Interaction and Second Language

Learning: Two Adolescent French

Immersion Students Working Together
MERRILL SWAIN SHARON LAPKIN
Ontario Institutefor Studies in Education Ontario Institutefor Studies in Education
University of Toronto University of Toronto
252 Bloor St. West 252 Bloor St. West
Toronto, Ontario Toronto, Ontario
M5S 1 V6 Canada M5S 1 V6 Canada
Email: main@oise.utoronto.ca Email: slupkin@oise.utoronto.ca

This article provides support for a theoretical orientation toward viewing dialogue as both a
means of communication and a cognitive tool. Data to support this position come from an
analysis of the language-related episodes isolated in the dialogue of two grade 8 French im-
mersion students as they carry out ajigsaw task. During the task, the students work out a story
line and write it out. As they do so, they encounter linguistic problems. To solve them, the stu-
dents use their first language (Ll) and second language (L2) in order to communicate to
each other and as tools to aid their L2 learning. The language-related episodes discussed pro-
vide evidence of language use as both an enactment of mental processes and as an occasion
for L2 learning. Variation in how other pairs of students in the class perform the task supports
existing evidence that the same task does not provide similar occasions for L2 learning to all
student dyads.

THEORETICAL PERSPECTIVE they work linguistically to achieve the needed

comprehensibility, whether repeating a message
We take the position in this article that lan- verbatim, adjusting its syntax, changing its
guage use is both communication and cognitive words, or modifying its form and meaning in a
activity. Language is simultaneously a means of host of other ways” (Pica, 1994, p. 494). The hy-
communication and a tool for thinking. Dialogue pothesis underlying this perspective is that the ac-
provides both the occasion for language learning tivity of negotiation leads to second language
and the evidence for it. Language is both process (L2) learning because it provides learners with
and product. comprehensible input (e.g., Krashen, 1985; Long,
When language use is considered as communi- 1983).Recent research such as that conducted by
cation, the concepts of input, comprehensible Mackey (1995) and Mackey and Philp (this issue)
input, and comprehensible output are appropri- provides supportive evidence for this view. Yet we
ate metaphors because they conjure up images of are still left with the issue of how comprehensible
messages. These messages are transmitted as out- input leads to L2 learning: What are the mecha-
put from one source and received as input else- nisms by which comprehensible input is con-
where. When there are difficulties in encoding verted into L2 knowledge and use?
or decoding these messages, language users mod- A complementary perspective is that language
ify and restructure their interaction to achieve serves not only a communicative function, but is,
message comprehensibility. “As they negotiate, itself, a psychological tool. Like any tool, it facili-
tates task performance by mediating between us
and the accomplishment of the task. The tool
The Modern LanguageJournal, 82, iii, (1998) may facilitate our performance of the task and
0026-7902/98/320-337 $1.50/0 may make some things possible that were not oth-
01998 The Moden LanguageJournal
erwise. It may qualitatively change the nature of
Merrill Swain and Sharon Lapkin 321
the activity and it may change the subsequent ratively produce a story may be considered di-
outcome. One only needs to think of the com- rectly revealing of mental processes. In the first
puter as a tool, and the profound ways in which it section of our literature review we therefore con-
is transforming our everyday activities, as for ex- sider a study (Goss, Ying-Hua, & Lantolf, 1994)
ample, in writing (e.g., Cummins, 1990). whose purpose was to demonstrate that the men-
In this article, we wish to explore this comple- tal processes used to solve a linguistic problem are
mentary perspective. Our exploration takes the manifested in dialogue. The present study p r e
form of examining the dialogue that occurs be- vides additional evidence for this position.
tween two learners as they attempt to solve the A further implication of the consideration of
linguistic problems they face while writing a language as a mediating tool, as Donato (1994)
short narrative. By taking the perspective that the has argued, is that “The focus [in SLA] should
students are using language as a psychological be . . . on observing the construction of co-
tool, we will need to examine their dialogue for knowledge and how this co-construction process
evidence of language being used as a tool in aid results in linguistic change among and within in-
of L2 learning (see also Platt & Brooks, 1994). dividuals during joint activity”’ (p. 39). That is,
That is, we will examine the data for examples of in joint activity, language serves to coconstruct
students’ use of language that mediates their knowledge. This knowledge can be inferred from
learning-for example, the use of language to the changes observed in linguistic performance.
generate and test hypotheses. This is still consid- In the present study, one of our goals is to try to
ered “output” (Swain, 1995), but it is output used trace the linguistic change that occurs as learners
for a cognitive function. It is speaking as a cogni- engage in the sort of collaborative dialogue to
tive activity, instantiated in dialogue. which Donato (1994) refers. We wish to suggest
The expectation that cognitive activity will be that collaborative dialogue provides the occasion
apparent in dialogue is supported by the work of for L2 learning (see Swain, 1997). Unlike the
Vygotsky and other more recent sociocultural claim that comprehensible input leads to learn-
theorists (e.g., Lantolf & Appel, 1994; Newman, ing, we wish to suggest that what occurs in col-
Griffin, & Cole, 1989; Wertsch, 1991),who argue laborative dialogues is learning. That is, learn-
that cognitive processes arise from the interac- ing does not happen outside performance; it oc-
tion that occurs between individuals. Language curs in performance. Furthermore, learning is
becomes a mediating tool by its first having been cumulative, emergent, and ongoing, sometimes
used by others in order to regulate behavior, in- occurring in leaps, while at other times it is im-
cluding cognitive behavior. Through a gradual perceptible.
process of internalization, one comes to be able Therefore, in the second part of the literature
to use the language of others (and the mental review, we consider two studies (Donato, 1994;
processes that interaction has constructed) to LaPierre, 1994) that offer evidence that the c e
regulate one’s own cognitive functioning. As construction of linguistic knowledge in dialogue
Leont’ev (1981) states: is language learning in progress. Those studies
suggest that the use of either the first language
Higher psychological processes unique to humans ( L l ) or the L2 as a mediational tool creates new
can be acquired only through interactionwith others, language or new knowledge about language and
that is, through interpsychologicalprocesses that only
consolidates existing knowledge (proceduraliza-
later will begin to be carried out independently by
the individual. When this happens, some of these tion) (see de Bot, 1996; Nobuyoshi & Ellis, 1993).
processes lose their initial, external form and are con-
verted into intrapsychologicalprocesses. (p. 56) RELATED RESEARCH

