Planning And Task Performance In A Second Language Language Learning And Language Teaching Rod Ellis

Planning And Task Performance In A Second
Language Language Learning And Language Teaching
Rod Ellis download
https://ebookbell.com/product/planning-and-task-performance-in-a-
second-language-language-learning-and-language-teaching-rod-
ellis-2002982
Explore and download more ebooks at ebookbell.com

Planning and Task Performance in a Second Language
<DOCINFO AUTHOR ""TITLE "Planning and Task Performance in a Second Language"SUBJECT "Language Learning and Language Teaching, Volume 11"KEYWORDS ""SIZE HEIGHT "220"WIDTH "150"VOFFSET "4">

Language Learning and Language Teaching
The LL&LT monograph series publishes monographs as well as edited volumes
on applied and methodological issues in the ﬁeld of language pedagogy. The
focus of the series is on subjects such as classroom discourse and interaction;
language diversity in educational settings; bilingual education; language testing
and language assessment; teaching methods and teaching performance; learning
trajectories in second language acquisition; and written language learning in
educational settings.
Series editors
Nina Spada
Ontario Institute for Studies in Education, University of Toronto
Jan H. Hulstijn
Department of Second Language Acquisition, University of Amsterdam
Volume 11
Planning and Task Performance in a Second Language
Edited by Rod Ellis

Planning and Task Performance
in a Second Language
Edited by
Rod Ellis
University of Auckland
John Benjamins Publishing Company
Amsterdam/Philadelphia

The paper used in this publication meets the minimum requirements
8
TM
of American National Standard for Information Sciences – Permanence
of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data
Planning and task performance in a second language / edited by Rod Ellis.
p. cm. (Language Learning and Language Teaching, issn 1569–9471
; v. 11)
Includes bibliographical references and index.
1. Second language acquisition. 2. Second language acquisition--
Ability testing. 3. Second language acquisition--Methodology. 4.
Language and languages--Study and teaching. 5. Language planning. 6.
Competence and performance (Linguistics) I. Ellis, Rod. II. Series.
P118.2.P59 2005
418’.0071--dc22 2004066032
isbn 90 272 1961 3 (Eur.) / 1 58811 613 1 (US) (Hb; alk. paper)
isbn 90 272 1962 1 (Eur.) / 1 58811 614 X (US) (Pb; alk. paper)
© 2005 – John Benjamins B.V.
No part of this book may be reproduced in any form, by print, photoprint, microﬁlm, or
any other means, without written permission from the publisher.
John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands
John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

JB[v.20020404] Prn:2/02/2005; 14:54 F: LLLT11CO.tex / p.1 (43-104)
Table of contents
Preface vii
Section I. Introduction
Chapter 1
Planning and task-based performance: Theory and research 3
Rod Ellis
Section II. Task rehearsal
Chapter 2
Integrative planning through the use of task-repetition 37
Martin Bygate and Virginia Samuda
Section III. Strategic planning
Chapter 3
What do learners plan? Learner-driven attention to form during
pre-task planning 77
Lourdes Ortega
Chapter 4
The effects of focusing on meaning and form in strategic planning 111
Jiraporn Sangarun
Chapter 5
The effects of strategic planning on the oral narratives of learners
with low and high intermediate L2 proficiency 143
Chieko Kawauchi

JB[v.20020404] Prn:2/02/2005; 14:54 F: LLLT11CO.tex / p.2 (104-145)
 Table of contents
Section IV. Within-task planning 165
Chapter 6
The effects of careful within-task planning on oral and written
task performance 167
Rod Ellis and Fangyuan Yuan
Chapter 7
Strategic and on-line planning: The influence of surprise
information and task time on second language performance 193
Peter Skehan and Pauline Foster
Section V. Planning in language testing
Chapter 8
Planning for test performance: Does it make a difference? 219
Catherine Elder and Noriko Iwashita
Chapter 9
Strategic planning, task structure, and performance testing 239
Parvaneh Tavakoli and Peter Skehan
Section VI. Conclusion
Chapter 10
Planning as discourse activity: A sociocognitive view 277
Rob Batstone
References 297
Index 309

JB[v.20020404] Prn:15/12/2004; 13:41 F: LLLT11PR.tex / p.1 (41-107)
Preface
The purpose of this book is to bring together a series of articles on the nature
of planning and its effects on task-based performance in laboratory, classroom
and testing contexts. The idea for the book originated in a colloquium on this
topic given at AILA Conference in Singapore in December 2002. Papers given
by Bygate and Samuda, Elder and Iwashita, Ellis and Fanguan, and Sanguran
were subsequently developed into chapters for this book. A number of other
researchers (Batstone, Foster, Ortega, Kawauchi, Skehan, and Tavakoli) were
later invited to submit chapters and did so.
Planning and its role in task-based performance are of both theoretical
interest to second language acquisition (SLA) researchers and of practical sig-
nificance to language teachers. In the case of SLA researchers, planning is
important because it links in with the current interest in the role of attention
in language learning. Whether learners plan strategically before they perform a
task or engage in careful within-task planning, opportunities arise for them to
attend to language as form, or as Ortega (Chapter 3) puts it ‘form-in-meaning’.
Thus, investigating planning serves as one way of studying what learners attend
to and what effect it has on the way they use language. Further, it is also hy-
pothesized that the kind of language use that learners engage in will influence
the process of acquisition itself. Its significance for language teachers lies in the
fact that planning is a relatively straightforward way of influencing the kind of
language that learners produce. It serves, therefore, as an effective device for
intervening indirectly in interlanguage development.
The predominant methodological paradigm in planning studies is exper-
imental. That is, the task performance of learners who engage in planning of
one kind or another is compared with a task performance where there is no
opportunity for planning. This paradigm continues to be reflected in several of
the studies reported in this book (e.g. the chapters by Kawauchi, Ellis and Yuan,
and Skehan and Foster). It has proved very fruitful in demonstrating that plan-
ning does indeed affect the way in which learners perform a task. Nevertheless,
this paradigm also has its limitations. It tells us nothing about what learners ac-
tually do when they are planning; it does not show us whether learners actually

JB[v.20020404] Prn:15/12/2004; 13:41 F: LLLT11PR.tex / p.2 (107-150)
 Preface
do what they planned to do; and, more crucially, perhaps, it fails to recog-
nize that planning and task-performance constitute social as well as cognitive
activities.
Clearly, then, there is a case for broadening the paradigm to incorporate
both a process element and to acknowledge the social nature of tasks. A num-
ber of the chapters in this book address planning as a process. Ortega extends
her earlier research on tasks to examine the strategies that learners use when
engaged in pre-task planning. Sanguran (Chapter 4) discusses how the in-
structions learners are given can influence the way in which they plan. Several
authors report the results of post-task questionnaires designed to investigate
how learners responded to the opportunities to plan. Skehan and Foster (Chap-
ter 7) undertake a detailed analysis of what they call ‘breakdown fluency’ with
a view to identifying process features of task performance that will provide evi-
dence of on-line planning. All of these studies extend the research on planning
in significant ways.
There is less evidence of any attention to the social aspect of planning
and task-performance. The prevailing tenor of this book is psycholinguistic.
In the concluding chapter, however, Batstone (Chapter 10) develops a convinc-
ing argument for a social perspective. He points out that learners can approach
tasks in two different ways – as requiring economical and efficient communi-
cation or as providing opportunities for them to engage in learning activities.
The idea that tasks always have a context and that this context will help to
shape how learners plan for and perform them is further supported in the two
chapters that address the role of task planning in a testing situation (by Elder
and Iwashita [Chapter 8] and Tavakoli and Skehan [Chapter 9]). The very dif-
ferent results of these two studies are perhaps best explained in terms of the
differences in the specific testing contexts.
It is to be hoped, then, that this book both reflects mainstream research
into the role of planning in task-based performance and also extends it.
Rod Ellis
Auckland, April 2004

JB[v.20020404] Prn:16/12/2004; 14:23 F: LLLT11S1.tex / p.1 (41-114)
Section I
Introduction
The last decade has seen a growing body of research investigating various as-
pects of L2 learners’ performance of tasks (see, for example, Bygate et al. (2001)
and Ellis (2003)). This research has focused broadly on a variety of design fea-
tures of tasks and implementation procedures and how these impact on such
aspects of language use as comprehension, input processing, meaning nego-
tiation and the fluency, complexity and accuracy of L2 production (Skehan
1996, 1998a). While task-based research has been able to identify a number of
variables that impact on performance (e.g. whether contextual support is avail-
able, whether the information is shared or split, whether the outcome is closed
or open, whether there is inherent structure to the task’s content), the results
have not always been consistent. This has led some researchers (e.g. Coughlan
Duff 1994) to argue that the ‘activity’ that results from a ‘task’ is necessarily
co-constructed by the participants on each occasion, making it impossible to
predict accurately or usefully how a task will be performed.
However, one implementation variable that has attracted considerable at-
tention and that has been shown to produce relatively consistent effects on
L2 production is task planning. A number of studies (e.g. Foster Skehan
1996) have shown that when learners have the opportunity to plan a task before
they perform it, the language they produce is more fluent and more complex
than when no planning is possible. Other studies (e.g. Yuan Ellis 2003) have
shown that unpressured on-line planning also has predictable effects, albeit
somewhat different from those arising from pre-task planning.
The choice of planning as the variable for investigation in this book is mo-
tivated both by its importance for current theorizing about L2 acquisition (in
particular with regard to cognitive theories that view acquisition in terms of
information processing) and its value to language teachers, for unlike many
other constructs in SLA, ‘planning’ lends itself to pedagogical manipulation.
The study of task planning, then, provides a suitable forum for establishing the
interconnectedness of theory, research and pedagogy in SLA (Pica 1997).
This introductory chapter has a number of purposes. It seeks to provide
a framework for the subsequent chapters by identifying and defining different

 Section I
types of planning. It examines the theoretical backgrounds that have informed
the study of planning in task-based performance. It reviews earlier research
that has investigated the effects of the different types of planning. It examines
a number of key methodological issues related to the study of the effects of
planning on task performance.

JB[v.20020404] Prn:15/12/2004; 16:04 F: LLLT1101.tex / p.1 (41-107)
Chapter 1
Planning and task-based performance
Theory and research
Rod Ellis
University of Auckland
Types of planning
All spoken and written language use, even that which appears effortless and
automatic, involves planning. That is speakers and writers have to decide what
to say/write and how to say/write it. Planning is essentially a problem solving
activity; it involves deciding what linguistic devices need to be selected in order
to affect the audience in the desired way. As Clark and Clark (1977) noted,
planning takes place at a number of differentlevels, resulting in discourse plans,
sentence plans and constituent plans, all of which have to be interwoven in the
actual execution of a language act.
Principal types of task planning
Figure 1 distinguishes two principal types of task-based planning – pre-task
planning and within-task planning. These are distinguished simply in terms of
when the planning takes place – either before the task is performed or dur-
ing its performance. Pre-task planning is further divided into rehearsal and
strategic planning. Rehearsal entails providing learners with an opportunity to
perform the task before the ‘main performance’. In other words, it involves task
repetition with the first performance of the task viewed as a preparation for a
subsequent performance. Strategic planning entails learners preparing to per-
form the task by considering the content they will need to encode and how to
express this content. In pre-task planning, the learners have access to the actual
task materials. It is this that distinguishes strategic planning from other types of
pre-task activity (e.g. brainstorming content; studying a model performance of

 Rod Ellis
Pressured
Unpressured
Planning
Pre-task planning
Within-task planning
Rehearsal
Strategic planning
Figure 1. Types of task-based planning
the task; dictionary search). Within-task planning can be differentiated accord-
ing to the extent to which the task performance is pressured or unpressured.
This can be achieved most easily by manipulating the time made available to
the learners for the on-line planning of what to say/write in a task perfor-
mance. In an unpressured performance learners can engage in careful on-line
planning resulting in what Ochs (1979) has called ‘planned language use’. In
pressured performance learners will need to engage in rapid planning resulting
in what Ochs calls ‘unplanned language use’ (although, of course, all language
use involves some level of planning). Ochs documents a number of linguistic
differences between the two types of discourse. For example, unplanned dis-
course tends to manifest non-standard forms acquired early whereas planned
discourse contains more complex, target-like forms.
While pre-task planning and within-task planning constitute distinctive
types of planning they should not be seen as mutually exclusive. As shown in
Figure 2, it is possible to envisage four basic combinations of the two planning
conditions. In condition 1, learners have no opportunity for either pre-task
planning or unpressured within-task planning. Given that learners (especially
with low proficiency) have a limited processing capacity and are likely to ex-
perience difficulty in accessing and encoding their linguistic knowledge, this
can be considered the most demanding condition. In condition 2, learners are
given the opportunity to pre-plan their performance (either by means of task
rehearsal or strategic planning) but are not allowed to plan their utterances
carefully on-line. In condition 3, the reverse occurs; learners are required to
start performing the task straight away but are given as much time as they wish
to take. Both of these conditions may ease the processing burden of the learner.

Planning and task-based performance 
Planning conditions Pre-task planning Unpressured within-task
planning
1 No No
2 Yes No
3 No Yes
4 Yes Yes
Figure 2. Planning conditions
Condition 4, where the learner has the opportunity for both pre-task planning
and unpressured within-task planning, can be expected to create the conditions
that help learners maximize their competence in performance.
Sub-categories of task planning
Both pre-task and within-task planning can be categorized further in ways not
shown in Figure 1 but which are of potential theoretical and practical signif-
icance. For example, learners can be left to their own devices when planning
a task (unguided planning) or they can be given specific advice about what
and how to plan (guided planning). In this case, they can be directed to attend
to linguistic form, to meaning or to form and meaning. Chapter 4 by San-
garun, for example, explores how directing learners to focus on some specific
aspect of language in their strategic planning of tasks influences subsequent
performance. Earlier studies (e.g. Hulstijn Hulstijn 1984) have explored the
effects of directing attention to form or meaning on within-task planning and
performance. Another option relevant only to strategic planning concerns par-
ticipatory structure, i.e. whether the planning is undertaken by the learners
working individually, collaboratively in small groups, or with the teacher (see
Foster Skehan 1999). As Batstone discusses in the concluding chapter to this
volume this can potentially affect the way a task is performed.
Clearly, which types and combinations of types of planning are of rele-
vance must ultimately be decided empirically. That is, each type/option needs
to be systematically examined to establish if it has any effect on the language
produced in a task performance. As we will see when we examine the previ-
ous research on planning and task-based performance this has been one of the
major goals of enquiry to date.

 Rod Ellis
Theoretical background to the study of planning in task-based research
I will consider three theoretical frameworks that have informed the study of
task planning in second language acquisition (SLA) research. These are (1)
Tarone’s (1983) account of stylistic variation, (2) models of speech produc-
tion and writing, and (3) cognitive models of L2 performance and language
learning. These theories explicitly or implicitly draw on three central constructs
involved in psycholinguistic accounts of language processing – attention and
noticing, a limited working memory capacity, and focus-on-form – so I will
begin by briefly outlining each of these constructs, as they have been applied in
SLA research.
L2 production as information processing: Some key constructs
Information processing models constitute the dominant approach to theoriz-
ing about language comprehension and production in cognitive psychology
today. While the current models differ in some major ways (see Robinson 1995
for a review of these), they all share a number of features; they all seek to ac-
count for how information is stored and retrieved; they all view information
processing as involving input, temporary storage of material attended to, long-
term storage of (some of) this material and mechanisms for accessing this
material from long-term memory. Lantolf (1996) has referred to this general
approach as the ‘computational model’ as it is based on an analogy between
the human mind and a computer.
There are a number of general principles that inform this model (Huitt
2003). One is the assumption of a limited capacity. That is, there are limits
on the amount of information that human beings can process from input or
for output. These limits cause bottlenecks in working memory and can lead
to language users prioritizing one aspect of language over another. A second
principle is that there is a control mechanism that language users will need to
access when they are confronted with a new task for which they do not pos-
sess proceduralized linguistic knowledge. This control mechanism draws on
explicit stored knowledge. As such, it uses up processing power and thus taxes
working memory. A third principle is that human beings process information
by means of both top-down processes that draw on encyclopedic knowledge
of the world and on situational context and bottom-up processes that involve
close attention to the linguistic signals in the input. These general principles
underlie the three central constructs discussed below.

