backup: adfs::0.$.chap7
23rd October 1993
Chapter 7: COLLABORATIVE INTERACTIONS *AT* COMPUTERS
In this chapter, I shall describe certain computer-based interactions
that I have observed within primary school classrooms. They all
involve circumstances arranged for children to work together *at* this
technology. By that, I mean circumstances in which small groups of
children (as it happens, usually pairs of children) are organised to
work on the same computer-based problem at the same time. As was
indicated in the last Chapter, such arrangements are typical of
current practice within early education. Moreover, psychologists have
been quite busy in finding ways to conceptualise the processes that
commonly arise within such collaborative problem solving. Again,
discussion in the last chapter dwelt on the various theoretical
distinctions that have been applied to learning organised in this way.
What, therefore, remains to be done? What can inspire yet more
classroom observations of peers working together at computers?
The arguments developed towards the end of the last Chapter should
suggest answers to these questions. I indicated there that prevailing
preferences of theory and method have encouraged a narrow view of what
is involved in collaborative learning. I believe the conceptual
vocabulary typically used to characterise this form of joint activity
requires rethinking - particularly if we are aiming to develop a
socially-grounded theory of cognition. I would like to check my own
efforts at such rethinking against a corpus of material from authentic
class-based collaborations. Hence the presentation of such material
in this chapter.
However, consolidating a socio-cultural perspective on cognition is
only part of my purposes in the present book. I am also concerned
with locating new technology as a particular context in which
the collaborative nature of learning might get organised. So, towards
the end of Chapter 4, I considered what might be the best way to
invest research and development effort in the interests of advancing
new educational technology: what strategy offers the greatest promise
for effectively integrating computers into educational practice? I
have since been arguing for a channelling of research effort in
one particular direction. That direction leads towards a greater
appreciation of how the technology can enter into educational practice
to mediate new forms of *collaborative* interactions - and, thereby
support learning. I am encouraging the question: "How can new
technology *resource* the collaborative nature of education?" So,
the empirical material reported in the present chapter will frame
computers as a distinctive basis for supporting the particular case of
learning within peer interaction. My argument will not be that the
technology is qualitatively different from other resources that might
be deployed within collaborative work: I shall not argue that it is
unique in this respect. However, it does seem to me to have
characteristics that are particularly powerful for collaborators, and
we should be sensitive to what they are.
In short, there is a reciprocity here that allows what I am describing
in this chapter to be viewed from either of two angles. On the one
hand, I am suggesting that attention to peer interactions around
computers is especially helpful for our efforts at conceptualising the
nature of collaboration very generally. On the other hand, I am
framing an account of collaborative interactions that has
particular implications for how we might best design and deploy
computers within classroom life.
My plan for this chapter is as follows. In the first section below, I
shall sketch an account of collaborative problem solving that focuses
on the creation of shared understandings and the particular role of
discourse in achieving this. I wish to argue that collaborations can
be analysed as states of social engagement: occasions upon which some
varying amount of effort is directed towards the creation of common
ground. Then, in later sections of this Chapter, empirical
observations of young children working together at various
computer-based tasks will be described. These observations will be
analysed so as to allow the development of two themes. Firstly, to
illustrate that effective collaboration involves a discursive
achievement: its analysis must go beyond the traditional technique of
counting and cataloguing discrete utterances as they are recorded
during peer interactions. Secondly, to argue that accounts of
collaboration should recognise that this state of social engagement
is, inevitably, situated. It arises within particular contexts of
problem solving and we must seek to understand the constraints and
opportunities that the structure of those particular contexts afford.
ON COLLABORATIVE DISCOURSE
The theoretical discussion within this section serves to preface
observations of classroom joint activity that follow later. First, I
am concerned with sharpening the sense in which we must view
collaboration as a discursive achievement growing out of human
intersubjective attitudes. The result should be a framework through
which some particular observations of collaboration can be viewed.
Collaboration and shared understanding
At the end of the last chapter, I expressed an unease with the scope
of concepts currently deployed to explain collaborative interactions.
These concepts tend to represent the discourse of collaboration as
comprising discrete utterances (or short exchanges) that have
circumscribed effects on the cognitions of the individuals concerned:
acting rather as "stimuli" for cognitive change. In this spirit, it
might be claimed that a collaborator could experience a cognitive
restructuring because of *conflict* stimulated by something that their
partner said. Or, it might be claimed that the pressure simply to
articulate the logic of one's position stimulates greater reflective
self-awareness. Such analyses of collaborative encounters see them as
occasions that increase the probability of certain potent events
occurring - certain social stimuli that provoke cognitive change in the
participants. I do not deny that both experiences of conflict or
opportunities to articulate a perspective should be part of
our concern when studying collaborations. Yet, this framework seems
to provide a distorted account of what typically happens on these
occasions: as if it misses something else significant that constitutes
the experience of joint problem solving. So, to stress conflict as
important seems contrary to how we describe a lot of collaborative
experience. It is often harmonious and constructive, not tense or
confrontational. While to stress the articulation of ideas seems to
make it a more passive, inward-looking experience than it often feels.
Moreover, the conceptual focus of both approaches tends to be on the
cognitive apparatus of collaborating individuals. So, accounts will
refer to metacognitions or to the cognitive structures of individual
collaborators; they are less likely to refer to a construction that is
inherently interpersonal and that might be best conceptualised in
social terms. Yet, this might be an approach a more socio-cultural
orientation should encourage.
In fact, the analysis of collaborative interactions has not yet been
greatly extended by socio-cultural ideas. Although, as I indicated in
the last chapter, the classically socio-cultural concept of
"internalisation" is one idea that has been applied in this context.
In his account of the internalising of exchanges within a zone of
proximal development, Vygotsky refers tellingly to the 'more capable'
peer as a possible partner for internalisation. Again, this seems to
distort our casual experience of what collaboration often involves: it
creates a constraint we may not always wish to accommodate. Although
the internalisation of moves socially-produced with an expert may
often be a part of what can happen, collaborations are experiences
that typically involve quite equable levels of expertise among the
participants. The socio-cultural appeal to processes of
internalisation seems to imply that peer interaction must be
approached as something less symmetrical: a kind of peer tutoring.
How, then, might a socio-cultural perspective contribute further
towards conceptualising the way that peers interact when they solve
problems together? This perspective should help define the cognitive
gains that arise from collaborating as being gains that are socially
constituted. It should help identify a cognitive resource that is
firmly located within the dynamic of the social interaction. My
suggestion here is that, in part, this might entail recognising the
importance of a human capacity for intersubjectivity, and how it
allows the possibility of creating structures of shared understanding.
There is certainly something missing from accounts of collaboration
that dwell only upon inventories of utterances - coded and categorised
for their pragmatic content. What, in particular, is not captured
is any sense of participants having used language to construct an
achievement of shared knowledge. Typically, analyses of these
interactions do not attend to how far collaborators are mutually aware
of such common ground, and how they draw upon it as a discursive
resource. There is one compelling reason for thinking this is an
important way to frame an account of what happens during collaborative
relationships: it reflects the way participants in such relationships
typically describe their experience when asked. Schrage (1990) has
assembled the reflections of a number of eminent collaborators as they
discuss the way they work - scientists, journalists, composers and so
on. They do not put their emphasis on the potential for productive
conflict, or the opportunity to clarify ideas through articulating
them publically. Such notions may be recognised as important products
of what goes on, but what successful collaborators are more likely to
refer to is the feeling of working towards constructing an object of
joint understanding - something that comes to exist between them as a
cognitive resource.
I am suggesting that students of educational interactions among peers
need to come to grips with this conception. Evidently, it suggests a
close link to ideas already reviewed in this book: particularly those
ideas concerned with collaborations developed in the course of
teacher-learner interactions. It naturally suggests the idea of
common knowledge as discussed in Chapter 5. Edwards and Mercer locate
this concept at the focal point of instructional activity: 'We can say
that the process of education, in so far as it succeeds is largely the
establishment of these shared mental "contexts", joint understandings
between teacher and children, which enable them to engage together in
educational discourse' (1987, p.69). Yet, little has been done to
identify the creation of shared mental contexts as an achievement
arising within other learning interactions - exchanges other than
those that take place between pupils and their teachers. In
particular, this idea has barely been considered as a way of analysing
the more symmetrical relationships that arise among classroom peers
who are engaged in joint problem solving.
Collaborating and conversing
At present, there is very little analysis of collaborative interaction
that adopts this perspective. Only Roschelle and Teasley appear to
have worked within a framework of the kind that I am describing here
(Roschelle, 1992; Teasley and Roschelle, 1993). They have studied
pairs of high school students working on a computer-based simulation
of principles from Newtonian mechanics. Their inclination, like my
own, is to characterise what transpires in terms of the partners
creating a "joint problem space". However, I believe there are
several respects in which an analysis of this kind needs to be
developed further. One is simple, but is necessary to any approach
stressing the situated nature of these events: the process of creating
joint problem spaces needs to be understood as something governed by
the structure that particular tasks present to collaborators. In
other words, these achievements need to be studied with attention to
how they are distinctively organised within the constraints of
particular contexts for acting. My assumption is that the nature of
what is socially constructed will vary in interesting ways that
reflect the structure of different problem tasks. I believe this
process is also likely to vary developmentally - reflecting different
histories of experience at joint problem solving. Finally, it is
likely to vary in ways that are associated with the period over which
joint activity is sustained. The implication of this last point being
that we need to respect the ecology within which learners' joint
problem solving is typically organised: in particular, the community
structure of classrooms. The observations described later in this
chapter represent small steps taken in these directions.
Roschelle and Teasley's work comes closest to the description of
collaborations made later in this chapter. Yet, I do not wish to
apply exactly the same analytic distinctions that they chose for
systematising their own observations of collaborative discourse. It
is worth elaborating on this difference, as it highlights some
important features of the conceptual framework within which this
research is located. Teasley and Roschelle (1993) note that
conversation is the process whereby their collaborators construct a
joint problem space. Thus, it may seem natural to borrow from
research traditions that are most centrally concerned with the
analysis of conversations. Accordingly, Teasley and Roschelle
organise their documentation of collaborative talk around the
participants techniques for: (i) introducing and accepting new
knowledge, (ii) monitoring ongoing activity for divergence of meaning,
and (iii) repairing such divergences. These are distinctions that are
commonplace within conversational analysis (Clark and Schaeffer, 1989;
Schegloff, 1991). However, adopting this analysis raises the question
of how we should regard collaboration as a distinct species of human
communication - how it is subsumed within that more general category
of communication we simply refer to as "conversation". This, in turn,
raises methodological questions. In particular, whether the
distinctions we make in analysing conversational cohesion are adequate
to capture the special forms of cohesion that might exist in
conversations that are also dubbed "collaborations".