In a joint problem-solving activity,what normally As indicated above, there are two aspects of
remains hidden in individuallyinternalized thought previous research related to the present study
may manifest itself in dialogue. This theoretical that we wish to consider: (a) dialogue as an en-
claim is discussed in Donato and Lantolf (1990), actment of mental processes, and (b) dialogue as
who suggest that cognitive processes, because they occasions for L2 learning. We will also refer to
are derived and constituted dialogically, “. . . can several studies conducted in functioning class-
be observed directly in the linguistic interactions rooms as corroborative evidence to our own of
that arise among speakers as they participate in the range of performances present in typical
problem-solving tasks” (p. 85). One of the pur- classroom activities.
poses of this article is to examine to what extent
the conversation of two students as they collabo-
322 The Modern LanguageJournal 82 (1998)
Dialogue as an Enactment of Mental Processes edge. The students relied on memory (language
mediated) and “feel” (an indication of relying
Goss et al. (1994) conducted a study to investi- on abstract linguistic principles, unmediated by
gate the validity of claims about what grammati- language).
cal judgments represent: Do “learners access This study (Goss et al., 1994) is important to
abstract grammatical knowledge or do they rely our work for its conclusion that the dialogue that
on some other knowledge source, such as mem- arises during collaborative problem-solving is an
ory of what someone (e.g., a language teacher) enactment of cognitive activity. The study s u p
has told them, memory of what they think some- ports our use of collaborative tasks and our in-
one has told them, their own folk knowledge, or terpretation of language-related episodes.
L1 equivalents, when carrying out such tasks?”
(p. 263). What is important to our argument here Dialogue as Occasionsfm L2 Learning
is Goss et al.’s methodology and their rationale
for using it. Rather than having L2 learners in- Two studies have looked specifically at the lan-
trospect about how they reached ajudgment, the guage learning evident in collaborative dialogue.
researchers asked learners to decide jointly on Donato (1994) studied what he referred to as “col-
the grammaticality of sentences. They rejected lective scaffolding.”LaPierre (1994) examined the
the use of think-aloud tasks because of their de- occasions for L2 learning in peer interaction-in
mand on learners to both solve a mental problem this case, interaction about the language that stu-
(cognitive level) and to report on the solving of dent dyads were producing. One of the goals of
the problem (metacognitive level) simultane- Donato’s (1994) study was to reveal how L2 learn-
ously. Under these circumstances, one of the ing is brought about on the social plane. In par-
processes (solving or reporting) is likely to break ticular, the study sought to:
down (Vygotsky, 1979). However, by using a col-
laborative procedure, learners use language as answer the question of whether learners can exert
they would normally, as a mediating tool: “. . . they a developmental influence on each other’s inter-
language system in observable ways. That is, rather
have a single goal, the solving of a problem, [and]
than to theorize that interaction has the potential to
they solve it through dialogic interaction with an- result in L2 development, this study attempts to ex-
other person” (Goss et al., p. 267). amine how social interactions in the classroom result
In addition, because of the social origin of cog- in the appropriation of linguistic knowledge by the
nition and cognitive processes, Goss et al. (1994) individual. (p.39)
note that “. . . the talk spontaneously generated
by individuals in collaborative problem-solving The students involved in the study were third se-
situations offers a window into intramental pro- mester students of French at an American uni-
cessing” (p. 166).In other words, the window into versity. The data analyzed consisted of a 1-hour
intramental processing is more transparent than session in which three students planned for an
the window provided through introspective tech- oral activity (the presentation of a skit) that
niques. With dyadic problem solving, the window would take place the following week. The stu-
is also potentially larger because it is likely to gen- dents had been told that they could not use notes
erate more talk. in their presentation, nor were they to memorize
In the Goss et al. (1994) study, elementary- and their lines, but they could make notes while
advanced-level students of Spanish were given a preparing if they wished. Donato examined the
grammaticality judgment task to perform either transcripts for examples of scaffolding, defined
individually or with a partner. Response patterns as a situation where, “in social interaction a
were similar across those who performed the task knowledgeable participant can create, by means
individually or jointly within the elementary of speech, supportive conditions in which the
group and within the advanced group. (In addi- novice can participate, and extend current skills
tion, error rates were slightly higher for those and knowledge to higher levels of competence”
who performed the task individually.) Goss et al. (p. 40). In all, 32 cases of scaffolded help were
concluded, therefore, that similar mental pro- identified in the hour-long planning session.
cesses are at work in joint activity and in individ- A key question here is whether this collective
ual activity when individuals from the same p o p scaffolding offered occasions for linguistic devel-
ulation respond to the same task. Their study opment in the individual learner. That is, could
suggests that language mediated students’judg- linguistic development be traced back to the col-
ments as revealed in their making use of trans- lective scaffolding episodes? To determine this,
lations and making explicit their metaknowl- evidence for independent L2 performance was
Merrill Swain and Sharon Lapkin 323
sought in the actual oral activity performed the results suggest rather forcefully that these lan-
following week. Of the 32 cases of collective scaf- guage related episodes, where students reflected
folding observed in the planning session, 75% of consciously on the language that they were pro-
the language structures involved in the scaffold- ducing, were the occasion for L2 learning.
ing were used correctly the following week. Thus,
through collaborative dialogue of this sort, learn- Clasmoom-BasedResearch
ers added to their own L2 knowledge and ex-
tended that of their peers. Learners provided for Two studies exist that are of particular rele-
each other the support needed to outperform vance to our work because they, too, were con-
their competence and, in the process, develop ducted in functioning classrooms and were fo-
their interlanguage (IL). Donato (1994) points cused on input and output: Foster (1993) and
out that his results are not surprising “in light of Jacob, Rottenberg, Patrick, and Wheeler (1996).
Vygotskyan theory which argues that individual These studies support the view that conducting
knowledge is socially and dialogically derived, research in an ongoing classroom context may
the genesis of which can be observed directly in provide a different perspective on the implemen-
the interactions among speakers during problem- tation of theoretical principles than research con-
solving tasks” (p. 51). ducted in a laboratory-type setting. If one goal of
The LaPierre study (1994;Swain, 1998) involved the research is to understand better the relation-
grade 8 early French immersion students and ship between theory and practice, then research
served as a pilot study to the one reported on, in conducted in a classroom setting will likely allow
part, in this article. In LaPierre’s study, it was hy- for a more realistic assessment of L2 learning
pothesized that when L2 learners engage in a task than that conducted in a laboratory-type setting.
in which they need to talk about the language Foster (1993) investigated (a) the amount and
they are producing (metatalk) in order to com- distribution of language produced by dyads and
plete the task, that metatalk may be a source of small groups working on language tasks, (b) the
L2 learning. The task the students engaged in degree to which these students negotiated for
was a story reconstruction task (dictogloss; see comprehensible input, and (c) the extent to which
Wajnryb, 1990). they modified their language in order to make it
Second language learning was tested by means comprehensible to others. Her study is different
of tailor-made dyad-specific posttests. Language- from most of those reported in the research lit-
related episodes were isolated from the tran- erature because it was conducted in her own
scripts of the students’ talk as they attempted to classroom (with part-time intermediate-leveladult
solve linguistic problems that they encountered English as a second language [ESL] learners).
while jointly reconstructing the passage. On the The study was designed to preserve the setting of
basis of these episodes, items were developed to an ongoing class during which students partici-
test the language discussed. Thus, every pair of pated in group tasks that were part of their sched-
students had a set of test items that reflected uled syllabus. With the exception of a grammar-
specifically what they had discussed in recon- based (optional information exchange) task, the
structing the passage. These tests were adminis- tasks were communicative meaning-based (op-
tered approximately 1 week after the students tional and required information exchange) tasks.
had completed the task. Of the tapes of the students’ interaction that
The results of LaPierre’s (1994) study show were not too noisy for transcription and where
that when students solved a linguistic problem the students correctly attempted the task, a p
that they had encountered, the solution corre- proximately 5 minutes of interaction per group
sponded to their responses 1 week later. More were transcribed and coded. Foster found that,
specifically, of the 140 episodes where, through overall, dyads with an obligation to exchange in-
collaborative dialogue, a correct solution was at- formation were most likely to talk and to negoti-
tained, approximately 80% of the relevant post- ate meaning. However, as she points out:
test items were correct. Furthermore, and equally
as telling, when students co-constructed an in- when the individual scores are taken into account, it
is clear that. . . many students preferred to contribute
correct solution (21 such episodes), approxi-
little to the interaction; only a few attempted to nego-
mately 70% of the answers on the posttest were tiate for comprehensible input; and even fewer pro-
wrong, although they matched the solutions pro- duced any modified output. The range in the indi-
vided by the pairs. In other words, the students vidual scores is so wide, and the lack of participation
tended to retain the knowledge that they had con- by some students is so striking as to make statistics
structed collaboratively the previous week. These based on group totals very misleading. (p. 25)
324 The Modern LanguageJournal 82 (1998)
Investigating another “everyday classroom,” THE STUDY
Jacob et al. (1996) reached a similar conclusion.
Desagn
The classroom that they observed was a sixth-
grade social studies one. They were interested in The data to be analyzed in this article were col-
exploring the extent to which a particular form lected in the context of a larger study involving
of cooperative learning (Johnson, Johnson, & four grade 8 French immersion classes, each un-
Holubec, 1986) provided the L2 learners in the dergoing a different treatment. The class fo-
class opportunities for learning academic Eng- cused on here was given ajigsaw task. In this task,
lish. Their study shows that a wide range of o p student dyads received a set of numbered pic-
portunities was available, including (a) the provi- tures (each member of the dyad got half the pic-
sion of assistance in decoding academic terms tures) that told a story. The students were to work
and instructions, (b) the clarification of pronun- out the story together and then write it out. Prior
ciation and meaning, (c) help with the conven- to doing the task, the class was given a short mini-
tions of written English, and (d) invitations to lesson (5 minutes) on French reflexive verbs.
contribute to the ongoing task. Jacob et al. con-
clude that TimeFrame and Activities
Cooperative learning gave L2 learners a wide range of
opportunities to acquire academic English . . . which The time frame and activities of the study are
included both input and output opportunities, with shown in Table 1. In Week 1, a pretest (described
L2 learners helping others as much as they were below) was given. In Week 2, a session was held to
helped. However, except for help with decoding aca- familiarize the students with the jigsaw task. The
demic terms, the various kinds of opportunities oc- students were first given a minilesson to focus
curred relatively infrequently. Moreover, there were their attention on the agreement of adjectives (in
some missed opportunities. . . (p. 253) this case, all colors) with nouns. Following this,
In the present study, we, too, observed consider- the students worked in pairs with a series of pic-
able variation from student pair to student pair tures, generating the story told by the pictures
in how they approached the task and how they and then writing it as a jointly constructed story.
carried it out. This is shown in the classroom- In the third week, a prerecorded minilesson
level data provided. about French reflexive verbs (5 minutes) was pre-
In our current research, our theoretical ori- sented on video. The video also showed two stu-
entation is towards understanding why collabora- dents working together to reconstruct a story
tive tasks might promote L2 learning. Our re- from a series of pictures (5 minutes); their inter-
search is concerned with whether these tasks in action was intended to serve as a model for what
fact do promote learning. Our practical interest the students were to do when they received their
is in the feasibility of implementing such proce- pictures. The modeling included dialogue about
dures in an ongoing classroom of active French linguistic form and grammatical rules. Next, the
immersion students. pictures (provided in Appendix A) were distrib-

