Language Teaching Research 11,3 (2007); pp.

281–299

The ZPD and whole class teaching:

Teacher-led and student-led interactional
mediation of tasks
Iju Guk Heuimangdae Elementary School and
David Kellogg Seoul National University of Education,
South Korea

Vygotsky’s ‘zone of proximal development (ZPD)’ has become associated with the
individual ‘scaffolding’ of learners. As a result, because teachers need to teach the
whole class, many public school teachers have had to dismiss the concept as
unworkable. Yet Vygotsky himself was chiefly concerned with public school teaching
and firmly rejected the idea of a ‘pedagogical duet’ between learner and teacher. He
also dismissed the teacher who attempted to provide the ensemble of learning content
by him or herself as a ‘rickshaw puller’, arguing instead that a teacher should be a
‘tram driver’, who organizes the social environment of learning. One way in which the
teacher might do this is by mediating a learning task for a single learner or a group of
learners, who then mediate the task for their classmates in group-work. We present
evidence that in this situation the way in which learners mediate tasks differs from the
way in which teachers do, and argue that this suggests learner-to-learner mediation is
in important ways closer to what Vygotsky termed ‘internalization’. We believe that
T-S and S-S interactional mediation do not create two different ZPDs but may instead
lie within a single, whole class ZPD.

I Introduction: How many learners in a ZPD?

Faced with whole classrooms of children rather than single learners, teachers
have often found it difficult to apply a concept Vygotsky apparently consid-
ered central to schooling, namely the ‘Zone of Proximal Development’ (ZPD)
(Wegerif and Mercer, 1997: 49; Mercer, and Fisher, 1997a: 20). Mercer and
Fisher conclude:
(T)he ZPD seems to us to have limited applicability in research directly concerned with the
quality of teaching and learning in classrooms. One obvious reason is that practical cir-
cumstances force most teachers to plan activities on the scale of classes or groups, not indi-
viduals. The notion of any group of learners having a common ZPD seems problematic.
(1997b: 209)

Address for correspondence: David Kellogg, Department of English Education, Seoul National
University of Education, 1650 Beonji Seochogu Seochodong, Seoul 137 742, South Korea; e-mail:
[email protected]

© 2007 SAGE Publications 10.1177/1362168807077561

282 The ZPD and whole class teaching

Askew et al. (1995) complain that Vygotsky’s ‘practical implications for

teaching are difficult to detect’, and thus operationalizing the ZPD as ‘scaf-
folding’ has been left to interpreters such as Bruner, whose work centered on
individual learners performing tasks such as assembling a toy pyramid. These
researchers tend to work one on one, rather as parents do, rather than with
whole classrooms, as teachers must.
Yet Vygotsky worked his whole life in such classrooms, and as a Soviet
educator, was centrally concerned with the problems of public schooling. So
it seems unlikely that an unrealizable student-by-student ‘scaffolding’ con-
ception should be the core of his pedagogical thinking. What seems much
more plausible is that Western epigones may have removed the social context
and the socialist content from Vygotsky’s original idea. Kinginger (2002)
describes three ways in which Vygotsky’s ‘great unfinished’ concept has been
co-opted by foreign language teaching in the USA: as a simple truism about
task appropriacy, as a scaffolding procedure, and as a framework for unpre-
dictable opportunities for metalinguistic learning.
How is it that a concept thought so central to Vygotsky’s work could receive
such varied interpretations? Poehner and Lantolf point out that it crops up in
at least three very different contexts with slightly different meanings (2005).
Wertsch (1985) and also van der Veer and Valsiner (1993) blame Vygotsky’s
lack of specification and deny that the ZPD was a central element of his think-
ing at all. Gillen (2000) claims that the ZPD did not even originate with him
but instead was part of a general enthusiasm for finding the social roots of phe-
nomena that accompanied the Bolshevik Revolution. Other writers, such as
Wells (1998) and Nassaji and Swain (2000), place the ZPD at the very core of
Vygotsky’s work. Surely the fact that no one since Vygotsky has been very
successful in operationalizing the concept is some evidence of its originality!
We take the somewhat perverse position that the ZPD both is and is not
original to Vygotsky, and that it is both central and peripheral to his thinking.
The general law on which the ZPD is based, that every higher psychological
function within the mind was once a concrete, social relationship between
minds, belongs to Pierre Janet, and in fact Vygotsky himself refers to it as
‘Janet’s Law’ (Valsiner and van der Veer, 2000: 370). But the distinction that
Vygotsky makes between ‘learning’ and ‘development’, the insistence that
the former leads the latter, and the application of this concept to a public
schooling context were Vygotsky’s own contribution. Similarly, the fact that
Vygotsky only mentions the ZPD sporadically and towards the very end of
his life suggests that it had not occupied a central place in his thinking, at least
not under that name. But the fact that Vygotsky considers the ZPD to be
absolutely essential to the understanding of school life, and the fact that
school life was the central focus of his work in psychology, suggest to us that
the ZPD, which is in every respect of a piece with the whole of his life’s work,
was earmarked to play a central role in his thinking and is therefore destined
to play a central role in the thinking of teachers who would follow his lead.
It is true, and it is not at all fortuitous, that the passage of Vygotsky’s work
on the ZPD that is mostly widely cited by Western scholars develops the
 Iju Guk and David Kellogg 283

concept as a response to a ‘thought experiment’ in individual testing rather than

collective teaching. He hypothesized two disabled 10-year-olds showing a
‘mental age’ of eight. With assistance, he surmised, one child might show that
he could solve problems up to a 12-year-old level, while the other could only
manage problems at the 9-year-old level, and the varying difference between
what they could do without assistance and what they could do with assistance
was held to be a better predictor of development then IQ tests. This testing con-
text was not central to Vygotsky’s oeuvre. Instead of a testing construct, it might
be more accurate to see the ZPD as a criticism of the limitations of testing (e.g.
1997b: 237), and most particularly his criticism of one of the pioneers of test-
ing, namely the American psychometric enthusiast Thorndike.

