to comprehensible input is characteristic of all cases of successful acquisition, first


or second language.

2. Greater quantities of comprehensible input result in better (or at least faster) acquisition.

3. Lack of access to comprehensible input (as distinct from incomprehensible or zero input)
results in little or no acquisition.

Long (1983a: 378) suggested that one way of demonstrating the relationship between
compre-hensible input and acquisition was to break the procedure down into three steps:

1. Demonstrate that linguistic/conversational adjustments promote comprehension of input.

2. Demonstrate that comprehensible input promotes acquisition.

3. Thereby deduce that linguistic/conversational adjustments promote acquisition.

The difficult and potentially significant step remains (2). Evidence for (1) is inferential and limited.
In a literature review, Chaudron (1985) isolated only one study (Kelch 1985) which demonstrated
that a slower rate of speech led to increased comprehensibility, and thereby, possible acquisition. In
all these studies, it is extremely difficult to isolate one variable such as input so that deterministic
claims can be made. Pica et al.s (1987) study attempted to investigate the important question of
which type of modification interactional or input was the most effective for eventual acquisition.
Their study compared the effects of pre-modified input with interactionally modified input. Results
suggested that interactionally modified input assisted comprehension, whereas the former had no
measurable effect.

A general problem with the input hypothesis is therefore that an empirical study would, firstly, need

23
to show that only comprehensible input promotes acquisition and secondly, that the acquisition is not
sporadic and unprincipled. The input hypothesis has been subjected to a number of other serious
theoretical criticisms:

1. Krashens notion of comprehensibility is misleading in the sense that it is not the only factor
making input learnable and usable (Pennington 1986: 56).

2. The i + 1 construct is not only difficult to define, it never actually is defined and appears to
represent a rather one-dimensional, incremental view of the learning process.

3. The relationship between comprehension and acquisition is not clearly spelled out. Input
functioning as intake for comprehension and input functioning as intake for acquisition must
be distinguished (Frch & Kasper 1986).

4. There is some evidence that simplified input may actually impede acquisition rather than
facilitate it. In the case of a type of simplified input the so called caretaker speech to
children Gregg (1984: 91) claims that no one has shown that the existence of caretaker
speech has any bearing at all on language acquisition.

5. There may be some instances of acquisition without new input. Some changes can be
expected in the learners interlanguage system that result from the learners manipulation of
his or her current lexical and syntactic knowledge.

6. There are other aspects of acquisition which would appear to be input-free. Firstly, features
which are a result of what Zobl (1983) has termed a projection device, where the
acquisi-tion of one feature appears to trigger the acquisition of another. The second aspect is
the loss of deviant interlanguage rules. In most cases, input cannot supply the learner with
the negative evidence that he or she would require in order to unlearn overgeneralised rules.

7. Linked to (6), task-based input theories do not adequately explain how learners distinguish
good input from bad. Another version of this criticism is that peer-group learners cannot do
this, and risk operating on deviant input. It is argued that corrective feedback from other
participants would be totally inadequate to distinguish such input.

8. How do we explain how variations in task input might result in task participants eventually
reaching the same conclusions about the target language?

The main problem with the input hypothesis has been its lack of explanatory power. If we expect the
hypothesis to explain the acquisition process, rather than just how it might contribute to acquisition,
then we would need to broaden the range of factors to be considered. Yule & Tarone (1991) noted
how the more recent use of interactive tasks by the comprehensible input analysts had created the
kind of interlanguage communication situation which represents already established territory for
analysts interested in communication strategies. This approach offers a broader look at the
comprehensible input question, avoiding the narrow confines of some comprehensible input
arguments.

... the assumption of a direct causal connection between input made comprehensible within modified
interaction in a second language and the acquisition of that language continues to be made. It is possible
that it is this (untested) assumption which has led input analysts to use an analytical frame-work
focusing on one side of the page the input side. Since the learner is assumed to be making pro-gress,

24
it is apparently not necessary to analyse the learners output in these studies. (ibid.: 169)

Yule & Tarone suggested a broadening of the input research base. They argued that analysts
inter-ested in the influence of input upon the process of second language acquisition, might benefit
from the insights gained from research on learner use of communication strategies. This other side
of the page perspective (op. cit.: 170) leads us to a more detailed reconsideration of the interaction
hypo-thesis itself.

The interaction hypothesis

The interaction hypothesis is, in many ways, the most relevant explanatory hypothesis for our
purposes, since it envisages a crucial role for input (particularly negative input) but also takes into
account the productive role of the learner in relation to both oral task input information and language
output from other interlocutors in the exchange. The role of the learner is specifically examined in
relation to the central concept of negotiated meaning. The interaction hypothesis has produced many
classroom-based studies of particular relevance to the present study.

Seligers (1977) research was designed to approach the general question of whether interaction leads
to acquisition and his study represents one of the earliest attempts to combine observational data with
a description of learners in terms of their interactional behaviour. This quantitative analysis consisted
of observations of only a small number of learners and an attempt to correlate their performance on
two English language tests. Seliger also attempted a qualitative analysis using a Language Contact
Profile questionnaire which was designed to provide information about the students use of English
outside the classroom. Seliger did find evidence of performance patterns and concluded that
learners who initiate interaction are better able to turn input into intake.

Long (1992) argued that acquisition is made possible, and is primarily facilitated, when interac-tional
adjustments are present. The main interactional features identified and their descriptions are given in
Table 2. These were used by Long in earlier research (1980, 1981) and were also employed by Pica
& Doughty (1985b).

Table 2: Interactional adjustment features.


Interactional feature Definition Example
Clarification requests Any expression that elicits A. She is on welfare
clarification of the preceding B. What do you mean by welfare?
utterance
Confirmation checks Any expression immediately A. Mexican food have a lot of
following the previous speakers ulcers?
utterance, intended to confirm that B. Mexicans have a lot of ulcers?
the utterance was understood or Because of the food?
heard correctly
Comprehension checks Any expression designed to A. There was no one there. Do
establish whether the speakers you know what I mean?
own preceding utterance has
been understood by the
addressee.
Self-repetitions The speaker repeats or A. Maybe there would be
1. Repairing paraphrases some part of her B. Two?
own utterance to help the A. Yes, because one mother goes
addressee overcome a to work and the other mother
communication problem. stays home.

25
2. Preventive The speaker repeats or A. Do you share his feelings.
paraphrases some part of her Does anyone agree with him?
own utterance to prevent the
addressee experiencing a
communication problem.
3. Reacting The speaker repeats or A. I think she has a lot of money.
paraphrases some part of one of B. But we dont know that.
her previous utterances to help A. But her husband is very rich.
establish or develop the topic of
conversation.
Other repetitions: The speaker repeats or A. I think the fourth family
1. Repairing paraphrases some part of the B. Not the fourth family, the third
other speakers utterance to help family.
overcome a communication
problem.
2. Reacting The speaker repeats or A. I think she has three children.
paraphrases some part of the B. This is the thing. She has three
other speakers utterance to help children.
establish or develop the topic of
conversation.

These interactional adjustments are used to represent core features of the process of negotiated
meaning and have been viewed as strategies which oral task participants use to prevent and repair
breakdowns in communication and sustain the conversation in order to complete the task. The list
remains essentially the same in other studies which have investigated the function of such strategies
in relation to the learners attempts to restructure task-based input (see Pica & Doughty 1985b).
However, there are other possible responses which a listener may make to an utterance regarded as
problematic, including attempts to avoid the perceived problem altogether. These types of response
can be considered in relation to the question of the functions of learner output on task.

The output hypothesis

Swains (1985, 1995) output hypothesis focuses on learner production, and is distinct from the
questions of comprehensible input and interaction so far discussed. Swain originally proposed the
output hypothesis as an extension to the interaction hypotheses, but it does appear to form a major
body of alternative theory in its own right in response to two main doubts concerning assumptions
underlying the input and interaction hypotheses. The first doubt concerns the assumption that it is the
exchanges themselves, in which meaning is negotiated, that are facilitative to acquisition as a result
of comprehensible input. The second doubt concerns the assumption that the key facilitator is input
rather than output. Swain cites studies on French immersion students in support of her contention that
the source of language acquisition for these students was comprehensible output itself. Swain
contended that the kind of output which results from interactive language tasks, for example,
provides learners with opportunities for contextualised, meaningful language use. Such opportunities
enable learners to test hypotheses about the target language and to move from a purely semantic
analysis of the language to a syntactic analysis of it. However, Swains conception of output is
difficult to separate from the interactive processes associated with the concept of negotiated meaning.
In both cases, it is the attempt to modify oral-task-enabling language, produced by task participants
themselves, which is conceptualised as either the trigger (Swain 1995) for more syntactic processing
or interactional modification (Long 1983a). Swain appears to have extended her original position to
highlight the triggering effect of output. The output hypothesis highlights production opportunities
as the key formative process for acquisition. On this hypothesis, oral task production contributes to

26
language learning firstly through the learners reception, and comprehen-sion, of task input, but more
importantly, through the learners output the attempts to produce samples of the target language to
complete the task.

Swains proposal is that acquisition may be enhanced in contexts where learners are pushed to
convey messages precisely, and in this regard is similar to Krashens input hypothesis, although, as
we have seen, Krashen has no explanatory theory in terms of the learners psycholinguistic
processing of the extra loading implied by i + 1. Ideas on pushed output have created renewed
interest in the effects of pushing learners to produce difficult, developmentally late acquired or
infrequently used forms. Pushed output exponents assume that one function of pushing output will
be to force learners to notice mismatches between their own and an interlocutors production
(Schmidt 1990) and that this noticing will then lead to learning.

Noticing and modification of output

Schmidt (ibid.) took the strong position that noticing is the necessary and sufficient condition for
converting input into intake, thus negating any role for subliminal or osmotic learning theories. Even
without implicit or explicit feedback provided from an interlocutor in relation to a learners output,
learners may still notice a gap in their knowledge when they encounter a problem in trying to produce
the target language. Pica et al. (1989) suggested that in the subsequent modification of output,
learners may also be engaged in the internalisation of new forms by testing hypotheses about the
target language. This is essentially Tarones (1988) position in relating interlanguage variation
theory to hypotheses of language acquisition through learner use of communication strategies. Swain
was concerned with what has been a neglected area of previous input theories the mental processes
involved when learners move from original to modified output as they move from the triggering
problem to its solution. Swain concluded that when they noticed a problem, learners either used task
input in a more focused way to try to solve the problem, or they worked out a solution through a
negotiation process which resulted in new, reprocessed output. Swain suggested that this process of
switching between first output and second output is part of the process of language learning. The
communicative need created by the task itself pushes learners into focusing on the form of their
linguistic output. Swain claimed that this process actually involves a move from semantic to
grammatical processing.

It is important to reiterate that under this hypothesis, output is not the only acquisitional resource.
When learners attempt target language responses to task demands and they notice a problem, they
modify their output. To initiate this knowledge restructuring by the learner, the feedback must be
informative in the sense that it must tell the learner something he or she does not already know. This
noticing of a gap in the learners knowledge may occur not through interlocutor output but through
reflexive self-questioning by the learner and the generation of alternatives and assessment of these
alternatives through further inspection or more considered thinking. The learner may then attempt to
modify the utterance, to delete part of it, to contradict it, to add to it, to elaborate it, to make
inferences from it, to summarise it or to paraphrase it. The process of negotiated output is thus a
constant fluid reinterpretation of meaning, involving strategic responses which attempt to add to the
message, detract from it or possibly to avoid the perceived problem source altogether. We might then
speculate that the nature of comprehension is itself a process of internal negotiation by the learner in
relation to multiple options of possible meaning.

A study by Pica et al. (1996) addressed some of these issues directly and had generally positive
conclusions in relation to the ability of peer-group oral tasks to stimulate interaction useful for this

27
process of internal and external negotiation. Although learners in the study provided deficient input
modification (compared to learnernative-speaker modifications) they were found to produce
comparable output modification to that provided by native speakers. Of particular interest to the
researchers was the apparently simpler strategy of segmentation, characteristic of
native-speakerlearner conversation. Here, the native speaker picks up on a segment of a prior
utterance by the learner and deliberately modifies it to conform more with a standard target form (in
some research, referred to as a recast see Carroll & Swain 1993). In doing so, the native speaker
confirms the learners original message and also displays a morphologically and lexically modified
version which might represent optimum useful input to the learner. However, the question of
participant interpretation again becomes important in any analysis of the function of such input. As
Carroll (1995: 84) stated:

If the learner cannot attribute a corrective information intention to the corrector, there is no reason to
assume that in normal discourse, he will construe repetitions or clarification requests as forms of
linguistic feedback. It follows that although indirect and tacit forms of negative evidence may be more
frequent in native speaker/non-native speaker discourses than overt forms, they are less likely to be
construed as feedback.

Task performance and deviance

In the previous consideration of creative construction theory, a more fundamental and immediate
problem emerges in relation to the positive role assigned to errors. Many teachers and researchers do
not consider peer-group oral tasks acceptable elicitation devices because, as we have seen, a large
proportion of task input can be expected from other learners. This input is viewed as an impoverished
stimulus; for example, White (1987) suggested that interlanguage negotiation is likely to be lacking
in just those features which the learner needs to acquire. This view is also supported by the study by
Pica et al. (1996), although their conclusion on the issue was that despite this relative lack of quality,
interlanguage negotiation still provides essential production opportunities so that learners can engage
in the fluid reinterpretation of task-related meanings that is the central process of negotiated
meaning.

Doubts remain however, and there is some evidence to support these doubts. In a study carried out in
French immersion classrooms in Canada (Lightbown & Spada 1990) the researchers claimed that the
students who engaged in extensive interaction with peers did appear to have their production errors
reinforced, even though their fluency increased. Concerns that this will always happen if peer group
interaction is encouraged can directly affect pedagogic decisions. Prabhu (1987) recorded a decision
by staff on the Bangalore ESP project not to use or encourage task use which might generate
interlanguage forms, since it was assumed that there would be an exchange of deviant forms with
the danger that the models presented would be subject to fossilisation. Prabhu also reported that in
the Bangalore project, group work was discouraged precisely because of the fear that learnerlearner
interaction would promote fossilisation of deviant interlanguage forms:

As a result, the effect of learner-learner interaction will largely be a firming-up of learners systems:
each learners output will reinforce the internal systems of the others without there being a
corresponding process of revision, or at least with less of a balance between firming up and revision than
when the teacher is party to the interaction. (ibid.: 82)

Consequently, peer group communication was actively discouraged in a syllabus that had otherwise
been noteworthy for its seminal work in relation to a procedural, task-based approach. Many teachers

28
and researchers still do not consider interlanguage a suitable model for classroom second language
development, except in relation to proficiency disparities and the resulting pushed input for one or
some learners. Group work is then regarded as acceptable if the learner is able to:

... communicate with someone who has sufficient proficiency in the target language to ensure an input
that is pitched not just at the learners level but, at times, slightly beyond it. (Ellis 1985: 115)

But the assumption of deviant or tainted input in peer group exchanges remains, and is evident in
calls for greater focus on accuracy in all language learning exercises. James (1994: 248) sum-marised
the fears in the Hong Kong context:

Many teachers and others are worried at the trends towards what they perceive as an inexorable
acceptance of pidginisation (declining standards) as a result of adherence to current classroom dogma.

The current classroom dogma referred to is the communicative approach, where fluency before
accuracy is often interpreted by proponents and critics as fluency at the expense of accuracy. If
language is assumed to be a fixed system, then, since interlanguage forms are not target language
forms, they are to be discouraged. This is a judgement that input from learners is unsuitable for
second language development simply because it is developmental and therefore contains
non-standard forms. Supporting this view, Aston (1986) gives an example of how, during the process
of task completion, learners of a monolingual background can be led to the negotiated acceptance of
non-standard rather than standard L2 forms. Unfortunately, the lack of rigorous longitudinal studies
into the question of deviance, coupled with essentially unresolvable philosophical differences, has
led to confusion on the central issue of definition.

There is some evidence from studies such as Porters (1986), and the present investigation, to counter
the assumption that learners only tend to exchange errors when they negotiate meaning. An earlier
study by Porter (1983) showed that in small-group work, only 3% of the errors were incorporated
into the speech of peers. Rivers (1987) provided early support for the notion that mistakes were part
of hypothesis testing and were therefore the way we all learn rather than the way we should not learn.
Porter further suggested that the findings on input provide evidence that teachers need not be overly
concerned about learners picking up errors from each other, or miscorrecting each other, since they
can also be suppliers of beneficial modifying feedback to other learners. This type of error correction
is reported in many studies (e.g. Bruton & Samuda 1980; Pica & Doughty 1985b; Gass & Varonis
1989). These studies document many instances of learners calling attention to each others errors as
they negotiate task meaning. Further, these learners not only called attention to errors but they
usually did so without miscorrection (see Rodgers 1988; Jacobs 1989).

However, the notion of output contamination through learner use of deviant forms remains one of the
strongest pedagogic arguments against the use of peer group oral tasks (Brown & Yule 1983a). It also
poses difficult questions as to when and how the teacher intervenes on questions of correct language
form in non-native-speaker interactions (Johnson 1988). Concerns with so-called deviance and
fossilisation underlie many theories about what language should be and how it should be learned:

Linguists should consider this paradox very carefully. There is no common-sense belief that linguistic
scientists so urgently wish to displace as the fetish of prescriptive grammar, and there is no
common-sense belief that has been so resistant to their efforts at displacement. (Cameron 1995: 81)

Summary

29
Learners appear to be able to offer other learners a potentially rich source of language input, although
it may still be the case that the type of modified input available is not so rich in features which might
be supplied to learners by native speakers.

Behavioural assumptions are often used in oral task design attempts to link specific input stimuli with
expected output performance. This is not to imply that behavioural theories of causation are
necessarily undesirable, only that they can never form an adequate principled basis for oral task
design. The role of automaticity in learning (rote learning) is well established and well accepted by
most learners, even if it is not thoroughly understood by theorists. However, in order to broaden
descriptions of learning processes it is necessary to accept that research into task performance must
also attempt to improve affective, motivationally-based descriptions what learners think about such
tasks and what they feel during the performance of such tasks.

30
Chapter Three: Oral task description

A concern with language learning tasks is never far away in discussion of language pedagogy, but the
design of these tasks, and their role in mastering a language are seldom considered in detail. There has
not been the same attention to tasks and their design as has been given to general statements of the
content of language teaching curricula in the last ten years.
(Candlin & Murphy 1987: 5)

Oral task classification

The purpose of this chapter is to describe the development of a research task typology which attempts
to provide generalisable descriptions for a range of representative published pedagogic oral tasks and
for oral task types widely utilised for oral evaluation both internationally and in Hong Kong.

A classification of oral task types used for the teaching and evaluation of language can be based on
many design and performance features. A task typology is essentially an organisational construct, or
a managerial convenience (Candlin 1987), but it can and should also be relatable to the kind of
communicative pedagogic practice which is revealed in published course material (see Appendix).
As Legutke & Thomas (1991: 71) stated:

There has been a proliferation of supplementary books offering games, problem-solving or


information-gap activities, role plays, simulations or scenarios to meet the demand for a vehicle of
communication in the classroom.

The organising principle can be based on many aspects of task usage: task content or activity, task
processes, predicted cognitive/behavioural and/or linguistic outcomes etc. An examination of earlier
suggestions for task typologies (Brown & Yule 1983a; Long 1983b; Waters & Hutchinson 1983;
McGrath 1984; Prabhu 1987; Legutke & Thomas 1991) indicates a wide variety of ap-proaches to
the problem, but a common usage of terms such as information gap, jigsaw and role play to
distinguish specific oral task types. There is a strong American tradition of task analysis in the
workplace which represents a significant amount of the material related to task analysis. For example,
McGraths (1984) TIP (time, interaction and performance) task typology proposes that all group
tasks can be categorised as one or another of four main types, each with two sub-types. The four main
task types, related to each other as the four quadrants of a circumplex structure, are identified by the
main performance process that each entails. These are:

generation of ideas or plans


choosing a correct answer or preferred solution
negotiating between conflicting views or interests
executing a decision.

From an information-processing point of view, these tasks differ in terms of the degree to which
effective performance on them depends on the transmission of information of values, interests,
personal commitments etc. (see McGrath & Hollingshead 1996). The identification of a main
performance process is similar to the basis for the typology in this study, where the following basic
performance processes are explored:

sharing information
making decisions

31
acting
sharing opinions.

This chapter describes the development of the task typology used in this investigation, a typology
which incorporates the task-type categories shown in Table 3 below. The typology also includes a
difficulty prediction, based on the design complexity of each task type, giving five basic task types,
each type having an easy and difficult version (e.g. A1 and A2).

Table 3: Summary of research typology task types.


Task Task type Research task topic areas
A1 two-way information exchange university course structure
A2 two-way information exchange university course structure
B1 jigsaw management
B2 jigsaw management
C1 simulation student committee
C2 simulation student committee
D1 scenario student counselling
D2 scenario postgraduate research
E1 convergent discussion/problem solving personal/social problem
E2 divergent discussion personal/social issue

Although an infinite variety of task designs is possible, an examination of the literature (see
Appendix) confirms that the tasks in Table 3 above are basic types which together represent a typical
range of tasks found in published coursebooks. Some of the task types used (types A, C and D) were
adapted for research purposes from Lynch & Andersons (1992) English for Academic Purposes
(EAP) tasks, with written permission from the authors. It is, of course, possible to design tasks that
contain more than one activity type an information gap followed by a discussion for example but
in this case the task would be more correctly regarded as a composite type (sometimes called a chain
task) where completion of one activity is necessary before proceeding to the next.

During the feasibility stage of the investigation, it was important to see whether other practitioners
would regard such a typology as pedagogically representative of oral task design possibilities they
were familiar with. Two workshops (Courtney 1995b, 1996a) were given during which practitioners
were invited to use the classifications to categorise a range of published oral tasks. In all cases,
practitioners were able to classify tasks according to the typology descriptions, with some types
identified as composite tasks consisting of more than one type. In order for this typology to be fully
understood, it is necessary to describe the main task types in more detail below.

Information-gap tasks

The basic problem when defining tasks as information-gap tasks (such as A1 and A2 in Table 3) is
that all oral tasks involve the process of bridging an information gap of some kind. There is a sense
in which all oral exchanges are initiated and sustained in order to share and exchange ideas and
feelings. If perfect understanding (an absence of an information gap) existed between interlocutors,
there would be no need for communication. However, the category of the information-gap task is
normally thought of as consisting of those kinds of task which involve a deliberate content or
procedural gap to be bridged transactionally by interlocutors who have been given different sets of
task-input information. They are sometimes termed one-way if only one interlocutor has informa-tion
to be passed orally to another, or two-way if both interlocutors have information necessary for task

32
completion. The information to be transferred is nearly always designed to be transactional and can
be textual, graphic or a combination of both. The information is usually referred to as static (Brown
& Yule 1983a), since it does not change during the performance of the task. It is also possible to
design tasks which contain input information which changes during the course of the task (updated
financial information during a business simulation for example). In this case, the information can be
referred to as dynamic.

Simulation tasks

The basic feature of any simulation is that it requires the participants to work co-operatively, and
often in a defined role, on a task which is designed to develop language and behaviour considered to
be authentic or relatable to a specific target domain. Simulations often involve chain tasks linked
together over a period of time, with participants transferring information from one task to the next.
Simulation tasks are in some senses the best known interactive tasks largely because of their wide
application in training environments. Real communicative purpose for authentic target situations is
the central claim for a methodology which is generally accepted as having high motivational value.

There are two main sub-types of simulation task. One is concerned with a representation of a real
world situation and has more in common with the role-play task discussed below. The second type
of task views simulations as operating realities (Crookall & Oxford 1990: 15) where the emphasis
is not on the reconstruction of an assumed reality outside the classroom, but on processes which are,
in themselves, considered relatable to a target domain.

The enormous expenditure on simulation exercises in business, commerce and science (see Twelker
1971; Telfer 1993) indicates that, for instructional and vocational training, simulation tasks are
probably the most widely used of all task types. Simulations for language teaching purposes in
tertiary EAP environments are less problematic in terms of their authenticity or relatability since
classroom and target environment are often the same. Representational simulations ideally require a
needs-analysis base which, according to Widdowson (1978: 80), ought then to be able to generate
genuine materials and authentic interactions. However, the linguistic performance skills in
simulation tasks as operating realities in their own right are, in a strong sense, separable from this
authentic base. As Arnold (1991: 241) puts it In a task, manipulation of language is coincidental to
the task. This coincidence may not always be manifest in target-like language, but may still result in
task-completion processes which are relatable to target domains and also involve task partici-pants in
beneficial language practice.

Role-plays and scenarios

Simulations often involve an element of role-play, but role-play tasks are not in themselves
simulations. The concept of an occupational or personal role links what takes place in the classroom
with what takes place outside it. Such roles are partially defined by the expectations and
responsibilities assigned by a wider society. Provided that the learners can identify with assigned
roles, role-playing provides opportunities for students to practise the kind of behaviour and
lan-guage which they will need to produce outside the classroom, practice which will help them
inte-grate these language and behaviour patterns with their own personalities. This provision is
important since, unlike simulations, theatrical postures are often demanded by the design of oral
role-play tasks. This theatrical requirement can be either situational (normal persona but in an
imaginary situation) or character (assumed persona in an imaginary situation).

33
A role-play is distinguished from a scenario (Di Pietro 1987) only in terms of the administration of
the task. Scenario tasks were originally conceived to involve a rehearsal stage where learners would
discuss possible performances and goals. Rehearsal can often produce more natural discourse than
the resulting task performance and therefore should not be seen only as preparation.

Opinion-gap (discussion) tasks

We tend to think of peer-group discussion as a modern pedagogic technique in language teaching.


Despite the strong possibility of a much longer undocumented historical pedigree, according to
Ansbacher (1951), the use of peer-group discussion tasks for personnel evaluation purposes was
developed by the German military in the 1920s. The technique, attributed to Johann Baptist Rieffert
was called the Schlusskolloquium or Rundgesprach and was aimed at showing behaviour between
what were termed equal partners. At the end of the war, the LDG (Leaderless Discussion Group)
was employed by Fraser (1946) amongst others, as a device for selecting management trainees.
American federal and State civil service examiners began trying out the technique at about the same
time (Bass 1950). The basic technique of the LDG was to ask a group of examinees to carry on a
discussion for a given period of time. No examinee was appointed leader and the examiners did not
involve themselves in the discussion. Examiners were primarily interested in assessing character
traits such as leadership potential and personality, but they also assessed co-operative and interactive
strategies. The features most prominent in assessments from this period are: individual prominence
(authoritarianism, confidence, aggressiveness, leadership, striving for recognition); group goal
facilitation (efficiency, co-operation, adaptability, focus on group solutions); and group sociability
(adaptability and a focus on group acceptance). Bass claimed that high inter-rater agreement and high
test-retest reliabilities were reported consistently, especially where descriptive behaviour checklists
were used as the rating technique. With regard to the type of task considered to be ideal, Bass
suggested using problems that are equally ambiguous to all participants, and that require the
initiation of structure for their solution. Where interest is in forecasting leader behaviour in real life,
the structure to be set up should approximate the real-life setting as much as possible.
Published oral discussion tasks still appear to follow this suggested formula. They are designed to
elicit long turns and an exchange of opinions rather than facts, although with ranking and prioritising
variants, the opinions may be based on factual input. In these variations, participants have to consider
given input options and rank them in terms of priority or desirability.

The development of the task typology

There is extensive literature on the subject of task classification, often dealing with the concept of
task from an organisational (and often behaviouristic) standpoint (see Fleishman & Quaintance 1984;
McGrath 1984; Telfer 1993). In relation to the classification of language learning tasks, Candlin
(1987) suggested an initial fourfold usage typology which divides tasks into those focused
respectively on learner training, information sharing, research and experimentation and learner
strategies. We can also attempt to classify oral tasks in terms of their design within the wider
language syllabus (e.g. whether or not they are linked, sequenced or graded) since there is always a
danger of reifying a decontextualised concept of task. However, even if the category of task is a
generally accepted psychological reality (Crookes 1986: 32) for the materials writer and syllabus
designer, there is little general agreement as to how the pedagogic contextualisation of this
psychological reality might be realised.

34
Many definitions of task appear to claim generalisability for identification and grading purposes, but
are couched in terms too broad to allow for individual classifying distinctions to be made as Breen
acknowledged:

... the notion of task is used in a broad sense to refer to any structural language learning endeavour which
has a particular objective, appropriate content, a specified working procedure, and a range of outcomes
for those who undertake the task. (Breen 1987a: 23)

Breen went on to make a useful distinction between task-as-workplan and task-in-process. This
distinction separates the notions of what the task is designed to do and what it actually does. Candlin
(op. cit.: 10) considered tasks to be problem solving activities, which in a strong sense they all are,
although Candlin appeared to be referring to both prescribed content problems such as what is the
meaning of the diagram? and procedural problems such as what do we have to do now?. A similar
problem arises with information-gap activities since there is a sense in which all oral tasks that
require an oral exchange for completion are information-gap activities. The gap may be intended to
be closed through a predominantly transactional task performance, where an exchange of
information about a picture or text is required, or a predominantly interactional performance, where
a personal exchange of knowledge between task participants is stimulated. Although the distinction
between transactional and interactional language use is one of emphasis rather than exclusivity (see
Brown & Yule 1983a for a fuller discussion), the distinction is sometimes employed in published oral
task types and those used for evaluation purposes.

An example of this type of classification can be found in the oral component of the Certificate in
Advanced English (CAE) examination, from the University of Cambridge Local Examinations
Syndicate (UCLES 1995). The CAE speaking paper is conducted by two examiners utilising a
sequence of dyadic pair group oral tasks, described in the following way (ibid.: 90):

Phase A: Interactional turns


Phase B: Transactional long turns with reciprocal information-gap activities.
Phase C: Transactional short turns, negotiating the outcome to a problem-solving task. Opinion
exchange task.
Phase D: Discussion task, interactional and transactional, widening the theme of phase C.

A further distinction can be made in terms of the source of the input information. Information-gap
tasks rely on the exchange of given but different information by task participants. However, when the
task design manifestly includes a procedural problem to be solved of the What do we do now? type,
it is often possible to dispense with lengthy textual or graphical task input. In place of the information
gap that motivates the exchange of simple information in closed tasks, the students now need to
overcome what is sometimes called an opinion gap (Rixon 1979) or a problem-solving gap
(Prabhu 1987). An example of this is the well-known lifeboat or desert island survival situation
(Klippel 1984), where task participants have to exchange opinions on how to rank a list of items in
order of importance to their survival.
Problem-solving is involved in all classroom discussion-type activities, and these can be organised in
a variety of ways (see Ur 1981; Klippel op. cit.; Pattison 1987). Problem-solving tasks enable the
classroom situation to be used as a setting for learners to express their own meanings in real, creative
communication (Rivers 1987). Oral tasks that involve this negotiated consensus of opinion have
often been termed convergent and tasks that do not require such a consensus, divergent (see also Duff
1986).

Given these divisions of oral tasks into three basic groups, is a more detailed descriptive typology

35
possible for research purposes? Berwick (1993) cautioned that the construction of a task typology is
as much a conceptual as an empirical problem, but the necessary analysis of tasks in order to
construct a typology requires attention to both aspects of the problem. The conceptual problem
centres on the elusive but important notion of psycholinguistic relevance (Crookes 1986: 7) which
appears to be largely a prediction of the features of task design which might be the most influential
variables in relation to participant motivation for task performance. But how do we predict these
variables? Is it better approached from an analysis of task design or task performance skills?

