Applied Linguistics 30/4: 474–509 ß Oxford University Press 2009

doi:10.1093/applin/amp042 Advance Access published on 30 November 2009

The Differential Effects of Three Types of

Task Planning on the Fluency, Complexity,

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

and Accuracy in L2 Oral Production

ROD ELLIS
University of Auckland and Shanghai International Studies University

The main purpose of this article is to review studies that have investigated the
effects of three types of planning (rehearsal, pre-task planning, and within-task
planning) on the fluency, complexity, and accuracy of L2 performance. All three
types of planning have been shown to have a beneficial effect on fluency but the
results for complexity and accuracy are more mixed, reflecting both the type of
planning and also the mediating role of various factors, including task design
and implementation variables and individual difference factors. A secondary
purpose is to outline a theory that can account for the role that planning
plays in L2 performance. The article concludes with a list of limitations in the
research to date.

INTRODUCTION
The study of the effect of planning on how L2 learners perform oral tasks is of
both theoretical and practical interest. Theoretically, it serves to test claims
regarding the nature of variability in learner language and the validity of
models of speaking such as that of Levelt (1989) as this has been applied to
L2 production (de Bot 1992; Kormos 2006). Practically it can help inform the
methodology of task-based teaching, where one of the options available for
implementing tasks concerns whether or not to allow students time to plan
and, if so, what kind of planning and for what length of time.
Ellis (2005b) distinguished three kinds of planning. A basic distinction is
drawn between pre-task (i.e. the planning that is done before learners perform
a task) and within-task planning (i.e. the planning that occurs on-line while
learners are actually performing a task). Pre-task planning can be further
divided into rehearsal (i.e. planning takes the form of an opportunity to per-
form the complete task once before performing it a second time) or strategic
planning (i.e. planning what content to express and what language to use but
without opportunity to rehearse the complete task). Within-task planning can
also take two forms. It can be pressured (i.e. learners are required to perform
the task rapidly by specifying a time limit) or unpressured (i.e. they are given
an unlimited amount of time to perform the task). Clearly, the distinction
between pressured and unpressured task performance is continuous rather
than dichotomous.
 R. ELLIS 475

Table 1: Definitions of fluency, complexity, and accuracy (based on Skehan

and Foster 1999: 96–97)
Aspect Definition

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Fluency The capacity to use language in real time, to emphasize
meanings, possibly drawing on more lexicalized systems.
Complexity/Range The capacity to use more advanced language, with the
possibility that such language may not be controlled so
effectively. This may also involve a greater willingness to
take risks, and use fewer controlled language subsystems.
This area is also taken to correlate with a greater likelihood
of restructuring, that is, change and development in the
inter-language system.1
Accuracy The ability to avoid error in performance, possibly reflecting
higher levels of control in the language as well as a conser-
vative orientation, that is, avoidance of challenging structures
that might provoke error.

The effects of planning on oral task performance have been investigated

with reference to three aspects of language production—fluency, complexity,
and accuracy. Notional definitions of these are shown in Table 1, based on
Skehan and Foster (1999). Operational definitions have varied considerably, as
will become evident in the subsequent sections of this article. These differences
in the operational definitions are problematic as they make comparisons across
studies difficult in some instances.
The three aspects of learner performance can be seen as constituting a lear-
ner’s language proficiency. That is, it is assumed that a proficient speaker will
be able to perform tasks fluently and accurately, using complex language.
Skehan (1998) has proposed that speakers of a language possess both a lex-
icalized system consisting of exemplars in the form of words and formulaic
sequences and a rule-based system consisting of knowledge of underlying,
abstract patterns, and that performance involves drawing variably on both
depending on the task conditions. A corollary of this claim is that proficiency
needs to be flexible to allow learners to vary how they draw on their lexica-
lized and rule-based systems according to the performance conditions.
A second corollary is that learners can compensate for inadequacy in one
system by drawing on the other. A third corollary is that there are likely to
be trade-offs as, for example, when a learner focuses on structural complexity,
accuracy is adversely affected. However, this last claim has been challenged
by Robinson (2001), who argues that such a trade-off is not inevitable as
complexity and accuracy can be positively correlated in task-based perfor-
mance (i.e. when a task is complex in terms of the number of elements
involved and there is an opportunity for pre-task planning). Both Skehan’s
and Robinson’s theories of task-based performance also acknowledge the role
476 THE EFFECTS OF THREE TYPES OF TASK PLANNING

that individual difference factors (such as the learner’s propensity for risk-
taking or working memory) have on performance.
My purpose in this article is threefold. First, I will examine a number of
studies that have investigated the effects of the different kinds of planning on

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

the fluency, complexity, and accuracy of learners’ performance of oral tasks
with a view to identifying the mediating variables and the impact they have on
performance. Secondly, I will consider the implications of these findings for a
theory of the role that planning plays in L2 oral performance. Finally, I will
point to a number of limitations in the research to date and suggest directions
for future research.

EXAMINING THE PLANNING STUDIES

Rehearsal
Rehearsal can be seen as a special type of pre-task planning. That is, perfor-
mance of a task at one time can be seen as providing planning for performance
of the same task at a second time.
Table 2 summarizes three studies that have investigated the effects of rehear-
sal on L2 learners’ fluency, complexity, and accuracy. There are two other
studies that have examined task rehearsal but these have not been included
as they measured its effects in other ways. Lynch and Maclean (2000, 2001),
for example, undertook a qualitative analysis of the discourse produced by 14
learners of English performing a poster-carousel task that involved them
responding repeatedly to the same or similar questions from fellow students
about a poster they had prepared. Nemeth and Kormos (2001) examined the
number of supports that learners provided for their claims and the frequency
with which lexical expressions of argumentation were used in a repeated
argumentative task. Neither of these studies was based on quantitative mea-
sures of fluency, complexity, and accuracy.
Clearly, the three studies summarized in Table 2 provide only limited evi-
dence regarding the effects that rehearsal has on performance. The studies
have examined two principal questions: (i) Does repeating a task have any
effect on performance of the same task? And (ii) does repeating the same task
have any effect on performance of a new task? All three studies addressed the
first question, while Bygate (2001) and Gass et al. (1999) addressed the second
question.
All three studies produced evidence to show that rehearsal of a task has
a beneficial effect on learners’ subsequent performance of the same task.
However, both Bygate and Gass et al.’s studies found there is no transference
of the rehearsal effect to a different task, even when this is the same type as the
original task.
The beneficial effects of task repetition are most clearly evident in fluency
and complexity. Both Gass et al. (1999) (if we take their holistic measure as
primarily a measure of fluency) and Bygate (1996, 2001) found that when
 R. ELLIS 477

learners repeated a task they gained fluency and complexity. In the case of
Bygate, the gap between the repetitions of the task was 10 weeks suggesting
that the effects were not dependent on the opportunity for immediate recall
but on traces established in long-term memory. Task repetition may have less

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

effect on accuracy. Bygate (2001) reported no statistically significant effect on
his general measure of accuracy. However, Gass et al. found that repetition led
to improvement in the use of one linguistic feature—choice of ser and estar.
These studies raise a crucial issue—the relationship between performance in
the L2 and acquisition. Given that the effects of repeating a task did not trans-
fer to the performance of a new task, there is no clear evidence that task
rehearsal assists acquisition. Bygate makes the point that for acquisition to
occur ‘massed’ repetition practice may be needed. Another explanation, how-
ever, is that for task repetition to have an effect on acquisition learners need
some kind of feedback on their initial performance of the task. An interesting
study by Sheppard (2006) supports this claim. Sheppard compared the fluency,
complexity, and accuracy of a control group of Japanese undergraduate stu-
dents who just repeated the same oral narrative task with experimental groups
that repeated the task but also received either input or feedback designed to
draw their attention to linguistic form between the first and second perfor-
mance. The control group improved in fluency (pruned syllables per minute),
complexity (grammatical phrases per T-unit but not lexical richness) but only
to a very limited extent in accuracy, results that were very similar to Bygate
(2001). The group receiving input between the task performances improved
in fluency, complexity (especially lexical richness)2, and accuracy, while the
group receiving feedback improved in fluency, complexity (lexical richness but
not grammatical phrases per T-unit), and markedly in accuracy. This study also
investigated the transfer of effects to a new task, which is arguably needed to
demonstrate acquisition. It found that there was no transfer in the case of the
control group but that there was in the two experimental groups for grammat-
ical complexity, although not for fluency and accuracy. It would seem then
that rehearsal alone has an effect on performance but that rehearsal in
conjunction with some other treatment has an effect on all three aspects of
language use and, in the case of complexity possibly also on acquisition.