It is certainly the case that conversational analysts have always been
deeply concerned with the "grounding" that must go on during
talk (Clark and Brennan, 1991; Clark and Schaeffer, 1989). This
is a concern, like our present one, with shared understanding. Thus,
one problem that has been studied is how talkers ensure that the
demonstrative references they make in their conversation are specified
well enough to be understood. This can be managed by relying on
conversational grounding. So, if a speaker makes some demonstrative
reference but their referent is actually ambiguous, then listeners may
cope with this by selecting a referent item on the basis of its
'saliance with respect to common ground' (Clark, Schreuder and
Buttrick, 1983, p.296). That is, they select a referent item that is
the most striking option; what defines "striking-ness" is an item's
relationship to the intimate context of mutual knowledge that has been
constructed between these particular conversants.
More generally, participants in conversation must each do work to
establish the mutual belief that what they have said is adequately
understood at that moment. So analysts of routine talk have exposed
important social-interactive techniques that are deployed to help with
this purpose: things that get done to construct the grounding from
which a sense of conversational *cohesion* can arise. Now, I have
been claiming here that the analysis of collaborations (as a
particular kind of interaction) must also concern itself with the
dynamics of such shared understanding. This might imply that analysts
of conversation will have already done the research that is necessary
to clarify how collaborative learning is managed. Or it might imply
that at least they already have available exactly the set of
conceptual distinctions we need. But although there is a way in which
it is appropriate to claim that conversations are occasions that
demand collaboration (Clark and Wilkes-Gibbs, 1986), there are also
good reasons to respect a common sense distinction here. Certainly,
there is no line that cleanly separates out from "conversation" a
class of interaction that we will invariably label as "*a*
collaboration" - a particular kind of conversation. Yet, there is
undoubtably a contrast to be respected and, because of this, we might
suppose that the analysis of collaborations requires additional
distinctions to those that we make in analysing routine conversation.
I believe this is so and will indicate why, before proceeding to study
some particular examples of collaborative talk.
Clark and Wilkes-Gibbs (1986) invoke a *Principle of mutual
responsibility* to express the sense in which all successful
conversations must depend upon collaboration:
The participants in a conversation try to establish, roughly by the
initiation of each new contribution, the mutual belief that the
listeners have understood what the speaker meant in the last utterance
to a criterion sufficient for current purposes. (p.33)
They illustrate the exercise of this Principle with a corpus of
conversational fragments in which speakers and listeners are seen to
be comfortably operating rich repertoires of interactive behaviour:
speakers' verbal 'presentations' are routinely met with various verbal
and non-verbal strategies of 'acceptance' by their listeners. The
progress and repair of any conversation is thus made possible through
mutual commitment to these collaborative processes. Moreover, the
overall structure of such exchanges are said to be managed in ways
that minimise joint effort: this amounts to the so-called *Principle
of least collaborative effort* (op.cit., p.26). Speakers will
experience pressures to respect the natural pace of conversation, as
well as pressures that arise from inherent uncertainty as to their
listener's existing knowledge. Such considerations may encourage
speakers to limit the effort they invest in making their reference
evident (while awaiting confirming feedback). We might say that a
capacity for intersubjective understanding allows them to pursue such
an optimisation strategy.
Yet, the Principle of mutual responsibility, as stated above, invites
further research. For example, studies are needed to clarify what
boundaries serve to define the notion of a "contribution" within
conversations. This issue has already been explored (Clark and
Schaeffer, 1989). However, we also need to investigate more of what
is meant by claiming a listener has understood something to 'a
criterion sufficient for current purposes'. This relates directly to
our present interest in collaborative learning. For Clark and
Schaeffer's phrase reminds us that conversations are indeed conceived
for various purposes, and that the particular ways in which mutual
responsibility is managed in talk will surely reflect this variety.
So, if our 'current purpose' is to have a casual conversation, then
the talk may be allowed to weave and turn in ways that indicate we
have set this criterion to be quite low. On such occasions, the
'effort' invested in establishing mutual understanding may be limited;
we may let uncertainties of reference slip passed; we may tolerate
inconsistency; we may encourage fluctuations of topic. However, if
our 'current purpose' is to *collaborate* - in the narrower, everyday
sense of that purpose - then our criterion may be set rather high.
What, in particular, we do and say to meet that criterion may be quite
distinctive.
Thus, any analysis of collaborative learning or problem solving will
need to go beyond considering only the local management of the
conversational exchange. In particular, the issue of 'accepting' a
conversational contribution (Clark and Schaeffer, 1989) becomes more
problematic: it may not simply amount to understanding what the
speaker intends, but deciding whether it is an acceptable contribution
to a particular common ground that is "working" in favour of the
collaborative purpose. So, our situation in considering peer-based
collaborations must be very much like that confronting Edwards and
Mercer in their analysis of classroom talk. There also, the
participants are motivated by the particular purpose of creating a
consensus of understanding. The established framework of distinctions
for analysing conversational cohesion will certainly apply - for
conversation is surely involved - but that framework will not do the
analytic work that we are most interested in getting done. That is,
it will offer few pointers to how mutual understanding is actively
pursued: how it is refined for the specific purposes of individuals
committed to a "collaborative" enterprise. The joint interest of
collaborators in creating a common product, or in reaching a
consensus, requires that they make a point of attending to this
development of mutual understanding. It must be a more central
concern of their conversation. In any event, the context of their
effort (the materials and resources etc.) will doubtless constrain how
they may proceed.
Proposed analysis of collaborations
I suggest two distinctive features that should be part of any analysis
of interactions that have been termed "collaborative". One is an
attempt to capture both the form and extent of a heightened *concern*
amongst collaborators for the construction of common ground. For, in
collaborative interactions, this purpose is brought into the
foreground of conversation. Thus, the sense of individuals
"collaborating" (as opposed to "merely" conversing) will arise the
more they explicitly reflect upon the creation of this shared
understanding. Analysis of collaborative interactions, therefore,
need to identify the ways in which this patent concern of the
participants is managed: an analysis needs to document the explicit
investment participants make. In other words, a traditional
conversational analysis - dwelling only on local monitoring,
divergence and repair - may miss the active investment collaborators
make in an organised convergence.
Another feature that I believe should be characteristic of any
analysis of collaborations is a sensitivity to how the structure of
the underlying shared task affords different opportunities for
creating this shared understanding. The social business of creating
common ground may be more or less constrained or facilitated in
different contexts for interaction. An analysis of what gets done
must attend to this dynamic - perhaps, in educational contexts, with a
view towards invigorating it. An important theme in my analysis
here is that shared understanding may be created to a greater or
lesser degree, depending on constraints naturally characteristic of
different settings for interaction.
This raises the question of how situations may be more or less
effective for collaboration. Moreover, allowing the idea that there
may be such a variation reminds us that simply putting learners
together for the purposes of joint activity may not necessarily be the
same as prompting them to collaborate. Curiously, most empirical
studies of peer interaction and learning seem to work with just such
an operational definition of collaboration. To be sure, they may
allow that different occasions can be more or less "successful", but
this judgement is typically derived from quantitative profiles
summarising various categories of collaborative action. An occasion
that is effective then becomes one that is rich in favoured categories
of talk - conflict, hypothesising, challenging or whatever. I am
suggesting that we take variation in the quality of interaction (as
"collaboration") more seriously. It may be helpful to regard
collaboration as involving a state of social engagement: one that is
more or less active on any given occasion. It is not something simply
to be taken for granted whenever joint activity is organised; rather,
it is a state of affairs to be diagnosed from the partners' detectable
commitment to constructing a shared understanding.
In summary, I am arguing that empirical analyses of collaborative
learning should, firstly, focus upon participants' access to a shared
understanding, including their explicit concern for elaborating such
mutual knowledge. Secondly, analysis should clarify how particular
contexts of problem solving provide distinctive resources to promote
this interactive achievement. I am not concerned with fixing the
precise semantics of "collaboration" and, in particular, do not wish
to dictate an exclusive field of use for this term. However, I
believe it is helpful to distinguish a state of social engagement that
is defined in terms of a striving after shared understanding; and
helpful to consider how such striving is effectively resourced.
In what follows, I shall describe observations of primary school
children jointly working at each of four computer-based tasks. The
interactions were all organised to blend into normal classroom routine
and the target activities were selected from among the range of those
that were familiar and in use within the schools. I hope that by
looking at rather different contexts, some sense of the variety of
possible shared understandings will emerge. That is, we shall be able
to reflect upon the following processes. How different task
environments afford distinctively different kinds of common ground;
how such mutual knowledge is negotiated and elaborated (beyond the
simple "repairs" of conversational analysis); and how the resource of
common ground is invoked as a platform for the construction of new
moves forward.
SOME GENERAL POINTS OF METHOD
The general procedure behind the observations reported below was as
follows. Schools were chosen where the use of computers was
commonplace and where pupils typically worked at them in pairs - or
small groups. For the present observations, same-sex pairs of pupils
were formed by class teachers and they worked in the normal area where
a computer was kept. This was usually a quiet corner of the classroom
or a reading area between classes.
Making observations in settings such as these is, I believe,
inherently problematic. I am sure that the details of what children
say and do on these occasions is highly sensitive to the social
context - including, the presence and identity of a research observer.
For such reasons, clear empirical relationships found in one study may
defy replication following seemingly trivial changes to features of
the working arrangement observed (cf. Littleton, Light, Joiner, Messer
and Barnes 1992). Indeed, this volatility might discourage any great
faith in parametric studies aimed at isolating "variables" relevant to
predicting collaborative performance. Practitioners may find more use
for broader characterisations of such performance: these may then be
drawn upon to guide the situated judgements that they have to make
within specific circumstances of teaching and learning.
However, the particular significance of these considerations here is
the practical one of refining an observational method. If, as
researchers, we wish to understand the organisation of activity within
classrooms, then our intrusions should disrupt the social order of
classrooms as little as possible. In the present case, the activities
chosen were familiar, being drawn from local resources and curricula;
the occasions of observation were those upon which pupils would
naturally be working in the manner arranged; the style of observation
was discrete although the act of observing was declared, explained and
its confidentiality was discussed. All material reported here was
drawn from sessions towards the end of school years, during which the
researcher would have become a familiar figure. The observer was not
himself present during the sessions that were used for research
purposes: these were video recorded in an open but unobtrusive manner.
Such recording was itself familiar and, with respect to the material
discussed here, on only two occasions did pupils make explicit
reference to being recorded (in each case, playful comments made
during the story composing activity). The adventure simulation task
(Granny's Garden) described below was not video recorded with a
camera. In this case, the observer was present in the background
making field notes, while the pupils' talk and the screen output
signal from their computer were both directed to video tape.
Pupil conversations were transcribed and annotated by reference to
video recordings and notes. Where sections of these transcripts are
reproduced here, the following conventions of presentation are
adopted: the speech is printed towards the left of the page while any
contextualising notes are printed towards the right. All cases
involve pairs of children and they are arbitrary identified as "A" and
"B". A numeral in front of these letters distinguishes a particular
pair - as, say, in the partnership 5A and 5B. The following symbols
are used within the discourse:
/ Short but distinct pause
// Pause of more then 2 seconds
. Gap of irrelevant talk between two segments
.
.
[ Point of overlap between speakers' talk, the overlapping
utterance being printed directly beneath.