Hoffman & Medsker (1983), working in the well established tradition of American business
management concerns with performance effectiveness, made a necessary distinction between the
term task-analysis which they preferred to reserve for a description of the skills required to perform
a specific (target) job, and instructional task-analysis, which isolates what must be learned in order
to be able to perform each task. Overlap can be expected to occur between any analysis of what might
be regarded as general performance skills and specific (task-related) skills revealed by instructional
task-analysis, taking the workplan stage of the task as a starting point. Specific state-ments
concerning necessary task completion skills and sub-skills would then only be meaningful after an
examination of the task performance. This process does allow for the necessary establish-ment of
what are essentially hypotheses about psycholinguistically relevant task design features.

The final task design features selected for the research typology used in the present study were
initially identified on this basis they were hypothesised as being pyscholinguistically relevant
influences on task performance outcomes. These features were interaction requirement (i.e. whether
or not task completion actually required interaction), goal orientation and task outcome (i.e. whether
the outcome was likely to be convergent or divergent), task complexity (defined in terms of input
information loading purely a task design feature) and interactant relationship in terms of the
all-important aspect of control of this input information.

Information possession rights and duties are an important design feature of oral tasks, relating as they
do to participant perceptions of status and power during task performance. Following Pica et al.
(1996), the category of interactant relationship was further elaborated into the sub-categories of
information (INF) holder, INF requester, INF supplier and INF requester-supplier. Interactants are
labelled X and Y, although in some pedagogic situations there may be more than one of each,
depending on the group arrangements.

Lastly, task topic was also included in the typology since, from the initial feasibility study, it was also
hypothesised as being an important variable in relation to task outcomes.

36
Table 4: The research task typology.
Interaction
Task type and
requirement and Goal orientation Difficulty prediction
description
information control
A1 Course structure
interaction necessary convergent, one correct A1 < A2
outcome
INF holder X or Y
two-way information
INF requester X or Y
exchange
INF supplier X or Y flowchart is partially
X and Y must exchange complete
information on the
structure of a credited
course
A2 Course structure
interaction necessary convergent, one correct A2 > A1
outcome
INF holder X or Y
two-way information
INF requester X or Y flowchart is blank
exchange
INF supplier X or Y
X and Y must exchange
information on the
structure of a credited
course
B1 Company
interaction necessary convergent, one correct B1 < B2
management
outcome
INF holder X or Y
jigsaw
INF requester X or Y four pictures
INF supplier X or Y
X and Y must exchange
pictorial interpretations
B2 Company
interaction necessary convergent, one correct B2 > B1
management
outcome
INF holder X or Y
jigsaw
INF requester X or Y eight pictures
INF supplier X or Y
X and Y must exchange
pictorial interpretations
C1 Student committee
interaction expected convergent, three C1 < C2
possible outcomes
INF holder X = Y
Simulation ranking
INF requester X = Y choose one candidate
from two
INF supplier X =Y
X and Y have to select a
candidate
C2 Student committee
interaction expected convergent, five C2 > C1
possible outcomes
INF holder X = Y
Simulation ranking
INF requester X = Y choose one candidate
from four
INF supplier X =Y
X and Y have to select a
candidate

37
D1 Seeing a tutor interaction expected
divergent, many D1 < D2
possible outcomes
Scenario role play INF holder X = Y
role more familiar
INF requester X = Y
X and Y take the role of
tutor and student INF supplier X =Y

D2 Doctoral research
interaction expected divergent, many D2 > D1
possible outcomes
scenario role-play INF holder X = Y
INF requester X = Y role less familiar
X and Y take the role of
student and supervisor INF supplier X =Y

E1 Student problems
interaction expected convergent, restricted E1 < E2
outcomes
two-way convergent INF holder X = Y
discussion
INF requester X = Y discussion points given
X and Y have a student INF supplier X = Y
problem to discuss
E2 Student issues
interaction expected divergent, many E2 > E1
possible outcomes
two-way divergent INF holder X = Y
discussion
INF requester X = Y discussion points not
given
INF supplier X = Y
X and Y have a social
issue to discuss

Task topic

Brown et al. (1984) used a simple notion of topic partially to define communication task types. Topic
was defined as what was being talked about, and therefore encompassed both procedural, content and
interactional concerns. Brown & Yules (1983a) categorisation of tasks was one of the first attempts
in the literature to attempt a principled classification of task design in relation to task content and
topic. The notion of a static task was used to define those tasks where the content remains unchanged
or static for the duration of the task. Typically, this might be where a speaker has to instruct a hearer
how to complete a map, so that both participants finish the task with a similar copy of the same map.
Dynamic tasks, on the other hand, were defined as those where speakers have to describe
relationships which change during the course of the task. Brown & Yule used a series of differing
accounts of a car crash as one type of dynamic task, designed to promote referential transactional
exchanges between task participants. In dynamic tasks, changes of character, location or time might
be involved and the actual language used has to allow the listener to distinguish which character the
speaker is referring to at any particular point in the task.

Abstract tasks are those where speakers have to express opinions on an abstract topic, where the
stimulus material does not contain the actual content to be communicated but requires task
participants to abstract it from prior experience. The speaker has to evaluate, select and structure the
information. Abstract tasks can be the most difficult type of task to perform, for a number of reasons.
The speaker has more communicative responsibility, i.e. he or she has to produce much of the input

38
information content and structuring and has to produce a response geared to a particular purpose. The
speaker also has to use imagination and language not presented in the stimulus material. The
influence of task topic is, according to Tarone (1988) one of the most under-researched areas in this
field. Topic selection and control are important analytical perspectives in conversation analysis,
although the definition of topic as a unit of discourse is problematic. How-ever, as van Lier (1988:
147) acknowledged:

The lack of definition that surrounds attempts to examine topic rigorously does not prevent
investi-gators from deriving much useful work while using the construct.

In this study, task topic is investigated in relation to task procedure and task content. Of particular
interest in relation to the research tasks is the analysis of topic management with respect to task
content, task procedure and off task concerns. Emphasising the importance of topic management, the
topicalisation hypothesis (see Table 2) postulates that acquisition is more likely when learners have
more control over the initiation and manipulation of a topic.

Task difficulty and complexity

As Long & Crookes (1993) reported, there is very little empirical support for the various models of
task difficulty which have been proposed. This is partly because the constructs themselves have not
been well defined operationally, and partly because there is no clear understanding of the cognitive
processes involved in the perception of difficulty by task participants. Prabhu (1987) used a
particular definition of task distance or the number of steps involved in problem solving, as a
measure of the relative difficulty of tasks. One of the earliest studies to attempt to provide adequate
research definitions was that of Brown et al. (1984), who identified several factors which appeared
from their research to be related to task difficulty. Task difficulty was defined experientially, that is,
in terms of the prediction of participant experience of difficulties whilst on task.

In order to clarify a fundamental confusion in the literature, this should be distinguished from task
design complexity, which is a feature of the prescribed requirements and organisation of the task.
This can be expected to be a major influence on experiential difficulty but should not be equated with
task difficulty itself.

The degree of complexity was summarised by Brown et al. (op. cit.: 64) in the following terms:

Increasing degree of complexity ---------------------->


fewer goals more goals
fewer paths to goals more paths to goals
fewer characters more characters
fewer relationships more relationships
familiar topic unfamiliar topic
less input text more input text
less visual input more visual input

Figure 2: Degrees of task complexity.

Any increase in the number of task goals or the amount of input text, for example, was hypo-thesised
as increasing the difficulty level of the task. The relative average length of the speaking turn required
for task completion was also hypothesised to be a major factor determining this experiential

39
perception of difficulty by students. This perception can also be linked to the important concept of
communicative stress.

40
Communicative stress refers to conditions under which a speaker feels more or less comfortable in
producing a turn. The number of task elements such as the amount and type of input instructions and
number of task goals are factors relating to task complexity and subsequently, perceptions of task
difficulty. Communicative stress or apprehension (McCroskey 1984) generally has a negative effect
on oral performance (Spielberger 1966). Learner factors such as cultural expectations and classroom
response styles might also be expected to be important in any consideration of communicative stress.
This might also be expected to result in an increase in the employment of certain types of strategies,
for example, avoidance strategies, as a response to these perceived task difficulties.

Factors related to the context or setting of task performance, such as the number of participants, and
individual learner factors, such as English proficiency level, can also be hypothesised to affect
learner perceptions of difficulty (see Carmichael 1996). The most obvious difference between the
conversation of dyads and three or four participants (trios and quartets) is that dyad members expect
utterances normally to be addressed to them. The interactional structure of trios and quartets is likely
to be much more complicated, since there is more potential competition for turns, and task
participants have a more complex notion of audience to address.

Previous studies have generally not attempted to include task complexity and difficulty as separate
variables. Segal (1982) suggested that the operational definition of task complexity has varied with
studies and has not been standardised. Anderson & Lynch (1988) defined difficulty in relation to
information and topic factors, and this is also representative of many other operational definitions of
task complexity. But one variable which appears to be consistently represented in research is the
number of task goals. A second dimension which is consistently hypothesised as significant is the
amount of specific, unequivocal information given to the decision maker (Driver & Streufert 1969).
Again, these dimensions are differentiated from those of difficulty (Shaw 1976) and the parallel
notion of participant uncertainty (Perrow 1967; Thompson 1967).

On the basis of these considerations, in this study, task complexity is operationalised in relation to the
following three dimensions:

(i) the information loading of the task characters, relationships, problems


(ii) the number of goals participants have to consider
(iii) communicative responsibility in relation to topic.

This is essentially task-related and therefore facilitates a grading of the research oral tasks in terms
of task as workplan design complexity. The main research task typology design components were
therefore:

1. Main interactive performance processes assumed at task design stage:

(a) exchanging information


(b) making decisions (prioritising)
(c) interpreting roles
(d) exchanging opinions
(e) solving problems

2. Interaction level necessary or expected (co-operation)

3. Who controls the task input information? (symmetry)

41
4. What type of goals? (convergence/divergence)

5. Task (as workplan) design complexity

(a) input information loading and processability


(b) number of goals
(c) communicative responsibility (topic familiarity and type, e.g. abstract, known role etc.).

The notion of the communicative responsibility level of the task is important. This is the degree to
which the speaker exercises control over the selection and use of information. Tasks carrying low
communicative responsibility are those where the basic task content data is supplied. Conversely,
tasks with a requirement for some degree of role-playing carry a higher level of communicative
responsibility, since task participants have to invent appropriate language and style as well as
com-plete the task.

Classifying and grading tasks in terms of the perceived power relations between task participants is
also an important consideration in terms of what is sometimes referred to as the symmetry of task
relationships (van Lier 1996). A sociolinguistic analysis would include participant perceptions of the
social power distribution within the task performance group, which might be linked to task
motivation in particular cases. For example, it may be that a task role which is viewed as sub-ordinate
by a task participant is not conducive to effective learning, since it proves socially inhibiting and
does not allow the learner to employ strategic competence fully.

The ten research tasks utilised in my quantitative investigation were divided into three broad design
groups, each one based on very different assumptions about the resulting task performances. These
three groups are: information gap tasks; simulation, role-play and scenario tasks; and opinion gap
(discussion) tasks. Tasks in each of these groups were then modified to represent two distinct levels
of design complexity. For example, the information-gap task (see Table 4) has two levels of design
complexity (A1 and A2). Both are information-exchange tasks where participants X and Y must
exchange information on the structure of a hypothetical credited university course. In the A1 version,
more essential task input is given (X has a partially completed flow chart of the course requirements).
The A2 version has a greater design complexity since X only has a blank flowchart and Y must
supply all the input by attempting to read and understand the input text.

Experiential grading of such tasks prior to performance in terms of task difficulty is very problematic,
since although we may expect a task to be difficult because of its design complexity, it may not be
perceived to be operationally difficult by task participants.

Task authenticity, relatability, memorability and salience

A final important consideration in the description or design of oral tasks is the notion of authenticity,
a central concept in cognitive learning theory. Cognitive theory, in most of its present forms, appears
to differ fundamentally from behavioural accounts of how language is learned, specifically on the
question of the necessity for authenticity in language learning tasks. Johnson (1988) identified the
opportunity to practise in authentic conditions as one of a set of necessary conditions for language
acquisition.
The notion of authenticity in the language classroom looks at first like something of a paradox if we
define the term in relation to assumed authentic language and behaviour outside the language

42
classroom. Widdowsons (1978) notion of authenticity forms a challenging philosophical basis for
the term, where it is taken to signal awareness of purpose and motivational commitment to that
purpose. Behavioural authenticity refers to behaviour or action which is intrinsically motivated, in
contrast to behaviour which is not authentic, or is prescribed. This is different from the kind of
intrinsic or extrinsic motivation which might be brought to authentic materials those which
appear authentic because they represent stereotypical target activities in a stylised form with high
face validity, often more in relation to content than process.

The research typology tasks here (see Table 4) were designed to include target language and
performance skills associated with the tertiary educational domain in Hong Kong, and were therefore
assumed to be relatable (Swales 1990: 76), but not necessarily authentic, to this domain. The
linguistic competence required to perform successfully in this target domain can be idealised as the
knowledge that a fluent English speaker possesses to use and interpret language appropriately in the
process of interaction in the domain of the Hong Kong tertiary learning environment. The notion of
communicative competence (Hymes 1972; Gumperz 1986) addresses the question of appropriateness
and takes us inevitably into sociological issues concerning the existence or not of definable and
teachable conventions of conduct in academic domains. Boswood & Mariott (1994: 8) made a case
for the existence of such domains in the Hong Kong context:

[a tertiary discourse community] ... allows its members to be characterised in terms of a range of
generalised behaviour patterns, constructed through a cognitive network of norms, beliefs and events.

The assumption seems to be that despite this general characterisation, specific instances of such
domains will vary from an idealised, prototypical structure, so genre analysis in this context is
concerned with establishing such a structure for the specific target domains of specific courses. This
description makes no reference to authentic task-enabling language it is a description based purely
on the assumed specificity of language-related behavioural patterns. It is in this conceptually vague
area that the notion of task relatability is possibly more useful than that of task authenticity, in terms
of attempting to establish working descriptions of target language domains, although there are
exceptions. For example, the target domain of an airliner flight deck can be fairly accurately
described in terms of prototypical language and behaviour, and task-based language instruction to
increase the safety and effectiveness of in-flight communications is often a required training
syl-labus component (e.g. Cathay Pacific in Hong Kong, see Telfer & Biggs 1988).

Authenticity in this context refers to the task content chiefly the domain-related input although
sometimes the term is extended to refer to a set of what are claimed to be authentic behaviours.
Brown et al. (1989: 33) propose authentic activity and situated cognition and invention as
alternatives to decontextualised didactic teaching in content classrooms. On their definition,
authentic activity involves a learners direct experience of the modes of thought and action of a
particular domain. Domain-related research such as Du-Babcocks (1996) analysis of turn-taking
behaviour and communication strategy usage in Hong Kong business settings is an essential basis for
the construction of this authentic activity so that it not only utilises authentic content but also
authentic processes.

The establishment of authenticity with respect to oral tasks presupposes, then, a description of both
target language requirements and domain language characteristics. In terms of target language
requirements, the oral tasks utilised in this study are relatable to target oral skills (argumentation,
seminar interaction etc.) associated with an academic genre. However, the question of domain
language characteristics remains potentially more troublesome. For example, Swales (1990) points to
the relative sparsity of research into oral aspects of the academic genre (but see Ferris & Tagg 1996)

43
when compared with products such as the research paper and the extensive literature surrounding
reflection on the process of writing.

Authenticity is often more obviously an issue in terms of certain types of task, such as scenarios and
simulations, where learner expectations of the behavioural aspects of role requirements may be
heavily coloured by stereotypical attempts at characterisations. Simulations are closely related to
role-playing and the term usually refers to an activity in which some kind of problematic situation is
contextualised in behaviour and language expectations related to a target domain such as the
prototypical business meeting, university committee etc. As we have seen, the key difference
between simulation and role-play is one of role distance. Participants are not asked to play the part
of somebody else (which might involve considerable role distance) but are expected to be
them-selves in a defined functional capacity as a chairperson or committee member etc. The
emphasis in simulations is on problem-solving within a defined situation, rather than concentrating
on adopting unfamiliar behaviour patterns.

There is still very little research evidence to support the main rationale of many types of authentic or
relatable performance tasks that of the transfer of specific genre-related skills from the class-room
to the target environment (Twelker 1971). However, it would be difficult to justify the enormous
expenditure on simulation exercises in business, commerce and science, if positive results were not
being obtained, even though these results may relate more to notions of behavioural appropriateness
rather than language competence.

EAP simulations are potentially less problematic than business simulations, since classroom and
target environment are essentially the same. Other types of simulations would require a needs-
analysis base on which to produce genuine materials and authentic interactions. The linguistic
performance skills in simulation tasks in particular are, in a strong sense, secondary to this authentic
base. However, this predominant concern with authentic behaviour should not allow considerations
of authentic task language to be dismissed. What makes language authentic in any target situation has
a lot to do not only with notions of appropriateness and behavioural mores, but also with
collocational vocabulary and phrase choices that signal membership of particular domains. This
signalling is evident not only in the processes whereby native speakers negotiate meaning in
conversational situations, but also in native speakers knowledge of the collocational restrictions
attached to many lexical and phraseological combinations, code usage restrictions which are
arguably a source of significant problems to learners.

44
45
Chapter Four: A quantitative comparison of oral task performance

Research design

This chapter details the quantitative investigation and resulting description of specific features of
participant interaction resulting from the performance of particular task types at specific levels of
complexity. The description of task output patterns is based on larger intact (normal classroom)
groupings than previous studies have attempted, and was a response to a perceived requirement for
larger and more comprehensive studies of task output patterns, conducted over a longer period of
time than has normally been the case (see Duff 1993: 59). The opportunity to do this was presented
by pedagogic involvement over a five-year period with the undergraduate language enhancement
programme at the Hong Kong University of Science and Technology (UST), and through a funded
research project for the Hong Kong Language Fund. Results from a pilot study in 1992, using only
two oral task types, endorsed the feasibility of larger-scale data collection (Courtney 1994). Using a
core of discourse features associated with negotiation of meaning on such tasks (see Long 1980, 1981;
Pica & Doughty 1985b), a quantitative descripion of task outputs was established.

Previous studies have usually restricted the number of respondents to fewer than twenty. This
in-vestigation utilised task performance data from 120 independent dyads (240 subjects) for five task
types at two levels of complexity per type (ten research tasks). It was important for statistical
pre-dictions of distribution that dyads were independent, i.e. that no learner participated in more than
one of the research performances. This was achieved through careful selection from recorded
performances over a three-year period, between 1993 and 1996, from two complete language
enhancement programmes. These programmes (LANG 001 & 002) were required language courses
for first-year undergraduates with a final grade of D or E in the Hong Kong Examinations Authority
Use of English A level examination. The courses were task-based English for Academic Purposes
courses, designed to involve students in the processes of academic writing and discussion. Similar
oral tasks to those in the research typology were formally included in the syllabuses of these courses
at the design stage, which ensured that the task performance setting for the research tasks was that of
the normal classroom environment for these intact groups. In preparation for this investigation, all
dyadic group work during the courses was regularly recorded so that task participants did not feel
under abnormal research conditions during research performances. The data investigation
necessitated the services of a transcriber and co-rater to deal with the large amount of transcription
involved, and the requirement for rating checks. A comprehensive data analysis was completed using
a multivariate analysis of variance (MANOVA) developed specifically for the research design, with
the assistance of the Department of Mathematics at UST, and computed with the assistance of SPSS
(Statistical Package for the Social Sciences) software, licensed to me for the duration of the project.

Research hypotheses

The strategies that learners use in order to complete oral tasks form an important part of this
investigation, since the employment of such strategies is central to the discussion of negotiated
meaning and its possible link to language acquisition. Further, if specific patterns of strategy
employment can be consistently identified from a large and representative sample of oral task
performances, then this would not only support the use of tasks designed to promote strategic
competence but also the inclusion of specific pre-task strategy instruction in language teaching
programmes and the use of specific oral task types for evaluation purposes.

46
The following hypotheses in relation to learner output on task, participant perception of difficulty
and participant management of interaction were derived from the initial research questions. For
clarity, both the null and alternative hypothesis are given.

Learner output

1. There are no significant differences between learner outputs and different task types. (null)

2. There are significant differences between learner outputs and different task types. (alternative)

3. There are no significant differences between learner outputs and tasks of different complexity.
(null)

4. There are significant differences between learner outputs and tasks of different complexity.
(alternative)

Perception of difficulty

5. There are no significant differences between learner perceptions of task difficulty and different
task types. (null)

6. There are significant differences between learner perceptions of task difficulty and different task
types. (alternative)

7. There are no significant differences between learner perceptions of task difficulty and tasks of
different complexity. (null)

8. There are significant differences between learner perceptions of task difficulty and tasks of
different complexity. (alternative)

Co-operative management of interaction

9. There are no significant differences between task types and learner employment of co-operative
communication strategies. (null)

10. There are significant differences between task types and learner employment of co-operative
communication strategies. (alternative)

11. There are no significant differences between learner employment of co-operative


communi-cation strategies and tasks of different complexity. (null)

12. There are significant differences between learner employment of co-operative communication
strategies and tasks of different complexity. (alternative)

47
Research definitions

The following definitions are provided for the core variables investigated in the quantitative
investigation, with an indication of how they were derived from the underlying constructs that they
represent, and how they relate to usage by other researchers in similar studies. The definitions are
operational, and relate specifically to the domain of the research investigation. The variables are
divided into two main categories those which facilitate a quantification of output in terms of turn
length, and those which facilitate a quantification of strategy usage in terms of participant
management of the interaction.

Utterances, turns, moves and repairs

Schegloffs (1972) turn classification framework included a sequential perspective of turn as a


sequence of utterances linked by turn topic. I have followed Labov (1972), Schegloff (op. cit.) and
Jefferson (1973) in regarding the utterance as the basic unit of analysis for the investigation of learner
strategy usage. This initially disregards the context of act (Austin 1962) for the purposes of discrete
analysis.

According to Harris original definition (1951: 14), an utterance could vary in size from a single
lexical item to a complete turn. Crookes & Rulon (1988) defined an utterance as a stream of speech
which constituted a single semantic unit. Since the goal for definition in the present investigation is
to facilitate a quantitative assessment of the amount and patterning of talk in relation to specific task
types, I have avoided definitional conjecture and equated single utterances or recognisable lexical
items as words for the purpose of this quantification. However, this decision does not permit any
observations about the quality of any of these utterances, or the turns which they collectively produce.

In order to facilitate turn quantification of task output performances, back channel responses have
been excluded from the analysis. This decision was taken because an analysis of back channel
responses would have made the data capture and processing impossibly lengthy and complex in
relation to the large number of task transcripts being investigated. This is not to underestimate the
possible importance of such responses in relation to participant strategy use, however.

My decision seems close to that of Rulon & McCreary (1986), who calculated the mean number of
words per turn or c-unit (communication unit). Rulon & McCreary adapted Lobans (1966)
definition of the c-unit, defining it as a phrase or sentence which communicates pragmatic or
semantic meaning regardless of grammaticality. This is a necessarily practical response to
interlanguage data where, as in all forms of oral discourse, interlocutors do not characteristically use
complete and grammatical sentences. The resulting count consists of all discrete utterances (words)
and recognisable lexical items in the transcripted segment of the task performance, minus the back
channel responses. It therefore represents an operational and quantitative measure of the amount of
output produced (in words) per speaker per task performance. Dividing this figure by the number of
turns gives an operational measure for the median length of turns produced by each participant
during the task performance. The large amount of data to be handled in this type of research design
effectively precludes a transcription process involving a more detailed prosodic and sequential
analysis, and this type of operational definition is adequate only for the goal of quantifying the total
amount of language output from each speaker during a task performance. It does not facilitate any
conclusions about the quality of the turn output, i.e. the potential of different types of turns in terms
of language practice opportunities. The initial analysis of the task perform-ance data was in relation

48
to the concepts of turn and turn topic. Topic management was defined in relation to the patterns of
participant topic nomination and shift between what appeared to be transactional content, manifestly
related to task content and design (CON TRANS) and what appeared to be interactional content
(CON INT), manifestly related to phatic or off task concerns. For the purposes of the quantitative
analysis, the category of CON TRANS was further divided into information specifically related to
task procedure and information specifically related to task content.

The move has been referred to as the phoneme of conversation (Goffman 1976: 272; Coulthard
1977: 745) and has featured as a construct in classroom analyses of interaction (e.g. Sinclair &
Coulthard 1975). There seems to be common agreement that two moves which belong together can
be referred to as an adjacency pair (Schegloff & Sacks 1973: 295; Coulthard 1977: 100; Richards
1980: 421) and this is regarded as one of the basic forms of an exchange. In the SinclairCoulthard
model of discourse, such moves would therefore constitute the framework of the task performance.
In this way, the move sometimes forms part of a task-based strategy (when an exchange appears to
mark or show apparent employment of a strategy under investigation), though we might want to
argue a distinction on the basis of conscious or unconscious interventions in the task performance
(Schmidt 1994). Rutherford & Sharwood Smith (1985) have argued that consciousness raising
drawing learners attention to the formal properties of language facilitates language learning.

The ways in which the moves of task participants follow one another are conceptualised as being
dependent on rules for turn-taking (Sacks et al. 1974), and also on mutual understandings between
task participants, which are related to the notion of contingency, where what is said in one turn is
dependent on, or related to, a previous turn. Some of the more evident features of turn-taking during
task performance are as follows:

1. Speaker-change recurs or at least, occurs.


2. One interlocutor talks at a time.
3. Occurrences of turn overlap are common but brief.
4. Transitions from one turn to the next without any time gap are common.
5. Both partners use turn allocation techniques.

The concept of a turn is a central and complex one in discourse and conversational analysis. From the
perspective of conversational analysis, an analysis of turn-taking can provide a picture of the
different moves of task participants. There are only two types of turn allocation, either self-selection,
or other-selection. When speaker A has made a move, he or she may either pass the initiative to B or
initiate a new exchange. This double move has been referred to as a back-pair move (Goffman
1976: 272) where the two moves are essentially back to back. This also assumes contingency and
co-operation in turn-taking, since it is assumed that what another speaker has said is intended as a
reply to the previous speaker, an idealised principle of co-operative behaviour in conversation (Grice
1975). When speakers perceive a difficulty in a conversational exchange, they may try to repair the
difficulty they experience or the assumed impending break in the communication. Repair strategies
are taken in this investigation as important indicators of the co-operative principle at work,
particularly other-repairs, carried out by the listener in a dyadic pair. Self-initiated repairs and
requests for assistance are also important considerations in attempting to understand the
conversational moves of the dyadic pairs.

49
Strategies for turn-taking can therefore be viewed in relation to co-operative principles and learner
communication strategies. Knowing when to take part in a conversational exchange, and how to
manage the turn once it has been obtained, is a fundamental oral skill. Turn-taking is itself closely
related to information control and topic nomination since one main reason for taking a turn is to
contribute or modify the topic being discussed with either new or reprocessed information. If we are
to consider the possibility of teaching communication strategies involving co-operation, turns, topic
management and repairs, then we must gain a clearer understanding of how these features operate in
the particular context being researched.

The information-related turn

Given the large amount of data to process from the present investigation, the notion of topic shift
between information manifestly related to task procedure and information manifestly related to task
content was a practical data management solution to the problem of facilitating primary topic-shift
distinctions. The category (CON INT) that of topic shifts away from task concerns, usually
specifically related to interpersonal concerns was also found to be necessary as the transcriptions
proceeded.

Turns off task

These were turns where the topic raised or discussed appeared to have no specific relation to task
content or procedure. One example from a transcript will suffice: What are you going to have for
lunch? This could initially be taken to indicate lack of interest by the speaker in pursuing the task
goals any further or perhaps an attempt to avoid a perceived difficulty in the task. This approach to
the analysis of off task transcript segments is similar to van Liers (1988) topicactivity distinction.
Turns on task were taken to be those turns which were either procedural or content turns as defined
above. Having isolated the turns that appeared to be concerned with task completion, it was then
possible to investigate the general category of turns on task in relation to repair activity.

Repairs and perception of task difficulty

Chaudron (1977) presented a comprehensive account of corrective actions by students in second


language classrooms, which he termed repair. The concept of repair can potentially cover a wide
range of actions, which might include statements of procedural rules, problems with compre-hension
of task input, second starts, prompting, clueing, helping, explaining and error correction. Repair in
the present study was treated as the generic term, and attempted correction of errors, for example,
regarded as one type of repair. In a dyadic group, an indication of a problem may be initiated either
by the same speaker (self-initiated) or by the other speaker (other-initiated). Repairs may be
attempted by the same speaker (self-repair) or by the other speaker (other-repair).

Procedural repair is concerned with the organisation, structure and completion of the task activity
itself. It is essentially related to planning and organising the task completion and typically results in
questions such as What do we do now? or Are there any more blanks to complete?

50
Content repair is concerned with the enabling language of task input and task completion, typically
the meaning in context of troublesome vocabulary items. This distinction is similar to that made by
Kasper (1985) in dividing her transcript data into language-centred phases and content-centred
phases.

In line with arguments already presented concerning the importance of context in the analysis of
speech acts, van Lier (1988) suggested that the single axis didactic-conversational framework was
not sufficient for the analysis of second language classroom data. He proposed a second axis which
distinguished repair which is designed to help, enable and support (conjunctive) from repair which is
designed to evaluate, challenge and contest (disjunctive). An adaptation of this model was
particularly useful for the present investigation in order to make the distinction between co-operative
and nonco-operative repair sequence types. My analysis is concerned with the identification of
problem sources perceived by the task participants, and the type and frequency of their responses to
these problem sources. The actual mechanism of repair types has not been investigated in terms of
the sequential aspects of repair behaviour (the extent to which the problem under repair is manifested
in a sequence of turns throughout the transcripted segment). A sequential analysis requires a
classification framework which includes the dimension of turn, where the following trajectories for
repair procedures are suggested: (a) same turn, (b) same-turn transition space, (c) next turn, (d) third
turn. One single repair procedure takes place within a sequence of minimally one turn and (normally)
a maximum of three turns (van Lier 1988).

The frequency of occurrence of specific repair patterns is a particular focus within the present
research design and a sequential analysis has not been attempted, except during some instances of
repair where the sequencing pattern reveals features necessary for coding decisions. There would, in
any case, be severe constraints on the number of transcriptions which could feasibly be analysed with
complete sequential descriptions of repair procedures.

Task performance coding

The following transcription coding for the transcripts relates to the constructs and operational
definitions outlined above. Each code is briefly explained in relation to the apparent problem source
it relates to and, for clarity, a prototypical example of a participant solution is given (see also
Chaudron 1988: 45; Allwright & Bailey 1991: 123). There are important questions of data
compromise associated with all transcription coding decisions, and in focusing on the appearance of
prescribed prototypical research constructs in the transcripts, other important data can be missed.
This is a general problem with experimental research, which is possibly best solved by regarding data
such as task performance transcripts as both deductive evidence and sources of further hypotheses
(inductive).