II Vygotsky vs. Thorndike

At the crux of Vygotsky’s disagreement with Thorndike is a largely unre-
marked aspect of his famous definition of the ZPD, far more central to the
concept than its testing context, to wit: the concept of mental age. The con-
cept of mental age is at once individual and generalizable within a classroom.
Vygotsky is scrupulous to exclude chronological age as a reliable guide to
this (1998: 199, but of course this is also implicit in his example of mentally
disabled children). He also discounts methods of determining mental age that
refer to parallel processes such as schooling, dividing childhood into pre-
school, primary school, and secondary school. Significantly, however, he con-
cedes that ‘since the processes of child development are closely connected
with the teaching of the child, and the separation of teaching into levels
depends on enormous practical experience, then naturally breaking childhood
up according to a pedagogical principle brings us extremely close to a real
division of childhood into separate periods’ (1998: 187). This is the basis of
the socio-cultural ‘objective’ ZPD that Chailklin (2003) refers to, where it
appears that Vygotsky does have something like classroom age grouping in
mind. So it is also hardly surprising when we read that Vygotsky explicitly
refers to the possibility of whole group zones of proximal development.

[S]ince teaching depends on immature, but maturing processes and the whole area of
these processes is encompassed by the ZPD of the child, the optimum time for teaching
both the group and each individual child is established at each age by the zone of their
proximal development.
(1998: 204, emphasis added)

Vygotsky’s criticism of Thorndike is not simply that his use of diagnostic

tests looks backwards to development rather than forwards to learning. Above
all, he criticizes Thorndike for:

the almost complete neglect of the social factor in education. The teacher remains the
highest authority, the prime mover of the pedagogical mechanism, the source of light and
284 The ZPD and whole class teaching

sermon. Education is addressed from the teacher to the pupil, remains deeply individu-
alistic all the time and-in the words of one author-reminds us of a pedagogical duet
between the teacher and the pupil.
(Vygotsky, 1997a: 150)

Vygotsky then proceeds to draw a rather colorful comparison between the

rickshaw puller and the tram driver. He points out that both the rickshaw
puller and the tram driver have an ‘animal’ component of labor as well as a
thinking and planning component, but that the proportions are reversed:
where ‘physical labor dominates in the rickshaw’ (1997a: 159), the tram
driver has reduced this component to virtually zero.
Vygotsky believes that the teacher’s labor may similarly be differentiated
into the provision of content and the organizing of the social environment.

The teacher’s labor, although it is not subject to the technical perfection which moves
and pushes it from the rickshaw to the tram-driver, has nevertheless the same two
aspects . . . [W]ith some exaggeration it may be said that the whole reform of contem-
porary pedagogics revolves around this theme: how to reduce the role of teacher when
he, just like the rickshaw-puller, plays the role of the engine and part of his own peda-
gogical machine as closely as possible to zero, and how to base everything on his other
role – the role of organizer of the social environment?
(1997a: 160)

Vygotsky might well see the teacher who limits teaching to the individual
‘scaffolding’ of learners as a mere rickshaw puller, playing the role of the
engine of the pedagogical machine and attempting to substitute him or herself
for the social environment of learning. In fact, although Vygotsky’s name has
become widely associated with scaffolding, in the whole of the Collected
Works, the word is used only once, and then it is to describe a construction
site in Berlin which a mentally retarded child passes on his way home (1997b:
205). Contrary to the claims made in Langford (2005: 126–27, 140–41), it is
never used as a metaphor for the ZPD.

III From rickshaw puller to tram driver

For the classroom teacher, what are the alternatives to being a rickshaw puller,
either in the form of a provider of comprehensible input or in the form of a
builder of individualized scaffolding? One obvious solution, strongly favored
by Fernandez et al., is for the teacher to organize the social environment into
group work in which, they argue, an ‘inter-mental development zone’ is cre-
ated, largely through the use of exploratory talk.

Where, as an example of an ‘asymmetrical’ interaction, a teacher might explicitly plan

how to show children an idealized version of a problem to help them understand it, in
symmetrical talk the idealized version often emerges in an unplanned way through
 Iju Guk and David Kellogg 285

attempts by children to share understandings and to explain solutions as they work

together. The children in symmetrical talk may not be consciously trying to scaffold the
development of each other’s understanding (as might a tutor) but the implicit ground
rules that they are following have this effect anyway (without needing any conscious
intention).
(Fernandez et al., 2001: 53)