Strategic planning
Table 3 summarizes a total of 19 studies that investigated the effects of strategic
planning on task-based performance in terms of fluency, complexity, and
accuracy. This does not constitute a sufficiently large sample to conduct
a meta-analysis of the kind reported in Norris and Ortega (2006a), so I will
provide a descriptive synthesis based on the identification of a number of key
variables in the studies.
Norris and Ortega (2006b) note that a ‘research synthesis always includes
an explicit articulation of how the relevant literature was searched and
how primary studies were selected for review’ (p. 6). In this case, I searched
Table 2: Studies investigating the effects of rehearsal on fluency, complexity, and accuracy
Study Learners Setting Research Method Dependent variables Results
questions

Bygate (1996) One learner of Laboratory What effects Learner asked to narrate Fluency: Only a small improvement
English does repeating a Tom and Jerry cartoon  Number of repeti- in accuracy was observed.
a narrative task on two occasions—3 days tions. Clearer evidence in
have on the apart. Complexity: changes in complexity
learners’ use  Syntactic complexity; and grammatical variety
of language?  Number of verb (e.g. an increase in
forms; and number of past tense
 Range of connectors. forms) and also in lexical
Accuracy: variety (e.g. an increase
 Percentage of errors; in type-token ratio). There
and was also a reduction in
 Lexical selection/ the number of inappro-
collocation. priate expressions.
Number of self-correcting
repetitions increased.
Gass et al. (1999) One hundred Laboratory Does task repe- Oral narratives based on  Holistic ratings of  The same content
and four under- tition yield more snippets from Mr Bean general proficiency group’s holistic ratings
478 THE EFFECTS OF THREE TYPES OF TASK PLANNING

graduate sophisticated videos. Learners recorded using magnitude of improved from tasks 1–3
students in language use? the stories as they estimation method. but were not sustained
fourth semester Will more accu- watched the videos.  Changes in ser and in task 4 (the new task).
Spanish course rate/sophisti- Three groups: estar.  The group differences
cated language  Same content group:  Lexical sophistica- on task 4 were not
use carry over recorded the same nar- tion—ratio of significant.
to a new rative three times (with sophisticated words  The same content group
context? 2- to 3-day intervals) (outside the most also improved more on
followed by a new nar- common 200 words) ser and estar than the
rative. to total words. different content group

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

  Different content group: but this improvement
recorded a new narra- did not generalize to
tive on four occasions. task 4.
 Control group: recorded  Less common words
their narrative twice (mainly closed content)
(2 weeks apart). Each used more frequently by
group recorded the same the same content group.
first and the same last
story.
Bygate (2001) Forty-eight ESL Laboratory What effect does Two sets of tasks: (i) a Fluency: Task repetition (times 1
learners in UK practising the narrative set and (ii) an  Number of unfilled and 5) led to increased
same type of interview set. pauses per T-unit. complexity on both the
tasks have on Three groups: Complexity: narrative and interview
learner perfor- Over a 10-week period all  Number of words per tasks and to decreased
mance of a new did time 1 and time 5 T-unit. fluency in the interview
task type? narrative and interview Accuracy: (but not the narrative)
What effect tasks. At time 5, they  Errors per T-unit. task. The increase in
would repeating repeated the time 1 task accuracy in both task
the same task and also did a new task. types was not statistically
have on perfor- Two experimental groups significant.
mance of a new each worked on one of Task-type practice had a
task? the task types, repeating very weak effect. Only
each task once. limited evidence that
Control group completed practising a task type
only time 1 and 5 tasks. aided performance of a
new task of the same type.
Little evidence of any
trade-off effects.
R. ELLIS
479

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Table 3: Studies investigating the effects of strategic planning on fluency, complexity, and accuracy
Study Learners Setting Research questions Method Dependent variables Results

Foster (1996) Thirty-two Classroom— What effect does Three tasks: Fluency: Planners paused less, were
intermediate groups planning time have (1) personal information;  number of pauses; silent less, and produced
ESL learners in on students’ flu- (2) narrative; and  total silence; and fewer repetitions than
the UK ency, complexity, (3) decision making.  repetitions. non-planners.
and accuracy? First 5 min analysed. Complexity: Planners used a greater
Planning time = 10 Three groups:  variety of past tense variety of past tense forms
min  guided planning; forms; and and produced more
 unguided planning;  clauses per c-unit. clauses per c-unit
and control. especially with guided
planning in more cogni-
tively demanding tasks
(i.e. decision making).
Unguided planning bene-
fited accuracy but only in
the personal information
and the decision making
task (i.e. not in the
narrative task).

Foster and As in Foster Classroom— General hypothe- As in Foster (1996). Fluency: Fluency: for tasks 1 and 3
480 THE EFFECTS OF THREE TYPES OF TASK PLANNING

Skehan (1996) (1996) groups sis—planning will  reformulations; planning (but not type of
have an effect on  replacements; planning) led to fewer
fluency, complex-  false starts; pauses and less silence but
ity, and accuracy.  repetitions; for task 2 guided planning
The effects of plan-  hesitations; had a greater effect than
ning will depend  pauses (>1 s); and unguided. In (iii) planning
on both type and  silence total. led to more repetition,
task complexity. Complexity: hesitation, and replace-
 clauses/c-units; and ments.

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

  variety of verb forms. Complexity: clear plan-
Accuracy: ning effects for subordi-
 error-free clauses; nation (detailed planning
and > undetailed planning >
 lexical errors. no-planning). Greater
variety of past tense forms
found on tasks 1 and 2
but not 3.
Accuracy: more error-free
clauses in planning condi-
tions for tasks 1 and 3 but
not for 2. No difference
between guided and
unguided planning.
Wendel (1997) Forty Japanese Laboratory— Planning will lead Learners watched two Fluency: Overall effect for planning
learners of one-on-one to increased flu- films and then retold  syllables per minute; found (overall increase of
English in a with the ency, complexity them under two  mean length of 29%); no effect for film
Junior College— researcher and accuracy in conditions: pauses. and no interaction
low intermedi- oral narratives.  10 min unguided Complexity: between planning and
ate to planning (with note-  T-unit (inclusive of film.
intermediate taking); and passive construc- Fluency: planning led to
 no planning. tions); and more syllables per minute
 lexical richness and shorter pauses.
(number of word Complexity: planning led
families). to higher T-unit scores but
Accuracy: not to more word families
 percentage of correct being used.
verbs; and Accuracy: effect of plan-
 clauses. ning not evident in cor-
rectness on verbs.
R. ELLIS

(continued)
481

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Table 3: Continued
Study Learners Setting Research questions Method Dependent variables Results

Wigglesworth One hundred Testing What effect does Five tasks varying in Analytic ratings of flu- Planning time resulted in
(1997) and seven adult planning have on cognitive difficulty. ency, grammar, and greater fluency, complex-
ESL learners in ratings and dis- One minute planning intelligibility. ity, and accuracy,
Australia. course analytic time. Fluency: especially in the high-
Twenty-eight measures of flu-  number of self- proficiency learners and in
divided into ency, complexity, repairs. the tasks with high cogni-
high and low and accuracy? Complexity: tive load. But was not
proficiency  amount of subordi- evident in all measures
nation. and also not in the
Accuracy: analytic ratings.
 suppliance of
plural –s;
 verbal morphology;
and
 indefinite article.
Skehan and Forty pre- Classroom— What effect does As in Foster (1996). Fluency: Planners paused less than
Foster (1997) intermediate groups planning have on Two variables examined:  number of pauses no-planners in the (i)
ESL learners oral task  planning time (10 min); >1 s. personal and (ii) narrative
in UK performance?  post-task (public Complexity: tasks but not in the (iii)
performance).  clauses per c-unit. decision-making task.
Accuracy: Planners produced more
482 THE EFFECTS OF THREE TYPES OF TASK PLANNING

 error-free clauses. complex language in tasks

(i) and (iii) but not (ii).
Planners were more accu-
rate than non-planners
but only in the narrative
task.
Some evidence that post-
task led to greater fluency
(in task (ii)) and greater
accuracy (in task (iii)).

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Mehnert (1998) Thirty-one Laboratory— What effect does Two tasks: Fluency: Fluency: planning time
under graduate one-on-one the length of plan-  instruction task (struc-  number of pauses; has a significant effect on
students of with the ning time have on tured); and  total pausing time; a number of measures
German—inter- researcher oral performance?  exposition (unstruc-  length of run; and (e.g. syllables per minute)
mediate level tured) task.  syllables per minute. in both tasks but espe-
Four planning groups: Complexity: cially in structured task.
 no planning;  number of subordi- All planning groups were
 1 min; nate clauses per more fluent than no-
 5 min; and T-unit; and planning group. More
 10 min.  number of S-nodes planning time resulted in
per T-unit. greater fluency.
Accuracy: Complexity: no effect for
 percentage of error- planning in any of the
free clauses; groups in both tasks.
 errors per 100 words; Accuracy: in the struc-
and tured task, 10 min of
 correct suppliance of planning resulted in fewer
three word order errors per 100 words than
rules. no-planning; in the
Density of speech: unstructured task the
 weighted number of difference lay between
lexical items as 1 min of planning and
percentage of no-planning.
total items. Density: in the structured
task only 10 min planning
resulted in greater lexical
density than other
planning times.

(continued)
R. ELLIS
483

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Table 3: Continued
Study Learners Setting Research questions Method Dependent variables Results

Ortega (1999)3 Sixty-four Laboratory— Does pre-task plan- Within-subject compari- Fluency: Planning produced signifi-
advanced level one-on-one ning increase the son of planning and no  pruned speech rate. cant increases on all
learners of with another complexity, accu- planning. Complexity: measures except type-
Spanish as learner racy, and fluency? Two narrative tasks based  words per utterance; token ratio.
an FL What do learners on a picture series. and
do when they plan? Ten minutes unguided  type-token ratio.
planning + note-making. Accuracy:
Retrospective semi-  target language use
structured interviews. of noun modifiers;
and
 target language use
of articles.
Rutherford Thirty-one ESL Language What effect does Two oral narrative tasks Complexity: No statistically significant
(2001) adult students laboratory— teacher-directed based on tightly structured  number of clauses effect found for planning
at mid- students per- planning have on picture stories’ question- per c-unit; and on either complexity or
intermediate formed tasks students’ oral pro- naire asking students what  length of c-unit. accuracy measures.
level individually duction and actual types of planning activity Accuracy: Descriptive statistics
nature of their they had engaged in.  number of error-free showed differential effects
planning? Five minutes guided plan- clauses; and for the two tasks in the
484 THE EFFECTS OF THREE TYPES OF TASK PLANNING

ning time (learners took  number of errors per planned and unplanned
notes). 100 words. conditions.
Wigglesworth Four hundred Testing— What effect do task Five task types. Analytical ratings of: The familiar task was
(2001) ESL learners at involving characteristics and Planning = 5 min.  grammar; easier when there was no
different levels both NS and conditions Planning was manipulated  fluency; planning. Also planning
of proficiency in NNS (including plan- in conjunction with two  cohesion; and had an adverse effect on
Australia interlocutors ning) have on oral other task characteristics:  communicative performance in both
task performance? familiarity and structure. effectiveness. structured and unstruc-
One task performed with- tured tasks.
out any manipulation.