Coding for negotiated meaning

Co-operative content repair (CCR)

Apparent problem source: the language generated by the task input content or participant interaction
during the attempt at task completion. A solution is attempted co-operatively.

51
CCR example from task A2.1
A24 glossary ... ah yes ... the vocabulary ... yes
B25 what is the course glossary?
A26 explaining the course component
B27 yeh I have the course component
Figure 3: Co-operative content repair.

Co-operative procedural repair (CPR)

Apparent problem source: the procedure of the task or the attempt at task completion. A solution is
attempted co-operatively.

CPR example from task A2.4


A66 I want to know about the time ... er ... where ... what
time stop this game?
B67 what time stop this game? I remember that the game
is for about thirty five minutes.
Figure 4: Co-operative procedural repair.

Evaluative content repair (ECR)

Apparent problem source: the language generated by the task input content or participant interaction
during the attempt at task completion. A solution is attempted by other-initiated performance
evaluation. Note that this evaluative response can also signal an unco-operative attitude to task
completion.

ECR example from task A2.6


A57 Examiner
B58 wrong pronunciation
A59 ok
Figure 5: Evaluative content repair.

Evaluative procedural repair (EPR)

Apparent problem source: the procedure of the task or attempted task completion. A solution is
attempted by other-initiated performance evaluation, although in this transcript extract, the question
of co-operation would have to be investigated through post-task retrospection since the B63 turn
below indicates a clear direction to help solve the procedural problem and the A64 contingent turn
appears to indicate a degree of placation.

EPR example from task A2.6


B63 no need to give me such detailed information ... only
fill the blanks
A64 ok ok ok but you tell me if you think the information is
not useful ... I try to give you
Figure 6: Evaluative procedural repair.

52
Comprehension check (COM)

A comprehension check is the speakers query of the task partner to see if they have understood what
the speaker has just said, typically Do you understand what Im saying? The apparent problem
source is content language or procedure.

COM example from task A2.3


A92 I just jot the information you give me but it is not so
clear ... you understand?
Figure 7: Comprehension check.

Confirmation check (CON)

A confirmation check is the speakers attempt to check understanding in relation to a previous


utterance (usually adjacent) of the task partner. A typical example would be Oh, so what youre
saying is ... The problem source is either content language or procedure.

CON example from task A2.7


A27 err ... term three you mean?
Figure 8: Confirmation check.

Clarification request (CLA)

A clarification request is a direct request for further information or help in understanding something
the task partner has said previously. Since it is a request for new information rather than a check on
prior information, it is different from the confirmation strategy above. A typical example would be
What do you think about the candidates experience? The problem source is either content
lan-guage or procedure.

CLA example from task A2.7


B7 what is this selection about?
Figure 9: Clarification check.

Repetition (REP)

A repetition is a morphosyntactic copy of a previous string of speech, either in part or completely,


and either self- or other-initiated. The apparent problem source is either content language or task
procedure. There are various terms in the literature which appear to refer to the same discourse
feature echos, repetition strategies or segmentations. They have been the subject of recent research
which identifies them as important strategies for peer group interlanguage development (see Pica et.
al. 1996).

REP example from task A2.7


A13 there is only one empty box
B14 only one empty box
Figure 10: Repetition.

53
Avoidance response (AVO)

A type of communication strategy, which indicates that the speaker has a temporary or permanent
inability to continue the task. In this sense, it is sometimes taken to indicate communication
breakdown. Typically, the speaker makes only a minimal response or openly admits
incomprehen-sion.

AVO example from task A2.5


A9 no I dont know ... I ... oh ... no ...
Figure 11: Avoidance response.

Overt expression of difficulty (OED)

This category appeared frequently in research transcripts as a type of avoidance strategy which acts
as a tension reliever or braking device on the progress of the performance. Typically, one speaker
will openly admit that he or she finds the task too difficult. The problem source may be either task
content or task procedure.

OED example from task A2.3


B84 its so complicated!
Figure 12: Overt expression of difficulty.

Research variables

There are five possible types of variable in this study. The task types are independent variables which
were hypothesised to have an effect on the dependent variables of turn and strategy features. The
control variables selected to be kept constant for the main quantitative study are given below.
Moderator variables are special types of independent variables, selected for their hypothesised
ability to affect any relationship between dependent and independent variables. In this study, selected
task design input elements, including the important element of task topic, were investigated as
possible moderator variables. Intervening variables are also hypothesised to affect the relation-ship
between variables but are not directly observable. Important affective concerns such as learner
motivation would come under this category in this investigation.

Output on task

This was initially defined in terms of turn features and measured by the number of words per task
performance segment (15 minutes). The quantification included words per turn, total turns, turns on
task, procedural turns, content turns and turns off task.

Learner perception of difficulty

This was initially measured using participant strategy usage associated with the process of negotiated
meaning, and taken as being marked by repair activity in the transcripts. The variables were coded
CCR, CPR, ECR, EPR, COM, CON, CLA, REP, AVO and OED.

54
Learner management of interaction

This was measured in the quantitative investigation through participant usage of co-operative
communication task completion strategies.

Moderator and control variables

In research tasks AE, the moderator variable specifically recognised was the number of task design
input elements. These were adjusted specifically for each type, theoretically to increase the task
complexity level, for example, increasing the number of candidates from two in task C2 to four in
task C2. Subject proficiency was not tested as a moderator variable, since it was assumed to be
controlled by the narrow range of Use of English scores accepted into the intact groupings by the
University administration. Gender was not controlled for in this investigation, since this would have
compromised the principle of using intact groupings (scheduled language classes were not
single-sex). Gender, however, can be hypothesised to be a moderator variable, although Fordes
(1996) study suggests that peer-group assessment ability is not strongly influenced by gender or by
proficiency, which therefore provides some support for the decision not to complicate the research
design further by formally including gender and proficiency as moderator variables. There is,
however, evidence from this study and others to suggest that proficiency would justify further
investigation as a moderator variable.

Triangulation secondary data

Methodological triangulation using retrospective interview data and post task questionnaires was
facilitated through qualitative case studies. This provided a necessary check for the research
constructs associated with co-operative task completion strategies, as well as facilitating an
investi-gation of the intervening variable of participant motivation on task. Following Cohen (1984),
the data capture was from interview procedures designed to elicit self-observation of specific
task-related language behaviour. Data capture was from delayed-retrospection but in relation to
participant interviews with a full transcript of the task performance and excerpts from the audio
recording for specific investigation concerns. Post-task motivation questionnaires completed the
methodological triangulation, and provided initial information for the post-task interviews.

Research procedure

Since the design of the quantitative study was experimental, in relation to task treatments, specific
subject and task controls were necessary to provide an element of standardisation in the outcomes
investigated. The research setting was the normal language classroom domain and research tasks
were designed to parallel pedagogic types being used in scheduled course materials.

Subject control variables

(i) General proficiency (Hong Kong Use of English A level results) D and E grades.
(ii) Ethnic origin all subjects were born and educated in Hong Kong.
(iii) Age all subjects were aged between 19 and 21.
(iv) Education subjects had received all their education in Hong Kong.

55
(v) Task performance setting all subjects were recorded under normal lesson conditions.

Standardised task implementation controls

Task complexity

Number of task goals, text/graphic input to task, task information loading, assumed cognitive
demand (static, dynamic, abstract), assumed degree of communicative responsibility.

Amount of context provided prior to task

Prior knowledge of the world (the situation or the cultural context that is assumed in the task input),
pre-performance task instruction, preliminary activity to introduce the task and set the context.

Processibility of language of the task

Task input instruction syntax and vocabulary (Fleisch readability rating, Microsoft Word).

Amount of help available to the learner

Assistance learner could get from instructor, other learners, books or other learning aids.

Output goals

Prescribed (except in divergent discussion tasks).

Time available

Input text reading time, time on task.

Setting

Task participation situation, task group size, mode of interaction.

Completion

A maximum time of thirty minutes was given in some cases where prescribed goals had not been
attained in a standard completion time of fifteen minutes.

Familiarity with task partner/personal learning style

Where possible, task dyads were made up of learners who did not know each other well.

56
Transcription procedure

Exact intervention points in research recordings for transcription purposes are always difficult to
establish. The final decisions have to be data-driven in relation to the research objectives. From an
examination of the pilot study data, transcription time was fixed at fifteen minutes from the start of
participant interaction on task (i.e. not including input reading time). This decision was made for the
following reasons:

1. To standardise task performances of varying length to the minimum length recorded on any
one task type.

2. Fifteen minutes is a standard length for many evaluated oral task performances e.g. the
HKEA Use of English oral examination task (part 2), the IELTS oral examination, the
UCLES Certificate of Advanced English.

3. Since dyad groupings are from intact classes, during normal English timetable allocations,
it was not feasible to take more lesson time on task recording, since this would have
jeopardised required core course material in many cases.

4. During data analysis at the pilot stage, fifteen minutes was found to be sufficient to capture
features of turn-taking and strategy use being investigated. Turn-taking data in particular
required a standardised performance length if data were to be comparable across tasks.

5. The resulting total transcript time represented a maximum feasible transcription time for
effective data management, coding and reliability rating purposes.

Data analysis

The analysis of the data from the quantitative study was completed with the Statistical Package for
Social Sciences (SPSS) software. SPSS was also used for the coding reliability analyses.

General procedures and assumptions

The quantitative investigation in this study is concerned with the variates of word, turn and attributed
strategy use. Quantitative variates can be either discrete, or continuous. The data of this investigation
is discrete, since it can be regarded as finite and the values are integers.

Multivariate analyses of variance (MANOVA) are used to compare statistical means between groups
when two or more dependent variables have to be considered simultaneously within the research
design. As such, they are an important method for descriptive research requirements in-volving
multiple variables. The results of the analyses can be converted into F ratios and tested for
significance using appropriate statistical tables or software such as SPSS. With this latter method, the
F value alone does not reveal significance. With computed statistics, the computed probability is the
indicator: the smaller the probability, the more significant the differences between means. Generally,
the probability figure depends on the observed figure for F and degrees of freedom of the F
distribution.
The F ratio is an established approach to research questions that require comparison between means.
It is equivalent to the variance between groups divided by the variance within the same groups. The

57
ratio departs further from 1.0 as the difference between groups increases relative to the variation
within the groups.

Various terms are used in the literature for the entities in a multivariate data set. The experimental or
observational units are referred to as subjects. The measurements or observations made on each unit
are collectively known as variates or variables and a distinction also has to be made between
measurement and observation by using the terms responses and attributes (Freund 1984: 515). In this
study, an important assumption is that there is no statistical connection between each of the subjects
and since it was ensured that no subject appeared in more than one observation, different units were
therefore considered to be independent for statistical purposes. If the values are not independent, then
much bigger samples are needed to assume that the sample means are normally distributed. Degrees
of freedom (df) are an essential part of the calculation of the task-based statistics since the df is a
measure of the independence of the data set, and in order to obtain an F ratio, we need the numerator
and the denominator degrees of freedom (N1). Reference is made in the research data analysis to
statistical values such as Fcrit and Fobserved. The critical value of F is the value that we might expect to
obtain from the sample through chance alone, and is obtained from a computed reference table for
values of F, whilst the observed value of F is the value calculated by SPSS.

Probability p (or alpha level, if prescribed) is the ratio of the expected outcomes to the number of
possible outcomes. The ratio varies between 0 and 1.0 and is sometimes expressed as a percentage,
or, in many language studies as an alpha decision level of p < .05 (a toleration of up to 5% error).
There is a weakening convention to form statistical inferences at just that point where a statistic has
been found to have a probability of less than a prescribed decision level. This is due to the in-creasing
use of statistical software and the acceptance that our confidence in results might vary in relation to
the type and purpose of the research. In behavioural research, there is a strong case for accepting as
significant higher values of p than we would want to in relation to research results in physics. In this
study, the value of p for each statistic has been calculated by the SPSS program. When this is done,
there is variation in the computed values of p, reflecting the low frequency counts in some data
categories (e.g. OED and AVO strategies). In these cases, it is not possible or desirable to be more
certain about the statistic.

Analysis of variance and hypothesis testing

This standard multivariate statistical approach presupposes that we already have a general idea of
how the oral task output data is expected to behave (sometimes referred to as a stochastic model of
the relationship between task and task outcomes). The test of an overall null hypothesis is made by
computing an F-ratio value comparing group means. The largest possible F value is the basis for the
significance test of the overall null hypothesis. If the variation between the groups becomes much
larger than the variation within the groups, the F ratio can lead to a decision to reject a null
hypothesis.

A multivariate analysis of variance (MANOVA) is a generalisation of analysis of variance to a


situation in which there are several measures (dependent variables) for each subject. In this study,
there are seventeen dependent variables for each task type. A MANOVA is therefore used to examine
whether these variables vary as a function of the research tasks. It allows us to test whether mean
differences among tasks are likely to have occurred by chance. Multivariate normality is an
important assumption that each variable, and all linear combinations of the variables, are normally
distributed. However, this assumption of multivariate normality cannot be directly tested, since it is
not possible to test an infinite number of linear combinations of variables for normality. In

58
uni-variate cases, the central limit theorem can be used, which suggests that the sampling distribution
of means approaches normality for large samples. MANOVA is robust to a modest violation of
normality if the violation is not caused by isolated, eccentric results (outliers). A further important
assumption is that the variancecovariance matrices within each group are sampled from the same
population.

An emergent problem in the research design derived from a potential violation of this assumption. If
one oral task has more variables than transcripted performances, the assumption of the homogeneity
of variancecovariance matrices is compromised. Since each of the ten research tasks had twelve
performance transcripts and seventeen dependent variables, it was decided to break down the
variables into two distinct data sets which could then be analysed separately. One data set consisted
of the dependent variable repair strategies, and the other of the word and turn dependent variables.
Since there were now effectively two independent MANOVAS, a Bonferroni adjustment was
required to test for inflated type 1 error. SPSS can provide this adjustment, which is essentially a
modified comparison procedure based on the task output statistics. The observed significance level
was adjusted for the fact that multiple comparisons were being made. To keep the overall alpha level
below 0.05, individual alpha levels were pre-set for this adjustment at 0.025 for each group. (I am
indebted to the Department of Mathematics, UST, for pointing out the need for this specific treatment,
and advising on its correct implementation.)

Analysis of negotiation of meaning

The discourse features associated with the concept of negotiation of meaning are relatively well
known from discussion in the literature. Negotiation sequences in the research transcripts were
assumed to be indicated or marked by repair strategy usage (see Duff 1986; Fotos 1994). A
description of negotiated meaning should include not only a description of the use of such strategies
as comprehension checks and clarification requests, but also a description of what Rulon &
McCreary (1986) have termed negotiation of content. This is defined as the learners attempt to
manage the topic of the exchange in order to understand not only the other interlocutor(s) but also the
task topic itself.

Analysis of topic management

Topic selection and control are important analytical perspectives, although the definition of topic as
a unit of discourse is still problematic, as Brown & Yule (1983b) report. Most definitions in the
literature describe orientation to topic as orientation to the main subject of the exchange Brown &
Yules notion of what is being talked about (ibid.: 71). With this starting point, the quantitative
analysis of the performance transcripts was divided into two main areas for the purpose of topic
management analysis:

(a) Procedural aspects of the task task-related meaning, goals to achieve, procedural
achievement strategies and constraints. Typical procedural appeals were of the What do
we have to do ...? type.
(b) Content aspects of the task negotiation about the enabling concepts and language that are
required to successfully complete the task. This is illustrated by lexical appeals for example,
which can be either direct, such as What is this word? or indirect such as I dont know
the word for this (note that in this case, it could also be classified as an avoidance strategy).

59
Sampling

The quantitative investigation here records the values of selected responses and attributes from a
sample of the first-year undergraduate population. In terms of sampling methodology, the sample
used can be classified as simple random, where every subject in the reference population of first-year
language enhancement students was equally likely to have been selected for inclusion in a research
dyad. The intact groupings were assigned by computerised class scheduling processes which formed
homogeneous groupings for all sections of the language enhancement programme. We can therefore
assume that individual subjects were independent and uncorrelated, which is a necessary assumption
for most methods of statistical analysis, and particularly important for multivariate analysis. Given
the restrictions of the research design, practical teaching concerns and data handling constraints, the
sample size was 240 subjects in intact groupings of 120 dyads. Each dyad performed one of ten
research task types, generating a total of 12 research task performances for each task type.

Coding reliability

There are specific problems of reliability with both quantitative and qualitative procedures for data
capture. The major problems derive from the set of variables associated with strategy use, since these
operationalise as highly inferential constructs. To compensate for this subjective potential, some
form of triangulation, usually data or methodological, must be employed, and coder reliability
measures need to be established before the main data analysis is completed so that adjustments in the
data handling procedure can be made if the reliability measurements are low.

A more pervasive problem is in relation to the generalisability of results. It might be argued that in
order to conduct the research, the interactive language tasks are too research specific not all oral
tasks are performed in the specific domain of this research. The representative nature of the task
types themselves may also be questioned, particularly by those who might argue that all task
performances are unique events.

Inter-rater or coding reliability is essential to establish in this type of research. It was estimated by
employing a similar method to that of Berwick (1993), which involved training a co-rater and
establishing inter-rater reliability measures at the end of the training process. This was the only
practical method, given the large amount of data generated by the investigation. Before the main data
analysis, both the co-rater and I coded the same random selection of one transcript for each task type
(10 transcripts) using the established task performance codes. After coding, and a series of meetings
in relation to coding perceptions, mutual agreement was eventually reached on all coding questions.
Inter-rater reliability was calculated by taking the frequency count scores from task-type data sets for
myself and the co-rater, and calculating a correlation coefficient between them using a Spearman
correlation for rank ordered data (see Hatch & Lazaraton 1991: 533).

This computation was performed on two data sets, the first set consisting of strategy markers
associated with content and procedural repair (CCR, CPR, ECR, EPR) and the second for task input
clarification and comprehension strategies (COM, CON, CLA, REP, AVO, OED). Spearman
cor-relation coefficients were .9383 and .9902 respectively. The high correlation on the input
clari-fication and comprehension set indicates a high degree of agreement on the occurrence of
strategies such as clarification requests in the transcripts. This is possibly because this construct is
more clearly marked in the transcript. The slightly lower, but still acceptable measure for repair

60
features is possibly due to some differences in perception in terms of the operational definitions of
the data set. For example, differences of opinion on whether the occurrence was essentially
evaluative or co-operative, and whether the topic management was procedural or related to the
enabling language of the task.

Results

The results of the quantitative study are given below, in relation to each set of research hypotheses.
Table 5 below gives the summary MANOVA statistics for all comparisons.

Table 5: All variables associated with usage of communication strategies.


Comparison Wilks lambda F df p
Overall 0.01895 6.1380 90,695.27 0.0001
Tasks A1/A2 0.6901 4.5361 10,101.00 0.0001
Tasks B1/B2 0.5574 8.0195 10,101.00 0.0001
Tasks C1/C2 0.6064 6.5549 10,101.00 0.0001
Tasks D1/D2 0.8720 1.4821 10,101.00 0.1570
Tasks E1/E2 0.7486 3.3919 10,101.00 0.0007

Wilks lambda is also given here as a measure of the proportion of total variability not explained by
group differences. Wilks lambda can also be transformed to a variable that has an F distribution i.e.
the figure for F can be derived from lambda.

Table 6: All variables for learner output on task.


Comparison Wilks lambda F df p
Overall 0.0458 6.8484 63,591.84 0.0001
Tasks A1/A2 0.8630 2.3583 7,104.00 0.0282
Tasks B1/B2 0.7105 6.0538 7,104.00 0.0001
Tasks C1/C2 0.6798 6.9975 7,104.00 0.0001
Tasks D1/D2 0.9152 1.3766 7,104.00 0.2230
Tasks E1/E2 0.8765 2.0931 7,104.00 0.0506

Hypotheses in relation to learner output

1. There are no significant differences between learner outputs and different task types. (null)

2. There are significant differences between learner outputs and different task types. (alternative)

From the operational definitions adopted, learner output was measured using the number of words
per task performance segment (15 minutes), words per turn, total turns, turns on task, procedural
turns, language turns and turns off task. Comparing all variables associated with learner output, the
observed value of F was much larger than a critical value of 1 at 6.8484 (p = .0001). Therefore, we
can reject the null hypothesis and accept the alternative that there are significant differences between
learner outputs and different task types.

Hypotheses related to learner perception of difficulty

5. There are no significant differences between learner perceptions of task difficulty and different

61
task types. (null)

6. There are significant differences between learner perceptions of task difficulty and different task
types. (alternative)

Learner perception of difficulty was initially measured with strategy use marked as repair activity in
the transcripts. This activity was initiated either by one participant or between participants when
difficulties were perceived. Comparing all variables associated with learner usage of negotiated
meaning strategies, since the observed value of F at 6.1380 was larger than an Fcrit of 1.0, we can
reject the null hypothesis and accept the alternative that there are significant differences between
learner perception of difficulty and different task types.

Hypotheses associated with learner co-operative management of interaction

9. There are no significant differences between task types and learner employment of co-operative
communication strategies. (null)

10. There are significant differences between task types and learner employment of co-operative
communication strategies. (alternative)

Learner management of interaction was measured using co-operative task completion strategies;
CCR (co-operative content repair) and CPR (co-operative procedural repair). Comparing these two
variables associated with learner management of interaction, the observed value of F for CCR was
larger than an Fcrit of 1.0 at 5.61. Therefore, we can reject the null hypothesis and accept the
alternative that there are significant differences between task types and learner co-operative
management of interaction.

Hypotheses associated with task complexity

Six of the investigation hypotheses were concerned with the effect of task complexity on task output
and learner perception of difficulty. Task complexity was altered by changing the number of task
goals, the text/graphic input to task, the task information loading, the cognitive demands (static,
dynamic, abstract) of the task or the degree of communicative responsibility. For example, the A2
task was hypothesised to be more difficult than the A1 version because it had four flowchart boxes to
complete rather than two. The hypotheses are given below (with the original numbering from page
46).

3. There are no significant differences between learner outputs and tasks of different complexity.
(null)

4. There are significant differences between learner outputs and tasks of different complexity.
(alternative)

7. There are no significant differences between learner perceptions of task difficulty and tasks of
different complexity. (null)

8. There are significant differences between learner perceptions of task difficulty and tasks of
different complexity. (alternative)

62
11. There are no significant differences between learner employment of co-operative
communi-cation strategies and tasks of different complexity. (null)

12. There are significant differences between learner employment of co-operative communication
strategies and tasks of different complexity. (alternative)

The different effects between the two versions of each task were tested separately, and results were
mixed. This was expected, in the sense that the design features selected to increase complexity were,
in each case, selected on a hypothetical basis. For the communication strategy and repair features,
there were significant differences between the two versions of tasks A, B, C and E at an Fcrit = 1.0.
However, comparing both versions of task D, no significant difference was found. Again, this was
not surprising, since the only difference between the two versions of the D task were a description of
roles in D2, which were assumed to be more distant in terms of participant familiarity. The
inconclusive nature of this result is reflected in the higher computerised probability figure for these
tasks (0.157).

For the group of variables associated with learner output, only task types B and C showed signi-ficant
differences between versions at p = 0.0001. The most significant performance differences were on
the C tasks, where the complexity factor tested was the number of candidates to consider for selection
(and hence an increased textual/cognitive input) in order to complete the task. This is interesting, and
supports other studies which have focused on task information loading as a significant factor in
participant perception of task difficulty.

In terms of task complexity and the use of co-operative strategies, the null hypothesis that there are
no significant differences between learner employment of co-operative communication strategies
and tasks of different complexity has to be accepted for C, D and E tasks. Co-operative procedural
repair (CPR) strategy use revealed expected significant differences in relation to task complexity on
type B (jigsaw) tasks (Fobs 40.3 at p = 0.0001), where the second version of this task involved
participants exchanging information about two different sets of graphical clues rather than working
on the same clues, as in version 1. For B tasks, the alternative hypothesis that there are significant
differences between learner employment of co-operative communication strategies and tasks of
different complexity, can be accepted.

Task-related output

In this investigation, task-related output has been defined in terms of the measurements associated
with two distinct statistical groupings. The first group is related to what was defined as learner output
(utterance and turn counts) and the second to participant strategy usage. In this study, multivariate
analysis of variance (MANOVA) procedures were used to examine the significance of differences
between the mean values generated from the performances of five types of tasks at two levels of
complexity. Comparing all tasks in relation to strategy use, overall significant differences were found
between performances on the range of task types (F = 6.138, p = 0.0001).

Words per turn and turn lengths

The statistics also facilitate descriptive comparisons of means between task types in terms of
utterance and turn characteristics. As expected, discussion-type tasks (D and E) show longer turn

63
lengths and a greater number of words per turn than jigsaw or information-gap tasks.
Information-gap tasks are characterised by greater amounts of repetition and clarification strategies.

Negotiation of meaning

For the group of strategy markers associated with input comprehension and negotiation of meaning,
repetition (REP) was the most important variable contributing to significant differences between
tasks. This result supports findings from Pica et al.s (1996) study investigating segmentation. The
other variables, in order of significance, were co-operative procedural repair (CPR), evaluative
con-tent repair (ECR), clarification checks (CLA) and comprehension checks (COM). Core
strategies associated with negotiation of meaning appear more frequently in information-gap and
jigsaw-type tasks. In general, these types of task seem to generate more negotiation of meaning in the
terms of this study, the more complex versions of each task type generating higher frequencies of
these figures in each case except D-type tasks.

Performance differences in relation to task complexity

Significant differences were found between the performances on A, B, C and E task types, when
comparing the easy and difficult versions of each task type. In each case, this significance
ap-peared to depend on two variables only. For A tasks, the F ratio showed a significant difference
between A1 and A2 task performance. The difference appears to depend on the strategy variables
CLA and OED. For B tasks, the F ratio showed a significant difference between B1 and B2 task
performance. The difference appears to depend on the strategy variables CPR and REP. For C tasks,
the F ratio showed a significant difference between C1 and C2 task performance. The difference
appears to depend on the strategy variables ECR and REP. No significant difference between
performances on either version of D tasks was evident. Lastly, for E tasks, the F ratio showed a
significant difference between E1 and E2 task performance. The difference appears to depend on the
strategy variables CON and REP. The REP strategy variable appears to be quite influential in terms
of explaining the differences between performances on all task types except D tasks.

For the group of turn measurements, turns off task were not affected by task differences after
adjustment for words, turns on task, procedural turns and language turns. The variable of turns on
task was the most important affecting significance when comparing the more complex and less
complex versions of tasks B and C. The results of a separate canonical variate analysis also indicate
graphically that for the word and turn group of variables there are significant differences between the
two versions of task B and task C.

64
These results suggest that in some cases, with notable exceptions such as on D tasks, increasing the
level of complexity by manipulating one design feature, did result in significant differences in learner
performances on task. Overall, all variables appear to be influential, with the exception of evaluative
procedural repair (EPR), avoidance (AVO) turns off task and total words per turn. These results tend
to support the general hypothesis that there are significant differences between learner output and
tasks of different complexity.

Performance differences in relation to task type

Pica & Doughty (1985b) found that significantly more negotiation of meaning occurred in
information-gap than opinion-gap activities, both in group work and in teacher-dominated lessons.
The results from this investigation support this finding, with evidence of more negotiated meaning
from information-gap and jigsaw activities than from opinion-gap, decision-making and discussion
activities.

Discussion of results

A fully developed descriptive framework must allow for a consideration of objections which focus
on the central difficulties of determining learner motivation in relation to strategy employment. A
fundamental problem for any descriptive framework of learner task-based performance is that of the
multiple function of utterances. Levinson (1983) suggested that it is impossible to characterise an
utterance as performing a certain speech act because it often has more than one function. While Tsui
(1994) saw this more in terms of strategy ambiguity than multifunctionality, it is still the case that a
quantification of utterances in terms of their assumed illocutionary force requires corrobora-tion
from the learners perspective. We need to know the function of particular discourse features
supposedly indicating negotiation of meaning during task performance in order to validate or
other-wise the underlying assumption of a relationship between linguistic form and illocutionary
force.

Our research task typology provides a basis for the comparison of dyadic task types at different
complexity levels, in terms of particular discourse features associated with the concept of negoti-ated
meaning. In this sense, the framework offers a measure for the evaluation of such tasks in terms of
their effectiveness in stimulating this type of interactional response. It also facilitates an investigation
of possible relationships between pedagogy and actual lesson outcome between task design and
learner performance. However, determinism should be avoided in resulting descriptions because it
would not be feasible to include all the possible intervening variables which might be hypothesised
to influence the outcome, and which would not be subject to measurement. The output patterns
already reported here show significance in relation to the type of output and specific oral task types,
performed in a particular performance domain. With regard to generalisability, our claim is that the
task types are generalisable to similar pedagogic domains, and that the broad consistency in their
respective performance outcomes does provide a pedagogic basis for the selection and design of
tasks intended to stimulate beneficial interactive language work between learners.

These findings also enable us to suggest a tentative connection between the design variables of task
complexity and the resulting output, although the results from the present investigation were mixed.
For example, the role of topic in the D-type tasks is clearly complex in terms of its influence on the
perception of task difficulty by task participants.

65
Fotos (1994: 343) conclusion that negotiation quantities were promoted differentially through the
manipulation of task design features, and that the greatest amount of interaction (negotiation of
meaning) was produced by the combination of an information gap and the requirement for a single,
agreed-upon task solution is also supported by the results of this study. The findings of Pica et al.
(1996) concerning repetition strategies are also supported. Repetition (REP) was the most important
variable contributing to significant differences between tasks. It may well be, as Pica et al. have
suggested, that repetition (or segmentation) is one of the most important strategies used by learners
to overcome perceived difficulty. In a previous survey of research on the amount of output produced
by manipulating similar variables, Long (1989) also identified information-gap tasks as the most
productive in this sense, supporting Doughty & Picas earlier (1986) findings. Shortreeds (1993)
finding of increased negotiation of meaning from more complex tasks is also supported by the results
of the present study.

This investigation acknowledges that deterministic approaches are probably going to be of limited
use, particularly when it is impossible to control for sociological and psycholinguistic factors.
How-ever, the quantitative study indicates that it is possible to establish a detailed descriptive
framework which, while not establishing causal connections, enables us to determine more precisely
just what it is that we might expect learners to do with and to learn from oral communication tasks.

66
67
Chapter Five: The problem of negotiated meaning

Introduction

The strategies used by participants during the performance of oral tasks can be seen as attempts to
prevent and repair breakdowns in communication in order to sustain task performance. They can also
be seen as part of a process to develop a mutually acceptable belief system about the task and its
performance (Wilkes-Gibbs 1997). In order to utilise these strategies effectively, the learner must
have sufficient strategic competence, which is the ability to use linguistic resources to get a point of
view across successfully to a particular audience. The concept can be broken down into two broad
conceptual areas (Yule & Tarone 1990):

(i) the overall skill of a learner in successfully transmitting information to a listener or


interpreting information transmitted;
(ii) the use of communication strategies by a speaker or listener when problems arise in the
process of transmitting information.