Fernandez et al. contend that such an ‘inter-mental development zone’ is a

different, non-volitional type of ZPD. They call openly for the ‘reconceptual-
ization’ of Vygotsky. Other workers in the field of language learning, for
example Donato (1994), Ohta (2001) and even Swain (2000), have also
shown how the ZPD may arise in the absence of a teacher. Goodman and
Goodman go so far as to state that teachers can neither create nor control
zones of proximal development (1990: 236).
We do not deny that the ZPD may arise in the absence of a teacher. In fact,
for most of pre-school (and indeed for most of human history) it must have
done so, and so we find that Vygotsky explicitly says that the child quite
spontaneously creates a ZPD in play. We certainly observe this on a daily
basis on the playground; for example, the counting rhymes which children
employ to choose who will be ‘it’ in a game progressively grow more diffi-
cult (and thus more difficult to cheat) as children grow older. However, we
note that for Vygotsky, school, and particularly primary school, represents a
continuation by other means of the processes of abstraction and self-
regulation begun in play (1993: 197). We thus still have the fundamental
question before us: how does the teacher organize the social environment of
learning? One obvious way in which this could happen is through the very
Vygotskyan idea of imitation (Lantolf, 2003). Let us suppose that teacher-led
mediation and group mediation, linked by procedural imitation, are not two
mutually exclusive alternatives. They can be seen as two waystages on an
unbroken line of mediated activity leading to internalization.
Mediation, here, simply refers to Vygotsky’s well-known observation that
all higher psychological processes, such as the conscious control of memory
and attention, are neither innately specified nor direct responses to the envi-
ronment but rather develop indirectly through the mediating action of tools,
signs, and of course the people who wield them. Even a small baby’s point-
ing gesture, according to this view, is neither pre-programmed nor immedi-
ately reactive. Instead, it is an exaptation, similar to the exaptation of mouths
and throats, originally developed for eating purposes and perverted for the
purpose of communication. The pointing gesture develops out of the grasping
gesture (which does in fact appear to be innate) only when this failed grasp-
ing gesture is interpreted as a pointing gesture by the baby’s caretakers.
Noticing this interpretation, the baby can then move the ‘meaning’ of the ges-
ture to the beginning of the baby’s action, where it serves as the intra-mental
cause of actions rather than their inter-mental effect. Through interactional
286 The ZPD and whole class teaching

mediation, the social interpretation of one gesture (grasping) becomes the

psychological motivation of another (pointing, which includes a semiotic
component). Vygotsky holds that the acquisition of language may proceed in
a likewise manner, though of course we now find that simple gestures become
differentiated into tools for acting on the outside world and symbols which act
on the mind itself. To make matters even more complicated, we find that tools,
symbols and other people hardly ever act alone; people use tools to record
symbols, and of course tool and symbol use must itself be mediated socio-
interactionally, and often with the help of other symbols (for example, when
we use spoken language to mediate the internalization of written language, or
the first language to mediate the acquisition of the second).
Vygotsky’s conception of ‘internalization’ has been criticized (Wertsch,
1985; 1998; Kramsch, 2002: 3) and even retranslated (as ‘interiorization’,
Lantolf, 2003). Wertsch’s critique is understandable, because his examples
(e.g. pole vaulting) are often activities that appear to resist internalization
(when a pole vaulter adjusts to a new fiber-glass pole it does not seem partic-
ularly illuminating to say that he has ‘internalized’ the pole and it would
indeed appear to make more sense to say that he has appropriated the tech-
nology or ‘mastered it’). But our own interest is something that is eminently
‘internalizable’, namely word meanings. It seems unnecessarily stand-offish
to say that a word meaning has been ‘appropriated’ or even ‘mastered’.
Worse, this terminology suggests to us the rather Western view of language as
intellectual property, not entirely unconnected to one-on-one ‘scaffolding’
and the revisionist reading of Vygotsky we wish to avoid. We are rather more
sympathetic to Lantolf ’s retranslation of the term as ‘interiorization’, partic-
ularly since it captures quite well Vygotsky’s view that in the process not only
the word meaning but also the architecture of the mind itself may be trans-
formed (which is why the ZPD is a zone of development and not simply
learning). However, while ‘internalization’ has an obvious linguistic counter-
part in ‘externalization’, it is not at all clear to us what ‘exteriorization’ of lan-
guage might mean (for example, whether or not it has to involve restructuring
the environment!). Therefore, out of theoretical humility (some would say
timidity), we shall retain the original term ‘internalization’.
It is argued by van der Veer and Valsiner (1991: 379) that ‘internalization’
is the aspect of the ZPD that remains empirically unverifiable; that it is impos-
sible to demonstrate that intra-mental knowledge of a particular logical
proposition or a higher psychological function (or even a strong showing on
one of Vygotsky’s memory tasks) is derived from a similar or even identical
construct in inter-mental knowledge; and that therefore the existence of the
ZPD can never be proven. Clearly, if this criticism were sustained it would
demolish not only the ZPD and Janet’s law, but Vygotsky’s whole monist con-
ception of the socially constructed individual. However, it seems to us, once
again, that it is precisely with word meanings that the objection cannot be sus-
tained. Not even the most extreme innatist views of language hold that the
genesis of word meanings in the child’s mind is unconnected to the words
 Iju Guk and David Kellogg 287

found in the child’s environment. Certainly, we feel safe in the assumption

that the word meanings and even the grammar structures in our data are the
same when they occur in T-S socio-interactional mediation and then appear in
almost the same task in S-S interaction involving the same learners.
However, the data we shall present are intended to illustrate the theoretical
argument that both teacher-led mediation and learner-to-learner interaction
may be considered waystages within a single whole classroom ZPD. It is not
intended to prove empirically that learning has or has not taken place in either
interactional format. Indeed, because the data is gathered in the course of
actual classroom teaching by one of the authors, it would be extremely diffi-
cult to control the variables and measure the outputs to a sufficient degree of
precision to provide any proof of learning at all. Even if we could do so, it
would not be sufficient for our argument, because, as Chaiklin (2003) notes,
the ZPD does not refer to learning in general, but only to learning which
results in development, and we have no way to demonstrate that the learning
we observe here (if that is what it is) is not developmentally inert. Our argu-
ment, then, must remain largely on theoretical grounds.
This puts us in good company. According to Bakhurst, Vygotsky’s own
approach was characterized by studying ‘the internalization of activities
which are first realized in public interaction with others’. Bakhurst further
notes that this theory-testing approach depended on intervening in the process
of internalization by offering different psychological tools, that Vygotsky
believed that the results were ‘often best presented by describing particular
cases in detail rather than giving statistical data for a large sample of sub-
jects’, and finally that he did not clearly distinguish between theoretical argu-
ments and empirical research (Bakhurst, 1991: 81). Vygotsky died before he
could fully work out the ZPD in the context of whole classroom teaching, and
our work is unfortunately closer to speculation than to imitation. But perhaps
it is not entirely wild speculation.