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Yuan and Ellis Forty-two Laboratory— What are the effects A oral narrative picture Fluency: Fluency: planning group
(2003) Chinese under- one-on-one of pre-task plan- task.  syllables per minute; produced more pruned
graduate stu- with the ning on fluency, No planning and pre-task pruned syllables per syllables per minute than
dents (English researcher complexity, and planning (10 min)—writ- minute no-planning group.
majors)—TOEFL accuracy in an oral ten notes. Complexity: Complexity: planning
373–520 narrative? Task performance  clauses per T-unit; group produced more
pressured. number of different clauses per T-unit than
verb forms; and no-planning group (no
 mean segmental difference on other two
type-token ratio. measures).
Accuracy: Accuracy: no statistically
 error free clauses; significant differences
and found.
 percentage of correct
verb forms.
Tajima (2003) Sixty-one Laboratory— Does provision of The task involved giving Fluency: Fluency:
Korean learners one-on-one planning time have directions by leaving a  number of moras  planning resulted in an
of L2 Japanese with the an effect on L2 oral message on an answer per second; increased speech rate
(post-beginner researcher performance? What phone.  number of pauses and in fewer pauses.
level) role do learners’ Students completed one per minute; and Complexity:
subjective feelings practice task and the  subjective rating  no effect found for
about the planning experimental tasks. on a 1–5 scale. grammatical complexity;
play? They were given 10 min Complexity: and
unguided planning time;  number of clauses  planners produced more
the mean planning time per AS-unit; and lexically complex lan-
used was 6:55 min. After type-token ratio. guage.
finishing the tasks they Accuracy: Accuracy:
completed a questionnaire  percentage of error-  overall planners were
asking them about the free clauses; and more accurate; and
planning they had  target language use no effect for particles
undertaken. of particles. found.
The positive effects for
planning were only evi-
R. ELLIS

dent in those learners

who evaluated the plan-
485

ning time positively.

(continued)
Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023
Table 3: Continued
Study Learners Setting Research questions Method Dependent variables Results

Sangarun Forty Thai Laboratory— What effects do Two tasks: Fluency: An overall effect for
(2005) Grade 11 EFL one-on-one three types of  an instruction task  syllables per minute; planning was found on
learners— with the planning (meaning  an argumentative task. pruned syllables per measures of fluency,
intermediate researcher focused/form Learners were given minute; and complexity, and accuracy.
level focused and form detailed planning instruc-  percentage of total But no difference found
and meaning tions and took notes. pausing time. among the three planning
focused) have on Fifteen minutes planning Complexity: conditions.
oral performance? time. They were asked to  sentence nodes per Overall, the learners
plan aloud. T-unit; and focused on meaning irre-
Planning performed as  clauses per T-unit. spective of the type of
think-aloud tasks. Accuracy: planning.
 percentage of error-
free clauses; and
errors per 100 words.
Kawauchi Twelve high- Language What effect does Three picture-based nar- Fluency: Proficiency:
(2005) intermediate laboratory— proficiency have on rative tasks from STEP  amount of speech Fluency—in unplanned
and 11 low profi- planned oral test. Learners performed (number of words); condition order was low <
advanced ciency performance? all two tasks in three and high < advanced but in
Japanese EFL students What is the effect cycles.  percentage of planned condition it was
486 THE EFFECTS OF THREE TYPES OF TASK PLANNING

learners performed of different types of Within-subjects design; repeated words. low < high = advanced.
tasks planning? learners performed same Complexity: Complexity:
individually task in both unplanned;  number of clauses Advanced group did not
and planned conditions. per T-unit; differ in planned and
Three types of planning  number of words per unplanned conditions.
activities (10 min): T unit; Accuracy:
writing;  number of subordi- Low proficiency signifi-
rehearsal; and nate clauses; and cantly different from other
reading—carried out after number of word two groups in both con-
unplanned task. types. ditions. Results varied

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

 Accuracy: according to verb cate-
 suppliance of past gories (e.g. only low pro-
forms for copula, ficiency group improved
regular and irregular on regular past tense)
verbs. Planning type:
No differences found in
the effect of the three
planning activities on any
of the measures.

Skehan and Sixty-one inter- Classroom— The study investi- Decision-making task Fluency: Fluency: both planning
Foster (2005) mediate level groups gated the effects of (assigning sentence for  end of clauses conditions produced fewer
ESL students guided vs. crimes committed). pauses; end of clauses pauses than
unguided planning Three groups:  mid-clause pauses; the no-planning group at
and also the effects no planning;  filled pauses; both times. Detailed plan-
of planning at two 10 min guided planning;  length or run; ners produced more filled
different times (first and  reformulations; and pauses than undetailed
5 min and second 5 10 min unguided  false starts. and no-planners but only
min) during a task planning. Complexity: at time 2.
performance  subordinate clauses Complexity: detailed
per AS-unit. planners produced more
Accuracy: subordinate clauses than
 error-free clauses and other two groups but only
clause length. at time 1.
Accuracy: detailed plan-
ners produced more
accurate speech than
no planning but only at
time 1.

(continued)
R. ELLIS
487

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Table 3: Continued
Study Learners Setting Research questions Method Dependent variables Results

Elder and One hundred Testing— What were the Two narrative tasks of the Analytical ratings of No effect for planning on
Iwashita (2005) and ninety- language effects of planning kind used in the Test of fluency, complexity, the scores obtained from
seven adult ESL laboratory time on students’ Spoken English. There and accuracy. the analytic ratings.
learners in performance of were also six other tasks Fluency: Also no effect on any of the
Australia narrative tasks in a completed under different  number of repeti- discourse analytic mea-
(TOEFL range testing situation as conditions. tions, false starts, sures.
427–670) measured by means Planning = 3 min + 75 s. reformulations, hesi- Also test-takers reported no
of ratings and dis- No planning = 75 s to read tations and pauses statistically significant dif-
course analytic instructions. divided by total ference in their perception
measures? Planning was manipulated speaking time. of task difficulty in the no-
in conjunction with other Complexity: planning and planning
task characteristics and  number of clauses conditions.
conditions. per c-unit.
Accuracy:
 percentage of error-
free clauses.
Tavokoli and Eighty adult EFL Testing—lan- What is the effect Narrative tasks of differing Fluency: Significant differences
Skehan (2005) learners in guage of task structure on complexity based on sets  speech rate; evident for:
Iran—elemen- laboratory— the fluency, accu- of pictures.  total silence;  planning (planners’ lan-
tary and inter- one-on-one racy, and complex- Five minutes allocated to  various repair mea- guage was more fluent,
mediate level with the ity of students’ look at the picture sets sures (e.g. false more complex and more
researcher performance of and do the planning. starts); and accurate);
488 THE EFFECTS OF THREE TYPES OF TASK PLANNING

narrative tasks?  length of run.  proficiency level (more

To what extent Complexity: proficient learners’ lan-
does proficiency  clauses per AS-unit. guage was more fluent,
mediate these Accuracy: complex and accurate);
effects?  error-free clauses. and
 task structure (number of
pauses, false starts, com-
plexity, and accuracy).
The interactions between
planning, proficiency, and
task structure were
non-significant.
Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023
Gilabert (2007) Forty-eight first- Laboratory What effect does Four wordless comic strips Fluency: Both planned conditions
and second-year (?)—no planning have on requiring oral narratives  pruned speech rate (i.e. (2) and (4)) resulted in
university stu- information learners’ fluency, performed under four (syllables per greater fluency and lexical
dents with a provided complexity, and conditions: minute). richness than the two
lower interme- accuracy? (1) no-planning/Here- Complexity: unplanned conditions (i.e.
diate proficiency In what ways does and-Now;  Guiraud’s index of (1) and (3)). There was no
level in English the planning vari- (2) 10 min planning/Here- lexical richness; and effect for planning on
able interact with and-Now S-Nodes per T-units. either complexity or accu-
Here-and-Now (3) no-planning/There- Accuracy: racy.
variable? and-Then  number of self- In contrast, the There-and-
(4) 10 min planning/ repetitions. Then conditions (i.e. (3)
There-and-Then and (4)) resulted in greater
accuracy than the Here-
and-Now conditions (i.e.
(1) and (2)) irrespective of
planning.

Guará-Tavares Fifty intermedi- Laboratory— What effect does Two picture-cued narra- Fluency: No statistically significant
(2008) ate Brazilian one-on-one planning have on tive tasks.  words per minute differences between control
learners of with the learners’ fluency, Task 1 performed by both (unpruned); and experimental groups
English enrolled researcher complexity, and control and experimental  words per minute on task 1. On task 2 the
in university accuracy? groups under no-planning (pruned); experimental group was
courses Is there a relation- condition.  number of silent significantly more accurate
ship between Task 2 performed by con- pauses per c-unit; and complex than the
working memory trol group under no- and control group.
and learners’ L2 planning condition and  percentage of total In the control group,
production under under the planning con- pausing time. working memory correlated
no-planning and dition by the experimental Complexity: significantly with accuracy
planning group.  number of clauses in task 2 (unplanned). In
conditions? Planning = 10 min. per c-unit. the experimental group,
All learners completed the Accuracy: working memory signifi-
Speaking Span Test as a  number of errors per cantly related with mea-
test of working memory. one hundred words; sures of fluency and
On-line and retrospective and complexity in task 2
R. ELLIS

reports of planning col-  percentage of error- (planned).

lected from the experi- free clauses.
489

mental group in task 2.

(continued)
Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023
Table 3: Continued
Study Learners Setting Research questions Method Dependent variables Results

Mochizuki and Fifty-six first- Classroom— What effect does Picture story re-telling Fluency: Fluency:
Ortega (2008) year high school learners guiding learners to task with audio narrative  mean number of trend for non-guided
students in worked in attend to a specific stimulus. One-way task. words per minute. planners to be more fluent
Japan (beginner pairs grammatical feature Three groups: Complexity: than no planning and
level) have on perfor- no-planning—students  mean length of T- guided planning group.
mance of this fea- re-told the story immedi- unit; Complexity:
ture in a task? ately after listening to it  mean number of no group differences in
and while looking at the clauses per T-unit; general complexity.
pictures; and Accuracy:
unguided planning—5  number of relative guided planners produced
min; clauses per T-unit. more and more accurate
guided planning—5 min; Accuracy: relative clauses than the
students received handout  frequency of use of unguided planners.
about how to make rela- relative clauses; and
tive clauses. degree of accurate
use of relative
490 THE EFFECTS OF THREE TYPES OF TASK PLANNING

clauses.

