Professional Documents
Culture Documents
The previous sections of this book have been concerned with questions of
general import to the study of second language acquisition. They have en-
deavoured to provide descriptions of how an L2 is acquired, and to explain
the structural processes that account for its universal properties and the social
and personal factors that account for variation in the rate and success of indi-
viduallearners. Although some attention has been given to the impact that
the acquisitional setting can have on outcomes (see Chapter 6), the primary
concern has been to account for the regularities in acquisition evident across
settings. Thus, no clear distinction has been drawn between the acquisition
that takes place in untutored and tutored settings.
Much of the research that has been reported in the earlier sections, how-
ever, has in volved classroom learners. This is particularly true of experi-
mental or correlational studies (for example, studies based on the
Competition Model or Gardner's research on motivation in Canada). One
reason is that the large numbers of learners needed for such research are more
accessible in educational than in naturalistic settings. However, this research
has not been concerned with classroom L2 acquisition per se; rather it has
used classroom learners to investigate questions of general significance to
SLA research.
In this section, we focus our attention specifically on classroom L2 acquisi-
tion. In so doing, of course, we will revisit many of the general issues raised
earlier in the book-for example, the rol e of interaction in shaping learning,
the difference between implicit and explicit knowledge, the role of negative
feedback, and the significance of acquisitional orders and sequences. In fact,
the classroom constitutes an ideal setting for examining the key theoretical
issues beca use it is possible to observe closely how input is made available to
the learner and what kinds of output learners produce in specific classroom
Contexts. It is also possible to engineer what input learners are exposed to and
what output they produce in order to investigate specific hypotheses about
how learning takes place.
The reason for focusing on the classroom, however, is not merely to shed
further light on how L2 acquisition takes place.lt is also motivated by a desire
to discover what classroom conditions are most likely to facilitate acquisi-
tion. In other words, it has a pedagogic purpose. However, this pedagogic
pUrpose is not quite the same as that addressed in methodological handbooks
for teachers, where the aim is to suggest specific techniques or activities that
')b't AaSSroom secona tanguage acqulsltlon
I•..
teachers can use. The research we wiU consider in this section invites us to 13 Classroom interaction and second
consider pedagogy not in terms of 'techniques' or 'activities', but in terms of
what kinds of classroom behaviours teachers need to engage in to promote language acquisition
learning-what questions to ask, when and how to correct learners' errors,
how to instigate negotiation for meaning in a classroom, etc. Furthermore,
whereas methodologists base their advice on what might be caUed 'sound
pedagogy', accumulated from centuries of experience of language teaching,
the pedagogical proposals drawn from L2 classroom acquisition research are
informed by theoretical concepts drawn from SLA research. It is not inten-
ded, however, to suggest that such an approach to language pedagogy is the
preferred one-a point that we wiU take up in the final chapter of this book. Introduction
It is possible to identify two major strands in classroom L2 acquisition re- The classroom affords the L2 researcher three different perspectives. The first
search, corresponding to two ways of viewing the classroom. According to perspective is that found in comparative method studies. These seek to com-
one view, what happens in the classroom provides opportunities for learning, pare the effect of different language teaching methods on L2 learning. Such
which can be explored by examining the relationship between inter- studies are 'product'-based because they rely entirely on measurements of
action-the means by which the opportunities are provided-and L2 language learning and make no attempt to examine the instructional and
learning. An alternative is to view the classroom as a place where attempts are learning 'processes' that take place inside the classroom.
made to intervene directly in the process of L2 learning-to ask, in other A second perspective involves going inside the 'black bOX,l of the class-
words, 'Do learners learn what they have been taught?'. Chapter 13 examines room itself. It views the classroom as a place where interactions of various
research belonging to the first strand, while Chapter 14 looks at the effects of kinds take place, affording learners opportunities to acquire the L2. AU-
direct intervention on learning.
wright (1984: 156) sees interaction as 'the fundamental fact of classroom
pedagogy' because 'everything that happens in the classroom happens
through a process of live person-to-person interaction'. This perspective has
drawn heavily on the research and theories dealing with the relationship be-
tween input/interaction and L2 learning (see Chapter 7). lt leads the re-
searcher to observe and describe the interactional events that take place in a
classroom in order to understand how learning opportunities are created.
IdeaUy, it should also lead to attempts to demonstrate the effects of different
types of interactional opportunity on L2 learning, but as we saw in Chapter 7,
few studies have investigated this relationship directly.
The third perspective involves investigating the effects of formal instruc-
tion. In this case, instruction is viewed as an attempt to intervene directly in
the language learning process by teaching specific properties of the L2. The
question that is asked is 'Do learners actuaUy iearn what they are taught?' Re-
searchers havebeen particularly interested in whether instruction directed at
specific grammatical items and rules has any effect on interlanguage develop-
mento This chapter wiU consider research that has adopted the first two per-
spectives, and the foUowing chapter wiU examine research relating to the
third.
First of aU, this chapter wiU consider some of the principal research meth-
ods that have been used to investigate the role of classroom interaction in lan-
guage learning. It wiU then review a number of comparative method studies.
The main part of the chapter is given over to a survey of research that has
566 Ctassroom secona language ueqU¡"'ntUH
examined different aspects of classroom interaction: teacher talk, error treat- without a process element to provide information about the actual events that
ment, teachers' questions, learner participation, task-based interaction, and take place inside the classroom, it is difficult to be certain that the method-
small group work. Finally, consideration is given to the relationship between program distinctions actuall~ result in d.ifferent ~lassr?om behaviours (see
classroom interaction and L2learning and, in particular, whether classroom Long 1984). The psychometnc approach lS also eV1dent m correlational stud-
environments are capable of providing the kinds of opportunities needed to ies that have examined the relationship between specific classroom behavi-
develop full L2 competence. ours (such as teachers' requests) and learning outcomes (for example, Politzer
A distinction can be drawn between 'classroom research' and 'classroom- 1980; politzer, Ramirez, and Lewis 1981). Chaudron (1988: 30) notes that
orientated research' (see Nunan 1991: 249). The former consists of studies these studies suffer from a failure to validate the categories used to measure
that have investigated learners inside actual classrooms, while the latter con- instructional features, as well as from a failure to establish theoreticallinks
sists of studies conducted outside the classroom (for example, in some kind of between the processes observed and learning outcomes.
experimental setting) but which have been motivated by issues of clear relev-
ance to classroom L2 acquisition. This chapter will consider both typesof Interaction analysis
research, but will try to make clear which is which.
Interaction analysis involves the use of a form or schedule consisting of a set
of categories for coding specific classroom behaviours. Long (1980b) refers to
Methods of researching learning in the second language classroom three different types of interaction analysis: in a category system each event is
A number of different research traditions in L2 classroom research can be coded each time it occurs, in a sign system each event is recorded only once
identified. Chaudron (1988) describes four (see Table 13.1). The psycho- within a fixed time span, while in a rating scale an estimate of how frequently
metric and interaction analysis traditions typically involve 'quantitative' and a specific type of event occurred is made after the period of observation. lni-
'explanatory' research, while the discourse analysis and ethnographic tradi- tially, schedule s were developed for content classrooms (see Flanders 1970
tions make use of more 'qualitative' and 'descriptive' methods. for an account of this early work), but these were rapidly adapted to the la n-
guage classroom (for example, Moskowitz 1967). Frequently, the categories
Typical issues Methods listed in a schedule reflected the researcher's assumptions about what behavi-
Tradition
ours were important and were not theoretically motivated. The Marburg
Language gain from different Experimental method-pre- and
psychometric schedule (Freudenstein 1977) included categories such as 'phases of instruc-
methods, materials, treatments. post-tests with experimental and
control groups. tion', assuming that the basic pattern of a lesson would follow the pattern
Extent to which learner behaviour Coding classroom interactions in 'warming up', 'presentation', 'learning', and 'using'. However, subsequent
Interaction analysis
is a function of teacher- terms of various observation schedules (for example, Fanselow 1977a; Allwright 1980) have attempted to
determined interaction. systems and schedules. produce more comprehensive and method-neutral sets of categories, while
Discourse analysis Analysis of classroom discourse Study classroom transcripts and some (for example, Allen, Frohlich, and Spada 1984) have been based on a
in linguistic terms. assign utterances to pre- theoretical understanding of the nature of L2 acquisition.2
determined categories.
Long (1980b) lists twenty-two interaction analysis systems to which sev-
Obtain insights into the Naturalistic 'uncontrolled'
Ethnographic eral more must now be added. This proliferation reflects differences in re-
classroom as a cultural system. observation and description.
search foci. It makes comparison across studies extremely difficult. AIso,
beca use the behaviour of the teacher and the learners is often treated separ-
Table 13.1: A summary o(Chaudron's (our research traditions in L2 classroom ately, information is lost about 'the sequential flow of classroom activities'
research (taken (rom Nunan 199Gb: 23)
~McLaughlin 1985: 149). Often interaction analysis runs the risk of produc-
lllg disconnected tallies of behaviours that obscure the general picture, as
there is no basis for deciding which combinations of features might be im-
The psychometric tradition portant. lnteraction analysis depends on a number of assumptions-for ex-
ample, that it is possible for an observer to 'read' the intentions of the teacher
The psychometric tradition is evident in the comparative method studies to be and students. Such assumptions can be questioned, thus casting doubt on the
considered in the next section and in program-product comparisons of the
reliability and validity of the measurements. However, this is more of a prob-
kind used to evaluate different types of immersion programmes (see Swain
lem with 'high inference' categories (for example, 'degree of teacher control
and Lapkin 1982 for a review). The main problem with such studies is,that
'-""'<><>1 vv'" ffH""Uc.UUft UftU ~ecunu tanguage acqulSltlon )6~
exercised over materials') than with 'low inference' categories (for example, room participants succeed in creating and managing the events in which they
'type of materials-text, audio or visual'), as Long (1980b) has pointed out. take part by repeated viewings of videotaped lessons and by soliciting the re-
actions of the participants to the events that took place. Microethnography,
which focuses on 'particular cultural scenes in key institutional settings'
Discourse analysis
(Erickson and Mohatt 1982, quoted in Mitchell1985) works in a similar way
Discourse analysis serves as a device for systematically describing the kinds of to constitutive cthnography, but restricts the object of study to contexts of
interactions that occur in language classrooms. Drawing on initial work on activity recognized as real by the participants (for example, 'getting ready for
content classrooms by Bellack et al. (1966) and by the Birmingham school of a break').