IV Some background and some data

Let us begin with some background. English is taught as a required foreign
language subject in Korean primary schools for two hours a week in the fifth
and sixth year grades. The lessons follow a single state-prescribed textbook,
with lessons based on the presentation of a listening text or the repetition of
a dialogue followed by a game or task-based activity. The teacher is often a
‘specialist’ teacher selected from the regular school staff rather than the chil-
dren’s general subject teacher. Our data show a ‘specialist’ teacher demon-
strating the game or task to one learner in a group of fifth year learners. This
learner, who is wearing a clip-on microphone, then goes to a group of chil-
dren and shows them how to carry out the task. We thus have the same task
presented twice, first in a T-S interactional format and then in an S-S interac-
tional format.
288 The ZPD and whole class teaching

In his most famous description of the ZPD, Vygotsky suggests a number

of ways in which a more capable adult can show children solutions to a
problem. For example, ‘some might run through an entire demonstration
and ask the child to repeat it, others might initiate the solution and ask the
child to finish it, or offer leading questions’ (1978: 85). The following
extract offers examples of all of these (though of course there is no way to
be certain that learning, much less development, is taking place). The
teacher and the students have just completed a classroom arm wrestling
tournament and are using the results, in the form of a competition ‘tree’
chart, to practice the grammatical comparative. The words in italics are in
Korean, given in the official government transcription, and followed by an
English translation.
Transcript 1: T-S interaction

1 T: Repeat after me. Dahye is stronger than Yeseul.

2 Ss: Dahye is stronger than Yeseul.
3 T: than Yeseul
4 Ss: than Yeseul
5 T: Keokkuro hamyeon mueoraguyo? (How would you say it the other way?) Yeseul
is . . .
6 Ss: Yeseul is weaker . . .
7 T: . . . weaker than Dahye.
8 Ss: Yeseul is weaker than Dahye.
Sure enough, the teacher ‘runs through a demonstration’ of how the students
can say the grammatical comparative and asks them to repeat after her in Line
1. After this, the teacher ‘initiates a solution’ with ‘Yeseul is . . .’ in Line 5 and
‘asks the children to finish it’, as Vygotsky suggests. Finally, the teacher
offers the leading question ‘Keokkuro hamyeon mueoraguyo? (How would
you say it the other way?)’
When we look at the how the learners mediate the same task in groups, we
notice a narrower repertoire of techniques.
Transcript 2: S-S interaction

1 S4: Eoryeoweo! (That’s difficult!)

2 YS: Ppalli hae bwa. (Hurry and give it a try.) You’re stronger than me.
3 S3: OK.
4 S4: You’re stronger than me.
5 S2: Eo. (Mmm.)
6 S3: Na mothaneundae. ( I can’t seem to get it.)
7 YS: Keureom neo ne hae. Hae bwa. Hae bwa. (OK, so you do yours! Give it a try.
Give it a try.)
8 S4: Na yikyeosseo. (I won.)
9 YS: Hae, mal hae. (Do it. Say it.)
10 S4: Mueo rago? (What do I say?)
11 YS: I am stronger than you.
12 S1: You are weaker than you.
 Iju Guk and David Kellogg 289

13 YS: Aniji, (That’s not it.) I am weaker than you.

14 S1: I am weaker than you.

Here we find practically no examples of initiating the solution and asking

the child to finish it, and only a very oblique leading suggestion (‘Na
yikyeosseo. [I won.]’). In giving ‘demonstrations’ of the solution phrase
(Lines 2, 4, 11, and 13), the children do not use leading questions or
commands.
There are, however, a number features present in the S-S interaction that
are absent from T-S interaction. In the first place, the exchange begins with a
metalinguistic remark on the difficulty of the problem, voiced almost as a
complaint. We find that students commonly initiate metalinguistic exchanges
(e.g. ‘Ppalihae bwa [Hurry up and do it]’, ‘Hae, malhae [Do it! Say it!]’) in
S-S interaction, while they seldom initiate these in T-S interaction. There also
appears to be a very significant amount of L1 metalanguage in S-S interaction
compared with T-S interaction, and there is, of course, a highly salient error
in the former but not the latter.
While we believe, with Vygotsky, that describing particular interactions in
detail is often more useful than presenting qualitative data, we also believe
that it is necessary to find out if these are general tendencies or simply char-
acteristic of these extracts. For this we need to look at the wider database. The
extracts above are drawn from a database of five different EFL lessons, each
given to two different classes in a primary school in Seongnam, South Korea,
totaling 2356 transcribed utterances, all of which include a teacher-mediated
task followed by group work. We will present findings and then offer tenta-
tive explanations based on re-contextualizations of the quantitative data in the
form of transcripts.
Over the years, Swain (2000) has put forward a very powerful argument for
the role of metalanguage, or ‘talk about talk’. Here, using and learning lan-
guage can come to together and the terrible ‘contradiction between . . . two
modes of processing language: that of learning it for future use, and that of
using it’ noted by Halliday can be transformed ‘from a constraint into a con-
dition which enables and even enforces the learning process’ (Halliday, 2003:
141). We turn, therefore, to the metalanguage first.