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

 R. ELLIS 491

relevant publications for experimental studies that compared L2 learners’ per-

formance of meaning-centred tasks on the three aspects of language use (flu-
ency, complexity, and accuracy) in terms of whether or not there was strategic
planning. Thus a key criterion for inclusion was the presence of a no-planning

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

as well as one or more planning groups. A second criterion was the use of
inferential statistics to identify statistical differences in group performances.
I will discuss the results of these studies in terms of the following parameters:
learners, setting, tasks, and planning:
1. The learners (second vs. foreign; proficiency): the studies were fairly
evenly divided between second-language and foreign-language learners.
Many of the studies provide only very crude indications of the proficiency
level of the learners. Most of the studies investigated learners who were at the
(roughly) intermediate proficiency level. The testing studies examined learners
with a range of proficiency levels. Three studies (Wigglesworth 1997;
Kawauchi 2005; Tavakoli and Skehan 2005) deliberately manipulated profi-
ciency as a variable.
Four studies (Ortega 1999; Tajima 2003; Sangarun 2005; Guará-Tavares
2008) investigated how learners responded to the opportunity to plan and
how they used their planning time. Clearly, learners’ orientation to the plan-
ning is a potentially crucial learner variable.
Only one study (Guará-Tavares 2008) investigated how individual learner
difference factors interact with planning to influence L2 production. Guará-
Tavares investigated the mediating role of working memory.
2. The setting (classroom vs. laboratory vs. testing): three different settings
can be identified in these studies; the classroom, laboratory, and testing.
Included in the laboratory setting are studies where the students performed
the task in a language laboratory and those where they performed them in a
non-instructional setting. Four of the five classroom studies were carried out
by Foster and Skehan. Foster (1996) makes the point that such studies have
greater ecological validity when the purpose of the study is to identify the role
that planning can play in language pedagogy. It should be noted that setting is
confounded with another variable—the length of planning time, as the testing
studies typically provided a much shorter time for planning than the classroom
or laboratory studies.
3. The tasks (interactive vs. monologic; simple vs. complex): a variable of
considerable interest to task-based researchers is the effect that the design of
the task has on L2 performance. Two design variables figure prominently in
the 19 studies—participatory structure (interactive vs. monologic) and task
complexity.
It should be noted that many of the tasks were narrative in nature but that
these can be performed in both an interactive or monologic mode (i.e. the
learner can be asked to tell a story to another person or can be asked to tell it
into a cassette recorder). Thirteen of the studies involved performing the task
in an interactive mode, while six (including all four testing studies) required
learners to perform them monologically.
492 THE EFFECTS OF THREE TYPES OF TASK PLANNING

Task complexity is a thorny issue as multiple variables are involved. Readers

should refer to Robinson’s (2001, 2003, 2007) Cognition Hypothesis for an
attempt to develop a theoretically based framework to account for task com-
plexity. However, as Samuda and Bygate (2008) noted, Robinson’s (and other)

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

frameworks are incomplete in a number of respects—for example, they do not
specify how to weight the different factors hypothesized to contribute to com-
plexity. Despite these problems, task complexity is clearly a key variable affect-
ing L2 performance and for this reason alone is attracting considerable
attention in its own right (see, for example, Michel et al. 2001; Gilabert
2007). The focus in this review, however, is not on task complexity per se
but on how this variable interacts with strategic planning.4 Complexity factors
that figure in the 19 studies in Table 3 are: (i) the degree of familiarity with the
task context; (ii) the degree of structure in the information to be communi-
cated; (iii) the number of distinct referents to be encoded; and (iv) temporal
reference (Here-and-Now vs. There-and-Then).
4. Planning (length; guided vs. unguided; form vs. meaning focused): of
obvious interest is whether the type of planning that learners are asked to
do affects their performance. One variable is the length of the planning. The
majority of the studies in Table 3 allocated 10 min. This appears to have
become the standard planning time. However, most of the studies do not
report whether the learners used the full 10 min available to them. Tajima
(2003) found considerable variance in the actual time his learners spent plan-
ning, with a mean of 6.55 min. The testing studies allocated a much shorter
time, reflecting the demands of the testing context. Wigglesworth (1997)
allowed only 1 min, Elder and Iwashita (2005) 3 min, and Tavakoli and
Skehan (2005) 5 min. One study (Mehnert 1998) explicitly manipulated
length of planning time.
The studies also varied in terms of whether the planning was unguided or
guided. In the unguided condition, learners were simply given the task, told
they would have to perform it, and asked to prepare to do so. In guided plan-
ning, their attention was directed to specific aspects of planning. In Sangarun’s
(2005) study, three types of guided planning were manipulated—form-, mean-
ing-, and form-/meaning-focused planning. In Kawauchi’s (2005) study, three
types of planning activity were investigated: (i) writing (i.e. the learners were
asked to write out what they wanted to say); (ii) rehearsal (i.e. they were told
to say out aloud what they wanted to say); and (iii) reading (i.e. they were
provided with a model passage to read). Mochizuki and Ortega (2008) asked
their guided planners to focus on the use of a single grammatical structure
(English relative clauses). The issue in such studies is what students actually
do when they are asked to plan, but this has been rarely investigated.
As the main focus of this article is the three aspects of language use, I will
organize the synthesis in terms of how the above variables impact separately
on fluency, complexity, and accuracy.
1. Fluency: fluency was operationalized by means of two kinds of mea-
sures—those that examined the temporal aspects of fluency (e.g. number of
 R. ELLIS 493

syllables per minute) and those that examined repair phenomena (e.g. false
starts, repetitions, and reformulations). In addition, the testing studies
(Wigglesworth 1997, 2001; Elder and Iwashita 2005; Tavakoli and Skehan
2005) and one of the laboratory studies (Tajima 2003) obtained impressionistic

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

ratings of fluency.
A general finding of all but two of the studies (Guará-Tavares 2008;
Mochizuki and Ortega 2008),5 irrespective of whether the learners were
second or foreign, was that strategic planning has a positive effect on fluency.
However, one of the testing studies (Elder and Iwashita 2005) failed to find any
effect, while another (Wigglesworth 2001) actually found that planning had a
negative effect. It should be noted, though, that both of these studies took
place in a testing context and one of them (Wigglesworth 2001) only used
ratings of fluency (i.e. there were no discourse analytic measures). Also, one
study (Foster and Skehan 1996) found that strategic planning led to increased
repair work (i.e. resulted in less fluency) in one of the tasks they used. A fair
conclusion, however, is that strategic planning can assist learners to speak
more fluently and that this is evident in both the temporal and repair dimen-
sions of fluency.
This last conclusion holds good for both second- and foreign-language lear-
ners. However, in the case of another learner factor—proficiency—the effect of
planning is more variable. Most of the studies examined learners of roughly an
intermediate level of proficiency. However, Ortega (1999) investigated
advanced learners and found an effect for planning. Wigglesworth (1997)
reported that strategic planning had a greater effect on fluency in the higher
proficiency learners in her study. However, Kawauchi (2005), who studied
learners at three proficiency levels (low, high, and advanced) reported that
strategic planning assisted the fluency of low- and high-proficiency learners
but not the advanced learners, possibly because in this study the advanced
learners had achieved full control over their linguistic resources and thus did
not need the opportunity to plan to utilize them fluently. Tavokoli and Skehan
(2005) compared low- and intermediate-level learners and found that the
more proficient learners performed the tasks more fluently. However, this
study reported no interaction between planning and proficiency. Mochizuki
and Ortega (2008) found that planning had no effect on fluency in their
very low-proficiency learners. To sum up, while both second- and foreign-
language learners benefit from strategic planning, no conclusion regarding
the value of planning for learners with different levels of proficiency is possible
yet. Future planning, studies need to provide more explicit definitions of
proficiency.
A third learner variable is the learners’ attitude towards the opportunity
to plan. Tajima (2003) reported that those students in his study who held a
positive attitude towards the planning performed more fluently than the non-
planners, whereas those that held negative attitudes did not. Ortega (1999)
also found that 12 out of 32 learners she interviewed did not consider the
opportunity to plan a boon. Reasons given by those learners who did not
494 THE EFFECTS OF THREE TYPES OF TASK PLANNING

value the opportunity to plan were that the task was simple enough to make
planning unnecessary, planning did not help to fill gaps in L2 knowledge, they
were not allowed to keep their notes when performing the task, and planning
made them feel self-conscious and anxious. However, Ortega did not report

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

whether there was any difference in the effects of planning on the perfor-
mance of those who valued it and those who did not.
Guará-Tavares (2008) examined the role played by learners’ memory in
planned and unplanned conditions. She reported that working memory cor-
related significantly with the measures of fluency in the planning group but
not in the same task performed by the no-planning group. However, because
her test of working memory was administered in English (the learners’ L2),
its results should be interpreted cautiously.
The general finding regarding setting is clear. Planning aids fluency in both
classroom and laboratory settings. However, it appears to have less or no effect
in a testing context. Tavakoli and Skehan (2005) did find an effect for planning
on fluency but, as Ellis (2005b) pointed out, in many respects, their study
resembled a laboratory study (i.e. the learners may not have oriented towards
the tasks as a test).
Whether the effect of strategic planning is influenced by the participatory
structure of the tasks cannot be determined from these studies. Strategic plan-
ning assisted fluency in all of the studies involving learner interaction but in
only four out of the six studies involving monologic performance of the tasks.
However, the two monologic studies that reported no effect for planning on
fluency involved a testing context and also provided only a very short time for
planning.
Of greater interest is task complexity. This was explicitly addressed in the
early Foster and Skehan studies. These included three tasks: (i) a personal
information task, (ii) a narrative task, and (iii) a decision-making task.
Foster and Skehan argue that (i) is the easiest because the information is
familiar to the learners and is well-structured. Task (iii) is considered to be
the most difficult because the information to be communicated is not familiar
and is not clearly structured for the learners. The effect of the strategic plan-
ning appears to be clearer on the simpler tasks. Skehan and Foster (1997), for
example, reported that planners paused less than no-planners in (i) and (ii) but
not in (iii). Foster and Skehan (1996) found that the planners actually repaired
more than the non-planners in (iii) (i.e. became less fluent in this dimension).
Mehnert (1998) also found that planning had a greater overall effect in the
more structured of her two tasks. Tavakoli and Skehan (2005) found a clear
effect for the structure of their tasks, with learners producing more fluent
language on the structured tasks. However, there was no interaction between
planning and task structure. Gilabert (2007) reported that planning aids flu-
ency in both a simple Here-and-Now task and a complex There-and-Then task.
Clearly, task complexity does interact with planning; planning appears to have
a greater effect on fluency in the case of less complex tasks. Why this might be
so will be considered later.
 R. ELLIS 495