? Question intonation
Italic type Emphatic speech
s-p-e-l-l Letters of a word spoken out in turn
TASK 1: ANAGRAM
Six pairs of girls and 6 pairs of boys aged between 7 and 8 were
observed using a program called Anagram. It was selected from a
package of basic literacy programs widely distributed to primary
schools at the time. Its apparent purpose was to encourage
pupils' attention to the structure of simple words by presenting
targets of jumbled letters ("anagrams") and offering the opportunity
to type the correct word underneath each successive target. The
program presents 16 such problems. A display of the current set of
possible words was available at the top of the screen for a pupil to
refer to; upon each correct solution, that word was removed from the
display. Thus, the activity can be summarised as a recurring sequence
of the kind: computer presents target - pupil types response - target
removed from display (once correct entry typed).
Given the typical routine of having children work in pairs, what form
of joint activity is encouraged by a simple structure such as that
described above? I do not wish to become entangled in discussing
whether the children observed can be said to have "collaborated", or
not. Exactly what we take this term to refer to should not be
prescribed here. However, the simplicity of the present task should
help us make sharper distinctions that may prove broadly helpful for
characterising joint activity. So, in describing reactions to this
task, I wish to identify a certain range of interactions that an
activity such as this seems naturally to afford. Later, I shall be
able to contrast such interactions with others that can be observed at
computer-based tasks having different characteristics. My view is
that a relatively rich, articulated interaction could be sustained by
an activity of the general kind described in this section - but that
the present structure is not at all effective. We might even worry
that the activity it supports can sometimes be counter-productive.
Of the 12 pairs of children observed, only two of these partnerships
acted in a way that seemed to invite confident use of the term
"collaborate". The basis for such an intuitive judgement lies in
convergent action and attention: children apparently oriented towards
the same thing. Barring only four trials upon which minor
distractions arose for one member of each partnership, the two pairs I
have mentioned mentioned reliably attended jointly to the target word,
and they also scanned the display of options together. However, the
same could be said for most of the other pairs, most of the time. So,
sustained attention to the computer screen does not, in itself, set
apart the pairs we sense as collaborating more. The single factor
that makes a striking difference to the atmosphere apparent within
these various pairings is the sense in which the activity is being
constructed as a sequence of *turns*. For 10 of the pairs, respecting
this structure was a significant issue. That is, they maintained a
rigid pattern of alternating responsibility for disposing of the
current target word.
In three of the pairs there were periodic exchanges in which the
partners argued about disruptions of their natural activity sequence.
In one case, a brief physical struggle arose. In the example below,
one of the girls completes a word that her partner has started
(perhaps getting impatient, or thinking her partner has become
distracted)
3A: Its my go (A moves to take over typing as B
pauses mid-word, searching for key)
3B: No it's not
3A: I know where the O is (B is now looking for this key)
3B: Got it // that's right (B completes last letter of target)
3A: I have another go 'cos you
had two of my goes
3B: I haven't
3A: Yes you did // my go now
3B: No / you just had a go
3A: You had two of my goes
3B: I didn't
3A: You did (A starts typing answer to new
target)
There are two interpretations of this concern to sustain turn taking
that I wish to confront. The first is the possibility that these
pairs are collaborating effectively: they just happen to assign the
(trivial) business of keying-in to a structure of turns. The second
is the possibility that turn-taking is a strategy to minimise each
individual's involvement in a task that might not be very enjoyable.
I shall appeal to detail of the interaction to indicate that neither
of these interpretations can be readily accepted.
One route towards tackling these questions lies in consideration of
what participants do during the periods when it is not their turn. In
this, there was some variation. With four of the ten turn-taking
pairs, both partners would occasionally call out the correct answer
during the other's turn and while the other was still looking for that
word. In two further pairs this happened, but only one partner acted
this way. Certainly, on the criteria of shared attention we must say
that, most of the time, they are focussed on the same material:
attending to the same target. However, the "collaborative attitude"
of these pairs remains different from that of the two pairs I
identified at the outset as working together effectively. With the
turn-taking pairs, it does not appear that the alternating structure
merely reflects some incidental need to ensure equable participation
in the keying-in (nothing here, or at other times, indicates that this
is seen as a particularly interesting thing to do). In other words,
these pairs are ensuring they get alternate "goes". With these pairs,
where one partner declares the answer during the other's control of
the keyboard this seems more in a spirit of being assertive, or
disruptive - rather than supportive. These contributions may be made
from a withdrawn seating posture, and not accompanied with active
screen pointing to identify the item. It is not unusual for the
keying-in partner to object to such contributions:
7A: Its "like" (Identifying the correct answer)
7B: Alright / I *know* (B keys in l-i-k-e)
7A: My turn
By contrast, the two pairs that give a sense of collaborating do not
display this tension. They are more likely to be in a similar posture
towards the computer. When one of them is keying in a word, if the
other verbally recites the letters, then it is in pace with their
partner's typing. They are likely to monitor reactions in the other,
or talk in the first person plural:
8A: Theres "come" // shall I put in "come"? (A pointing to this word)
8B: Yes / go on
.
.
.
8B: What happens if we get one wrong?
This is not a kind of harmony that is so apparent in the turn-taking
pairs. However, the difference has to be carefully made. This is not
a difference based simply upon some pupils enjoying the task and
others not. Certainly, if some pupils were not properly engaged, then
turn-taking might arise as a way of limiting their enforced
participation. In fact, all pairs were quite animated; they seemed to
like the exercise and individuals sustained their attention when not
taking their own turns. However, in some cases (just the two
collaborating pairs), a common pattern of attention organised by this
task is complemented by an active effort among the partners at
coordination. These efforts are subtle and intermittent (given the
simple nature of the task) but, in small ways, they serve to monitor
and consolidate a shared understanding. So, a demonstrative gesture
may accompany a spoken identification of the correct target - to
ensure mutual recognition. Guesses are sometimes put to partners
before steps are taken to key them in - to ensure endorsement of the
choice. Control of the keyboard is casually managed - because
keyboard disputes are typically symptomatic of the "having of goes"
and (in the case of these collaborators) *both* partners are
effectively having one extended, coordinated "go".
Thus, a small number of pairs convey a sense of collaborating. The
talk and action of these partners is more concerned to monitor, update
and confirm the topics of a potentially shared understanding.
Of course, this is a very modest, undifferentiated task. However,
that may make it a good case to reason from. For, it is just on
the borderline of what naturally might support collaboration. Among an
arbitrary population of pairings, it promotes a variety of
interactions - and some of those patterns of interacting we may want
to say are more "collaborative" than others. In considering these
interactions, I would like to sum up what is taking place in terms of
an investment that is, or is not, made in defining mutual objects of
attention. The case where there is limited investment of this kind
requires us to think about why the effort may be avoided. The case
where it does occur requires us to think about characterising what is
happening in terms of a certain kind of discursive achievement.
On resisting shared reference
There are three particular observations that are suggested by some
children's resistance to adopting a collaborative attitude around this
task. Firstly, the interactions clearly remind us that setting up the
social arrangements for a collaboration by no means ensures that it
will ensue - in the sense of a striving for constructing shared
reference. The participants were motivated, they all conversed in a
lively manner and the task got completed with good humour. Thus,
coordinating the seating and, thereby, facilitating the talk does not,
alone, ensure a coordination of the understandings. Secondly,
these interactions direct our attention to variations in task
design. Part of the shortfall in our expectations for joint activity
must be traced to structural characteristics of the activity itself.
This issue is important to address in relation to computer-based tasks
because they are so interactive. Their design may incorporate
distinct contingencies involving events presented to users and those
users' various inputs in return. In the present case, the rigid
pattern of self-contained word problems may be particularly
well-matched to the emerging strategy of turn taking.
However, this response to the contingencies of the program is not
inevitable. We have seen in the present case that not all pairs of
children adopted it. This variation might be researched in a number
of ways - including reference to the history of interaction enjoyed by
particular pairs, or to similarities in ability on the underlying task
and so on. However, one factor that might be found to moderate
whether such contingencies evoke turn taking strategies or not is the
culture of classrooms or schools.
This is my third point suggested by the present observations: the
existence of collaboration has to be understood against a background
of an institutional culture. This is highlighted by other behaviour
in some of these pairs not yet mentioned. In two pairs, one of the
partners in each case periodically announced during a "go" being taken
by the other that they knew the current answer. Their reluctance to
actually say the word might be taken as reluctance to spoil the
other's turn. However, the manner of these intrusions did not suggest
this. They were announcements made with a more satisfied tone: as if
declaring a certain superiority. The link to institutional life
arises if we suppose that this kind of interaction reflects a
broader pressure to work independently: to become conscious of one's
own understandings as personal achievements.
A related point occurs more forcefully in the interactive style of two
further pairs. In these cases, there was an underlying impression
that one partner was quicker and more fluent at this task than the
other. Indeed, on their own turns, they each reliably found the
targets more quickly. In one of these pairs, this more able
individual conspicuously sighed and heaved on several of the
protracted turns taken by the other. While, in the other pair, on
several trials the more able partner adopted a distinct (and weary)
tutoring tone:
11A: Ummm (looking for key "b")
11B: Its somewhere near the h // (giving a hint)
Don't forget they're capitals (screen letters are lower case,
B assumes is A is confused
by upper case keyboard)
Through these examples, I am implying that a pervasive theme within
classroom culture may be the idea that, as a pupil, one is generally
supposed to solve problems independently. Computer-based joint
activities do not necessarily subvert this (although I shall argue
later that they have a special *potential* for doing so). The present
style of software may even present a task structure that reinforces
the idea of competitive and parallel (rather than collaborative) joint
activity. In that respect it may be counter-productive: providing a
different experience of joint work than that which teachers may be
intending to cultivate.
There is a further concealed feature in this kind of
task/culture/pairing blend that invites investigation. Where a
pattern of discrete problems prompts a turn-taking strategy and where
the turn takers display asymmetry of ability, then the superiority of
one individual relative to the other may become visible and potent -
to them both. This is not just a product of the sequential
organisation of the problems. In addition, these tasks often put a
premium on speed of response and they furnish feedback of a clear and
evaluative kind. In terms of events reported here, we may only
speculate. However, the slower individuals within the asymmetrical
pairs described above may have their attention rather explicitly drawn
to this discrepancy in ability. Of course, it is true that there are
many ways in which such differences are signalled within classroom
life. However, the direct and relentless feedback characteristic of
these computer activities may prompt a more vivid interpersonal
comparison - especially when received in the setting of a close
collaboration. Moreover, there is much evidence to suggest that
pupils of this age are able and inclined to judge their own
intellectual abilities in terms that are relative to their classroom
peers (Ruble, 1988). Some activities may furnish particularly
striking data that can feed such comparisons.