TABLE 1
Research Time Frame and Events
Week 1 Week 2 Week 3 Week 4 Week 5

Pretests Informal Videotaped Tapes Posttests (oral

developed from instructor-led lesson and transcribed and and written)
pilot study were training session instructions: class-specific administered
administered Modeling of posttests
task developed
performance

Task done in Task done in

pairs pairs and tape-
recorded

(Focus on (Focus on
adjective reflexive
agreement) verbs)
Mendl Swain and Sharon Lupkin 325
uted so that one student in each pair received are few materials for the teaching of grammar
pictures numbered 1, 3, 5, and 7 and the other that have been produced expressly for the im-
student received pictures 2, 4, 6, and 8. Taking mersion classroom, so L1 teaching materials and
turns, the students first generated the story orally teacher-made activities are frequently used. As
and then wrote it out. As the students worked in Kowal and Swain (1997) note, “It is highly likely
pairs, their conversations were tape-recorded. that students entering the intermediate grades
This was a large class of 35 students and all (grades 7-9) will have been exposed to an eclec-
recording had to be done simultaneously in the tic language-teaching approach consisting of
classroom (no other space was available in this learner-centred activities fortified with a regular
overcrowded school). Even with our best (we dose of traditional, prescriptive grammar activi-
think heroic) efforts to tape-record under these ties” (p. 288).
conditions, the data from five pairs of students In this article, we examine in depth the lan-
were lost due to loud background noise and stu- guage-related episodes (for a definition, see the
dents playing with the tape recorders. section on ”Data Analysis and Results”) occur-
In the fourth week, the tapes were transcribed. ring in the conversations of one pair of students
Based on the content of the taped oral interac- as they carry out the jigsaw task based on the pic-
tions, additional test items were developed to be tures in Appendix A. From those pairs for which
included in the posttest. In the fifth week, the we had a complete data set, we chose a pair of stu-
posttest, which included all pretest items and the dents whose written story was much better than
new posttest items, was administered. average, yet where a difference in proficiency lev-
els suggested that a possible “expert/novice” re-
Participants lationship might exist. Listening to the taped in-
teraction, however, suggests that neither student
The students were grade 8 students who had dominated during their pair work and that both
been in an early French immersion program contributed in important ways to the collabora-
since kindergarten. All of their initial instruc- tive activity.
tion, through grade 3, had been in French. After The information used to select the student pair
that, instruction in English was introduced and (pretest scores and their teacher’s ratings of over-
by grade 8, approximately 50% of their instruc- all ability in French) is given in Table 2. Table 2
tion was in English, 50% in French. During the also shows the average score and rating on these
time spent in French, the teaching of academic measures for the students in the class for whom
content in French, along with French language we also have usable recordings, as well as the
arts, was the instructional priority. ranges observed within the class.
In spite of the overall experiential nature of L2 The classroom teacher had rated each student
learning in immersion, formal grammar instruc- in his class on a 7-point scale, with 7 representing
tion occurs in French immersion classrooms. the highest proficiency level. For the pair that we
The focus in grammar lessons appears to be on selected (given the pseudonyms Kim and Rick),
isolated rules, paradigms, and the manipulation the teacher’s ratings of overall ability in French
of form rather than on relating form to function indicated that Kim was stronger than Rick; the
(Allen, Swain, Harley, & Cummins, 1990). There pretest scores reflect the teacher’sjudgment. In

TABLE 2
Test Scores and Teacher Ratings for Kim and Rick and Their Class

Note. a The pre-and posttest scores cannot be compared directly because the posttest includes additional
class-specific items. 7-point scale with 7 representing the highest proficiency level.
326 The M o h LanguageJournal 82 (1998)
fact, Kim obtained the highest pre- and posttest 10 days for a total of 39 pairs across all classes),
scores in the class and was one of two students to and because the identification of language-re-
be given a rating of 7 by her teacher. Rick is rated lated episodes in the conversations of these L2
as average with respect to the rest of the class in speakers turned out to be a complex and time-
overall ability in French. consuming task, only the clearest and most obvi-
ous examples were incorporated into items for
Pre- and Posttests the posttest.