Planning and task-based performance 
1. Attention and noticing
In a number of seminal articles in the 90s, Schmidt (1990, 1994) advanced the
hypothesis that conscious attention, or what he called ‘noticing’, is essential for
language learning. He states ‘although unattended stimuli may have subtle but
undeniable effects on humans (as in sublimal perception experiments), it is
widely argued in psychology that learning without attention to what is to be
learned is impossible’ (Schmidt 1994:17). He goes on to argue that in the case
of learning attention must necessarily be conscious as ‘all demonstrations of
detection without conscious registration . . . demonstrate only the processing
of what is already known, not learning’. This is a view that has not gone un-
challenged, however. In particular, Tomlin and Villa (1994) have proposed that
three components of attention can be distinguished; alertness (a general readi-
ness to deal with incoming stimuli), orientation (the aligning of the attentional
mechanisms to some specific aspect of language) and detection (the actual
process by which a specific feature of language is attended to focally). They
claim that none of these components necessarily involves consciousness and
that even detection can occur without any conscious registration of the stim-
uli attended to. More recently, Schmidt (2001) has been less dogmatic about
whether (conscious) attention is required, writing ‘the question of whether
all learning from input requires attention to that input remains problematic,
and conceptual issues and methodological problems have combined to make
a definitive answer elusive’ (p. 29). He continues to assert, however, that in-
tentional, conscious attention is beneficial for learning as it can help learners
process features of language that otherwise would not be noticed.
Much of the discussion of noticing (as conscious attention) in language
learning has focussed on its role in input processing and, as such, might be
seen as having little relevance to theorizing about how task planning aids acqui-
sition. Task planning, whether of the pre-task or within-task type, may involve
learners attending to the linguistic input provided in the task materials (e.g.
in a text reformulation task), but in many tasks (e.g. those that involve a pic-
torial rather than verbal input) it clearly does not. Planning primarily entails
learners accessing their own implicit and explicit knowledge of the L2 for use
in production, as suggested by Ochs’ (1979) account of planned language use.
The question arises, then, as to whether noticing has any role to play in output-
processing. Swain (1985b, 1995) claims that it does. According to the Output
Hypothesis, production requires learners to process syntactically, which in-
volves bottom-up rather than top-down processing and requires attention to
form. Similalarly, Robinson (2001b) suggests that output as well as input re-
quires attention and that the degree of attention will depend on the complexity

 Rod Ellis
of the task they asked to perform, with more complex tasks requiring more at-
tention. Providing learners with the opportunity to plan a task, therefore, may
aid performance. However, as we will see later, there is some disagreement as to
how pre-task planning affects attention. One view is that it encourages greater
attention to form during task performance, resulting in increased accuracy
and complexity. An alternative view, promulgated by Robinson, is that pre-
task planning simplifies the task and thus obviates the need to attend closely to
form during performance but assists automatic access to stored language and
so leads to greater fluency.
2. Limited working memory capacity
There are number of models of working memory (see Miyake Shah 1999).
One of the most commonly cited in the task planning literature is that of Bad-
deley (e.g. Baddeley Hitch 1974; Baddeley Logie 1999). This identifies
three components of working (or short-term) memory; the central executive
or supervisory attentional system, the phonological loop, and the visual spatial
sketchpad. Two of these seem relevant to a role for task planning (i.e. not the
visual spatial sketchpad).
The central executive system governs the relationship between working
memory and long-term memory, allocating attention to specific long-term
memory systems. This system is limited in capacity, and thus the extent to
which language learners are able to attend to a specific system will depend on
the extent to which other systems are automatized. For example, if learners
use up available processing space in lexical searches the attention they can pay
to grammar will be limited. Providing learners with the opportunity for pre-
task planning or for unpressured within-task planning can ease the burden on
working memory, allowing learners the opportunity to engage in controlled
processing and to process multiple systems linearly.
The phonological loop is comprised of two sub-components – the phono-
logical store, which affords a temporary representation of material drawn from
the input or long term memory, and a mechanism that allows for articulatory
rehearsal, which enables decaying material introduced into working memory
to be sustained. Planning is likely to draw extensively on this component, al-
lowing learners to maintain one set of material while drawing on another set to
modify or refine it. For example, learners will be able to access linguistic mate-
rial from their interlanguage grammars and maintain this in the phonological
loop while they edit it through reference to their explicit knowledge of the L2.
In other words, the phonological loop is likely to play a central role in monitor-

Planning and task-based performance 
ing (discussed below). In short, planning is seen as a means of helping learners
overcome the limitations in capacity of their working memory.
3. Focus-on-form
The term ‘focus-on-form’ has been variably used in the SLA literature. It helps
to distinguish three related but different senses of the term, depending on
whether the perspective is a pedagogic one, a discoursal one or a psycholinguis-
tic one. In the context of language pedagogy, focus-on-form refers to attempts
to intervene in the process of acquisition by inducing learners to pay atten-
tion to linguistic form while they are primarily concerned with decoding or
encoding message content. These attempts can be planned (i.e. a specific form
is selected for attention) or incidental (i.e. specific forms are attended to as the
need arises). In discoursal terms, focus-on-form refers to the pre-emptive and
reactive devices that interlocutors use to draw attention to form while learners
are engaged in performing some task that gives priority to message conveyance.
These devices can consist of ‘queries’ (i.e. questions about linguistic form)
or various types of implicit and explicit corrective feedback (e.g. reformula-
tions of learners’ incorrect utterances, known as ‘recasts’, and metalinguistic
explanation). In psycholinguistic terms, ‘focus-on-form’ refers to the mental
processes involved in selective attention to linguistic form while attempting
to communicate. ‘Noticing’, discussed above, serves as a cover term for these
processes.
SLA researchers argue that L2 acquisition, especially in the case of adult
learners, requires a focus-on-form. There are two rationales for this claim. The
first relates back to the idea that learners have a limited working memory ca-
pacity and therefore experience difficulty in attending to meaning and form
at the same time (see, for example, VanPatten 1990). Because it is ‘natural’ for
learners to give priority to meaning, they may overlook certain linguistic fea-
tures, especially those that are non-salient, redundant or do not contribute to
meaning. As a result they need to be induced to attend to the formal aspects
of the language. The second, more controversial claim is that interlanguage de-
velopment can only take place if learners attend to form while they are engaged
with meaning. As Doughty and Williams (1998) put it ‘the fundamental as-
sumption of FonF instruction is that meaning and use must already be evident
to the learner at the time that attention is drawn to the linguistic apparatus
needed to get the meaning across’ (p. 4). They propose that there is a ‘cognitive
window for the provision of focus on form’ of up to 40 seconds; that is, learn-
ers are able to hold material in working memory for this length of time during
which they have the opportunity to attend to the form of what they have tem-

 Rod Ellis
porally stored. Doughty (2001) suggests that ‘roving attention’ enables learners
to pay attention to form without interruption of their original speech plan.
The theoretical and empirical bases for these proposals are reviewed in detail
in Doughty (2001).
Providing learners with the opportunity to plan a task performance con-
stitutes a means of achieving a focus-on-form pedagogically. It mitigates the
limitations of their working memory by allowing learners the ‘cognitive win-
dow’ needed to attend to form while they are primarily concerned with message
conveyance. In other words, it creates a context in which learners have the op-
portunity to map form onto meaning by accessing linguistic knowledge that is
not yet automatized.
Theoretical bases for task planning
The three constructs discussed above all figure to a greater or lesser extent in
the theories of language use/acquisition that I will now consider. The three
theories to be considered are presented chronologically, reflecting their origins
in the history of task-based research. In each case I will outline the theory and
then consider how it has been applied to task planning.
1. Tarone’s theory of stylistic variation
Tarone’s theory draws heavily on Labov’s account of stylistic variation in na-
tive speakers. Labov (1970) argued that ‘there are no single style speakers’; that
is, individual speakers manifest variation in their use of language because they
are able to draw on a variety of ‘styles’. Further, he argued that ‘these styles
can be ranged along a single dimension according to the amount of attention
that speakers pay to their speech’ (i.e. focus on form). Depending on the situa-
tion, speakers vary in the extent to which they monitor their speech. Attention
through monitoring is greatest in speech that reflects a careful style and least
in the vernacular style found in everyday speech. Labov was able to show that
what he called ‘style shifting’ was probabilistic but also systematic and therefore
predictable. That is, speakers tended to use one variant in one style and another
variant in another style to a greater or lesser extent depending on whether the
social context encouraged them to pay attention to what they said.
Drawing on this theory of intra-speaker variability, Tarone (1983) pro-
posed what she called the Capability Continuum for L2 learners. This consists
of a continuum of styles, ranging from the ‘careful’ to the ‘vernacular’, which
Tarone saw as comprising the learner’s L2 knowledge. To explain how L2 devel-
opment takes place, Tarone proposed two ways in which new forms can enter

Planning and task-based performance 
interlanguage. In one way, forms originate in the learner’s vernacular style and
then spread to the more careful styles over time. In the other way, forms appear
initially in the learners’ most careful style, manifest only when the learner is
paying close attention to speech production, and then spread to the less formal
styles where they replace earlier, more primitive forms. Subsequent empirical
work (e.g. Tarone 1985; Tarone Parrish 1988) was directed at showing how
the choice of forms was strongly influenced by the nature of the task learners
were asked to perform. However, contrary to expectations, these studies did not
always show that the more target-like forms occurred with greater frequency in
tasks designed to elicit a careful style.
Viewing learners’ L2 knowledge as a ‘capability continuum’, then, can ex-
plain how planning assists L2 production and acquisition. In the case of un-
pressured online planning, as in conditions 3 and 4 in Figure 2, learners will
be able to attend to their speech and thus access their careful style. This will
be reflected in greater accuracy (i.e. a more target-like performance). However,
the provision of opportunity for careful on-line planning may not in itself pro-
mote acquisition. In this respect, pre-task planning followed by the pressured
performance of a task (i.e condition 2 in Figure 2) may be more effective. Pre-
task planning allows learners to access their careful style but then requires them
subsequently to perform the features they have accessed in real time where
close attention to speech is not possible, thus encouraging the spread of these
features from the careful to the vernacular style.
Nevertheless, the theory lacks explanatory power. First, it does not ac-
count for why some forms are more target-like in the learner’s vernacular
style. Second, the role of attention is not clearly specified. Third, the key no-
tion of ‘spread’ is underdeveloped. The theory originated in a social account
of language variation but planning is essentially a psycholinguistic construct.
Current research on the role of planning has turned to theories that offer a
fuller psycholinguistic account of L2 production.
2. Models of speech production and writing
By far the most influential theory where studies of task planning are concerned
is Levelt’s (1989) model of speech production. Many of the later chapters (i.e.
Chapters 2, 4, 6, 7 and 9) draw on this model. The model was developed to
account for the speech production of native speakers but de Bot (1992) has
adapted it for bilingual speech production.
Levelt’s (1989) model identifies three autonomous processing stages: (1)
conceptualizing the message, (2) formulating the language representation, and
(3) articulating the message.

 Rod Ellis
The conceptualizing stage involves three sub-stages. First, the speaker de-
cides upon the communicative goal. In the second substage (macro-planning)
the speaker developsthe communicative goal into a series of sub-goals and then
identifies a speech act for each sub-goal that will achieve the intended effect.
In the third sub-stage (micro-planning), the speaker retrieves the information
needed to realize each of the subgoals and organizes it by determining ‘the
information perspective of [an] utterance, its topic, its focus, and the way in
which it would attract the addressee’s attention” (Levelt 1989:5). The product
of the micro planning is a preverbal message that is not linguistic in nature but
contains, nonetheless, all information needed to convert the preverbal message
into language. This preverbal message is then forwarded to the formulator.
Formulation involves establishing language representations of the prever-
bal messages by retrieving lexical items from the speaker’s mental lexicon. Each
lexical item is comprised of two kinds of information: ‘lemma’ and ‘lexeme’.
The lemma contains information about the meaning and syntax of each lexi-
cal item, while the lexeme contains information about its morphological and
phonological properties. Thus, retrieving a lexical item serves to prompt the
syntactic building procedure required for grammatical encoding. This results
in a ‘surface structure’ (i.e., ‘an ordered string of lemmas grouped in phrases
and subphrases of various kinds’ (Levelt 1989:11)), which is then processed by
the phonological encoder, resulting in a phonetic or articulatory plan (i.e., “an
internal representation of how the planned utterance should be articulated”
(Levelt 1989:12)). Levelt (1989) calls this ‘internal speech’.
Finally, this internal speech is transferredto the articulator. The articulator
retrieves chunks of internal speech that are temporarily stored in an articula-
tory buffer and then “unfolds and executes [them] as a series of neuromuscular
instructions” (p. 27). This leads, ultimately, to the production of overt speech.
These three stages are regulated by a self-monitoring process consisting of
three subsystems. The first subsystem inspects whether the preverbal message
matches the speaker’s original intention. It does this before the message is sent
on to the formulator to be converted into internal speech. The second subsys-
tem inspects the internal speech before it is articulated as overt speech. Finally,
the third subsystem inspects the overt speech that has been generated.
Levelt (1989) also identified two characteristics of speech production
which are relevant to task planning; (1) controlled and automatic processing
and (2) incremental production. According to Levelt, some of the compo-
nents of the speech production process (specifically, the conceptualizer and
the monitor) operate under controlled processing, while other components
(specifically, the formulator and the articulator) operate automatically in the

Planning and task-based performance 
main. In addition, he proposed that speech production processes can take place
in parallel.
De Bot (1992) considers the adaptations to Levelt’s model needed to ac-
count for speaking in an L2. He suggests that in the case of the conceptualizer,
macro-planning is not language specific but micro-planning is (i.e. the pre-
verbal message specifies which language (or languages) are to be used to encode
the message). De Bot argues that there are separate systems for the L1 and
L2 as far as the processing components of the formulator are concerned, al-
though the two systems are likely to be connected in at least some areas. In
contrast, given the cross-linguistic influences evident in L2 pronunciation, he
considers the existence of two separate systems for articulation ‘very improb-
able’ (p. 17). We might also note that whereas L1 speakers are able to carry
out the processes involved in formulation and articulation (but not concep-
tualisation) without attention, L2 learners (especially those with limited L2
proficiency) are more likely to need to activate and execute their linguistic
knowledge through controlled processing. Thus, they are likely to experience
problems during the formulation and articulation stages, as these processes are
demanding on working memory.
Levelt’s model is explicitly designed to account for speech production.
However, available theories of writing (e.g. Bereiter Scardamalia 1987; Hayes
Flower 1980; Grabe 2001; Grabe Kaplan 1996; Kellog 1996; Zimmerman
2000) posit a very similar set of processes to those proposed by Levelt. There
is also general acceptance that these processes will be broadly similar in both
L1 and L2 writing. Kellog’s (1996) model, for example, distinguishes three ba-
sic systems involved in written text production. Each system has two principal
components or processes. Formulation entails (1) ‘planning’, where the writer
establishes goals for the writing, thinks up ideas related to these goals and orga-
nizes these to facilitate action, and (2) ‘translating’, where the writer selects the
lexical units and syntactic frames needed to encode the ideas generatedthrough
planning and represents these linguistic units phonologically and graphologi-
cally in readiness for execution. Execution requires (3) ‘programming’, where
the output from translation is converted into production schema for the appro-
priate motor system involved (e.g. handwriting or typing) and (4) ‘executing’,
the actual production of sentences. Monitoring consists of (5) ‘reading’, where
the writer reads his or her own text (‘a necessary but not sufficient condition for
writing well’, p. 61) and (6) ‘editing’, which can occur both before and after exe-
cution of a sentence and can involve attending to micro aspects of the text such
as linguistic errors and/or macro aspects such as paragraph and text organiza-
tion. The extent to which a writer is able to engage in monitoring will depend

 Rod Ellis
in part on whether the writer has the time to adopt a ‘polished draft strategy’
or is engaged in pressured text production. Kellog, like Levelt for speaking,
emphasises that writers simultaneously activate formulation, execution, and
monitoring processes, although the extent to which this is achievable depends
on working memory.
Kellogg also suggests how the different components of the model relate to
working memory. He argues that the central executive, a multi-purpose system
responsible for problem-solving (see above), mental calculation and reasoning,
is involved in all the sub-processes with the exception of executing, which, he
argues, is usually accomplished without the need for controlled processing. It
should be noted, however, that this assumes an adult, native-like automaticity
in handwriting or typing, which may be lacking in some L2 learners, especially
those whose first language (L1) employs a different script. It is possible, there-
fore, that the central executive may be called upon by some L2 writers during
execution. Kellog suggests that the visuo-spatial sketchpad, which stores and
processes visual and spatial information in working memory, is only involved
in planning. Finally, he proposes that the phonological loop, which stores and
processes auditory and verbal information, is required for both translating
and reading. The key feature of Kellog’s model is that the central executive
has limited capacity, with the result that a writer may have to make decisions
about which writing process to prioritise when under pressure to produce text
rapidly. This is reflected in a trade off of attention directed at the different pro-
cesses. Formulation demands are seen as critical, taking priority over execution
and monitoring.
These models provide a basis for considering what components of lan-
guage production (spoken or written) learners focus on while planning and
also for examining what effects planning strategies have on actual production.
Rehearsal, for example, may provide an opportunity for learners to attend to
all three components in Levelt’s model – conceptualisation, formulation and
articulation – so it would seem reasonable to assume that this type of pre-task
planning will lead to all-round improvements when the task is repeated, as
found by Bygate (1996). Strategic planning can be considered likely to assist
conceptualisation in particular and thus contribute to greater message com-
plexity and also to enhanced fluency, as found by Wendel (1997). Unpressured
within-task planning may prove beneficial to formulation and also afford time
for the controlled processing required for monitoring. As a result, accuracy
might increase. In other words different types of planning can be predicted to
ease the pressure on the learner’s limited working memory in different ways,

Planning and task-based performance 
variably affecting the competition and trade-offs evident in different aspects of
language production, as claimed by Skehan and Foster (1997).
The main advantage of these models of language production, then, is that
they offer a detailed description of what is involved in speaking and writing
and thereby afford relatively precise hypotheses about the effects that planning
will have on task performance. In one respect, however, they are more lim-
ited than Tarone’s theory of stylistic variation; they do not account for how
linguistic change takes place, for, as Levelt (1989) has pointed out, they consti-
tute steady-state models. Thus, while the models can explain the relationship
between planning and language use they do not address how language use
contributes to language acquisition.
3. Cognitive models of task-based performance and learning
Skehan’s (1998b) ‘cognitive approach’ is based on a distinction between an
exemplar-based system and a rule-based system. The former is lexical in na-
ture and includes both discrete lexical items and ready-made formulaic chunks
of language. The linguistic knowledge contained in this system can be easily
and quickly accessed and thus is ideally suited for occasions calling for fluent
language performance. The rule-based system consists of abstract representa-
tions of the underlying patterns of the language. These require more processing
and thus are best suited for more controlled, less fluent language performance.
They are needed when learners have to creatively construct utterances to ex-
press meaning precisely or in sociolinguistically appropriate ways.
Skehan also distinguishes three aspects of production; (1) fluency (i.e. the
capacity of the learner to mobilize his/her system to communicate in real time,
(2) accuracy (i.e. the ability of the learner to perform in accordance with tar-
get language norms) and (3) complexity (i.e. the utilization of interlanguage
structures that are ‘cutting edge’, elaborate and structured). He suggests that
language users vary in the extent to which they emphasize fluency, accuracy
or complexity, with some tasks predisposing them to focus on fluency, others
on accuracy and yet others on complexity. These different aspects of produc-
tion draw on different systems of language. Fluency requires learners to draw
on their memory-based system, accessing and deploying ready-made chunks
of language, and, when problems arise, using communication strategies to get
by. In contrast, accuracy and, in particular, complexity are achieved by learn-
ers drawing on their rule-based system and thus require syntactic processing.
Complexity is distinguished from accuracy in that it is related to the ‘restruc-
turing’ that arises as a result of the need to take risks whereas accuracy reflects
the learner’s attempt to control existing resources and to avoid errors.