The controversy concerning the amount and type of communication strategies we should recognise
is one that cannot be settled here. A broad division is evident between theorists who favour
typologies consisting of only two types reduction and achievement strategies (Frch & Kasper
1983) and those who conceptualise only compensatory strategies (see Bongaerts & Poulisse 1989
for a fuller description). In this study, I have identified and utilised selected reduction and
achieve-ment strategies, including confirmation checks, comprehension checks, clarification
requests, re-petitions, expansions and enquiries about task content and procedure. These strategies
have been closely related to the concept of negotiated meaning and to learner strategic competence.

Tarone (1984) described the interactional element of strategic competence in much the same way as
negotiated meaning as the mutual attempt by two interlocutors to agree on a meaning. However,
this definition is a performance definition, not one which relates to a theoretical competence base,
since, as we have seen, Tarone has been mainly concerned with the effect of observed strategy forms
on a learners interlanguage development. Canale & Swains (1980) concept of socio-linguistic
competence seems close to that of strategic competence, and formed part of their model of
communicative competence, which also consisted of separate conceptual components in relation to
grammar and discourse. Sociolinguistic competence entails mastery of speech act conventions,
norms of stylistic appropriateness and the use of language to establish and maintain social relations
within a perception of participant role and status relationships.

Learner perceptions of the usage and distribution of social power within an interaction are an
important consideration in the analysis of strategic competence and negotiation of meaning. For
example, Knowles (1975) and van Lier (1994) have both suggested that when learners perceive
themselves to be in subordinate learning roles (perhaps as a result of a perception of unequal social
status in an exchange) effective learning is not promoted. In oral task performance, we cannot
assume that task participants do not have conscious or unconscious perceptions of the distribution of
social power within the group.

In the performance of oral communication tasks, this perception is also likely to be knowledge-
related that is, related to perceptions by participants of interlocutor knowledge of the target
language and the task itself the background knowledge which the learner brings to the
inter-pretation of discourse (Brown et al. 1989: 38). This perceptual knowledge could be expected to

68
undergo modification throughout the task performance, as participants gain experience of each other
and the task. We can expect peer assessments of respective inequalities in proficiency and motivation
to influence the method and the motivation of their negotiation of meaning (see also Brown & Yule
1983a: 34 on this same point).

How then, might we characterise task interaction from the point of view of task participants? Is it
largely a constant calculation and recalculation of respective advantage, or is it to be seen more as a
manifestation of sociolinguistic competence or the lack of it? Firstly, there is a strong sense in which
all meaning in all conversational exchanges can be said to be negotiated. The work of Grice (1975)
was seminal in this respect, and contains two concepts which are relevant both to the notion of
negotiated meaning and relevant to its analysis in discourse. The first is that the speakers intention
should be a fundamental consideration in terms of propositional meaning, and the second is that
orderliness and co-operation in a conversation provide a principled basis for understanding its
dynamics. This orderliness (Fairclough 1995: 28) is due to the conformity of interlocutors to an
implicitly understood framework of discoursal rights and obligations, involving turn-taking, topic
control, rights to question and obligations to answer (see also Sinclair & Coulthard 1975). The
concept of negotiated meaning as a co-operative attempt by two interlocutors to reach a consensus on
interpretation rests, initially at any rate, on this principled theoretical base. Grices four maxims, or
principles, of co-operative behaviour concerning the quantity, quality, relation and manner of talk
have more recently been conflated (Sperber & Wilson 1986) to the single principle of relevance to
the ongoing discourse, a route Grice sometimes appeared to take himself (op. cit.: 45):

Our talk exchanges do not normally consist of a succession of disconnected remarks, and would not be
rational if they did. They are characteristically, to some degree at least, co-operative efforts; and each
participant recognizes in them, to some extent, a common purpose or set of purposes, or at least a
mutually accepted direction.

Grice refers to the four specific maxims as being under a more general co-operative principle (ibid.:
45). Despite the term maxim, it is not always clear whether these categories are conversational ideals
or taxonomic devices, but these assumed implicit agreements, if adhered to, must also leave room for
differing interpretations of such conversational principles on the part of task participants. The basis
for these differing interpretations is, firstly, the participants own personal views on the issue(s)
being negotiated, as well as those which each listener attributes to each speaker. The attribution of
meaning is also a personal response to a specific communication situation and it is never possible for
two interlocutors to reach exactly the same understanding with regard to the meaning of task
performance features. In this sense, negotiated meaning ought to be seen more as the establishment
of a mutually agreed consensus often, by task participants implicitly agreeing to disagree in order
to finish the task.

This conception seems close to more recent discussion concerning collaborative theory
(Wilkes-Gibbs 1997) where, in trying to reach perceived task goals, participants are seen as trying to
add to their common beliefs about the task and the task process in a way that is mutually intelligible.
This is seen as both an opportunistic and a strategic process where task participants are constantly
calculating gain and loss in the collaborative act.

This perspective might then cast some doubt on the idea that interlocutors are always co-operatively
negotiating the meaning of task input. They may also be following personal agendas which have
more to do with expediency accepting a loose consensus of meaning rather than a mutually
agreed exact definition, in order to move on to other concerns. van Lier (1996: 46) drew attention to
the consideration that comprehensibility in this context is often (mistakenly) taken as a monolithic

69
property of the language, which is either present or absent. Task participants negotiate a clearer
mutual conception of task-related problems, but there is no reason to suppose that they ever achieve
an ideal state of full comprehension of the task, however that might be finally realised. Their
respective personal agendas might also result in what Bly (1993) termed unco-operative
negotiation, where a mutual concern for task completion is not shared to the same degree. This point
has been taken up by some critics of the task-related concept of negotiated meaning, who see it as an
interpersonal rather than a transactional process (see Aston 1986, 1993) and are therefore critical of
a basic assumption of comity.

These concerns might lead us to redefine the concept of negotiated meaning to state that the goal of
negotiated meaning for task participants is not one final, finite interpretation, but a constant clearing
of the path, as it were, to mutual acceptance and agreement. Pennington (1986: 56) appeared to
support this view by extending the concept of negotiated meaning to include the idea of
acces-sibility:

Accessibility, while incorporating the notion of comprehensibility, is a different notion, in that its
central concepts are motivation and memorability rather than ease of understanding. Information can be
said to be more accessible to students to the extent that it motivates them to learn, whether because of the
inherent nature of the content e.g. that it is particularly relevant in the learning situation or in their lives
or because of the way in which the material is presented.

This interpretation not only frees up the debate from a concern with either one meaning or multiple
interpretations, but also allows the concept of interlocutor motivation to be brought more fully into
the discussion. This is similar to van Liers (1996: 48) definition of the notion of accessibility which
gives prominence to participant motivation in negotiated exchanges, and where a theory of language
engagement replaces that of comprehensible input.

Strategies, moves and speech acts

When two learners monitor their own performance on task, their reaction to perceived problems
through the mechanism of repair has been termed negotiation of meaning and can be initiated by the
original identifier of the trouble source (self-repair) or by the other member of the dyad (other-repair).
Ellis (1984) used the term medium-oriented for repair work which appears to focus on the form
and/or functions of the target language. The term activity-oriented was used where the repair work
appears to be focused on the organisation and structure of the task. These broad orientations can be
assumed to influence the type of repair work undertaken by learners with respect both to what in the
performance is judged to be important to reflect on retrospectively, and to how repair adjustments
might be made. In our study, two broad categories of content repair and procedural repair are
considered.

Following Oxfords (1990) typology of learner strategy usage, this investigation utilises a further
broad distinction between direct strategies and indirect strategies. The former are assumed to be
evident to the learner and under the learners control, and include memory strategies, cognitive
strategies and reduction and achievement (compensatory) strategies which are specifically related to
strategic competence. Indirect strategies, not under the learners control, include metacognitive
strategies, related to learning style, and social and affective strategies. The possibility of research
access to direct strategies in particular was assumed in the qualitative part of this investigation (not
reported here). Direct strategies can be divided into interactional or non-interactional receptive
communication strategies, and would be subsumed under Oxfords (ibid.) cognitive and

70
compensation categories.

The term communication strategy relates to a mutual attempt between two interlocutors to agree on
a meaning in situations where requisite meaning structures do not seem to be shared (Tarone 1983).
Little is known in psychology about what constitutes a strategy (see Selinker 1972) but there has
been much speculation in applied linguistics concerning the term (see Frch & Kasper 1984; Yule
& Tarone 1991). Tarone (1978) originally argued that strategies are used whenever the learner
detects a deficiency in his or her strategic competence, or perceives a deficiency in linguistic
competence with regard to some aspect of the target language. This results in the conscious and
unconscious use of strategies to compensate for the mismatch between the interlanguage stage of the
learner and the target language goal.

Tarones definition of communication strategies is interactional defining the central function as


facilitating the negotiation of meaning. Frch & Kasper (1983) originally viewed strategies
psycholinguistically, as potentially conscious plans for solving a perceived problem in reaching a
particular communicative goal. In a later paper (1984), they argued that the interactional definition
as exemplified by Tarones (1978) definition of communication strategies, obscured some types of
strategy use by defining strategies too narrowly in relation to the concept of the negotiation of
meaning. Tarones definition of communication strategies employed an interactional aspect. Under
certain conditions, Communication strategies are used by a speaker to try to obtain agreement with
a listener on some aspect of negotiated meaning and are therefore primarily directed at task
participants rather than solely task language concerns.

The concept of participant strategy use is central to this study and an interactional definition has also
been adopted in which a communication strategy is taken to represent a mutual attempt by two
interlocutors to agree on meaning in a situation on task where initial agreement is not evident.

A further important distinction was made by Frch & Kasper (1983) between repair work and
communication strategies. In their view, repair work should be distinguished from the notion of
communication strategy since self-repairs, for example, reveal that the speaker has run into some
difficulty in executing his or her plan, and as such should be considered as post-execution plan shifts.
However, some disagreement with this position is evident from a consideration of the notion of
appealing, which Schegloff et al. (1977) regarded as an important co-operative communi-cation
strategy. As such, the frequency of occurrence of such features could be expected to indicate further
attempts to negotiate meaning around a task problem source.

This investigation assumes that repair-work procedures are types of communication strategies, since
it would be impossible to identify plan shifts accurately from either transcripted or introspective data.
Having identified a problem area, when a learner is interested in a negotiated solution to a perceived
problem, he or she employs an interactional or apparent self-repair strategy to attempt to solve the
problem. A non-interactional strategy is often employed in the interests of what van Lier (1988: 184)
termed avoidance of face threats. This occurs when one of the participants does not want to admit
that he or she is experiencing a communication problem. An extreme example of this type of
response would be avoidance behaviour, although even with retrospective interviewing techniques
this is difficult to identify in transcripts, since it may not be marked by overt forms, but may manifest
itself as an apparent lack of motivation for the task in general. If, however, interlocutors do adopt
overt interactional strategies for repair, then, as Luk (1992) has pointed out, this is often handled in
stages initially by general requests followed by a specific repair in some cases and then more
specific requests. There is also a third strategy, which is often apparent simply claiming ignorance.

71
This overt admission of problems appears to have two main functions. The first is to constitute a
minimal reply without initiating a repair sequence; the second is actually, in effect, to conceal a more
complex comprehension or production problem.

A serious threat to the validity of any research in this area of discourse analysis is the observation that
such strategic moves may have multiple functions, and also multiple realisations. These are also
central problems for speech act theory (see Leech 1983; Levinson 1983; Flowerdew 1990). Speech
act, a term which derives from the early work of the philosopher J. L. Austin (1962), is
conceptualised as a communicative activity, defined with reference to the intentions of the speaker
while speaking. Several categories of speech act have been proposed, and these are all considered to
be the minimal unit of speaking which can be said to have a function (Richards 1980: 417).

The potential multifunctionality of speech acts within conversations means that speech act analysis
is complex. This complexity is increased through the prominence given in speech act theory to the
issues of social power briefly discussed above. Here, the notion of symmetry (Luckmann 1990) is
important. Conversation is seen as one type of dialogue which is characterised by a tendency toward
communicative symmetry (ibid.: 58) where the perception by participants of social power
distinc-tions in exchanges (Levinson 1983) tends towards an acceptance of equal status. The
contention that when speakers interact, they do so on the basis of a calculation of the power and status
arrange-ments of the interaction also implies that learners will be motivated to select certain moves
rather than others, and that this selection is based partly on a conscious appraisal of symmetry or
otherwise.

The possibility of multifunctionality with respect to task participant strategies, such as confirmation
checks and clarification requests, has been pointed out by several researchers, notably Chun et al.
(1982) and Aston (1986, 1993). Work by Chaudron (1983) demonstrates the potential of detailed
analyses of different realisations of these devices. This is an important area for future research,
involving as it does the important notion of the effect of utterance, or perlocutionary force (Lyons
1977), which allows a distinction to be made between the intended and the actual effect of an
utterance. A task participant may try to persuade another to adopt a solution X, but may in fact cause
solution Y to be taken. An utterance intended by a speaker to act as a comprehension check, for
example, may be understood by the listener as an evaluative judgement or perhaps as phatic
information, motivated more by a need to establish and maintain social contact.

However, whilst accepting this potential for multifunctionality, this investigation assumes a
congruent relationship between task output form and task output function. It assumes that task
parti-cipant concerns with contingency and goal completion are factors likely to support a general
prag-matic interpretation of utterance meaning in terms of what Sperber & Wilson (1986) refer to as
the relevance of utterance to the ongoing task related discourse as well as the context of
performance.

Proficiency and strategy use

The research tasks here were designed on the assumption that the subjects in the quantitative
investigation had similar general proficiency levels, as defined by the results of the HKEA Use of
English examination. Some models of task-based learning and strategy use link motivation to
pro-ficiency (cf. Littlewood 1992). A number of researchers have investigated this relationship.
Yeung & Heyworths (1990) Hong Kong study suggested that learner competence and proficiency
were actually more influential variables than task design considerations. A study of learnerlearner

72
interaction by Varonis & Gass (1985) found that task participants who shared the same proficiency
levels and learner background tended to negotiate less during task performance. This is an interesting
result, but was not supported by the results of the case study investigation in this study. Tarone (1978)
found that less able learners preferred reduction strategies to achievement strategies. Ellis (1985)
supported Tarones findings and suggested that learners progress from predominantly reduction
strategy usage to achievement strategy usage as they develop along an interlanguage continuum.
Bialystok (1983) found that advanced learners exhibited specific patterns of inter-language strategy
use. Ng (1995) found that contrary to the assumption that less advanced learners would tend to use
more reduction strategies, lower-intermediate subjects in her investigation used the same amount as
the advanced learners. However, she noted that both members of an upper-intermediate dyad used
only interlanguage-based achievement strategies (as opposed to code-switching), irrespective of task
type. This finding supports claims that more advanced learners prefer this type of achievement
strategy.

Of significance to this investigation is van Liers (1996) contention that the most productive
task-based work in terms of language development comes from interaction between peers of equal
proficiency. This is essentially because task participants will be more likely to engage themselves
fully with an oral task if their perception of the rights and duties implicit in task performance tends
towards symmetry. However, equality of perceived status does not always imply symmetry, as the
previous discussion of the possibility of unco-operative task behaviour has shown. Apart from a
tendency of some participants to dominate the completion process, other problems may occur when
particular participants seek early closure to a problem (Rudduck 1991).

There may be a more general relationship between proficiency and strategy use on task: several
studies have emphasised the role of simplified output between lower proficiency learners (Pica 1987;
Holliday 1988). Such output is characterised by the predominant use of simpler strategies, such as
clarification requests, repetition or segmentation, in order to increase the comprehensibility of the
exchange. We might set up a general hypothesis that learners would always engage in more
simplified strategy use than native speakers in their responses to peer signals during interaction.
Recent research on this question (Pica et al. 1996) did support the hypothesis that peer groups would
use simpler strategies such as segmentation more than native speakers would when conversing to
learners, and that this usage was also task-specific.

The communicative stress associated with specific task performances can also be linked to
pro-ficiency, since we might expect that the more proficient the learner, the less communicative
stress he or she will experience in relation to task performance. Language anxiety (Gardner &
MacIntyre 1993) refers to the stress learners feel when using the target language in specific situations
and high levels of anxiety may inhibit learners when performing oral tasks where monitoring by a
teacher is expected (see Lai 1993). Language anxiety, or communicative stress, is a demotivating
factor in oral performance, experienced by speakers who feel that they are not able to provide
language responses to meet the perceived demands of the situation. As such, the concept can be used
both as an indication of task difficulty and a basis for task grading (Lynch & Anderson 1992). In
terms of a learners developing interlanguage, proficiency can be taken to mean a transitional
competence which also encompasses behavioural as well as linguistic skills. A learner may not be
proficient in terms of equivalence to a normative comparison with a native speakers competence,
but may be proficient in the successful employment of strategic competence relevant to the task in
hand. So-called proficiency shortfalls, manifested as errors can also be seen as evidence of
developing proficiency under a definition of the latter which views a learners degree of proficiency
as a system created by the learner (see Stern 1983: 355). Although this investigation acknowledges

73
the im-portance in theory of proficiency as a factor in motivation, the concept has been used in the
sense of a developing interlanguage system rather than normatively, as a measure of linguistic
competence in relation to a native speaker ideal. As such, proficiency was not included as a formal
research variable but as a subject control in the quantitative investigation.

Evidence for negotiated meaning

Evidence for negotiated meaning has previously been sought in relation to a quantification of
participant strategy use on task without attempting to gather data on participant motivation from
participants themselves. A number of researchers (Pica & Doughty 1985a, 1985b; Doughty & Pica
1986; Duff 1986; Rulon & McCreary 1986) have sought to confirm the hypothesis that small-group
work results in more negotiation of meaning than teacher-directed lessons. Rulon & McCreary (op.
cit.) examined the differences between group oral task activities and teacher-fronted activities and
confirmed that contextualised two-way oral tasks can produce significantly more strategy use
associated with negotiation of meaning than when the teacher leads the discussion. They also
compared strategy use between two types of oral task, attempting to differentiate between tasks that
appeared to focus on meaning that was not necessarily related to previous classwork, and tasks that
were contextualised into the syllabus so that task participants were more familiar with the task
content and vocabulary. It was hypothesised that the time spent actually negotiating task-based
meaning would be reduced with the contextualised tasks, and the results did support this hypothesis.

One study which attempted to investigate aspects of negotiated meaning and participant perception
of the process was Slimanis (1987) research which investigated conversational adjustments made on
task by highly motivated learners. Slimani concluded that the use of strategies associated with
negotiated meaning did not have a direct, observable beneficial effect on those who employed them
(ibid.: 314). Her measure of acquisition (uptake) was unusual and highly inferential, since the
subjects themselves made the assessment. Her methodology is interesting however, for the element
of introspective enquiry which it entailed, an element which was lacking in other studies during the
same period (but see Hawkins 1985).

The findings reported by Long & Porter (1985) support the contention that peer-group interaction
promotes this type of negotiation process and that the resulting interaction is a richer input source
than learnernative-speaker interaction in terms of the amount of language produced and the type of
strategy usage employed. A more recent study, by Pica et al. (1996), provides further support to these
findings and to claims that peer negotiation does address some of the theoretical acquisition needs of
learners, particularly in relation to the production of modified task output. Results from the study
reveal comparable production of such output when learners negotiated with native speakers or with
each other. These are significant research findings in support of the contention that peer-group oral
tasks are pedagogically valuable in terms of the opportunities they offer for negotiated meaning,
particularly in relation to the simplified feedback which learners are capable of supplying to each
other.

The results of these studies are difficult to interpret in terms of generalisations. Group work has not
conclusively been shown to result in more negotiated meaning than lockstep teaching, despite
inferential evidence that this is the case. However, it has been shown that certain types of peer-group
tasks are more likely to result in differences in the amount of interaction produced than in
teacher-controlled pedagogic tasks (see Bruton & Samuda 1980; Long & Porter 1985; Samuda &
Rounds 1988; Brown 1991).

74
Interaction and task type

Chaudron (1988) stated that the results of studies of interaction between learners had revealed no
clear trends in differences between different classroom organisational structures. However, an
important qualification to this position, which is of significance for the present investigation, was
that the language task itself did appear to be a critical factor. In the same year as Chaudrons study,
Tarone (1988) outlined a range of task design, participant and task setting features which might be
hypothesised as having an influence on the amount of interaction produced:

(i) the subjects interpretation of task instructions


(ii) the order of tasks or items within the task
(iii) task difficulty and complexity
(iv) feedback given to subjects
(v) topic
(vi) time
(vii) interlocutor familiarity.

Long (1981) found evidence that more interactional modifications occurred in two-way than in
one-way tasks. Two-way tasks are tasks in which all the participants are given information they
have to share in order to complete the task. One-way tasks are those where only one participant is
responsible for providing input information to the other participants in order to complete the task.
This finding stimulated a number of similar studies, using the same general classification of task type
as either one-way or two-way. An example of this type of study in the Hong Kong context was that
by Ng (1995), who investigated the effects of one-way and two-way tasks on learner strategy use.
Her results confirmed Longs original finding that there was a tendency to employ more strategies in
two-way tasks rather than in one-way tasks, where learners do not have so much opportunity to
negotiate meaning.

However, in a study involving intermediate level learners at university, Gass & Varonis (1985) did
not find any significant differences in the performance outcomes of one-way and two-way tasks.
They did, however, find evidence to support their conclusion that peer-group oral tasks offer
non-native speakers a positive opportunity to receive comprehensible input and produce
comprehensible output through negotiation. Tong-Fredericks (1984) compared learner output on
three types of communication task, a problem-solving task, a role-play task, and what was termed an
interaction, task which involved opinion exchange. The problem-solving task produced the highest
number of turns per minute, but resulted in very disorderly turn-taking. Self-corrections were more
frequent on the role-play and interaction task. No differences in fluency, as measured by speech rate,
were found. Tong-Fredericks showed that different tasks elicited different responses, which could
promote acquisition in different ways. This also meant that different task types would consistently
produce these different responses, but no evidence was offered to support this implication.

Duff (1986) used what she termed convergent and divergent problem-solving tasks, and her
study compared a variety of interaction variables within learner dyads on two types of these tasks
problem-solving and debate. Convergent tasks required learners to reach a consensus, whereas
divergent tasks required them to exchange different information without the need for consensus. The
convergent tasks were found to produce more negotiation of input, but the divergent tasks resulted in
greater complexity of output. The notion of output complexity is important here. No formal
categories for its assessment were set up in the present study, but, as Skehan (1989) pointed out,
output complexity in terms of syntactic and lexical range, can be an important research and

75
pedagogical variable.

Shortreed (1993) did find significant differences between task type and the amount of negotiated
meaning produced. The study researched task-related strategy use in order to test the hypothesis that
task participants would simplify their speech and use a higher frequency of interactional
modifications in accordance with their level of proficiency, and the task complexity. The constructs
used to distinguish the tasks with regard to their relative complexity were firstly, the amount of
shared reference across the tasks, and secondly, the differing levels of production required to
complete each task successfully. Shortreed (ibid.: 111) reported that:

The most noteworthy result of this study was the difference found in the amount of negotiation work
across both tasks. In particular, all four groups used significantly more interlanguage and repair
strat-egies on task 2 (the more complex task) than on task 1. This was reflected in sentence level topic
encoding strategies where the frequency of questions, declaratives and imperatives differed
signi-ficantly across tasks ...

Samuda & Rounds (1993) investigated an information-gap task for the turn-taking behaviour it
generated, and took the number of turns generated per group as one of the indicators of negotiated
interaction. In a much larger study, researchers compiling the Edinburgh map task corpus (Anderson
et al. 1993) used a problem-solving task to investigate interaction and task-based strategy use. A
longitudinal investigation was possible with this project, which is important, since the design of the
other studies mentioned so far did not permit this theoretically desirable approach. To date, the
researchers report that it has been possible to identify several interactive skills which characterise the
dialogues of the more successful communicators. The map task project also permits exploration of
how task-based collaboration develops, and secondly, how it is achieved. When the same speaker is
observed performing the same tasks with familiar and unfamiliar partners, some account can be given
of strategies which are used generally by a given speaker and those which appear as unfamiliar as
participants attempt to build the basis of their collaboration.

Types of objection to the concept of negotiated meaning

There appear to be three main types of objection to the concept of negotiated meaning discussed
above. The first questions whether the concept has any validity at all that the meaning of meaning
in this context is far from clear. The second, related, type of objection acknowledges that negotiated
meaning is a possibility in learner exchanges but suggests that it doesnt always occur and that when
it does, it is not always evidence of co-operation. The third type of objection questions its value to the
learner if it does occur.

Does negotiation of meaning exist?

Objections to the existence of the concept of negotiated meaning largely depend on differing
conceptions of learner motivation when apparently involved in strategy use on task. The possibility
of unco-operative negotiation of meaning has already been mentioned, where a lack of mutual
co-operative scaffolding results in non-completion of the task. There is also another objection to the
whole notion of negotiated meaning, which centres on the contention that speech acts are, or may be,
multifunctional. Learners may pretend to comprehend for phatic or socio/cultural reasons. The
cultural context of this study was that of Hong Kong society during the run-up to the resumption of
sovereignty by China, and there are many controversial allusions in pedagogic literature to

76
fundamental differences in learning styles and culturally-based conceptions of language and
language usage between English and Chinese speakers. For example, Scollon (1993) hypothesised an
underlying Confucian assumption of Chinese learners of the importance of self-knowledge rather
than public display. If this propensity for purely receptive learning roles is a cultural reality, it might
also be expected to result in task participants avoiding signalling their perception of communication
problems rather than attempting to repair these perceived problems co-operatively. We might also
conjecture, however, that this response is more universal, relating more to personality type than
cultural background. Difficult interactive classroom tasks designed to promote negotiation of
meaning may fail to do so because certain types of learner might feel that the performance of these
tasks is stressful (see also Brown et al. 1984: 35).

Aston (1986: 128) was one of the first seriously to question the assumption that evidence of
negotiation of meaning, or what he termed trouble-shooting procedures was, in fact, evidence of
co-operative attempts to achieve task completion. He pointed out that task participant strategy use
could also be seen as essentially phatic concerns with the circumstances of unshared participant
backgrounds, and therefore might be more correctly viewed as attempts by task participants to
interact with each other to get to know each other rather than to interact for task completion reasons.
The meaning being negotiated might also be seen more in terms of acceptable form (rather than
comprehensible form). Task participants might be attempting to increase its accessibility, its
memorability and its relevance (salience) in relation to the learning situation (Pennington 1986).

Varonis & Gass (1985) also admitted to research difficulties in distinguishing exchanges involving
the negotiation of meaning from exchanges involving strategies such as keeping the conversation
going, which are more correctly viewed as phatic phenomena. Other researchers have tried to
indicate how the central question of task participant motivation might be better established. Yule &
Tarone (1991) were among the first to argue that the features associated with negotiation of mean-ing
might in fact be more profitably considered through conversation analysis:

Negotiated input must be not just the result of moves by one speaker, but the result of co-operative moves
by both speakers. The communication strategy framework (of conversation analysis) is one which
focuses precisely upon such moves by both participants. (ibid.: 167)

This type of analysis inevitably has to attempt to examine the speakers personal assessment of the
listener.

Does negotiation of meaning always occur?

The second type of objection questions whether learners always negotiate meaning in any social
interaction, and focuses on particular variables which could be perceived to be influential barriers to
such negotiation. Negotiation may not occur if topics are very familiar, to the point of being boring
or because they are so familiar that participants are confronted with no significant communication
problems to repair. Conversely, negotiation might not take place if the task is perceived to be too
difficult, resulting in substantial use of avoidance strategies and rapid task abandonment. Pica (1988)
suggested that in classroom situations, learners do not negotiate for one of two reasons: the first is
when they have no interest in the topic of the task; the second, when they do not want to display their
lack of strategic competence to attempt the perceived problems of the task. This reluctance to display
a lack of knowledge or ability may have a cultural dimension since Hong Kong students are evidently
very concerned with issues of face (see Littlewood et al. 1996). However, the paucity of evidence
from our transcripts concerning unco-operative negotiation may indicate that students in Hong Kong

77
see non-completion of tasks as a more important threat to face than their assessments in relation to
interactional symmetry on task.

It may also be the case that negotiation of meaning is not always co-operative and should not be
assumed to be (Aston 1993). Bly (1993) suggested that any model of conversation ought to account
for unco-operative as well as co-operative conversation. Co-operative speakers collaborate to reach
mutual understanding in a process variously termed scaffolding or grounding (Hatch 1978a).
However, sometimes one of the speakers is unco-operative and in this circumstance, Bly claimed it
was helpful to conceptualise the grounding process not as linguistic hypothesis testing on the part of
the listener, as some researchers have suggested (e.g. Tarone 1983), but as goal-related negotiation
of meaning in which both partners have different task goals. When an unco-operative partner plays
an active role in this process, it is likely that a co-operative partner will not successfully resolve
repair problems which arise. In short, the extent to which two participants in a task dyad construct
mutual agreement is ultimately constrained by their differing world views and life experience. In this
sense, co-operation is better viewed as a general trend during task performance toward com-promises
in many different aspects of that performance.

Is negotiation of meaning useful?

The third type of objection concerns the assumed value of the phenomenon to language learning and
acquisition. There are two main related objections. Firstly, negotiation of meaning may inhibit
learning, facilitating communicative performance without facilitating acquisition of new linguistic
features (Prabhu 1987). Further, because of the interactional support the learner receives, he or she
may not experience any need to learn new features (Sato 1986). An important corollary of this
argument, however, is that although linguistic competence is not developed, strategic competence
may be enhanced. It may still be the case that even if negotiation of meaning does not facilitate
acquisition, it may help learners to make better use of existing resources.

Secondly, negotiation of meaning may only be apparent and may not result in comprehension. In a
study which incorporated some retrospective investigation, Hawkins (1985) tried to determine
whether apparently appropriate responses made by learners actually signalled comprehension. The
subjects performed four types of communication tasks and retrospective data was collected from
learners in order to determine whether what appeared to be appropriate responses in the context of
task performance did, in fact, represent comprehension. Hawkins concluded that for 50% of the
responses for which retrospective data from the learners was available, comprehension had not taken
place.

The study is also interesting for its use of introspective data, which was collected from the subjects
by a procedure used by McCleary (1980) and Erickson (1981). The tapes of the original task
performances were played back to the subjects, who were asked to stop the recorder at any time and
comment. The researcher also stopped the recorder and asked questions at specific points. Hawkins
concluded that the concept of comprehension in peer-group discourse is an elusive phenomenon
which requires more exact and complex description, and that the retrospective conversational
inter-action between the subjects and the researcher is important as a source of explanatory data.

Negotiated meaning and motivation

Central to arguments concerning adequate description of task performance is the problem of learner

78
motivation. Motivation for language learning in general has been seen as instrumental, social or
intrinsic, depending on whether the emphasis is on learning for practical reasons (Ames 1986),
culturally integrative purposes (Schumann 1988) or self-goals more connected to personal
fulfil-ment and self-esteem (Deci & Ryan 1992). In an investigation of Hong Kong Form 4 and Form
6 students general motivation for language learning, Richards (1994) concluded that there was no
direct causal link between motivation and linguistic competence, but that motivation was linked with
certain actions which might eventually result in an improvement in proficiency. Given instruc-tional
input and task objectives, a learner clearly needs general motivation for a task and a sufficient level
of proficiency in order to interact successfully in the task performance. An important aspect of the
qualitative case study research was also to try to ascertain which task types are intrinsically more
motivating for Hong Kong students.