The length of the planning has been found to have a clear effect on
fluency. All but one of the studies that allowed 10 min for planning reported
significant effects on measures of fluency. Two of the testing studies
(Wigglesworth 2001; Elder and Iwashita 2005) that allowed a much shorter

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

time did not. However, Wigglesworth (1997), which allowed just 1 min,
and Tavakoli and Skehan (2005), which allowed 5 min, reported enhanced
fluency. Mehnert (1998) found a clear relationship between the length of
the planning time and fluency; the longer the planning, the greater the
fluency.
The type of planning also appears to have an effect. The key factor here is
whether the planning is guided or unguided. Foster and Skehan (1996) found
that guided planning led to greater fluency on their narrative task. Other
studies that provided guided planning (e.g. Wendel 1997; Kawauchi 2005;
Sangarun 2005) also reported positive effects on fluency. However, one
study (Mochizuki and Ortega 2008) found no advantage for guided planning
where fluency was concerned but this may have been because the learners
were instructed to focus on a single grammatical structure. It is also clear that
unguided planning can be effective. Foster and Skehan (1996) found that the
type of planning made no significant difference in the case of the personal and
decision-making tasks. Yuan and Ellis (2003) and Gilabert (2007) reported
clear effects on fluency for unguided planning. More work is needed to find
exactly when guided planning works better than unguided. One possibility is
that it may depend on the nature of the task. Another is that it may depend on
the proficiency level of the learners.
Two studies (Kawauchi 2005; Sangarun 2005) investigated different types of
guided planning. However, while these studies reported clear effects for plan-
ning on fluency, they did not show that one type of guided planning worked
better than another.
2. Complexity: complexity was measured primarily in terms of amount of
subordination (e.g. mean number of clauses per T-unit). Another measure of
grammatical complexity included the number of different verb forms used.
Lexical complexity was measured by means of type-token ratio and the
number of different word types.
Overall, the results for complexity are somewhat more mixed than those for
fluency, although there is plenty of evidence that strategic planning does lead
to more complex language. Thirteen studies reported a statistically significant
effect for planning on grammatical complexity. However, six studies (Mehnert
1998; Rutherford 2001; Tajima 2003; Elder and Iwashita 2005; Gilabert 2007;
Mochizuki and Ortega 2008) found no effect. Also, Skehan and Foster (1997)
found no effect on their narrative task. Four studies (Wendel 1997; Ortega
1999; Tajima 2003; Yuan and Ellis 2003) that investigated lexical complexity
failed to find that planning had any effect but Gilabert reported that planning
resulted in significantly greater lexical richness. Overall, strategic planning
appears to have a greater effect on grammatical complexity with only
Gilabert’s study showing otherwise.
496 THE EFFECTS OF THREE TYPES OF TASK PLANNING

Whether learners are learning the L2 as a second or foreign language does

not appear to interact with planning to influence complexity. Both types of
learner benefit. However, as was the case for fluency, Kawauchi’s (2005) study
suggests that strategic planning may not lead to greater complexity in

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

advanced learners. Guará-Tavares (2008) found that working memory was
significantly related to complexity in her planning group but not in her
no-planning group.
Whether the learners performed the task in a classroom setting or a labora-
tory setting also does not influence whether planning leads to greater com-
plexity. However, in a testing setting, planning appears to have less or no
effect.
Task factors interact with planning to influence complexity but the results
are not clear. Wendel’s (1997) study, where the narrative task was performed
with the researcher, found an effect but Rutherford’s (2001) language labora-
tory study, where the learners performed the task monologically, did not.
However, other studies that involved a monologic performance (Yuan and
Ellis 2003; Kawauchi 2005) reported that planning assisted complexity.
Regarding task complexity, Foster and Skehan (1996) found that strategic
planning led to greater grammatical richness in their personal information
and narrative tasks but not in the decision-making task. In contrast, Skehan
and Foster (1997) reported that planning resulted in greater grammatical com-
plexity in the personal and decision-making tasks but not in the narrative task.
However, in Gilabert’s (2007) study no results were reported to show that task
complexity influenced the effect that planning had on lexical richness. Clearly,
more research is needed to establish how task factors interact with planning to
influence complexity.
Regarding planning variables, there is no clear evidence that the length of
the time for planning affects complexity. Mehnert’s (1998) study—the only
one to investigate this variable—reported no effect for planning on complexity
in any of her time conditions. Nor is the type of planning influential. Neither
Sangarun (2005), Kawauchi (2005) or Mochizuki and Ortega (2008) found
any difference in the effect of the different types of planning they investigated.
However, there is some evidence that the degree of guidance provided
influences outcomes. Foster and Skehan (1996) found that detailed planning
resulted in greater grammatical complexity than undetailed, a finding that
they replicated in Skehan and Foster (2005). However, this was only true
for the first 5 min of the task performance, suggesting that, as the task
progresses, the effects of the guided planning reduce.
3. Accuracy: previous studies have produced somewhat mixed results
regarding the effects of strategic planning on grammatical accuracy. Ellis
(1987), for example, reported no effect, whereas Crookes (1989) did. These
mixed results are also reflected in the studies in Table 3. Thirteen of the studies
found that planning enhanced accuracy but six reported no effect.
One factor that does appear to influence the results for accuracy is the
learner’s proficiency. Kawauchi (2005) found that planning had much less
 R. ELLIS 497

effect with advanced level learners than with learners of low proficiency. For
example, she reported that only the low-proficiency learners improved on
regular past tense as a result of the opportunity to plan. This study suggests
the importance of controlling for learner proficiency when investigating the

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

effects of planning on accuracy. Learners’ attitudes towards planning also have
an effect on accuracy (Tajima 2003). Interestingly, working memory was not
related to accuracy in Guará-Tavares (2008) study.
Another potential influential factor is task type. Gilabert (2007) found no
interaction between task type and planning where accuracy was concerned.
The tasks in this study were narrative tasks. Foster and Skehan (1996) reported
that planning led to greater accuracy in their personal and decision-making
tasks but not in their narrative task. However, Skehan and Foster (1997)
reported that the effect was evident in the narrative task but not in the
other two tasks. Interestingly, Skehan and Foster’s (2005) study found that
the effect of planning on accuracy was not uniform throughout their decision-
making task (i.e. it was only apparent in the early stages of performing the
task). Mehnert (1998) reported an interaction between the kind of task and
planning time. In the case of her structured task, 10 min of planning time were
needed to produce an effect, whereas in the unstructured task 1 min sufficed.
This result seems counter-intuitive (i.e. one might have expected that with a
less-structured task more planning time would be needed). However, Mehnert
did not control for within-task planning, which as we will see later can influ-
ence accuracy. It is possible that those learners in Mehnert’s study who had
only 1-min strategic planning compensated by engaging in careful on-line
planning with the result that accuracy increased. To date, then, no conclusion
is really possible about how task type influences the effect that planning has on
accuracy. It is likely that it is not the task itself that induces a variable focus but
rather how the learners orientate to the task—an under-researched aspect of
task-based studies.
Most of the studies that investigated types of planning failed to find that this
variable had any effect on accuracy. An exception is Mochizuki and Ortega
(2008), who directed learners’ attention onto a specific grammatical structure
(relative clauses) in the guided planning condition. This resulted in the stud-
ents using relative clauses more frequently and also more accurately. This
study suggests that when the guided planning is focused it can have an
effect on accuracy.

General comments
This review has shown that, overall, strategic planning has a clear effect on
task performance but also that few conclusions regarding the role played by
specific learner and task variables are possible. These variables have not been
carefully controlled or systematically manipulated. However, I would argue
that such a review is valuable in that it has helped to identify the variables
that researchers have investigated to date.
498 THE EFFECTS OF THREE TYPES OF TASK PLANNING

One issue that remains unresolved is whether ratings based on descriptors of

the three dimensions that afford only ‘sketchy characterizations of perfor-
mance’ (Elder and Iwashita 2005: 232) are capable of detecting the effects of
strategic planning. None of the testing studies that used such ratings reported