 Rod Ellis
Whereas Skehan’s research assumes that learners possess a limited process-
ing capacity such that trade-offs between fluency, accuracy and complexity
(especially these last two) are likely to occur, Robinson’s (2001c) research is
premised on a multiple-resources view of processing – that is, that learners, like
native speakers, have the capacity to attend to more than one aspect of language
at the same time. According to this view, structural complexity and functional
complexity are not in competition, as Skehan claims, but are closely connected
such that increasing the cognitive complexity of a task is hypothesized to lead
to greater linguistic complexity and accuracy as a result of increased output
modification and input incorporation.
In Robinson’s theory, task complexity is determined by two sets of fea-
tures, ‘resource directing’ (e.g. whether or not the task requires reasoning) and
‘resource depleting’ (e.g. whether or not there is opportunity for strategic plan-
ning). These two factors ‘interact and affect task production in measurable
ways’ (p. 31). Optimal attention to form arises when the task is resource di-
recting and not resource depleting, as would be in the case when a task requires
reasoning and there is no opportunity for strategic planning. Robinson argues
that such a task is likely to enhance complexity and accuracy at the expense
of fluency. In contrast a simple task that has no reasoning demands and allows
opportunity for strategic planning is likely to promote fluency but not accuracy
or complexity.
It is clear, then, that Skehan’s and Robinson’s models afford contradictory
predictions as to the effects of planning on language performance. Accord-
ing to Skehan’s model, strategic planning provides an opportunity for learners
to access their rule-based system and thus makes them less reliant on their
exemplar-based system. It may also assist them in taking the risks needed to
access ‘cutting edge’ interlanguage features rather than relying, conservatively,
on more fully acquired features. Thus, it is predicted to enhance linguistic
complexity to the detriment of accuracy. In contrast, in Robinson’s model,
strategic planning is seen as a resource-depleting factor that works hand in
hand with negative resource-directing factors to determine the overall com-
plexity of the task and the extent to which learners attend to form when they
perform the task, resulting potentially in increased fluency but decreased ac-
curacy and complexity. However, as Robinson (2001b) admits the majority
of studies of strategic planning have not supported his claim as they indicate
a positive effect on complexity and, sometimes, on accuracy (see the section
following). Neither Skehan nor Robinson consider the effects of unpressured
on-line planning but presumably this can be hypothesized to work in similar
ways to strategic planning (but see Skehan and Foster’s chapter in this book).

Planning and task-based performance 
Type of planning Message content
(Conceptualisation)
Formulation Monitoring
1. Pre-task planning Yes Yes No
2. Unpressured on-
line planning
No Yes Yes
Figure 3. Planning and task performance
What insights do these various theories provide about how planning (1)
affects task performance (spoken or written production) and (2) L2 acquisi-
tion? As shown in Figure 3, planning can impact on both the content learners
communicate when performing a task and on their choice of language. In
the case of the latter, planning is seen as important because of the role it can
play in helping learners to access their L2 knowledge through controlled pro-
cessing and, according to Skehan’s theory, in promoting selective attention to
form and monitoring. However, in accordance with the above discussion, the
two principal types of planning – pre-task planning and unpressured on-line
planning can be seen as impacting somewhat differently on these dimensions
of performance. Thus, whereas pre-task planning contributes to the concep-
tualization of message content while also assisting controlled processing and
selective attention to form, unpressured on-line planning has little impact on
message content but facilitates language choice in formulation by allowing for
controlled processing and selective attention to form and also monitoring.
While the theories are informative about how planning influences the per-
formance of tasks, they are less convincing about how it contributes to acqui-
sition. Extrapolating from performance to acquisition requires acceptance of a
number of underlying assumptions:
1. Interlanguage development occurs while learners are primarily focused on
message conveyance (i.e. performing tasks).
2. Interlanguage development is facilitated by selective attention to form.
3. Because learners have a limited working memory capacity, attention to
form requires opportunity for controlled processing.
4. As a result of the opportunity for the selective attention made possible by
controlled processing, learners are able to access more ‘advanced’ linguistic
forms during the formulation stage of production and to achieve greater
accuracy through monitoring than is possible in automatic processing.
5. One aspect of language use that fosters acquisition is the production of
language that is complex and accurate (cf. Swain’s Output Hypothesis).

 Rod Ellis
These assumptions appear inherently reasonable, but, as we will see when I
review the extant research on task planning, there is as yet very little empirical
evidence in support of them. In particular, there is a notable lack of support
for assumption 5, which is fundamental to the claim that planned language use
assists acquisition.
Previous research on task planning
In line with the preceding typology of planning types, I will review the previous
research on task planning by considering studies that have investigated pre-
task planning and unpressured on-line planning. Studies of task-planning in a
testing context will be considered separately.
Pre-task planning
1. Rehearsal
The research on rehearsal suggests that it has a beneficial effect on learners’ sub-
sequent performance of the same task but that there is no transference of the
rehearsal effect to a different task, even when this is the same type as the orig-
inal task. Bygate (1996) compared one learner’s retelling of a Tom and Jerry
cartoon on two separate occasions, three days apart. He found that rehearsal
enhanced complexity, with the learner using more lexical verbs (as opposed
to copula), more regular past tense forms (as opposed to irregular), a wider
range of vocabulary and cohesive devices (e.g. words like ‘then’, ‘so’ and ‘be-
cause’), and fewer inappropriate lexical collocations on the second occasion.
There were also more self-correcting repetitions on the second telling of the
story. Bygate (2001) reports a larger study that sought to investigate the ef-
fects of practicing specific types of task (involving narrative and interview) on
both a second performance of the same task and on performance of a new task
of the same type. The study showed that the second performance manifested
greater fluency and complexity and also that the opportunity to practice that
particular type of task helped. However, the practice did not appear to assist
performance of a new task of the same type. In other words, disappointingly,
there was no transfer of practice effect. Gass et al. (1999) report very similar
findings in a study that compared learners’ use of L2 Spanish in tasks with the
same and different contents. In this study an effect for task repetition on rat-
ings of overall proficiency, accuracy in the use of ‘estar’ (to a lesser extent) and

Planning and task-based performance 
lexical complexity (type-token ratio) was found. However, again there was no
transfer of these effects to a new task.
Lynch and McLean (2000; 2001) made use of a unique task that involved
rehearsal. In the context of an English for specific purposes course designed to
prepare members of the medical profession to give presentations in English,
they designed a ‘poster carousel’ task. This required students to read an aca-
demic article and prepare a poster presentation based on it. Students then
stood by their posters while other members of the group visited and asked
questions. Altogether, each ‘host’ had six ‘visitors’. Given that visitors tended
to ask the same questions, there was substantial opportunity for retrial. Lynch
and Mclean document how recycling output resulted in both greater accuracy
and fluency. However, they noted that different learners appeared to benefit in
different ways with level of proficiency the key factor. Thus, whereas a learner
with low proficiency appeared to benefit most in terms of accuracy and pro-
nunciation, a learner with higher proficiency used the opportunity for retrial to
improve the clarity and economy of her explanations of a complex idea. Lynch
and McLean also report considerable variation in the learners’ awareness of the
changes they were making in their production.
Task rehearsal, then, seems to have beneficial effects on learner perfor-
mance. As Bygate (1999) suggests, learners are likely to initially focus on mes-
sage content and subsequently, once message content and the basic language
needed to encode it has been established, to switch their attention to the se-
lection and monitoring of appropriate language. Bygate suggests that rehearsal
may afford learners the extra processing space they need ‘to integrate the com-
peting demands of fluency, accuracy and complexity’. Bygate and Samuda, in
Chapter 2, provide further evidence of this. However, it may not be inevitable
that learners switch attention from content to form on the second perfor-
mance. Nemeth and Kormos (2001) found that repeating an argumentative
task influenced the number of supports the participants provided for their
claims but had no effect on the frequency with which lexical expressions of
argumentation were used. Also, before any strong claims can be made for re-
hearsal it will be necessary to show that the gains evident from repeating a task
transfer to the performance of new, similar tasks.
2. Strategic planning
The role of strategic planning has attracted considerable attention from re-
searchers. An effect on all three dimensions of production – fluency, accuracy
and complexity – has been found. Each dimension will be considered sepa-

 Rod Ellis
rately. First, though, I will consider research that has investigated what learners
do when they plan strategically.
To date, only two studies have investigated what learners actually do when
they are given the opportunity to plan. Wendel (1997) interviewed his learners
immediately on completion of the tasks. They varied somewhat in what they
reported doing during the planning time but all of them said they had focussed
on sequencing the narrative events in chronological order. Only 3 reported at-
tending to grammar but even these admitted it did not help them much when
it came to telling the stories. As one learner put it: ‘I feel like I’m pushing to tell
you what’s going on in the film. I focus on story, not grammar’. Wendel con-
cluded that it is not useful for learners to try to plan the details of grammatical
usage off-line. Ortega (1999) used retrospective interviews to investigate what
learners did while they performed a narrative task. She found that they adopted
an identifiable approach in their planning (e.g. they worked on the main ideas
and organization first and then on the details), they attended to both content
and linguistic form, and they made a conscious effort to plan at the utterance
level. Ortega also reports that the learners varied considerably in the emphasis
they gave to form and content, a point that she elaborates on further in Chap-
ter 3. These two studies suggest that, when planning strategically learners tend
to prioritize content. However, Ortega’s study suggests that, not surprisingly,
they do also attend to form.
Several studies indicate that strategic planning helps to enhance fluency.
Studies by Foster (1996), Foster and Skehan (1996), Skehan and Foster (1997),
Wendel (1997), Mehnert (1998), Ortega (1999) and Yuan and Ellis (1993) all
report that giving learners the opportunity to plan results in greater fluency
(i.e. a faster speaking rate and fewer dysfluencies). Foster (1996) and Foster and
Skehan (1996) report that planners paused less frequently and spent less time
in total silence than non-planners in all three tasks they investigated. However,
the effect on fluency was stronger on the more difficult narrative and decision-
making tasks than on the easier personal task. Skehan and Foster (1997), using
similar tasks, replicated the result for total pauses. Wendel (1997) found that
the planners in his study produced more syllables per minute and showed a
lower mean length of pause in two narrative tasks. Ortega (1999) found a faster
speech rate in learners of L2 Spanish on a story-telling task when they had an
opportunity to plan strategically. Yuan and Ellis (2003) also report a clear effect
for strategic planning on fluency. Foster (2001) found that planning resulted in
learners producing a greater amount of speech whereas it led to native speakers
producing less. Interestingly, Foster reports that the percentage of learner talk
accomplished by means of lexicalised sequences did not change from the un-

Planning and task-based performance 
planned to planned condition (i.e. it remained steady at about 17%) whereas
that of the native speakers did change (from 32% in the unplanned to 25% in
the planned). Her study suggests that planning opportunities may be used dif-
ferently by learners and native speakers when the former lack an extensive store
of lexicalized chunks and thus are forced to rely more on rule-based procedures
in both planned and unplanned talk. Planning enables learners to access their
rule-based procedures more speedily but not, so it would seem, to alter the
balance of their use of formulaic and rule-based resources.
A question of obvious interest is what effect the amount of time allocated
for planning has on fluency. A reasonable assumption is that the length of plan-
ning time is positively correlated with the degree of fluency. Mehnert (1998)
set out to investigate this, allocating different groups of learners 0 minute, 1
minute, 5 minutes and 10 minutes of planning time. In general, she found that
fluency did indeed improve in relation to the length of planning time. How-
ever, the main effect for fluency was evident between the non-planners and the
planners; the differences among the three planning groups were mostly non-
significant. Thus, providing learners with longer planning time did not have a
major effect on the fluency of their speech.
In most of these studies, learners were simply given the task materials and
told to plan what they wanted to say. However, a number of studies examined
the effects of different kinds of strategic planning. Foster and Skehan (1996)
investigated the effects of more guided planning. They compared the effects of
‘undetailed’ and ‘detailed’ planning, where the learners were given metacog-
nitive advice about how to attend to syntax, lexis, content, and organization.
The results showed that, in line with the overall effect of planning on fluency,
for the narrative task the guided planners were notably more fluent than the
unguided planners, but that there was no marked difference for the personal
and decision-making tasks. This study suggests that the type of planning in-
teracts with the type of task to influence fluency. Foster and Skehan (1999),
however, found that asking learners to focus on form or meaning had no dif-
ferential effect on fluency. Much may depend on the precise instructions given
to the learners, as Sanguran (see Chapter 4) suggests. The study she conducted
did find that focussing on form, meaning or form/meaning combined had an
effect on fluency. Skehan and Foster also investigated the source of planning,
comparing the effects of (1) teacher-led planning, (2) individual learner plan-
ning and (3) group-based planning on task performance. Where fluency was
concerned, (2) proved most effective. However, as Batstone points out in Chap-
ter 10, the ineffectiveness of the group-based planning may reflect the way in
which the groups were constituted.

 Rod Ellis
In contrast to fluency, the effects of strategic planning on accuracy are quite
mixed. A number of studies reported that strategic planning led to increased
accuracy. In Ellis (1987), I found that planning that provided opportunities for
both strategic and on-line planning resulted in more accurate use of the regular
past tense. Mehnert (1998) reported a significant difference in the accuracy of
1-minute planners over non-planners. However, the 5-minute and 10-minute
planners performed at the same overall level of accuracy as the 1-minute plan-
ners. Other studies found no effect (e.g Crookes 1989; Wendel 1997). Yuan
and Ellis (2003), using a general measure of accuracy, also found that strategic
planning had no effect, a result that contrasted with that which they reported
for unpressured on-line planning (see below). A number of studies found that
strategic planning assisted accuracy only on some structures, some tasks and
in some conditions. Ortega (1999) reported mixed findings – planning led to
greater accuracy in the case of Spanish noun-modifier agreement but not in the
case of articles. Foster and Skehan (1996) reported that both undetailed and
detailed planners produced fewer errors than the non-planners on a decision-
making task, that only the undetailed planners were more accurate than the
non-planners on a personal task, while no effect for planning on accuracy was
evident on a narrative task. Skehan and Foster (1997) found that planning (un-
detailed) led to greater accuracy on the personal and narrative tasks but not on
the decision-making task. Foster and Skehan’s (1999) study of the effects of
source of planning found that accuracy was greatest when the planning was
teacher-led. However, rather surprisingly, directing learners’ attention to form
as opposed to content during planning had no effect on accuracy.
It would appear from these results that whether strategic planning has any
effect on accuracy may vary depending on a variety of factors, including the
extent to which particular learners are oriented towards accuracy, the learners’
level of proficiency, the type of task, the length of planning time available, and
the particular grammatical feature. Also, with the exception of Yuan and Ellis
(2003), these studies made no attempt to control for on-line planning. Thus,
it is possible that the different results reflect whether learners were able to or
chose to engage in monitoring while they performed the task. In terms of the
Levelt model, strategic planning can be expected to aid conceptualisation but
the impact of this may depend on the readiness of learners to shift attention to
formulation when performing the task. If they do this, then strategic planning
may lead to greater accuracy but if they do not do this no effect will be evident.
Thus, the effect of strategic planning on accuracy may be linked to the kind of
on-line planning that occurs subsequently during task performance. Clearly,
though, more research is needed to identify how planning interacts with task