Motivation is a complex concept, where receptivity (Allwright & Bailey 1991: 15770) to the social
context of the language learning event would seem to be just as important as the prior motivation
which learners might bring to this event. Apart from studies into general motivation for language
learning, there has been very little research into specific motivation for the selection and usage of
particular procedural or enabling strategies during task performance. This is not surprising, when the
question of the psychological plausibility of the concept of communication strategies is still very
much in question (Kellerman & Bialystok 1997). This is not to imply that answers have not been
suggested, only that the models presented also have difficulty accounting for non-strategic language
use (ibid.: 44). This is an indication of a greater definitional problem, where, if all language is
regarded as strategic, a model of strategy motivation must account for strategy selection in terms of
a specific set of cognitive processing skills, so far unidentified.

If our pre-hypothesis is that strategy motivation is under the conscious cognitive control of the
learner, then we might want to support Bialystoks (1990) hypothesis that it is possible to ascertain
the relative amounts of cognitive analysis and control demanded by various kinds of communicative
strategy usage. This pre-hypothesis appears to have some general support amongst communicative
strategy researchers, although Frch & Kasper (1983) referred to communicative strategies as
potentially conscious, allowing for the development of automatic or routinised strategy usage. A
basic assumption of the present study and many others incorporating the concept of communicative
strategies is that the motivation for their selection and use by learners is accessible to research
enquiry. This enquiry has usually taken the form of observing and measuring rather than attempting
to find out about possible causation. One notable exception was the study by Hawkins (1985), which
incorporated retrospective investigation in relation to task performance outcomes, and this approach
was influential in relation to my own research design. The subjects in the case studies for the
qualitative investigation (not reported here) were interviewed to attempt to determine their general
motivation for language learning and their specific motivation for particular instances of apparent
strategy usage.

79
Strategic competence

While sociolinguistic competence is seen as concerned with culturally contextualised rules of


discourse, strategic competence has been viewed largely as the ability to use linguistic resources
effectively. This effectiveness has been measured in terms of compensatory or repair activity in a
conversational exchange, largely because this constitutes a measurable and observable competence
which manifests itself when task participants experience a communication problem. This can be seen
from Tarone & Yules definition (1989: 104), where speakers are hypothesised as using
communication strategies when the following three (necessary) conditions occur:

1. A speaker desires to communicate a meaning X to a listener.


2. The speaker believes the language form he or she wants to use to communicate meaning X cannot
be produced.
3. The speaker chooses to: avoid (not attempt to communicate meaning X), or attempt alternate means
to communicate meaning X such as mime, word coinage, circumlocution, etc. (The speaker stops
trying alternatives when it seems to him or her that there is shared meaning).

Strategic competence has been further defined in relation to a specific category of communication
strategies, namely resource, expansion and achievement strategies (Tarone 1984; Paribakht 1985;
Bongaerts & Poulisse 1989). These strategies do not call for the communication partners assistance.
Learners try to overcome perceived problems by using their own resources. However, if we confine
the concept of strategic competence to those strategies which are brought to bear on essentially
observable referential misunderstandings, then, as Kasper (1997: 346) has pointed out, we ignore the
pragmatics of interpersonal rhetoric which extends the concept of strategic competence to include
not just task-based referential goals, but also participant relational goals an important aspect of the
necessary social contextualisation of task performances.

How might the employment of strategic competence help develop this particular competence fur-ther?
Nobuyoshi & Ellis (1993) suggested that pushing learners to modify their output results in an
increase in ability to deploy existing grammatical knowledge more accurately and Pica et al. (1989:
64) suggested that in modifying task-related output, learners are testing hypotheses and
experi-menting with new structures and forms in creative ways. This hypothetical link between
output and hypothesis testing by the learner, was also used by Tarone (1993: 17) to explain how a
learners interlanguage system developed, and in doing so, she appeared to come very close to
equating this type of output with the acquisition of further strategic competence:

... it is in those interactional contexts where the learner needs to produce output that the current
interlanguage system cannot handle that the learner pushes the limits of the interlanguage system to
make it handle that output, thus keeping the system permeable and open to change.

Failure to repair a perceived problem successfully may result in a breakdown of communication and
learners may simply elect to avoid or ignore the problem, rather than reach a point where their
strategic competence is revealed as inadequate. The source of the problem may be procedural,
lin-guistic, or located in the interactional nature of the task setting and more concerned with
inter-personal factors such as perceived threats to face for example. Indeed, the mechanism of repair
can be viewed as specifically organised in such a way as to avoid such face threats and thereby form
part of the learners management of task completion strategies. It follows that the employment of
types of task-related strategies can then be taken not only as a measure of individual strategic
competence but also as an indication of the perceived level of difficulty of the task.

80
Strategic competence and the context of performance

The strategic competence of task participants is employed in a task-related context or setting and in
direct or indirect response to that context. Context is defined here as the linguistic, situational, social
and cultural environment of task-related actions. Communication breakdown and evidence of
strategic competence can be linked to the idea of contextualised action, as well as contextualised
referential problems on task. Conflicts or potential conflicts between personal agendas the actional
and relational goals of task participants are resolved not just with clarification strategies for
example, but with strategies intended to repair what we might see as impending interpersonal
communication breakdown the avoidance of face threats through the use of politeness strategies,
for example, in order then to achieve more task-related goals. Just as the use of a word in a linguistic
context is determined by collocational or selectional restrictions, so the choice of an action during
task performance can be hypothesised as being determined by a participants inter-pretation of the
social context of the task performance. This understanding is unique to that parti-cipant i.e. it is not
identical for speaker and listener. This mismatch is the basis of all information-gap activity, and is
the reason why all conversational interactions can be considered to be negotiated. Listeners build a
particular interpretation which depends on their own interests from among the many possible
meanings of task-related utterances. These interests, in turn, relate to the parti-cipants knowledge,
a personal reference point or base, composed not only of declarative and pro-cedural knowledge in
relation to the oral task and its enabling language, but also consisting of the participants knowledge
of the world and the task performance situation or setting. Participant interaction has to be viewed as
partially related to this private knowledge, however, as Young (1988: 122) has pointed out, the term
interaction is generally used by second language acquisition theorists in a rather restricted sense to
mean oral exchanges in the target language between a learner and one or more interlocutors. The
activity is usually task-focused and there is a pre-hypothesis that all subjects share the same
definition of the task (Todd 1981: 220) and that this definition remains constant during the task
performance. However, as Mohan & Marshall Smith (1992: 81) acknowl-edge, a more dynamic
model might be required, one that takes into account

[t]he learners understanding of the context or environment of his or her actions ... as the learners task
expands ... the context is socially constructed by the co-operative work ... and the context both
illuminates and is illuminated by communication within the task.

Since the interpretation of context will be different for both speaker and listener, their interpretation
of the changing context of task performance, through the changing pattern of task-related strategy
employment, will be dynamic rather than static i.e. that interpretation itself is a dynamic,
negoti-ated process where speakers and listeners are willing to accept mutually convenient
interpretations of task-related problems and solutions, rather than working together towards one
fixed and im-mutable interpretation. It is in this sense that participant motivation for the choice of
particular interpersonal and task-related strategies is again the key area for research focus.

Strategy marking in discourse

When learners perceive a problem on task therefore, this may be indicated by the appearance of
strategy forms in task transcripts and the appearance of such phenomena then serves to signpost or
mark the problem source, which may be procedural, linguistic, or located in the interactional nature
of the task setting. The appearance of strategy marking only relates to the conscious use of
communicative strategies, and, in the definitional terms of this study, only to those communicative
strategies under investigation in relation to a particular concept of negotiated meaning. These

81
com-municative strategies are specifically those associated with task interaction which have been
defined as overtly strategic (Kasper & Kellerman 1997a: 8).

The question of what linguistic forms we can reliably associate with strategy use in discourse is
complex, and ultimately becomes one of the operational definition of communication strategies and
how they are assumed to be marked in discourse. Here, a degree of correlation between linguistic
form and illocutionary force has to be assumed.

Trying to form a general definition of strategy markers in relation to units of language use is
problematic, since use of a particular speech act will be dependent on the context of performance.
There is certainly no simple one-to-one relationship between syntactic units and information units
(see also Brown & Yule 1983b: 167) although the operational definitions of strategy usage in this
study have initially appeared to have assumed otherwise in relation to manifest evidence of
participant intention from the research transcripts. In the case of confirmation strategies, for example,
this evidence has been in the form of recognisable questions such as So what you are saying is ...
etc. The syntactic evidence of the use of a confirmation strategy here has, at the very least, to be
verified by post-task retrospective interviewing.

Given the amount of data involved in a large search for task-related outcome patterns, operational
definitions have to be based on essentially broad criteria for three main reasons. The first relates to
the general problem of quantification in an area of discourse which is characteristically imprecise.
The second constraint on definitional precision is in relation to the marking, or flagging, of strategy
use. If we only look for evidence of prototypical strategy forms, then we risk ignoring other potential
types that may be just as influential in the process of meaning negotiation. For example, transcripts
might also increasingly evidence other types of discourse features such as hesitation at the approach
to a problem source (Kasper & Kellerman 1997a) and this could be viewed as a common type of
strategic response to a perceived problem with spoken discourse (Channell 1994). The transcript
would obviously have to be sufficiently and specifically detailed to reveal this, and this is one
significant problem with tailoring transcription coding to exclude data on the grounds of
preconceived purpose or practical concerns. An apparent lack of precision with respect to
communicative strategy marking can lastly be linked to the possibility of multifunctionality with
respect to any speech act. However, it is the context dependency of these speech acts which underlies
the important presupposition that, in a research context such as this, where task performances are
subject to the same controls, strategy markers are reasonable evidence of participant strategy
employment i.e. we have to initially assume that there is a significant correlation between linguistic
form and illocutionary force in order to establish a coding protocol.

82
83
Chapter Six: Research and pedagogic conclusions

... it is not only possible but also valuable to combine procedures of naturalistic enquiry and
experi-mental research, if experimental results are to be interpretable. (Allwright & Bailey 1991: 45)

An important responsibility facing the qualitative research community is to ensure that qualitative
methods and the studies that employ them are relevant and accessible to the practitioner as a consumer
or producer of qualitative research. (Lazaraton 1995: 467)

Oral task types and the negotiation of meaning

A consideration of the interactional definition of negotiated meaning has led us to redefine the
concept as a dynamic process of task input clarification in relation to the demands of the task and the
social context of performance. It incorporates both the idea of making task input more
comprehensible and more accessible, a process achieved through constant hypothesis formation and
evaluation on the part of task participants. This interpretation frees up the debate from a narrow
concern with the negotiation of one type of task-related meaning, and allows the concept of
interlocutor motivation to be brought more fully into the discussion.

It is also easy to lose sight of experiential dimensions such as memorability and salience (Pennington
1986) when considering questions of task inputs and outcomes, and when raising what have been
seen as mechanistic metaphors in association with many analyses of task-based interaction. Salience
is defined in terms of the attributes of the task that are noticeable and therefore memorable.
Memorability and salience in relation to task design and perceptions of task relevance could be
strong motivational features, encouraging learners to use task input in a more focused way to try to
solve problems, or to work out solutions which result in new, reprocessed output.

It is in this sense that the concepts of input and output can be too narrowly defined, the first in task
design terms and the second in terms of goal achievement. Output is better conceptualised as a
process of learner revisits to aspects of the task and its performance. My use of the term revisits
encompasses both the notion of recasting (Carroll & Swain 1993) and segmentation (Mendonca &
Johnson 1994). This cyclical revision process, where previous troublesome items are returned to in
the ongoing discourse may be a significant feature of the process of negotiated meaning, where the
concept of the participants linear progression through the task might be better conceptualised as a
cyclical process of continued comparison and restructuring, a continual revisiting of the central
concerns of the task participants with respect to the task and the performance setting.

There appear to be two main considerations here: the first is the motivation of the individual learner
in relation to the task performance; the second, the type of task. In a jigsaw task, for example, since
no one participant holds all of the necessary completion information, participants must interact to
find a solution. During the interaction we could expect overt evidence of negotiated meaning. In an
information-gap task, since only one participant has the necessary information, the other participant
has to interact to find a solution. However, in a problem-solving task, it is possible for one of the
participants to dominate the exchange and only request information from the other, rather than
negotiate it. When this type of task is open-ended, or divergent, there is even less incentive for
weaker learners to participate, as can be seen from the transcripts of some of the E1 and E2
discussion tasks, where negotiation opportunities are reduced to a minimum by interlocutors
swapping ideas at a relatively easy level of topic and language management.
From the five task types in the research typology, information-gap and jigsaw tasks produced the

84
highest frequency of occurrence of the core discourse features associated with negotiated meaning.
These tasks did appear to encourage participants to attempt to construct mutual understandings of
task content and language-related meaning. These findings are similar to those from a study by
Doughty & Pica (1986), who found that significantly more negotiation of meaning occurred in
information-gap than in decision-making activities. It may be the case, therefore, that when learners
express a preference for particular task types, such as discussion tasks, they are selecting a task type
which might not be as useful for them in terms of the hypothesised benefits of negotiated meaning.
It would be dangerous to assume, however, that the usefulness of tasks in terms of interlanguage
development is restricted only to the opportunities they offer for repair strategy employment. The
fact that research subjects generally expressed a preference for discussion-type tasks is interesting,
and may indicate that their increased motivation for such task types makes these useful to them in
other ways. This preference may be linked to a perception of a decreased level of
information-processing involved to complete the task. Information-gap tasks appear to require
relatively high levels of input processing on the part of task participants, whereas discussion tasks
appear to carry higher communicative responsibility. This responsibility might be stressful where the
content is abstract, but preferable when the content is familiar, since participant topic control is
higher. In this case, it is possible that the benefits of this reduced atmosphere of communicative stress
may outweigh the possible disadvantages of the task type in terms of its relative inability to promote
the type of strategy use associated with negotiated meaning.

An analytical investigation of these questions in terms of task design and negotiated meaning is
limited without the modifying perspective of learner motivation. For example, some subjects in this
investigation stressed that task type was actually less important to them than the frequency with
which oral practice was done in the classroom. They wanted more specific opportunities to use their
English in class. Their perception that speaking activities in general were useful was an important
element in their personal motivation. Discussion tasks in seminar formats have therefore proved very
popular in an advanced language course I compiled and taught at UST. Such seminars can be
structured to link different task types. Students may be performing discussion tasks (convergent or
divergent) following a phase of information exchange. If the topic is well prepared beforehand, is
perceived to be of relevance to their studies or future careers, and task roles are clearly defined,
students will feel less anxious in seminar task performance than in other types of interactive task.
There may also be a place for tasks which are designed to create a degree of anxiety, although these
are certainly not prominent in published course or evaluation materials. Information-gap tasks and
discussion tasks where the input is changed or extended over the length of the performance, as in
some business simulation tasks, might create opportunities for learner negotiation of both task input
and process in a way that might prove very motivating to some types of learners. This type of more
stressful task performance setting might also be expected to improve salience and memorability for
these learners.

The case studies also reveal the importance of individual learner differences in the question of
negotiated meaning. This investigation has sought patterns of task-related talk and has found
evidence to support general claims in relation to the connection between task type and a relatively
narrow definition of task outcome. The case study interviews also revealed a greater variation in
motivation and individual cognitive style than the operational definitions might suggest. For example,
there appeared to be very personal approaches to what might be termed general planning strategies
in relation to the task the individual perceptions of how best to complete the activity in relation to
goal and setting requirements. This variation might also be more evident in terms of strategy
motivation on task if a closer retrospective focus on particular strategies had been possible. In terms
of individual personality attributes which might also influence strategic com-petence on task for

85
example, persistence, imagination and tolerance for uncertainty case study subjects also revealed
variations which might also have been important moderator variables.

Functions of negotiated meaning

The research subjects did appear to be working co-operatively to complete the tasks and to make task
input more comprehensible through their signals of perceived problems with input concerns. The
case study retrospective introspection revealed that in most observed instances of strategy use,
strategies such as comprehension requests and clarification requests etc., were interpreted by learners
in the same way as by me, i.e. these features were attempts to negotiate co-operatively between
interlocutors to improve mutual understanding of the task input and output. These investigations not
only support the existence of the concept of negotiated meaning as defined, but also do not offer
support to the contention that the phatic purpose of such interactional modifications is predominantly
something other than mutual co-operative attempts to complete the task, despite the possibility for
other outcomes. I say predominantly, because there was certainly evidence from some case study
interviews that apparent co-operation was also being used as a strategy in itself to achieve other ends.

We could extend the concept of negotiated meaning to include different sequences of negotiated
interaction, since negotiation can be about acceptable management of task completion, acceptable
output form and acceptable task input meaning. The concept of the negotiation of acceptable output
form is accessible through the category of evaluative content repair (ECR) in this investigation.
However, this construct failed to achieve any significance in the data analysis, largely because task
participants, as expected, did not engage in much overt critical peer evaluation of language form in
the task performances. If task instructions specifically call for a focus on accuracy, a different result
may occur.

Attitudes to oral discussion and criticism might be expected to vary between cultures and affect
participant motivation to criticise the performance of other task participants. For example, Mak &
Reid (1992) claimed that criticism in an educational context is contrary to the value system of
Chinese culture and, from a wide-ranging international survey of oral reticence, Asker (1997)
claimed to have found evidence that students from both Hong Kong and Micronesia were
significantly different from those from other countries in their attitude to peer-group discussions.
Communicating in groups appeared to be more acceptable to Hong Kong students, although dyadic
communication was less popular.

While causation in relation to task-based repair procedures is impossible to establish with certainty,
the results from both our quantitative and qualitative investigations support (but do not locate) a
central contention of a relationship between oral task type and the strategies associated with
negotiated meaning, with the important caveats that the relationship is likely to be highly
domain-specific, and that the multivariate analysis employed in the quantitative study could only
reveal significant differences between task types and learner output not the nature of the suggested
relationship. The central descriptive problems in relation to the functions which these strategies
actually perform (manifest functions) have to be distinguished from a consideration of the functions
which participants actually intended them to perform (latent functions). The data in both these areas
is highly inferential but the investigation supports the notion of a congruent relationship between
participant intention with respect to strategy use and transcript appearance.
Strategies and negotiated meaning

The strong interactional claim in terms of strategy use and negotiated meaning is that such strategies

86
are essentially co-operative, or collaborative in nature. In the case of clarification checks (CLA), for
example, one participant explicitly invites the other to provide more information on previous task
content. This is done with the expectation that the other participant can and will help to find a
solution to the problem. Appeals for such help can also be indirect where, for example, a task
participant switches code, which is then taken as a signal by the listener that a problem has been
encountered. Reference has often been made in the literature reviewed as a basis for the investigation,
to this co-operative principle (Grice 1975) where task participants are assumed to employ direct and
indirect strategies to conclude the task co-operatively. This co-operation has to be viewed initially as
an ideal which is both productive and receptive, and contingent on the mutual acceptance of common
beliefs and concerns in relation to the task and its performance.

However, the possibility of very different conceptions of task goals on the part of task participants
clearly exists. Where participants view each other as having equal rights and status in task
performance, co-operation is more likely to be positive. This point is indirectly supported from
research into learner perceptions of the learning process. Baileys (1983) research through learner
diaries revealed that learner anxiety seemed to be strongly linked to a competitive interpretation of
the language learning class. Where participants perceive face threats (Goffman 1976) from
participation in unequal exchanges, such as with native speakers, forms of unco-operative strategy
use might also be expected. This is one reason why peer-group oral tasks might be preferred by
learners, and why these might be as useful or more useful for interlanguage development than
exchanges with native speakers, since there is a relative absence of such face threats. There is strong
evidence from the case studies (not reported here) that this assumption is supported and that learners
themselves view these processes as beneficial partly for this reason. Participants were able to employ
co-operative content repair techniques to overcome linguistic shortcomings and offer accurate
assistance, particularly with lexical items. There were very few manifest occurrences of
unco-operative repair procedures, and very few examples of the employment of avoidance strategies,
results which were possibly influenced by learners knowing that they were being recorded (there
were also very few examples of expected avoidance strategies such as code-switching).

It may be hypothesised that if structural modifications do occur in participant interlanguage as a


result of task strategy employment, it happens at a level which is not evident to a researcher and not
manifested in the performance itself. This may be when learners use strategies such as
subvocalisation, or when they silently attend to the perceived deviant structure, and subsequently
remodel their hypothesis about that structure without indicating this revision to their task partners.
However, this investigation has generally assumed that correction moves on task are predominantly
overt, and this assumption is supported by other studies (Gass & Varonis 1985, 1989; Rodgers 1988;
Samuda & Rounds 1988; Jacobs 1989). The type and variety of strategies which the research task
participants employed for task completion appeared to be somewhat restricted within the
investigated range. This observation supports similar findings from the study by Pica et al (1996),
where learners tended to use simplified strategies such as repetition and segmentation more than
other types. Learners would simply repeat items that were troublesome and often would receive
further simplified morphosyntactic information about the item from their peer interlocutor. This
models, at a lower level of complexity, the process that characterises native speaker modification of
non-native speaker input.

87
The present study corroborates the high frequency of usage of what are essentially repetition
strategies associated with certain types of task (e.g. pure information exchange tasks such as A and
B types). In terms of the group of variables associated with negotiation of meaning, repetition (REP)
was the most important variable contributing to significant differences between tasks. The other
variables were co-operative procedural repair (CPR), evaluative content repair (ECR), clarification
checks (CLA) and comprehension checks (COM). These core strategies appeared more frequently in
information-gap and jigsaw-type tasks, leading to the conclusion that these types of task seem to
generate more negotiation of meaning, as defined, than other task types. The more complex versions
of each task type generated higher frequencies of these figures in each case, except D-type tasks. This
result is also supported from the study by Pica et al. (1996) where it was found that learners engaged
in more negotiated meaning of task input on information-gap tasks, and these tasks elicited a higher
amount of segmentation and clarification requests in particular. These re-searchers also suggested
that the clarification requests were more conducive to learner modi-fication of interlanguage than the
confirmation checks, since listeners tended to react to this latter strategy with a simple yes.
Clarification requests such as What did you say? tended to make listeners respond with a
restructuring of utterances. This is an interesting area for further research since, because of the
specific research objectives, this study has initially regarded all task-based strategy use associated
with negotiated meaning as qualitatively the same in relation both to interlanguage adjustments or
modifications and the assumed benefit of this process for language acquisition. Such recasting by
speakers provides implicit feedback and has been the subject of much recent research into the effects
of implicit and explicit feedback to learners (see Carroll & Swain 1993). Motivation for strategy
employment generally appeared to be predominantly instru-mental in relation to a co-operative
attempt to achieve task completion goals. This is not to say that the case study subjects did not choose
to use certain strategies for more interpersonal reasons, only that evidence for this is not apparent
either from the transcripts or from post-task retrospection.

Oral task types and topics

A large-scale survey of the experiences, proficiency and perceptions of Hong Kong undergraduates
(Littlewood et al. 1996) has generally endorsed the classroom methodology advocated here. Eighty
percent of the respondents in that survey expressed a preference for language learning materials
which contain a lot of communication exercises, e.g. discussions (ibid.: 74). This general stated
preference for discussion tasks was endorsed during my post-task case study interviews, although it
is possible that in some cases, students might have subsumed all active oral pedagogy under the
umbrella designation of discussion.

This preference seems to be for a number of reasons. Hong Kong students are familiar with the
discussion task type because of their preparation for the Hong Kong Examinations Authority oral
examinations prior to university entrance (Andrews & Fullilove 1994). Subjects in the case studies
also expressed a preference for the activity since it gave them a chance to express their own opinions
something which didactic teaching styles both at secondary and tertiary level prohibited them from
doing. Some subjects also stated that they preferred the discussion task type since, unlike in some
other task types particularly information-exchange and jigsaw types there was usually no
requirement to establish the nature of the information gap before they could feel they were actually
able to practise their oral English.

Task-type preference under the generic category of discussion is greatly influenced by affective
factors such as prior experience of this type of pedagogy, perceived relevance to learner interests, and
awareness of the possible benefits of interactive task-based methodology. Results from the case
study interviews and from research visits to Hong Kong secondary schools (Courtney 1997) also

88
support Askers (1997) finding that Hong Kong students have a preference for discussion-type tasks
as a setting for oral performance because this task type is perceived to create the least personal
anxiety. Asker used a measure of the subjects willingness to communicate (cf. McCroskey &
Richmond 1990) and found that, as in the present investigation, role-play tasks were generally not a
preferred means of oral practice, largely because of the increase in personal anxiety related to the
increased communicative responsibility involved, the high degree of exposure and the abstract and
theatrical nature of the activity. Simulation tasks were popular where these did not involve any
role-play, and where the sequencing of tasks within the syllabus allowed ample input preparation for
the activity. Some subjects stressed that task type was less important to them than the frequency with
which oral practice was done in the classroom, since they wanted more specific opportunities to use
their English in class.

Teaching formats such as seminars were popular, since students are essentially performing a range
of task types, principally type E discussion tasks, but also information-gap tasks, in terms of content
presentation. The topic must be well prepared beforehand, of relevance to their studies or futures, and
roles should be clearly defined. In the normal interactive discussion group, students can feel some
anxiety because of the lack of role definition. In a seminar discussion group, presentation and
discussion responsibilities are normally more clearly defined.

In the research tasks, task content topic was fixed for research purposes and learner topic
pre-ferences were established from post-task retrospective interviews. The more control learners feel
they have over topic selection and management, the more successful the resulting performance is
likely to be in terms of learner motivation to complete the task. Case-study subjects stated a
preference for video input as pre-task preparation for discussion-type tasks. This is possibly because
of the higher degree of autonomy given to learners in relation to their selection of specific content
from this form of task input (see Christopher & Ho 1996) and the motivational value of what was
seen as real life input.

From my experience of running discussion groups for students involved in prescribed language
enhancement courses at UST, as well as for debating team students and those involved in voluntary
language improvement, it seems to be the case that where complete topic choice is offered, areas
chosen by groups are initially often so broad that content has to be broken down into more
manageable sub-topics. Learners probably prefer topics that have been sharpened in this way, so that
the question of topic choice has always to be seen in relation to the question of topic viability. It is
here that the role of the language teacher as adviser and facilitator is crucial.

Pedagogic recommendations

Content concerns often take precedence over process concerns in the design of language syllabuses,
largely because content appears to be more definable. This is also reflected in the marked
pre-occupation among learners with the end product of a learning sequence (Legutke & Thomas
1991: 205). Where process is a prominent concern, there are two main pedagogic reasons for the
inclusion of peer-group oral tasks in the language classroom. The first is related to the creation of
speaking opportunities, under the common assumption that practice makes perfect, but this is also
linked to fundamental beliefs that pedagogic resources should not be seen as only those controlled by
the teaching process. This is also what Kasper (1997: 353) refers to as the reinvention of the status
of the learner or non-native speaker as an interactional resource, rather than as a debilitating
condi-tion. Many available course materials do reflect this belief (see Savignon 1983; Richards &
Rodgers 1986; Brown 1994), but, as I have indicated, there is also considerable resistance to the use

89
of what are perceived to be deviant resources in the form of peer-group interactions. Such practice
opportunities present learners with platforms for feedback from other learners, which is not only
different from that received from native speakers, but, for Hong Kong learners, far in excess of any
such input they might receive in their normal study and living environments. Studies already
examined in relation to this question (e.g. Long 1981; Swain 1985; Pica 1988) provide positive
evidence that peer-group interaction can give learners linguistic data for the development of their
language competence and the development of their current interlanguage performance ability. In the
Hong Kong context, these arguments are all the more important since opportunities for interaction
with native speakers are generally infrequent or impossible.

The second main reason for the inclusion of oral tasks in the language syllabus is in relation to the
promotion of learner interaction for the behavioural benefits which might result from such
inter-action the promotion of interpersonal skills and the cognitive benefits of negotiation of
meaning and the creation of new knowledge. Tasks can initially be designed to stimulate both
co-operative interaction and meaningful conversational exchanges between learners within a
language syllabus that encompasses broader curriculum goals as well as specifications for the
realisation of target domain requirements. Ideally, these tasks should be related to criterial
performances and skills, established through needs and genre analyses. In order for the tasks to be
embedded in this way, they must also be clearly defined in terms of grading and sequencing which,
as this investigation has indicated, presents a major definitional problem. Sequencing is possible in
terms of design intention complexity or goal orientation for example but actual performance
outcomes may re-quire adjustments to be made. In this sense, a task-based syllabus is also an
emergent syllabus, subject to the same requirement for negotiated consensus on procedure as the
performance of the constituent tasks themselves. The value of such tasks in the wider language
syllabus depends on a number of factors, which we can roughly divide into two major linked areas
of task design and task motivational value. We can make a distinction between convergent practice
(focusing), which pro-vides the learner with opportunities for performing a specific target structure,
and divergent practice, which allows the learner to practise a wider range of competencies. Focusing
can be seen as a more gradable activity in the sense that learners can be focused on particular aspects
of the form or meaning of the target language system. A particularly important area here is pragmatic
knowledge, which enables learners to create or interpret discourse by relating utterances to the
intentions of language users in specific domain settings. This knowledge would include knowledge
of ideational functions (describing, classifying), manipulative functions (affecting our world),
instrumental func-tions (getting people to do things) and regulatory functions (controlling others).

Practice also varies according to the degree of control over language output (or input) which the
learners are given, or feel that they have. This gives us a theoretical continuum between controlled
and narrowly focused (convergent) practice using structural activities such as drills, clozes etc., and
controlled but unfocused (divergent) practice using, for example, a discussion task. It would be
erroneous to regard either focused or unfocused tasks as treatments to be applied in a clinical sense
to alleviate perceived linguistic deficiencies. The use of tasks will always be allied to the creation of
types of language opportunity or types of communicative interaction, and although specific task
types have been identified in this study as appearing to promote certain types of opportunity or
certain types of interaction, there could never be a generalisable correlation between task design and
task outcome. While keeping these questions of the relation of research to peda-gogic
implementation in mind, we may, however, make general assumptions that, for example,
information-gap tasks appear to offer the greatest opportunities for negotiated meaning and eventual
language acquisition, while, although they may be popular with students, discussion tasks generally
result in less negotiated meaning, and may, where they involve theatrical role-play, prove

90
demoti-vating.

In making pedagogic recommendations, we recognise that the notion of language practice through
task-based pedagogy is a complex multilayered conceptualisation, which embodies the general
principles of task construction described here. Memorability and other considerations of learner
perceptions of task relevance could also be strong motivational features in task performance, so that
task-based pedagogy should ideally involve the learners themselves in the negotiation of the syllabus
content (Yalden 1987). If learners are merely presented with tasks as one-off panaceas for perceived
language deficiencies, without being given any opportunity to become engaged in the task design
process, either through prior consultation on the reasons for implementation or post-performance
feedback, then a major related source of possible learning is ignored.