In the example of this simple language program, we have seen that the
nature of the social interactions supported can reflect structural
features within the contingencies of the program and can reflect
aspects of classroom culture. There is certainly a sense in which we
may want to say that - despite comparable levels of animated talk -
children may or may not collaborate when we put them together in this
way. What it is that is "resisted" within these turn-dominated
interactions, is the construction of mutual understanding through
shared reference. In the example of Anagram, the potential for
working this way is present, but limited. I turn next to
summarise what does constructed by those pairs for whom that potential
is, in some part, realised - and why it is necessarily "limited".
On cultivating shared reference
As I have already remarked, there is a straightforward sense in which
all these children have a shared object of attention. They are all
attentive to the computer display that presents the problem: they all
remain on-task in this sense, even when they do adopt a turn-taking
framework of coordination. However, orientation to the same display
defines a very limited sense of sharing reference. It may be what
leads us to claim that they remain *task*-engaged; it does not provide
an adequate basis for claiming that they are *socially*-engaged around
that task. What is required beyond this joint reference to an
external display is a *distribution* of attention across it. In
particular, in the case of the simple anagram program, what is
required is, first, a process of search (for the target solution) then
followed by a process of keying-in the response (itself, comprising a
sequence of letter searches). Thus, the collaborative achievement of
shared reference depends upon an active concern for coordination.
There must be a concern to coordinate reference with respect to the
sequencing of attention and action that the task contingencies invite.
So, in some of these anagram pairs, there is an apparent investment in
ensuring mutual recognition and identification of the target word:
8B: It's "one" isn't it? (This is the target word)
8A: Yea / right / o-n-e
That is, language (and perhaps gesture) is deployed to monitor that
both partners are attending to the same task feature at a particular
moment. A similar kind of investment may be evident during the point
where a chosen word is being typed in:
2A: OK / l // i // k // e (each letter is spoken by
A as soon as preceding one
is typed in by B)
2B: Right // what's this (Start of new trial)
This effort towards sustaining shared reference is brought into relief
by those turn-taking children who, during their partner's turn,
boast that they know the answer but are not going to say it. This is
a strategy that certainly reveals a momentarily shared frame of
external reference; but it also reveals a (considered) failure to
coordinate socially the distribution of attention within that frame.
These partners are doing the opposite of "collaborating" in this
sense: they are withholding the communicative contributions that could
ensure a fusion of reference at a particular moment - and which might
create a platform for subsequent joint action.
This notion of a "platform" - a shared position from which partners
investigate further - is one that I wish to elaborate. In the case of
Anagram, the opportunities for collaborators to construct such a
resource are very modest. They amount to no more than a possibility
of creating a mutuality of reference and attention that can govern the
next response in a cycle of action. There is barely any sense in
which that mutuality can influence subsequent decisions or can
otherwise accumulate to the advantage of the collaborators. If
mutuality were to accumulate in the course of joint problem solving,
then it could be characterised as an internal, private object of
shared understanding. In the next set of classroom observations, I
consider a task that seems particularly rich in its potential for
conjuring up this sort of shared object. That task is the requirement
for partners to compose a story at a word processor.
TASK 2: COMPOSING A STORY
Computer programs of the kind described above are less popular now
than they were at the time my observations were made. Currently,
there is more enthusiasm for generic, tool-like software. Text
processing, in particular, is an activity that surveys suggest is now
widespread within primary education (Becker, 1991). Moreover, the
notion that children might use a word processor as a collaborative
activity has attracted particular attention. In fact, in Britain,
collaborative writing is now a National Curriculum attainment target
(DES, 1989a). Practitioner accounts suggest that collaborative text
processing works well (eg. Crawford, 1988). While more formal
observational research suggests an advantage for organising joint
writing with computers over its organisation with more traditional
media (Davies, 1989; Dickenson, 1986).
Children's early explorations of narrative structure is typically a
socially-organised affair (McNamee, 1979). So, it is natural to
invite pupils to compose stories as a collaborative task. Moreover,
the computer furnishes a setting for them that is well adapted to
joint activity. The product of writing is made equally visible to the
partners; there is no stigma attached to poor handwriting; any editing
that might emerge from discussion can be comfortably executed. In
fact, the example is a good illustration of how a technology
restructures an activity such as to afford richer possibilities of
collaboration (Daiute, 1985). A socio-cultural attitude towards
educational research will find the opportunities implied in such a
situation especially attractive to study.
Here, I shall report observations of 5 pairs of 10-year old children
composing a story together at their classroom computer (an Acorn BBC
Master, equipped with the word processor 'WordWise'). The procedure
was one with which pupils were familiar. Their teacher supplied them
with a one-sentence idea which they were required to use as a trigger
for their own story. In this case the sentence began: "the bus driver
forced on his brakes, but it was too late...."
The children worked at this task for a single session of around an
hour - in the longer cases incorporating a morning break period.
Details of this session length, the amount of talk within it, and the
length of the stories written are given in Table 7.1. It is clear
that there was a lot of talk and that it was fairly evenly distributed
within all the pairs: the dominance of the most talkative partner
varying across pairs from 57% and 68%. One option for systematising
this talk would be to code and categorise the individual utterances.
This might generate profiles for each pair describing, for example,
how many assertions, questions, challenges, endorsing remarks (and so
on) were made within the session. As stated before, I believe this is
a style of analysis that has some value but it is not the one that
will be adopted here. I am more concerned to capture how far partners
are socially "engaged", in the sense of being involved with
constructing and exploiting shared objects of understanding. This
requires examining the talk as coherent discourse - rather than
categories of discrete verbal utterances.
-------------- insert Table 1 about here -----------
How would a shared object of understanding be defined in the context
of composing a story? It might be identified with the text the
children write: the collection of sentences on their computer screen.
This is certainly available as an external, shared point of reference.
So, it might function rather as the display in the Anagram program
described above: an object over which children might chose to
coordinate the distribution of their attention. However, the children
make surprisingly little explicit reference to this product. In only
two pairs was there any re-reading of text further back than the
current sentence. And in these cases (two and three examples
respectively), this seemed to function only as a loose search for
reminders. It was typically read in a quiet, distracted voice (as if
awaiting some inspiration) and there was no explicit discussion of the
material reviewed in this way. While it may have triggered the next
suggestions that a partner made, such suggestions were not discussed
with any reference to text that had been read. On the other hand,
there was quite frequent re-reading of individual sentences during
mid-composition (on average, 38% of sentences got some review of this
kind). However, the purpose seemed to be more one of choosing the best
syntax, rather than furnishing an occasion for developing content.
Yet the transcripts of these interactions do often convey a strong
sense of children focussing on a shared object known to them both.
So, their discussion seems economical in a way that must be
presupposing of particular mutual knowledge. Moreover, as with the
Anagram task, there is also an impression of some pairs being more
engaged in this way than others. What I believe creates this
impression is an underlying variability in how much investment is put
into negotiating the narrative detail. In other words, for some
pairs, the talk is serving to create a relatively articulated (if
implicit) object available for shared reference. This covert
object comprises an accumulation of mutual understanding that
underpins the sentences selected for the final story.
We may be alerted to the fact that there is such an investment
"behind" the story by looking at particulars in the product itself.
Here are a few sentences from the beginning of two of them:
One day in September a coach tour was about to begin on Junction road.
It was going to London for three days with the new Zebra coaches...
One day we were taking some old ladies to London. They wanted to see
Max Bygraves. In one and a quarter hours they stopped to have some
grub. They had a cup of coffee and a glass of wine (very
romantic)....
The elements of this story have their own history. Some of it is to
be linked with very generally-shared cultural experience; while some
of it is more intimately linked to local knowledge. "Junction Road",
"Zebra Coaches", "Max Bygraves" and "old ladies going to London" are
all highly evocative - that is, they promote distinctive narrative
possibilities. Moreover, in these cases at least, such elements have
arisen from discursive processes in which their significance is
mutually established and agreed. Junction Road is a familiar route
near these pupil's school; here are the authors making a commitment to
this as one story ingredient and, in doing so, recognising its
narrative potential:
1B: You could put / like we were going along //
what road?
1A: Oak Tree Road / Oak*wood* Road // (Fictional roads?)
Say we were going along *Junction* Road
1B: Junction / yea
1A: Go on / make it Junction then the lollipop (Traffic control person)
lady can fall in.
1B: Alright
1A: They were going along Junction Road (Reciting possible format
for B to type in)
A similar process is illustrated in the following discussion from
another pair. Here they are concerned to establish characterisation
within the story. In terms of the way the narrative develops, their
discussion and decisions firm up the shared idea of a day trip
involving fun-loving old ladies:
5A: Where could we be taking them? //
To London?
5B: To Scarborough
5A: I know / think of a really old pop star
what grannies would like.
5B: I know / Max Bygraves.
5A: Whose he?
5B: That man who's on Family Fortunes //
says he can sing.
5A: What about that man // oh forgot what he
is called now
5B: Alvin Stardust
5A: I know / Russ Abbott // 'cos old ladies
would like him
5B: Oh no / he can't sing
5A: I know but he's funny
5B: I know // don't put Russ Abbott (Privately, has selected
earlier Bygraves;
this gets typed in)
At later points in the discussion of what to write, it is quite common
for the consequences of earlier understandings to work through; as,
for example, those established in the extract above. They surface as
an influence on current planning. For example:
5A: They sang Old King Cole
5B: We won't want any pop songs / 'cos
they're old ladies
5A: I know
5B: Think of one of Alvin Stardust's songs // (A sings)
What's that called?
In the following extract, an existing, already negotiated context is
invoked to support an interpretation of what could happen once the bus
had fallen down a deep hole in the road.
1A: Suddenly we heard a scream // the (Making suggestions
Lollipop Lady had fallen down // for typing)
No / we all looked up
1B: We wouldn't see anything except the roof
1A: There wouldn't be a roof / it didn't cave
in on top of us
1B: Were we on the top deck or the bottom deck?
1A: The top // but we stuck our head out the
window and looked up //
Right / so we stuck our head out (starts typing)
Characteristically, what gets agreed as the text to input conceals a
body of mutual understanding created in the talk preceding this moment
of typing ('it didn't cave in on top of us'). While further narrative
detail is agreed on the spot, in order to make sense of the current
proposal ('Were we on the top deck or the bottom?').
The process of negotiating a shared understanding is made all the more
visible by the children's sentence-by-sentence composition. There is
very little explicit long term planning. Each development in the
story is worked out at the start of each new sentence - although, as
noted above, care is taken to ensure that new developments cohere with
what has gone before. There are only a few cases of a new sentence
being launched without some advance declaration by the author for
their partner to register. (The few exceptions concern pair 4, one of
whom becomes impatient to finish the story in order that he can start
something else he is wanting to do.) Exactly what happens at these
points of composition reveals something of the collaborative process
in each case - as defined in terms of a concern for constructing
mutual knowledge.
So far, I have raised the idea of a common object of understanding
that is actively created within these pupils' collaborative discourse.