1 S-S interactional mediation has a slightly lower proportion

of utterances involving metalanguage
In our examination of the T-S Transcript 1 above, we saw only two utterances
that were metalinguistic out of the eight utterances (numbers 2 and 8) versus
eight out of 14 in the S-S transcript. In contrast, the whole database shows a
small margin of difference, as shown in Table 1.
There are, however, considerable differences in the functions of metalan-
guage, as Table 2 shows.
290 The ZPD and whole class teaching

Table 1 Metalanguage use in two types of mediation (T-S and S-S)

Type of mediation Metalanguage Non-metalanguage

(T-S or S-S) use use Total

N % N % N %

T-S interaction 303 20.85 1150 79.15 1453 100.0

S-S interaction 172 19.09 729 80.92 901 100.0
Total 475 20.18 1881 79.82 2354 100.0

• Inquiring about word meanings, e.g. ‘Mueo rago? (What do I say?)’

• Confirming word meanings, e.g. ‘You’re stronger than me.’ ‘Eo. (Mmm.)’
• Leading questions or initiating a solution, e.g. ‘Keokkuro hamyeon mue-
oraguyo? (How would you say it the other way?) Yeseul is . . .’
• Negotiation, e.g. ‘Na mothaneundae. ( I can’t seem to get it.)’ ‘Keureom neo
ne hae. Hae bwa. Hae bwa. (OK, so you do yours! Give it a try. Give it a try.)’
• Demonstrating a solution, e.g. ‘Repeat after me. Dahye is stronger than
Yeseul.’
• Translation of L2 into L1 (or vice versa), e.g. ‘Ol su issnyago? (You mean can
I come?)’ (See Line 4, Transcript 4, below for the contextualized example.)
The whole database shows that T-S interaction is in fact rich in instances of
the techniques which Vygotsky mentions: demonstrating whole utterances
and initiating solutions (eliciting through incomplete phrases or leading ques-
tions). S-S interaction, on the other hand, appears to richer in the negotiation
and confirmation of word meanings.
These differences emerge even more clearly when we return to the tran-
scripts. In the following T-S interaction, the teacher is getting a group of chil-
dren to stand up, walk around, and invite each other to imaginary birthday
parties. She presents the task by doing it rather than merely talking about it;
nevertheless her example contains a good deal of metalanguage (utterances 6,
9, and 10), as she attempts to expand the learner’s ‘yes’ into ‘I’d love to’.
Transcript 3: T-S interaction

1 T: YS, er . . . This Saturday is my birthday. I’ll have a birthday party in my house.

Would you like to come to my house?
2 YS: Yes.
3 T: Yes?
4 T: You can say ‘Yes, I’d love to’.
5 YS: Yes, I’d love to.
6 T: You don’t look, Jibe byeollo an ogo sipeo hanun geo katae byojeongi. (You look
like you don’t really want to come.) Yes, I’d love to Ireokke hanbeon hae boseyo. (Try
it like this for once.) You can say ‘Yes, I’d love to’.
7 YS: Yes.
8 T: I’d love to.
9 YS: I’d love to.
Table 2 Functions of metalanguage in types of interaction

Type of WHO Function of metalanguage Total

interaction
Inquiring about Confirming word Leading questions or Demonstrating Negotiating Translation
word meanings meanings initiating a solution a solution

T-S S 34 17 7 3 7 3 71
T 3 17 95 112 . 11 238
T 37 34 102 115 7 14 309
S-S S 35 25 56 14 29 7 166
T . . . 6 . . 6
T 35 25 56 20 29 7 172
Iju Guk and David Kellogg
291
292 The ZPD and whole class teaching

This embellishment (‘I’d love to’) is passed on in the S-S phase of interac-
tional mediation by the imitation of the ‘more able peer’, HC, who is sent
from the teacher-led group to a learner-led group.

Transcript 4: S-S interaction

1 HC: This Saturday is my birthday. I will have a birthday party. Would you like
come . . . Would you like to come to my house?
2 KH: Ol su issnyago? (You mean can I come?)
3 HC: Eo. (Yeah.)
4 KH: Yes.
5 HC: I’d love to.
6 KH: I’d love to. (proud of his new utterance): Nega haneun geo chal bwassji? (Did
you see that?) Au deul-a! (Just call me Elder Brother!)

Although metalanguage is certainly present (utterances 2 and 3, and most

ostentatiously in utterance 6) in S-S mediation of the task, it takes quite a dif-
ferent form from what we observed in the T-S interaction. In both T-S tran-
scripts (Transcript 1 and Transcript 3 above), we see that teachers tend to use
English to refer to language (e.g. ‘Repeat after me’ in Transcript 1 and ‘You
can say it like this’ in Transcript 3). In the even-numbered S-S transcripts, on
the other hand, the learners tend to use Korean, (e.g. ‘Ppalli hae bwa. [Hurry
and give it a try.] You’re stronger than me’ in Transcript 2 and ‘Ol su iss-
nyago? [You mean can I come?]’ in Transcript 4.) When HC provides assis-
tance with new language in Line 5, there is no explicit ‘listen and repeat’ or
‘you can say it like this’ metalanguage at all. The feedback move (6, where
KH compliments himself on his new knowledge and fishes for compliments
from his peers) is also oblique. It is, of course, not simply inter-mental but
also intra-mental: the child is evaluating himself!
In contrast to KH’s self-congratulation, there is no clear indication from
YS that the new language is appreciated. In contrast to KH’s comprehension
check (‘Ol su issnyago? [You mean can I come?]’), we have no evidence
from YS that ‘I’d love to’ is actually understood. A teacherly response to this
data might be to remark that there is rather more evidence of learning in the
S-S interaction than in the T-S one. This teacherly response is not necessar-
ily a theoretically naïve one; the S-S interaction is marked by a greater use
of metalinguistic reflection and a greater integration of L1 and L2. However,
the key issue for us is here is whether or not the mediational means is under
the control of the learner. It is this increasing control over the mediational
means that allows the learner to progress from other-mediation to self-
mediation (for example, repeating utterances privately), and ultimately to
internalization, that is, to no mediation at all (for although internalized word
meanings remain ‘mediated’ in the sense that they bear the social stamp of
their origins, they no longer require the active use of mediators such as tools,
signs and other people to activate).
 Iju Guk and David Kellogg 293