We conducted a pilot study with the set of pic- DATA ANALYSIS AND RESULTS
tures shown in Appendix A with a different class
of grade 8 immersion students than those involved Language-Related Episodes
in the main study. Based on the transcribed tape-
recorded interactions of the students in the pilot Although the tapes of each pair of students had
classroom, and on the assumption that the con- been transcribed immediately following their
tent of some of these interactionswould be similar recording, a second transcriber later checked
between the pilot students and the main study stu- each tape at a more leisurely pace. Because the
dents, a pretest was constructed for use in the pair considered in this article consisted of a male
main study. The three item types used are illus- and a female student, we were able to identify the
trated in Figure 1. speakers and relate each one’s part in the dia-
The first item type (Type A) was used because logue to test performance.
discussion about the gender of nouns occurred Below, we examine Kim and Rick’s dialogue in
quite frequently in the students’ conversations. terms of language-related episodes (LREs) in an
Type A items require students to choose the mas- attempt to understand their conversation as L2
culine or feminine form of the indefinite article learning. A LRE episode is defined as any part of
(un/une) or to indicate that they do not know (Je a dialogue where the students talk about the lan-
ne sazs pas). guage they are producing, question their lan-
The second item type (Type B) was intended to guage use, or correct themselves or others (Swain
capture movement along a continuum of not & Lapkin, 1995).*
knowing something or not being certain of some- For all the student pairs, LREs were classified
thing to greater certainty. Thus, Type B items in- as either “lexis-based”or “form-based.”The lexis-
volve a “certaintyscale”in which students evaluate based LREs involved students seeking French vo-
the grammaticality of a given sentence by indicat- cabulary or choosing among competing French
ing that it is definitely wrong (certainement incor- vocabulary items. The form-based LREs involved
rect), probably wrong (probabkmmt incmrect),prob students focusing on spelling or an aspect of
ably correct (p-obablement c m e d ) , or definitely French morphology, syntax, or discourse, usually
correct (certainementcmd).Students also had the in the context of writing out their story rather
option of indicating that they do not know (Je ne than in the initial telling of it. The interrater re-
saispas). Aspects of many of the sentencesjudged liability obtained in the identification and cate-
in this way had, in one way or another, been a gorization of LREs for Kim and Rick was loo%,
point offocus for some students in the pilot study. following considerable discussion about the iden-
The third item type (Type C) was initially con- tification and categorization of the LREs of all
structed in order to allow for the measurement of the student dyads.
aspects of vocabulary knowledge. Type C items The results are discussed in the following
are multiple choice in format (4 choices). Many order: First, in order to place Kim and Rick in
of the distractors had appeared in the conversa- context and to indicate the variation in task per-
tions of the students in the pilot study. In the item formance amongst student dyads, we provide
depicted in Figure 1, students must recognize general information about what the other stu-
l’oreilleras the correct word for “pillow.” dent dyads in this class did on the same task. Sec-
As mentioned above, based on the content of ond, we examine in detail selected LREs from
the conversations of the student pairs that were excerpts of Kim and Rick’s conversation.
tape-recorded in the third week of the study, ad-
ditional test items were developed and added to Variationin Task Performance
the pretest items in order to form the posttest ad-
ministered in the fifth week. Because the tran- Table 3 presents information about aspects of
scriptions and new item development were done the performance of student pairs ( n = 12). The
under considerable time pressure (approximately stories written collaboratively by each pair of stu-
Merrill Swain and Sharon Lupkin 327
FIGURE 1
Examples of Test Item Types
Type A. Pour chaque mot francais ci-dessous, choisissez la forme correcte de I'article indifini (un, une) et
cochez la case approprike. Si vous ne savez pas, cochez la case Je ne sais pas P droite.
(For each French word below, choose the correct form of the indefinite article [a, an] and mark
the appropriate box. If you do not know, mark the box I do not know at right.)

I un I une I mot I Jenesaispas

I
I I I couverture I I
gant

chandail
I
I I I
I
table

Type B. Pour chaque phrase ci-dessous, indiquez si la phrase est correcte ou incorrecte selon I'image. In-
diquezjusqu'P quel point vous &tescertain de votre reponse en cochant la case appropriie.
Dans chaque groupe il y a au moins une phrase correcte, mais il est aussi possible d'avoir plusieurs
phrases qui sont correctes dans chaque groupe.
(For each sentence below, indicate whether the sentence is correct or incorrect according to the
picture. Indicate how certain you are of your answer by marking the appropriate box.
In each group there is at least one correct sentence, but it is also possible to have several correct
sentences in each group.)

1. correct correct
1. Le clown sort de la boite.
12. Le clown se sort de la boite. I I

Type C. Choisissez la meilkre rt;ponsedanschaque groupe. Cochez la case approprike P droite de la phrase.
(Choose the best answer in each group. Mark the appropriate box to the right of each sentence.)

1. Voili mon horloge. [I

2. Voila mon riveille-matin. []
3. Voila mon r&ve-matin. [I
4. VoilP ma cloche. [I A

Note. English translations (in parentheses) not supplied to students.

328 T h M o r i a LanguageJournal 82 (I998)
dents were rated by two French immersion teach- As a further indication of variation among
ers. They rated the stories on a 5-point scale with pairs, one pair spent approximately 17 minutes
1 representing very poor performance and 5 r e p on task and generated 2 LREs, whereas another
resenting excellent performance on five dimen- pair spent approximately 4 minutes on task and
sions: content, organization, vocabulary, mor- generated 5 LREs. In spite of this variation, we
phology, and syntax. In addition, the number of found a correlation between time on task and
idea units (maximum of 21) was counted. Table 3 number of LREs of .78 ( p = .002).
shows the ratings given to these aspects of Kim It is interesting to note that, although the cor-
and Rick’s story, the range of ratings for the stu- relation between pretest scores (total for each
dent pairs in their class, and the average ratings. pair) and the number of LREs was not significant
As can be seen, Kim and Rick’s story was highly (.41; p = .lo), the correlation between posttest
rated. It is provided as it was written, with a trans- scores (total for each pair) and the number of
lation, in Appendix B. LREs was significant (.62; p = .04). This suggests
The average time students spent on task was ap- that, quantitatively at least, the number of LREs
proximately 10.2 minutes (SD = 6.9). Kim and Rick and the posttest scores are positively related. The
spent the most amount of time on task: 23 min- qualitative analysis of individual LREs provided
utes. The least amount of time spent on task was in the next section suggests that the LREs may
3.5 minutes. The average number of LREs was 8.8 have positively influenced the posttest scores.
(SD = 8.0), with a range of 26 to 1. Kim and Rick
produced 23 LREs. The average number of form- Kim and Rick’s LREs
based LREs was 4.8 (SD= 4.5), with a range of 15
to 1. Kim and Rick produced 15. The average In the qualitative analyses that follow, a selec-
number of lexis-based LREs was 4.0 (SD = 3.7), tion of Kim and Rick’s LREs are discussed in
with a range of 12 to 0. Kim and Rick produced 8. terms of the three strands of research as they