I have also illustrated how investment in this process results in the
construction of a platform: a position from which further composition
might economically proceed. Given this framework, we should next turn
to considering how the creation of such a platform is actually managed
through structures of talking. I shall explore this theme under
three headings below. The first identifies some of the particular
discursive strategies employed to refine shared narrative knowledge
that is genuinely mutual. Pair 5 are chosen to illustrate this theme
as they seem the most strikingly committed to getting this mutuality
right. Other pairs illustrate different themes that surface within
the constructive process, and these are covered under two further
headings below. In particular, I consider the possibility of
conflicting goal structures among partners; and also the notion of
differential ownership, as this might arise in relation to shared
knowledge.
Working to create shared knowledge
Pair 5 displayed the most active investment in negotiating and
refining the understandings behind their written narrative. Each
sentence was reliably preceded by some exchange that established the
next set of options in the story - although further negotiation would
also develop once the writing of a sentence had got started. On no
occasion did a member of this pair start typing a sentence without
declaring what they intended to write. It was also rare for a partner
not to react to these proposals: only two of the 16 sentences written
were begun without some verbal exchange around the proposed content -
and in each of these cases discussion started shortly after a few
words had been typed. In fact, the general impression created by most
pairs was one of composing to the following pattern: at each new
sentence, one partner would announce an idea; this was endorsed or,
otherwise, some discussion followed until a final form of words was
typed. Often this amounted to no more than a phrase, so that a
further period of announce-and-discuss would arise in mid-sentence.
In either case, it was also common for minor editorial suggestions to
be made during the typing of what had been agreed.
The pervasive tendency to declare text before it was typed conveyed
an impression of partners recognising the joint nature of the
task. However, it is from discussion around those declarations that
we get an impression of a richer *collaborative* engagement. For it
is there that shared understanding gets defined and is allowed to
accumulate. The simplest reaction to a partner's narrative idea is to
endorse it - either implicitly by silence (rare), or by some form of
explicit agreement. Other reactions become forms of elaboration or
challenge. So, the simplest way in which a declared idea for writing
becomes elaborated is through what we might term "associative
elaboration". In such cases, the idea becomes a prompt for a partner
to generate a modification by means of some natural semantic
association.
5A: We could be taking // it could be a
school trip / or an old ladies outing
5B: Yea / a grannies' party
5A: Right // one day (Starts typing this)
.
.
5B: They had a cup of coffee
5A: And a glass of water
5B: And a glass of wine // and some wine (Writes this)
In the first exchange, a substitution for a proposed idea was
generated (grannies' party); in the second, the elaboration took the
form of an extension to the original proposal (*and* a glass of water)
which is then itself substituted (wine). There is no reasoning
invoked to justify the changes, but the associations create a
continuity across the partners' contributions. This suggests a
concern to converge upon the same object of shared understanding; the
acceptance of elaborations serving to sustain the construction of a
mutuality.
At other times, contributions were challenged rather than elaborated.
However, if partners were still working to sustain shared reference,
then we would expect to find these challenges accompanied by
commentary that makes some sense of rejecting the idea in question.
5B: Where could we take them to
5A: I know Scarborough
5B: Whitby
5A: No
5B: Thats too far away for them isn't it
5A: I know / sunny Spain
5B: You wouldn't get to Spain on a bus I don't think
The Scarborough-Whitby exchange is an associative elaboration
(both towns being seaside resorts in Northern England). Whitby
received the stark challenge "No". Interestingly, this prompted a
self-reviewing response from the original proposer: the rejection was
rationalised in terms of the place being "too far away". In this way,
some possible tension may have been avoided. While the next
suggestion (possibly frivolous) was also rejected, this time with the
listener furnishing the reviewing comment that makes some sense of
denying the speaker's suggestion.
Responsibilities for challenging a proposal and, then, reviewing the
basis for a challenge can be the reverse of what is normally expected
in argument. In the example below, the challenge came from the author
of the proposal ('hang on a minute..') while the reviewing commentary
came from the listener who found reasons for persisting with their
partner's original idea:
5A: We stopped to have our lunch (Proposal for text)
5B: And then had a little walk / hang on
a minute some of the old ladies couldn't
have a walk / some of them would be on
wheelchairs
5A: No 'cos they would be on walking wheels
5B: Walking wheels?
5A: No / like crutches (Matter seems to be
settled by this)
In a traditional analysis of such talk, 5B's self-challenge
('hang on a minute..') might be coded as an utterance with a
particular pragmatic content: perhaps some species of articulating
one's own reasoning. Such a coding might then increment 5B's standing
(or the standing of the pair) on a researcher's summary statement of
their interaction. There is some value in this traditional analysis,
but it fails to address how such collaborative talk is effectively
occasioned. It fails to reveal how the talk arises from participation
in a set of particular circumstances. The talk itself is surely
unexceptional: we can be sure children deploy the rhetoric of
challenging, reviewing, reflecting and so on, as they argue in the
playground. The interesting question concerns how this rhetoric
becomes mobilised to support a schooled task - and, moreover, why it
becomes vigorously mobilised within some pairs but less so in others.
Here I am suggesting that the focus of our approach to such questions
should be on the nature of an evolving shared understanding and, then,
on a given partnership's varying commitment to investing in this. The
dialogue above illustrates how reasoning (in this case, about
narrative) may be motivated by a concern for respecting and
contributing to a jointly-constructed cognitive object - the context,
events and characterisations that lie behind and inspire a written
story. So, the same concern motivates my final example of discursive
effort clearly visible in these interactions.
Joint thinking of the kind promoted here will often take the form of
an exchange in which there is a convergence upon a solution, but not
by challenging and re-casting some initial suggestion in the manner
illustrated above. In these further cases, the reasoning is more of a
"calculation". The participants are seeking to refine a suitable
status or "value" for some story ingredient that they are toying with.
In the following case, the focus is literally quantitative in this
sense:
5B: So it took two hours (Planning to write)
5A: It didn't because
[
5B: it took two hours to (Thinks A has
get *half* way there misunderstood)
5A: It doesn't because my nanna lives in London
5B: And how long does it take to get all the way
5A: About five hours / right?
5B: So it'll have to be two and a half
5A: But we stopped for half hour lunch
5B: Did you // so that means it was four and a half
hours // so that would be two and a quarter
hours (Writes this value)
5A's contributions here have some force: she has been to London and
claims to remember the journey time. More often, these kind of
convergences can not be directed by reference to such external
authority. The criteria for decisions will be looser - more what is
amusing, or interesting, or consistent with the narrative so far. Of
course, in such cases, the argument may well turn on their common
access to a shared understanding about that narrative. Such
circumstances suggest that joint story writing is both a good and a
poor model system for considering schooled collaborations. It is good
because it does involve creating a form of joint understanding that is
particularly vivid. For the achievement is continuous with something
familiar in more playful experiences - fantasy-making in general. It
is a poor model system because conflicts over what should be said next
are not so easily resolved. There can often be no legislating
reference to some logic inherent in the joint knowledge that has been
created: a conflict can not necessarily be resolved by appealing to
some privileged way for developing an existing narrative.
Pair 5 seemed successful as collaborators because they managed
effectively the resource of such shared understanding. They
freely entered into exchanges that served to enrich the detail of this
common knowledge. Moreover, they used the resulting structure as a
platform for their joint progress: they appealed to it in order to
establish consistency and coherence for their ideas. In particular,
this effort was realised in an equable manner. In the next section, I
shall refer to dialogue from other pairs where the symmetry of making
contributions was not so striking. This raises the issue of partners
adopting differing responsibility for creating a structure of shared
knowledge.
On owning shared knowledge
Pairs 1 and 3 also engaged in some of the constructive effort
illustrated above for Pair 5. They were certainly as talkative and as
task-oriented (in neither pair was there any reference to off-task
concerns). Moreover, they were as concerned to declare, elaborate,
challenge, review and calculate the content of a shared knowledge.
Their talk exemplified the general processes that dealt with such
things and which were illustrated in the last section. What was more
characteristic of these pairs, however, was an asymmetry of
responsibility for the content of that common knowledge.
In Pair 3, the first sentences were characterised by give-and-take of
a kind that created for both partners a stake in the narrative:
3A: We have to name like three important people (= central characters)
3B: There were three *important* people going? (= special characters?)
3A: No / not like that / like *main* people (disambiguates)
3B: Like the bus drivers name // the bus driver
was called Alex
3A: No / put like there were three boys / yea
three girls like Vicki, Jean and Shelley
3B: They were going because they won a tickey
3A: Yea / they could have won a competition
Here A is concerned to establish the idea that there are to be three
principle characters and, then, who they should be. B's attempt to
include the driver is denied in favour of focussing on three children.
While B's later idea that these children had won the trip is then
accepted. In constructing later sentences, however, B's contributions
increasingly follow the fate of the bus driver example above. The
extracts below are from sentences 3, 8 and 10 respectively:
3A: It had gone 32 miles when they had the
first stop
3B: When the bus broke down
3A: When they had their first stop
3B: Yea
.
.
3B: At a cafe
3A: No / a restaurant // 'cos they wouldn't stop
at a cafe they wouldn't all fit in
.
.
3B: What did they eat?
3A: they had
3B: Egg
3A: A three course meal / including
3B: A cup of tea afterwards // no including VAT
3A: No // including the drink
3B: Including the drink of cocoa (A writes: 'Including the
drink with it')
This final case is typical; all B's specific contributions (egg, tea,
VAT, cocoa) fail to get incorporated. Perhaps the general idea of
drinking was hers, but that was probably already implicit in A's 'meal
including....' For most of the sentences composed by this pair, B's
contributions were either acknowledged and rejected or re-cast.
Although often this might involve some sense-making by A, as in the
second of the three three extracts above.
The interaction of Pair 1 was similarly asymmetrical, although the
dynamic of their exchange was different. B makes one suggestion early
on that is taken up: in the first sentence she proposed that the
characters are on the bus because they are 'going to the pictures'.
However, thereafter, although she remained engaged with the writing
and the composition, all the contributions originated from A. This is
not because A re-casts B's ideas (as in Pair 3 above) but because B
makes very few specific suggestions. Instead, she participated by
querying all A's input. It is not clear if this persistence occurs
because A never gives ground to negotiate an option, or whether B has
adopted a sceptical attitude for some other reason of her own.
1A: We had already begun to fall down the hole
1B: It wouldn't have been that deep / we'd soon
hit the bottom
1A: Not if it landed on a load of things
.
.
1A: When we hit the bottom everyone began to panic
1B: Not everybody would // the driver wouldn't
1A: I would if I was him
.
.
1A: When she got to the top / she pulled the rope down
1B: This lollipop lady isn't going to do all these
heroics
1A: She would
1B: Probably she'd bring down the food and somebody
else would do the rest
1A: No she'd go up and then throw a rope down to us
/ and hold it at the top
1B: I don't think she'd have enough strength / this
lollipop lady
A (amicably) persisted in her suggestions, justifying them by appeal
to the variety of possible motives and circumstances that a
still-developing narrative can furnish. B also persisted in
challenging most of these suggestions. The result is a sense of both
partners oriented to shared knowledge but (as with Pair 3) a concern
for ownership that leads one partner to take responsibility for most
of the final content. In Pair 3, this is accepted harmoniously; while
in Pair 1, the situation creates more of a tense atmosphere.