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

any differences between the unplanned and planned performances. However,
the one laboratory study that included a holistic measure of fluency (Tajima)
did find that it distinguished the planners and the non-planners. Derwing et al.
(2004) examined the correlations between temporal measures of fluency and
ratings of fluency in low-proficiency learners of L2 English interacting one-on-
one with the researcher in a laboratory context and found a strong relationship
in unplanned speech (this study did not examine planned speech) between the
ratings and analytic measures. It would seem, then, that impressionistic ratings
are able to discriminate degrees of fluency in task-based performance but
maybe not in a testing context.
Overall, it would seem that strategic planning has less of an effect in a testing
context than in a teaching or laboratory context. Only one of the three testing
studies (Wigglesworth 1997) was able to show any differences in the discourse
analytic measures of fluency, complexity, and accuracy (I am excluding the
study by Tavakoli and Skehan, 2005, for the reason given above). In Ellis
(2005c) I speculated that the explanation for why planning is ineffective in
a testing context is because learners perceive that they are being assessed and
that this leads them to focus on accuracy at the expense of fluency and com-
plexity through careful on-line planning. In other words, the testing context
negates the advantages that normally accrue from strategic planning. Another
explanation is that it is the length of time that is the problem; if the testees had
been given a full 10 min to plan, effects might have been evident. However,
Mehnert (1998), in a laboratory study, reported effects for as little as 1 min of
planning so time may not be the reason.
Overall, strategic planning appears to have a more consistent effect on flu-
ency than on either complexity or accuracy. Only Guará-Tavares (2008)
Mochizuki and Ortega (2008) and the testing studies (Tavakoli and Skehan
2005, excluded) failed to find any effect on fluency. Presumably planning what
and how to say something enables learners to perform a task with few pauses
and with less need for repair. By accessing linguistic resources prior to the
performance of the task less effort is needed in accessing them during the
task. However, as Ortega (1999) pointed out, the demands of on-line produc-
tion may sometimes override the benefits accrued from strategic planning.
This may be why the effects of planning on complexity and accuracy are
more variable.
Skehan (1996) proposed that because learners have limited processing
capacity, they will need to prioritize and that this leads them to focus on
one aspect of performance at the expense of others. The ensuing trade-off
appears to affect complexity and accuracy. That is, if learners choose to take
risks, accuracy will suffer. But if they focus on controlling their resources to
ensure accuracy, then complexity suffers. This can explain why some studies
 R. ELLIS 499

found that strategic planning aided complexity but not accuracy and other
studies vice versa. Robinson’s (2003) Cognition Hypothesis, however, makes
a different claim, namely that a no planning condition will reduce attention
and memory resources with negative consequences for both complexity and

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

accuracy, a claim supported by the results of Gilabert’s (2007) study. Overall,
however, the results are more supportive of Skehan’s position. This can also
explain why Guará-Tavares (2008) found that working memory aided com-
plexity but not accuracy—presumably, learners used their limited processing
capacity to focus on producing more complex language in this study.
Clearly, other studies are needed to investigate the mediating role played by
such individual difference factors as working memory, language aptitude, will-
ingness to communicate, and risk-taking. These factors may interact with plan-
ning in different ways to influence the fluency, complexity, and accuracy of
learners’ production. We are told so little about the actual learners in many of
the studies (e.g. Gilabert’s study).
One learner difference variable that has been examined is language profi-
ciency. Kawauchi’s (2005) study is an important one because it suggests that
any benefit that might be derived from strategic planning is only evident in
learners of low or intermediate proficiency, not in advanced learners. In other
words, strategic planning is of value when learners still have difficulty in
accessing and controlling their linguistic resources. However, two studies
that claimed to investigate advanced learners (Wigglesworth 1997; Ortega
1999) reported that planning helped them. The problem here is that there is
no agreed measure of ‘advanced proficiency’.
It is worth noting again that just about all of the studies (including Tavokoli
and Skehan, 2005, which also manipulated proficiency in the design of the
study) investigated non-advanced learners. Ortega (1999) and Tajima (2003)
also showed that another difference factor that appears to be significant is the
attitudes that the learners hold towards the planning.
Interestingly, the effects of planning are evident whether the study is carried
out in a classroom context or a laboratory context. Only when learners per-
ceive they are being tested does strategic planning seem to have little or no
effect.
Task variables are almost certainly important but no clear conclusions can be
drawn yet. One task variable that has an impact on performance is the degree
of structure of the information in the task. Several studies investigated this
and showed that, in general, greater structure leads to greater fluency, com-
plexity, and accuracy. What is less clear is to what extent and in what ways
task structure interacts with planning. Skehan and Foster (1997) and Mehnert
(1998) both reported an interaction between these two variables, but Tavakoli
and Skehan (2005) and Gilabert (2007) found that although both variables had
an independent effect, there was no interaction.
Several studies have investigated whether the type of strategic planning has
any effect. The key difference appears to be whether the planning is guided or
unguided with Foster and Skehan (1996) and Mochizuki and Ortega (2008)
500 THE EFFECTS OF THREE TYPES OF TASK PLANNING

showing that this affects performance. Other ways of manipulating planning

(such as attempts to direct attention to form or to meaning), however, do not
appear to be effective. Ortega (2005) provides an explanation for this. In a
detailed analysis of the strategies adopted while learners were planning two

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

narrative tasks, she concluded that learners did not so much attend to form
and meaning separately but rather ‘engaged in solving form-in-meaning
problems’ (p. 106). It is possible, then, that learners make up their own
minds about how best they can plan and will not be unduly influenced by
the task-designer’s instructions unless these are very specific (as in Mochizuki
and Ortega 2008).

Within-task planning
A number of early studies (e.g. Hulstijn and Hulstijn 1984; Ellis 1987) com-
pared task performance under pressured and unpressured conditions but did
not specifically address the relative effects of these conditions on fluency,
complexity and accuracy.6 In fact, to date, only one study has attempted to
investigate this (Yuan and Ellis 2003). Other studies (e.g. Shehadeh 1999;
Kormos 2006) have examined specific performance phenomena (i.e. monitor-
ing and learner self-corrections) that are closely associated with on-line plan-
ning but did not seek to investigate whether the task conditions themselves
influenced these and so will not be considered here.
Yuan and Ellis (2003) compared university Chinese EFL learners performing
a task under a no-planning condition when they were pressured to complete
the task rapidly and a within-task planning condition when they were given as
much time as they wanted to complete the task. This study also had another
condition (strategic planning), as summarized in Table 3 above. The learners
in the no-planning condition spoke for a mean of 186 s, while those in the
within-task planning condition took notably longer (a mean of 243 s), sug-
gesting that the two groups performed the task in accordance with the instruc-
tions. The results were as follows. For fluency there was no statistically
significant difference between no-planning and within-task planning groups.
For complexity, within-task planning resulted in greater syntactical complexity
than no-planning but there was no difference in syntactical or lexical variety.
For accuracy, within-task planning resulted in more correct clauses and correct
verbs than no-planning. It is perhaps surprising that there was no difference in
fluency as one might have expected the within-task planning group to speak
more slowly than the no-planning group. Presumably the effort of speaking
within a time limit without any opportunity to plan prior to the task inhibited
the fluency of the no-planners. This study, then, suggests that when learners
are given ample time to perform a task they use it to formulate their utterances
more carefully, leading to increased complexity and accuracy.
In Yuan and Ellis’ (2003) study, within-task planning was operationalized
experimentally by allowing learners as much time as they needed to perform
the task (i.e. by removing any pressure they might feel about the need to
 R. ELLIS 501

complete the task rapidly and smoothly). Clearly, this can lead to a number of
different operations. Skehan and Foster (2005) suggest the following:
1. snatching time during on-line planning to plan what language to use;

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

2. buying time to reconceptualize the content of what will be said;
3. monitoring ongoing performance in order to self-correct; and
4. reacting to a communication problem that has arisen on-line.
Samuda and Bygate (2008) pointed out that we do not know what the learners
were actually doing while they performed the task—a point, however, that is
equally pertinent in the case of strategic planning. Ideally, we need measures
that will demonstrate that on-line planning is actually taking place. Skehan
and Foster (2005) suggest the following: false-starts, mid-clause pausing, filled
pauses, and reformulation. Their study (summarized in Table 3) compared
learners’ production in terms of these features during the first and second 5
min of performing a narrative task. Interestingly, dysfluency and repair
reduced over time, suggesting that as a task progresses learners need to
spend less effort on on-line planning. In other words, it is at the beginning
of a task that the opportunity for unpressured on-line planning is most
beneficial.

Summary
Of the three types of planning examined in this article, strategic planning has
received by far the most attention from researchers. There does not appear to
be a clear theoretical basis for this preference and it may simply reflect the
common sense assumption that strategic planning will assist fluency, complex-
ity, and accuracy in oral performance and the relative ease of designing studies
to investigate this type of planning. There is, however, clearly a case for more
thorough study of rehearsal and within-task planning.
It is not easy to reach clear conclusions about the effects of planning as there
has been no systematic investigation of key variables. The studies reviewed in
the previous sections suggest the following:
1. Rehearsal results in greater fluency and complexity (and to a lesser extent
accuracy), but these effects do not transfer to the performance of a new
task unless there is some kind of additional intervention, suggesting
that simply repeating a task may not have a measurable impact on
acquisition.
2. Strategic planning clearly benefits fluency but results are more mixed
where complexity and accuracy are concerned, possibly because there is
a trade-off in these two aspects (i.e. learners will tend to prioritize either
complexity or accuracy). A number of variables have been shown to have
an impact on the effect of strategic planning, in particular the learners’
proficiency (the effects are less evident in very advanced learners),
the degree of structure of the information in the task (planning is of
502 THE EFFECTS OF THREE TYPES OF TASK PLANNING

greater benefit with less well-structured tasks), and, in one study, work-
ing memory. However, different types of strategic planning seem to have
a negligible effect on production.
3. Within-task planning may benefit complexity and accuracy without

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

having a detrimental effect on fluency.
To date, no study has examined the joint effects of pre- and within-task
planning.