linguists (Sinclair and Coulthard 1975; Coulthard and Montgomery 1981; The ethnographic tradition has been particularly evident in research into
Sinclair and Brazil 1982), discourse analysts give attention not only to the bilingual classrooms (for example, Phillips 1972; Trueba, Guthrie, and Au
function of individual utterances but also to how these utterances combine to 1981; Wong-Fillmore 1985; Cathcart 1986) and the diary studies discussed
form larger discoursal units. They aim to account for the joint contributions in Chapter 11. Gaies (1983a) gives three advantages of such research: (1) it
of teacher and student and to describe all the data, avoiding the kind of 'rag can account for learners who do not participate actively in class, (2) it can
bag' category found in many interaction analysis schedules. provide insights into the conscious thought processes of participants, and (3)
McTear (1975) has shown how the Birmingham framework can be ad- it helps to identify variables which have not previously been acknowledged.
apted to account for the discourse structure found in language lessons. Re- Disadvantages include the time-consuming nature of the work needed to
searchers at CRAPEL in the University of Nancy (e.g Gremmo, Holec, and collect data, the difficulty of generalizing results, and the danger of ignoring
Riley 1978; Riley 1985) have also made use of discourse analysis to show superordinate variables relating to the learners' social contexto
how 'natural' discourse is distorted in the language classroom as the result of These four traditions have produced a wealth of research into L2 class-
the teacher's dominance. Other researchers (for example, ElIis 1980 and room processes and products. It is to the results (and insights) which this re-
1984a; Van Lier 1982 and 1988) ha ve developed frameworks based on di s- search has provided that we now turno
course analysis to characterize the different types of interaction that can oc-
cur in the L2 classroom. More commonly, researchers have used the
techniques of discourse analysis to develop comprehensive accounts of spe-
Comparative method studies
cific areas of discourse. Chaudron (1977), for example, pro vides an analysis The aim of comparative method studies is to establish which of two or more
ofteacher feedback, while Long and Sato (1983) offer a discourse-based ana- methods or general approaches to language teaching is most effective in terms
lysis of teachers' questions. In the case of these latter studies, discourse cat- of the actuallearning (the 'product') that is achieved after a given period of
egories have served as a basis for quantitative research, but, as Chaudron time. Many of the earlier studies were 'global' in nature, conducted over
(1988) notes, much of the research within the discourse analysis tradition has weeks, months, and even years. The later ones have tended to examine differ-
been descriptive in nature, concentrating on the development of comprehens- ences resulting from shorter periods of exposure to different methods.
ive analytical systems. The 1960s were characterized by what Diller (1978) called 'the language
teaching controversy'. This pitted the claims of rationalist approaches to lan-
guage teaching against those of empiricists. Methods such as the traditional
The ethnographic tradition
grammar-translation method and the cognitive-code method emphasized
The ethnographic tradition involves the kind of detailed descriptive work ad- the provision of explicit knowledge through rule explanation and of learning
vocated by Van Lier (1988). It emphasizes the importance of obtaining mul- simultaneously through all four skills, while so-called 'functional' methods
tiple perspectives through triangulation. Long (1980b) identifies a number of such as audiolingualism and the oral approach emphasized inductive rule
ethnographic approaches. Participant ethnography involves the researcher learning through listening and extensive oral practice. At the time, it seemed
taking a regular part in the activities under study. Non-participant ethno- logical to investiga te which method produced the better results.
graphy sets the researcher outside the classroom events being observed. It In an early study, Scherer and Wertheimer (1964) compared the grammar-
uses a variety of data collection techniques: note-taking, interviewing, ques- -translation method and the audiolingual approach by following the progress
tionnaires, ratings of personal opinions, and written documents (for ex- of different groups of college-leve! students of L2 German taught by each
ample, teachers' handouts and students' homework). Constitutive method and tested at the end of the their first and second year of study. The
ethnography (see Mehan 1979) aims to make explicit the way in which class- results showed that students in the grammar-translation group did better in
.)/U Classroom second language acqulsltlOn Uassroom mteractlon and second language acquisition 571
reading and writing while the students in the audiolingual group did better at f r a review) conducted a number of studies designed to compare the effects
listening and speaking. In other words, each method resulted in learning o 1earning of Total Physical Response (TPR), a method devised and pro-
'products' that reflected the instructional emphasis. onoted by Asher, and other methods, in particular the audiolingual approach.
A subsequent large-scale study known as the Pennsylvania Project (Smith ~ her claims that the results of these studies show that TPR results in greater
1970) compared the effects of three methods on beginning and intermediate s:ort-term and long-term retention of new linguistic m~te~ial and better un-
French and German classes at the high-schoollevel. The three methods were derstanding of novel utterances. He also suggests they mdICate that learners
(1) 'traditional' (i.e. grammar-translation), (2) 'functional skills' (essentially are able to transfer the training in listening they received to other
the audiolingual approach), and (3) 'functional skills plus grammar'. Student skills-speaking, reading, and writing. Furthermore, Asher reports that stu-
achievement in the four skills of listening comprehension, speaking, reading, dents taught by TPR were more likely to continue studying the foreign la n-
and writing was evaluated at mid-year and at the end of the year using a bat- guage and displayed more positive attitudes. Krashen (1982: 156) concludes
tery of standardized tests. The results in general showed no significant differ- his survey of Asher's studies by claiming that 'the TPR results are clear and
ences between the three methods, except that the 'traditional' group was consistent, and the magnitude of superiority ofTPRis quite striking'. A num-
superior to the other two groups on two of the reading tests. After two years, ber of caveats are in order, however. First, Asher had a vested interest in find-
the 'traditional' group again surpassed the 'functional skills' group in reading ing in favour of TPR; second, the period of instruction was relatively
ability but did significantly worse on a test of oral mimicry. No differences short-only 20 hours in some studies; third, only beginners were investig-
were found in the students' performance on the other tests. ated. We do not know, therefore, whether the method is equally effective in
Hauptman (1970) compared the effects of instruction based on a 'struc- the long term or with advanced learners. Nevertheless, the TPR studies stand
tural' approach (where the grammatical structures were sequenced according out in comparative method studies as providing evidence that support the su-
to level of difficulty) and a 'situational' approach (consisting of dialogues ar- periority of a particular method.
ranged by situation). In this case the learners were third to sixth grade stu- More recent discussions of language teaching methodology have emphas-
dents of Japanese. In general, the situational group performed better than the ized the importance of providing opportunities for learners to communicate.
structural group on situational test items, and did as well on structural test The few studies that have investigated the effectiveness of communicative
items; however, this difference was found to be significant only for high apti- language teaching, however, have led to the same kind of indeterminate res-
tude/high IQ students in the situational group. Design problems (such as the ults found in the early global method studies. Palmer (1979) compared the ef-
short length of the teaching programme-only three weeks-and the fact fects of 'traditional' instruction and 'communicative' instruction involving
that one persqn did all the teaching) make it difficult to reach firm conclusions extensive peer-communication based on language games. The subjects were
from this study. Thai learners of English. No significant differences between the groups were
In general, therefore, these studies failed to provide convincing evidence found. One possible explanation for this unexpected finding, according to
that one method was superior to another. One possible explanation, as sug- Krashen (1981), was that while teacher talk was in the target language in the
gested by Clark (1969) in his detailed discussion of the Pennsylvania Project, traditional group, it was in the learners' Ll in the communicative class. Ham-
was that the distinctions between the different methods were not in fact clear. mond (1988) compared groups of students in a Spanish program at two uni-
Clark comments: versities. Eight experimental groups were taught by means of the Natural
Approach (Krashen and Terrelll983) and 52 control groups were taught by
If ostensibly different teaching methods tend in the course of the experi-
means of the grammar-translation method, which emphasized the deductive
ment to resemble one another in terms of what actually goes on in the class-
learning of grammar. Although in general the experimental groups outper-
room, the likelihood of finding significant differences in student
formed the control groups in both a mid-term and a final examination, many
performance is accordingly reduced.
of the differences were not significant. This study did show, however, that
As Allwright (1988) has carefully documented, this led to the development of students in a communicative classroom did no worse in learning grammar
instruments for observing what actually takes place during the course of in- than those in a traditional program-a point we will take up in a later section.
struction (for example, Jarvis 1968) and a focus on describing language les- The difficulties of conducting effective comparisons between 'communic-
sons at the level of technique rather than method (for example, Politzer ative' and 'non-communicative' classrooms are also evident in a carefully
1970). planned study by Allen, Swain, Harley, and Cummins (1990). This was based
Comparative method studies were not abandoned, however. Asher af\d as- on Stern's (1990) distinction between 'experiential' and 'analytic' teaching
sociates (for example, Asher, Kusudo, and de la Torre 1974; see Asher 1977 strategies (see Table 13.2). Allen et al. used an interaction analysis schedule
'''' .. · __·r r_r.rorr_ro_
Ctassroom znteractlon and second language acquisition 573
Method • ---- ....•••.Practice opportunities This system was not developed to account for language lessons, but as
McTear (1975) has shown, it fits remarkab!y well. Only small changes are
Atmosphere ~ ~~~~R "
eceptlvlty necessary. For example, students in the L2 classroom often produce an
additional response after the follow-up move in IRF exchanges:
Figure 13.1: The relationship between plans and outcomes (from Al/wright and
Bailey 1991: 25) T: What do you do every morning?
s: 1clean my teeth.
Our concern here is with the nature of classroom interaction. Although
T: You clean your teeth every morning.
classroom interactions are not usually designed in advance of a lesson, they
S: 1clean my teeth every morning.
have been found to manifest distinct and fairly predictable characteristics.