Planning and task-based performance 
design variables, implementational procedures and individual learner factors.
The variable impact of pre-task planning (rehearsal) on accuracy as a result of
the learner’s orientation during performance is explored by Bygate and Samuda
in Chapter 2.
The results are clearer for complexity. As for fluency, strategic planning
has a definite, positive effect; planners produce more complex language than
non-planners. Crookes (1989) reports that 10 minutes of planning time led to
learners producing more complex sentences and a wider range of lexis. Foster
and Skehan (1996) found that detailed planners used significantly more subor-
dination than undetailed planners who, in turn, produced significantly more
subordination than the non-planners. This was broadly true for all three tasks.
Skehan and Foster (1997), however, found that the planners’ production was
more complex on only two of the tasks. On the narrative task, where plan-
ning led to greater accuracy, no effect for complexity was evident, suggesting a
trade-off between these two aspects of production. Wendel (1997) found that
his planners used more complex grammatical structures but not more lexically
rich language. Mehnert (1998) also found a positive effect but only for the 10-
minute planners - the 1-minute and 5-minute planners performed at the same
level as the non-planners. Ortega (1999) reports that mean number of words
per utterance (a complexity measure) was significantly higher in the planning
condition. Yuan and Ellis (2003) also found that strategic planning had a pos-
itive effect on complexity. With regard to the source of planning, Foster and
Skehan (1999) found that individual learner planning worked best for com-
plexity as it did for fluency. Again, in this study, whether the learners focused
their planning on form or content had no differential effect on complexity.
These studies indicate that giving learners the opportunity to plan can in-
crease the complexity of their production. They also suggest that this effect
can be enhanced if (1) learners have a reasonable length of time to plan, say
10 minutes, (2) they are given guidance in how and what to plan and (3)
they plan individually rather than in groups. It should be noted, however, that
the measures of complexity used in these studies did not distinguish between
propositional complexity (i.e. the content of the learners’ messages) and formal
complexity (i.e. the actual language used). Here too further research is needed.
What general conclusions are possible from these studies? The first is that
strategic planning has a stronger effect on fluency and complexity than accu-
racy. This suggests that when learners plan strategically they give more atten-
tion to drawing up a conceptual plan of what they want to say rather than
to formulating detailed linguistic plans. Even when asked to engage in form-
focussed planning they may not do so, preferring to use the time given them

 Rod Ellis
to sequence ideas and to work out the semantic linkages among propositions.
Alternatively, it is possible that even when learners do attend to form when
planning, they find it difficult to carry over the forms they have planned into
the performance of the task, as suggested by Bygate and Samuda in Chapter
2. The second conclusion is that trade-off effects are evident. When learners
plan they have to choose what aspect of production to focus on; focussing on
fluency and complexity is at the expense of accuracy and vice-versa. Finally,
there is some evidence to suggest that strategic planning has a greater effect on
production in general when the task is cognitively demanding. If a task is easy
learners are able to perform it fluently using accurate and complex language
without the need for planning.
Unpressured on-line planning
Giving learners time to plan on-line and to monitor their output appears to
have a clear impact on accuracy. Hulstijn and Hulstijn (1984) asked learners
of L2 Dutch to perform short oral narratives under four conditions involv-
ing combinations of two variables; time (i.e. the learners were told to speak as
quickly as they could or to take as much time as they wanted) and focal atten-
tion (i.e. learners were instructed to focus on form or on meaning). They found
that time pressure by itself did not affect the accuracy of word order but that in
combination with a focus on form it had a profound effect. This study, then,
suggests that when learners use the time at their disposal to attend to formu-
lation and to monitor the use of their grammatical resources their production
becomes more accurate. However, if they use the time to plan content no effect
on accuracy is observed.
In Ellis (1987), I compared learners’ performance on written and oral nar-
rative tasks based on pictures. In the case of the written task (task 1) the learners
were given as much time as they wanted to write the narrative. In the first oral
task (task 2) they were asked to retell the same narrative but without recourse
to their written versions. In the second oral task (task 3) they were given a dif-
ferent set of pictures and instructed to tell the story with minimal opportunity
for prior-planning. Figure 4 summarizes the kinds of planning opportunities
afforded by these three tasks. I found that the learners’ use of the regular past
tense forms (but not the irregular past tense or copula past tense forms) was
most accurate in task 1 and least accurate in task 3, with task 2 intermedi-
ate. The difference between task 1 and 2 can be explained in terms of on-line
planning; accuracy was greater when there was no time pressure. However, as

Planning and task-based performance 
Task On-line planning/monitoring Strategic planning
1 Yes Yes (Probably)
2 No Yes
3 No No
Figure 4. Types of planning opportunities in Ellis (1987)
Crookes (1989) and others have pointed out, tasks 1 and 2 also differed with
regard to medium.
Building on Ellis’ study, Yuan and Ellis (2003) set out to compare the ef-
fects of pre-task and on-line planning on learner performance of a narrative
task in a more systematic way. In the pre-task planning condition learners were
given 10 minutes to prepare the task and then performed it under time pres-
sure. In the on-line planning condition, the learners were given no chance to
prepare but were allowed to perform the task in their own time. There was also
a control group that had no preparation time and was required to perform the
task under time pressure. The results indicated that opportunities for unpres-
sured on-line planning assisted both accuracy and complexity but, as might be
expected, inhibited fluency.
These three studies suggest that the time learners are given for on-line plan-
ning improves the accuracy of their production. However, the effects may only
be evident when learners are drawing on their rule-based system. In both Hul-
stijn and Hulstijn (1984) and Ellis (1987) the effects of time pressure were only
evident on grammatical structures that are clearly rule-based (i.e. Dutch word
order rules and English regular past tense); they were not evident in structures
that are more lexical in nature (i.e. irregular and copula past tense forms).
Planning in a language testing context
The study of the effects of planning on the performance of tasks in a testing sit-
uation is of considerable importance given that testers in general are concerned
to elicit the ‘best performance’ from a testee (see McNamara 1996). If planning
time can affect aspects of a test-taker’s performance then arguably it ought to
be considered when designing the test.
Three research studies have investigated the effects of pre-task planning
in a testing situation. Wigglesworth (1997) examined the performances of 107
adult ESL learners performing five tasks that were part of the Australian As-
sessment of Communicative Skills (Access) test. The candidates performed the
tasks in a planned and unplanned condition. The performances were rated

 Rod Ellis
by two trained raters using an analytic rating scale to measure fluency, gram-
mar (or in one task vocabulary) and intelligibility. The performances of 28
candidates, who were divided into high and low proficiency groups, were tran-
scribed and analyzed using measures of complexity, fluency and accuracy.
Wigglesworth reported no significant differences in the rating scores for the
planned and unplanned conditions but significant differences in the analytic
discourse measures for complexity, fluency and accuracy, especially in the high
proficiency candidates and especially in tasks with a high cognitive load. She
concludes that at least for some learners and in some tasks planning time can
help to improve the performance of test-takers but that this effect is not evident
in external ratings.
In a second study, Wigglesworth (2001) sought to further investigate one
of the findings of the previous study, namely that the effects of planning time
were not evident in the scores obtained from raters. The study examined the
effect of a number of test task variables, one of which was planning, on adult
ESL learners’ performance on five tasks that were routinely used to evaluate
achievement in the Australian Adult Migrant Education Program. In this study
an effect for planning was found on the test-takers’ ratings but the effect was
not as great as might have been expected. Planning proved to have a detrimen-
tal effect on tasks that were familiar to the candidates and on both structured
and unstructured tasks. Wigglesworth notes that these results are inconsistent
with the findings of task planning research in non-testing situations and sug-
gests that this may reflect the fact her study used external ratings rather than
discourse analytic measures. However, Iwashita, Elder and Mcnamara (2001)
used both analytic discourse measures and ratings to examine the effects of
three minutes of planning time on the task performance of 201 ESL students
and failed to find evidence of any effects on either the discourse measures or
the rating scores. Elder and Iwashita reproduce this finding in Chapter 8 and
examine a number of possible explanations.
It is possible, then, that the testing context constrains the beneficial effects
of planning. This suggests, more generally, that the ‘psychological context’ of
a task constitutes an important dimension that needs to be taken into account
in planning studies (see Batstone’s discussion of this possibility in Chapter 10).
The main conclusion to be drawn from these studies, however, is that there is a
need for further research into the effects of planning in a test situation. It seems
clear, however, that whatever effect planning time has on task performance it
may not be reliably measured by an external rating. This is problematic where
assessment is concerned, as it is not practical to calculate discourse analytic
measures in testing situations.

Planning and task-based performance 
Final comments
This review of the research suggests that the effects of planning in a testing
context may be somewhat different from those reported for laboratory or class-
room contexts. One reason may be that learners feel pressured in a testing
context with the result that their on-line planning is hurried. To date no studies
have examined whether there are any differences in on-line planning in testing
and non-testing contexts. This is a fairly obvious next step.
The results of the research certainly suggest that pre-task and unpressured
on-line planning may be somewhat different. Whereas opportunities for on-
line planning result in more accurate and complex language use, probably
because learners have the chance to monitor linguistic form, opportunities for
pre-task planning generally favour fluency and complexity, possibly because it
leads to an emphasis on conceptualizing what has to be communicated rather
than how to say it.
As I noted in the concluding comments to the previous section, researchers
have focussed their attention on investigating how different types of planning
(in combination with different types of tasks) impact on learner production.
They have not attempted to show how or even whether the planning of tasks
assists language acquisition. Thus any claims regarding planning and acquisi-
tion can only be theoretically based. Clearly, the absence of empirical support
for the key assumptions listed at the end of the previous section constitutes a
major lacuna in the research to date.
Methodological issues
The task planning research to date raises a number of methodological issues.
Perhaps the key one concerns how acquisition as opposed to language pro-
duction can be investigated. Other issues are how to ensure that learners carry
out the type of planning specified in the research design and how to measure
learners’ actual production when they perform the task. These issues will be
considered below.
Investigating the effects of planning on acquisition
The term ‘acquisition’ assumes that there is some change in the learner’s
L2 knowledge representation. Evidence for change can be found in (1) the
learner’s use of some previously unused linguistic forms, (2) an increase in

 Rod Ellis
the accuracy of some linguistic forms that the learner can already use, (3) the
use of some previously used linguistic forms to perform some new linguistic
functions or in new linguistic contexts and (4) an increase in fluency (i.e. in
the speed with which linguistic forms are used in communication).
The usual method for obtaining these kinds of evidence of change is the
standard experimental design involving an experimental group that completes
a pre-test, a treatment and post-tests (immediate and delayed) and a control
group which receives the tests without the treatment. In the case of task plan-
ning research, the treatment consists of the opportunity to plan and perform a
task. Such a design, as we have already seen is rarely employed. To the best of my
knowledge, the only studies that have made use of such a design are Bygate’s
(2001) and Gass et al.’s (1999) studies of task rehearsal. Bygate’s study asked
learners in the experimental groups to perform two tasks prior to the treat-
ment (which in turn consisted of three opportunities to repeat tasks similar to
one of the pre-treatment tasks) and the same two tasks following the treatment
together with two new tasks of the same type. In this way, Bygate was able to
assess to what extent the treatment resulted in changes in the way the learners
(1) performed the same task they had completed before the treatment and (2)
a similar task to the pre-treatment task. Such a design is promising as it does
allow the researcher to pinpoint changes that occur as a result of the treatment.
It contrasts with the standard design used in task planning research (see, for
example, Foster Skehan 1996; Yuan Ellis 2003), which typically involves
an experimental and control group performing the same task under different
planning conditions (e.g. strategic planning as opposed to no planning). Such
a design cannot address acquisition.
There is, however, a major limitation to the kind of design that Bygate
employed. It does not provide data that can easily speak to the effects of task
planning on the acquisition of specific linguistic features. That is, it can only
provide evidence of general linguistic change, as in types (2) and (4), but not
of specific linguistic changes, as in types (1) and (3). To obtain evidence of
the effects of task planning on specific linguistic features it is necessary to tar-
get specific features for study. This cannot be readily achieved by means of the
kinds of unfocused tasks that have figured in task planning research to date.
However, it may be achievable through the use of focused tasks. Whereas unfo-
cused tasks allow learners to choose from a range of forms focused tasks aim to
induce learners to use specific forms. In Skehan’s (1998b) terms they are ‘struc-
ture trapping’ in that they make the employment of the specific forms, natural,
useful or, ideally, essential (Loschky Bley Vroman 1993). The advantage of
such tasks is that they allow researchers to construct pre- and post-tests to mea-

Planning and task-based performance 
sure whether learners knew the targeted forms prior to performing the task and
what the effect of planning tasks is on learning. The only planning studies that
have investigated specific linguistic forms to date are Ellis (1987), which tar-
geted past tense forms, and Hulstijn and Hulstijn (1984), which targeted word
order rules in Dutch. Somewhat disappointingly, more recent studies have been
based on unfocused tasks.
Investigating learners’ planning strategies
In a typical task planning study, learners are asked to carry out planning in
accordance with instructions. Below, as an example, is the description of the
‘guided planning – content focus’ condition in Foster and Skehan (1999):
The students were introduced to the idea of a balloon debate. The teacher then
presented ideas that each character might use to defend his or her right to stay
in the balloon and students were encouraged to add ideas of their own.
Here is a description of the unpressured on-line planning condition in Yuan
and Ellis (2003):
The on-line planners were required to tell the story by producing at least four
sentences for each of the six pictures after seeing the pictures for only 0.5
seconds. They were given unlimited time to enable them to formulate and
monitor their speech plans as they performed the task.
Such instructions raise a number of methodological issues. The most obvious
one, given the evidence that pre-task and on-line planning have been hypoth-
esized to have somewhat different effects on learners’ performance of a task,
is the need to ensure that learners receive instructions relating to both kinds
of planning. In the case of studies investigating pre-task planning this has not
usually occurred. That is, the learners are given instructions relating to how to
conduct strategic planning/rehearsal but are left to perform the actual task in
any way they choose. Thus, it is possible that the learners interpret the task per-
formance conditions very differently, with some engaging in unpressured and
others in real-time on-line planning. This may be one explanation why stud-
ies of pre-task planning have produced such mixed results for accuracy (see
previous section).
There is also an obvious methodological need to establish whether learn-
ers actually carry out the planning instructions they were given. That is, do
they conform to the prescribed planning conditions? Again, few studies have
attempted to establish this. However, more recently, a number of researchers
have attempted to describe the different strategies learners actually use during

 Rod Ellis
the pre-task planning phase of a study. All three studies in the section dealing
with pre-task planning in this book (Section 3) do this. The data used for such
an investigation includes the actual notes that learners make while planning
strategically (see Ellis Yuan 2004) and post-task interviews with individual
learners (Ortega 1999: Chapter 3 in this book). Such research is important not
just to ensure that learners plan as intended but also because it can serve as a ba-
sis for drawing up guidelines for the design of effective planning instructions.
Sanguran, in Chapter 4, makes a useful advance in this direction by formu-
lating an explicit set of assumptions that guided her in the preparation of the
planning instructions she used in her own study.
Somewhat different kinds of evidence are needed to demonstrate what
kind of planning – pressured or unpressured – learners engage in on-line.
While it may be possible to establish this through post-task interviews (al-
though learners may have difficulty remembering their on-line decisions even
if stimulated recall techniques are used), clearer evidence may be forthcoming
by inspecting the fluency properties of the texts learners produce as a result of
performing the task. Yuan and Ellis (2003) considered two such properties –
the number of syllables produced per minute and the number of pruned sylla-
bles per minute (i.e. after dysfluencies had been discounted). They were able
to show that learners in the unpressured on-line planning condition spoke
significantly more slowly than learners in the pressured on-line planning con-
dition. In this way, they were able to demonstrate that the unpressured on-line
planners had performed as required.
Measuring learner production
Learner production can be measured either by means of external ratings or
by means of discourse analytic measures. In general, language testers have
preferred the former and SLA researchers the latter.
External ratings are based on scales that specify (1) the specific competency
being measured and (2) levels of performance for each competency (often re-
ferred to as ‘bands’). In the case of ratings of task-based performance, the target
competency can be specified either in behavioural terms that reflect the degree
to which the learners have successfully completed the task (see, for example,
Norris, Brown Hudson 2000) or in linguistic terms. In the case of the latter,
learners’ linguistic competency can be described either holistically (e.g. for the
highest ‘band’ the descriptor might be ‘speaking proficiency equivalent to that
of an educated native speaker’) or an analytic measure, where different dimen-
sions of performance (for example, fluency, complexity and accuracy) are rated

Planning and task-based performance 
separately. In Ellis (2003:298–302) I summarise the various options relating to
external ratings.
In the case of discourse analytic measures, two types of measures are
possible – measures of specific linguistic features and measures of general di-
mensions of oral and written discourse. There are a variety of well-established
methods for deriving measures of specific linguistic features (e.g. error analysis,
obligatory occasion analysis, frequency analysis and form-function analysis –
see Ellis and Barkhuizen (2004) for a detailed account of these methods as they
have been used in SLA). In the main, however, researchers have not used these,
preferring instead general measures of learner production.
These general measures have been based on Skehan’s model of L2 pro-
ficiency, which distinguishes two basic dimensions – meaning (fluency) and
form with the latter further sub-divided into complexity and accuracy. Skehan
(see Skehan Foster 1997; Tavokoli and Skehan’s study in Chapter 9 in this
book) has been at pains to establish the independence of these dimensions by
factor analysing scores obtained from a battery of measures. While the anal-
yses do not always produce entirely similar results (e.g. in Skehan Foster
1997 the analysis resulted in three distinct factors easily identifiable as fluency,
complexity and accuracy while in Tavokoli and Skehan the analysis produced a
somewhat different set of factors – temporal aspects of fluency, repair fluency
and complexity/accuracy combined) they do broadly confirm Skehan’s model.
Thus, the general measures employed by Skehan and his co-researchers, have
an established theoretical base.
There are nevertheless a wide range of measures of fluency, complexity and
accuracy to choose from (see Figure 5 for a summary of the various measures
employed in the studies reported in the subsequent chapters in this book). In
one respect this is useful as, arguably, multiple measures of each dimension
may yield a more valid assessment than single measures. The downside is that
when researchers differ in their choice of measures it becomes difficult to com-
pare results across studies. Ideally, work is needed to establish measures that
provide the most valid assessment of each dimension (using, for example, a
factor analytic approach such as that employed by Skehan), which can then be
employed across studies. It is also worth noting that it may prove necessary
to develop separate measures for spoken and written production, most obvi-
ously for fluency. Most of the measures used to date have been developed for
oral production, as this has been the focus of the bulk of the planning studies.
However, Ellis and Yuan (2004) developed measures of written production and
Wolfe-Quintero et al. (1998) offer a comprehensive list of measures of all three
dimensions for writing.