Even from the learner’s point of view, it is easy to see how S-S socio-
interactional mediation is a logical step in this direction. After all, YS might
reasonably object that the language the teacher has mastered is too difficult
for her, but KH’s friends may be more inclined to think that if other group
members can use a particular word meaning then there is no reason in princi-
ple while they cannot master it themselves. From the teacher’s point of view,
the move from ‘rickshaw pulling’ T-S interaction to more ‘tram-driving’ S-S
interaction certainly saves labor. But a common teacherly expectation to the
move from teacher-fronted interaction to group work to object that it will
result in a lower proportion of L1 use, and a higher proportion of error. Once
again, we find that this response is far from naïve, and that it has an interest-
ing theoretical dimension.

2 S-S interactional mediation has a lower proportion but higher

absolute number of learner utterances in English
Swain and Lapkin (2000) found that students use L1 for three significant pur-
poses:

(1) moving the task along, which means figuring out the order of events, retrieving
semantic information; understanding pieces of information; developing an understand-
ing of the story, task management, (2) focusing attention, which means vocabulary
search, focus on form; explanation; framing; retrieving grammatical information, and
(3) interpersonal interaction, which means off task (include L1 vernacular use) and dis-
agreement.
(2000: 257–58)

Like Swain and Lapkin, we found few examples of the third, off-topic, cat-
egory of L1 use. This perhaps explains why there is a fairly similar propor-
tion of English use in T-S and S-S interaction. The lack of off-topic L1 use
is perhaps due to the fact that the S-S work is immediately preceded by a
clear T-S example, which the children imitate, alter (substituting whole utter-
ances for piece-by-piece constructions and L1 use for L2) but never entirely
abandon.
Clearly, the children here are heavily involved in L1 use for focus on form,
as we saw in our section on metalanguage. But L1 is also extensively used for
other tasks, including task management, as Table 3 confirms. Interestingly, in
spite of the fact that there is a larger proportion of learner utterances in
English in T-S interaction (53.51%), there are still absolutely more English
utterances (422) in the S-S condition, presumably because the children get to
talk more. The difference in code mixing is also interesting. In these extracts,
the teacher and the children and then the children without the teacher are
planning a musical performance. The teacher is very demonstrative, and
uses English.
294 The ZPD and whole class teaching

Table 3 Learner use of English in T-S and S-S mediation of learning tasks

Type of Use of English or other codes Total

Mediation
(T-S or English Non-English
S-S)
Korean Code-mixing Total

N % N % N % N % N %

T-S 335 53.51 273 43.61 18 2.88 291 46.49 626 100.0
S-S 422 47.15 420 46.93 52 5.92 472 52.85 895 100.0
Total 757 49.77 693 45.56 72 4.67 765 50.23 1521 100.0

Transcript 5: T-S interaction

1 T: Fold the paper like this. (Children do so.)

2 T: Let’s write the title. (They do so.)
3 T: I want to whatever . . .

The children in this class, in contrast, use much less explicit language, and L1
at that:
Transcript 6: S-S interaction

1 KS: Malhae. (Talk.)

2 S1: Neobuteo malhae. (You first.)
3 S2 Yigeo? (Here? [pointing to the paper])
4 S3 Yigeo? (Here?)
5 KS: Eo. (Unh-hunh . . . ) I want to play the drum.

As we can see, the teacherly fear of L1 use in S-S interaction is hardly

unfounded. But we shall argue that the L1 use in Transcript 6 should not
be contrasted with L2 use, because only the teacher uses the target lan-
guage in Transcript 5, but rather with silence. Through this we can see,
once again, more evidence that the language is appreciated, or at any rate
understood (Lines 4–5, ‘Yigeo? [Here?]’ ‘Eo. [Unh-hunh]) I want to play
the drum.’).
There also appears to be more indexical language, including deictic
expressions such as ‘Here’, which depend on shared context and expressions
in which speakers are negotiated (e.g. Turns 1–2, ‘Malhae. [Talk.]’
‘Neobuteo malhae. [You first.]’). This is not surprising, since the size of the
group makes it easier to point meaningfully and the more symmetrical
quality of the interaction makes speaking order negotiable. But both of these,
together with the code switching noted earlier, can be seen as indications
that the mediation of the word meanings is gradually passing from teacher
to learner control, and the interaction as a whole is moving closer to
internalization.
 Iju Guk and David Kellogg 295

3 S-S interactional mediation features less accurate

but longer learner utterances
A further indication that we are moving closer to the internal state of the L2
system when we look at S-S interaction can be seen in the presence of errors.
As Table 4 shows, the learners tend to be more accurate in the T-S condition
than in the S-S one. However, English utterances in S-S interaction also
appear to be longer and more complex than in the T-S condition: student utter-
ances in English were 3.11 words long on the average in S-S interaction, com-
pared with only 2.78 words long in T-S interaction. Levene’s test shows that
the variances may be assumed to be the same, and because the sample is
large, this result is significant according to the independent samples t-test
(t ⫽2.757; df ⫽755; p ⬍.01).
Turning again to the transcripts for an explanation, we find that the teacher
often breaks down sentences to focus on particular word meanings, as in this
task, where the children are casting parts for an imaginary class musical:
Transcript 7: T-S interaction