Range Average

Highest Lowest

Ratings of Story Writtena

Content 5b 5 1 2.9

I Organization I 4.5 I 4.5 I 1 I 3.1 I

Vocabulary 5 5 1 3.1

Morphology 5 5 1 2.9

1 Syntax
Number of Idea Units
(Max. = 21)
1 5

16
1 5 1 1 1 2

I . 8

12.5 1
1

I Time on Task (in Minutes) I 23 I 23 I 3.5 I 10.2 I

#LREsc 23 26 1 8.8

#Form-Based LREs 15 15 1 4.8

#Lexis-Based LREs 8 12 0 4.0

Menill Swain and Sharon Lapkin 329
appear in the literature section of this article: (Alarm clock?)
(a) dialogue as an enactment of mental processes, 57 Kim: R h e i l k a t i n .
(b) dialogue as occasions for L2 learning, and (c) (Alarm clock.)
LREs and their implications for classroom-based Turn 66 Rick: Se riveilk a cause. . . du son. . .
research. (Wakes up because . . . of the
sound. . .)
Dialogue as an Enactment of Mental Processes. Sev-
67 Kim: Rheille-matin.
eral excerpts from Kim and Rick‘s transcribed
(Alarm clock.)
interaction will be examined as instantiations of
68 Rick: A cause du. . .
the following mental processes that mediate L2
(Because o f . . .)
learning: (a) generating alternatives (or hypoth-
69 Kim: Du rheille-matin qui sonne? Does
esis generation), (b) assessing alternatives (or hy-
that sound OK?
pothesis testing), and (c) applying rules or ex-
(Of the alarm-clock that rings?
tending knowledge to new L2 contexts (Swain &
Does that sound OK?)
Lapkin, 1995).We will also point to other ways in
70 Rick: Or what about . . .Jacqueline se h e
which Kim and Rick‘s language use mediates
a cause du. . . du rheille-. , . yeah,
their learning; that is, ways in which their lan-
qui sonne.
guage use serves as a tool supporting L2 learning
(Or what about . . . Jacqueline
by consciously singling out the L2 as an object to
[the girl in their story] gets up be-
be monitored, reflected upon, and manipulated.
cause of the. . . of the alarm-. . .
ExcerptA contains parts of two LREs. The first
yeah, that rings.)
relates to the use of rheille-matin: Turns 2-4, 9,
71Kim: OK. Or you can say du rheille-
55-72, and 92-95 were all considered as part of a
matin or du sonnement du rheille-
single LRE. Turns 68-72 and 92-93 also consti-
matin.
tute part of a second LRE focusing on k sonnement.
(OK. Or you can say of the alarm
Excerpt A: clock or the ring of the alarm
Turn 2 Kim: On p a t pas &terminer qu ’est-ce que clock.)
c’est. 72 Rick: No, rheille-matin pi sonne.
(One can’t figure out what it is.) (No, alarm clock that rings.)
3 Rick: Rheille-matin. Turn 92 Rick: Sur la rh-. . .The-matin.
(Alarm clock.) (On the alarm clock.)
4 Kim: Et il y a un rheilbmatin rouge. . . 93 Kim: Sur le rheille-matin pour arriter k
SUT une tabk brune, et k r h e i k sonnement.
matin dit six heures, et c’est tout. (On the alarm clock to stop the
(And there is a red alarm clock ring.)
. . . on a brown table, and the 94 Rick: Rive-matin?
alarm clock says six o’clock, and (Alarm clock?)
that’s all.) 95 Kim: REVEILLE-matin.
(Alarm clock.) [Stresses compe
This exchange continues for another four turns, nent meaning “wake.”]
with Kim using “rheille-matin” three more times
and Rick once more; Rick then, for some reason,
We do not know why Rick sometimes used “rhe-
switches to “ h / krhe-matin”3 in Turn 9. His hesi- matin” after he, himself, initially suggested using
tation in producing it suggests his uncertainty.
“rheille-matin“ to Kim (Turn 3). However, it is
Turn 9 Rick: Elk est en train de dormir u p i s que clear from the dialogue that Rick is uncertain as
. . . la rhe-matin est encore sonnb Et to which is the correct vocabulary item. This is in-
le r&. . . rhe-matin dit six hares un. dicated by the pauses prior to, or even during,
(She is sleeping after the alarm the use of “rhe-matin” (Turns 9 and 92) and
clock rang again. And the alarm “rheille-matin” (Turn 70); by his need for reassur-
clock says one minute after six ance before writing “rheilk-matin“ (Turn 56);
o’clock.) and finally by overtly asking if “rhe-matin”is OK
(Turn 94) and getting immediate feedback from
This uncertainty continues:
Kim that it should be “REVEILLE-matin.” In writ-
Turn 55 Kim: . . . il y a un rheille-matin. ing the story, Rick correctly uses “rheille-matin”
(. . . there is an alarm clock.) three times, although it is misspelled each time as
56 Rick: Rheilbmatin? “rheil-matin.”
330 The Moa2r-n LanguageJournal 82 ( I 998)
There was a relevant multiple-choice pre- and prefers his own rendition (Turn 72) and writes it
posttest item. Students saw a picture of an alarm down. This, too, is an example of collaborative
clock and were asked to choose the best response dialogue, where Kim and Rick stretch their IL in
from: the generation of new vocabulary and alternative
subordinate structures.
1. Vuilri mon hmloge.
Excerpt B is interesting because Kim proposes
2. VoiE mon rheilbmatin.
“gurCon” (boy) and Rick suggests a refinement,
3. Voilri m m rtvematin.
“gars,”which represents a more informal register
4. Vuilri ma cloche.
that is certainly appropriate in this context.
As a pretest item, Kim correctly chose rheilb
Excerpt B:
mutin and Rick chose rtvematin. In the posttest,
both students chose the correct response.
Kim: . . . eUe vuit un garcon p i . . .
(. . . she sees a boy who . . .)
In this collaborative dialogue, we are able to
Rick: unGARS.
observe change in Rick’s use of the correct term
(a GUY) [emphasis on “guy”l
for “alarm clock.” It is not a one-time shift from
Kim: OK, un gum
wrong to right, but a wavering between alterna-
tives. The source of his learning is not only input, (OK, a guy.)
although Kim used “rheilbmatin” 17 times dur- In this LRE (Excerpt B), although no metalin-
ing their entire conversation. Nor was the source guistic terminology is used explicitly, Rick and
of his learning only output, although it may have Kim are nevertheless talking about register varia-
been Rick‘s attempt to write it (Turn 56) that fo- tion. This LRE illustrates how a task of this sort
cused his attention on his own uncertainty about draws on the students’ understanding of the rela-
which term to use. We wish to argue that it is the tionships among meaning, form, and function in
joint construction of knowledge that resulted context. Excerpt C relates to the meaning and
from Rick‘s questioning and Kim’s responses syntax of the verb suiwe (to follow):
that, in part, accounts for Rick’s shift from in-
Excerpt C:
correct to correct usage. Here Rick’s questions
Rick: EUe se. . . et elk se. . . how do you say fol-
serve as hypotheses and Kim’s responses serve to
low?
confirm or disconfirm them.
(She [reflexive pronoun] . . . and she
The sonnement LRE (in Turns 66-72 and con-
[reflexive pronoun] . . . how do you say
tinued in Turns 92 and 93) is particularly inter-
follow?)
esting because le sonnemat is not a word in
Kim: Hmmm?
French (le son or la sonnerie are the relevant words
Rick: How do you say follow?
here). Although the word sonnement does not
Kim: Suit.
exist in French, Kim, in creating this word, a p
(Follows.)
plies a productive morpho-phonological rule in
Rick: Suit. Elk se suit or elle suit ?
French ( - m a t is a suffix which marks many mas-
(Follows. She follows [reflexive form]
culine nouns). Elsewhere in the transcript, Rick
or she follows? [nonreflexive form])
questions whether it is “la smnement” or “le son-
Kim: EUe se. . , eUe LE suive.
nement,” and Kim immediately assures him that it
(She [reflexive pronoun] . . . she follows
is “LEsonnemat.” Here we see Kim and Rick a p
HIM.)
plying rules to new contexts, albeit incorrectly.
Rick: EUe le?
They solve a lexical problem in much the same
(She [follows] him?)
way as native speakers might coin a new word, by
Kim: EUe LE suive.
using their existing language knowledge as a tool
(She follows HIM.)
to create new knowledge.
Rick: Jusqu ’ci 17cole.
In addition, Turns 69-72 provide clear exam-
(To school.)
ples of language used as a tool to regulate Kim’s
and Rick’s cognitive activity. Here Kim and Rick In asking which of two alternative forms to use
use their L1 to help them to consider what they ( “Elle se suit” or “Elle suit”),Rick is consciously
are trying to express in their L2 by setting up drawing attention to this linguistic structure and
their L2 as an object to be reflected upon and articulating two different hypotheses, from which
manipulated. Kim asks “Does that sound OK?” Kim chooses the correct one, supplying the cor-
(Turn 69). Rick responds with “Or what rect pronoun object as she does so. Kim’s intona-
about . . .?”(Turn 70). Kim replies ”OK. Or you tional emphasis on the direct object pronoun le
can say ‘X’ or ‘Y”’ (Turn 71). Rick, however, may help to make the syntactic frame of suiwe
M m d l Swain and Sharon Lapkin 331
more salient for Rick. Rick writes the syntactic (And . . . and, yeah, and goes to the
structure correctly and on the relevant posttest bathroom.)
items (see Table 4), although not certain, cor- Rick: Salk de bain. . , pour se la-
rectly marks Le chat se suit (The cat follows itself) (Bathroom. . . to wa-)
as probably wrong and Le chat ks suit (The cat fol- Kim: OU elk se lave.
lows them) and Le chat suit ks chiens (The cat fol- (WHERE she gets washed.)
lows the dogs) as definitely correct.
Rick chooses to write Kim’s solution. Each of
What Kim does in this LRE is a particularly in-
these grammatical alternatives also serves a d i s
teresting feat from a processing perspective. She
course function, allowing Kim and Rick to se-
seems to know that suiure is a transitive verb and
quence the elements of their narrative.
therefore takes a direct object pronoun rather
Excerpt E relates to the personal care verb se
than a reflexive pronoun. She thus switches from
brosser (to brush). French has many verbs (comb-
using the reflexive pronoun to a direct object
ing hair, brushing teeth, washing one’s face, etc.)
pronoun, k, emphasizing it for Rick‘s benefit and
that must be expressed by the reflexive form of
perhaps also for her own benefit. In doing so,
the relevant verb.
however, Kim “loses control” over the correct
form of the verb. This would appear to be an ex- Excerpt E:
cellent example of cognitive overload: processing Rick: . . . et brosse.
capacity is limited and in carrying out this com- (. . . and brushes.)
plicated processing operation, Kim is unable also Kim: Et SE brosse ks dents. . . ks ch-. No, wait a
to process the verb form correctly, falling back on second. Isn’t it elk se brosse b dents?And
the form of the high frequency -er verb type it’s E p e i p . Elk se p e i p .
rather than using the correct present tense of (And brushes [emphasizes the reflex-
suiure, which she had just produced.4 ive] her teeth . . . her hair. No, wait a sec-
In Excerpt D, two alternatives are generated ond. Isn’t it she brushes her teeth? And
and assessed. Each of the subordinate clauses it’s combs [again emphasizes the reflex-
that the students generate and assess (“pour se ive]. She combs her hair.)
laver“ [to wash herself] and “mi elk se lave” [where
In this excerpt, as in Excerpt A, we again see how
she gets washed]) is acceptable and accurate in
language is mediating task performance. It is so
this context.
clear here because of Kim’s use of English. Kim
Excerpt D: tells herself (not Rick, although it may have an ef-
Kim: Et. . . et, yeah, et s’en va au sulk de bain. fect on Rick) to “wait a second.” This act of self-