 Rod Ellis
Type of
measure
Specific
measure
Description Study
1. Fluency Production rate The number of syllables pro-
duced per minute of speech/
writing
Ellis and Yuan;
Kawauchi; Elder and
Iwashita; Sanguran
Breakdown flu-
ency
The ratio between number of
words reformulated and total
words produced
Ellis and Yuan
Number of repetitions Kawauchi; Elder and
Iwashita
Total silence
Number of pauses greater than
1 second
Number of filled pauses
Length of run
Skehan and Foster;
Tavakoli and Skehan
2. Complexity Syntactic com-
plexity
Ratio of clauses to some general
unit (e.g. T-units, c-units or AS-
units)
Ellis and Yuan;
Kawauchi; Elder and
Iwashita; Sanguran;
Skehan and Foster;
Tavakoli and Skehan
Length of unit (e.g. T-unit) Kawauchi
Number of subordinate clauses Kawauchi
Complex
grammatical
structures
Use of comparatives and condi-
tionals
Sanguran
Syntactic vari-
ety
Total number of different gram-
matical verb forms used in the
task
Ellis and Yuan
Lexical variety Mean segmental type/token ra-
tio
Ellis and Yuan
3. Accuracy Overall
grammatical
accuracy
Error-free clauses Ellis and Yuan; Elder
and Iwashita; Skehan
and Foster; Tavakoli
and Skehan
Error-free clauses of different
lengths
Skehan and Foster
Number of errors per 100 words Sanguran
System-based
grammatical
accuracy
Correct verb forms
Past-tense markers
Ellis and Yuan
Kawauchi
Figure 5. Discourse analytic measures used in the studies reported in this book

Planning and task-based performance 
A final question concerns the length of the learner texts to which the mea-
sures are applied. In many cases, researchers do not use the full texts produced
by learners but instead elect to use only part of the texts, typically the first five
or ten minutes. The problem here, as Skehan and Foster’s chapter in this book
indicates, is that planning may have a markedly different effect on the first few
minutes of production in comparison with later. Learners may have difficulty
sustaining careful formulation and monitoring over a lengthier period of time.
Skehan and Foster’s study raises the awkward possibility that the findings of
the research to date, which have typically been based on relatively short learner
productions may not be generalizable to extended discourse.
Conclusion
Task planning has proven a rich vein for empirical study, as attested by the
large number of studies that have investigated this implementational variable
(larger than have investigated any other task variable) and by the current col-
lection of studies. Why has task planning proven such a fruitful arena for SLA
research? Is it just another fad in SLA, like the error evaluation studies in the
70s and 80s, that will soon lose its attraction? I think not. First, the study of
task planning, as I have tried to show in this chapter has a strong theoretical
basis drawing on a set of constructs (controlled processing, limited capacity
memory, focus-on-form) and a number of well-established theories of L2 use
and acquisition. Research, such as that reported in the subsequent chapters of
this book, can both draw on this theory and help to test it. In a sense, then,
the study of task planning lies at the very centre of current research in SLA.
Second, the research is of obvious pedagogical relevance. Planning, whether of
the pre-task or within-task kind, is a variable that teachers can easily manip-
ulate in their day-to-day teaching. While teachers should not look to research
as the only determinant of lesson design they can certainly benefit from the in-
sights and ‘provisional specifications’ (Stenhouse 1975) that the task planning
research offers them. Thus, for both theoretical and practical reasons I expect
task planning to continue to attract attention in the years ahead.
This book constitutes an advance on the research to date. It addresses a
variety of issues, some previously examined, others new:
– the role of task rehearsal in helping learners to elaborate content and to
integrate the different strands of their L2 proficiency;
– the actual strategies learners employ during pre-task planning;

 Rod Ellis
– the way in which learners orientate to the opportunity to engage in strate-
gic planning;
– the extent to which learners’ attention to form and meaning can be manip-
ulated through pre-task planning;
– the effect of different types of planning (pre-task vs. on-line; detailed vs.
undetailed);
– the interaction between strategic planning (a task implementation vari-
able) and task design features (such as the introduction of a surprise
element into a task);
– the effects of learners’ L2 proficiency on their ability to make use of the
opportunity for pre-task planning;
– the relative effects of unpressured on-line planning on oral and written
production in an L2;
– the extent to which learners are able to sustain the effects of planning on
performance over an extended period of time;
– the effect of context (e.g. a language test) on task performance subsequent
to planning;
The range and variety of these issues testify to the richness of task planning as
an area of SLA enquiry.

Section II
Task rehearsal
The chapter in this section examines the effects on task-performance of having
learners repeat a task – of what was called ‘rehearsal’ in Chapter 1. Bygate and
Samuda’s paper is important both methodologically, theoretically and peda-
gogically.
As noted in Chapter 1, the bulk of the research that has investigated the
effects of planning on task performance has examined learner productions in
terms of fluency, accuracy and complexity. There is an obvious need to ex-
tend analysis to the macro properties of learner discourse. Bygate and Samuda
show that one way of doing this is by examining what they call ‘framing’. This
is a cover term for a heterogeneous collection of linguistic resources used by
speakers to convey ‘perspective’ (e.g. the speaker’s attitude to what is being
communicated) and to ‘preview’ (e.g. by providing an advance organizer of
what is to come). In effect, framing fleshes out the bare factual bones of a
discourse. The analysis of learner narratives they present in terms of framing
demonstrates that this constitutes a significant addition to the tools in cur-
rent use. Bygate and Samuda’s analysis also points to the value of combining
group-based statistical analysis with a qualitative, case study approach.
Their chapter is important theoretically because it provides a thoughtful
account of how different kinds of planning (strategic planning, on-line plan-
ning and rehearsal) contribute to task performance. Bygate and Samuda argue
that rehearsal offers the learners certain processing opportunities not available
in the other types of planning, in particular the ability to integrate their linguis-
tic resources. Repeating a task enables learners to reorganise and consolidate
information into a richer, discoursally more sophisticated performance.
Finally, Bygate and Samuda suggest that rehearsal is a useful pedagogic
procedure not just because of the opportunities it affords learners to develop
their L2 discourse skills but also because rehearsal arises in naturally occurring
communicative activities (i.e. it has situational authenticity). The challenge
facing teachers is to introduce task repetition in ways that students will find
motivating.

Chapter 2
Integrative planning through the use
of task-repetition
Martin Bygate and Virginia Samuda
Lancaster University
Introduction
This paper addresses an intriguing language teaching and learning puzzle: how
to lead students to integrate prior knowledge into performance. Associated
with this is the question of how best to help them to identify new knowledge
needed for their development.
It is generally accepted that learning involves restructuring (McLaughlin
1990; Skehan 1998b). The term relates to a number of distinct aspects of the
learning process, characterised in Ellis 1990 as noticing, comparing and inte-
grating. Noticing of new elements in the input will often signal a change in
the perceptual processes; noticing will bring with it an interpretation of the
new input, and a change in the interpretations of existing knowledge. How-
ever, material that has been noticed (whether through explicit instruction, or
through other cognitive/perceptual processes), although it is in principle avail-
able to the learners, may not in fact be drawn on. That is, a common learning
and teaching problem is to get learners to integrate knowledge that is available
to them into their active language use.
Indeed, one problem with communicative teaching is that this integration
can fail to take place: as long as the learners are able to produce language which
will achieve their communicative purposes, they may not do the additional
work needed to extend their active repertoire. (Brumfit’s 1984 separation of
accuracy work and fluency work also seemed not to attend to this issue). Some
aspects of this problem have been addressed in recent work on pedagogic tasks
in L2 learning (see for instance Samuda 2001). In this paper we explore the
possibility that doing a communication task a second time can help learners to

 Martin Bygate and Virginia Samuda
achieve integration of what they already know into what they do. The princi-
ple underlying this exploration is that repetition of a task enables two different
experiences of the same task demands. The differences between the two expe-
riences are seen as being due to different states of knowledge on the part of
the speaker, and capable of enabling change. The first encounter with a task is
likely to be the more creative encounter: the learner has to respond to a new de-
mand. This is likely to mean that the learner has rather a lot of new work to do:
for instance deciding how to do the task, what messages to produce, and how
to produce them. In comparison, on repeating a task, the learner has valuable
experience to draw on: after all, s/he has already internalised the information
content, organised it into communication units, found relevant language to
convey the meanings, and pronounced it. Hence on the second occasion the
learner is likely to be under less pressure than on the first encounter, provided
of course that the task is performed under the same conditions (with no addi-
tional time pressure, for example). Because of this, it is likely that at the first
encounter the learner is more likely to rely on the most automated aspects of
his/her language, than at the second. In contrast, at the second encounter, the
learner is not only cognitively prepared, but furthermore, her/his vocabulary
and grammar (especially vocabulary) are ‘primed’, so that there is a chance
that on the second occasion the learner will generate more sophisticated out-
put. This might involve such things as providing more backgrounding, and
selecting a wider range of ways of formulating the message. In other words, an
initial encounter with a task can be seen as creating a holistic representation of
the task, along with the experience of handling it in real time. This represen-
tation and the accompanying experience can be stored, creating a kind of plan
which can be drawn on on a second occasion, enabling the learner to integrate
a broader range of their resources into their performance - that is, to perform
more adequately their competence as it were (Clark 1974).
Types of planning
In the literature, two types of planning, strategic planning, and on-line planning
are the ones widely identified (see Chapter 1). In this chapter, we are making
a case for seeing task repetition as a form of planning. We argue that it is the
experience of processing the task as a whole together with certain elements of
both pre-task and on-line planning that is important. In this paper we call this
integrative planning.

Integrative planning through the use of task-repetition 
Strategic planning
Strategic planning typically involves focussed or unfocussed instructions to
students to plan their performance on an upcoming task, for 2–10 minutes
(for example Foster Skehan 1996; Mehnert 1998). Results strongly support
the theory that planning does affect performance: amount and type of strate-
gic planning has effects on performance, notably on fluency and complexity,
(Crookes 1989; Foster Skehan 1996; Ortega 1999) but also on accuracy (Ellis
1987); and different effects have been shown to occur with different types of
task. The underlying theory has not been elaborated in detail, but the belief is
that strategic planning reduces the processing load of subsequent on-line per-
formance: speakers may have mentally organised the content; and/or worked
on the formulation of aspects of the communication. This preparation is held
in memory and enables learners to produce more complex messages, both in
content and in form, to produce them more fluently and to be more accurate.
There are however some limitations in concentrating on this type of plan-
ning. Firstly, it isn’t at all clear how this kind of planning can affect learners
over an extended period of learning (see Skehan Foster, this volume). We
see the impact on a specific performance, but the connection between the re-
ported effects for a specific task and longer term learning is yet to be theorised
and researched. Secondly, although sometimes strategic planning does natu-
rally occur before certain speech events, and is very frequent before writing,
strategic planning seems untypical of many oral activities. Hence, while it has
clear potential as a pedagogic device, it is not a target condition for normal
everyday speech production. Students will usually need to be able to perform
adequately without strategic planning, and not depend on having this facility.
The third and perhaps most important problem is that although it is not clear
how far ahead speakers can plan, in many speech contexts, the amount of dis-
course that can be pre-planned is bound to be limited. A speaker may be able to
plan the rough content – and some expressions – for the first two or three min-
utes of talk, but it is unlikely that they would be able to map out much further
ahead in any detail, mainly because of working memory limitations. In other
words, the construct of strategic planning is unclear in terms of its functioning.
Another issue concerns the focus of strategic planning. So far, research
results have generally shown that strategic planning influences fluency and
complexity more than accuracy. Although the reasons for this are not clear,
a probable explanation is that in strategic planning speakers are more likely to
focus on the substance than the expression of their talk, leading to reduced fo-
cus on accuracy. That is, prior to a task, macro-planning is on the whole more

 Martin Bygate and Virginia Samuda
manageable and more productive than detailed micro-planning of utterances.
Useful language is harder to predict and harder to keep in mind than a content
plan. Hence speakers are more likely to use strategic planning time to ‘boot
up’ reserves of ideas, but less likely to check whether they have all the language
needed to express them. Given this information ‘charge’, speakers would then
tend to get into more informationally complex talk than they would without
planning, accounting for the increases in complexity. This would be consistent
with the findings in Yuan Ellis 2003 that their strategic planning group pro-
duced greater lexical variety, but less accurate grammar (Yuan Ellis 2003:23):
lexical variety may well be pushed by the pre-loading of information content.
Regarding the impact of strategic planning on fluency, there are also dis-
tinct pressures which are likely to lead to increased fluency. One is that the
pre-task marshalling of ideas is likely to reduce the incidence of on-line hes-
itation in finding them; and the second is that some speakers at least will be
motivated to speak faster in order to avoid loss of planned material from work-
ing memory. If this surmise is correct, content planning would tend to lead
to greater fluency at a likely cost to accuracy. Following this reasoning, both
the pressure towards complexity, and the pressure towards fluency would each
tend to derive from the same source – attention to planned content. By the
same token both pressures could often be expected to lead towards an increase
in errors (it is perhaps worth recalling that typical advice to elementary second
language writers is to keep the message simple to avoid errors). In addition,
we anticipate that the increased focus on content and fluency may well be at
the expense of the speaker’s capacity to explore their grammatical range and
monitor for accuracy.
All this is not to say that strategic planning makes it impossible for speak-
ers to activate relatively unused (or less automated) language. But the scope
for doing this, and especially of remembering it at the point of need in the
appropriate utterance, is likely to be limited. Incidentally, this view suggests a
potential separation between lexical and grammatical processing, which can be
related to the two kinds of planning. That is, within a strategic planning condi-
tion, speakers will be more ready to attend to vocabulary than grammar, in line
with VanPatten’s (1996) view that the listener privileges the lexical rather than
grammatical elements of speech (an insight also developed for both production
and comprehension by George 1972).
Hence although it is clear that strategic planning can be a valuable ped-
agogical resource, the research results suggest that the procedure may have
introduced a bias into the processing of speech, which could be detrimental to
the focus on form. Strategic planning is likely to bias towards macro-planning

Integrative planning through the use of task-repetition 
and away from grammatical work. This bias does not work in favour of the
integration of available resources into students’ performance. That is, if our
interpretation is correct, then strategic planning would not simply increase on-
task capacity, as intended in the research paradigm – it may also bias how that
capacity will be used. So, in summary, although strategic planning may help
access general ‘declarative’ knowledge prior to performance (which is clearly
desirable in terms of language development), some of the knowledge that is ac-
cessed may turn out to be either irrelevant or else forgotten when the speaker is
engaged in producing specific utterances. To the extent that strategic planning
helps to activate the learners’ knowledge structures prior to talk, memory lim-
itations may constrain the extent to which these are actually engaged during
the talk itself. Rather, attention may focus on information content, which, if
accessed, may bias against attention to form.
On-line planning
The second type of planning which has been researched has been referred to as
‘on-line planning’ (Yuan Ellis 2003). This is defined as the kind of planning
which occurs during performance. It consists mainly of processes of message
conceptualisation, lexico-grammatical searches, and monitoring, all at the level
of particular utterances – that is, at the micro- rather than the macro-level. As
has been shown elsewhere in this volume (see Chapter 6), this dimension is
operationalised through allowing students more on-line performance time, on
the assumption that without time pressure, they will engage in more covert
planning activities than students performing under time pressure.
In contrast to strategic planning, this type of planning is likely to tax work-
ing memory less, since it occurs during the planning and production of specific
utterances. Hence it may be more open than strategic planning to the range
of different types of operation which speakers might need to engage in during
speech production. As Levelt’s 1989 model (see Chapter 1) suggests, these oper-
ations involve speakers not only in creating plans, but also in monitoring them
prior to production. Hence if on-line planning time is used, speakers are likely
to be better able to attend to the conceptualisation, the formulation and even
the articulation of their messages. So that whereas in the context of strategic
planning a speaker is unlikely to be able to produce or to recall many detailed
plans (for instance at lexico-grammatical or articulatory levels), in contrast,
on-line planning time may help speakers do precisely this. It may also give
them space to self-correct after production.