It is also quite clear that the central feature of any oral interactive task is an information gap, either in
a manifest or latent sense. This would be an overt and specific design feature in type A and B tasks,
and only generally implied in the content of other types. If there is no information gap, then there is
no meaningful reason to communicate either in transactional or interactional terms. Oral tasks should
also be used which are designed to elicit extended discourse rather than short or one- word answers,
in order to provide opportunities for attention to, and a focus on, enough of the language system for
substantial strategy employment and fluency practice to result from task-related language processing.
This processing should ideally be unrehearsed, mirroring genuine communi-cative situations. We
have seen that scenario tasks were originally designed to be rehearsed prior to a final performance,
but, arguably, the rehearsal phase of the performance is of greater benefit to language learners since
it is less prescribed, and therefore more authentic as an activity. Goal orientation is important, since
authentic communication requires a communicative purpose, even though co-operative intentions
may not be shared by interlocutors.

Oral tasks should also require learners to deal with language spontaneously, since genuine
communication involves the ability to attend to many linguistic and extra-linguistic factors at once.
This spontaneity may be compromised by prior awareness training of the pedagogic rationale behind
the usage of oral interactive tasks, but the Hong Kong learners in this investigation appeared to
respond well to an indication of outcome expectations and what would constitute an acceptable task
performance. They also responded favourably to an indication of the likely benefits to inter-language
development, which we can link to certain types of oral task performance. Here again, the issue of
student preference for discussion tasks and simulations can be raised in relation to the evidence that
specifically designed information-gap tasks promote a potentially richer situation for language
processing.

At the beginning of this discussion, I outlined an important contention from Dornyei & Thurell (1994:
40) that one of the main problems with task-based oral work in the classroom is that teachers feel
they can only make a random, intuition-based selection of general communicative activities. The
evidence presented in this study should give practitioners more confidence in selecting specific oral
task types for specific pedagogical purposes. Similarly, the objection that learners are capable of
playing havoc with even the most carefully designed and much used task (Breen 1987a: 23) is one
that is possibly true in terms of their potential to do so but not an objection which has any empirical
support from this research. It may well be that the negative result implied by the word havoc is
misplaced, and that we should actually be encouraging not just negotiation of meaning but
negotiation of the task design itself. This might be realised through the project concept, as a series of
linked and graded tasks which can provide an excellent communicative syllabus in itself (Candlin
& Edelhoff 1982; Yalden 1987; Legutke & Thomas 1991; van Lier 1996).

91
Another pedagogic implication of this research is that materials designers should not assume that all
language learners will need exposure to the same types of oral task. EAP programmes should be
offered that are context-specific wherever possible, and needs analyses should be concerned
pri-marily with developing rich descriptions of target domain contexts. For example, at UST, we
have an advanced course for Business majors, built around the regular inclusion of specific types of
oral interactive tasks, sequenced in terms of content and graded in terms of design complexity. A
weekly discussion task was designed to focus students on content issues relatable to their majors, but
also oral skills required for the more interactive learning styles in the final year at University and the
first year of local company training schemes. There is clearly a movement toward less formal, more
interactive styles of teaching (Lynch 1995). An internal report from the USTs Educational
Tech-nology Department (HKUST 1993) on teaching styles at UST also showed that many lecturers
increasingly favour an interactive, discussion-based approach to their lessons, and students report a
growing use of interactive methods by their subject instructors. The identification of such changes in
target processes and procedures is essential data for the continual modification of a task-based
syllabus.

Task-based syllabuses

Syllabus types have been broadly described as either synthetic or analytic (Wilkins 1976). In theory,
synthetic syllabuses break up components of the target language (which is assumed to be a definable
system) into graded chunks, to be learned in a sequence. In contrast, an analytic syllabus offers the
learner target language exemplars which have not been controlled for structure or lexis, and which
require the learner to work on and analyse the input. This division is also reflected in Whites type
A and type B syllabus distinctions, where type A focuses on what is to be learned and type B on how
it is to be learned. Figure 13 below shows how oral task characteristics can be broadly specified
under the two main syllabus types, the testingproduct type A and the developmentalprocess
type B (White 1988). Type A represents what Breen (1987b) has termed the formal syllabus and is
organised in such a way that it reflects the organisation or logic which is seen to be an inherent
feature of the language system. Hymes (1972) concept of communicative competence emphasised
that it represents how we relate our linguistic competence to our social use competence and that the
two systems are interdependent. The functional syllabus was a result of this, and moves us towards
Whites type B. It focuses on the learners knowledge of speech acts, or the purposes a learner may
achieve through language use in different domains. Using Hallidays (1978) analysis of the textual,
ideational and interpersonal functions of language, the formal type A syllabus focuses on a textual
knowledge system. The functional syllabus, building specifically upon ideas derived from speech act
theory, represents a paradigm shift towards interpersonal knowledge and the uses to which language
can be put in social situations. This has extended to a concern with the pragmatic uses of language,
and the negotiative power of the learner in relation to his or her strategic competence.

Task-based syllabuses are essentially process syllabuses. They seek to provide opportunities for
learners to gain knowledge of how correctness, appropriateness and meaningfulness can be
simultaneously achieved during relatable communication within classroom situations. Task-based
syllabuses organise and present what is to be achieved through teaching and learning, in terms of how
a learner may engage his or her communicative competence. According to Breen (1987b), the
task-based syllabus has the following features:

1. Communicative competence is represented as the undertaking and achievement of a range


of tasks.

92
2. Task completion relies directly on the communicative competence contributions of
learners.
3. It emphasises the learning process as appropriate syllabus content.

Breen points out that a task-based syllabus is really two syllabuses running parallel: one consists of
communication tasks, and the other of metacommunicative tasks, or learning-for-communication
tasks. The task-based syllabus brings together Hallidays analysis of the textual, ideational and
interpersonal functions of language.

Type A syllabus Type B syllabus

Task focus content skills


Objectives prescribed negotiated
Language focus formal accuracy communication
Discourse transactional interactional
Knowledge displaying creating
Information familiar new
Interaction ritualistic open-ended
Participation teacher-focused student-focused
Relationships hierarchical egalitarian
Roles set varied
teaching transmission facilitation
learning mimesis interpretation

Figure 13: Task characteristics and syllabus type (adapted from Pennington 1995: 8).

Such a task-based syllabus can incorporate many of the learner-centred features which are likely to
be more facilitative of communicative language practice than more traditional type A syllabus
approaches.

The specific global products which task-based syllabus designers might attempt to achieve are
co-operative interaction, critical thinking and learner awareness of the requirement to negotiate and
form hypotheses in order to create new knowledge about language form and use. However, there
would seem to be no reason, other than a desire for theoretical purity, why these two broad
para-digms should be viewed as mutually exclusive. In theory, task-based language learning
syllabuses are developmentalprocess type B syllabuses because they operate on the basis of a theory
of lan-guage learning which assumes learners should be encouraged to take responsibility for the
process-ing of type B input. However, in practice, this may not always be the case. In a study of
so-called communicative classes, Nunan (1987) found that accuracy issues predominated, and that
there were no signs of curriculum negotiation. Long & Crookes (1992: 43) also observed that
translation of theory into practice may not be as prevalent as we might think:

While several classroom studies have been conducted of various issues in TBLT [task-based language
teaching], no complete program that we know of has been implemented and evaluated which has fully
adopted even the basic characteristics of TBLT...

The origins of a task-based approach must be found at the curriculum level, where statements and
assumptions about language learning ultimately inform syllabus design. In this sense, we might also
talk of the task-based curriculum, although it is often only at the level of syllabus and lesson that a
presumption of the centrality of task as an organising principle becomes clear, and perhaps
questionable. Some task-based syllabuses might not be process-based at all, but actually based on

93
assumptions about language, which results in tasks being constructed in relation to theoretical or
stereotypical assumptions about form and function, rather than about task-related action (see Breen
1987b). Because of this, some ESP practitioners have seen the need for an increased emphasis on
actual or prototypical domain process rather than surface descriptions of domain settings in teaching
materials superficially styled as Business or Technical (see Arnold 1991; Courtney 1995a). Such
syllabuses may also consist of tasks which are not specifically related either to each other or to a
greater organising principle of content, as in the Bangalore syllabus (Prabhu 1987). These types of
task address what are seen by the task designers as essentially discrete situational or functional needs
of language.

Although an emphasis on the processes involved in task completion is an organising principle for a
developmentalprocess task-based syllabus, a consideration of task design and purpose is still central
to a TBLT syllabus (see Crookes 1986; Crookes & Long 1987). Some of the characteristics of TBLT
syllabuses are evident in tertiary language enhancement programmes, which contain pedagogic tasks
ideally derived from a needs analysis of the target tasks that learners will be re-quired to perform,
such as speaking in seminars and chairing meetings etc. This needs analysis, or gathering of
baseline data (Allwright & Bailey 1991: 74) is necessary to try to establish a rela-tionship between
classroom activity and target language and behaviour. Swales (1990: 76) gives examples of the kind
of target tasks which might be relatable to the University genre. His own example of academic
communications (ibid.: 77) involved not only relatable procedures, but also the use of relatable task
content.

This use of relatable content in a TBLT syllabus is similar to content usage in the adjunct model of
syllabus design, where much of the carrier material for English courses is derived from a targeted
content course. In this sense, task- and content-based syllabuses have much in common. This
con-tent can also include relatable behaviour as well as relatable information: a proficiency-based
curri-culum derives not from an analysis of the language code, but from an analysis of target
communi-cative behaviours, that is from an analysis of the tasks the learner will have to accomplish
in English. This allows for the development of a syllabus that is planned around the performance of
communicative tasks as outcomes.

Synthetic and analytic syllabuses need not be mutually exclusive and no practical dilemma is posed
by the use of tasks with a structural focus in an otherwise analytic task-based syllabus. For example,
learners can negotiate meaning in relation to grammatical problems, linked to their syllabus or
proficiency requirements. However, inclusion of this type of task poses theoretical threats to a
process syllabus, and may also pose practical threats by the appearance of discreteness which
grammar tasks may have. As we have seen, despite the necessity for research discreteness, pedagogic
tasks are not isolated events but, ideally within a process syllabus, linked to attempt to realise both
syllabus and curriculum goals.

The process syllabus can be distinguished from the task-based syllabus in one main way. It is not
directly concerned with organising the subject matter of language. The process syllabus designer
provides a framework which enables teachers and learners to organise their own activities and
content in an on-going and adaptive way.
Oral tasks in the Hong Kong secondary school curriculum

The need for more active participation by students in class oral work has been stated at many points
in this discussion. The findings of another investigation in Hong Kong add support to previously
reported findings:

94
The results of our research confirm those of previous studies ... in revealing an important shortcoming
in the preparation that many students receive for their university career. This is their lack of experience
in using English as a vehicle for spoken communication and active, exploratory learning in class. It
appears that their previous education has provided them with inadequate opportunities to practice
spoken English and has at the same time socialised them into adopting passive roles in the classroom.
As a result, many of them lack confidence in this domain and cannot perform without feelings of anxiety.
(Littlewood et al. 1996: 78)

As part of the data-gathering exercise for my study, I visited six Hong Kong secondary schools to
assess the feasibility of increasing the amount of oral practice opportunities available to students,
particularly in relation to the use of oral task types. There are serious resource constraints on the
inclusion of more interactive oral work in the present Form 6 and 7 English curriculum. Firstly, there
are the important practical factors of teacher qualifications and training, and class size. The attempts
to reduce class sizes by introducing split-class teaching into secondary classrooms (Hong Kong
Education Department 1990b) are not widespread enough, nor were they frequent enough within
participating schools, to make much impact (Courtney 1997). As the Hong Kong Education
Department (1990a) Guidelines state (p. 3), It does not necessarily mean that smaller classes would
automatically bring about small group teaching. However, it is clear from my research visits that in
schools where the split-class facility proposed by the Education Commission has been instituted, the
opportunity for effective use of oral interactive tasks is considerably enhanced, as is the effective
monitoring of such tasks. Where such reduced numbers were not possible, useful oral practice could
still be obtained, provided that classroom organisation was adequate.

The greatest influences on oral work in the Hong Kong secondary environment have probably been
the Hong Kong Examinations Authoritys oral examinations (Andrews & Fullilove 1994). The Use
of English oral examination at Form 7 level is weighted at 18% of the total examination mark, so the
backwash effect on classroom preparation is extensive. Beginning at Form 6, students are pre-pared
for the two main oral evaluation activities, individual oral presentation and then group discussion.
The HKEA discussion task could be classified as research typology type E1, a conver-gent
opinion-gap task.

The need for more active participation by students in class oral work has often been stated in previous
Education Department Guidelines (see Hong Kong Education Department 1996). At present, a
typical exposure pattern for English oral practice would be: Form 5 students have ten forty-minute
English lessons per six-day cycle, from which, two lessons are given to oral English, viz.
approximately 1 hour 20 minutes per week for oral practice; Form 6 students have eight lessons per
six-day cycle, with approximately 1 hour 30 minutes allocated to oral English work. However, in all
the schools I visited, although most of the English teachers devoted a great deal of extra time after
school to further training for oral examinations or related activities such as debates, oral tasks were
almost solely used in relation to examination preparation, and because of the target examination
format, only convergent discussion type tasks were practised.

This narrow and infrequent use of peer-group tasks was partly because of class-size constraints.
However, an average class size of forty students could feasibly be arranged into twenty pair-groups
without a change in classroom seating, or groups of four students by asking alternate rows to turn
round to face the students behind them. Both methods of class rearrangement were found to work
very well with minimal disruption during research visits. A good structure for a forty-minute lesson
based on the performance of one oral interactive task is:

Phase 1 input 10 minutes

95
Phase 2 set up 5 minutes
Phase 3 task performance 15 minutes
Phase 4 feedback 10 minutes

This gives at least fifteen minutes of oral practice to each student in the class, and the opportunity for
further oral interaction during the input and feedback phases. In mixed-ability classes, proficiency
pairings can be considered in cases of very low oral ability. The teacher can use prearranged
numbering to allocate students to weakstrong proficiency groupings, whether in pairs or groups of
three or four students. By pre-assigning students personal numbers, and asking them to pair or group
on the basis of these, negative comparisons can be minimised, since groupings will appear random to
students, and organisation is enhanced.

The predominant washback effect from the HKEA oral examinations was already evident in all of the
schools visited. Preparation for these examinations formed the basis for specific class oral work. The
examinations consist of a combination of simple interview techniques, oral presentation and group
discussion. Exclusive concentration on these formats would certainly deny students experience with
other types of oral interactive tasks. From the classroom trials with selected and adapted versions of
tasks from the research typology, it seems possible to include a wider range of suitably
contextualised tasks in the existing secondary English teaching syllabus. Task topic content can be
adjusted to suit student or subject content concerns, and task types can be linked together over
several weeks if necessary.

Research task types B1 and B2 were used with classes. Type B is an information-gap task which has
a jigsaw feature each participant has different information to offer for the group to finish the task.
Given the close proximity of pairs in the classroom situation, this was modified into an opinion-gap
task on the basis of common input. Both participants were given the same set of pictures and asked
to complete a short written summary of each in order to work out the overall story. The topic was
modified from that of Business and Management in the tertiary version, to the relative success of two
local clothing shops. This topic and task encouraged co-operative interaction in order to produce a
final written output and the generation of another product from the interaction gives the teacher a
good basis for later monitoring of the performance. Written output can either be used as the
foundation of another linked task, or can form the starting point for a teacher-directed phase to
exploit the original task input further. In this task version, the written output was speci-fically guided
with given vocabulary to practise conditional constructions. This type of information-gap task
characteristically produces short co-operative turns, usually clarification or repetition of previous
items.

This pattern emerged as participants rapidly tried to establish the nature of the gap, and exactly how
to complete the tasks. Often, this was in relation to the graphic input, typically in terms of the items
in the pictures as well as the dynamic nature of the story they represented. Strategies students used to
complete them were repetition of vocabulary and grammar items, as well as clarification on task
content and procedural points. These simple strategies are particularly useful for lower proficiency
learners, since they do not require very much conscious pre-planning.
Different task types can be combined to form linked activities. Tasks can also continue over several
lessons or an entire course a business simulation for example. Many problem-solving tasks involve
ranking followed by convergent discussion. In one of the schools visited, debating was an activity
which was encouraged, but this was apparently not a general practice. Debating is an activity which
lends itself to a chain-task approach and can be readily adapted to the classroom situation. It not only
offers students oral opportunities, but also encourages the development of critical thinking. This

96
process begins in the necessary research stage of any debate activity. Students can be asked to bring
information into an initial debate research session, and this can then form the basis of an initial
information-exchange task. This can be closed to enable students to make a list of possible arguments
or counter-arguments in the time allowed for the task activity, which can then lead to a discussion
task in a larger group, where the arguments are further added to and refined. At this stage, group work
can result in an all-class summary of the various positions available and the teacher can complete
coverage of the main arguments. Further lessons can deal with debating skills, or the skills of
preparation for presentation to the class. Candidates for the final teams can be presented to the class
as the basis for selection through simulation ranking tasks. Practice speeches by the teams can be
evaluated by using pair or group evaluation tasks and the final lesson on the process can involve the
whole class in a formal debate of the issues, with selected adjudicators, or other schools can be
involved in an inter-school competition. Follow-up lessons can deal with issues such as audience,
argumentation techniques and question formation and reply. The whole process of debate preparation
is a vehicle for effective oral work for the whole class, which can encompass the entire range of oral
interactive tasks and help develop the all-important facility of critical thinking about selected social
and international issues.

With problem-solving tasks that require a solution to, say, a grammar problem, pre-teaching could
take most of the previous lesson in terms of covering the general topic. These types of tasks are
relatively easy to invent following a lesson with a particular structural focus. They should usually
involve three phases:

1. Consideration of authentic published or corpus examples exemplifying the structure.


2. Discussion of a possible rule or rules apparently governing usage, or try to classify the
examples into a given framework.
3. Reaching a consensus on the solution.

Lastly, it has been suggested that we might instruct learners in the use of appropriate strategies during
oral task performance. If task strategy instruction is to be effective, a focus on the benefits of
interactive pedagogy is essential. Many students have indicated to me in informal conversations that
classroom oral interaction among peers is often viewed by them as playing games or not engaging
in real work, and this idea needs to be effectively countered. There certainly is the potential for
such a view from learners who are used to a mechanistic and examination-oriented curriculum.
Task-based learning requires a teaching paradigm shift not only on the part of the teacher, but also
the learner. My discussions with Form 6 and 7 teachers in Hong Kong schools (Courtney 1997)
support the view that the majority of learners perceive peer-group techniques as play rather than
work. This finding supports that by Futaba (1994), who found that pair work on communication
tasks was perceived to be not as useful as teacher-fronted lessons.

This view may also be linked to a perception by some teachers that classroom conversation is almost
synonymous with doing nothing (van Lier 1996), since it is viewed as a distraction from the
content-based exercises which make up the syllabus particularly if this latter is synthetically
organised. Conversation can be seen as a break from work, rather than the serious business of work
itself, and in teaching environments where much of the study time is spent in silent preparation, the
notion of deliberately stimulating conversation can appear to challenge the authority of the teacher.
Apart from their implications for teacher training, these perceptions endorse the need for more
specific learner awareness training in relation to strategy use and the likely benefits which might be
expected to accrue from oral task performance and practice in specific discourse strategies.

97
Oral tasks in the Hong Kong tertiary curriculum

Task-based and content-based organising principles for the language curriculum can be directly
linked. Indeed, a syllabus which consisted of a selection of oral task types, for example, without an
organising principle which had some content continuity, would not only suffer from lack of face
validity, but might also suffer from erratic motivation on the part of the learners. In one sense, the
design of a task-based syllabus within an academic context is made easier since the teaching setting
can be considered to be part of the target genre. This does not preclude needs analysis, which should
also take into account the discourse communities that learners hope to join in terms of post- graduate
research or industrial and commercial employment. The next stage includes evaluations of the design
and effectiveness of particular tasks to be embedded in the syllabus materials, using a methodology
similar to that employed in this investigation. This is a consideration of what Swales (1990: 72) calls
the place of task within a syllabus constructed in relation to genre considerations and the
assumptions regarding task performance and second language acquisition. The implications for EAP
oral pedagogy are that genre-specific oral syllabuses and oral tasks may be necessary, and that EAP
teachers need to prepare students for more interactive learning styles.

Brumfit (1984) emphasised the importance of contextualising oral tasks within the wider language
curriculum, stressing that tasks in themselves can never replace a syllabus, and that adequate
preparation of students for task performance is essential (cf. Ferris & Tagg 1996). Materials
designers should also not assume that all university students will need exposure to the same types of
oral task. EAP programmes that are context-specific should be offered wherever possible.

The identification of target task types is a key part of the process towards a task-based syllabus and
facilitates the construction of pedagogic tasks which form part of the syllabus. The construction of
these target task types for specific domains such as the tertiary learning environment has to be
governed by needs analyses which identify appropriate language skills which either form part of the
academic domain or can be expected to facilitate progress to a domain target through the learning of
particular enabling skills.

Specific-purpose oral tasks

One problem with advocating the use of specific tasks for specific purposes is that if they are not
contextualised in the wider syllabus, the assumed result is likely to produce formulaic and stilted
discourse output, unrelated to any broader description of a target discourse community. Arguably, all
pedagogic language learning tasks are, or should be, designed and used for clearly defined, specific
purposes, and the purposes may be either domain-related in terms of the task product or task-related
in terms of the process of task completion. The example of the use of interactive discussion tasks for
training purposes in relation to the domain of the airliner flight deck (Telfer 1993) might be taken as
the clearest end of a descriptive continuum, where the target product is safe and unambiguous cockpit
communication facilitated by the process of simulation and discussion.
The results from the present investigation, and from other studies (Mackey et al. 1992; Pica et al.
1993), suggest that different types of tasks can be expected to elicit different types of outcomes. In
this sense, oral tasks can be designed to elicit and promote specific types of oral performances,
perhaps in relation to longer turn lengths (Brown & Yule 1983a) other-evaluation or repair, different
types of information clarification techniques, or more effective topic management instruction. Since
the oral task performances in this investigation do show patterns which are statistically significant in
terms of participant strategy use, the results also suggest that oral tasks can be selected for specific

98
procedural requirements, which may in themselves be transferable process-oriented skills (effective
topic management, for example).

Other types of discussion tasks have been proposed for specific pedagogic purposes involving
peer-group oral responses or reviews in relation to tasks previously undertaken, or perhaps as part of
an on-going process of peer- and self-criticism of task performances. These discussion groups can be
set up with three distinct stages, offering decreasing levels of learner support and increasing levels
of communicative responsibility instructor-centred, instructor-supported and lastly, independent
(see Christopher & Ho 1996). This process essentially creates a task on a task, where each related
task can be seen as having both language acquisition value and content-specific value above that
assumed in the original pedagogic design. Cazden (1988: 126) refers to this point with the metaphor
of discourse as catalyst to characterise these essentially generative processes.

Tasks can also be selected or designed for specific content requirements rather than for their
interactional potential. This content might be subject-related where language enhancement itself is
given an adjunct role, or it may contain explicit language content itself, in the form of specific
grammar and vocabulary. It might also contain content input specifically related to a target domain
such as the career interview (Courtney 1996b) which elicits relatable process as well as relatable
products. Such tasks can be described as focused rather than specific (Fotos 1994).

Focused oral tasks

Loschky & Bley-Vroman (1990) argue that it is necessary to re-establish a link between grammar
and communication if we are to create classroom tasks that have the potential to influence the
learners competence base. They suggest that learners begin to restructure their interlanguage
particularly when they notice gaps in their knowledge, and that in order for this to happen, the
relevant aspects of grammar must be crucially involved in successful task performance. Successful
creation of appropriate oral tasks therefore requires that the task designer exercise a great deal of
control over the language used in the task. They claim that the two essential features of these focused
grammar tasks are that the grammar items are necessary to task completion, and that clear feedback
on successful goal completion is available.

The performance of a specifically focused task can result in the frequent use of particular
grammatical structures, depending on the input instructions and goals of the task. However, the
characteristics of any oral task type are often such that a particular structure is likely to arise naturally
(Chaudron 1985; Tarone & Parrish 1988). However, there is a danger in grammar-focused tasks that
they can often result in far less communicative activity in terms of negotiated meaning than other task
types. This is largely because, in order to complete the task, the interaction is more strictly focused
on language form a requirement for choosing words to complete a cloze, for example. This is
possibly more a design problem than an inherent weakness in adopting a structural focus as the topic
focus of the task. Fotos (op. cit.) approach appears to avoid these potential problems since the oral
task instructions are such that learners have to negotiate a solution which requires considerable
interaction and creativity.

Oral task evaluation

While this study has been largely about a type of evaluation of oral tasks for research purposes it
has not, so far, addressed the important question of the evaluation of oral task performance for

99
educational or vocational purposes. The relevance of these questions can be seen from the pre-ceding
discussions in relation to the possible connections between task design and performance features and
second language acquisition. Even if we adopt the apparently neutral position that oral tasks only
provide opportunities for oral performances, we still need a principled way of evaluating those
performances, viz. which type of tasks might produce the best opportunities for certain types of
performances. In short, if there is a relationship between oral task design and oral task performance,
as this investigation has suggested (and as oral testers implicitly assume), what are the best types of
tasks for oral evaluation purposes?

To answer this question in any detail, we need firstly to consider the question of testing purpose or
intention, and whether a relationship between specific oral task design intention and oral task
outcomes is adequate enough for acceptable oral test validity. A discussion task may present
candidates with an opportunity to interact fluently, but is general oral fluency what we want to test?
We need to know what it is that our oral task is trying to test, and to devise tasks that are relatable
both to the testing purpose and the testing target. This returns us again, finally, to a consideration of
the underlying tensions of accuracy and fluency in the context of task.

Firstly, my presupposition is that oral task evaluation within a process syllabus is more effectively
undertaken using a performance-referenced testing methodology, where direct rather than indirect or
system-referenced tests are employed. Perhaps this is not so much an assumption as a reasoned
connection between oral English tests in Hong Kong and what they are designed to test. Direct oral
tests for example would involve the elicitation of an outcome which would ideally demonstrate
particular target domain skills directly giving a sales presentation for example, or coping with a job
interview. However, an enduring problem in oral testing is that tests of learner competence are often
used to test learner performance. A related problem is that tests of target performance ability run the
risk of being tests only of ability in the examination context.

The extent to which the task types and the task performance measurements such as used in this
investigation can be used for assessment purposes has been a continuing issue in language testing
theory (see Bachman 1990). A review of some of the current oral testing provisions in Hong Kong
(Table 6) shows wide usage of these task types and provides further support for the study claim that
the research typology oral task types are generalisable to current pedagogic and evaluation usage.

Peer-group tasks possess a number of potential advantages for the evaluation of oral performance.
Most importantly, they provide a basis for assessment of oral outcomes that are nearer to authentic
contingent oral discourse. Related to this point, peer-group oral tasks are potentially less stressful for
candidates, since they interact with each other rather than the assessor. This feature also removes the
need for an assessor to act as interlocutor as well, thereby freeing the assessor to attempt a more
objective scoring of the test (and to also assess more candidates an essential practical consideration
in the local evaluation context).

100
Table 7: Oral examinations related to study research task types.
Typology task designation Task type
A information gap (one-way/two-way)
B jigsaw
C simulation ranking
D scenario role-play
E1 convergent discussion (problem-solving)
E2 divergent discussion
(opinion exchange)

Oral examination Oral task types with research typology Task format
designation
Hong Kong Use of English oral Stage 1 task type A (one-way) Stage 2 4 candidates
task type E1 2 examiners
CET(6) Chinese College English Stage 1 task type A (one-way) 1 examiner
Test oral level 6 (under Stage 2 task type E1 1 interlocutor
development) Stage 3 review questions 4 candidates
Trinity College London oral Stage 1 task type A (two-way) 1 examiner
Stage 2 task type E2 1 candidate
Stage 3 task type E2 + A
Stage 4 task type E1 + A
Institute of Linguists oral Stage 1 task type A (two-way) 1 examiner
Stage 2 task type A + D 1 candidate
Stage 3 task type E2
International English Language Stage 1 task type A (two-way) 1 examiner
Testing Service (IELTS) oral Stage 2 task type D 1 candidate
Stage 3 task type E2
TOEFL oral Stage 1 task type A (two-way) + D 1 examiner
Stage 2 task type E1 2 candidates
Stage 3 task type A (one-way)
Cambridge (UCLES) Level 3 Stage 1 task type A (two-way) 1 examiner
First Certificate in English (FCE) Stage 2 task type A (one-way) + B 1 interlocutor
oral Stage 3 task type E1 2 candidates
Stage 4 task type E2
Cambridge (UCLES) Level 4 Stage 1 task type A (two-way) 1 examiner
Certificate in Advanced English Stage 2 task type A (one-way) 1 interlocutor
(CAE) oral Stage 3 task type C 2 candidates
Stage 4 task type E2
Cambridge (UCLES) Level 5 Stage 1 task type A (one-way) 1 examiner
Certificate of Proficiency in English Stage 2 task type B 1, 2 or 3
(CPE) oral Stage 3 task type E2 candidates

Long & Crookes (1992) have claimed that criterion-referenced direct tests are the most appropriate
for evaluating oral task performances. However, as Robinson & Ross (1996) point out, the most
common application of criterion-referenced testing to second language programmes is based on
domain-referenced sampling of language items identified as being, to some significant degree,
domain-specific. It is obviously essential that we should be able to identify which items are both
representative of the domain, and specific to it, and domain-specific processes would also have to be
considered. The establishment, with some degree of credibility, of an adequate definition of an
assumed prototypical oral performance in a specific domain remains a central pursuit of the needs
analysis which must eventually inform task design. This definition should ideally include both
process and product relatable to an empirical reality in the case of product and a psychological,
cognitive reality in the case of process.

101
The criterion for success on tasks designed to test domain-related proficiency is the achievement of
a prescribed threshold, but this threshold seems often quite arbitrary, and related to subjective
band-scale assessments (such as the IELTS oral interview see Table 7). Such evaluation also
implicitly assumes that the language items and processes constituting the threshold definitions, are,
in themselves, teachable within the language curriculum. This is an important consideration in a
curriculum where language teachers may still be adopting a synthetic approach within an analytical
syllabus, with consequent loss of opportunity to develop language-related behaviour as well as the
language itself. This problem has already arisen at Forms 6 and 7 level in Hong Kong, where the
teaching of discussion skills is often done with an emphasis on memorisation of selected phrases for
the oral examination, an approach which tends to discourage the development of the strategic
competence required for threshold task compliance in the examination situation (see Andrews &
Fullilove 1994).

Ungeneralisable, non-system-referenced oral task evaluations are initially problematic in terms of


their relationship to target domains, although once established, their predictive ability of linguistic
and social performance in these domains is likely to be higher. In order to use any of the present
research tasks as part of a formal assessment programme, actual or prototypical descriptions of the
target domain should guide the construction of test items. The typology employed in this study
represents a generalisation of dyadic oral task activity and should ensure that task types could be
designed as relatable or representative vehicles for the testing of discrete or multiple aspects of
appropriate domain performance. For example, needs analysis has established that seminar
dis-cussion is a required oral performance activity for undergraduates at UST (Noakes & Wong
1997). A discussion task (type E) with relatable task content could provide the basis for the kind of
relatively unfocused or divergent outcome which would enable an evaluator to make an assessment
in terms not only of task content and topic management, but also of behavioural outcomes thought
desirable to co-operative and positive discussion appropriate at tertiary level. If the test requires less
emphasis on the acquisition of genre-based behavioural skills, and more on the summative status of
language acquisition itself (both competence and performance), an information-gap task might be a
more suitable vehicle, eliciting a greater variety of strategy usage and turn-taking opportunities from
a more linguistically focused task.