These will be considered in terms of (1) the structure and general character- The exchange structure is, therefore, IRF(R). It should be noted, however,
istics of classroom discourse, (2) types of language use, (3) turn-taking, and that this structure is only likely to arise in classroom discourse which is
(4) differences between classroom and naturalistic discourse. teacher-controlled. Although IRF(R) exchanges tend to domina te, other
kinds can also be found. Van Lier (1988) points out that it is easy to overstate
the lack of flexibility evident in L2 classroom discourse. He found that al-
Structure and general characteristics though the discourse is often strictly controlled by the teacher, learners do
Classroom discourse has an identifiable structure. Mehan (1979) distin- sometimes initiate exchanges, and 'schismic talk' (talk that deviates from
guished three components in general subject lessons: (1) an opening phase, some predetermined plan) also occurs, at least in some classrooms.
where the participants 'inform each other that they are, in fact, going to con- A number of interaction analysis schedules have also attempted to account
duct a lesson as opposed to some other activity', (2) an instructional phase, for the general characteristics of classroom discourse. Some of the most fre-
where information is exchanged between teacher and students, and (3) a clos- quently cited are Fanselow (1977a), Allwright (1980) and Allen, Frohlich and
ing phase, where participants are reminded of what went on in the core of the Spada (1984). The last mentioned was used in the comparison of experiential
lesson. However, language lessons often seem just to start and stop (i.e. they and analytic teaching approaches discussed in the previous section. We will
focus on it here.
consist entirely of the instructional phase), perhaps beca use the content is
'Ianguage' itself. The Communicative Orientation in Language Teaching (COLT) (Allen,
In one of the best known accounts of classroom discourse, Sinclair and Frohlich and Spada 1984) differs from the systems that preceded it in that it
Coulthard (1975) develop a hierarchical model by identifying the following was not only informed by current theories of communicative competence and
'ranks' in the structure of a lesson: (1) lesson, (2) transaction, (3) exchange, communicative language teaching but also by research into Ll and L2 ac-
quisition.3 The authors comment:
(4) move, and (5) act. Overall, a 'Iesson' has on!y a weakly defined structure,
consisting of 'an unordered series of transactions'. A 'transaction' consists of The observational categories are designed (a) to capture significant features
a 'preliminary', one or more 'media!', and a 'terminal' exchanges. Ir is most of verbal interaction in L2 classrooms and (b) to provide a means of com-
easily identifiable by means of 'boundary exchanges', signalled by framing paring some aspects of classroom discourse with naturallanguage as it is
and focusing moves. The element of structure that is most clearly defined, used outside the classroom (1984: 232).
however, is that of 'teaching exchange', which typically has three phases,
involving an 'initiating' move, a 'responding' move, and a 'follow-up' move, ~he system is in two parts. The first part, 'A description of classroom activi-
as in this example: tles', is designed for use in real-time coding. Ir consists of a set of general cate-
gories broken down into narrower sub-categories, in line with earlier
T: Ask Anan what his name is? (initiating) systems. The main unit of analysis is 'activity type'; examples of the kinds of
S: What's your name? (responding) activities that might be identified are drill, translation, discussion, game and
T: Good. (follow-up). dialogue. Each activity is then described in terms of participant organization
576 Classroom second language acquisition Classroom interaction and second language acquisition 577
(whether whole class, group work, or individual work), content (the subject The aim of interaction analysis systems such as COL T is to identify signific-
matter of the activities), student modality (the various skills involved in the ant aspects of L2 classroom discourse and to develop specific categories that
activity), and materials (type, length, and source/purpose). This part, there- allaw for quantification. By so doing they provide a basis for investigating
fore, relies on pedagogic rather than interactional constructs. However, the which interactional features are important for language acquisition. How-
second part, 'Communicative features', reflects a more discoursal perspective ever, as we noted earlier in this chapter, they ha ve a number of drawbacks.
on the classroom. Coding is based on an audio recording of the classes ob-
served. Seven communicative features are identified relating to use of the tar-
get language, whether and to what extent there is an information gap, Types of language use
sustained speech, whether the focus is on code or message, the way in which Other researchers have sought to describe classroom interaction by identi-
incorporation of preceding utterances takes place, discourse initiation, and fying the different types of language use or interaction found in L2 class-
the degree to which linguistic form is restricted. Table 13.3 illustrates Part raams. We will now examine a number of such accounts.
Two of the system by comparing extracts from two lessons, one reflecting a Allwright (1980) provides what he calls 'a macro-analysis of language
'stereotyped routine', and the other communication that is closer to natural teaching and learning' by identifying three basic elements:
language behaviour. Unlike other interaction analysis systems, COL T has
1 Samples, instances of the target language, in isolation or in use
been used quite extensively in classroom research (Spada 1987; Harley,
2 Guidance, instances of communication concerning the nature of the tar-
Allen, Cummins, and Swain 1990). It has been successfully used to distin-
guish the types of interaction that occur in different L2 classrooms. get language
3 Management activities, aimed at ensuring the profitable occurrence of
(1) and (2) (1980: 166).
A Analytical instruction
These elements are not mutually exclusive, as instances of 'guidance' and
Utterance Communicative features 'management activities' automatically provide 'samples'. They are held to
vary according to their relative proportion, their distribution between teacher
T: What's the date today? L2/pseudo-request/minimal speech
and learner, their sequencing, and the language used (target or other).
8,: April15th L2/predictable information/ultraminimal speech/limited form
T: Good. L2/comment/minimalspeech Other accounts of types of classroom use distinguish between interaction
T: What's the date today? L2/pseudo-request/minimalspeech where the focus is the code itself (a key feature of the language classroom) and
82: April15th. L2/predictable information/ultraminimal speech/limited form interactions which centre on genuine meaning exchange. McTear (1975), for
T: Good. L2/comment/minimal speech instance, identifies four types of language use based on this general
distinction:
embedded in some ongoing activity such as model-making (activity); (2) research that seeks to establish relationships between classroom processes
framework goals associated with the organization and management of class- and L2 learning statistically.
room events; and (3) social goals. 1 discussed interactional sequences taken
from an ESL classroom in Britain to illustrate how the type of goal influences
the discourse, and then speculated about the learning opportunities each type Turn-taking
affords. 1pointed out that interactional events with core goals are likely to re-
Research which has specifically examined turn-taking in the L2 classroom
strict learners to a responding role, whereas framework and social goals pro-
has drawn extensively on ethnomethodological studies of naturally occurring
vide opportunities for them to initiate discourse and to perform a wider range
conversations (for example, Sacks, Schegloff, and ]efferson 1974). These
of language functions. Kaneko (1991) used this framework to examine the ef-
identified a number of rules that underlie speaker selection and change: only
fect of language choice (target or the learners' Ll) on learning. She found that
one speaker speaks at a time; a speaker can select the next speaker by nom-
the items that ]apanese high school students reported learning in English les-
inating or by performing the first part of an adjacency pair (for example, ask-
sons occurred most commonly in interactional events with core goals where
ing a question that requires an answer); a speaker can alternatively allow the
there was an element of spontaneous language use.
next speaker to self-select; and there is usually competition to take the next
In Van Lier's (1982; 1988) framework, there are four basic types of class-
turno Classroom researchers frequently highlight the differences between
room interaction, according to whether the teacher controls the topic (i.e.
turn-taking in natural and classroom settings. McHoul (1978), for instance,
what is talked about) and the activity (i.e. the way the topic is talked about). has shown that classroom discourse is often organized so that there is a strict
Type 1 occurs when the teacher controls neither topic nor activity, as in the allocation of turns in order to cope with potential transition and distribution
small talk sometimes found at the beginning of a lesson or in private talk be-
problems and that who speaks to whom at what time is firmly controlled. As a
tween students. In Type 2 the teacher controls the topic but not the activity; it result there is less turn-by-turn negotiation and competition, and individual
occurs when the teacher makes an announcement, gives instructions, or de- student initiatives are discouraged.
livers a lecture. Type 3 involves teacher control of both topic and activity, as
Turn-taking in language classrooms does not differ from that in general
when the teacher elicits responses in a language dril!. In Type 4 the teacher
subject classrooms. Lorscher (1986) examined turn-taking in English lessons
controls the activity but not the topic, as in small-group work where the pro-
in different types of German secondary schools and found that turns were al-
cedural rules are specified but the students are free to choose what to talk most invariably allocated by the teacher, the right to speak returned to the
about. In a further development of this framework, Van Lier (1991) adds a teacher when a student turn was completed, and the teacher had the right to
third dimension,: the function that the language serves. He follows Halliday interrupt or stop a student turno Lorscher argues that these rules are deter-
(1973) in distinguishing three types of function: ideational (telling people
mined by the nature of the school as a public institution and by the teaching-
facts or experiences), interpersonal (working on relationships with people), learning process.
and textual (signalling connections and boundaries, clarifying, summarizing, The most extensive discussion of turn-taking in the L2 classroom can be
etc.).
found in Van Lier (1988). In an attempt to identify participation in classroom
These broad frameworks have been developed through ethnographic stud-
discourse characterized by learner initiative, Van Lier identifie~ a number of
ies of language classrooms and, as such, were not intended to serve as schemes turn-taking behaviours that he considers indicative of such initiative. Ex-
for coding specific classroom behaviours. Often it is not clear what unit of amples, under the headings provided by Van Lier, are:
classroom discourse they are intended to relate to-in Ellis (1984) 1 con-
sidered them in terms of interactional 'sequences', loosely defined as a unit of Tapie
discourse with a unitary topic and purpose, while Van Lier illustrates his
framework with reference to instructional activities. Their advantage is that The turn is off-stream (i.e. discontinuing), introduces something new, or
they provide a tool for understanding classroom interaction and how it might denies/disputes a proposition in a previous turno
affect learning. Their disadvantage is that they do not permit precise quanti-
fication, and thus cannot be easily used in experimental or correlational
JOV \.-tU:>sroomsecona language acqulsition Classroom interaction and second language acquisition 581
Self-selection . stitutional roles, the tasks are concerned with the transmission and recep-
Selection originates from the speaker. In n of information and are controlled by the teacher, and there is a focus on
~~owledge as a product and on accuracy. Natural discourse is characterized
Allocation by more fluid roles. established. t~rough inter~ction, tasks that enco~rage
equal participation lI1 the negotlatlOn of meaI1l~g, an.d a focus on the lI1ter-
The turn selects one specific next speaker.
actional process itself and on fluency. One way lI1whlch the two worlds can
be brought together is through communicating about learning itself, as
Sequence
suggested by Breen (1985).