TABLE 4
Kim and Rick’s Performance on Certainty Scale Items Relating to the Reflexive Construction
Posttest Item Correct Response Actual Response
Kim Rick

re: Excerpt C

Le chat les suit. definitely correct definitely correct definitely correct

Le chat se suit. definitely wrong definitely wrong probably wrong
Le chat suit les chienr. definitely correct definitely correct definitely correct
re: Excerpt E

Je me coupe les on@. * definitely correct definitely correct probably correct

(same on pretest) (same on pretest)
Elk se lave k h a p . definitely correct definitely correct definitely correct
(same on pretest) (same on pretest)
Jacques se p e i p les cheueux. definitely correct definitely correct definitely correct
Je me coupe mes on@. definitely wrong definitely wrong probably correct
re: Excerpt G

Le clmun sort de la h-te. definitely correct definitely correct definitely correct

Le clmun se sort de la h-te, definitely wrong definitely wrong probably correct
Note. *This is an analogue forJe me brosse les cheoewc/les dents, in that reflexive verbs relating to personal care b e
have syntactically in the same way in French. Therefore this and the other items in this set relate to Excerpt E.
332 The M o h LanguageJournal 82 (1998)
regulation gives her time to work out the correct (Is that the pillow?)
form. English (her L1) frames her French and Kim: No, this is l’meiller.
sets it up as an object to be manipulated and re- (No, this is the pillow.)
flected upon. Her “Isn’t it elle se brosse les dents? Rick: Pillow?
And it’s SE peigne” is not language being used to Kim: Yeah, pillow’s meiller.
communicate with Rick; it is language being
There was no pretest item, although the above di-
used to hypothesize and to confirm.
alogue would suggest that Kim would have got-
In their story, Rick has correctly written these
ten the item correct and Rick might not have. On
personal care verbs in their reflexive form. On
the posttest item (see Item C in Figure l),both
the four relevant posttest items (see Table 4),
students choose the correct response. The word
Kim is both correct and confident of her re-
oreiller is also used correctly in their written text
sponses. Rick, however, is correct and certain of
(see Appendix B).
his correctness on only two of the four items,
Excerpt G focuses on the verb smtzr, which does
those using the very same verbs (se laver and se
not exist in the reflexive form in French.
peigner) that had been the focus of their attention
at some point in their conversation, and that Rick Excerpt G
had written correctly in their narrative. With the Rick: Un bras. . . wait. . . m’canique. . . sort?
sentences ‘ye me coupe les/ms 0n@s,”5 which in- (An arm. . .wait. . . a mechanical [arm]
troduced a new verb, Rick was less certain. Per- comes out?)
haps his posttest responses indicate a tentative Kim: Smt, yeah.
extension of knowledge to new L2 contexts. (Comes out, yeah.)
Dialogue as Occasions for L2 learning. We have al- Rick: Sesort?
ready cited LREs for which there were both pre- (Comes out?) [incorrect reflexive
and posttest items (rheille-matin and analogues form]
for se brosser) and for which only posttest items ex- Kim: No,sort.
isted (suiweandother reflexive verbs). In the for- (No, comes out.) [correct nonreflexive
mer cases, it will be recalled, there is evidence of form]
learning on Rick’s part, because he moves from As Rick produces the first utterance in this LRE,
an incorrect to a correct response from pre- to he orders himself, and possibly Kim, to ”wait,”
posttest.6 In this section, we examine other ex- giving himself time to work out what follows. His
amples where learning has occurred. The first of rising intonation suggests that once again he is
these is found in Excerpt E testing a hypothesis. As shown in Table 4,both
In Excerpt F, Rick falters as he realizes that he Kim and Rick judge correctly ‘‘LA clown smt de la
does not know the word for “pillow:” bo2e” (The clown comes out of the box) as defi-
Excerpt F: nitely correct (see Test Item B in Figure 1). Kim
Rick: Et elle est encore au. . . au. . .uh . . . a l’autre correctly marks the related item “Le clown se sort
bout du lit avec, avec ses pieds sur le. . . sur de la boite” as definitely wrong. Rick, however,
la. . . how do you say “pillow”? marks it as probably correct. We consider it note-
(And she is already at the other end of worthy that he did not mark it as definitely cor-
the bed with, with her feet on the. . . on rect. In other words, the choice of the response as
the . . . how do you say “pillow”?) “probable” rather than “definite” suggests that
Kim: Oreiller. Rick has moved toward recognizing smtir as
(Pillow.) grammatical and se sortir as ungrammatical as a
Rick: Avec ses pieds sur l’oreiller. result of the collaborative reconstruction of this
(With her feet on the pillow.) part of the narrative.
Later, Kim uses the word “oreil&” which gives LREs and Their Implicationsfor Chsroom Practice.
Rick the opportunity to check his comprehen- One of the LREs in particular gives rise to an im-
sion and, as he is writing, to write it down: portant pedagogical issue related to collabora-
tive work. In Excerpt H, Rick suggests s’en aller
Kim: Quelque chose uh . . . est sur 1’. . . quelque and Kim proposes marcher to express the notion
chose est sur l’oreiller. of walking to school:
(Something . . . is on the . . . something
is on the pillow.) Excerpt H:
Rick: Is that E’oreiller? (pointing to something Kim: [Elk voit un] gars
in the picture.) (She sees a guy)
Ma’U Swain and Sharon Lapkin 333
Rick: . . .gars, pi s’en va a l’icob constructed knowledge, as we saw in the son-
(guy, who is going to school) nemat LRE, where a productive rule of word for-
Kim: qui march vers l’icole. . .march mation in French was applied and its gender in-
(who is walking towards school, walking) ferred from a characteristic suffix. Kim and Rick
also use their L1, a mediational tool fully avail-
Both verbs exist in the French lexicon, but in this
context, it is nonnative-like to use marcher.7 Per- able to them, to regulate their own behavior, to
focus attention on specific L2 structures, and to
haps because marcher is the alternative that Kim
generate and assess alternatives.
(the usual “expert”in this pair) suggests, and be-
Kim and Rick‘s interaction supports the value
cause it conforms to their L1 usage, it is the al-
and unique role of collaborative activity in the
ternative they agree on and write in their story.
classroom as articulated by Wells (1996):
Here is an example where the solution that the
pair reaches is wrong even though a correct al- . . .it is not necessary for there to be a group member
ternative has been suggested. who is in all respects more capable than the others.
Without teacher feedback, both students will . . . in tackling a difficult task as a group, although no
either remain uncertain about the functional dis- member has expertise beyond his or her peers, the
group as a whole, by working at the problem together,
tinction between marcherand s’en a l h o r theywill
is able to construct a solution that none could have
have “learned” the wrong lexeme, given the con- achieved alone. In other words, each is “forced to rise
text. For this reason, we believe that the sort of above himself“ and, by building on the contributions
task used in our data collection must be followed of individual members, the group collectively con-
by opportunities for teacher feedback on the structs an outcome that no single member envisaged
recorded oral dialogue or the written product, at the outset of the collaboration. (p. 10)
or both.
From a research perspective, this study provides
empirical data to suggest that collaborative dia-
CONCLUSIONS AND DISCUSSION logue (consisting of one or more LREs) is a use-
ful concept for understanding L2 learning. In
The data offered in this article provide support terms of methodology, we think it would be valu-
for a theoretical orientation toward viewing dia- able in future work to combine an analysis of stu-
logue as both a means of communication and a dents’ collaborative dialogue with follow-upinter-
cognitive tool. In the LREs discussed here, we see views in order to derive a more fine-grained under-
Kim and Rick, as theyjointly develop the story line standing of the mental processes discussed above.
and write it out, using language to co-construct In the data analyzed here, we were able to link
the language they need to express the meaning each student to his or her test performance be-
they want and to coconstruct knowledge about cause of the gender difference in the student
language. Their dialogue serves as a tool both for dyad and because it was clear which student was
L2 learning and for communicating with each writing. In the majority of cases, it was impossible
other. In Kim and Rick’s language use, we can to identify the individual students in their dyads,
see simultaneously the process of language learn- so that test data could not be interpreted in rela-
ing and the communicative outcome. tion to individual contributions to dialogues. It
Like the students in the Goss et al. (1994) will be important to incorporate technical im-
study, Kim and Rick use language as they would provements in future studies.
normally and we are able to observe evidence of Kim and Rick work effectively in their dyad,
cognitive processes instantiated in their talk. and both are strong students. From a pedagogi-
They continually generate alternatives, assess al- cal perspective, one wonders if their relative suc-
ternatives, and apply the resulting knowledge to cess would extend to student dyads with different
solve a linguistic problem. Often Rick generates proficiency levels. Detailed analyses of such stu-
alternatives (e.g., “le gars” in Excerpt B, the hesi- dent dyads (e.g., two low-proficiency students)
tation over reflexive/nonref lexive forms), and are needed to help teachers make principled
Kim helps him to assess these. In many cases, she decisions about how to group students effec-
provides correct information to Rick about the tively. Affective variables also appear to be key:
linguistic form or rule he has brought to their On several occasions, Rick looks to Kim for s u p
joint attention. Although Kim generally takes the port (e.g., the rheille-matin example) as well as
lead, Rick makes equally important contribu- for information.
tions (it is he, for example, who suggests the use One of the striking features of our data is their