Evidently, for these pairs, there is a perfectly proper sense in which
the knowledge base supporting their talk can be said to be "shared".
It is known to both partners and it, thereby, provides the necessary
basis for new narrative ideas that either of them may propose.
However, we also see with these pairs how it need not be "shared" in
the sense of co-constructed. While the talk of both partners may
address a common object of reference, the form that object comes to
take arises from contributions that are more the responsibility of one
partner. I can not pursue empirically the significance of this
asymmetry; however, I believe it requires us to consider a feature of
pupil collaborations that has been neglected by developmental
psychologists. That feature is the *affective* dimension of this
joint activity.
There is not enough relevant data to allow reasoning about the
emotional tone of the interactions described here so far. For
example, it is true that Pair 5 convey an exuberance that might be
associated with their equable investment in creating joint knowledge -
an achievement that then can support them in story-telling. However,
this observation is merely one of correlation: it is not clear whether
effective co-construction is directly arousing in this way, or whether
it is a mere correlate of something else that is. On the other hand,
the idea surely resonates with our intuitions about successful
collaborative thinking. Building a resource of mutually familiar
understanding is often an emotionally engaging experience; perhaps
because of a certain intimacy that it affords. The achievement allows
us to communicate on a topic in a heavily presupposing way - something
like what we do when reflecting more privately on the same topic.
This observation suggests *negative* affect could arise within these
interactions. In particular, this might occur when collaboration
requires that a partner must deploy shared understanding to which they
have *not* been allowed to make a fair contribution. This might be
the experience of partner B in Pair 1 described above.
In the interactions studied here, there is one further basis for a
less than successful construction of shared knowledge. This concerns
the motives that participants may have for engaging in the joint
activity involved.
On the goals for shared knowledge
In this section I shall briefly refer to the interactions of Pairs 2
and 4. They also made a more modest investment in shared understanding
but not for reasons of any asymmetry in the making of contributions.
For these pairs, the achievement of mutual narrative understanding
seemed a less important goal of the activity. Evidently this is a
straightforward way in which collaborations may be less successful:
the setting does not motivate the construction of mutual knowledge.
In the case of Pair 2, both partners contributed to the development of
a story: that is, the constructions written can be traced about
equally to each person's contribution. All of what was written was
publically declared before any typing started. However, these
contributions were not refined and enriched within the talk. On only
three occasions was there any challenge or elaboration of a proposal.
In one case this was to maintain consistency with narrative events
that had been established earlier. Much of the talk was about the
mechanics of writing the story. Thus, 52% of the dialogue concerned
the keyboard, spelling, checking how much had been typed, or reciting
words in synchrony with the typing. We can not claim that these
pupils were not engaged with the task (they talked continuously and
rarely referred to off-task concerns) and we can not claim that they
did not each make contributions to the overall activity. However,
their contributions were not oriented towards the construction of a
social resource that that might empower the story-telling. It is not
reasonable to speculate here as to what their primary goal might have
been. It seemed to be more related to the delivery of a product that
was too narrowly defined - understood only in terms of being a certain
length, grammatically accurate and so on. Thus, the creative
possibilities of a collaborative orientation to shared understanding
was poorly motivated in this pair.
In Pair 4, the shortfall had more to do with *competing* goal
structures among the participants. During the first half of the
session, they developed shared understanding of the story much as Pair
5. However, partner B increasingly presses contributions that would
bring the story to a quicker end. At the start of the extracts
below, A refers to the fact that B is anxious to finish and do
something else ('your animal thing'):
4B: That can be the end // we got out of the hole
4A: No / just because you want to do your
animal thing
4B: But what can we put?
4A: We can put we went down caves / we left the
people in a safe cave
4B: But we already got them out
.
.
4B: We are not going to put them in caves //
They are not dosey you know
4A: No / we are going to put them
[
4B: They could walk home
or jump in the bus
4A: We put them in a safe cave / right?
4B: Thats going to make it a funny story
The dialogue proceeds in the manner illustrated above. The challenges
from B entail alternatives, the effect of which seems to be to make
the story come to a speedier conclusion. Evidently, the prospects for
creating a resource of shared knowledge do depend upon both partners
framing the goals of the exercise in similar terms. Where this has
not happened and where there is some incompatibility of the goals that
have been set (as here), then the experience of collaborating may be
less positive.
Summary comments on joint story composition
Compared to the Anagram task discussed earlier, composing a story
involves a more elaborate investment in creating common ground. The
task provides a rich opportunity to achieve some intimacy and scope
of shared knowledge. One suggestion has been that this achievement is
emotionally gratifying. However, we have also considered obstacles to
satisfactory building of shared knowledge, and they may generate more
negative experiences among the collaborators. For example, some
partners may work to claim greater ownership over the knowledge that
is shared. Others pairs may fail to end up working from a rich common
knowledge, because they differ in their understanding of the purposes
driving their joint activity.
In all, the close consideration of these interactions highlight three
important points. (1) Collaboration is clearly seen to involve an
active concern for the construction of mutual understandings. The
extent to which collaborators are engaged in this has been overlooked
in more traditional psychological analyses. (2) The object of shared
understanding that can emerge within a collaboration needs to be
properly understood. For example, the case a shared *narrative*
understanding can be problematic because partners may have to deal
with an uncertainty as to how a new narrative proposal is properly
legitimatised in reference to the existing structure. (3) There are
various obstacles to the construction of shared understanding that is
endorsed and respected by both partners. Such obstacles may be
inevitable but they may also be more effectively anticipated with
suitable attention to task design, task description, the clarification
of purposes and, perhaps, social knowledge about the partners
themselves.
The present account highlights the importance of collaboration as
involving a concern for shared objects for reference. In Anagram, the
screen display furnished such an object: an external stimulus that
could serve to organise the distribution of partners' attention. In
telling a story the shared object becomes something less visible -
certainly, something beyond the screen display. However, some caution
is needed in pursuing an analysis of this sort. It might not be
helpful to reify what it is that can get created within a
collaborative interaction. It might not be helpful to cast this
achievement in terms that inevitably suggest a slightly mysterious
sort of "object", suspended somewhere between the collaborators. Such
language may be of some shorthand use - as long as it is not developed
in too literal a spirit. It certainly helps keep in mind one
important idea: that effective collaborators are able to coordinate
the focus of their interest or attention - that they have available a
distinctively shared point for reference in their deliberations.
Perhaps this is why a metaphor is useful here. It helps us keep in
mind some key feature of what we are trying to understand. That
feature may offer helpful implications for how we organise research:
in this case, towards clarifying the discursive management of joint
attention. By the same token, Edwards and Mercer can usefully invoke
a variant of their own on the basic metaphor:
Overt messages, things actually said, are only a small part of the
total communication. They are only the tips of icebergs, in which the
great hidden mass beneath is essential to the nature of what is
openly visible above the waterline. (1987, p.160)
Perhaps 'icebergs' serve to distract us from the coding and counting
of utterances. The metaphor encourages, instead, the framing of what
is said at a particular moment in terms of a more extended discursive
effort: a context that is not adequately specified in terms of events
visible only at the moment of interaction. This is a helpful
evocation.
Referring to 'objects' (including those of Artic proportions) can,
therefore, be productive. However, as observers, our access to what
has been achieved remains located in whatever discourse and action we
are able to witness. Analysis should not become preoccupied with
abstracting from such events some independent cognitive 'object' of
social origin. One way in which what is typically achieved during
collaboration can be expressed more precisely is by incorporating
reference to an essential mutuality. Partner 'A' knows things
relevant to the problem at hand and arising from previous
collaborative action; 'A' also knows that partner 'B' knows that 'A'
knows these things; moreover, 'B' knows that 'A' knows that 'B' knows
them also....and so on. Thus, what they have achieved is based upon
the possibility of such intersubjective understanding. However, the
achievement is still not an inventory of social actions to be
comprehensively reproduced by researchers: talking of cognitive
objects here does not take us that much further forward. We may make
most progress if, as researchers, we concentrate our attention on how
language and action is deployed (i) in the interests of elaborating
and refining this common understanding and (ii) in ways that take such
understanding for granted and, thereby, help focus more precisely what
gets done next.
TASK 3: ADVENTURE SOFTWARE
The first task (Anagram) described in this chapter illustrated a
rather poor structure for collaboration. The program did not readily
offer pupils a resource of mutual knowledge. It provided no
opportunity for the *accumulation* of shared experience. If such an
achievement was possible at all, it was through orchestrating joint
attention towards the modest events that defined a particular anagram
problem. However, the cycling pattern of *self-contained* trials
provided no strong motive for such coordination and, instead, pupils
often adopted an attitude of alternating "goes". In contrast, the
story-telling of the task discussed above provided a more effective
vehicle for this co-construction of understandings. The final
achievement in this case is richer, but also more subtle. It is not
some stimulus array located between the collaborators. The object of
joint attention - the narrative knowledge - is something more private.
Fortunately, preschool children's experience with socio-dramatic play
will equip them very well to coordinate their interests towards
creating such a resource.
Yet, the familiarity and appeal of narrative formats does not mean
that creating shared knowledge around a story is going to be
unproblematic. Appropriating story-making to become an activity for
the classroom requires the imposition of certain extra demands that
might not arise in more playful arenas. In particular, there is a
pressure for closure: a pressure on the story-writers to persist
towards the production of a single object. Tension may then arise
because of a looseness to the authority that any developing narrative
can claim over the individual collaborators who are constructing it.
So, if their individual contributions diverge, there may be no easy
appeal to the *necessity* of a particular narrative route.
There is a popular category of early educational software that seems
to exploit some of the potency of narrative understanding, while
creating for decision-making a clearer source of authority: one that
collaborators might productively refer to in their discussions. Such
software uses the metaphor of an "adventure". Normally, this involves
programming a problem-solving activity to be represented in some
narrative format. Thus, pupils follow through a story, becoming
participants by responding to various challenges or puzzles that are
arranged to occur. Their responses may determine the course of the
story, or their own fate within it. The example to be discussed here
is the program "Granny's Garden". This has been an extremely popular
activity within British primary schools (Bleach, 1986) and there are
numerous accounts of classroom practice that claim the program can be
a good stimulus for collaborative work (eg., Farish, 1989; Hill and
Browne, 1988). Pupils move though a sequence of text-and-graphics
displays by making various keyboard responses. Sometimes these
responses will be the answer to puzzles that have been posed in the
current display. These puzzles usually involve reasoning or
remembering in relation to events or objects in the story. This
particular story entails searching for a number of royal children
kidnapped and hidden by a wicked witch. The pupils' journey starts in
Granny's Garden and goes through four distinct sections with their own
motifs and themes. Each section concludes, hopefully, with the
discovery of one of the four missing children - or, where things go
wrong, with the witch sending pupils back to the start of that
section.
The same pupils who worked on the Anagram task (described earlier)
were subjects for the observations made here. Twelve pairs of 7-year
old children used the program on five separate sessions. The final
two of these sessions provided material for the analysis reported
below. Their speech was preserved on a video recorder along with the
corresponding computer screen displays. This talk was transcribed and
annotated with material recorded by an observer present in the
background of these sessions.