The turn is independent of sequence. However, although the potential exists for natural discourse to occur in the
Van Lier goes on to provide an example of how this system can be used to classroom, studies show that it seldom does. Pica and Long (1986) found that
assign a 'participation index' for each of the participants in an interaction. He there was very little negotiation of meaning in elementary ESL classrooms in
uses this type of analysis to challenge McHoul's claim that 'only teachers can Philadelphia in comparison to native speaker-non-native speaker conversa-
direct speakership in any creative way' (1978: 188), pointing out that in his tions outside the classroom, as evident in significantly fewer conversational
L2 classroom data many of the teacher's utterances were undirected and the adjustments by the teachers. Politzer, Ramirez, and Lewis (1981) report that
learners frequently did self-select. He also notes that explicit turn-taking 90 per cent of all student moves were responses, testifying to the limited na-
sanctions are rare in adult L2 classrooms and that there is a considerable de- ture of opporrunities to participate that learners are afforded in class-
gree of tolerance of unintelligibility. Ir would seem, therefore, that the rigid room s-a point we will take up later. These and other studies (see Glahn and
rules of turn-taking described by Lorscher are not always evident. Holmen 1985; Kasper 1986) testify to the restricted nature of pedagogic dis-
Studies of turn-taking in the L2 classroom make explicit comparisons be- course, although other studies, such as Enright (1984) show that there can be
considerable variation between classrooms.
tween classroom and naturalistic discourse, the underlying assumption being
that the former is less conducive to successful L2 acquisition than the latter. The teacher's control over the discourse is the main reason for the preval-
Van Lier offers an interesting explanation for why this might be the case. He ence of pedagogic discourse. Researchers at CRAPEL (for example, Gremmo,
suggests that if turn-taking is rigidly controlled, the learners have no need to Holec, and Riley 1977; 1978) have argued that in the classroom setting dis-
attend carefully to classroom talk to identify potential transition points when course rights are invested in the teacher. Ir is the teacher who has the right to
they can take their turno As a result, they lack 'an intrinsic motivation for participate in all exchanges, to initiate exchanges, to decide on the length of
listening' and so are robbed of input. However, to date there have been no exchanges, to close exchanges, to include and exclude other participants, etc.
studies investigating the effect of different types of turn-taking on acquisition. When teachers elect to act as 'informants' or 'knowers' (Corder 1977b), they
are likely to make full use of their rights, and as a consequence the learners are
placed in a dependent position. As a result there is a preponderance of teacher
The difference between classroom and naturalistic discourse acts over student acts (typically in a 2:1 ratio), beca use teachers open and
The discourse that results from trying to learn a language is different from close each exchange.
that which results from trying to communicate. Edmondson (1985) draws on In the opinion of some, pedagogic discourse constitutes a 'falsification of
Labov's idea of the Observer's Paradox (see Chapter 4), to suggest that there behaviour' and a 'distortion' (Riley 1977), but other researchers see it as inev-
is also 'the teacher's paradox', which states: itable and even desirable (for example, Edmondson 1985). To date, there are
~ore arguments than evidence, although as we will see in Chapter 14, formal
We seek in the classroom to tea'ch people how to talk when they are not be-
II1struction does appear to result in faster learning and higher levels of ulti-
ing taught (1985: 162). mate achievement.
Thus, there is a tension between discourse that is appropriate to pedagogic
goals and discourse that is appropriate to pedagogic settings. However, be- Teacher talk
cause the classroom affords opportunities to communicate as well as to learn,
there are 'co-existing discourse worlds'. Kramsch (1985) suggests that the na- The bulk of L2 classroom research has focused on specific aspects of interac-
ture of classroom discourse will depend on the roles the participants adopt, tion, which we will now consider, beginning with teacher talk. Chaudron
the nature of the learning tasks, and the kind of knowledge that is targeted. (1988: Chapter 3) provides a comprehensive survey of studies of teacher talk.
Instructional discourse arises when the teacher and the students act out His main conclusions are summarized in Table 13.4.
~mJJ' '"''"',,, "t<t-unu tUnguage acqu¡s¡t¡on Classroom interaction and second language acquisition 583
messages. As Vigil and Oller (1976) have pointed out, it is useful to distin- Courchéne 1980; Chaudron 1986; Fanselow 1977b; Lucas 1975). The main
guish 'cognitive' and 'affective' feedback; the former relates to actual under- conclusions are that certain types of errors are much more likely to be treated
standing while the latter concerns the motivational support that interlocutors than others: discourse, content, and lexical errors receive more attention than
provide each other with during an interaction. 'Repair' is a somewhat nar- phonological or grammatical errors; that many errors are not treated at all;
rower term used by ethnomethodologists such as Schegloff, ]efferson, and that the more often a particular type of error is made, the less likely the
Sacks (1977) to refer to attempts to identify and remedy communication teacher is to treat it; and that there is considerable variation among teachers
problems, including those that derive from linguistic errors. 'Correction' has regarding how frequently error treatment takes place. Edmondson (1985)
a narrower meaning still, referring to attempts to deal specifically with lin- has also pointed out that teachers sometimes correct 'errors' that have not in
guistic errors; it constitutes an attempt to supply 'negative evidence' (see fact been made!
Chapter 10) in the form of feedback that draws the learners' attention to the Another issue concerns who performs the treatment. Studies of repair in
errors they have made. naturally-occurring conversations have shown a preference for self-initiated
In an attempt to give greater precision to the term 'treatment', Chaudron and self-completed repair (see Chapter 7). In classroom contexts, where, as
(1977) distinguishes four types: we have seen, discourse rights are unevenly invested in the teacher, other-
initiated and other-completed repair can be expected. This is what Van Lier
1 Treatment that results in learners' 'autonomous ability' to correct them-
selves on an item. (1988) predominantly found. Other patterns of repair can also occur, how-
ever. Kasper (1985) found that in the language-centred phase of an English
2 Treatment that results in the elicitation of a correct response from a
learner. lesson in aGrade 10 Danish gymnasium, the trouble sources were identified
by the teacher but they were repaired either by the learners responsible for
3 Any reaction by the teacher that clearly transforms, disapprovingly re-
them or by other learners. In the content phase of the same lesson, self-
fers to, or demands improvement.
initiated and self-completed repair was evident, although the learners were
4 Positive or negative reinforcement involving expressions of approval or
disapproval. inclined to appeal for assistance from the teacher. As Van Lier (1988: 211)
points out, the type of repair work is likely to reflect the nature of the context
Type (1) cannot be determined within the context of a single lesson as it re- which the teacher and learners have jointly created. AIso, the teacher's per-
quires evidence that the feedback has had some effect on acquisition. Type (4) sonal teaching style may influence the type of repair work that occurs, as Nys-
is limiting because it restricts the object of enquiry to occasions when the trom's (1983) study of the correction styles of four teachers in bilingual
teacher draws explicit attention to learner performance. Most studies have classrooms demonstrates.
examined (2) and (3). Probably the main finding of studies of error treatment is that it is an
A second issue concerns learners' attitudes towards error treatment. Cath- enormously complex process. This is evident in the elaborative decision-
cart and Olsen (1976) found that ESL learners like to be corrected by their making systems that have been developed (Long 1977; Day et al. 1984; Chau-
teachers and want more correction than they are usually provided with. dron 1977) and also in the extensive taxonomies of the various types of teach-
Chenoweth et al. (1983) found that learners liked to be corrected not only ers' corrective reactions (AlIwright 1975; Chaudron 1977; Van Lier 1988).
during form-focused activities, but also when they were conversing with nat- Chaudron's system, for example, consists of a total of 31 'features' (correct-
ive speakers. This liking for correction contrasts with the warning s of ive acts that are dependent on context) and 'types' (acts capable of standing
Krashen (1982) that correction is both useless for 'acquisition' and dangerous independently). These descriptive frameworks provide a basis for examining
in that it may lead to a negative affective response. Krashen may be partly teachers' preferences regarding types of error treatment. Studies have shown,
right, though, as Cathcart and Qlsen also report that when a teacher at- for instance, that repetitions of various kinds are a common type of corrective
tempted to provide the kind of corrections the learners in their study said they feedback (Salica 1981; Nystrom 1983). The frameworks are indicative of the
liked, it led to communication which the class found undesirable. ~areful descriptive work that needs to be done before it is possible to set about
Of considerable interest is the extent to which teachers should correct Investigating the effects of different kinds of error treatment on acquisition.
learners' errors. Here, there are widely diverging opinions, reflecting the dif- Further evidence of the complexity of the decision-making process during
ferent theoretical arguments about the need for negative feedback (see the error treatment is the inconsistency and lack of precision that teachers mani-
section on 'Error evaluation' in Chapter 2). Irrespective of what teachers fest. Long (1977) notes that teachers often give more than one type of feed-
should do, there is the question of what they actually do do. Chaudron (1988) back simultaneously, and that many of their feedback moves go unnoticed by
reviews a number of studies which have investigated this (Salica 1981; the students. Teachers are likely to use the same overt behaviour for more
•••...••••••••••••••••
\J\J, ••••••••••.•••••.
\J/¡.U- ¡'W,¡,ov¡,w5c. WL-LjW¡'';'¡'L-¡Vr¡ Classroom interaction and second language acqUlsltlon ) 1) /
than one purpose. A teacher repetition can occur after a learner error and
control which it gives the teacher over the discourse. Thus, a question is likelY
serve as a model for imitation, or it can function as a reinforcement of a Cor-
to occupy the first part of the ubiquitous three-phase IRF exchange. Ques-
rect response. Teachers often fail to indicate where or how an utterance is de-
tions typically serve as devices for initiating discourse centred on medium-
viant. They respond positively even when the learners continue to make the
orientated goals, although they can also serve a variety of other functions.
error. They correct an error in one part of the lesson but ignore it in another. Much of the work on questions has centred on developing taxonomies to
They may give up on the task of correction if learners do not seem able to describe the different types. In one of the earliest taxonomies, Barnes (1969;
cope. Nystrom (1983) sums it all up with this comment: 1976) distinguished four types of questions he observed in secondary school
teachers typically are unable to sort through the feedback options available classrooms in Britain: (1) Factual questions ('what?'), (2) Reasoning ques-
to them and arrive at the most appropriate response. tions ('how?' and 'why?'), (3) Open questions that do not require any reason-
ing, and (4) Social questions (questions that influence student behaviour by
According to some (for example, Allwright 1975), inconsistency is inevitable means of control or appeal). Barnes made much of the distinction between
and even desirable, as it reflects the teacher's attempts to cater for individual tWOtypes of reasoning questions: those that are c/osed in that they are framed
differences among the learners, but according to others (for example, Long with only one acceptable answer in mind, and those that are open beca use
1977), it is damaging. These differences in opinion reflect the relative em- they permit a number of different acceptable answers. Barnes also points out
phasis given to 'cognitive' and 'affective' aspects of feedback. that many questions have the appearance of being open, but, in fact, when the
Given the amount of research that has been devoted to error treatment, the teacher's response to a student's answer is examined, turn out to be closed; he
lack of studies that have investigated its effect on acquisition is disappointing. calls these pseudo-questions.