Other documents randomly have
different content

avec le comte de Maistre. C'est ainsi que je connus accidentellement
Martainville, homme provoquant et intrépide. J'avais eu occasion de
le voir un an avant dans un duel où il avait été héroïque; il ne me
connaissait que de visage; il ne savait pas mon nom, quoique j'eusse
pris parti pour lui dans sa querelle.
Il craignait en ce moment d'être assassiné par les nombreux
ennemis que lui suscitaient ses invectives mordantes contre les
adversaires des Bourbons. Il me fallut insister longtemps, donner le
nom du comte de Maistre, être reconnu comme par des sentinelles à
travers des guichets pratiqués dans des couloirs, pour parvenir avec
mon dépôt jusqu'à lui.
Une fois cette glace rompue, je trouvai dans Martainville un brave
et jovial combattant de l'épée et de la plume, qui adorait dans le
comte de Maistre un étranger de la même religion politique que lui.
Chateaubriand, Bonald, Lamennais (intolérant au nom du Ciel et
absolutiste au nom des hommes alors), étaient à Paris, à cette
époque, avec Martainville, les correspondants et les patrons de ce
grand écrivain, dont on veut faire aujourd'hui, à Turin et à Paris, un
agitateur de l'Italie, précurseur de M. de Cavour, et, qui sait? peut-
être un destructeur du pouvoir temporel des papes. Ô pauvre
imagination humaine! tu ne vas jamais si loin que la bouffonnerie
des partis! Si les ombres rient dans l'éternité, l'âme beaucoup trop
rieuse de celui qui fut ici-bas le comte de Maistre doit bien rire en
voyant son nom servir d'autorité à une révolution.
Mais maintenant que nous avons le portrait de cet homme devenu
l'entretien du monde, voyons en peu de mots sa vie, et mêlons-y ses
œuvres; car l'homme, la vie et l'œuvre se tiennent indissolublement
dans le philosophe, dans le politique et dans l'écrivain.
Nous avons une excellente abréviation de la vie du comte de
Maistre écrite par son fils. C'est le fils qui connaît le mieux le père; la
piété filiale est le génie d'un biographe. Nous ne jugerions pas les
œuvres du père sur les paroles du fils, mais, quant aux circonstances

de la vie domestique, il n'y a pas de plus sûrs et de plus honnêtes
témoins que les enfants.
Nous faisons toutefois nos réserves sur deux ou trois actes de la
vie publique du comte de Maistre, actes que nous caractériserons
tout autrement que ne les caractérise son fils. Si la piété filiale a son
culte, elle a aussi son fanatisme; nous nous en défendrons: c'est le
droit de la postérité.
XI
Le comte Joseph de Maistre était né à Chambéry en 1754. Son
père, président de ce qu'on appelait le sénat de Savoie, eut dix
enfants. Joseph de Maistre était le premier-né. Élevé à Chambéry et
à Turin, sa naissance le prédestinait à la magistrature provinciale
dans son pays. D'abord substitut, puis sénateur (c'est-à-dire juge) à
Chambéry, il y épousa mademoiselle de Morand, fille d'une condition
égale à la sienne.
Trois enfants qui vivent encore, portés tous les trois à de hautes
fortunes en France par la renommée paternelle dans l'aristocratie
européenne, furent le fruit de ce mariage. Ces fortunes attestent la
vigueur des opinions aristocratiques et religieuses, solidaires depuis
Chambéry jusqu'à Paris et à Pétersbourg. Les opinions ennoblissent,
les orthodoxies deviennent parentés entre les petites et les grandes
noblesses. Une des filles du modeste gentilhomme de Chambéry se
nomme la duchesse de Montmorency en France.
M. de Maistre exerçait honorablement ses fonctions de
magistrature provinciale dans sa petite ville au moment où la
Révolution française éclata. Son fils prétend qu'il était libéral; peut-
être?
En 1793, après l'invasion de la Savoie par M. de Montesquiou, le
comte de Maistre se retira à Turin avec ses frères, qui servaient dans

l'armée sarde. Revenu peu de jours après à Chambéry, il y vit naître,
dans les angoisses de l'invasion française, sa troisième fille,
Constance de Maistre, qu'il ne devait pas revoir avant vingt-cinq ans.
Il laissa sa femme à Chambéry, pour y préserver leur petite fortune,
et il émigra à Lausanne. Ses biens paternels, très-modiques, furent
séquestrés, mais il portait avec lui une meilleure fortune; ce fut à
Lausanne qu'il écrivit, comme un pamphlet de guerre contre la
Révolution française, l'ouvrage qui commença sa réputation parmi
les émigrés de toute date dont la Suisse, l'Allemagne et l'Angleterre
se remplissaient alors. C'était une captivité de Babylone pour toutes
les aristocraties de l'Europe, un peuple dans un peuple, qui avait ses
doctrines, ses passions, sa langue à part.
M. de Maistre parla dès les premiers jours cette langue de
l'émigration avec une habileté magistrale, une vigueur et une
originalité qui créèrent son nom. Ses Considérations sur la France
éclatèrent de Lausanne à Turin, à Rome, à Londres, à Vienne, à
Coblentz, à Pétersbourg, comme un cri d'Isaïe au peuple de Dieu. Le
style de Bossuet était retrouvé au fond de la Suisse. Le début seul
annonce un philosophe dans le publiciste. Quelle théorie de la
monarchie!
«Nous sommes tous attachés au trône de l'Être suprême par une
chaîne souple qui nous retient sans nous asservir.
«Ce qu'il y a de plus admirable dans l'ordre universel des choses,
c'est l'action libre des êtres libres sous la main divine. Librement
esclaves, ils agissent tout à la fois volontairement et fatalement. Ils
font réellement ce qu'ils veulent, mais sans déranger les plans
généraux. Chacun de ces êtres occupe le centre d'une sphère
d'activité dont le diamètre varie au gré de l'éternel Géomètre qui sait
étendre, restreindre ou diriger sans contraindre la nature.
«Dans les ouvrages de l'homme, tout est pauvre comme l'ouvrier;
les vues sont bornées, les moyens roides, les ressorts inflexibles, les
résultats monotones. Dans les ouvrages de Dieu, les richesses de

l'infini se montrent à découvert jusque dans le moindre élément. Sa
puissance opère en se jouant; entre ses mains tout est souple, rien
ne lui résiste; pour lui tout est moyen, même l'obstacle, et les
irrégularités produites par l'opération des êtres libres viennent se
ranger dans l'ordre général.»
Cela continue ainsi pendant plusieurs pages, pages plus
semblables à une ode d'Orphée célébrant la Divinité dans ses lois
qu'à un pamphlet de publiciste dépaysé contre la révolution qui
l'exile. Les pages de l'Histoire universelle de Bossuet n'ont pas plus
de cette moelle de grand sens dans les choses. C'est un Bossuet
laïque.
XII
À l'instant le monde de l'émigration et des cours fut attentif et
saisi; tout le monde lettré se dit: «Écoutons! Voilà un prophète de
consolation qui nous vient des montagnes.»
Il continue, il console ses coexilés par une magnifique théorie de
l'irrésistible puissance de la Révolution qui broie tout devant elle, ses
amis comme ses ennemis. Il y voit un de ces fléaux divins auxquels
il est presque impie de résister, tant ils sont divins dans leur force.
C'est une pierre qui roule d'en haut; sa loi est d'écraser ce qui
l'arrête. Il disait plus vrai qu'il ne croyait dire. La Révolution avait
une mission qu'elle ignorait elle-même; mais cette mission n'était
pas tant de renverser le passé que de courir vers un avenir nouveau
de la pensée et des choses. C'était une marée équinoxiale de l'océan
humain; de Maistre n'y voyait qu'un accès de fureur et de crime.
Fureur et crime y prévalurent, en effet, trop inhumainement de 1791
à 1794; la Révolution en a été punie par la stérilité. La fureur et le
crime ne sèment pas, ils ravagent; mais, une fois le sang-froid
revenu à l'esprit révolutionnaire, il reprenait un grand sens humain
que le philosophe du passé ne pouvait ni ne voulait comprendre.

«La Révolution, ajoute-t-il, mène les hommes plus que les
hommes ne la mènent.» Quelle admirable intuition! et quelle preuve
plus sensible qu'elle est menée elle-même par une force occulte vers
un but inaperçu encore par ses amis et par ses ennemis!
«Les révolutionnaires, dit-il, réussissent en tout contre nous parce
qu'ils sont les instruments d'une force qui en sait plus qu'eux.»
Quelle était donc cette force omnisciente? pouvait-on répondre au
publiciste. Si ce n'était pas la fatalité, que vous répudiez avec raison
comme un blasphème, c'était donc un dessein supérieur à
l'intelligence humaine; une force supérieure à l'intelligence humaine,
qu'est-ce autre chose que Dieu?
«Votre Mirabeau, ajoute-t-il, n'est au fond que le roi des halles. Il
a prétendu en mourant qu'il allait refaire, avec ses débris, la
monarchie, et, quand il a voulu seulement s'emparer du ministère, il
en a été écarté par ses rivaux comme un enfant.»
Cela était vrai de Mirabeau vicieux, factieux et populaire; mais
combien faux de Mirabeau philosophe, orateur et législateur, quand il
avait dépouillé ses vices avec son habit de tribun! Il était alors le
prophète inspiré de la vraie Révolution, comme le comte de Maistre
était le prophète inspiré de la contre-révolution. Aussi, ce qu'il y a à
admirer dans ce premier ouvrage de Joseph de Maistre, ce ne sont
pas les vérités, ce sont les vues. Du haut de ses rochers il a le
regard de l'aigle; il voit plus loin que le vulgaire, mais il ne voit pas
toujours vrai. Il commence sa vie par un magnifique sophisme,
comme Jean-Jacques Rousseau, son compatriote. Le sophisme de de
Maistre devait aboutir à la servitude, mensonge à la dignité morale
de l'homme, comme le sophisme de liberté de Jean-Jacques
Rousseau devait aboutir à l'anarchie, mensonge de la société
politique.
Ce fut un malheur pour Joseph de Maistre d'avoir commencé sa
course au milieu de l'émigration et sur son terrain; il ne voulut plus
revenir sur ses pas. Il mourut le plus honnête et le plus éloquent des

hommes de parti, au lieu de vivre et de mourir le plus honnête et le
plus éloquent des philosophes chrétiens. La vérité pure ne lui plaisait
pas assez; il lui fallait le sel de l'exagération pour l'assaisonner au
goût de sa caste. Inde labes!
XIII
Le livre, à partir de là, devient foudroyant contre les
révolutionnaires quels qu'ils soient, savants, lettrés, modérés,
régicides, justement enveloppés, s'écrie-t-il, dans le nuage de la
vengeance céleste contre ceux qui attentent à la souveraineté. C'est
un dithyrambe à la Némésis révolutionnaire, la hache excusée de
tout pourvu qu'elle frappe! «Il y a eu, dit-il, des nations condamnées
à mort, comme des individus coupables, et nous savons pourquoi.»
Tout à coup il se tourne inopinément contre les royalistes qui
demandent la contre-révolution, la conquête de la France, sa
division, son anéantissement politique. Il fulmine contre cette idée à
son tour. «Si la Providence efface, c'est pour écrire,» dit-il. Il veut
que la réaction de la France contre la France vienne d'elle-même, de
la France; et en cela il se montre à la hauteur des pensées d'en
haut. Il finit par une prophétie qui n'était que de la logique en
comptant sur la versatilité des peuples et surtout des Gaulois, en
annonçant la restauration des Bourbons sur le trône. Seulement, s'il
était prophète pour l'événement, il n'était pas prophète pour le
temps; car ce qu'il annonçait pour demain est arrivé à vingt-cinq ans
de distance, et, avant de restaurer les Bourbons, la France a relevé
un trône militaire et absolu pour un des généraux qui l'aidèrent à
vaincre l'Europe.
Tel est le livre, nul comme prophétie, violent comme philosophie,
désordonné comme politique (relisez le chapitre sur la glorieuse
fatalité et sur la vertu divine de la guerre; cela est pensé par un
esprit exterminateur et écrit avec du sang). Mais ce livre est un

éclair de foudre parti des montagnes des Alpes pour illuminer d'un
jour nouveau et sinistre tout l'horizon contre-révolutionnaire de
l'Europe encore dans la stupeur. Ni Vergniaud, ni Mirabeau lui-même
n'avaient eu de pareils éclairs dans la parole ni de pareilles vigueurs
dans l'esprit. M. de Maistre regardait le premier face à face
l'écroulement du monde religieux et politique avec le sang-froid d'un
esprit partial, sans doute, mais surhumain. Le style, nouveau aussi
par sa sculpture lapidaire, était à la hauteur de l'esprit. Ce style bref,
nerveux, lucide, nu de phrases, robuste de membres, ne se
ressentait en rien de la mollesse du dix-huitième siècle, ni de la
déclamation des derniers livres français; il était né et trempé au
souffle des Alpes; il était vierge, il était jeune, il était âpre et
sauvage; il n'avait point de respect humain, il sentait la solitude, il
improvisait le fond et la forme du même jet; il était, pour tout dire
en un mot, une nouveauté. La nouveauté, c'est le symptôme des
gloires futures. Cet homme était nouveau parmi les enfants du
siècle.
XIV
Ce fut le sentiment de l'Europe en le lisant. Un vengeur nous est
né! s'écrièrent l'ancien régime, l'ancienne politique, l'ancienne
aristocratie, l'ancienne foi. Mais ce vengeur rajeunissait par la
jeunesse de son style la vieillesse des choses.
Ce livre, répandu comme un secret parmi l'émigration, fit du
gentilhomme savoyard le favori sérieux de la contre-révolution, des
camps et des cours. On dit au roi de Sardaigne: «Comment négligez-
vous ce prodige que Dieu vous envoie pour vous illustrer et pour
vous sauver? Les grandes puissances seraient jalouses de ce don du
Ciel. Hâtez-vous d'en décorer vos conseils.» On l'appela, en 1797, à
Turin. La faible monarchie sarde fut écrasée dans les guerres de
1799 entre la France et l'Autriche. Le roi de Sardaigne se réfugia
dans son île, sur un débris de trône. Le comte de Maistre, qui n'avait

rien à espérer de l'Autriche que l'abandon et de la France que la
proscription, suivit le roi en Sardaigne. On lui donna, sous le titre de
régent de la chancellerie, la direction très-insignifiante des tribunaux
de cette petite île.
Bientôt l'homme parut trop grand pour l'emploi. Cet écrivain qui
embrassait le monde d'un regard ne pouvait se résigner à l'étroitesse
d'horizon d'une petite cour insulaire sur un écueil de la Méditerranée,
peuplé d'habitants presque sauvages. Il fatiguait la cour et les
ministres des secousses de son imagination. Son génie oratoire et
inquiet froissait la routine et la médiocrité de la cour de Cagliari. On
le voit clairement dans sa correspondance, il importunait les Sardes
et les Piémontais favoris de la cour. Ne pouvant nier son mérite, on
l'envoya pérorer ailleurs. Lui-même étouffait dans cette bourgade
décorée du nom de capitale. La Sardaigne anéantie et ruinée ne
pouvait avoir une diplomatie sérieuse en Europe; un peu d'intrigue
et quelques supplications aux grandes cours étaient sa seule
politique. Le roi, évidemment importuné lui-même des imaginations
trop grandioses du comte de Maistre, le nomma son ministre
plénipotentiaire à Pétersbourg.
C'était un honneur dans la forme, au fond c'était un exil. Son fils
présente comme un sacrifice douloureux à la monarchie l'acceptation
du comte de Maistre de ce poste; on peut croire cependant que
l'ambition très-haute du comte de Maistre fut heureuse de cette
mission à une telle cour. Il lui fallait les grandes scènes, les grands
auditoires; il avait besoin d'espace comme tout ce qui veut rayonner
de loin. Les appointements (vingt mille francs), conformes à la
pénurie de cette pauvre cour de Cagliari, étaient insuffisants sans
doute, mais ils étaient cependant bien au-dessus du traitement d'un
sénateur de Chambéry.
XV

Le comte arriva à Pétersbourg plein de pensées vagues pour son
roi, pour la Russie, pour lui-même. Sa tête fermentait de
restauration; il voulait relever la maison de Savoie par les Russes,
peut-être même par les Français. On va voir bientôt dans sa
correspondance qu'il savait au besoin s'accommoder avec la
Révolution pourvu qu'elle rétablît et qu'elle agrandît le trône de son
monarque.
L'empereur Alexandre et l'aristocratie russe l'accueillirent, non
pour son titre, mais pour son nom. Les Considérations sur la France
avaient popularisé ce nom jusqu'à la cour de Russie. Il devint en peu
de temps le favori des salons de Pétersbourg. Il y était gracieux,
enjoué, souple, éloquent, étrange et sérieux à la fois. Son éloquence
à chaînons rompus et à brillantes fusées de génie était surtout,
comme celle de madame de Staël, une éloquence confidentielle de
coin du feu; il n'avait pas assez de gravité et de solidité pour une
tribune, il avait assez d'inspiration, de grâce et de décousu pour un
tête-à-tête. De plus, son rôle à Pétersbourg était de plaire et de
flatter. Les Savoyards naissent courtisans par la situation subalterne
de leur province à Turin. Le grand Savoyard plaisait généralement et
flattait à merveille. Les ministres étrangers, même les ministres de
France en Russie, ne voyaient en lui qu'un représentant du malheur
et du détrônement. On ne craignait pas l'ascendant de Cagliari sur le
monde; on admirait l'esprit de son représentant. Son existence, un
peu amère sous le rapport de la fortune, était très-douce sous le
rapport de la société. De plus, quoi qu'il en dise çà et là dans ses
lettres à sa cour et dans ses lettres familières, il était loin d'être
insensible aux rangs, aux titres, aux décorations, aux faveurs de
cour. Le titre d'ambassadeur d'un roi à la cour de Russie, bien que ce
roi ne fût plus qu'un naufragé du trône sur un îlot d'Italie, caressait
agréablement son orgueil. Je l'ai assez vu pour ne pas croire à ce
désintéressement d'amour-propre. Cet amour-propre n'enlevait rien
à sa vertu, mais il transpirait souvent dans sa correspondance.
J'en eus un jour une preuve bizarre qui ne s'effacera jamais de
mon souvenir. Les petites circonstances sont quelquefois les

meilleures révélations du caractère.
À l'époque de mon mariage, qui fut célébré à Chambéry, le comte
Joseph de Maistre fut choisi par mon père absent pour le représenter
au contrat et pour me servir ce jour-là de père. Le contrat se signait
dans une maison de plaisance nommée Caramagne, à quelque
distance de la ville, chez la marquise de la Pierre, centre de la
société aristocratique de Savoie. Le comte d'Andezenne, général
piémontais, gouverneur de Savoie, servait de père à ma fiancée. Une
nombreuse réunion de parents et d'amis remplissait le salon. On lut
le contrat, et on appela les témoins à la signature. Le gouverneur de
la Savoie fut appelé le premier par sa qualité de père de la fiancée et
par son rang de représentant du souverain dans la province. Il signa
et chercha à passer la plume à la main du comte de Maistre.
Le comte, que nous venions de voir dans le salon, tout couvert de
son habit de cour et de ses décorations diplomatiques, avait disparu.
On le chercha en vain dans le château et dans les jardins; nul ne
savait par où il s'était éclipsé. On fut obligé de laisser en blanc la
place de sa signature; mais, une fois le contrat signé, il reparut,
sortant d'un massif de charmille où il s'était dérobé pendant la
cérémonie. Nous lui demandâmes confidentiellement la raison de
cette disparition, qui avait contristé un moment la scène.
«C'est, dit-il, qu'en qualité d'ambassadeur du roi et de ministre
d'État je ne voulais pas inscrire mon nom au-dessous du nom d'un
gouverneur de Savoie. Demain j'irai signer seul et à la place qui
convient à ma dignité.» Et il alla, en effet, le lendemain signer le
registre. Les uns admirèrent cette grandeur de respect pour soi-
même, les autres cette politesse. Quant à moi, j'admirai cette force
du naturel qui place l'étiquette plus haut que le cœur.
XVI