In carrying out such a task collaboratively, what possibilities exist
for pupils to create shared knowledge, and how may it then be called
upon to support collaborative reasoning? As claimed for the
story-writers above, the knowledge that collaborators share within
this activity goes beyond their joint access to the transient events
portrayed on screen displays at a particular moment. It is knowledge
that has accumulated during the course of coordinating activity
towards such events, but over some period of time. It is narrative
knowledge but, unlike that of the story-*writers*, it is less of the
collaborators' own making. Yet, this knowledge is not imposed either:
not to be defined merely in terms of the story designed by the
programmer. In other words, a particular pair of pupils will know
about their own circumscribed experiences of engaging with this story.
They will know about particular decisions they themselves made in the
narrative sequence; they will know about particular consequences and
their reactions to them. Here are two pupils calling upon such shared
knowledge:
3A: Where shall we go? (Of four locations)
3B: The stairs?
3A: No / 'cos the snake remember
3B: We've never been in the backroom (An untried location)
3A: I know // if there's anything like you have to
pick up / we won't // in case we get caught
Certainly, there are constraints to what is understood; they arise
from contingencies programmed in the adventure. However, there is
also an open-endedness. These pupils have had a *particular* set of
encounters with that narrative and, from such experience, they now
share distinctive memories and concerns. In the example above, this
is reflected in apprehensions they have about snakes appearing, and in
projections they make that any objects accidentally encountered might be
usefully avoided.
For purposes of appreciating more of where this joint knowledge was
constructed, the talk of these pairs was separated according to its
relation to the contingencies of the task. Distinct sections of talk
were marked on the transcripts and, across the pairs, the mean
percentage of the conversational investment at each point was noted.
So, on the average, 19% of talk was concerned with input matters
(i.e., reading from the screen) and 14% was concerned with output
matters (keying-in responses). 6% of talk made very general
references to the activity or its execution (i.e., evaluative comments
or comments not relevant to the current problem). Remaining were
remarks about the current problem or the question currently posed (44%
of talk) and, finally, remarks about the outcome or consequences of
decisions (17% of talk).
Mutuality was most clearly created within the discourse making up
these last two categories. For example, "outcome" talk is
particularly effective in establishing overlapping versions of the
events comprising the adventure. There was a great deal of evaluative
and emotive reaction in this talk. In the example below, the children
are reacting to a character in the story who asks for the children's
favourite food and then (invariably, it seems) comments that the food
makes him sneeze:
9A: I know why he sneezes at every food we mention
9B: Why?
9A: 'Cos he sneezes wherever he is / he sneezes
whatever he's doing or wherever he is
9b: What if we put pepper so he really would sneeze?
Their spontaneous commentary on the narrative serves to increment
their joint understanding of events, contexts and characters. This is
similarly illustrated in the further examples below:
10A: Here's the Raven coming
10B: He's flying down to see what we are up to
2B: I don't like bees // they sting you
2A: But this one is our friend
2B: I know but
[
2B: but this one is our friend
4A: Hey // it looks like Brian from the
Magic Roundabout
In the above, Pair 10 are watching a bird that crosses their screen
and who is introduced early in the adventure as a creature that will
help them: the children's talk serves to personalise further their own
participation in the story. Pair 2 are reacting to a bee who has
appeared with the possible intention of helping them in relation to a
particular problem. Again, they seem to be establishing their
attitude to this character: locating him in relation to their joint
engagement with the story. 4A is forging a particular link to a
popular character known to them both through video tapes.
From encountering the narrative events, from developing an
interpretative commentary about them, and from a history of deciding
about how choices are to be made, these pairs will construct their
distinctive version of the narrative. Moreover, it will be mutually
understood: each member of a pair can project this understanding into
their partner and exploit the mutual knowledge when they come to
reason about decisions to be taken at any particular juncture in the
adventure. This is very similar to the analysis of story composition
offered earlier in this chapter. Except that, here, the format
involves pre-established relationships and contingencies in the story
and these carry a certain authority in terms of decisions that have to
be made. This is particularly felt where children are re-tracing a
section of the narrative (perhaps having been returned by the witch to
an earlier starting point - by way of penalty). More simply, it will
be felt where options for a current decision are constrained by what
has most recently been done, and this will have to be jointly
recalled:
6A: Where do we go now?
6B: Not the cupboard 'cos we been there
// not some stairs / no way (Acts scared)
6A: Only got three left then // kitchen?
6B: If you want to (Not hopeful)
6A: No I think the witch is there // let's go
upstairs / 'cos we got an apple remember
6B: What will that do / because the snakes upstairs
6A: And we throw the apple at it
6B: And what will that do?
// the witch will come wont it?
6A: Yeah (Sounds deflated)
If the children are to reason about the problem in the adventure,
then part of what must be achieved is an agreement about what has
already happened to them. So, as in the example above, the talk will
sometimes be deployed in the service of a collective remembering
(Middleton and Edwards, 1990). The above example also illustrates how
*inferred* aspects of shared memory may inform the reasoning that
takes place. Thus, the remark "'cos we got an apple" leaves a lot
unstated. Partner A's assumption is that B knows that apples promise
a kind of insurance because - from past experience that they have
shared and know to have shared - throwing apples at threatening agents
can get you out of trouble. (B, however, also recalls that the witch
will sometimes appear on these occasions and successfully reminds A of
this possibility as being more likely - at this point.)
There is a significant general point within that last example. If we
study discourse in order to clarify that shared knowledge is in place
and that it is informing the present solution of problems, then we
face a curious difficulty. For, where there is a well-developed
shared knowledge, an important consequence will be an economy in what
does then need to be said. This collaborative achievement allows
participants to assume that the context or background to their remarks
is known; so they need say less. Thus, our difficulty as observers,
is that this knowledge-building achievement must be traced by us more
in terms of what is *not* said. This may be felt as an awkward
challenge by researchers whose training encourages them to index the
social purposes of talk by coding and counting distinctive kinds of
utterances.
As an exercise in building and reasoning from shared knowledge, the
adventure game format has a lot to recommend it. The appeal and
accessibility of the narrative structure effectively motivates mutual
engagement in the task. Moreover, the pre-set nature of the
underlying adventure provides an external authority for evaluating the
decisions that collaborators make. Under these circumstances, it
seems less likely that asymmetries in the working arrangements will
arise. The following passage is relevant to these observations. The
children are deciding which of a number of foods they should throw, in
order to tame a particular dragon:
5A: Oh / try chips
5B: I don't think we've any chips left
5A: We should have been counting //
We should have a piece of paper
5B: We can't use chips
5A: Try them / and see what happens
5B: Well / it's your fault if we get caught you know
The third person plural references indicate a strong degree of mutual
engagement in the activity. Their problem with choosing chips
indicates how much decision making in this kind of activity relies
upon fairly straightforward remembering: recalling either what
happened on a previous occasion or what they have just done. In this
sense, the activity is very limited - the underlying contingencies in
the narrative are not programmed to vary between occasions of use. On
the other hand, the extract above illustrates how reasoning in this
framework can still sometimes be a potent experience: by encountering
limits in their capacity to jointly remember events, these children
reflect on the value of developing a strategic approach based upon
pencil and paper. Finally, the extract illustrates that a fracturing
of the collaboration can still occur. Where they are forced into
making a sheer guess, the assertive partner faces it being "your fault
if we get caught".
The Adventure format is recognised in early education as a structure
that is engaging and that supports joint work. It illustrates an
ingenious transformation of children's playful interest in narrative -
borrowing it to support the organised presentation of schooled
problems. I have indicated how this format allows pupils to create
shared understandings; this can then serve as a resource for the
management of joint reference during problem solving. In the longer
run, an analyses such as this should contribute to refinements in the
design of classroom materials (eg., educational software) - so as to
make them more effective for collaborators. I will conclude with two
observations relevant to that practical aim. The first concerns how
we may go about further clarifying the nature of the joint knowledge
arising in activities like Granny's Garden. The second concerns the
recurring theme of managing asymmetry in the joint activity.
For any activity of the present kind, it will be important to evaluate
the extent to which the potency of shared understandings depends upon
them having been co-constructed. For example, it would be possible
for two pupils first to use an Adventure program separately. They
might then come together and reason in a lively and effective way,
based upon inferences about what the other *ought* to know. A
collaborator would naturally make assumptions about their partner's
knowledge of this program - in the same manner that they would make
assumptions about a vast array of worldly experience that a partner
might have, some of which might be relevant to the current problem
solving. On the other hand, the dialogue recorded in this study
displayed a good deal of more intimate reference. The partners would
often refer to what happened last time *they* did something, or they
would make idiosyncratic evaluative comments that could not be
be part of knowledge shared with a new partner. An interesting
challenge for research will be to evaluate how important this more
intimate sharing of experience is in sustaining an effective
collaborative effort. Children working with the same partner across a
series of problems may develop more flexible strategies and greater
success than children working with a different partner at each
occasion (Goldberg and Maccoby, 1964). Others studies suggest that
partners can require several sessions to develop an effective problem
solving style (Forman and Cazden, 1985).
My second observation arising from this activity concerns how the
structure of the program affords more or less equable patterns of
engagement. Asymmetries of this kind were less evident than in the
other tasks described in this chapter. The fact that these were the
same children who acted rather un-collaboratively on Anagram reminds
us of how important those structural details are for determining the
dynamic joint work. However, in 7 of these 12 pairs there was a
reliable (occurring on every session) asymmetry in the responsibility
for reading text from the screen. Moreover, the dominance of screen
reading correlated significantly (r = 0.80) with dominance in
controlling the keyboard for moving the adventure sequence forward.
Subsequent to the collaborative sessions described above, individual
children were asked to work through the first part of the activity
alone. The median time to move through each frame of the story was
recorded and correlated with their dominance when collaborating. This
was also significant (r = 0.68), such that the fast individual readers
were those who tended to control the display during collaborations.
In a very straightforward way, this relationship suggests how
computer-based tasks may have structural features that constrain or
facilitate certain patterns of joint activity around them. The
framework of socio-cultural theorising tends to encourage analysing
cognitive practices in terms of these affordances. What is needed is
a more lively empirical interest in exploring them.
TASK 4: FACTOR SNAKE
The examples discussed above illustrate how various circumstances of
collaboration present their own distinctive possibilities for the
creation of shared knowledge. The goals of any given problem-solving
task and the contingencies within it will structure a setting in
terms of such possibilities. Thereby, they will permit a resource of
mutual knowledge to be developed with more or less effectiveness. I
am suggesting that the potency of any setting for the support of
collaborative learning depends upon its potential in this sense. If
partners are able to create a well-articulated object of shared
reference, then they will have equipped themselves with a real
platform for exploratory discussion. In studying such occasions, the
significance of particular instances of rhetorical talk between
collaborators needs to be analysed in respect of its relationship to
this shared knowledge - and not just enumerated into a profile of
utterance categories.