of “le gars” in Excerpt B). Applying their knowl- variability. Like Foster (1993), we found some
edge to new contexts may also create new jointly pairs of students whose low production of LREs
334 The Modern LanguageJournal 82 ( I 998)
and limited time on task suggest that students a p matin,” means alarm clock, and we have used that
proach the task differently and will profit differ- meaning in the translations. A number of people who
entially from the collaborative activities imple- commented on a first draft of this article suggested
mented in classrooms.In future studies, it will be that Rick may have been influenced by the relevant
pretest item in which one of the distractors (derived
important to interview students shortly after
from the pilot data) was Voild m m rhe-matan.This is cer-
such activities have been completed to discover,
tainly a possibility and does not detract from our argu-
for example, what aspects individual students ment. Indeed, Rick may have taken away from the
find appealing or unappealing,conducive or un- pretest the knowledge that &ematin is a possible way to
conducive to learning. Also, it may be that col- say “alarm clock.” The fact that our readers were keen
laborative tasks such as the one used in this study on this explanation for Rick’s use of rhematzn suggests
should be implemented selectively or only with that, implicitly, they have accepted the possibility that
several dyads at a time, so that the teacher can even one single test item of many may lead to change in
monitor on-task behavior and provide linguistic a learner’s linguistic performance. This parallels our
guidance. We hope that the kind of information claim that learning occurs in dialogue. As we wish to
presented in this article will better equip teachers argue, at least some learning occurs in performance,
not afterwards.
to provide the sustained intervention needed to
There are three verb conjugations in French: first
foster continuing target language development. conjugation -er verbs such as donnq second conjuga-
tion -ir verbs such as finit; and third conjugation -re
verbs such as suivre. Of the three verb conjugations, the
ACKNOWLEDGEMENTS first is the most frequent and is often generalized to
verbs of the second and third conjugations (Harley &
Swain, 1978.)
This research was made possible by a grant (#410- The second of these ( j e me c a p e mes ongtes) is un-
93-0050) to Merrill Swain and Sharon Lapkin from the grammatical because the possessor of the nails is ex-
Social Sciences and Humanities Research Council of pressed in the pronoun me, and may not be expressed
Canada. Many people contributed to the data collec- again in the possessive adjective mes.
tion, transcription, and analysis of the main study in As each LRE represents the collaborative working
which the two students described in this article partici- through of a linguistic problem that Kim and Rick en-
pated. We would particularly like to thank Doug Hart, counter, it is of note that in Kim and Rick‘s 23 LREs, 21
Joan Howard, ShelleyTaylor, and Iva Baltova, who have resulted in a correct solution, 1 in a wrong solution
been unfailing in their help, even when called upon at (see Excerpt H), and 1 in no solution at all. Of the 23
untimely moments. Iva Baltova also drew the cartoons LREs, there were 15 relevant posttest items. Of those
shown in Appendix A. For comments on a draft version 15, Kim was always accurate in her response. Rick was
of this article, we are grateful to Alister Cumming, Rick either accurate or marked a “probable” category (see
Donato, Birgit Harley, Jim Lantolf, Tere Pica, Miles Item Type B in Figure 1). That is, Rick never marked
Turnbull, and the editors of this special issue, Sue Gass “definitely correctn when the accurate response was
and Alison Mackey. A special thank you is also ex- “definitely wrong,” and he never marked “definitely
tended to Gordon Wells, who allowed Merrill to sit in wrong” when the accurate response was “definitelycor-
on his course on Sociocultural Theory and engage in rect.”
collaborative dialogue with him and the class, while she ’ In English, “she walks to school” is grammatical,
was on sabbatical during 1996. and it is this structure that is likely being incorrectly
transferred to the French elk march a l’icok. In French,
the correct structure would be elle s’en ua lZcok. For a
discussion of this issue and of what French immersion
NOTES
students generally do when using verbs of motion in
French, see Harley (1989).
* “Jointactivity”can include interacting with text. In
some theories this would be called “individual learn-
ing.” We would argue, along with Wertsch (1991) and REFERENCES
others, that interacting with text is a social activity.
Usually we were able to identify which of the two
students were speaking (although the similarity of Allen, P., Swain, M., Harley, B., & Cummins, J. (1990).
voices at this age sometimes created problems in iden- Aspects of classroom treatment: Towards a more
tification). However, we were not usually able to relate comprehensive view of second language educa-
accurately the student speaking in a dyad to his or her tion. In B. Harley, P. Allen, J. Cummins, & M.
set of test scores. Being able to do so was a fortuitous re- Swain (Eds.), The development of second language
sult of the dyad we selected. p r o f i h q (pp. 57-81). Cambridge: Cambridge
3 The word rhemeans dream and the compound T h e - UniversityPress.
mutin is not a word. We believe that Rick, in using “&e- Cummins, J. (1990). From the inner city to the global
M a ' l l Swain and Sharon Lapkin 335
village: The microcomputer as a catalyst for col- conversation and the negotiation of comprehen-
laborative learning and cultural interchange. sible input. Applied Linguisticr, 4, 126-141.
Language Culture and Cum'culum, 1, 1-13. Mackey, A. (1995). Steppzng up the pace: Input, interaction
de Bot, K. (1996). The psycholinguistics of the output and interlanguage deuelopment: An empirical study of
hypothesis. Language harning, 46,529-555. questions in ESL. Unpublished doctoral disserta-
Donato, R. (1994). Collective scaffolding in second lan- tion, University of Sydney, Australia.
guage learning. In J. Lantolf & G. Appel (Eds.), Mackey, A,, & Philp, J. (1998). Conversational interac-
Vygotskian approaches to second language research tion and second language development: Recasts,
(pp. 33-56). Norwood, NJ: Ablex. responses, and red herrings? Modern Language
Donato, R., & Lantolf,J. (1990). The dialogic origins of Journal, 82,338-356.
L2 monitoring. In L. F. Bouton & Y. Kachru Newman, D., Griffin, P., & Cole, M. (1989). The con-
(Eds.), Pragmatics and language learning (Vol. 1, struction zone. Cambridge: Cambridge University
pp. 83-97). Urbana-Champaign, I L Division of Press.
English as an International Language. Nobuyoshi, J., & Ellis, R. (1993). Focused communica-
Foster, P. (1993). Discoursal outcomes of small group tion tasks and second language acquisition. ELT
work in an EFL classroom: A look at the interac- Journal, 47,203-211.
tion of non-native speakers. T h a w Valley Univer- Pica, T. (1994). Research on negotiation: What does it
sity WorkingPapers in EngZish Language Teaching, 2, reveal about second-language learning condi-
1-30. tions, processes, and outcomes? Language Learn-
Goss, N., Ying-Hua, Z., & Lantolf,J. (1994). Two heads ing, 44,493-527.
may be better than one: Mental activity in sec- Platt, E. & Brooks, F. B. (1994). The "acquisition-rich
ond-language grammaticality judgments. In E. environment" revisited. Modern LanguageJournal,
Tarone, S. Gass, & A. Cohen (Eds.), Research 78,497-511.
methodology in secon&language acquisition (pp. 263- Swain, M. (1995). Three functions of output in second
286). Hillsdale, NJ: Erlbaum. language learning. In G. Cook & B. Seidlhofer
Harley, B. (1989). Transfer in the written composition (Eds.), Principle and practice in applied linguistics:
of French immersion students. In D. W. Dechert Studies in honour of H. G. Widdowson (pp. 125-
& M. Raupach (Eds.), Transferin languageproduc- 144). Oxford: Oxford University Press.
tion (pp. 3-19). New York: Ablex. Swain, M. (1997). Collaborative dialogue: Its contribu-
Harley, B., & Swain, M. (1978). An analysis of the verb tion to second language learning. Reoista Canaria
system used by young learners of French. Inter- a% Estudios Zngleses, 34, 115-132.
language Studies Bulletin, ?, 35-79. ~
Swain, M. (1998). Focus on form through conscious re-
Jacob, E., Rottenberg, L., Patrick, S., & Wheeler, E. flection. In C. Doughty &J. Williams (Eds.),Focus
(1996). Cooperative learning: Context and o p on f m in classroom second language acquisition (pp.
portunities for acquiring academic English. 6481). Cambridge: Cambridge University Press.
TESOL Qarferly, ?O, 253-280. Swain, M., & Lapkin, S. (1995). Problems in output and
Johnson, D. W., Johnson, R., & Holubec, E. (1986). Cir- the cognitive processes they generate: A step to-
cles of learning: Cooperation in the chwoom. Edina, wards second language learning. Applied Linguis-
MN: Interaction Book. tics, 16, 371-391.
Kowal, M., & Swain, M. (1997). From semantic to syn- Vygotsky, L. (1979). Consciousness as a problem of psy-
tactic processing: How can we promote it in the chology of behavior. Soviet Psychology, 17,3-35.
immersion classroom? In R. K. Johnson & M. Wajnryb, R. (1990). Grammar dictation. Oxford: Oxford
Swain (Eds.), Immersion education: International University Press.
perspectives (pp. 284-309). Cambridge: Cam- Wells, G. (1996). The zone of proximal dmebpment and its
bridge University Press. implications f m learning and teaching. Unpublished
Krashen, S. (1985). The Input Hypothesis: Issues and im- manuscript, Ontario Institute for Studies in Edu-
plications. London: Longman. cation, University of Toronto, Ontario, Canada.
Lantolf, J., & Appel, G. (Eds.). (1994). @gotskian a p Wertsch,J. V. (1991). Vices of the mind:A sociocultural ap
proaches to second language research. Norwood, NJ: proach to mediated action. Cambridge, MA: Har-
Ablex. vard University Press.
LaPierre, D. (1994). Language output in a cooperative
learning setting: Determining its e f f t s on second lan-
guage.learning. Unpublished master's thesis, On-
tario Institute for Studies in Education, Univer-
sity of Toronto, Ontario, Canada.
Leont'ev, A. N. (1981). The problem of activity in psy-
chology. In J. V. Wertsch (Ed.), The concept of ac-
tivib in Sovietpsychology (pp. 37-71). Armonk, NY:
Sharpe.
Long, M. (1983). Native speaker/non-native speaker
336 Thz Modern LanguageJournal 82 (1 998)