THEORETICAL PERSPECTIVES
These studies have been informed by a number of theoretical perspectives.
Some of the early studies drew on theories of variability (e.g. Tarone 1983)
to explain how different tasks affect learners’ grammatical accuracy, but the
later research, including many of the studies reviewed above, has drawn on
Levelt’s (1989) model of speaking. Planning, of course, is just one of several
variables that can affect L2 production, so ultimately a theory is needed that
will consider planning in relation to these other variables. Both Skehan (1998)
and Robinson (2003) have developed such a theory. However, in the discus-
sion that follows I have chosen to focus on Levelt’s model, partly for reasons of
space, but also because it seems to me to provide the clearest and fullest
account of how planning can influence production.
Levelt’s model distinguishes three overlapping processes: conceptualization,
formulation, and articulation. It also allows for speakers to ‘monitor’ both prior
to and after an utterance has been produced. Levelt also identified two char-
acteristics of speech production, which are relevant to task planning: (i) con-
trolled and automatic processing and (ii) incremental production. According to
Levelt, some of the components of the speech production process (specifically,
the conceptualizer and the monitor) operate under controlled processing,
while other components (specifically, the formulator and the articulator) oper-
ate largely automatically. However, what is true for native speakers may not be
true for learners, who are likely to experience problems with formulation and
articulation (De Bot 1992).
Levelt’s model, in conjunction with a view of the learner as a limited infor-
mation processor who experiences difficulty in attending to all aspects of lan-
guage production at the same time, provides a basis for explaining a number of
the key findings of the planning studies. Different types of planning can be
predicted to ease the pressure on the learner’s working memory in different
ways, variably affecting the competition and trade-offs evident in different
aspects of language production. Both rehearsal and strategic planning are
likely to assist conceptualization and thus facilitate fluency. They may also
have some impact on formulation and articulation as learners are likely
to have accessed relevant linguistic resources, while planning and will
find it easier to access them again during formulation and articulation.
However, if learners are primarily oriented towards fluency, strategic planning
 R. ELLIS 503

may not benefit either complexity or accuracy. Another possibility, if learners

are viewed as having a limited processing capacity, is that they will find it
difficult to attend to both complexity and accuracy and so will prioritize one
of these. This is what a number of the studies seem to show. While all of them

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

demonstrate a positive effect for strategic planning on fluency, several of them
found that it supported either complexity or accuracy but not both. One study
(Gilabert 2007), however, found that it benefited neither. It can also be
hypothesized that very advanced learners will function more like native
speakers. That is strategic planning will enhance fluency but have a much
more limited effect on complexity and accuracy as such learners have fewer
problems with formulation and articulation, as shown in Kawauchi’s (2005)
study. In contrast, unpressured within-task planning may prove beneficial to
formulation and also afford time for the controlled processing required for
monitoring. As a result, complexity and accuracy increase, as shown in
Yuan and Ellis (2003).
Such a general account of the effects of planning in terms of Levelt’s
production processes is useful and convincing but it is incomplete. What is
also needed is an explanation of the role played by individual difference
factors. If learners do engage in the same kinds of processes as a result
of opportunities to plan, they almost certainly do not all engage in them
to the same extent. Batstone (2005) convincingly argued that the way in
which learners orientate to a task will influence how they perform it. This
may explain why a testing context seems to produce different kinds of perfor-
mance from a laboratory or teaching context. It also explains why Tajima
(2003) found that only students with a positive attitude for planning benefited
from it. Other individual difference factors that are likely to influence the
impact of planning are working memory, language aptitude, willingness to
communicate, and anxiety. However, with the exception of Guará-Tavares’
(2008) study, to date there have been no studies that have examined how
these factors interact with planning to influence production.
Below is a framework depicting the various factors involved in investigating
the role that planning plays in L2 performance (Figure 1). There are four sets
of variables: (i) task variables (e.g. the complexity of the task design),
(ii) individual difference factors (e.g. L2 proficiency and learner attitudes),
(iii) planning variables, and (iv) L2 production variables. The model hypothe-
sizes that the task and individual difference variables influence how learners
plan and also mediate the effect that the planning has on L2 production. A full
theory, however, will need to explain how the factors shown in Figure 1
interact to influence the different aspects of L2 oral production and we are
clearly a long way from being to formulate such a theory, with Robinson’s
(2001, 2003) Cognition Hypothesis constituting the most complete to do so
to date.
A full theoretical account of the role of planning also needs to consider
whether and in what ways planning assists the development of fluency
and the acquisition of linguistic knowledge. As pointed out earlier, there has
504 THE EFFECTS OF THREE TYPES OF TASK PLANNING

Task design

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Planning L2 production

Individual
differences

Figure 1: A framework for investigating the role of planning in task-based

production

been no research that has investigated the long-term effects of planning in

task-based performance. The development of fluency and the acquisition
of linguistic knowledge are arguably separate phenomena. Schmidt (1992)
pointed out that ‘there is . . . little theoretical support from psychology for
the common belief that the development of fluency in a second language is
almost exclusively a matter of the increasingly skilful application of rules’
(p. 377). Instead, it depends on extending exemplar-based knowledge, as
claimed by Skehan (1998). Thus the development of fluency can take place
independently of the acquisition of linguistic resources, as illustrated in
Schmidt’s (1983) often cited case study and Towell’s (2002) longitudinal
study of 11 learners of L2 French.7 Rehearsal and strategic planning (but not
unpressured on-line planning) can be hypothesized to assist the development
of fluency because they help learners to improve access to their exemplar-
based knowledge and also to build stronger connections between exemplars.
Within-task planning may assist learners to automatize their grammatical
knowledge.
In order to theorize the relationship between planning and the acquisition of
linguistic resources, it is important to distinguish three senses of acquisition:
(i) the acquisition of new linguistic features, (ii) the restructuring of existing L2
knowledge, and (iii) the development of greater control (accuracy) over exist-
ing linguistic features. I hypothesize that planning has very little effect on (i),
although in interactive tasks it is possible that it will contribute to the kinds of
interaction that have been shown to foster the acquisition of new linguistic
forms (e.g. the negotiation of meaning and form). By and large, however, the
effects of planning are likely to be seen in restructuring and greater control.
Skehan (1998) has argued that planning can assist restructuring by the effect it
has on complexity. In this respect, all three kinds of planning can contribute.
Pre-task planning and, in particular, unpressured within-task planning can be
hypothesized to enhance increased control over existing resources through the
effect they have on complexity. Implicit in all these hypotheses is that more
complex production will lead to acquisition.
 R. ELLIS 505

CONCLUSION: LIMITATIONS AND FUTURE RESEARCH

There are some obvious limitations in the studies carried out to date. One of
the most serious is the lack of information about what learners actually do

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

while they plan. It cannot be taken for granted that learners do actually plan
(either strategically or on-line) when given time to do so, especially if they
do not see the planning time as useful (as was the case with some of
Tajima’s, 2003, and Ortega’s, 1999, learners). One or two studies have exam-
ined the planning strategies that learners adopt during planning but to date
only Ortega (1999) has attempted to relate these to their actual performance
of the task.
More attention needs to be paid to within-task planning. The strategic
planning studies have not controlled for within-task planning (i.e. we do
not know whether the learners had ample time for on-line planning or not).
Some of the differences in the results reported by these studies may
reflect differences in on-line planning. It would also be helpful to investigate
the joint effects of pre-task and within-task planning. No study has done this
to date.
Another major limitation is the lack of a longitudinal study of the effect of
strategic planning. This is obviously desirable to establish whether any benefit
gained from planning a task at one time carries over to a later time. The only
longitudinal studies were those that investigated rehearsal (Bygate’s, 2001,
study lasted 10 weeks). The fact that this was unable to show that the benefits
of task repetition extended to a new task should serve as a warning call to all
planning studies.
There are three other notable methodological problems. The first is that
none of the studies reviewed in this article collected baseline data of native
speakers performing the task. The only planning study that I am aware of
that did collect such baseline data is Foster’s (2001) study of formulaic
chunks.8 A related issue is the failure of many of the studies to provide
precise specifications of the proficiency of their learners. We need to know
what effects accrue from planning for learners of different proficiency levels
in comparison with native speakers. The third problem is the absence of
studies that have looked at learners’ extended performance of a task.
Researchers have tended to focus on just the language produced during
the early stage of task completion. Skehan and Foster (2005), however,
demonstrated that there are clear differences in the effect that planning
can have on the early and subsequent stages, with the effect less evident
in the later stage. Thus, it is possible that the impact of strategic planning has
been overestimated.
Finally, as noted earlier, it is important to examine how individual learner
factors such as learners’ attitudes towards planning, their working memory
capacity, and their degree of language anxiety affect how they rehearse a
task, engage in strategic planning, or carry out on-line planning.
506 THE EFFECTS OF THREE TYPES OF TASK PLANNING

NOTES
1 Complexity, as defined by Skehan and and this may have interfered with
Foster (1999), includes both gramma- conceptualization. Guará-Tavares

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

tical complexity/variety and lexical reports that the learners focused pre-
variety. However, in many of the stud- dominantly on formulation. In con-
ies reviewed in this article, complexity trast, Sangarun (2005) asked her
is narrowly defined as grammatical learners to plan aloud but in their L1
complexity, typically measured in and reports they focused mainly on
terms of amount of subordination. meaning. Mochizuki and Ortega
2 The finding that rehearsal leads to (2008) may have failed to find an
increase in lexical richness in Gass effect for fluency because the learners
et al. (1999) stands in contrast to the they investigated were of a very low
finding that strategic planning was level of proficiency. Planning of any
found to have no effect on lexical kind may have little effect on fluency
variety/richness in a number of studies with such learners.
(but see Gilabert 2007). 6 Hulstijn (1989), using the same data as
3 Ortega’s (1999) study was extended for the 1984 paper, did examine the
in Ortega (2005). The later study effects of time pressure, along with
included learners of two pro- focus of attention (form vs. meaning)
ficiency levels—low-intermediate and on fluency, finding that both had a
advanced. The main difference in the significant effect on response duration
effects of strategic planning on task
and speech rate but not on the
performance of the two groups
number of self-corrections.
was that it led to increased lexical
7 In a longitudinal study, Towell (2002)
complexity in the case of the low-
undertook a qualitative analysis of one
intermediate learners and more accu-
low- and one high-performer of L2
rate noun-modifier accuracy in the
French. This showed that the low-per-
case of the advanced learners.
former increased her phonation/time
4 Robinson (2001, 2003) views planning
ratio and length-of-run by pausing
as a factor that influences task com-
less, especially at syntactic boundaries
plexity (he refers to it as a ‘resource
dispersing factor’). I have adopted a rather than within them. The high-
different position in this article. Task performer, however, still outper-
complexity is considered solely in formed the low-performer on length-
terms of task design variables. Planning of-run as a result of her greater ability
is viewed as a task implementation to produce subordinating structures.
variable. Thus the model of task- Thus, whereas the two learners grew
based performance that underlies this closer in terms of fluency, they
review is one where task design and remained distinct in terms of
task implementation variables jointly complexity.
influence task performance (not task 8 Foster (2001) compared native-speak-
complexity). This is made explicit in ers’ and L2 learners’ use of ‘lexicalized
Figure 1 later in the article. sequences’ in planned and unplanned
5 Guará-Tavares (2008) failed to find conditions. She found that whereas
that planning had any effect on flu- the native speakers produced fewer
ency. The explanation for this seems such sequences in the planned condi-
to lie in the fact that the learners tion, there was no change in the
were asked to plan aloud in English learners.
 R. ELLIS 507