A discussion task similar to the type E tasks in this study has been used since 1993 as the main
evaluation vehicle in the oral component of the HKEA Use of English examination. Great care is
taken in the construction of the battery of test tasks for this examination to ensure that the candidates
are not disadvantaged by task input (content) material which is beyond their current linguistic or
social experience. Part One of the examination is designed to familiarise them with this material
before they proceed to the group discussion in Part Two (see Andrews & Fullilove 1994). Evaluation
is holistic and normative, but raters are made aware of desirable performance features, centred on the
notions of co-operation and appropriate turn-taking and turn-holding techniques.

Discussion tasks may prove difficult for many candidates, simply because of the degree of
com-municative responsibility required. This can be linked to the requirement for adequate levels of
strategic competence rather than pre-arranged functionalnotional routines which can be rehearsed.
However, the arguments for the benefits of more natural arenas for language practice apply equally
to evaluation. Convergent discussion tasks probably have greater value than divergent tasks for
evaluation purposes, because of the element of output structuring which can be indicated in the task
input instructions. Many of my case study subjects referred to this aspect of such tasks and viewed
such output support as conducive to better performance which, in turn, offers the potential for better
oral assessment.

102
However, formative task-based assessment is possibly preferable to summative assessment. This is
because task performances can vary for a variety of reasons, so it is important to build up a profile of
a learners spoken ability in relation to a number of different types of task performance. Evaluation
in a syllabus context (as opposed to one-off testing) should be an ongoing process which includes
not just oral task performance, but the affective learning environment and the learners response to it
(e.g. motivation and learning participation outside task). An effective test suite partly to achieve this
ideal would include a range of tasks from the research typology (UCLES 1995). Each task should be
so constructed that a consistency of input and, ideally, a consistency of likely performance output
patterns, is produced. As I have indicated, consistent output is more problematic than consistent task
input, but all task-based oral tests assume such consistency with respect to criterial or band-scale
description of both linguistic and performance aspects in relation to an assumed or defined domain
description.

However, there is an obvious danger in using attainment of target-domain behavioural prescriptions


as a criterion of success or otherwise on oral tasks designed to evaluate language proficiency,
parti-cularly where that proficiency includes a measure of strategic competence. For this reason,
scoring procedures must be related more to the process of goal completion and not the actual
attainment of task goals. A scoring technique which uses the notion of compliance with
content-based goals can produce highly subjective results, since its quantitative approach often
neglects the more important quality of the interaction how effectively and naturally the information
is exchanged. Ideally, oral evaluation would involve discourse analysis of a selection of the turn and
strategy features utilised here, but practical constraints would severely inhibit this approach.

A review of oral test formats currently used in Hong Kong reveals a widening use of performance-
based peer-group formats. However, it may be the case that administrative constraints adversely
compromise the validity of such tests, not in terms of design but in terms of rater recruitment and
reliability. Evaluation of oral test performances at a superficial level of discourse is often the result,
with test suites being restricted to only one interactive task type. Ideally, candidates would be
assessed in relation to a range of oral task types, designed to elicit different types of assessable
out-comes in relatable contexts. Some of these task types might be expected to facilitate fluency
rather than accuracy, as a result of design intention to allow an evaluative focus on strategic
competence and language process rather than concerns of accuracy in relation to discrete language
items. Here again, the prior agreement of raters is essential, in order to ensure that both accuracy and
fluency are adequately defined in relation to expected task performance.

Fluency and accuracy

The concept of fluency in relation to peer-group oral task performance is a complex one. The concept
assumes that speakers set out to produce discourse that is comprehensible, easy to follow and free
from errors and breakdowns in communication (Richards 1990: 75). Unfortunately, native speakers
do not often meet these requirements, employing hesitation and so called vague lan-guage
(Channell 1994) in a way which should lead us to question many definitions of fluency. Successful
task completion strategies are also likely to include the imprecise language which characterises
native speaker communication.

Accuracy is often seen as a component of fluency, but a preoccupation with precise grammatical
accuracy in the evaluation of task-based performances is likely to be on the basis of a misleading

103
ideal for second language learners. Normal native speaker communication involves interruptions,
false starts and grammatically incomplete sections. Since vagueness in this sense is probably due to
a deliberate choice by speakers in relation to their contribution to the communicative message then
this would seem to support the view that we should regard words and expressions such as you know
and thing as part of an overall communicative strategy. The usage of vague language would come
under the category of compensation strategies, the employment of which should not always be
regarded as evidence of a lack of fluency or competence deficiencies, but as evidence of strategic
competence attempts by oral task participants to achieve successful communication by rethinking
their approaches to a perceived problem.

With regard to the accuracy fears identified earlier in relation to the concern that peer feedback was
likely to be inaccurate, very little evidence was found in the task transcripts of participants
misinforming each other in relation to the enabling language of the task performance (although there
was some consolidation of mispronounced real names from task input). It must be said, however, that
there was also very little evidence of task participants attempting to inform each other in relation to
language use something we perhaps could not expect unless proficiency differences or task content
and instructions made employment of evaluative strategies (particularly ECR) more necessary.

Littlewood et al. (1996) found a strong preference (70%) from a large sample of Hong Kong
under-graduates for correction of their perceived oral errors. The finding that the desire of Hong
Kong undergraduates to participate in active communication in English is almost as strong as their
desire to have their oral errors corrected is not surprising:

In the views of most students, then, there is not necessarily any conflict between the demands of fluency
and accuracy in communication; they wish to achieve a high level of performance in both aspects of
language use. (ibid.: 79)

Accuracy concerns might therefore be best addressed prior to task performance but, in terms of the
morphosyntactic form of task-related output, there are objections to the idea that prior instruction
could help to produce more accurate output (cf. Matthei & Roeper 1983: 162):

It is hard to see how we can manipulate the input to the speech production mechanisms ... We can, for
example, ask our subjects to describe pictures or the actions depicted in little movies and thus gain some
control over what our subjects will talk about. But we cannot manipulate other critical variables, like
what syntactic form our subjects sentences will take and what words they will choose to put in their
sentences.

The findings of this investigation do suggest that through specific oral task design and
imple-mentation, learners can be engaged in interactional activity which can include a specific focus
on form. An affective situation can be created in which learners interact with other learners in
parti-cular ways and where particular input structures (Loschky & Bley-Vroman 1990) are
manipulated during task performance. Tasks can also be utilised that promote self- and other- repair
sequences, and learners can be made aware of the benefits of attempting to use more evaluative
strategies on task, in an appropriate and co-operative way.

There has been much controversy surrounding the question of performance tasks and accuracy. In a
communicative task-based oral syllabus, a balance has to be maintained between the demands for
fluency practice and the equally important demand for attention to appropriate and accurate language
form. Together with these concerns, the tasks embedded in a task-based syllabus have to retain two
important features of the target language a speakers right to choose the form of an utterance, and

104
the resulting unpredictability of that utterance. Learners should also be made aware of the
unpredictability of target domains, and task and syllabus design should seek to sensitise learners to
the main factors likely to affect genre descriptions and related discourse. If such sensitivity in task
construction is missing, possibly through a lack of understanding in relation to either domain
requirements or task purpose, then the accuracy of the tasks themselves, as domain- specific design
constructs, will be adversely affected.

While this study has found no evidence of a connection between oral task performance and the
promotion of learner inaccuracy, it remains true that research in this area has shown very mixed
results, with some evidence from studies of immersion programmes in Canada that increased peer
interaction does lead to reinforcement of production errors. However, we have has also reviewed
evidence that peer-group work can also assist the production accuracy of learners, although this may
be in a somewhat restricted way (see Gass & Varonis 1989; Pica et al.1993). Our findings support
those from these studies, that it is very rare for learners overtly to incorporate other learners errors
in their own task-based production.

Skehan (1989: 21) distinguishes among fluency, complexity and accuracy in task completion. In this
context, fluency concerns the users skill in producing language in real time almost certainly
drawing upon a lexicalised repertoire of exemplars to achieve this. Complexity emphasises
developmental stages, subordination, elaborateness, risk taking and so on, i.e. factors which suggest
that the learner/test-taker is prepared to use language at the cutting edge of interlanguage
development and perhaps attempt, through communication, to stretch the underlying linguistic
competence that is available. Finally, accuracy is concerned with the avoidance of error and a
concern on the language users part to use a slightly more restricted repertoire of language in such
a way that it can be controlled more effectively even if it may, as a result, be less complex. Skehan
sees the central issue that influences the balance that is achieved between the goals of fluency,
complexity and accuracy as the way language users draw upon attentional resources. The language
user needs to cope with a variety of processing conditions and adapt the way resources are used and
different communication goals and priorities are set. This is also Bachmans (1990) position, where
the strategic competence necessary for this mediates between communicational intentions, context
of situation and the competence base itself the background knowledge of the language system.
Skehans conception of strategic competence offers more precision as to how the labels of
goal-setting, assessment and planning will be translated into reality. Skehan postulated stable and
generalised capacities for handling these attentional demands. He points out that oral testers need to
consider how to sample processing conditions in a more effective manner. This leads to his schema
for the set of factors which underlie task performance:

(i) language complexity syntactic and lexical


(ii) cognitive complexity familiarity with task content and type
(iii) processing reasoning operations required, nature of input material, degree of
organisation of input material.

These factors come into play during three stages of pre-, during and post-task activities which lead
us to the following framework for task implementation:

Table 8: Framework for task implementation.

105
Stage Goal Typical techniques
Linguistic introduce new forms to the Explicit and implicit teaching
interlanguage repertoire and consciousness raising
Cognitive Reduce cognitive load Plan linguistically and
cognitively observe similar
tasks
Push learners to express more Plan
complex ideas Observe
Task calibration Increase or reduce difficulty Introduce surprise
Provide visual support
Manipulate pressure Change time conditions
Scope for control
Post task Consciousness raise Salience of post-task public
performance

Interestingly, this implementation framework also incorporates aspects of pushed output theory and
Breens (1987b) recommendations for the use of metacognitive enabling tasks.

Conclusion

There is undeniably a strand of behaviourism in all attempts to link task input stimuli with output
performance, and while behaviourism is not a totally discredited theory, it does have limited
explanatory power. Ultimately, the value of descriptive research into task performance should be
seen more in terms of improving affective, motivationally-based descriptions what learners feel
about the oral task and their performances in order to improve the oral tasks themselves.

Improved description is also necessary in relation to the research constructs we use to investigate the
effects of task. There is a crisis of representation in applied linguistics, as there is in modern
sociology how best to represent the social and cognitive processes that we have so much data about
and which functional models have failed to provide explanatory power for. As I have tried to show,
the concepts of task input, negotiated meaning and performance output are often used metaphorically
to symbolise different task-related or learner-related processes which are themselves hypothesis
about the nature of task-based interaction. The metalanguage of task is in this way a constant threat
to the psychological reality of the construction. It is all too easy for task and syllabus designers to
reify not only the perceived components of the construct of task, but also to discuss task itself as a
convenient organising metaphor for an information-processing model which assumes a modular
rather than a developmental and experiential model of language acquisition. Information-processing
models are misleadingly deterministic in the sense that they assume that defined input must produce
a definable output. This simplified approach is often seen in oral testing, where the lack of definition
with regard to the assumed processes of language production in the testing and target domains results
in testing tasks that are neither related to testing purpose or relatable to target domains.

I have tried to outline what I see as the limitations of these constructs and have also indicated that
task is best seen as only a component of a wider syllabus, rather than as a kind of syllabus in
miniature. In this context, the effect of a specific task performance on a learners developing
interlanguage is likely to be accumulative and subtle in ways that require non-deterministic research
approaches to illuminate. The opportunities presented to learners in a task-based approach are more
likely to encompass more of the characteristics of effective learning situations than many other types
of pedagogy. This is because oral tasks present opportunities for learners to negotiate actively
towards accurate, appropriate and acceptable meaning within a learning context which itself is more

106
likely to represent the contingent but still essentially unprescribed nature of naturally occurring
conversation.

A task-based language syllabus which gives prominence to focused task-based interaction therefore
incorporates more of the learner-centred features which this study has shown to be more facilitative
of communicative language practice than other more synthetic approaches to syllabus design. Such
a syllabus should promote contingent interaction, co-operative negotiation opportunities, the
development of strategic competence and learner awareness of the possibilities and potential of
task-based peer-group oral work in the language classroom or language laboratory. All this is also
largely dependent on training both teachers and learners for task process management, an
observation which has been implicit in much of the foregoing discussion.

The process of task and syllabus design also has to be governed partly by the identification of
required prototypical oral performances, and partly by practical considerations concerning how best
to effect the necessary compromises between the often conflicting demands of subject teachers,
language learners, administrators, language theorists and employers.

In terms of linking task designs and syllabus types to language acquisition, we should be content to
establish not proof of a connection between task and syllabus to language acquisition in the
posi-tivistic sense, but an accumulation of inferential evidence both supportive and contrary
based on improved description, particularly from learners themselves. This evidence should be from
both observation and participation within a task-based syllabus, and attempt to investigate learner
moti-vation for task interaction, and learner motivation within specific task performances. As this
study has tried to show, the social context of performance and the learners attempts continually to
re-define his or her position in this dynamic context are possibly the richest areas of enquiry in task
performance analysis. Investigation of these areas necessarily broadens the debate from an
essen-tially narrow, behaviouristic argument, to one with sociological, sociolinguistic and
psychological dimensions.

In order to obtain this broader perspective, there would seem to be room for both input-output
theorists and those investigating the social contextualisation of oral task performances. Qualitative
enquiry is essential for the validation and illumination of quantitative investigation where the social
context of performance is taken to be one that changes as the learners perceptions of the task
changes. As such, the investigation of this dynamic context requires the realisation that the concept
of negotiated meaning, so central to the description of learner interaction on task, must be seen as an
attempt by task participants to negotiate not just task-based meaning but also an acceptable
consensus in relation to their own beliefs and attitudes concerning the process and value of task
interaction with specific partners. It is hoped that this type of qualified, bi-methodological
investigation, represents what Yule & Tarone (1991: 170) recommended as a return to the more
humble approach of describing both input and learner performance in interaction, and refraining
from making claims about acquisition which are based upon untested assumptions.

References

107
Allwright, R. & Bailey, K. M. 1991. Focus on the language classroom. An introduction to classroom research
for language teachers. London: Cambridge University Press.
Ames, C. 1986. Effective motivation: The contribution of the learning environment. In R. S. Feldman (Ed.),
The social psychology of education (2nd ed. 1990, pp. 23556). Cambridge: Cambridge University Press.
Anderson, J. R. 1985. Cognitive psychology and its implications. New York NY: W. H. Freedman.
Anderson, A. & Lynch, T. 1988. Listening. Oxford: Oxford University Press.
Anderson, A., Clark, A. & Mullin, J. 1989. The development of referential communication skills: Interac-tions
between speakers and listeners in extended dialogues. Paper presented at the 3rd EARLI Conference,
Madrid.
1991. Introducing information in dialogues: How young speakers refer and how young listeners
respond. Journal of Child Language 18, 66387.
Anderson, A., Bader, M., Gurman Bard, E., Boyle, E., Doherty, G., Garrod, S., Isard, S., Kowtko, J.,
McAllister, J., Miller, J., Sotillo, C., Thompson, H. & Weinert, R. 1993. The HCRC map task corpus.
University of Edinburgh and University of Glasgow.
Andrews, S. & Fullilove, J. 1994. Assessing spoken English in public examinations why and how? In Boyle
& Falvey, pp. 5785.
Ansbacher, H. L. 1951. The history of the leaderless group discussion technique. Psychological Bulletin 48,
38390.
Arnold, E. 1991. Authenticity revisited: How real is real? English for Specific Purposes 10, 23744.
Asker, B. 1997. Identifying orally reticent students. [Online.] Available at: sol.aston.ac.uk/lsu/lsub8ba. html.
Aston, G. 1986. Trouble-shooting in interaction with learners: The more the merrier? Applied Linguistics 7,
2, 12843.
1993. Notes on the interlanguage of comity. In G. Kasper & S. Blum-Kulka (Eds), Interlanguage
pragmatics (pp. 22450). New York NY: Oxford University Press.
Atkinson, J. M. & Heritage, J. 1984. Structures of social action: Studies in conversational analysis.
Cambridge: Cambridge University Press.
Atkinson, P. 1996. Sociological readings and re-readings. Singapore: Avebury.
Austin, J. L. 1962. How to do things with words. Oxford: Oxford University Press.
Bachman, L. 1990. Fundamental considerations in language testing. Oxford: Oxford University Press.
Bachman, L. & Palmer, 1996. Language testing in practice: Designing and developing useful language tests.
Oxford: Oxford University Press.
Bailey, K. M. 1983. Competitiveness and anxiety in adult second language acquisition: Looking at and
through the diary studies. In Seliger & Long, pp. 67103.
1985. Classroom-centered research on language teaching and learning. In M. Celce-Murcia (Ed.),
Beyond basics: Issues and research in TESOL (pp. 96121). Rowley MA: Newbury House.
Bailey, N. 1989. Theoretical implications of the acquisition of the English simple past and past progressive:
Putting together the pieces of the puzzle. In S. M. Gass, C. Madden, D. Preston & L. Selinker (Eds),
Variation in second language acquisition: Psycholinguistic issues (pp. 10924). Clevedon: Multilingual
Matters.
Barnes, D. 1976. From communication to curriculum. Harmondsworth: Penguin.
Bass, B. M. 1950. The leaderless group discussion technique. Personnel Psychology 3, 1732.
Beebe, L. M. 1988a. Five sociolinguistic approaches to second language acquisition. In Beebe (1988b), pp.
4377.
(Ed.). 1988b. Issues in second language acquisition: Multiple perspectives. New York NY: Newbury
House.
Berwick, R. 1993. Towards an educational framework for teacher-led tasks. In Crookes & Gass, pp. 97124.
Bialystok, E. 1983. Some factors in the selection and implementation of communication strategies. In Frch
& Kasper (1983), pp. 10018.
1990. Communication strategies: A psychological analysis of second-language use. Oxford:
Blackwell.
Bialystok, E. & Sharwood Smith, M. 1985. Interlanguage is not a state of mind: An evaluation of the construct
for second-language acquisition. Applied Linguistics 6, 10117.
Bly, B. 1993. Unco-operative language and the negotiation of meaning (conversation). PhD, Stanford

108
University.
Bohannon, J. N. & Stanowicz, L. 1988. The issue of negative evidence: Adult responses to childrens language
errors. Developmental Psychology 24, 6849.
Bongaerts, T. & Poulisse, N. 1989. Communication strategies in L1 and L2: Same or different? Applied
Linguistics 10, 3245.
Boswood, T. & Mariott, A. 1994. Ethnography for specific purposes: Teaching and training in parallel.
English for Specific Purposes 13, 1, 321.
Boyle, J. & Falvey, P. (Eds). 1994. English language testing in Hong Kong. Hong Kong: Chinese Uni-versity
Press.
Breen, M. 1987a. Learner contributions to task design. In Candlin & Murphy, pp. 2346.
1987b. Contemporary paradigms in syllabus design. Parts 1 & 2. Language Teaching 20, 2, 8192; 3,
15774.
Breen, M. & Candlin, C. N. 1980. The essentials of a communicative curriculum in language teaching. Applied
Linguistics 1, 89112.
Brown, G., Anderson K., Shilcock R. & Yule, G. 1984. Teaching talk. Cambridge: Cambridge University
Press.
Brown, G. & Yule, G. 1983a. Teaching the spoken language. An approach based on the analysis of
conversational English. Cambridge: Cambridge University Press.
1983b. Discourse analysis. Cambridge: Cambridge University Press.
Brown, H. 1994. Teaching by principles. Englewood Cliffs NJ: Prentice Hall.
Brown, J. D. 1988. Understanding research in second language learning. Cambridge: Cambridge Uni-versity
Press.
Brown, J. S., Collins, A. & Duguid, P. 1989. Situated cognition and the culture of learning. Educational
Researcher 18, 1, 3242.
Brown, R. 1991. Group work, task difference and second language acquisition. Applied linguistics 12, 1, 112.
Brumfit, C. J. 1984. Communicative methodology in language teaching. Cambridge: Cambridge University
Press.
Bruton, A. & Samuda, V. 1980. Learner and teacher roles in the treatment of oral error in group work. RELC
Journal 11, 4963.
Bygate, M. 1988. Units of oral expression and language learning in small group interaction. Applied
Linguistics 9, 1, 5982.
Cameron, D. 1995. Verbal hygiene. London & New York NY: Routledge.
Canale, M. & Swain, M. 1980. Theoretical bases of communicative approaches to second language teaching
and testing. Applied Linguistics 1, 1, 147.
Candlin, C. N. 1987. Towards task-based language learning. In Candlin & Murphy, pp. 522.
Candlin, C. N. & Edelhoff, C. 1982. Challenges: Teachers guide. London: Longman.
Candlin, C. N. & Murphy, D. F. (Eds). 1987. Language learning tasks. Lancaster practical papers in English
language education (Vol. 7). Englewood Cliffs NJ & London: Prentice-Hall International.
Carmichael, S. 1996. Negotiation of meaning in small-group discussions. In Ho & Carmichael, pp. 8298.
Carroll, S. E. 1995. The irrelevance of verbal feedback to language learning. In L. Eubank, L. Selinker & M.
Sharwood Smith (Eds), The current state of interlanguage. Studies in honor of William E. Rutherford (pp.
7388). Amsterdam & Philadelphia PA: John Benjamins.
Carroll, S. E. & Swain, M. 1993. Explicit and implicit negative feedback: An empirical study of the learning
of linguistic generalisations. Studies in Second Language Acquisition 16, 20530.
Cazden, C. B. 1988. Classroom discourse: The language of teaching and learning. Portsmouth NH:
Heinemann.
Channell, J. 1994. Vague language. Cambridge: Cambridge University Press.
Chaudron, C. 1977. A descriptive model of discourse in the corrective treatment of learners errors. Language
Learning 27, 2946.
1983. Research on metalinguistic judgments: A review of theory, method and results. Language
Learning 33, 34377.
1985. Comprehension, comprehensibility and learning in the second language classroom. Studies in
Second Language Acquisition 7, 21632.

109
1988. Second language classrooms: Research on teaching and learning. Cambridge: Cambridge
University Press.
Chomsky, N. 1965. Aspects of the theory of syntax. Boston MA: MIT Press.
Christopher, E. & Ho, S. 1996. Lights, camera, action: Exploring and exploiting films in self-access learning.
In R. Pemberton, E. S. L. Li, W. W. F. Or & H. D. Pierson (Eds), Taking control: Autonomy in language
learning (pp. 185200). Hong Kong: Hong Kong University Press.
Chun, A. E., Day, R. R., Chenoweth, N. A. & Luppescu, S. 1982. Errors, interaction, and correction: A study
of nativenon-native conversations. TESOL Quarterly 16, 53747.
Clark, J. A., Scarino, A. & Brownell, J. 1991. Latest developments in the targets and target-related assessment
(TTRA) project. Paper presented at the Institute of Language in Education International Conference, Hong
Kong, December.
Cohen, A. 1984. Studying second language learning strategies: How do we get the information? Applied
Linguistics 5, 2, 10112.
Coulthard, M. 1977. An introduction to discourse analysis. Harlow: Longman.
Courtney, M. 1988. Some initial considerations for course design. English for Specific Purposes 7, 195203.
1991. Language across the technical curriculum. English Studies 6, 506.
1994. The effect of instruction on undergraduate oral performance. Report for the University Grants
Committee. Hong Kong: Language Centre, Hong Kong University of Science and Technology.
1995a. Task-based language teaching for academic purposes. In Wong & Green, pp. 8395
1995b. Evaluating discussions. Proceedings of the Third MELTA Biennial International Conference
(pp. 1119). Kuala Lumpur: MELTA.
1996a. Talking to learn selecting and using peer group oral tasks. English Language Teaching
Journal 50, 4, 31825.
1996b. Preparing for interviews. In D. Gardner & L. Miller (Eds), Tasks for independent language
learning (pp. 1045). Alexandra VA: TESOL.
1997. Oral interactive task pedagogy and evaluation. Language Fund Research Report No. E/064/94A.
Hong Kong: Author.
Courtney, M. & Wong, M. Y. 1996. Peer group discourse. Description and causality. In Ho & Carmichael, pp.
7381.
Crookall, D. & Oxford, R. (Eds). 1990. Simulation, gaming and language learning. New York NY: Newbury
House.
Crookes, G. 1986. Task classification: A cross disciplinary review. Honolulu HI: Center for Second Language
Classroom Research, University of Hawaii (Technical Report No. 4).
Crookes, G. & Gass, S. M. (Eds). 1993. Tasks and language learning: Integrating theory and practice.
Clevedon: Multilingual Matters.
Crookes, G. & Long, M. H. 1987. Task-based second language teaching: A brief report. Modern English
Teacher (Tokyo) 24, 5, 268; 6, 203.
Crookes, G. & Rulon, K. A. 1988. Topic and feedback in native speaker/non-native speaker conversation.
TESOL Quarterly 22, 67581.
Crookes, G. & Schmidt, R. 1991. Motivation: Reopening the research agenda. Language Learning 41,
469512.
Cross, T. G. 1977. Mothers speech adjustments: The contribution of selected child listener variables. In C. E.
Snow & C. A. Ferguson (Eds), Talking to children: Language input and acquisition (pp. 15188).
Cambridge: Cambridge University Press.
Day, R. R. (Ed.). 1986. Talking to learn: Conversation in second language acquisition. Cambridge MA:
Newbury House.
Deci, E. L. & Ryan, R. M. 1992. The initiation and regulation of intrinsically motivated learning and
achievement. In A. K. Boggiano & T. S. Pittman (Eds), Achievement and motivation: A
social-developmental perspective (pp. 936). Cambridge: Cambridge University Press.
Demetras, M. J., Post, K. N. & Snow, C. E. 1986. Feedback to first language learners. The role of repetitions
and clarification questions. Journal of Child Language 13, 27592.
Di Pietro, R. 1987. Strategic interaction: Learning languages through scenarios. Cambridge: Cambridge
University Press.

110
Donmoyer, R. 1990. Generalizability and the single case study. In E. W. Eisner & A. Peshkin (Eds),
Quali-tative inquiry in education: The continuing debate (pp. 175200). New York NY: Teachers
College Press.
Dornyei, Z. & Thurell, S. 1994. Teaching conversational skills intensively: Course content and rationale. ELT
Journal, 48, 1, 409.
Doughty, C. & Pica, T. 1986. Information gap tasks, an aid to second language acquisition? TESOL Quarterly
20, 30525.
Doyle, W. 1983. Academic work. Review of Educational Research 53, 2, 15999.
Driver, M. & Streufert, S. 1969. Integrative complexity: An approach to individuals and groups as
information-processing systems. Administrative Science Quarterly 14, 2, 27285.
Du-Babcock, B. 1996. Top management and turn taking in professional communication: First versus second
language strategies. Perspectives 8, 1, 139.
Duff, P.A. 1986. Another look at interlanguage talk: Taking task to task. In Day, pp. 14781.
1993. Tasks and interlanguage performance. In Crookes & Gass, pp. 5795.
Ellis, R. 1982. Discourse processes in classroom second language development. PhD, University of London.
1984. The role of instruction in second language acquisition. In D. N. Singleton & D. G. Little (Eds),
Language learning in formal and informal contexts (pp. 6788). Dublin: IRAAL.
1985. Understanding second language acquisition. Oxford: Oxford University Press.
1990. Instructed second language acquisition. Learning in the classroom. London: Blackwell.
1992. Second language acquisition and language pedagogy. Clevedon: Multilingual Matters.
Ellis, R. & Rathbone, M. 1987. The acquisition of German in a classroom context. London: Ealing College of
Higher Education.
Erickson, F. 1981. Some approaches to inquiry in schoolcommunity ethnography. In H. T. Trueba, G. P.
Guthrie & K. H. Au (Eds), Culture and the bilingual classroom: Studies in classroom ethnography (pp.
1735). Rowley MA: Newbury House.
Frch, C. & Kasper, G. (Eds). 1983. Strategies in interlanguage communication. London: Longman.
1984. Two ways of defining communication strategies. Language Learning 34, 1, 4563.
1986. The role of comprehension in second language learning. Applied linguistics 7, 3, 25774.
(Eds). 1987. Introspection in second language research. Clevedon: Multilingual Matters.
Fairclough, N. 1995. Critical discourse analysis: The critical study of language. London & New York NY:
Longman.
Fanselow, J. F. 1977. Beyond Rashomon conceptualizing and describing the teaching act. TESOL Quarterly
11, 1, 1739.
Ferris, D. & Tagg, T. 1996. Academic oral communication needs of EAP learners: What subject-matter
instructors actually require. TESOL Quarterly 30, 1, 3157.
Fillmore, W. 1982. Instructional language as linguistic input: Second language learning in classrooms. In L.
Wilkinson (Ed.), Communicating in the classroom (pp. 13152). New York NY: Academic Press.
1985. When does teacher talk work as input? In Gass & Madden, pp. 12036.
Flanders, N. A. 1970. Analyzing teaching behavior. Reading MA: Addison-Wesley.
Fleishman, E. A. & Quaintance, M. K. 1984. Taxonomies of human performance. New York NY: Academic
Press.
Flowerdew, J. 1990. Problems of speech act theory from an applied perspective. Language Learning 40, 1,
79105.
Forde, K. 1996. The effects of gender and proficiency on oral self and peer assessments. In B. Ma (Ed.),
English Language Studies Working Papers 1 (pp. 3447). Hong Kong: City University of Hong Kong.
Fotos, S. S. 1994. Integrating grammar instruction and communicative language use through grammar
consciousness raising tasks. TESOL Quarterly 28, 2, 32351.
Fraser, J. M. 1946. An experiment with group methods in the selection of trainees for senior management
positions. Occupational Psychology 20, 6367.
Freund, J. E. 1984 Modern elementary statistics. 6th ed. Englewood Cliffs NJ: Prentice Hall.
Fulcher, G. 1996. Does thick description lead to smart tests? A data-based approach to rating scale
con-struction. Language Testing 13, 2, 20838.
Futaba, S. 1994. Second language acquisition through negotiation. A case of non-natives who share the same

111
first language. PhD, University of Pennsylvania.
Gardner, R. C. & MacIntyre, P. D. 1993. A students contribution to second language learning. Part II:
Affective variables. Language Teaching 26, 111.
Gass, S. M. 1988. Integrating research areas: A framework for second language studies. Applied Linguistics
9, 198217.
Gass, S. M. & Madden, C. G. (Eds). 1985. Input in second language acquisition: A book of readings. Rowley
MA: Newbury House.
Gass, S. M. & Varonis, E. M. 1985. Task variation and non-nativenon-native negotiation of meaning. In Gass
& Madden, pp. 14961.
1989. Incorporated repairs in NNS discourse. In M. Eisenstein (Ed.), The dynamic interlanguage (pp.
7186). New York NY: Plenum.
Goffman, E. 1976. Replies and responses. Language and Society 5, 257313.
Green, C. F. 1995. Developing discussion skills. Towards a learner-centred approach. In Wong & Green, pp.
11425
Gregg, K. 1984. Krashens monitor and Occams razor. Applied Linguistics 5, 79100.
Grice, H. P. 1975. Logic and conversation. In P. Cole & J. Morgan (Eds), Syntax and semantics 3, (pp. 4158).
New York NY: Oxford University Press.
Gumperz, J. G. 1986. Interactional sociolinguistics in the study of schooling. In J. Cook-Gumperz (Ed.), The
social construction of literacy (pp. 4568). Cambridge: Cambridge University Press.
Halliday, M. A. K. 1978. Language as a social semiotic. London: Edward Arnold.
1989. Spoken and written language. Oxford: Oxford University Press.
Harris, Z. 1951. Methods in structural linguistics. Chicago IL: University of Chicago Press.
Hatch, E. M. 1978a. Discourse analysis and second language acquisition. In Hatch (1978b), pp. 513.
(Ed.). 1978b. Second language acquisition: A book of readings. Rowley MA: Newbury House.
Hatch, E. M. & Lazaraton, A. 1991. The research manual. Design and statistics for applied linguistics. New
York NY: Newbury House.
Hawkins, B. 1985. Is an appropriate response always so appropriate? In Gass & Madden, pp. 16278.
1988. Scaffolded classroom interaction and its relation to second language acquisition for language
minority children. PhD, University of California at Los Angeles.
Ho, S. P. S. & Carmichael, S. (Eds). 1996. Exploring language. Hong Kong: Language Centre, Hong Kong
University of Science and Technology.
Hoetker, J. & Ahlbrand, W. P. 1969. The persistence of the recitation. American Educational Research
Journal 6, 1, 14567.
Hoffman, C. & Medsker, K. 1983. Instructional analysis the missing link between task analysis and
objectives. Journal of Instructional Development 6, 4, 1027.
Holliday, L. 1988. Let them talk: A study of nativenon-native interaction in conversation. Working Papers in
Educational Linguistics (University of Pennsylvania) 4, 89100.
Hong Kong Education Commission. 1995. Enhancing language proficiency: A comprehensive strategy.
Report Number 6. Hong Kong: Government Printer.
1996. Quality school education. Report Number 7. Hong Kong: Government Printer.
Hong Kong Education Department. 1990a. Guidelines on class teaching of English in secondary schools.
Hong Kong: Government Printer.
1990b. Split-class teaching in secondary classrooms. Hong Kong: Government Printer.
1996. English Section Advisory Inspectorate Report. Hong Kong: Government Printer.
Hong Kong Examinations Authority. 1992. Guidelines for candidates. Use of English oral examination. Hong
Kong: HKEA.
Hong Kong University of Science and Technology. 1993. Teaching styles. Hong Kong: Educational
Tech-nology Centre, Hong Kong University of Science & Technology (Report 4/93).
Hymes, D. H. 1972. On communicative competence. In J. B. Pride & J. Holmes (Eds.), Sociolinguistics:
Selected readings (pp. 26993). Harmondsworth: Penguin.
Jacobs, G. 1989. Miscorrection in peer feedback in writing class. RELC Journal 20, 6876.
James, G. 1994. Oral testing at tertiary level. In Boyle & Falvey, pp. 24556.
Jefferson, G. 1973. A case of precision timing in ordinary conversation. Semiotica 9, 1, 4796.