Sa correspondance avec sa famille et ses amis, à dater de son
arrivée à Pétersbourg, ne laisse rien dans l'ombre de son âme et de
son esprit, de sa vie publique et de sa vie domestique. Le comte de
Maistre, qui était autant homme de conversation qu'homme de
plume, était par conséquent un correspondant exquis, car les lettres
ne sont au fond que la conversation écrite. Ces deux volumes de
correspondance, tantôt intime comme les soupirs d'un exilé vers sa
patrie, sa femme, ses enfants, ses frères, tantôt politique, sont une
des meilleures parties de ses œuvres. Elles ont été complétées
récemment par la publication indiscrète de ses dépêches à la cour de
Sardaigne. L'homme se trahit quelquefois dans ces trois volumes. On
a dit qu'il n'y avait point de grand homme pour son valet de
chambre; on peut dire, après avoir lu ces innombrables lettres, qu'il
n'y a point de secret pour la postérité. Le comte de Maistre s'y met à
nu tout entier à son insu, et, bien que l'homme y soit toujours
brillant et charmant dans sa nature, il disparaît souvent sous le
diplomate de peu de scrupule. L'adorateur inflexible de l'ancien
régime n'y disparaît pas moins sous l'adorateur de la victoire
révolutionnaire, quand la victoire révolutionnaire donne une chance
à la fortune de son parti. Il est toujours honnête homme, sans
doute, mais il n'est rien moins que l'homme d'une seule pièce qu'on
a voulu nous faire de lui. Il sait très-bien se retourner quand la roue
tourne. Il sait très-bien aussi donner à la fortune le nom majestueux
et divin de Providence. Quand la Providence tourne la page du livre
du destin, lui aussi il tourne la page, comme un traducteur obéissant
du texte sacré. Il continue à prophétiser, sans se troubler des
contradictions qu'une si haute prétention de confident et de
commentateur de la Providence fait encourir à son don de prévision.
Dangereux métier que celui d'augure! Malgré sa piété très-sincère, il
y a une certaine impiété à se mettre au niveau de l'Infini et à parler
sans cesse au nom de Dieu. Il avait trop lu la Bible; le ton d'oracle
avait vicié en lui l'accent modeste de ce grain de poussière pensant
qu'on appelle un homme de génie.
Nous en trouvons une preuve étonnante dès les premières pages
de sa correspondance. Il vient de fulminer, ainsi qu'on l'a vu, contre

la Révolution, ses œuvres, ses hommes. La légitimité est son
principe, l'ancien régime est son dogme; les Bourbons, solidaires,
selon lui, de la maison de Savoie, sont ses dieux terrestres; il a un
culte pour leurs malheurs, il a une correspondance avec leur chef
Louis XVIII. Il croit et il espère en eux comme dans la Providence
des trônes et des peuples; il est l'ami de leurs représentants ou de
leurs favoris, le comte d'Avaray et le comte de Blacas. Une pensée
contraire à la restauration du principe de la légitimité serait une
trahison de sa religion politique, une apostasie de son cœur.
Tout à coup Bonaparte s'assied sur un trône de victoires; les
puissances européennes le reconnaissent, l'usurpation se fait
dynastie, l'avenir paraît s'aplanir et s'étendre sans limites devant la
fortune d'un soldat heureux. Les royalistes sont consternés. Écoutez
M. de Maistre dans ses lettres à Madame de Pont, émigrée
désespérée à Vienne.
«Tout le monde sait qu'il y a des révolutions heureuses et des
usurpations auxquelles il plaît à la Providence d'apposer le sceau de
la légitimité par une longue possession. Qui peut douter qu'en
Angleterre Guillaume d'Orange ne fut un très-coupable usurpateur?
et qui peut douter cependant que Georges III, son successeur, ne
soit un très-légitime souverain?» (Quelle doctrine que celle en vertu
de laquelle l'usurpation de la veille est la légitimité du lendemain!
Quelle morale que celle où le temps transforme le crime en vertu!)
Il continue:
«Si la maison de Bourbon est décidément proscrite, il est bon que
le gouvernement se consolide en France. J'aime bien mieux
Bonaparte roi que simple conquérant. Cela tue la Révolution
française, puisque le plus puissant souverain de l'Europe (Bonaparte)
aura autant d'intérêt à étouffer cet esprit révolutionnaire qu'il en
avait besoin pour parvenir à son but. Le titre légitime, même
seulement en apparence, en impose à un certain point à celui qui le
porte. N'avez-vous pas observé, Madame, que dans la noblesse, qui

n'est, par parenthèse, qu'un prolongement de la souveraineté, il y a
des familles usées au pied de la lettre? La même chose peut arriver
dans une famille royale. Il n'y a certainement qu'un usurpateur de
génie qui ait la main assez ferme et même assez dure pour rétablir...
Laissez faire Napoléon... Ou la maison de Bourbon est usée et
condamnée par un de ces jugements de la Providence dont il est
impossible de se rendre raison, et, dans ce cas, il est bon qu'une
race nouvelle commence une succession légitime, etc.»
On voit avec quelle souplesse de logique le fidèle de l'ancien
régime se convertit aux volontés de la Providence et les justifie
même contre son propre dogme. «Il n'y a, écrit-il quelques lignes
plus bas, qu'une bonne politique comme une bonne physique: c'est
la politique expérimentale!» Quelle amnistie à toutes les infidélités!
XVII
À quelques jours de là on trouve dans une lettre à son frère ces
délicieuses mélancolies du regret des temps passés:
«Moi qui mettais jadis des bottes pour aller à Sonaz (château près
de Chambéry), si je trouvais du temps, de l'argent et des
compagnons, je me sens tout prêt à faire une course à Tobolsk,
voire au Kamtschatka. Peu à peu je me suis mis à mépriser la terre;
elle n'a que neuf mille lieues de tour.—Fi donc! c'est une orange.
Quelquefois, dans mes moments de solitude, que je multiplie autant
qu'il est possible, je jette ma tête sur le dossier de mon fauteuil, et
là, seul au milieu de mes quatre murs, loin de tout ce qui m'est cher,
en face d'un avenir sombre et impénétrable, je me rappelle ces
temps où, dans une petite ville de ta connaissance (Chambéry), la
tête appuyée sur un autre dossier, et ne voyant autour de notre
cercle étroit (quelle impertinence, juste ciel!) que de petits hommes
et de petites choses, je me disais: «Suis-je donc condamné à vivre
et à mourir ici comme une huître attachée à son rocher?» Alors je

souffrais beaucoup; j'avais la tête chargée, fatiguée, aplatie par
l'énorme poids du rien. Mais aussi quelle compensation! je n'avais
qu'à sortir de ma chambre pour vous trouver, mes bons amis. Ici
tout est grand, mais je suis seul; et, à mesure que mes enfants se
forment, je sens plus vivement la peine d'en être séparé. Au reste, je
ne sais pas trop pourquoi ma plume, presque à mon insu, s'amuse à
te griffonner ces lignes mélancoliques, car il y a bien quelque chose
de mieux à t'apprendre.
«Je ne puis écrire autant que je le voudrais, mais jamais je ne
vous perds de vue. Vous êtes tous dans mon cœur; vous ne pouvez
en sortir que lorsqu'il cessera de battre. À six cents lieues de
distance, les idées de famille, les souvenirs de l'enfance me ravissent
de tristesse. Je vois ma mère qui se promène dans ma chambre avec
sa figure sainte, et en t'écrivant ceci je pleure comme un enfant.»
Délicieux!
XVIII
Ces sensibilités de cœur contrastent toujours en lui avec les
duretés de l'esprit. L'écrivain était acerbe, l'homme était bon; c'est le
contraire de tant d'autres, tels que Jean-Jacques Rousseau, hommes
très-humanitaires dans leurs écrits, très-personnels dans leur
conduite. M. de Maistre n'aurait pas jeté un chien de sa chienne à
cette voirie vivante où Jean-Jacques Rousseau jetait ses enfants.
Ses lettres suivent pas à pas les événements et les commentent à
sa manière.
«Après la bataille d'Iéna, dit-il, j'avais écrit à notre ami, M. de
Blacas: Rien ne peut rétablir la puissance de la Prusse. J'ai eu,
depuis que je raisonne, une aversion particulière pour le grand
Frédéric, qu'un siècle frénétique s'est hâté de proclamer grand
homme, mais qui n'était au fond qu'un grand Prussien. L'histoire

notera ce prince comme un des plus grands ennemis du genre
humain qui aient jamais existé. Sa monarchie était un argument
contre la Providence. Aujourd'hui cet argument s'est tourné en
preuve palpable de la justice éternelle. Cet édifice fameux, construit
avec du sang et de la boue, de la fausse monnaie et des feuilles de
brochures, a croulé en un clin d'œil, et c'en est fait pour toujours!»
Voyez le danger des oracles! un demi-siècle après cet anathème la
Prusse balançait l'empire en Allemagne et prospérait insolemment
malgré les vices très-réels de son origine, et malgré, qui sait? peut-
être à cause du machiavélisme de son fondateur et de ses cabinets.
Ceci s'adressait au comte d'Avaray, favori de Louis XVIII, alors
réfugié à Milan sous la protection de la Russie.
Tournez la page; vous lirez sur Bonaparte les lignes suivantes pour
justifier la paix conclue par la Russie avec l'usurpateur du royaume
de Louis XVIII.
«Je sais tout ce qu'on peut dire contre Bonaparte: il est
usurpateur, il est meurtrier; mais, faites-y bien attention, il est
usurpateur moins que Guillaume d'Orange, meurtrier moins
qu'Élisabeth d'Angleterre. Il faut savoir ce que décidera le temps,
que j'appelle le premier ministre de la Divinité au département des
souverainetés; mais, en attendant, Monsieur le Chevalier, nous ne
sommes pas plus forts que Dieu. Il faut traiter avec celui à qui il lui a
plu de donner la puissance.»
Allez plus loin, vous lirez des lettres à Louis XVIII lui-même, roi
bien digne par son esprit d'un tel correspondant.
Allez encore, vous arrivez bien inopinément à une des plus
étranges péripéties de caractère et d'imagination qui puissent
confondre le don de prophétie dans un homme assez hardi pour se
l'arroger. Nous voulons parler de la tentative d'un rapprochement
personnel du comte de Maistre avec Bonaparte.—Pour quel but? Il
est facile de le conjecturer quand on a lu ses lettres familières et les

lettres officielles plus récentes destinées à excuser sa démarche
auprès de la cour de Sardaigne; et enfin par quel intermédiaire? par
l'amitié du duc de Rovigo (Savary), accusé alors, à tort ou à droit, de
l'exécution sanglante du duc d'Enghien. Le comte de Maistre, qui
venait, deux lettres plus haut, d'anathématiser le meurtre du duc
d'Enghien, se rapprochant avec déférence de Savary qui venait
d'assister à l'exécution de la victime! Et le ministre du roi de
Sardaigne se concertant, à l'insu de son maître, avec le ministre de
Bonaparte pour opérer un rapprochement intime et secret entre
l'homme de Vincennes et le roi de Cagliari!
La plume tombe des doigts. Laissons le comte de Maistre faire lui-
même cette étonnante confession. «Ne vous fiez pas aux princes,»
dit l'Écriture. Ne vous fiez pas aux prophètes politiques, dit cette
correspondance. Lisez, car, si vous ne lisiez pas, vous ne croiriez pas.
XIX
On a vu, par les lettres précédentes, que l'envoyé oisif du roi de
Sardaigne à Pétersbourg flottait entre la résistance et
l'acquiescement à la fortune de Napoléon, et qu'il commençait à
prendre au sérieux cette fortune qu'il avait d'abord prise en
moquerie ou en haine.
On a vu de plus que l'envoyé du roi de Sardaigne s'ennuyait de
son oisiveté. Qu'avait-il à faire en effet à Pétersbourg qu'à recevoir
de loin les rumeurs des champs de bataille, des négociations, des
congrès, des entrevues d'Erfurt ou de Tilsitt entre les princes, et à
transmettre à sa cour les mille et mille commérages politiques des
salons de Pétersbourg, commérages vagues, souvent faux, sur
lesquels il échafaudait des dépêches, des plans, des combinaisons
plus propres à amuser sa cour de Cagliari qu'à la servir?

L'envoyé de Sardaigne n'avait en réalité là qu'un seul rôle: écouter
aux portes et faire de l'esprit sur ce qu'il avait entendu par le trou de
la serrure. Le métier n'allait pas à une tête si forte et si active. Il
rêvait un rôle plus conforme à sa stature; il n'aspirait à rien moins
qu'à rendre à son ombre de gouvernement un trône réel sur le
continent, per fas et nefas. On va le voir. Il voulait imposer son nom
à la reconnaissance de la maison de Savoie par un de ces services
officieux, éclatants, qui font d'un sujet le restaurateur de son prince;
ou plutôt il ne savait pas bien précisément encore ce qu'il voulait à
cet égard, car la résurrection du Piémont lui paraissait radicalement
impossible tant que Napoléon serait sur le trône, et cependant c'était
désormais à Napoléon qu'il allait s'adresser pour relever la
monarchie de Sardaigne sur le continent. Il s'agissait donc dans sa
pensée d'un de ces desseins confus, chimériques, équivoques, qui
ont besoin du succès pour être avoués. Or, puisqu'à ses propres
yeux il était impossible, Napoléon vivant, de rendre Turin, le Piémont
et la Savoie au roi de Sardaigne, c'était donc un autre royaume qu'il
fallait obtenir de Napoléon en indemnité pour cette cour. Mais, pour
que cette indemnité d'un royaume détaché par Napoléon lui-même
de ses conquêtes pût être donné au roi de Sardaigne, il fallait deux
choses: d'abord consentir à être l'obligé et pour ainsi dire le
complice du conquérant distributeur d'empires. Que devenait
l'honneur de la maison de Savoie?
Il fallait de plus accepter, après l'avoir sollicité, un de ces
royaumes arrachés par le conquérant à une autre maison régnante
pour en gratifier la maison de Savoie devenue usurpatrice à son tour.
Que devenait la légitimité?
On voit que tout cela n'était ni très-digne, ni très-logique, ni très-
moral. Les politiques n'ont pas de scrupules, mais les prophètes, qui
parlent sans cesse au nom de la morale divine, sont tenus d'en avoir.
M. de Maistre en manquait ici.
Quoi qu'il en soit, le comte de Maistre inventa dans sa féconde
imagination, une belle nuit, un plan de restauration, ici ou là, de la

cour de Sardaigne. Ce plan, il se garda bien de l'avouer à personne,
de peur qu'on ne soufflât sur sa chimère: les aventureux craignent
les conseils.
Ce plan consistait à séduire Savary, l'envoyé de Napoléon en
Russie, par les empressements de sa politesse et par les agréments
de son esprit; puis, après avoir séduit l'envoyé, de séduire le maître,
de convertir Napoléon à la contre-révolution par la puissance d'un
entretien tête à tête avec le vainqueur du monde, de l'éblouir, de le
fasciner, de le magnétiser, de le dompter à force d'audace et
d'éloquence, de le convaincre de la nécessité de rétablir la maison
de Savoie dans quelque grand établissement monarchique sur le
continent; puis, après ce triomphe du génie sur Napoléon, de revenir
à la cour de Cagliari en apportant à son souverain un royaume ou un
autre.
XX
On comprend, sans qu'il soit besoin de le dire, que l'envoyé du roi
de Sardaigne en Russie se garda bien de consulter sa cour sur une si
étrange hallucination de sa propre politique; la cour proscrite, mais
scrupuleuse, de Cagliari aurait, au premier mot, désavoué et rappelé
son ministre. Comment, en effet, la maison proscrite de Savoie
aurait-elle avec dignité mendié un trône à son proscripteur? et
comment cette maison royale, représentant dans son île la fidélité
malheureuse à la légitimité des trônes, aurait-elle pu se démentir en
expulsant elle-même une autre maison royale de ses possessions,
par la main de Napoléon, pour se déshonorer en acceptant ses
dépouilles?
Or, nous l'avons dit, on ne pouvait prendre cette indemnité de la
maison dépouillée de Savoie que sur d'autres dépouilles. Et, de plus,
comment le roi de Sardaigne, allié et protégé de la Russie, de
l'Angleterre, de l'Espagne, de l'Autriche, de la Prusse, parent enfin de

la maison de Bourbon, aurait-il justifié aux yeux de ces alliés
naturels ses relations secrètes avec Napoléon, le jour où cette
négociation ou cette intrigue viendrait à transpirer du cabinet de M.
de Maistre dans le monde?
C'était là une de ces manœuvres équivoques qui perdent plus que
la fortune d'une cour, qui perdent son caractère. Le comte de Maistre
en eut le pressentiment sans doute, car il garda un profond silence,
silence très-répréhensible, envers sa cour sur ces aventures de
diplomatie très-compromettantes pour ceux dont il était censé être
le diplomate. Quand un homme représente son souverain, l'homme
disparaît sous le ministre. Il ne lui est pas permis de dire: J'agis,
comme homme privé, dans un sens inverse de mon rôle et de mon
devoir comme ministre de ma cour. Si l'on veut agir comme homme
privé et d'après ses propres inspirations au lieu d'agir selon ses
instructions, il faut commencer par donner sa démission de son titre
d'envoyé de sa cour. Alors on est libre, on n'engage que soi; mais en
restant ministre, et en agissant comme homme, on engage sa cour
et on forfait à sa mission. Voilà les principes.
Le comte de Maistre les faussait en prétendant agir comme
homme et rester revêtu de son caractère d'envoyé de son roi.
On conçoit l'étonnement et la juste colère qui saisirent les
ministres et le roi à Cagliari quand les ministres et le roi apprirent
avec stupeur cette incartade de zèle et cette folie de fidélité dans
leur ministre à Pétersbourg. De ce jour data, pour M. de Maistre,
réprimandé et mal pardonné, une défiance et un éloignement de sa
cour à son égard qui ne lui permirent jamais de monter jusqu'où son
génie pouvait prétendre en Piémont.
Lisons de sa propre main le récit de cette incroyable échauffourée
de zèle.
XXI