1 T: And then, drum yiji? (Is it?) I want to play drums.

2 NU: I want to play drum.
3 T: drums
4 NU: drums

In contrast, the learners tend to leave the utterances whole, and rather than
correct errors, they often copy and complicate them. Here GS attempts to
focus on the word ‘drum’ as the teacher had been observed to do, but the
result is not what he expects (perhaps because there is no actual error in this
case).
Transcript 8: S-S interaction

1 S1: I want to play the drum.

2 S2: I want to play the drum.
3 S3: Ya, dasihae! (Hey, try it again!)
4 GS: Drum.

Table 4 Percentage of error-free utterances in T-S and S-S mediation of tasks

Accuracy Total
Type of mediation
(T-S or S-S) Error-free At least one error

N % N %

T-S 323 96.42 12 3.58 335

S-S 375 88.86 47 11.14 422
Total 698 92.21 59 7.79 757
296 The ZPD and whole class teaching

5 S2: I want to play the drum.

6 GS: I want to play the dance.
7 S1: I want to play the dance.
8 GS: Dasi. Neo meonjeo hae. (Again. You go first!) I want to sing.
9 S3: Norae pureurago? (You want to sing a song?)
10 GS: Dasi malharago. (I want you to speak again!) I want to play a dance. Malharago.
(Say it again!)
11 S1: I want to play a dance.
12 S3: I want to play the sing.

In Line 6, GS tries a different form of analysis, not breaking down the sentence
and focusing on the problematic part, but rather repeating the grammatical pat-
tern and varying the lexis (‘I want to play the dance’). The result is an error,
cheerfully repeated by his peer in Line 7. GS, now unsteady in his role of sur-
rogate teacher, introduces another verb, ‘sing’, which eventually, in Line 12,
serves to create a new error, ‘I want to play the sing’. The emphasis here is
clearly on participation in the group rather than acquisition of accurate forms!
A rather teacherly way to look at this is to note with some dismay that the
participants are not internalizing the form mediated by the teacher (‘I want to
play drums’). But for the purpose of our discussion there are two other points
arising that are of equal interest. The first is that the children show a certain
preference for whole utterances, even when they are analyzing sentences into
grammatical constituents, and this feature, fairly consistently observed
throughout the data, would explain why their utterances tend to be longer in
the S-S condition. But the second point arising is that while the T-S version
of the utterance is undoubtedly closer to the task target, the learner’s forms
are closer to the children’s level of unassisted performance. On the evidence
of this data, ‘I want to play the dance’ is, like it or not, closer to internaliza-
tion than ‘I want to play drums’. This does not, of course, in itself indicate
that this is part of the ZPD; on the contrary, it seems likely that this type of
‘learning’ is not destined to turn into development, at least not if the teacher
has her way.

V Conclusion: One ZPD or two?

A comparison of T-S and S-S interaction that focuses on a single obvious fea-
ture, say, L1 use, might suggest that T-S and S-S interactions are taking place
in different languages and according to different rules, and therefore they
belong in different zones, only one of which can be the true ZPD. The same
conclusion follows if we look too narrowly for surface features, such as the
‘demonstration of a solution’, or ‘initiation of a solution’ or ‘leading ques-
tions’ mentioned by Vygotsky in his famous definition. We might easily be
led to the conclusion that there are two different kinds of ZPD being built, one
based on ‘asymmetrical scaffolding’ (Wertsch, 1998) and the other on sym-
metrical, collaborative dialogue (Fernandez et al., 2001).
 Iju Guk and David Kellogg 297

There are, in fact, a small number of points of complete difference in the

two interactions. For example, the children never use English commands such
as ‘Listen and repeat’, nor do they break utterances down into single words or
even single letters as the teacher does. On the other hand, there are far fewer
instances of children repeating utterances to themselves when the teacher is
present, and very few student initiates of any kind. While the teacher seems
more concerned with the construction of grammar, the children are interested
in co-constructing discourse.
But for the most part the differences we observe are quite small. First,
there is marginally more metalanguage in T-S interaction than in S-S inter-
action, and this appears to be more often in L2. Second, there is a smaller
proportion of learner utterances in English in S-S interaction than in T-S
interaction, although English utterances in the S-S condition are more
numerous in absolute numbers than those found in T-S interaction. Third,
learner utterances in English tend to be more accurate but shorter in the T-S
condition.
Instead of interpreting these differences in T-S mediation and S-S medi-
ation as indicating two different types of ZPD, we wish to suggest that they
create different ends of the same whole class ZPD, with T-S interaction
relying on projection of word meanings and grammar structures from an
adult, and S-S interaction relying on co-construction of discourse in whole
utterances. The teacher in the data appears to have a strong language ori-
entation, while S-S interaction tends to centre more on the task. Similarly,
the teacher tends to be analytical and break down language into very small
pieces (putting together sentences word by word and words letter by
letter), while students are more synthetic and work with larger chunks. But
this merely suggests to us that T-S interaction lies close to the upper end
of the ZPD, that is, the end of the ZPD concerned with inter-mental medi-
ation and assisted performance, which is at first asymmetrical, while the
S-S interaction represents the lower, unassisted end, bordering on inter-
nalization.
It is possible, even likely, that different forms of mediation do lead to dif-
ferent forms of knowledge, some of which are more clearly inter-mental (for
example, the teacher telling a learner to ‘Listen and repeat’, or demonstrating
a solution to a task which a learner must watch in order to imitate) and others
of which are more intra-mental (such as a learner repeating language to him-
self or herself, or creating and sharing his or her own solution to a task).
However, these different forms of knowledge may co-exist and even be
superimposed on each other (Yi and Kellogg, 2006). Because the more intra-
mental forms of knowledge only gradually survive and replace the more
inter-mental forms, a ZPD that includes both teacher mediation and com-
plete internalization of complex linguistic structures cannot be a short-lived
affair. This is one reason why Vygotsky refers to a zone of development and
not simply learning. By the same token, it is unlikely to have a maximum
population of two.
298 The ZPD and whole class teaching

Acknowledgements
Both authors are sincerely grateful for the widely divergent yet equally con-
structive criticisms we received from the reviewers, and the patient mediation
of the editor, all of whom we completely absolve from the diverse weaknesses
that remain.