Chaudron (1986) reports that only 39 per cent of the errors treated in the im- Kearsley (1976) provides an extensive taxonomy of question types based
mersion classroom he studied resulted in successful student uptake (i.e. were on conversational data. Long and Sato (1983) made use of this framework in
eliminated in the next student utterance). This might suggest that error treat- their study of ESL teachers' questions, but found ir necessary to make a num-
ment is often not successful, but it is possible that by raising learners' con- ber of changes to include new categories to accommodate questions not ac-
sciousness it contributes to acquisition in the long run (see Chapter 14 for a counted for by Kearsley's categories, and to eliminate other categories not
discussion of the possible delayed effect of formal instruction). Ramirez and exemplified in their classroom data. Their taxonomy can be found in Table
Stromquist (1979) report a positive correlation between the correction of 13.5. Ir centres on the distinction between echoic questions, which ask for the
grammatical errors and gains in linguistic proficiency, but this study did not repetition of an utterance or confirmation that it has been properly un-
examine the relationship between the correction and elimination of specific derstood, and epistemic questions which serve the purpose of acquiring in-
errors. formation. The latter type include referential and display questions, which
At the moment, therefore, opinions abound about what type of error treat- Long and Sato discuss in some detail. This distinction is similar but not ident-
ment is best, but there is little empirical evidence on which to make an in- ical to the open/closed distinction of Barnes. Referential questions are genu-
formed choice. Currently, two recommendations seem to find wide approval. inely information-seeking, while display questions 'test' the learner by
One is that error treatment should be conducted in a manner that is compat- eliciting already known information, as in this example:
ible with general interlanguage development (for example, by only correcting
T: What's the capital of Peru?
errors that learners are ready to eliminate). The other is that self-repair is S: Lima
more conducive to acquisition than other-repair, as it is less likely to result in T: Good.
a negative affective response (Van Lier 1988). These recommendations await
empirical support. Referential questions are likely to be open, while display questions are likely
to be closed, but it is possible to conceive of closed referential questions and of
open display questions.
Teachers' questions Other taxonomies have focused on other aspects of teachers' questions.
Teachers, whether in content classrooms or in language classrooms, typically Koivukari (1987), for example, is concerned with depth of cognitive pro-
ask a lot of questions. For example, in three hours of language-content teach- cessing. Rote questions (those calling for the reproduction of content) are
ing, J ohnston (1990) observed a total of 522 questions of various types. Long considered to opera te at the surface level, while two kinds of 'comprehension'
and Sato (1983) observed a total of 938 questions in six elementary level ESL questions (those calling for the reproduction of content and those calling for
lessons. One reason for the prevalence of questioning is undoubtedly the the generation of new content) operate at progressively deeper levels.
•••..••.•••
..,...,., \.PoJ"" .JvvVIl-U H.4H5UUót;:; ULLJUl~ltlUn
Classroom mteractlOn and second language acquiSltlon .) ~~
Type Sub-category Example In one sense the teacher's question is referential as it concerns an area of the
student's private life she has no knowledge of, but in another sense it can be
1 Echoic a comprehension AII right?; OK?; Does everyone understand considered display, as it is clearly designed to elicit a specific grammatical
checks 'polite'? structure and was evaluated accordingly. This question suggests the need for
b clarification requests What do you mean?; I don't understand; What?
e confirmation checks
an additional category, such as Barnes' (1969) 'pseudo-question'. But it can
S: Carefully
T: Carefully?; only be considered 'pseudo' if we know that the teacher intended the question
Did you say 'he'? to be of the display variety-something that we cannot be sure about, as
2 Epistemic a referential Why didn't you do your homework? teachers may ask a referential question only to treat the students' responses in
b display Wha1's the opposite of 'up' in English? the same manner as they would their responses to a display question. The
e expressive l1's interesting the different pronunciations we above exchange is certainly an example of what McTear (1975) called
have now, but isn't it?
d rhetorical 'pseudo-communication'. Its prevalence in many classrooms makes the
Why did I do that? Because ...
coding of individual questions problematic to a degree not always recognized
by researchers.
Table 13.5: A taxonomy of the fltnctions of teachers' questions (from Long and Studies of teachers' questions in the L2 classroom have focused on the fre-
Sato 1983; based on Kearsley 1976)
quency of the different types of questions, wait time (the length of time the
teacher is prepared to wait for an answer), the nature of the learners' output
Hakansson and Lindberg (1988), in an analysis of questions in Swedish, dis- when answering questions, the effect of the learners' level of proficiency on
tinguish questions according to form and cognitive level as well as their com-
questioning, the possibility of training teachers to ask more 'communicative'
municative value and orientation. There are three formal categories: nexus
questions, and the variation evident in teachers' questioning strategies. Much
questions (questions that can be answered 'yes' or no'), alternative questions of the research has been informed by the assumption that L2learning will be
(that provide the responder with an alternative to select from), and x-ques- enhanced if the questions result in active learner participation and meaning
tions (where there is an unknown element, as in wh-questions). At the cog- negotiation.
nitive level, questions are distinguished according to whether they relate to Barnes (1969) found that the secondary school teachers he investigated
cognitive memory (i.e. they require some kind of reproduction of informa- showed a clear preference for closed reasoning-type questions. Open reason-
tion), convergent thinking (i.e. they require the 'analysis and integration of ing-type questions, where the teacher accepted a number of answers, were
given or remembered data within a tightly structured framework', 1988: 77) rare. Barnes argues that the teachers saw their role as 'more a matter of hand-
or divergent/evaluative thinking (i.e. they require data to be generated freely ing over ready-made material, whether facts or processes, than a matter of en-
and independently). The categories relating to communicative value reflect couraging pupils to participate actively and to bring their own thoughts and
the referential/display distinction, while those relating to communicative recollections into the conversation' (1969: 23). Closed questions are indicat-
orientation concern whether the question is focused on the language itself ive of a transmission style of education; Barnes argues in favour of a more in-
(the medium) or on real-life topics (the message). This latter distinction is terpretative style. Long and Sato (1983) adopt a similar position. They found
viewed as continuous rather than dichotomous. Hakansson and Lindberg's that the ESL teachers in their study asked far more display than referential
taxonomy is probably the most comprehensive yet devised for the language questions (476 as opposed to 128). This contrasts with native-speaker beha-
classroom. viour outside the classroom where referential questions predomina te (999 as
Whereas there are few problems in assigning teachers' questions to formal opposed to 2 display questions in the sample they studied). They conclude
categories, difficulties do arise with functional, communicative, or cognitive that 'ESL teachers continue to emphasize form over meaning, accuracy over
categories. These are 'high inference' and often call for substantial inter- communication' (1983: 283-4). Other studies (for example, White and
pretative work on the part of the analyst. It is not always clear-cut, for ex- Lightbown 1984; Early 1985; Ramirez et al. 1986; Johnston 1990; White
ample, whether a question is referential or display, as this example from 1992) also indicate that display/closed questions are more common than re-
White (1992) illustrates: ferential/open questions in the L2 classroom.4
T: How long have you worn glasses? How long have you had your glasses? Teachers also seem to prefer instant responses from their students. White
S: 1have worn these glasses for about six years. and Lightbown (1984) found that the teachers in their study rarely gave
T: Very good. Same glasses? enough time for students to formulate answers before repeating, rephrasing,
590 Classroom second language acquisition Classroom interaction and second language acquisition 591
or redirecting the question at another student. The shorter the wait time, referential questions with a high-Ievel class and more display questions with
however, the fewer and the shorter the student responses. a low leve! class, but the other teacher in this study followed the opposite
One way in which teachers' questions might affect L2 acquisition is in pattern .....
terms of the opportunities they provide for learner output. In Cha pter 7 (page Given the indetermmate nature of the findmgs of many of the studIes, It
282) we considered the comprehensible output hypothesis, according to might seem premature to prescribe questioning strategies in teacher educa-
which 'pushed output' helps learners to reconstruct their interlanguages. A tion. A number of studies, however, have investigated the effect of training
key issue, therefore, is whether teachers' questions cater for such output. teachers to ask specific types of questions. Brock (1986) and Long and
Brock (1986) found that responses to referential questions (mean length = 10 Crookes (1987) found that instructor s given training in the use of referential
words) were significandy longer than responses to display questions (mean questions did respond by increasing the frequency of this type of question in
length = 4.23 words) in four advanced ESL classes at the University of their teaching. Koivukari (1987) found that training led to teachers using
Hawaii. Similar results have been obtained by Long and Crookes (1987), more 'deep' comprehension questions and fewer superficial rote questions,
Nunan (1990a), and White (1992), suggesting that the findings are fairly and was also able to demonstrate that an experimental group who benefited
robusto However, as White illustrates, it does not follow that al! display from this treatment showed improved comprehension scores.
questions produce short responses. It is obvious that 'there is no meaning left Finally, several studies have pointed to the necessity of acknowledging in-
to negotiate' in exchanges such as the following: dividual variation in teachers' questioning strategies. White (1992), for in-
T: What's this? stance, found very different patterns of questioning in his two teachers. Other
S: !t's a cup. studies by Long and Sato (1983), Long and Crookes (1987), Koivukari
T: Good. (1987), and Johnston (1990) also report extensive differences, although not
(Long and Crookes 1987) all these researchers bothered to draw the reader's attention to them expli-
cidy. Banbrook and Skehan (1990) provide illustrative evidence to argue that
However, this is not the case in the following exchange based on a reading there is both intra- and inter-teacher variation. They identify three sources of
comprehension lesson:
intra-teacher variation: '(a) general teacher variation, (b) variation that takes
T: Did anyone manage to find some reasons for this?
place over the phases of the lesson and (c) variation in question asking ... that
S: With the decline of religion there is no pressure on woman to get is the consequence of the teaching tasks or activities engaged in' (1990: 150).
married.
They argue that although variation in teachers' questions is well attested, its
(White 1992:.26)
parameters are not yet well understood.
In both cases, however, the exchange began with a display question. White These comments by Banbrook and Skehan lead us back to one of the cent-
argues that there is a need for a more delicate categorization of display ques- ral issues in L2 classroom research-the extent to which it should be 'qualitat-
tions to allow for the different types of student response evident in these two ive' and descriptive as opposed to 'quantitative' and experimenta!. Many of
exchanges. Banbrook (1987) provides similar examples to show that ref- the studies discussed above have be!onged to the latter paradigm, but one
erential questions can also elicit responses of varying lengths and complexity. might feel along with Van Lier (1988) that:
It should also be noted that the length of the student's response to a question
the practice of questioning in L2 classrooms, pervasive though it is, has so
is only one of several possible mea sures of learner output. Of equal interest,
far received only superficial treatment ... An analysis must go beyond sim-
perhaps, is the length of the sequence initiated by a question. In this respect,
ple distinctions such as display and referential to carefully examine the pur-
Long and Crookes (1987) report that display questions e!icited more student
poses and the effects of questions, not only in terms of linguistic
turns than referential questions. Clearly, more work is needed to tease out the
production, but also in terms of cognitive demands and interactive purpose
relationship between question type and learner output.