Given our interest here in new educational technology, the relevant
questions must now concern how computer-based tasks can be best
designed to resource this 'common knowledge building' enterprise. I
will conclude this chapter by reporting on an development exercise of
my own that was inspired by this concern. My aim was to take an
existing program of modest effectiveness and refine it for the
specific purpose of better supporting collaborative engagement. The
program (Factor Snake) was described in some detail within Chapter 5.
In that context, it served to illustrate the limitations of a
computer-based experience that was not assimilated into the mainstream
of teacher-class collaborative talk. In the present context, I am
considering more its limitations in respect of supporting another form
of collaborative exchange: that between pupils at the time that they
are using the program.
The program could be regarded as an animated version of Dienes Blocks:
these are familiar classroom materials for visually representing
numbers as rectangular arrangements that illustrate their factor
structure. For example, the number 12 could be represented as various
matrices - 1 x 12, 2 x 6, 3 * 4 etc. (see Figure 5.1). The Factor
Snake program represented a number by continually winding and
unwinding (snake-like) a set of small squares: this sequence covered
all the possible factor matrices for a given number. The speed of
this animation was such as to make it difficult to count more than
about 7 squares in any given line. Guessing the value of larger
numbers therefore depended on attention to the factor structure of a
matrix. The pupils thereby executed some arithmetic calculating (eg,
4+4+4 or 3 x 4).
This program was used periodically across a school year in one
particular top infant class (children aged between 6 and 7 years in
their third year of schooling). The present observations were made
towards the end of that year and involved two groups, each comprising
10 pupils. Group 1 used the program in a standard format throughout.
Group 2 did also, except on the final three sessions of the year, when
these pupils were provided with an extra programmed resource (to be
described below). Comparisons reported here between the groups are
based upon the very last session of each, during which video
recordings were made.
The standard format for the program was as follows. Children had to
estimate a succession of values taken by this animated snake: we may
call these efforts "trials". The set sise for possible snake values
was incremented on each new trial, thus making the task progressively
more difficult. Correct estimates resulted in a screen-based score
being increased by the value of the current snake target. Incorrect
guesses resulted in feedback and a resetting of the score: target set
sise was also reset and a new sequence of trials initiated.
Estimating involved using a mouse-driven screen pointer to select a
number (from a screen bank showing 1-59) corresponding to the current
snake target. During the year, the program had been set to deliver
target numbers only in rectangular matrix form. For these final
sessions, numbers were described by non-rectangular matrices. These,
therefore, would be unfamiliar shapes; they demanded calculating a
matrix value and adding the "extra" number of squares appearing as an
incomplete bottom row. In summary, children had to deploy their
emerging knowledge of factors to make estimates of
visually-represented numbers that could not always be directly
counted.
The extra resource provided for Group 2's final sessions might be
called a "workpad". In relation to the building of shared knowledge,
it was designed to provide an external support for such efforts. It
was to be a tool towards which collaborators could direct their
attention and action. By placing their pointer on a particular screen
icon, a 4-cm square writable area appeared on the display. When the
pointer was in this area, clicks on the mouse button caused one small
fixed square to be drawn at that point; these squares were the same
dimension as those making up the snake (that continued to wind and
unwind at all times). Pressing again on a workpad square drawn in
this way would erase it. The net effect is illustrated in Figure 7.1:
it must be imagined that a pupil is aiming to make a (static)
reproduction of one of the factor matrices that constitutes the
current snake. Evidently, the construction of this replica requires
the children to attend directly to the parameters of the target
matrix.
Initial impressions from recording of the interactions suggested a
noticeable difference between the groups. This is certainly reflected
in the overall amount of talking. While Group 1 were quite lively,
there was 30% more talk among Group 2 children. However, this
difference will not be our main interest: it might be expected, given
that the workpad created more involved contingencies for Group 2 to
talk more about. What is of greater interest is the effect of this
feature on the *distribution* of talk within the pairings.
Consideration of this returns us to the issue of turn-taking as
discussed for Anagram above.
It might be thought that the structure of this task mitigated against
a turn-taking pattern: if a partner adopting such a strategy gets a
wrong estimate on their turn, then the scoring is terminated and the
activity is restarted. This is to the disadvantage of both partners.
Accordingly, these pupils did not organise their "goes" on this basis.
Instead most of them exchanged control of the mouse (and most of the
estimating) as soon as an error had been made (and the score,
therefore, reset). Under such a system, it is even in the interests
of turn-taking partners *not* to be helpful: for they may want their
partners to make and error and, thereby, effect the changeover. In
reality, no collaboration was as ruthless as this analysis might
imply. Perhaps the passive partner realised that the activity was
bound, eventually, to go beyond the ability of the currently active
member: goes rarely lasted longer than 5 minutes. In any case seven
out of the ten Group 1 pairs showed concern with respecting turns; as
did nine out of the ten in Group 2.
However, the atmosphere among Group 2 pairs was more in keeping with
what we might expect from "collaborators". While there was this
pervasive concern to distribute responsibility for controlling the
computer, these children's discussion was far less suggestive of a
rigid turn-taking regime. This difference between the groups was
evaluated by considering more closely the final two runs of estimates
made by each pair in each Group. Only comments directed towards
estimating a current target were considered (teasing "I know it" kinds
of comments were omitted). For each run, a partner's contribution can
be expressed as a percentage of this total talk; then the mean of the
higher figure taken from each of the two runs is an indication of the
overall asymmetry of talking within a typical run of estimates. This
was reliably lower for Group 2: indicating a more equable distribution
of constructive talk in the condition where a workpad was made
available.
To the casual observer, this difference is visible in the form of a
more sustained engagement by Group 2 members with *all* the successive
problems they witness. In terms of the concepts of mutual
understanding being developed in this chapter, this difference also
may be expressed by claiming that Group 2 children had possession of a
more potent resource for developing and exploiting shared reference.
Talk within this Group was typically of the form:
6A: Wait 'till it splits up
6B: Is it 12?
6A: Not quite sure
6B: OK try it in the box (Means workpad)
6A: 3 fours and a two // 3 fours and a two
6B: It's 14 / I think
6A: Put another four
6B: Hope this is right
6A: Josie / just do 3 fours and a two
6B: I don't like this (Nervous of making
the estimate)
6A: We did better than this last time anyway
The pair are using 'the box' to replicate one configuration of the
number 14. Superficially, the impression is that they are working at
it "together" - certainly, the third person plural references were
more typical of Group 2 pairings. In this case, the partners refer to
the fact that they 'did better [score] than this last time [sequence
of problems]': so they characterise the overall enterprise in a manner
of joint responsibility, even though there remains a concern to
alternate control of the mouse.
We may say that the workpad has furnished an external point for the
fusing of partners' attention and their action. It is evidently a
more compelling focal point than, say, the display presented by the
program Anagram. This is because the pairs of children are able to
manipulate the workpad and discuss interpretations of what happens
when they do so. The device certainly serves to support shared
reference by effectively coordinating the moment-to-moment actions of
responding to the estimating problems. However, it may also support
the building of a more subtle form of shared understanding: a
mutuality that develops as each particular pair accumulates their own
experience in using it. Consider again the fragment of discourse
reproduced above. Once transcribed, such talk generally reads as
rather sparse. Recurring comments of the form '3 fours and a two'
only make real sense when encountered as situated in relation to the
screen display. Yet, an interactive workpad may be particularly
effective in making concrete and communicable just this kind of sparse
representational talk about numbers. The collaborators have access
to a (shared) device for supporting talk about structural features
of a number. We may suppose that young children's problems in
collaborating around the abstractions of mathematics are partly to do
with them (normally) *not* having a concrete and shared resource for
instantiating the abstract (Turkle and Papert, 1991; Wilenski, 1991).
The workpad here serves to support collaboration by offering a
concrete, shared reference for the "manipulation" of numerical
abstractions.
CONCLUDING COMMENTS
In introducing this discussion of collaborations *at* computers, I
expressed doubts about the strategy that researchers have favoured for
analysing such interactions. Analyses based only on coding,
categorising and counting utterances fail to do justice to
collaborative encounters as states of social engagement. My
particular concern in this chapter has been to illustrate an approach
to such encounters that would respect this dimension of engagement.
In doing so, I have identified structures of shared reference as
central to what participants may strive to achieve within joint
problem solving.
The nature of this mutual knowledge may be very different for
different kinds of task. Sometimes it may be entirely represented by
an external stimulus array (Anagram). In such a case, attention must
be distributed across it in a coordinated fashion if there is to be
productive collaboration. At other times, it may be expressed in an
abstract structure that is largely created within conversation (the
narrative of story-writers). We may suppose that all such objects of
joint attention will be potent insofar as they readily afford
manipulation and exploration by the collaborators. Narrative formats
work for this purpose - up to a point - but other abstract joint
knowledge may be more volatile. Thus, the trick to successfully
supporting much collaborative work in classrooms may involve
confronting pupils with abstract material within concrete and
manipulable representational formats. Indeed, studies of group work
practices in classrooms already hint that this is significant. For
example, Bennett (1991) reviews studies that grapple with the
correlates of effective group work. He finds that young pupils are
selective about when they are reticent. They do not collaborate
easily around abstract material such as that encountered in maths
work; yet they are quite forthcoming in situations that offer more
exploratory possibilities: 'It seems as if, given the opportunity to
talk about action, the children will take it' (1991, p.591).
Often computers may turn out to be a special resource for creating
such opportunities.
As was discussed in Chapter 3, the design of computers (their
localised input and output devices) demand a narrow focussing of
attention and action. Their interactivity also offers rich
possibilities for exploratory manipulation. In particular, the
powerful graphic capabilities of new technology can render abstract
material manipulable in concrete formats. Emihovich and Miller (1988)
have noted the advantage of such representations for integrating talk
and activity among pupils and *teachers*. Here, we are considering
that the same capabilities can be important in supporting shared
reference among pupils themselves - as they collaborate. My own view
is that computers can be especially effective in this arena. However,
this is not the same as simply noting that computers seem to engage or
animate pupils when they use them in their classrooms. The analyses
in this chapter highlight an underlying variety in this engagement;
the important challenge is to determine when and how the creation of
shared understanding is embedded in these lively interactions.
The studies reported here were conceived to explore the dynamics
of shared knowledge during collaborative computer work: how
collaborators invested in its creation, what form that creation took,
and how it could be exploited as a platform for reasoning. Thus, very
little of the research has conformed to procedures of
*experimentation*. Yet, there is room for experimental work in this
area and such research would be welcome. There is a much-cited study
by Malone (1981) in which a compulsive computer game is dissected to
determine which of its components serve to motivate users so
effectively. The concern of this study is evidently with the solitary
user. But, as reviewed earlier, much educational application of new
technology arranges more peer-based practices of working. So, it may
be useful to apply Malone's analytic approach to the study of
software features relevant to social patterns of use. Thus, we may
come to understand more about how structural details of educational
software support or constrain the possibility of collaborative work.
I am sure such research will have to dwell upon the special potential
of this technology for cultivating rich frameworks of shared
understanding.