APPENDIX A
Pictures Used by Dyads in Jigsaw Task

', 1 2

1 5

"The tricky alarm clock"

01. Baltova (1994)
Merrill Swain and Sharon Lapkin 337

APPENDIX B
Kim and Rick's Written Story
(exactly as written)

L'histoire de Jacqueline
mai 3 1995

Le soleil se I h e au matin et c'est une tres belle journte. Jacqueline est en train de dormir dans son lit sous
sa couverture bleu. A cBtt du lit il y a une petite table brune et en dessus il y a un rheil- matin rouge qui dit
6:OO du matin. Jacqueline se reveille P cause du rheil-matin qui sonne. Sa tCte avec les cheveux frissts est P
l'autre bout du lit avec les pieds sur l'oreiller. Elle sort sa jambe de la couverture pour pousser un boutton
sur le rtveil-matin pour arr&terle sonnement. h i s elle se met 2 dormir apr6s que le sonnement est arrCtt.
Un bras mtchanique sort du rtveil matin pour chatouiller le pied de Jacqueline avec une plume qui est tenue
par une main jaune. Elle se l&velentement et s'en va ii la salle de bain oh elle se 1Pve et se brosse les dents et
se peigne les cheveux. h i s elle semble desatreuse apr& de dormir. Elle est p&te pour aller P l'tcole quand
sur la rue elle voit un gars qui marche vers l'tcole et elle le suive jusqu'a l'tcole.

English Translation of Kim and Rick's WrittenStory

The Story of Jacqueline

May 3,1995

The sun rises in the morning and it's a very beautiful day. Jacqueline is sleeping in her bed under her blue
blanket. Beside the bed there is a small brown table and on it there is a red alarm clock that says 6 a.m. Jacque-
line wakes up because of the alarm clock that is ringing. Her curly-haired head is at the other end of the bed
with her feet on the pillow. She sticks her leg out from under the blanket in order to push a button on the
alarm clock to stop the ringing. Then she begins sleeping after the ringing has stopped. A mechanical arm
comes out of the alarm clock to tickle Jacqueline's foot with a feather that is held by a yellow hand. She gets
up slowly and goes to the bathroom where she washes and brushes her teeth and combs her hair. Then she
looks disastrous after sleeping. She is ready to go to school when in the street she sees a guy who is walking
toward school and she follows him to school.

Upcoming Conference Sessions on Publishing

THE MLJEDITOR INVITES READERS T O SESSIONS ON PUBLISHING IN THE MLJ.

American Council on the Teaching of Foreign Languages

Chicago, Illinois
November 20-22,1998
actflhq@aol.com
www.actfl.org

Central States Conference on Foreign Language Teaching

Little Rock, Arkansas
April 15-18,1999
rmcheatham@ualr.edu