(1988: 224).
Very few studies have examined the re!ationship between teachers' choice
of questions and the learners' proficiency leve!. In Ellis (1985d), I found no It might also stretch to examining the questions asked by learners. Studies by
difference in the use a teacher made of open and closed questions with two Midorikawa (1990) and Robson (1992) suggest that, when learners are given
learners over a nine month period, but I did find evidence to suggest that the the opportunity to ask questions, they automatically elect to use open, ref-
cognitive complexity of the questions changed, with more questions requir- erential-type questions. Rost and Ross (1991) have shown that certain types
ing some form of comment as opposed to object identification evident at the of learner question (for example, 'forward inference' questions in which the
end of the periodo White (1992) found that one of his teachers used more learner asks a question using information that has already been established)
J / L vIU:>:>rUUm secona language acqulsltlon
Classroom interaction and second language acquisition 593
rJ)
rJ) e: rJ) e:
facilitate comprehension and, also, that such questioning strategies are train- ui e: o
able. _;:; ~:8 ~ 1JQ)
c"¡;::
o
~0._ E'"
Q) ~
society. Teachers from such a background ask questions beca use they provide C/)~2ro~ E~~~.~ o.c: ·Q-~ur§
c .•.....cQ3a> °cnctf!.....t o.:!::: ~1a5
_0.0.
.!:+-'+-,o
(f.IUc:_
.2'3 o I.....c () e () :J ctS
~ ;: e: ••... eO IDt:"'OID enQ.
a means by which an expert (the teacher) and a novice (the learner) can jointly ~ ~.~ 8 ~ g) 8. ~ m .g- ~>- '" o ~ Q)
-rJ):¡:¡ .2 en (/)
ID 1:: ~ (/) ~ :~ ~.2> E g e: ~8~~
construct a proposition across utterances and speakers. Poole suggests that
this may be why such questions are so ubiquitous and why they are difficult to
2!
:; .~.~
~.º8 ~~~ §~
e: '"
-o:E
:=c
~ o ~~~
O)::J::J
.- U) en ~~Q5§
~Ot:¡::¡
rJ) ca~
+-'0')
Ern~
~'- c},...Q)a>·_c
g-o rJ)
O
'"
Q) Q)
'"
CO-o (1j tU
""O~::.;::
get rid of; teachers are being asked to reject a strategy that they feel to be cul-
CI)
turally warranted. A corollary of this line of reasoning is that display ques- c;¡ e:
o o 'O
Oi
~-g
•..•
e: e:
"'0 Oh
Q) ~
tions may not be the norm in classrooms where the participants belong to a tU
~ ::J '" Q) '" o (;
~
cñ
rJ) O
culture that does not utilize such caretaking practices. C>
e
tf
~ :~ g'i':'e:
=CiSO
E oe:
en.22
rJ)-Q)
_Q)e:
~ Q) ';::
<::
;::,.
'E
Q)"-
~
.a2
rJ)
~
-"'5"'0
0..0
cnrocn..c::
Q) •
Q)"'o.
rJ) E E
~ g' o
~~§~
~j¿~~
'"
';::
cu U
~
tU
o e: e:
o
Q) ()
0..0 >-~o
o o. o ea ~ g} () •...
S
rJ).c:
Q) Q) ~
~
O
+-'
('Ij
en
c::
en
Q)
ffi ~ ~.~ ~
~t5ge o.E~; .2-50,,:-::,
•...
Given that learners are often restricted to a responding role, it is not surpris- rJ) ~rr o.
E..- o ::J - E·- <::
prehensible output hypothesis in Chapter 7), then it would seem that learning
'" .D li e: O .•..
V¡
.=:!
rJ) '0 "á5-+-'C
.ae:Q)
1J
Q)
o::J
:~
o
.- e:
rJ) O "3
may be inhibited in the classroom. The assumption that participation is im- (J)+::
O rJ) '"
2.Q -g
(ti t5 1...
1J::J
EQ)1J 1J
rJ) e: Q) V¡
portant for learning also underlies several of the studies of motivation dis- .- ('Ij oo 0::J-
eo. ~ '" ::J <::
e e: ~ e 1i) :~ en Q)
~O" -!:)
o .2.& (/J
-e:rJ) ~ó)~m e: ~ ea
~
cussed in Chapter 11. We will consider participation here both from the point :¡:; 0.- ID ~.s ee rJ)
Q)
Q)
O)
::JO)
O
tU
o
"
"'O .~
~ ~o~~ Q)
E 1JQ)e:
Q) .•.•.•.•• ()
Q)
rJ)
rJ)
~~
Q)
0--=_05 \Q
E~5~ e: e:_ ~.c: rJ)
::>
~2Q)'S o Q)
o.~rJ)
rJ) e:
(J) O'c <Y")
There is no clear evidence that the extent to which learners participate pro- tU
Q) 5 ~~ ~ .g ci ~ ~ rJ) Q)
0....1
rJ) ••••• ~.~ 8- ,....
~ g.:~~ o. '"
ductively in the classroom affects their rate of development. Table 13.6 sum- :E ~¿88 Q).c:
0:0 Q)Qj
0:rJ)
::J
Z(J)~
rJ)
~
marizes a number of correlational studies that have examined the .c:~N rJ) rJ) .r: ~
o 1J '" =E
0.0
~ rJ) e:
relationship between amount of learner classroom participation and L2 'O 'c~
O) ~~~ fie
achievement/proficiency. The results are mixed. Whereas studies by Seliger e: LL o", e: rJ) [1! ~ ~~ .
(1977),5 Naiman et al. (1978), and Strong (1983; 1984) report positive cor-
.¡:
:;; :::I~~~
ea·-
'+-
Q)
-e.."1
'" o
rJ) Q)
e:
:;;
-oU)-ow
CC:Nct
m
2! ~ 0C/)j!J. e'crs ~
!.........J
mOOO5-
('Ij
•.
a>
e:
Q) '"
ea
.D
<D
,., '" O) Q) ea
causes learning' or 'proficiency causes participation' explanation is correct 'C
Q)
.'2' E e: Q)
::>
Qj 'ro e >-
'" >-
when a significant relationship is discovered. As we will see in Chapter 15, u¡ (J) z i'i5 O ¡¡:¡
0'1.4'>'>1 uUln :>"'/.-una lunguage acqulsltlon Classroom interaction and second language acquisition 595
when studies investigating the effects of quantity of practice directed at spe- individuallearner differences that affect the degree of anxiety experienced.
cific grammatical structures are considered, there are grounds for believing Gaies (1983b) also argued that negotiation is evident in teacher-dominated
that practice does not make perfect, suggesting that the preferred interpreta_ lessons. He studied the kind of feedback that learners provide in tasks that
tion ought to be that proficiency causes participation. That is, the more profi- required the teacher to describe graphic designs and found evidence of con-
cient the learners are, the more they get to participate. siderable variation among the learners. Ir is important, therefore, not to over-
emphasize the restrictive nature of learner participation in teacher-controlled
Quality of learner participation interaction.
-'- O)
en
""""
(/J
sociolinguistic context in which the task is carried out. Attempts to elassify os o (/J (/J
~ eo E
m m e
-O>
...... ~
o
task variables in this way are helpful, but we are a long way from developing a ;: (/J ~'SO)
o ~
'Ñ >-""
eo
ID 2:;¡ (/)
Z
.5
..c:.~ :J
.¡¡;
O)
m
~.s .
(/J E
o
taxonomy that is both complete and psycholinguisticalIy justified. om o oo .<:
oe
c:
'E
Z
O)
(/J
a. (/J
oa. e 'Ue eO>
Q)
O (Ij._
+-'
1:J
EqualIy important in the study of tasks is the choice of measures of inter- ff 'Ó
e'" ~~:a >-
Q)
"5
ti;
a. "C
f' Q)
~ ~~ ~
Q) ffi ::..
action (the dependent variable). A number of studies have examined the .o em ,Q .~ -g.~ o
>- e
(/J
(/J .crorn
O g>-g
quantity of interactional modifications found in the negotiation of meaning '"
;: ~ ~ ""
~$ ~ "5 ..5
:5
6 O) .0:
(/J
.!B .5 e (f)
a.
.0:
ffi ~
:5c~
ctI
g'
(see Chapter 7) on the grounds that these are important for comprehension, ~ o e e .<:O)C:
o';:: o rn 1J o O Q)
and thereby also for acquisition. One criticism that might be levelIed at such e :2:
"C
O) O)
O>
m
O) a.-o
O) e
o ~ ~.~ e
o '> ti; . ~~~ :g -§ .E 'g. 2
studies is that they ignore important qualitative differences in the way mean- eO)
eo 'Ó
O)
O)
.->
"C
.:.:
(/J
(/J
o e E
-O)
c: .- 2 ~
ro
ea c: -o
o o c :;:
ing is negotiated (Ehrlich, Avery, and Yorio 1989; Newton 1991). A number e e:;¡ o"""':' .5 $9 ee.<: Cñ ;: - (/J m
& m Eos O> '"'
.- Q)~
:.~ , -o :J .::t:. .S?
.•..•
:5 c: m
of studies have also examined to what extent interlocutors provide feedback
15
.?-
§:
,Q
""
Q).S
o>c: 'c ;: §~~ 'O ~~ 00-g
.•... ~
_
o
\,,)
~
that non-native speakers then act on by incorporating corrected features into
e
m "C
(/J
.!B
m
:;¡-
(/J.<:
m m
O)o
' .E~~
o>~(/J
(/J
O)
Q).::t:. (ti
!.".., o
CJ)
ctI .•..•
.•...
e
<:u
.•..•
c: E ~ c:O""
ID
.g m .8 O)~
""~ -o .-e .- E (/J g-ci
e ~;1:J g .::;
their own production. Other aspects of interaction which ha ve been investig- '0, N O>
e :.::¡:c: c: "C ffi(1)2 Cñ
<ti
'" ~
:5 ~ ~
Q)~ E "'-1
.¡¡;
~-d Q)~ o e o e:- .5.E ....¡
ated inelude the number and size of turns, repair (and, more narrowly, self- o
15
e'~ ~ E
O)
o rJ)(3 (/)
ctS
1-
(/J
"O
m -O)
ea.