REFERENCES
Batstone, R. 2005. ‘Planning as discourse in M. Bygate et al. (ed.): Researching Pedagogic
activity’ in R. Ellis (ed.): Planning and Task Tasks, Second Language Learning, Teaching and

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Performance in a Second Language. John Testing. Longman.
Benjamins. Foster, P. and P. Skehan. 1996. ‘The influence
Bygate, M. 1996. ‘Effects of task repetition: of planning on performance in task-based
Appraising the developing language of lear- learning,’ Studies in Second Language Acquisition
ners’ in J. Willis and D. Willis (eds): Challenge 18/3: 299–324.
and Change in Language Teaching. Heinemann. Gass, S., A. Mackey, M. Fernandez and
Bygate, M. 2001. ‘Effects of task repetition on M. Alvarez-Torres. 1999. ‘The effects of
the structure and control of oral language’ task repetition on linguistic output,’ Language
in M. Bygate, P. Skehan, and M. Swain Learning 49: 549–80.
(eds): Researching Pedagogic Tasks, Second Gilabert, R. 2007. ‘The simultaneous manipula-
Language Learning, Teaching and Testing. tion of task complexity along planning time
Longman. and (+/ Here-and-Now): Effects on L2 oral
M. Bygate, P. Skehan and M. Swain (eds). 2001. production’ in M. Garcia-Mayo (ed.):
Researching Pedagogic Tasks, Second Language Investigating Tasks in Formal Language Learning.
Learning, Teaching and Testing. Longman. Multilingual Matters, pp. 44–68.
Crookes, G. 1989. ‘Planning and interlanguage Guará-Tavares, M. G. 2008. Pre-task Planning,
variation,’ Studies in Second Language Acquisition Working Memory Capacity and L2 Speech
11: 367–83. Performance. Unpublished Doctoral Thesis,
De Bot, K. 1992. ‘A bilingual production model: Universidade Federal de Santa Catarina, Brazil.
Levelt’s ‘‘Speaking’’ model adapted,’ Applied Hulstijn, J. 1989. ‘A cognitive view of interlan-
Linguistics 13: 1–24. guage variability’ in M. Eisenstein (ed.): The
Derwing, T., M. Rossiter, M. Munro and Dynamic Interlanguage: Empirical Studies in
R. Thomson. 2004. ‘Second language fluency: Second Language Variation. Plenum Press.
Judgments on different tasks,’ Language Hulstijn, J. and W. Hulstijn. 1984.
Learning 54: 655–79. ‘Grammatical errors as a function of processing
Elder, C. and N. Iwashita. 2005. ‘Planning for constraints and explicit knowledge,’ Language
test performance: Does it make a difference?’ Learning 34: 23–43.
in R. Ellis (ed.): Planning and Task Performance Kawauchi, C. 2005. ‘The effects of strategic
in a Second Language. John Benjamins. planning on the oral narratives of learners
Ellis, R. 1987. ‘Interlanguage variability in nar- with low and high intermediate proficiency’
rative discourse: Style shifting in the use of in R. Ellis (ed.): Planning and Task-Performance
the past tense,’ Studies in Second Language in a Second Language. John Benjamins.
Acquisition 9: 1–20. Kormos, J. 2006. Speech Production and Second
Ellis, R. 2005a. Planning and Task Performance in a Language Acquisition. Lawrence Erlbaum.
Second Language. John Benjamins. Levelt, W. 1989. Speaking: From Intention to
Ellis, R. 2005b. ‘Planning and task-based Articulation. The MIT Press.
research: theory and research’ in R. Ellis Lynch, T. and J. Maclean. 2000. ‘Exploring the
(ed.): Planning and Task-Performance in a benefits of task repetition and recyling for
Second Language. John Benjamins. classroom language learning,’ Language
Ellis, R. 2005c. ‘Planning in language testing’ Teaching Research 4: 221–50.
in R. Ellis (ed.): Planning and Task-Performance Lynch, T. and J. Maclean. 2001. ‘Effects of
in a Second Language. John Benjamins. immediate task repetition on learners’ perfor-
Foster, P. 1996. ‘Doing the task better: How mance’ in M. Bygate, P. Skehan, and M. Swain
planning time influences students’ perfor- (eds): Researching Pedagogic Tasks, Second
mance’ in J. Willis and D. Willis (eds): Language Learning, Teaching and Testing.
Challenge and Change in Language Teaching. Longman.
Heinemann. Mehnert, U. 1998. ‘The effects of different
Foster, P. 2001. ‘Rules and routines: A consid- lengths of time for planning on second lan-
eration of their role in the task-based language guage performance,’ Studies in Second Language
production of native and non-native speakers’ Acquisition 20: 52–83.
508 THE EFFECTS OF THREE TYPES OF TASK PLANNING

Michel, M., F. Kuiken and I. Vedder. 2001. Schmidt, R. 1983. ‘Interaction, acculturation
‘The influence of complexity in monologic and the acquisition of communication compe-
and dialogic tasks in Dutch L2,’ IRAL 45: tence’ in M. Wolfson and E. Judd (eds):
241–60. Sociolinguistics and Second Language Acquisition.

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Mochizuki, N. and L. Ortega. 2008. ‘Balancing Newbury House.
communication and grammar in beginning- Schmidt, R. 1992. ‘Psychological mechanisms
level foreign language classrooms: A study of underlying second language fluency,’ Studies
guided planning and relativization,’ Language in Second Language Acquisition 14: 357–85.
Teaching Research 12: 11–37. Shehadeh, A. 1999. ‘Non-native speakers’ pro-
Nemeth, N. and J. Kormos. 2001. ‘Pragmatic duction of modified comprehensible output
aspects of task-performance: The case of argu- and second language learning,’ Language
mentation,’ Language Teaching Research 5: Learning 49: 627–75.
213–40. Sheppard, C. 2006. The Effects of Instruction
Norris, J. and L. Ortega. 2006a. ‘The value and Directed at the Gaps Second Language Learners
practice of research synthesis for language Noticed in their Oral Production. Unpublished
learning and teaching’ in J. Norris and PhD Thesis, University of Auckland.
L. Ortega (eds): Synthesizing Research on Skehan, P. 1996. ‘A framework for the imple-
Language Learning and Teaching. John mentation of task-based instruction,’ Applied
Benjamins. Linguistics 17/1: 38–62.
J. Norris and L. Ortega (eds). 2006b. Skehan, P. 1998. A Cognitive Approach to
Synthesizing Research on Language Teaching and Language Learning. Oxford University Press.
Learning. John Benjamins. Skehan, P. and P. Foster. 1997. ‘Task type and
Ortega, L. 1999. ‘Planning and focus on form in task processing conditions as influences on for-
L2 oral performance,’ Studies in Second Language eign language performance,’ Language Teaching
Acquisition 21: 109–48. Research 1: 185–211.
Ortega, L. 2005. ‘What do learners plan? Skehan, P. and P. Foster. 1999. ‘The influence
Learner-driven attention to form during pre- of task structure and processing conditions on
task planning’ in R. Ellis (ed.): Planning and narrative retellings,’ Language Learning 49/1:
Task Performance in a Second Language. John 93–120.
Benjamins. Skehan, P. and P. Foster. 2005. ‘Strategic and
Robinson, P. 2001. ‘Task complexity, cognitive on-line planning: The influence of surprise
resources, and syllabus design: A triadic frame- information and task time on second language
work for examining task influences on SLA’ performance’ in R. Ellis (ed.): Planning and
in P. Robinson (ed.): Cognition and Second Task-Performance in a Second Language. John
Language Instruction. Cambridge University Benjamins.
Press. Tajima, M. 2003. The Effects of Planning on
Robinson, P. 2003. ‘The Cognition Hypothesis, Oral Performance of Japanese as a Foreign
task design, and adult task-based language Language. Unpublished Dissertation, Purdue
learning,’ Second Language Studies 21: 45–105. University.
Robinson, P. 2007. ‘Criteria for classifying and Tarone, E. 1983. ‘On the variability of interlan-
sequencing pedagogic tasks‘ in M. Garcia- guage systems,’ Applied Linguistics 4: 143–63.
Mayo (ed.): Investigating Tasks in Formal Tavakoli, P. and S. Skehan. 2005. ‘Strategic
Language Learning. Multilingual Matters. planning, task structure, and performance
Rutherford, K. 2001. An Investigation of the testing’ in R. Ellis (ed.): Planning and Task-
Effects of Planning on Oral Production in a Performance in a Second Language. John
Second Language. Unpublished MA Thesis, Benjamins.
University of Auckland. Towell, R. 2002. ‘Relative degrees of fluency: A
Samuda, V. and M. Bygate. 2008. Tasks in comparative case study of advanced learners of
Second Language Learning. Palgrave MacMillan. French,’ International Review of Applied
Sangarun, J. 2005. ‘The effects of focusing Linguistics 40: 117–50.
on meaning and form in strategic planning’ Wendel, J. 1997. Planning and Second Language
in R. Ellis (ed.): Planning and Task Performance Narrative Production. Unpublished Doctoral
in a Second Language. John Benjamins. Dissertation, Temple University, Japan.
 R. ELLIS 509

Wigglesworth, G. 1997. ‘An investigation (eds): Researching Pedagogic Tasks, Second

of planning time and proficiency level on Language Learning, Teaching and Testing.
oral test discourse,’ Language Testing 14/1: Longman.
21–44. Yuan, F. and R. Ellis. 2003. ‘The effects of

Downloaded from https://academic.oup.com/applij/article/30/4/474/225516 by Glasgow University Library user on 15 May 2023

Wigglesworth, G. 2001. ‘Influences on pre-task and on-line planning on fluency,
performance in task based oral assessments’ complexity and accuracy in L2 monologic
in M. Bygate, P. Skehan, and M. Swain oral production,’ Applied Linguistics 24/1: 1–27.