«Au moment ou je m'occupais de ces idées, écrit-il plus tard au
ministre des affaires étrangères à Cagliari pour s'excuser, il arrive ici
un favori de Napoléon (Savary). Cet homme se prend de quelque
intérêt pour moi. Il est présenté dans une maison où je suis fort lié,
M. de Laval, Français résidant à Pétersbourg et chambellan de
l'empereur Alexandre. Je me demande s'il n'y aurait pas moyen de
tirer parti des circonstances en faveur du roi. Les hommes
extraordinaires (Napoléon) ont tous des moments extraordinaires; il
ne s'agit que de savoir les saisir.
«Les raisons les plus fortes m'engagent à croire que, si je pouvais
aborder Napoléon, j'aurais des moyens d'adoucir le lion et de le
rendre plus traitable à l'égard de la maison de Savoie. Je laisse mûrir
cette idée, et plus je l'examine, plus elle me paraît plausible. Je
commence par les moyens de l'exécuter, et à cet égard il n'y a ni
doute ni difficulté. Le chambellan, M. de Laval, dont il est inutile que
je parle longuement, était, comme je vous le disais tout à l'heure,
fait exprès. Il s'agissait donc uniquement d'écarter de cette
entreprise tous les inconvénients possibles, et de prendre garde
avant tout de ne pas choquer Napoléon. Pour cela je commence par
dresser un Mémoire écrit avec cette espèce de coquetterie qui est
nécessaire toutes les fois qu'on aborde l'autorité, surtout l'autorité
nouvelle et ombrageuse, sans bassesse cependant, et même, si je
ne me trompe, avec quelque dignité. Vous en jugerez vous-même,
puisque je vous ai envoyé la pièce. Au surplus, Monsieur le Chevalier,
j'avais peu de craintes sur Bonaparte. La première qualité de
l'homme né pour mener et asservir les hommes, c'est de connaître
les hommes. Sans cette qualité il ne serait pas ce qu'il est. Je serais
bien heureux si l'empereur me déchiffrait comme lui. L'empereur
Alexandre a vu, dans la tentative que j'ai faite, un élan de zèle, et,
comme la fidélité lui plaît depuis qu'il règne, en refusant de
m'écouter il ne m'a fait cependant aucun mal. Le souverain légitime
intéressé dans l'affaire (le roi de Sardaigne) peut se tromper sur ce
point; mais l'usurpateur est infaillible.

«Tout paraissant sûr de ce côté, et m'étant assuré d'ailleurs de
l'approbation de la cour de Russie, et même de la protection que les
circonstances permettaient, il fallait penser à l'Angleterre.» Il confie
son idée à l'ambassadeur d'Angleterre en Russie; celui-ci,
évidemment embarrassé de la confidence, la lui déconseille aussi
poliment qu'il peut.
«Je comptais commencer la conversation avec Bonaparte,
continue-t-il, à peu près de cette manière: Ce que j'ai à vous
demander, avant tout, c'est que vous ne cherchiez point à m'effrayer,
car vous pourriez me faire perdre le fil de mes idées, et fort
inutilement, puisque je suis entre vos mains. Vous m'avez appelé, je
suis venu; j'ai votre parole. Faites-moi fusiller demain, si vous
voulez, mais écoutez-moi aujourd'hui.
«Quant à l'épilogue que j'avais également projeté, je puis aussi
vous le faire connaître. Je comptais dire à peu près: Il me reste,
Sire, une chose à vous déclarer: c'est que jamais homme vivant ne
saura un mot de ce que j'ai eu l'honneur de vous dire, pas même le
roi mon maître; et je ne dis point ceci pour vous; car que vous
importe? Vous avez un bon moyen de me faire taire, puisque vous
me tenez. Je le dis à cause de moi, afin que vous ne me croyiez pas
capable de publier cette conversation. Pas du tout, Sire! Regardez
tout ce que j'ai eu l'honneur de vous dire comme des pensées qui se
sont élevées d'elles-mêmes dans votre cœur. Maintenant, je suis en
règle; si vous ne voulez pas me croire, vous êtes bien le maître de
faire tout ce qu'il vous plaira de ma personne; elle est ici.
«Comment donc cette idée a-t-elle été si mal accueillie à Cagliari?
Je crois que vous m'en dites la raison, sans le savoir, dans la
première ligne chiffrée de votre lettre du 15 février, où vous me dites
que la mienne est un monument de la plus grande surprise. Voilà le
mot, Monsieur le Chevalier; le cabinet est surpris. Tout est perdu. En
vain le monde croule, Dieu nous garde d'une idée imprévue! et c'est
ce qui me persuade encore davantage que je ne suis pas votre
homme; car je puis bien vous promettre de faire les affaires de S. M.

aussi bien qu'un autre, mais je ne puis vous promettre de ne jamais
vous surprendre. C'est un inconvénient de caractère auquel je ne
vois pas trop de remède. Depuis six mortelles années, mon
infatigable plume n'a cessé d'écrire chaque semaine que S. M.,
comptant absolument sur la puissance ainsi que sur la loyauté de
son grand ami l'empereur d'Autriche, et ne voulant pas faire un pas
sans son approbation, etc. C'est cela qui ne surprend pas! Dieu
veuille bénir les armes de M. de Front plus que les miennes! Quand
j'ai vu qu'elles se brisaient dans mes mains, j'ai fait un effort pour
voir si je pourrais rompre la carte. Bonaparte n'a pas voulu
m'entendre; si vous y songez bien, vous verrez que c'est une preuve
certaine que j'avais bien pensé. Il a jugé à propos, au reste, de
garder un silence absolu sur cette démarche; car je n'ai nulle preuve
qu'il en ait écrit à son ambassadeur ici, et je suis sûr qu'il n'en a pas
parlé au comte Tolstoï à Paris.
«Je n'ai demandé, ajoute-t-il, qu'une simple conversation avec
Napoléon comme simple particulier. (Nous avons montré que le
simple particulier n'existait pas dans le ministre, à moins qu'il n'eût
donné sa démission.) Il n'y avait que moi de compromis, dit-il
encore, car on était maître de m'emprisonner ou de m'étrangler à
Paris.»
XXII
Nous venons de retrouver dans les Dépêches publiées récemment
à Turin des traces plus explicites de cette affaire. Elle fut la grande
faute de la vie publique du comte de Maistre. Écoutez son entretien
secret avec Savary, et lisez quelques phrases du Mémoire que le
comte de Maistre adresse à cet aide de camp de Napoléon pour être
communiqué à Napoléon lui-même. On ne croirait pas, avant d'avoir
lu, que la confiance dans la toute-puissance de son propre génie eût
porté si loin un homme de tant de sens. Il faut croire en soi quand

on est une intelligence supérieure, mais il ne faut pas y croire
jusqu'à la folie, sous peine de tenter des choses folles.
«2 octobre 1807.
«Mardi je vis le général Savary chez M. de Laval. Après les
premières révérences, je lui dis que j'étais extrêmement mortifié de
ne pouvoir me rendre chez lui, mais que la chose n'était pas
possible, vu l'état de guerre qui subsistait en quelque manière entre
nos deux souverains.
«En effet, lui dis-je, le vôtre chasse les représentants ou les
agents du roi, et il refuse expressément de le reconnaître pour
souverain.
«Il me répondit poliment:—C'est vrai.
«Il engagea d'abord la conversation sur les émigrés, sur la justice
et l'indispensable nécessité des confiscations, etc.; car il croyait que
je voulais parler pour moi, et la veille il avait dit à M. de Laval qu'il
ne voyait pas quelles espérances je pouvais avoir pour mon maître,
mais qu'il en avait de très-grandes pour moi.
«Il me semble, lui dis-je, Général, que nous perdons du temps, car
il ne s'agit nullement de moi dans cette affaire. Supposez même que
je n'existe pas. Je n'ai rien à demander au souverain qui a détruit le
mien.
«Il parut un peu surpris. Alors il tomba sur le Piémont.—Pourriez-
vous concevoir, Monsieur, l'idée d'une restitution? etc. Ce fut encore
une tirade terrible. Je le laissai dire, car il ne faut jamais arrêter un
Français qui fait sa pointe. Quand il fut las, je lui dis:—Général, nous
sommes toujours hors de la question, car jamais je ne vous ai dit
que je voulusse demander la restitution du Piémont.
«—Mais que voulez-vous donc, Monsieur?

«—Parler à votre empereur.
«—Mais je ne vois pas pourquoi vous ne me diriez pas à moi-
même...
«—Ah! je vous demande pardon, il y a des choses qui sont
personnelles.
«—Mais, Monsieur le Comte, quand vous serez à Paris, il faudra
bien que vous voyiez M. de Champagny.
«—Je ne le verrai point, Monsieur le Général, du moins pour lui
dire ce que je veux dire.
«—Cela n'est pas possible; Monsieur, l'Empereur ne vous recevra
pas.
«—Il est bien le maître, mais je ne partirai pas, car je ne partirai
qu'avec la certitude de lui parler.
«Il en revint toujours à sa première question:—Mais qu'est-ce que
vous voulez? Enfin, Monsieur, la carte géographique est pour tout le
monde; vous ne pouvez voir autre chose que ce que j'y vois.
Voudriez-vous Gênes? la Toscane? Piombino? Il courait toute la
carte.
«—Je vous ai dit, Monsieur le Général, qu'il ne s'agit que de parler
tête à tête à votre empereur, oui ou non.
«Je vous exprimerais difficilement l'étonnement du général, et
vraiment il y avait de quoi être étonné. Cette conversation
mémorable a duré, avec une véhémence incroyable, depuis sept
heures du soir jusqu'à deux heures du matin. Un seul ami présent
mourait de peur que l'un des deux interlocuteurs ne jetât l'autre hors
des gonds; mais je m'étais promis à moi-même de ne pas gâter
l'affaire, et, pourvu que l'un des deux ait fait ce vœu, c'est assez.

«Le général Savary m'a dit en propres termes:
«On ne l'inquiétera point dans sa Sardaigne; qu'il s'appelle même
roi s'il le juge à propos; ce sera à son fils de savoir ensuite ce qu'il
est.
«Voilà une des gentillesses que j'ai entendues. Je ne vous détaille
point cette conversation; il faudrait un volume, et le livre serait trop
triste. Ce que je puis vous dire, c'est que je me suis avancé dans la
confiance du général, car en sortant il dit au chambellan qui
l'accompagnait: Je suis vif; si par hasard j'ai dit quelque chose qui
ait pu affliger le comte de Maistre, dites-lui que j'en suis fâché.
«Le résultat a été qu'il se chargerait d'un Mémoire que je lui remis
peu de jours après. Dans ce Mémoire je demande de m'en aller à
Paris avec la certitude d'être admis à parler à l'empereur sans
intermédiaire; je proteste expressément que jamais je ne dirai à
aucun homme vivant (sans exception quelconque) rien de ce que
j'entends dire à l'empereur des Français, pas plus que ce qu'il
pourrait avoir la bonté de me répondre sur certains points; que
cependant je ne faisais aucune difficulté de faire à monsieur le
général Savary, à qui le Mémoire était adressé, les trois déclarations
suivantes:
«1o
Je parlerai sans doute de la maison de Savoie, car je vais pour
cela; 2o
je ne prononcerai pas le mot de restitution; 3o
je ne ferai
aucune demande qui ne serait pas provoquée.
«Si je suis repoussé, je suis ce que je suis, c'est-à-dire rien, car
nous sommes dans ce moment totalement à bas. Si je suis appelé,
j'ai peine à croire que le voyage ne produira pas quelque chose de
bon, plus ou moins.»
Savary montre, dans cette entrevue, la rudesse, mais le bon sens
d'un soldat. Il ne flatte pas le rêve, mais il écoute l'homme. Il
expédie même son Mémoire à Napoléon.

«Mon Mémoire est parti, dit plus bas le comte. Le vent de l'opinion
l'a emporté, accompagné, favorisé plus qu'il ne m'est permis de vous
le dire. Si j'ai vécu jusqu'à présent d'une manière irréprochable, j'en
ai recueilli le prix dans cette occasion. Malheureusement tout s'est
borné à la personne, à l'exclusion de l'objet politique.»
XXIII
Ce Mémoire, que nous avons sous les yeux, est en tout une
aberration de zèle. Qu'on en juge par quelques citations.
«Je n'ai point la prétention de déployer à Paris un caractère
public; le roi mon maître ignore même (je l'assure sur mon honneur)
la résolution que j'ai prise. La grâce que je demande est donc
absolument sans conséquence. Arrivé en France, je n'ai plus de titre;
le droit publie cesse de me protéger, et je ne suis plus qu'un simple
particulier comme un autre sous la main du gouvernement. Il semble
donc que dans cette circonstance la politique ne gêne aucunement
la bienfaisance. Sa Majesté Impériale appréciera d'ailleurs mieux que
personne le mouvement qui m'entraîne.
«Au reste, quoique je connaisse les formes et que je sois très-
résolu à m'y soumettre, quoique j'aie la plus grande idée des
ministres français et que la confiance qu'ils ont méritée les
recommande suffisamment à celle de tout le monde, néanmoins je
dois répéter ici à M. le général Savary ce que j'ai eu l'honneur de lui
dire de vive voix: c'est que mon ambition principale, en me rendant
à Paris, serait, après avoir rempli toutes les formes d'usage, d'avoir
l'honneur d'entretenir en particulier Sa Majesté l'Empereur des
Français. Pour obtenir cette faveur, rien ne me coûterait; mais, si je
ne puis y compter, le courage m'abandonne. Si l'on peut voir au
premier coup d'œil quelque chose de trop hardi dans cette ambition,
la réflexion prouvera bientôt que le sentiment qui m'anime ne peut
s'appeler audace ni légèreté, et que l'homme qui prend une telle

détermination y a suffisamment pensé. Je sens d'ailleurs et je
proteste que c'est une grâce, et que je n'y ai pas le moindre droit;
mais, pour la rendre moins difficile, ou pour rendre au moins la
demande moins défavorable, je ne fais aucune difficulté de faire à M.
le général Savary les trois déclarations suivantes:
«1o
Si l'Empereur des Français avait l'extrême bonté de
m'entendre, j'aurais sans doute l'honneur de lui parler de la maison
de Savoie;
«2o
Je ne prononcerais pas le mot de restitution;
«3o
Je ne ferais aucune demande qui ne serait pas provoquée.
«J'ose croire que ces trois déclarations excluent jusqu'à
l'apparence de l'inconsidération, et, quand même mon désir serait
repoussé, j'ose croire encore que Sa Majesté l'Empereur des Français
n'y verrait rien qui choque les convenances, rien qui ne s'accorde
parfaitement avec la juste idée qu'il doit avoir de lui-même.»
XXIV
L'empereur Napoléon ne répondit même pas à une demande
d'audience si extraordinaire et qui ne pouvait que l'embarrasser. Il ne
pouvait sacrifier ses départements du Piémont incorporés à l'empire
à une conversation éloquente avec un homme d'excentricité. Il ne
pouvait improviser un trône pour M. de Maistre sans détrôner ou un
autre souverain des vieilles races, ou un nouveau souverain de sa
propre maison. Le rêve eut un triste réveil.
Tout fut connu. La cour de Cagliari, de plus en plus surprise, ne
ménagea pas les termes dans sa réprimande à son ministre en
Russie. Nous voyons le contre-coup de ces mécontentements très-
graves de la cour de Cagliari à l'amertume des répliques du comte

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com

Planning And Task Performance In A Second Language Language Learning And Language Teaching Rod Ellis

More Related Content

Similar to Planning And Task Performance In A Second Language Language Learning And Language Teaching Rod Ellis (20)

Recently uploaded (20)

Planning And Task Performance In A Second Language Language Learning And Language Teaching Rod Ellis