O)
oo
o .!B.o
:;::
.~
o
.E
o2'~
¡¡¡ c:
;:'" e8
(/J
~~.s c:
E :;:
O
corrections), speaking speed, reference (exophoric v. anaphoric), topic, in- >- O) <ti
eo,"C O> e o O)
ea ctI en
V)
"O
(/)
m
(/J
e-
O)
O)
O>
e c:
O)
c:
.-
o~a.
:¡:¡ox (/J
O)
O)
~ ~~~~
O O O 2
<:u
-e
:1:: "C :;>
"Co
~ C/J.~
c: (/J
15 ,Q0 e o
m .-
"Oro~
e"C1- ~~
m c:
c: c5 c: ~ m
o o
~
Table 13.7 summarizes the main studies that have investigated task vari- .2
-oID ",,& .Q.!!2 ID V)
J!!
"'5
O)
o
em o e m
o
om oo
.oC:
e
(/J
~
Oí
:;:;
O .~ ffi~1::
ea u el) Q)
-O>
(ij'(ñ "ª~~ g
'O oo..c: ID:J
~ .':!
"-
ables. From this it can be seen that several studies have examined the effects <ti
~~ <;=
'c ~ ~ ~'§~ O>~;: g
~ E "C'-
o '-..c o E :s
e "c.<:
Ci> ID O
(; ~ ~'5
o.. ID O
0)-
o o ~ 2c: ~ "'0._
Q) :;:~ +-'
~
\,,)
posed to those that require a two-way exchange. Examples of the former in- ::E ~~ <DO
(/)Z Z
(/)<D
Z~ -' :2: :2w~á3 ct~ o o ~g-
~O~5 .•... ~
'...:::
<:u
elude giving instructions and telling a personal story, while an example of the a. a. ...... a. N ~ <:u
latter is types of communication games in which the participants each hold m ~ m "C m
:if?
.•..•
O> (i':': O> O) O> C()
eo ~~ e § ~2~ g? g? @"M' .~ C: .::;
part of the information needed to complete the task. The essential difference
lies in whether the exchange of information is optional 01' required. A number
~
E
0-
.=..-c
~.º
.2
Oí
E
;:~
.Q ~
~.~
m o o
_(/J_
g> E'~
á3 ~
c.:J
:¡=o-
.::t:. á
'5 ~
tU o
'¿;
.~
,Q .;:¿u o ~~ +->
UJ_
.::::t:.
~ E
c- o
+-' :c ~ ~ ~~ ;g.~ ~~ .•..•
V)
of studies (Long 1980a; Doughty andPica 1986; Newton 1991) show that .0:
'"'
(/J
m
~
O)
:5
'"' 2 f2 . ID .5 ~ -g
O) ~
a>
(/J
:¡e fO'.s.-gE
<:u
;:,
two-way tasks result in increased negotiation of meaning. Gass and Varonis m 5-g~ ~ ~o~ cag> .::;
~~ m .::t:.":::
o o
i
;: fU' ¿; ~.:: ID._
~
o "S: .2 6
.- m m > ;:
g ~.~
{/) C"')
~:;~;:~.s~~
(j) (fJ
V)
(1985a), however, found no difference in the number of indicators of non- <ti
(5C rn.::::t:.-;:a6 <:u
Q)
:c ~ (/J
, .<:
O) ~ <v ~ ~ ~ ~
(/J
> C·¡;:
O> O)
~~2;~-g~é\J~
understanding (a measure of meaning negotiation) in the two types of tasks. '"
(/J
> E- (/J
> ~'Ec'E(/J g>~ ~ .Q ~.~ ~'5. ~ ~ T~ .•...
.¡:
'" >- ~~
.0- '"' o ID.2 ID > :E~~ g: 2 E ~~ ~ 9-.§ ~
VJ
They suggest that the distinction between one-way and two-way tasks is bet- > m m
~ ~ro ~ f; (0)-
.:.: ;: (/J e5: ;: (/J m-M E"C~ wr3>
a. c:.- c:
Q)1ij-c
o 'S
Il..:.: _~ ~e Eo >O
e ;:e (/) Q).~- C> ID ID r.:.:
ter seen as continuous rather than dichotomous, which 'makes comparison a <ti
e (/J _(/J
0)"" 0)""
e - o"C
~~~
o.!B -0- Su.!: 3-~ ~ ~E-g
Q).::::t:.
'" (/J
IV)
1- o.!B s.s Su.S () 0$9
CJ)
ES1ffi sE o s'u) Q) ,....
complex process' (1985a: 159). However, Long (1989) considers the results <:u
of the research sufficiently robust to elaim that 'two-way tasks produce more I
(/)
"C
em :¡;
Z (/J (/J (/J ...J ~
negotiation work and more useful negotiation work than one-way tasks' <D (/J "C
m
"O
m
(/J "C
-'
(1989: 13). ~ "O
m >- >- "C m
>-
¡B
"O
e
m
'"' "O
M (/)
"O (/J
"O
m
'"'
"C
"C
(/)
(/)
w e O)
(/)"C
(/J "C
(/) Z Z (/J
'O o
This is one of three conelusions regarding task-types that Long arrives at in <ti
Z m Z c: m
-t)'Z
(/)
Z
"O
m
(/J
'O I'"'Z .;¡; Z '"' (/J
a.~(/J ~(/J
his own survey of the research. The second is that 'planned tasks "stretch" in- Q) (/)"C
Z(/)
I
(/)
(/)"C ch
Z.!!! Z
(/)
Z Z ch "C '"
o e O)
o ~
"CO)
:.o ch
Z Z Z (/)
a,~ m E
terlanguages further and promote destabilization more than unplanned
::1
~ZZ Z Z ~ LO Z Mm
en
<D
Z O')~ ••.
Z C\J
Z C\J~ ~~
tasks' (1989: 14). This is supported bytask-based studies oflearnervariability
m
lO
(see Chapter 4) which show that learners tend to produce more complex and am
(/J
a:>
O')
"C
C:lO -g ~ a;
a;
O')
.:.: m a:> ea c.o a;
more target-like language when they ha ve time to plan their output. It should >.
a:>
~
O')
.g
"C~
c: (/J (/JO')
O)~
<O
c:o >.CO
T"""
.:::t:. ~
O')
eo O')
m 'e: ""c: ~ 1: ~ .2 c:
;: (/J
be noted, however, that it is not yet elear whether performing in planned tasks "C
::1
O>
e g>~ (/J o o o o ~ O)
e
-'o o~m dS (/J
m
~
m e-s :s ~ .~ ~ O) o
subsequently helps learners to perform better in unplanned tasks. ' Uí I-LL~ (!» oa: O O Il. CD ID Z ...,
598 Classroom second language acquisition Classroom interaction and second language acquisition 599
Long's third conclusion is that 'closed tasks produce more negotiation difference in the overall quantity of interactional adjustl11ents in a one-way
work and more useful negotiation work than open tasks' (1989: 16). An open task perforl11ed in a lockstep setting and in small-group work, but when they
task is one where the participants know there is no predetermined solution·
replicated this study using a two-way task (Pica and Doughty 1985b), they
examples include free conversation, debate, ranking activities, and suggestin~ did find significant differences. The overall results of these two studies are
preferred solutions to problems. A closed task requires the participants to shown in Table 13.8. They indicate that there is an interaction between the
reach a single, correct solution or one of a small, finite set of correct solutions·
participation pattern and the type of task. Thus, group-work only results in
examples include identifying the differences between two pictures and identi~ more negotiation of l11eaning if the task is of the required-information ex-
fying the perpetrator of a crime. Studies by Duff (1986) and Berwick (1990) change type.
lend support to Long's conclusion. Long claims that closed tasks result in Participation pattern
Task
more topic and language recycling, more feedback, more feedback incorpora- Teacher-fronted Group
tion, more rephrasing, and more precision. However, as F. Jones (1991) n % n %
points out, more open-ended tasks may afford learners greater interactive 347 49 145 40
freedom to practise conversational skills such as topic selection and change. Optional informatíon exchange
385 45 400 66
The work so far has helped to identify a number of task variables that affect Required information exchange
participant performance, but it is probably premature to reach firm conclu- % = % of T -units and fragments combined.
sions. In particular, it is still not clear which task variables are the most im-
Table 13.8: Comparison o( modified conversational interaction generated by
portant ones, and even less clear what effect combining different sets of
optional vs required in(ormation exchange tasks on teacher-(ronted and group
variables has. Also, little is known about how the performance elicited by dif- participation pattems ((rom Pica and Doughty 1988: 51)
ferent tasks affects acquisition (see Chapter 7). Thus, despite Long and
Crookes' (1987) claim that 'it should be possible to build up a multi- As reported in Chapter 7, several other studies confirm the value of in-
dimensional classification, organizing tasks in terms of their potential for terlanguage talk as a source of opportunities for meaning negotiation (for ex-
second language learning on the basis of psycholinguistically and psycho- ample, Gass and Varonis 1985a and Porter 1986). Rulon and McCreary
logically-motivated dimensions', it is likely to be a long time before such a sys- (1986) investigated the effect of participation pattern on the negotiation of
tem becomes available. content, defined as 'the process of spoken interaction whereby the content of
a previously encountered passage (aural or written) is clarified to the satisfac-
tion of both parties' (1986: 183). They found little difference between small-
Small-group work and interaction
group and teacher-Ied discussions with regard to length of utterance, syn-
Another area of L2 research of considerable pedagogic interest is interaction tactic complexity, or interactional features, but they did find that significantly
in small groups. Group work is often considered an essential feature of com- more negotiation of content occurred in the small-group discussions.
municative language teaching (see Brumfit 1984). Long and Porter (1985) From these studies it seems reasonable to conclude that interaction be-
summarize the main pedagogical arguments in favour of it. Ir increases lan- tween learners can provide the interactional conditions which have been hy-
guage practice opportunities, it improves the quality of student talk, it helps pothesized to facilitate acquisition more readily than can interaction
to individualize instruction, it promotes a positive affective climate, and it involving teachers. However, a word of caution is in order. Both Pica and
motivates learners to learn. In addition to these pedagogic arguments, a psy- Doughty, and Porter, show, not surprisingly, that interlanguage talk is less
cholinguistic justification has been advanced: group work provides the kind grammatical than teacher talk. Ir is possible, therefore, as Plann (1977) has
of input and opportunities for output that promote rapid L2 acquisition. suggested, that exposure to incorrect peer input may lead to fossilization.