VI References
Askew, M., Bliss, J. and McRae, S. 1995: Scaffolding in mathematics, science and
technology. In Murphy, P., Selinger, M., Bourne J. and Briggs, M., editors,
Subject learning in the primary curriculum: Issues in English, Science and
Mathematics. London and New York: Routledge in association with the Open
University, 209–17.
Bakhurst, D. 1991: Consciousness and revolution in Soviet philosophy: From the
Bosheviks to Evald Ilyenkov. Cambridge: Cambridge University Press.
Chaiklin, S. 2003: The ZPD in Vygotsky’s analysis of learning and instruction. In
Kozulin, A., Gindis, B., Ageyev, V.S. and Miller, S.M., editors, Vygotsky’s edu-
cational theory in cultural context. Cambridge: Cambridge University Press,
39–64.
Donato, R. 1994: Collective scaffolding in second language learning. In Lantolf, J.P.
and Appel G., editors, Vygotskian approaches to second language research.
Norwood, NJ: Ablex. 33–56.
Fernandez, M., Wegerif, R., Mercer, N. and Rojas-Drummond, S. 2001:
Reconceptualizing ‘scaffolding’ and the ZPD in the context of symmetrical col-
laborative learning. Journal of Classroom Interaction 36(2): 40–54.
Gillen, J. 2000: Versions of Vygotsky. British Journal of Educational Studies 48(2):
183–98.
Goodman, Y.M. and Goodman, K.S. 1990: Vygotsky in a whole language perspec-
tive. In Moll, L.C., editor, Vygotsky and education: Instructional implications
and applications of socio-historical psychology. Cambridge: Cambridge
University Press, 223–50.
Halliday, M.A.K. 2003: On language and linguistics. London and New York:
Continuum.
Kinginger, C. 2002: Defining the ZPD in US foreign language education. Applied
Linguistics 23(2): 240–62.
Kramsch, C. 2002: Introduction: ‘How can we tell the dancer from the dance?’ In
Kramsch, C., editor, Language acquisition and language socialization:
Ecological perspectives. London: Continuum.
Lantolf, J.P. 2003: Intrapersonal communication and internalization in the second
language classroom. In Kozulin, A., Gindis, B. Ageyev, V.S. and Miller, S.M.,
editors, Vygotsky’s educational theory in cultural context. Cambridge:
Cambridge University Press, 349–70.
Langford, P.E. 2005: Vygotsky’s developmental and educational psychology. Hove
and New York: Psychology Press.
 Iju Guk and David Kellogg 299

Mercer, N. and Fisher, E. 1997a: The importance of talk. In Wegerif, R. and

Scrimshaw, P., editors, Computers and talk in the primary classroom.
Clevedon: Multilingual Matters, 13–21.
—— 1997b: Scaffolding through talk. In Wegerif, R. and Scrimshaw, P., editors,
Computers and talk in the primary classroom. Clevedon: Multilingual Matters,
196–210.
Nassaji, H. and Swain, M. 2000: A Vygotskian perspective on corrective feedback in
L2. Language Awareness 9: 34–51.
Ohta, A.S. 2001: From acknowledgment to alignment: a longitudinal study of the
development of expression of alignment by classroom learners of Japanese. In
Kasper, G. and Rose, K., editors, Pragmatics in language teaching. Cambridge:
Cambridge University Press. 103–20.
Poehner, M.E. and Lantolf, J.P. 2005: Dynamic assessment in the language class-
room. Language Teaching Research 9(3): 233–65.
Swain, M. 2000: The output hypothesis and beyond: Mediating acquisition through
collaborative dialogue. In Lantolf, J., editor, Language learning and sociocul-
tural theory. Oxford: Oxford University Press, 97–114.
Swain, M. and Lapkin, S. 2000: Task-based second language learning: The uses of
the first language. Language Teaching Research 4(3): 251–74.
Valsiner, J. and. van der Veer, R. 2000: The social mind. Cambridge: Cambridge
University Press.
van der Veer, R. and Valsiner, J. 1993: Understanding Vygotsky: A quest for synthe-
sis. Oxford and Cambridge MA: Blackwell.
Vygotsky, L.S. 1978: Mind in society. Cambridge, MA: Harvard University Press.
—— 1993. Collected works, volume 2. London, Boston, Dordrecht, and Moscow:
Kluwer Academic/Plenum.
—— 1997a: Collected works, volume 3. New York and London: Plenum.
—— 1997b: Collected works, volume 4. New York and London: Plenum.
—— 1998: Collected works, volume 5. New York and London: Plenum.
Wegerif, R. and Mercer, N. 1997: A dialogical framework for researching peer talk.
In Wegerif, R. and Scrimshaw, P., editors, Computers and talk in the primary
classroom. Clevedon: Multilingual Matters, 59–61.
Wells, G. 1999: Dialogic inquiry. Cambridge: Cambridge University Press.
Wertsch, J.V. 1985: Vygotsky and the social formation of mind. Cambridge, MA:
Harvard University Press.
—— 1998: Mind as action. Oxford: Oxford University Press.
Yi, J. and Kellogg, D. 2006: Beneath higher ground: Vygotsky, Volosinov, and an
archaeology of reported speech in primary English writing. Language
Awareness 15(1): 38–52.