112
Johnson, D. W. 1981. Studentstudent interaction: The neglected variable in education. Educational
Researcher 10, 1, 510.
Johnson, K. 1988. Mistake correction. ELT Journal 2, 8996.
Johnson, R. K. 1979. Communicative approaches and communicative processes. In C. Brumfit & K. John-son
(Eds), The communicative approach to language teaching (pp. 192205). Oxford: Oxford Uni-versity
Press.
Johnson, R. K. & Morrow, K. (Eds). 1981. Communication in the classroom. Harlow: Longman.
Kasper, G. 1985. Repair in foreign language teaching. Studies in second language acquisition 7, 2, 20015.
1997. Beyond reference. In Kasper & Kellerman (1997b), pp. 34560.
Kasper, G. & Kellerman, E. 1997a. Introduction: Approaches to communication strategies. In Kasper &
Kellerman (1997b), pp. 113
1997b. Communication strategies: Psycholinguistic and sociolinguistic perspectives. London & New
York NY: Longman.
Kelch, K. 1985. Modified input as an aid to comprehension. Studies in Second Language Acquisition 7, 1,
20015.
Kellerman, E. & Bialystok, E. 1997. On psychological plausibility in the study of communication strategies.
In Kasper & Kellerman (1997b), pp. 3148.
Klippel, F. 1984. Keep talking: Communicative fluency activities for language teaching. Cambridge:
Cambridge University Press.
Knowles, M. 1975. Self-directed learning. Chicago IL: Follett.
Krashen, S. D. 1985. The input hypothesis. London: Longman.
Labov, W. 1972. Language in the inner city: Studies in the Black English vernacular. Philadelphia PA:
University of Pennsylvania Press.
Lai, C. 1993. Communication failure in the classroom: An exploration of causes. Hong Kong: English
Department, City University of Hong Kong (Research Report 26).
Larsen-Freeman, D. 1976a. The acquisition of grammatical morphemes by adult learners of English as a
second language. PhD, University of Michigan.
1976b. An explanation of the morpheme accuracy order of learners of English as a second language.
Language Learning 26, 1, 12535.
Lazaraton, A. 1995. Qualitative research in applied linguistics: A progress report. TESOL Quarterly 29, 3,
45573.
Leech, G. 1983. Principles of pragmatics. London: Longman.
Legutke, M. & Thomas, H. 1991. Process and experience in the language classroom. London: Longman.
Levinson, S. C. 1983. Pragmatics. Cambridge: Cambridge University Press.
Lightbown, P. 1983. Exploring relationships between developmental and instructional sequences in L2
acquisition. In Seliger & Long, pp. 21745.
1985. Input and acquisition for second language learners in and out of classrooms. Applied Lin-guistics
6, 26373.
Lightbown, P. & Spada, N. 1990. Focus on form and corrective feedback in communicative language
teach-ing: Effects on second language learning. Studies in Second Language Acquisition 12, 4, 42948.
Littlejohn, A. P. 1982. Teacherless language learning groups: An experiment. Unpublished MS, Department
of Linguistics, Lancaster University.
Littlejohn, A. & Melouk, M. (Eds). 1988. Research methods and processes. Lancaster: Department of
Lin-guistics and Modern English Language, Lancaster University.
Littlewood, W. 1992. Teaching oral communication: A methodological framework. Oxford: Blackwell.
Littlewood, W., Liu, N. F. & Yu, C. 1996. Hong Kong tertiary students attitudes and proficiency in spoken
English. RELC Journal 27, 1, 7088.
Loban, W. 1966. Language ability: Grades seven, eight and nine. Washington DC: Government Printing
Office.
Long, M. H. 1980. Input, interaction and second language acquisition. PhD, University of California at Los
Angeles.
1981. Input, interaction, and second language acquisition. In H. Winitz (Ed.), Native language and
foreign language acquisition (pp. 25978). New York NY: New York Academy of Sciences (Annals,

113
379).
1983a. Native speaker/non-native speaker conversation in the second language classroom. In M. A.
Clarke & J. Handscombe (Eds), On TESOL 82: Pacific perspectives on language learning and teaching
(pp. 20725). Washington DC: TESOL.
1983b. Does second language instruction make a difference? A review of the research. TESOL
Quarterly 17, 35982.
1983c. Native speaker/non-native speaker conversation and the negotiation of meaning. Applied
Linguistics 4, 12641.
1988a. Instructed interlanguage development. In Beebe (1988b), pp. 11541.
1989. Task, group, and task-group interactions. In S. Anivan (Ed.), Language teaching methodology for
the nineties (pp. 3150). Singapore: SEAMEO/RELC.
1992. Input, focus on form, and second language acquisition. Unpublished paper from AAAL. Personal
communication.
Long, M. H. & Crookes, G. 1992. Three approaches to task-based syllabus design. TESOL Quarterly 26,
2756.
1993. Units of analysis in syllabus design: The case for task. In Crookes & Gass, pp. 15573.
Long, M. H. & Porter, P. 1985. Group work, interlanguage talk and second language acquisition. TESOL
Quarterly 19, 2, 20728.
Long M. H. & Sato, C. J. 1984. Methodological issues in interlanguage studies: An interactionist perspective.
In A. Davies, C. Criper & A. Howatt (Eds), Interlanguage (pp. 25379). Edinburgh: Edinburgh University
Press.
Loschky, L. & Bley-Vroman, R. 1990. Creating structure-based tasks for second language development.
University of Hawaii Working Papers in ESL 9, 1, 161212.
Luckmann, T. 1990. Social communication, dialogue and conversation. In I. Markova & K. Foppa (Eds), The
dynamics of dialogue (pp. 4561). New York NY: Harvester Press.
Luk, V. 1992. Interactional listening tasks: A comparative study of strategy and practice teaching ap-proaches.
Edinburgh Working Papers in Applied Linguistics 3, 7891.
Lynch, A. J. 1995. Training lecturers for international audiences. In J. Flowerdew (Ed.), Academic listening:
Research perspectives (pp. 199218). Cambridge: Cambridge University Press.
Lynch, T. & Anderson, K. 1992. Study speaking. A course in spoken English for academic purposes.
Cambridge: Cambridge University Press.
Lyons, J. 1977. Semantics. Cambridge: Cambridge University Press.
Mackey, A., Doughty, C. & Pienemann, M. 1992. Are tasks effective in eliciting morphosyntactic features?
Unpublished data, University of Sydney. Summarised in S. Robinson & S. Ross, The development of
task-based assessment in English for Academic Purposes programs. Applied Linguistics 17, 4 (1996),
45575.
Mak, D. & Reid, A. 1992. The experience of validation at Hong Kong Polytechnic. In A. Craft (Ed.), Quality
assurance in higher education (pp. 191205). London: Falmer Press.
Matthei, E. & Roeper, T. 1983. Understanding and producing speech. New York NY: Universe.
McCleary, L. 1980. Tuning in on the monitor and listening to the critic. Unpublished MS. Berkeley CA:
University of California.
McCroskey, J. C. 1984. The communication apprehension perspective. In J. A. Daly & J. C. McCroskey (Eds),
Avoiding communication: Shyness, reticence, and communication apprehension (pp. 1338). Beverly
Hills CA: Sage.
McCroskey, J. C. & Richmond, V. P. 1990. Reliability and validity of the willingness to communicate scale.
Communication Quarterly 40, 1, 1625.
McGrath, J. E. 1984. Groups, interaction and performance. Englewood Cliffs NJ: Prentice Hall.
McGrath, J. E. & Hollingshead, A. B. 1996. Groups interacting with technology. London. Sage.
McLaughlin, B. 1987. Theories of second language acquisition. London: Edward Arnold.
McNeill, A. 1990. Vocabulary learning and teaching: Evidence from lexical errors in the spontaneous speech
of ESL learners. Institute of Language in Education Journal 7, 14153.
Medawar, P. B. 1984. Aristotle to Zoos. A philosophical dictionary of biology. London: Weidenfeld &
Nicolson.

114
Mendonca, C. O. & Johnson K. E. 1994. Peer review negotiations: Revision activities in ESL writing
instruction. TESOL Quarterly 28, 4, 74569.
Mohan, B. & Marshall Smith, S. 1992. Context and co-operation in academic tasks. In D. Nunan (Ed.),
Collaborative learning and teaching (pp. 8199). Cambridge: Cambridge University Press.
Murphy, D. F. 1993. Evaluating language learning tasks in the classroom. In Crookes & Gass, pp. 13961.
Newmark, L. 1971. A minimal language teaching program. In P. Pimsleur & T. Quinn (Eds), The psy-chology
of second language learning (pp. 12536). Cambridge: Cambridge University Press.
Ng, I. W. Y. 1995. Task type and proficiency level as variables determining the use of communication
strategies in spoken non-native English discourse. In Wong & Green, pp. 96113.
Noakes, N. & Wong, K. M. K. 1997. English language needs for the School of Engineering. Hong Kong:
Language Centre, Hong Kong University of Science and Technology.
Nobuyoshi, J. & Ellis, R. 1993. Focused communication tasks and second language acquisition. ELT Journal
47, 20310.
Nunan, D. 1987. Communicative language teaching: Making it work. ELT Journal 41, 13645.
1991. Language teaching methodology. New York NY: Prentice Hall.
Oxford, R. L. 1990. Language learning strategies: What every teacher should know. New York NY: Newbury
House.
Paribakht, T. 1985. Strategic competence and language proficiency. Applied Linguistics 6, 2, 13244.
Pattison, P. 1987. Developing communication skills. Cambridge: Cambridge University Press.
Pavesi, M. 1986. Markedness, discourse modes and relative clause formation in a formal and informal context.
Studies in Language Acquisition 8, 3855.
Pennington, M. C. 1986. A discussion of Robinson. In Second language teaching and educational tech-nology:
A state of the art symposium. Philadelphia PA: University of Pennsylvania/Defense Intelligence Agency.
1995. Eight case studies of classroom discourse in the Hong Kong secondary English class. Hong
Kong: Department of English, City University of Hong Kong (Research Report 42).
Perrow, C. 1967. A framework for the comparative analysis of organizations. American Sociological Review
32, 2, 194208.
Pica, T. 1987. Interlanguage adjustments as an outcome of NSNNS negotiated interaction. Language
Learning 37, 4, 56393.
1988. Communicative language teaching? An aid to second language acquisition? Some insights from
classroom research. English Quarterly 21, 2, 7080.
Pica, T. & Doughty, C. 1985a. The role of groupwork in classroom SLA. Studies in SLA 7, 1, 23348.
1985b. Input and interaction in the communicative language classroom: A comparison of teacher
fronted and group activities. In Gass & Madden, pp. 11532.
Pica, T., Holliday, L., Lewis, N. & Morgenthaler, L. 1989. Comprehensible output as an outcome of lin-guistic
demands on the learner. Studies in Second Language Acquisition 11, 6390.
Pica, T., Kanagy, R. & Falodun, J. 1990. Choosing and using communication tasks for second language
instruction and research. Paper presented at the TESOL Convention, San Francisco, March. In Crookes
& Gass (1993), pp. 935.
Pica, T., Lincoln-Porter, F., Paninos, D. & Linnell, J. 1993. What can second language learners learn from
each other? Only their researcher knows for sure. Unpublished data summarised in T. Pica, Questions
from the language classroom. Research perspectives. TESOL Quarterly 28, 1 (1994), 4979.
1996. Language learners interaction: How does it address the input, output and feedback needs of L2
learners? TESOL Quarterly 30, 1, 5981.
Pica, T. & Long, M. 1986. The linguistic and conversational performance of experienced and inexperienced
teachers. In Day, pp. 8598.
Pica, T., Young, R. & Doughty, C. 1987. The impact of interaction on comprehension. TESOL Quarterly 21,
73758.
Pienemann, M. 1989. Is language teachable? Psycholinguistic experiments and hypothesis. Applied
Lin-guistics 10, 1, 5279.
Porter, P. A. 1983. Variations in the conversations of adult learners of English as a function of proficiency
level of the participants. PhD, Stanford University.
1986. How learners talk to each other: Input and interaction in task-centered discussions. In Day, pp.

115
20022.
Prabhu, N. S. 1987. Second language pedagogy. Oxford: Oxford University Press.
Richards, J. C. 1980. Conversation. TESOL Quarterly 14, 41332.
1990. The language teaching matrix. Cambridge: Cambridge University Press.
Richards, J. C. & Rodgers, T. S. 1986. Approaches and methods in language teaching: A description and
analysis. New York NY: Cambridge University Press.
Richards, S. 1994. Motives and methods: Motivation, strategy choice and language use among secondary
school students learning English in Hong Kong. Hong Kong: Department of English, City University of
Hong Kong (Research Report 39).
Rivers, W. 1987. Interactive language teaching. Cambridge: Cambridge University Press.
Rixon, S. 1979. The information gap and the opinion gap. English Language Teaching 33, 2, 4350.
Robinson, P. & Ross, S. 1996. The development of task-based assessment in English for Academic Pur-poses
programs. Applied Linguistics 17, 4, 45576.
Rodgers, T. 1988. Co-operative language learning: Whats news? In B. K. Das (Ed.), Materials for lan-guage
learning and teaching (pp. 115). Singapore: SEAMEO/RELC.
Rudduck, J. 1991. Innovation and change. Milton Keynes: Open University Press.
Rulon, K. A. & McCreary, J. 1986 Negotiation of content: Teacher-fronted and small-group interaction. In
Day, pp. 18299.
Rutherford, W. & Sharwood Smith, M. 1985. Consciousness-raising and universal grammar. Applied
Lin-guistics 6, 27482
Sacks, H., Schegloff, A. & Jefferson, G. 1974. A simplest systematics for the organization of turn taking for
conversation. Language 50, 696735.
Samuda, V. & Rounds, P. L. 1988. Relating group behavior and task structure in group work. Paper presented
at the 22nd annual TESOL convention, Chicago, March.
1993. Critical episodes: Reference points for analyzing a task in action. In Crookes & Gass, pp. 12539.
Sato, C. J. 1986. Conversation and interlanguage development: Rethinking the connection. In Day, pp. 2345.
Savignon, S. 1983. Communicative competence: Theory and classroom practice, texts and contexts in second
language learning. Reading MA: Addison-Wesley.
Schachter, J. 1984. A universal input condition. In W. E. Rutherford (Ed.), Language universals and second
language acquisition (pp. 16783). Philadelphia PA: John Benjamins.
1986. Three approaches to the study of input. Language Learning 36, 2, 21125.
Scarcella, R. & Higa, C. 1981. Input negotiation and age differences in second language acquisition.
Lan-guage Learning 31, 2, 40937.
Schegloff, E. 1972. Notes on conversational practice: Formulating place. In D. Sudnow (Ed.), 1972 studies in
social interaction (pp. 12131). New York NY: Free Press.
Schegloff, E. & Sacks, H. 1973. Opening up closings. Semiotica 8, 289327.
Schegloff, E., Jefferson, G. & Sacks H. 1977. The preference for self-correction in the organization of repair
in conversation. Language 53, 2, 36182.
Schiffrin, D. 1987. Discourse markers. Cambridge: Cambridge University Press.
Schmidt, R. 1990. The role of consciousness in second language learning. Applied Linguistics 11, 21958.
1994. Deconstructing consciousness in search of useful definitions for applied linguistics. AILA Review
11, 1126.
Schmidt, R. W. & Frota, S. N. 1986 Developing basic conversational ability in a second language: A case
study of an adult learner of Portuguese. In Day, pp. 237326.
Schwartz, J. 1980. The negotiation for meaning: Repair in conversations between second-language learners of
English. In D. Larsen-Freeman (Ed.), Discourse analysis in second language acquisition research (pp.
10139). Rowley MA: Newbury House.
Schumann, J. K. 1988. Research on the acculturation model for second language acquisition. In T. J. Quinn
& T. F. McNamara (Eds), Issues in second-language learning: General and particular (pp. 8394).
Deakin University Press.
Scollon, S. 1993. Metaphors of self and communication English and Cantonese. Perspectives 5, 2, 4163.
Segal, U. A. 1982. The cyclical nature of decision making. Small Group Behaviour 13, 3, 33348.
Seliger, H. W. 1977. Does practice make perfect? A study of interaction patterns and L2 competence.

116
Lan-guage Learning 27, 26378.
Seliger, H. W. & Long, M. H. (Eds). 1983. Classroom oriented research in second language acquisition.
Rowley MA: Newbury House.
Selinker, L. 1972. Interlanguage. IRAL 10, 20931.
Shavelson, R. J. & Stern, P. 1981. Research on teachers pedagogical thoughts, judgments and behavior.
Review of International Research 51, 4, 45598.
Shaw, M. E. 1976. Group dynamics: The psychology of small group behavior. 2nd ed. New York NY:
McGraw-Hill.
Shortreed, I. 1993. Variation in foreigner talk input: The effects of task and proficiency. In Crookes & Gass,
pp. 96122.
Sinclair, J. & Coulthard, M. 1975. Towards an analysis of discourse. London: Oxford University Press.
Skehan, P. 1989. Individual differences in second-language learning. London: Edward Arnold.
1998. Task-based approaches to language testing. In E. S. L. Li & G. James (Eds), Testing and
evaluation in second language education (pp. 3449). Hong Kong: Language Centre, Hong Kong
University of Science and Technology.
Slimani, A. 1987. The teaching/learning relationship: Learning opportunities and learning outcomes: An
Algerian case study. PhD, Lancaster University.
Smith, D. R. 1970. The fallacy of the communication breakdown. Quarterly Journal of Speech 56, 4, 14.
Snow, C. & Hoefnagel-Hohle, M. 1978. Age differences in second language acquisition. In Hatch (1978b), pp.
14557.
Sperber, D. & Wilson, D. 1986. Relevance: Communication and cognition. Oxford: Oxford University Press.
Spielberger, C. D. (Ed.) 1966. Anxiety and behavior. New York NY: Academic Press.
Stern, H. H. 1983. Fundamental concepts of language teaching. Oxford: Oxford University Press.
Stevick, E. 1976. Memory, meaning and method. Rowley MA: Newbury House.
Swain, M. 1985. Communicative competence: Some roles of comprehensible input and comprehensible
output in its development. In Gass & Madden, pp. 23553.
1995. Problems in output and the cognitive processes they generate: A step towards language learning.
Applied Linguistics 16, 3, 37191.
Swales, J. M. 1990. Genre analysis. Cambridge: Cambridge University Press.
Tarone, E. (1978). Conscious communication strategies in interlanguage: A progress report. In H. D. Brown,
C. A. Yorio & R. H. Crymes (Eds), On TESOL 77: On teaching and learning ESL (pp. 194203).
Washington DC: TESOL.
1983. On the variability of interlanguage systems. Applied Linguistics 4, 14263.
1984. Teaching strategic competence in the foreign language classroom. In S. Savignon & M. Berns
(Eds), Initiatives in communicative language teaching (pp. 23457). Reading MA: Addison Wesley.
1988. Variation in interlanguage. London: Edward Arnold.
1993. Second language acquisition in a variationist framework. Unpublished MS. Minneapolis MN:
University of Minnesota.
Tarone, E. & Parrish, B. 1988. Task related variation in interlanguage: The use of articles. Language Learning
38, 2144.
Tarone, E. & Yule, G. 1989. Focus on the language learner. Oxford: Oxford University Press.
Taylor, B. 1983. Teaching ESL: Incorporating a communicative, student centred component. TESOL
Quarterly 17, 1, 6778.
Telfer, R. (Ed.). 1993. Aviation instruction and training. Cambridge: Cambridge University Press.
Telfer, R. & Biggs, J. B. 1988. The psychology of flight training. Ames IA: Iowa State University Press.
Thompson, J. D. 1967. Organizations in action. New York NY: McGraw Hill.
Todd, R. 1981. Methodology: The hidden context of situation in studies of talk. In C. Adelman (Ed.), Utter-ing,
muttering: Collecting, using and reporting talk for social and educational research (pp. 21127). London:
Grant, McIntyre.
Tong-Fredericks, C. 1984. Types of oral communication activities and the language they generate: A
com-parison. System 12, 13346.
Tsui, A. 1994. English conversation. Oxford: Oxford University Press.
Twelker, P. A. 1971. Simulation and media. In P. J. Tansey (Ed.), Educational aspects of simulation (pp.

117
13184). London: McGraw-Hill.
UCLES. 1995. Cambridge Certificate in Advanced English handbook. Cambridge: University of Cambridge
Local Examinations Syndicate.
Ur, P. 1981. Discussions that work. Task-centred fluency practice. London: Cambridge University Press.
van Lier, L. 1982. Analysing interaction in second language classrooms. PhD, Lancaster University.
1988. The classroom and the language learner. Ethnography and second-language classroom
re-search. London & New York NY: Longman.
1989. Reeling, writhing, drawling, stretching and fainting in coils: Oral proficiency interviews in
conversation. TESOL Quarterly 23, 489508.
1994. Language awareness, contingency and interaction. AILA Review 11, 6982.
1996. Interaction in the language curriculum. London: Longman.
Varonis, E. M. & Gass, S. 1985. Nonnative/nonnative conversations: A model for negotiation of meaning.
Applied Linguistics 6, 1, 7190.
Wagner-Gough, J. & Hatch, E. 1975. The importance of input data in second language studies. Language
Learning 25, 297307.
Waters, A. & Hutchinson, T. 1983. ESP at the crossroads. In J. Swales, Episodes in ESP (pp. 178203).
London: Pergamon, 1985.
Wells, G. 1985. Language development in the pre-school years. Cambridge: Cambridge University Press.
Wenden, A. & Rubin, J. (Eds). 1987. Learner strategies in language learning. London: Prentice-Hall.
White, L. 1987. Against comprehensible input. The input hypothesis and the development of second-language
competence. Applied Linguistics 8, 95110.
White, R. V. 1988. The ELT curriculum: Design, innovation and management. Oxford: Basil Blackwell.
Widdowson, H. G. 1978. Teaching language as communication. Oxford: Oxford University Press.
1983. Language purpose and language use. London: Oxford University Press.
Wilkes-Gibbs, D. 1997. Studying language use as collaboration. In Kasper & Kellerman (1997b), pp. 23874.
Wilkins, D. A. 1976. Notional syllabuses. London: Oxford University Press.
Wong, K. P. Y. & Green, C. F. (Eds). 1995. Thinking language: Issues in the study of language and lan-guage
curriculum renewal. Hong Kong: Language Centre, Hong Kong University of Science and Technology.
Yalden, J. 1987. Principles of course design for language teaching. London: Cambridge University Press.
Yeung, S. & Heyworth, R. 1990. Strategies of more competent and less competent problem-solvers in a
problem-solving task of sorting a scrambled passage. CUHK Education Journal 18, 1, 1126.
Young, D. 1986. The relationship between anxiety and foreign language oral proficiency ratings. Foreign
Language Annals 19, 43945.
Young, R. 1988. Input and interaction. Annual Review of Applied Linguistics 9, 12234.
Yule, G. & Macdonald, D. 1990. Resolving referential conflicts in L2 interaction. The effect of proficiency
and interactive role. Language Learning 40, 53956.
Yule, G. & Tarone, E. 1990. Eliciting the performance of strategic competence. In R. C. Scarcella, E. S.
Anderson & S. D. Krashen (Eds), Developing communicative competence in a second language (pp.
17994). [Rowley MA]: Newbury House.
1991. The other side of the page: Integrating the study of communication strategies and negotiated input
in SLA. In R. Phillipson, E. Kellerman, L. Selinker, M. Sharwood Smith & M. Swain (Eds),
Foreign/second language learning. A commemorative volume for Claus Frch (pp. 16271). Clevedon:
Multilingual Matters.
Zobl, H. 1983. Markedness and the projection problem. Language Learning 33, 293313.

118
Appendix: Bibliography of published sources of oral interactive tasks

Alexander, L. G. 1968. For and against. London: Longman.


Anderson, A. & Lynch, T. 1988. Listening. Oxford: Oxford University Press.
Brown, G. & Yule, G. 1983. Teaching the spoken language. An approach based on the analysis of conversational
English. London: Cambridge University Press.
Brown, G., Anderson, A., Shilcock, R. & Yule, G. 1984. Teaching talk. Cambridge: Cambridge University Press.
Byrne, D. 1967. Progressive picture composition. London: Longman.
Carver, T. K. & Fotinos S. D. 1986. A conversation book. English in everyday life. Englewood Cliffs NJ: Prentice-Hall.
Christison, M. A. & Bassano, S. 1981. Look whos talking!: A guide to the development of successful conversation
groups in intermediate and advanced ESL classrooms. San Francisco CA: Alemany Press.
Davis, P. & Rinvolucri, M. 1990. The confidence book. Colchester: Pilgrims/Longman.
Di Pietro, R. 1987. Strategic interaction. Cambridge: Cambridge University Press.
Fein, E. 1984. Speak for yourself. Rowley MA: Newbury House.
Finocchiaro, M. 1970. Lets talk. New York NY: Regents.
Folse, K. S. 1993. Talk a lot. Ann Arbor MI: University of Michigan Press.
Fried-Booth, D. 1987. Project work. Oxford: Oxford University Press.
Golebiowska, A. 1990. Getting students to talk. London: Prentice Hall.
Heaton, J. B. 1966. Composition through pictures. London: Longman.
Hong Kong Examinations Authority. 1992. Guidelines for candidates. Use of English oral examination. Hong Kong:
HKEA.
HKUST 199194. LANG 001, 002 and 204 course materials. Hong Kong: Language Centre, Hong Kong University of
Science and Technology.
James, K. 1984. Speak to learn. London & Glasgow: Collins.
Jones, L. 1981. Functions of English. Cambridge: Cambridge University Press.
Klippel, F. 1984. Keep talking: Communicative fluency activities for language teaching. Cambridge. Cam-bridge
University Press.
Knowles, P. L. & Sasaki, R. 1981. Fluency squares for Business and Technology. Englewood Cliffs NJ: Prentice Hall
Regents.
Lynch, A. J. 1983. Study listening. Cambridge: Cambridge University Press.
Lynch, T. & Anderson, K. 1992. Study speaking: A course in spoken English for academic purposes. Cambridge:
Cambridge University Press.
Mathews, A. & Read, C. 1981. Tandem. London: Evans.
Mawer, J. 1992. Business games. Hove: Language Teaching Publications.
Morgan, J., & Rinvolucri, M. 1988. The Q book. Singapore: Longman.
Nolasco, R. & Arthur, L. 1987. Conversation. A resource book for teachers. Oxford: Oxford English Series.
Nunan, D. 1989. Designing tasks for the communicative classroom. Cambridge: Cambridge University Press.
Pifer, G. & Mutoh, N. W. 1977. Points of view. Rowley MA: Newbury House.
Porter-Ladousse, G. 1983. Speaking personally. Cambridge: Cambridge University Press.
Potter, J. 1992. Use of English. Oral 1 and Oral 2. Hong Kong: Witman.
Richard-Amato, P. A. 1988. Making it happen: Interaction in the second language classroom. New York NY: Longman.
Rinvolucri, M. 1984. Grammar games. Cambridge: Cambridge University Press.
Rooks, G. 1981. The non-stop discussion workbook: Problems for intermediate-advanced students of English. Rowley
MA: Newbury House.
Sadow, S. A. 1982. Idea Bank: Creative activities for the language class. Rowley MA: Newbury House.
Smallwood, I. M. & Burns G. J. 1994. Impact. Hong Kong: Macmillan.
Spaventa, L. (Ed.). 1980. Towards the creative teaching of English. London: Allen & Unwin.
Stone, L. 1988. Task-based activities: A communicative approach to language laboratory use. Philadelphia PA: Media
Learning Center, Temple University.
1991. Task-based II: More communicative activities for the language lab. Philadelphia PA Media Learning
Center, Temple University.
UCLES. 1996. Cambridge English as a Foreign Language examination suite. Cambridge: University of Cambridge
Local Examinations Syndicate.
Ur, P. 1981. Discussions that work: Task-centred fluency practice. Cambridge: Cambridge University Press.
1988. Grammar practice activities: A practical guide for teachers. Cambridge: Cambridge University Press.
Yu, A. 1991. Incorporating communicative language lab activities in a business communication course. In Stone, pp.
6267.

119

Vous aimerez peut-être aussi