A number of studies have compared the interaction in teacher-centred les- However, two of Porter's findings give reason to believe that this may not be
sons with that found in group work. Long, Adams, McLean, and Casta nos the case. She reports that when learners corrected each other's error s, they did
(1976) report that students working in small groups produced a greater so wrongly only 0.3 per cent of the time and also that only 3 per cent of the
quantity of language and alsó better quality language than students in a errors the learners produced could be attributed to repetition of a fellow-
teacher-fronted, lockstep classroom setting. Small-group work provided learner's error. In general, therefore, learners do not appear to be unduly dis-
more opportunities for language production and greater variety of language advantaged by exposure to deviant input frol11 other learners. In one respect,
use in initiating discussion, asking for clarification, interrupting, competing though, interlanguage talk may be inferior. Porter found that non-native
for th~ Boor, and joking. Pica and Doughty (1985a), however, found no speakers did not provide sociolinguistically appropriate input and suggests
600 Classroom second language acquisition Classroom mteractlOn ana secona language acqul~¡llUn ov ~
that learners may not be able to develop sociolinguistic competence frOI11 as opposed to 17.5 per cent). Where mixed ability pairings are involved,
each other-a point that is taken up in the following section. therefore, success (and perhaps also acquisition) is more likely if the lower-
These studies were guided by Long's hypotheses regarding the importance levellearner is in charge of the key information that has to be communicated.
of interactionally modified input. Bygate (1988), however, like Hatch To sum up, the research suggests that learners will benefit from interacting
(1978a), emphasizes the role of learner output in collaborative discourse Con- in small group work. They will ha ve more opportunity to speak, to negotiate
struction. He suggests that group work facilitates acquisition by affording meaning and content, and to construct discourse collaboratively. This may
learners opportunities to build up utterances through the use of satellite units. help acquisition. They will also be exposed to more ungrammatical input.
These are words, phrases 01' clauses that constitute either moodless utterances Whether this has an adverse effect is not yet clear. Ir is possible, however, that
that lack a finite verb 01' some kind of syntactically dependent unit, as in the small group work does not aid the development of sociolinguistic compet-
following example: ence. Less is known about the ideal composition of small groups, but studies
SI: at the door suggest that mixed gender and mixed proficiency pairs may be optima!.
S2: yes in the same door 1 think
S3: besides the man who is leaving Summary
S2: behind him.
In this section we have considered a number of aspects of L2 classroom dis-
course. Given the diversity of classroom settings and the breadth of research
Bygate found numerous examples of such units in the speech produced by
learners working in small groups. He argues that the use of satellite units al- now available, it is difficult to arrive at a general picture. The following state-
ments constitute an attempt to do so, but should be treated circumspectly:
lows for flexibility in communication, allows the learner time to prepare
messages, and allows messages to be built up collaboratively thus helping to 1 L2 classroom discourse often manifests a well-defined structure, con-
extend learners' capabilities. However, as Bygate provides no comparable sisting of IRF(R). Different types of language use are evident, reflecting
data from teacher-controlled lessons, it is not clear whether satellite units are in particular whether the focus of an interaction is on learning the L2 01'
a special feature of group work. FéErch's (1985) examples of vertical con- communicating in it. L2 classroom discourse is likely to differ from nat-
structions in interactions controlled by the teacher suggest that they might ural discourse in a number of ways, including the rules and mechanisms
not be, although it is probable that they are at least more frequent in in- for turn-taking.
terlanguage talk. 2 Teachers adapt the way they talk to classroom learners in similar ways to
Other stud.ies of non-native speaker-non-native speaker interaction have foreigner talk. They also modify their language in accordance with the
focused on the participant variables that affect the quality of the talk. One learners' proficiency leve!.
variable that has been found to be important is sex (see Chapter 6). Another 3 Error treatment constitutes an interactional event of considerable com-
variable that has been investigated is the proficiency level of the learners. Por- plexity. T eachers do not correct all their students' errors and they are of-
ter (1986), in the same study referred to above, investigated the input and out- ten inconsistent in whether they correct and which errors they correcto
put in non-native speaker dyads of mixed proficiency levels. Not surprisingly, Learners often fail to incorporate the teacher's corrections in their sub-
intermediate learners got more input and better quality input from advanced sequent responses.
than from other intermediate learners. Conversely, advanced learners get 4 Teachers vary considerably in the number and the type of questions they
more opportunity to practise when they are communicating with intermedi- ask. Several studies have shown that display/closed questions predomin-
ate learners. Porter feels that mixed pairings offer something to both sets of ate over referential/open questions. The latter type may result in more
learners. In a very careful study involving a task that required the resolution meaning negotiation and more complex learner output.
of a number of 'referential conflicts' (the subjects were given maps that dif- 5 Studies investigating the relationship between learner participation and
fered in a number of ways), Yule and McDonald (1990) examined the effects L2 proficiency have produced inconsistent results. Both a 'participation
of proficiency in mixed-Ievel dyads, where in some interactions the sender of causes proficiency' and a 'proficiency causes participation' explanation
the information was of low proficiency and the receiver high proficiency need to be considered, with the latter more likely.
(L>H) and in others the opposite (H>L). They found that the L>H inter- 6 The quantity and quality of interaction varies according to task. AI-
actions were at least twice as long as the H>L interactions. Furthermore, though much work needs to be done to establish how specific task char-
negotiated solutions to the referential problems were much more likely to acteristics affect interaction, there is some preliminary evidence to
take place in the L>H condition than in the H>L (a 67.5 per cent success rate suggest that tasks which require information exchange, allow learners to
602 Classroom second language acquisition Classroom interaction and second language acquisition 603
plan their output, and have a limited number of possible outcomes result
1 guage teaching. This project, which was conducted in a number of
in interaction which facilitates acquisition.
s~:ondary schools in Bangalore and Madras with beginner learners of L2
7 Small-group work has been found to provide more opportunities for
English, was evaluated by Beretta and Davies (1985). Although not al! the
meaning negotiation than lockstep teaching, if the tasks are of the 're- sults showed an advantage for the project schools over the control schools,
quired-information exchange' type. The quality of interaction also ap- r\ich were taught by means of the strucrural-oral-situational method,
pears to be enhanced if the learners comprising the pair/group are ~avies and Beretta interpret the resu/ts as 'being, on the whole, positive' and
heterogenous with regard to sex and proficiency leve!. Group work may conclude that 'they provide tentative support for the CTP'. Finally, Light-
not be the best way to develop sociolinguistic competence.
bown (1992) reports on an interesting project in New Brunswick, in which
Ir needs to be emphasized that the studies considered in this section have been Canadian French children in grades 3-6 were taught English by listening to
descriptive in nature. They provide no direct evidence to show any link be- tapes and following the written texto Results at the end of the third year of the
tween particular aspects of classroom discourse and L2 learning. In many project showed that 'students in this program have succeeded in learning at
cases, however, they have been informed by theories of L2 acquisition, and it least as much English as those whose learning had been guided by the teacher
is on the basis of these that claims regarding the facilitative or debilitative ef- in a more traditional program' (1992: 362). They were as good even at speak-
fects of certain kinds of interaction have been based. ing English. The program also resulted in very positive student attitudes.
These studies, therefore, demonstrate that the communicative classroom is
The relationship between classroom interaction and second effective in promoting L2 acquisition. However, it should be noted that in all
languagelearning these studies the learners were at an elementary leve!.
Other studies suggest that communicative classrooms may not be so suc-
We shall now consider the few studies that have investigated the relationship cessful in promoting high levels of linguistic competence. Krashen (1982) has
between classroom interaction and L2 learning. We will begin with studies claimed that immersion classrooms have succeeded in developing very high
that have examined whether successful L2learning is possible in a favourable levels of L2 proficiency, but there is growing evidence that they are not quite
classroom environment and then move on to look at studies that have tried to so successful as Krashen claims. Researchers ha ve for some time recognized
establish direct links between features of interaction and learning. that immersion learners generally fail to acquire certain grammatical distinc-
tions (see Chapter 7). Other studies also suggest there may be limitations on
Second language learning in the communicative classroom what can be achieved in communicative classrooms. Spada and Lightbown
(1989) found that an intensive ESL course (5 hours a day for 5 months),
A number of scholars have proposed that the most effective way of develop-
which was taught by means of communicative methods emphasizing tasks
ing successful L2 competence in a classroom is to ensure that the learners have
leading to natural interaction, produced little evidence of syntactic develop-
sufficient opportunities to participate in discourse directed at the exchange of mento For example, the students were only 50 per cent accurate in their use of
information (see Krashen 1982; Swain 1985; Prabhu 1987). According to
plural-s and only 20 per cent in the case of V + -ing. My study of the requests
this view, the failure of many classroom learners derives from the lack of com-
produced by two classroom learners (Ellis 1992a) also suggests that the com-
prehensible input and/or comprehensible output. One way of investigating municative classroom may not be well-suited to the achievement of socio-
this claim is by studying to what extent a communicative classroom environ-
linguistic competence. 1 argue that classroom learners may experience
ment results in successful L2 learning.
interpersonal needs to perform speech acts such as requests, and also express-
There is now convincing evidence that learners can learn 'naturally' in a
¡ve needs that lead them to do so in a varied ways, but that they do not experi-
communicative classroom setting. We have already considered Hammond's
ence any sociolinguistic need to modify the way they perform requests in
(1988) study (see page 571). Terrell, Gomez, and Mariscal (1980) showed
accordance with situational factors. This is beca use 'the classroom consti-
that elementary learners of L2 Spanish can successfully acquire various ques-
tutes an environment where the interactants achieve great familiarity with
tion forms simply as a result of being exposed to questiol1s in the input. 74 per each other, removing the need for the careful face-work that results in the use
cent of 7th grade students' questions and 82 per cent of the 8th and 9th grade
of indirect request-types and extensive modification' (1992a: 20).
students' questions were correctly formed even though no explicit instruction One interpretation of the research on communicative classrooms is as
had been provided. Prabhu (1987) developed a programme known as the fol!ows:
Communicational Teaching Project (CTP), which had as its aim the
development of linguistic competence through a task-based approach to
"---"'~W~~I VVII~ .J\,... •••..Vf~i,..i. ~b•...•..
~••."'~6~·L•.. 1~···....
••....•.. ·••·•.'-",· Gtassroorn mreracfton ana secona tanguage acqulsltLOn bU)
6 Learner participation