"Conversational Patterns of Hearing Impaired children and their Hearing Peers: Using Digitile Software in an Integrated Setting"

advertisement
Conversational Patterns of Hearing Impaired Children and
their Hearing Peers: Using Digitile Software in an Integrated
Setting.
Candidate number: 27723
Word count: 6000 (approx.)
April 2011
University of Sussex, UK
27723
Acknowledgements
I would like to thank Nicola Yuill, my project supervisor, for her support and guidance
throughout this piece of research. I am grateful to Sophie Sharpe, who collaborated in
the collection of the data. Thank you to Victoria Bonnet for her technical advice and
reliability coding and to Rafaela Steinborn who also contributed to the reliability coding.
Thank you to Alison Marchbank (teacher of the deaf at the school) and her colleagues for
their advice on sample selection and task design and for facilitating the data collection. I
am also grateful for the professional advice of Jaqueline Bristow and Chris Browning.
Finally, many thanks to all the children who participated in the study.
School organizing, code adaptation, data analysis and reporting were conducted by the
researcher.
2
27723
Abstract
The present study demonstrates the differences in pragmatic language skills displayed
by children in dyads of mixed hearing ability compared to matched hearing ability. It
also explores how an application for collaborative learning on ipads can be best used to
maximise contingent and equitable conversation. 5 dyads consisting of 1 hearing
impaired child and 1 hearing child (mixed dyads) were compared to 5 dyads consisting
of 2 hearing children (matched dyads). All participants completed two collaborative
tasks on an ipad: one unstructured and one structured. Results revealed significant
differences between dyad types in the contingency of their conversation but not in the
equity of their verbal participation. Interaction effects indicted that, as predicted, more
structured tasks had a beneficial effect for mixed dyads but not matched dyads. Results
are discussed in the light of the growing literature on how single display groupware can
be used to mediate social interaction in educational settings.
3
27723
Introduction
Pragmatic language skills underpin the fundamental workings of our social discourse.
Where syntax and semantics describe the grammatical structure and lexical meaning of
words, pragmatics refers to the way language is used in its context. Since our use of
spoken language is inherently social, its contextual meaning is closely linked to the way
we interact and understand those around us. The academic study of conversation as a
shared set of communicative norms emerged largely from the work of Grice (1975), and
has lead to a consideration of language pragmatics as the “interface between cognitive,
social and linguistic development” (Adams, 2002, p.974). The domain encompasses a
wide range of verbal and non-verbal behaviours, from sensitivity to the subtle nuances
of conversation to an understanding of the simple rules of turn taking in a verbal
exchange. The present research focuses on the pragmatic skills that enable us to
maintain an equitable conversation and make smooth transfers between conversational
turns. This ability to co-construct a dialogue, consisting of two or more perspectives,
binds language and social understanding at a deep cognitive level.
Harris (1996) provides a theoretical framework in which to understand this
relationship. It is proposed that the development of language- and more specifically,
conversational skill- is a necessary precondition for developing what is known as
“theory of mind”. Sometimes referred to as “mind reading” ability, this refers to the
social understanding of others’ beliefs and desires. The relationship between language
and theory of mind development has since been well established (Astington and Baird,
2005). With specific regard to language pragmatics, Ensor and Hughes (2008) found that
both the content and the connectedness of conversations between mothers and their
preschool children were significantly and independently correlated with theory of mind
abilities 24 months later. “Connectedness” occurs when conversational turns are
semantically related and contingent. The implication is that contingent conversation
facilitates the sharing of mental states and the deeper social understanding that follows.
This link manifests itself in the development of social understanding of deaf children
from non-native signing families. There is robust evidence that these children (who have
missed out on early conversational experience) lag behind their hearing peers in theory
4
27723
of mind, displaying similar deficits in false belief understanding to those on the autistic
spectrum (Peterson and Siegal, 1995; Woolf, Want and Siegel, 2002). This applies to oral
deaf children who use cochlear implants and hearing aids (Peterson, 2004).
Furthermore, research indicates that this relationship is bidirectional. Silvestre,
Ramspott and Pareto (2007, p. 51) posit that, “knowledge of one's own mental states
and experiences and those of others constitutes one of the fundamental pillars for the
development of pragmatic abilities.” This makes intuitive sense, given what we know
about the centrality of perspective taking in pragmatic language skill, and bears out in
empirical research. In a study by Surian, Tedolsi and Siegal (2010), oral deaf children
and non-native signers displayed less sensitivity to conversational maxims than hearing
children but native signers (who had early access to conversation in sign language)
performed just as well as their hearing peers. Most, Shina-August and Meilijson (2010)
analysed conversations of hearing impaired (HI) and hearing (H) children with a
familiar adult. They found that contingency was one of several pragmatic skills that HI
children were more likely to use inappropriately than their hearing peers. They
attributed their results to difficulties in theory of mind and reduced exposure to varied
pragmatic situations for HI children. At a more functional level, Jeanes, Neinhuys and
Rickards (2000) found that throughout middle to late childhood, deaf students struggled
with making specific and appropriate requests for clarification.
However, the evidence for the differences in pragmatic skills of HI children compared to
their normally hearing peers is far from conclusive. Duncan (1999) examined the
conversational skills of HI and H children in a naturalistic integrated setting. No
significant differences were found between the conversational skills of children with
hearing loss and those with normal hearing, leading the researcher to conclude that
difficulties experienced by HI children were down to extraneous factors inherent in the
hearing loss rather than more fundamental difficulties in pragmatic skills. A key
difference in the methodology of Duncan (1999) compared to studies such as that of
Most et al. (2010), is that the conversations analysed were with hearing peers rather
than a familiar professional adult. Research of this nature is particularly important as it
reflects the behaviour of HI children in their natural environment and is high in
ecological validity. However, because the analysis was made comparatively (that is,
skills of HI children were measured in relation to their hearing conversation partners
5
27723
rather than to hearing children with a different conversation partner or in absolute
terms), it would be expected that pragmatic behaviours would appear the same because
the conversation partners would influence each other. Conversation is, by its very
nature, a collaborative activity and caution should be taken when drawing conclusions
about separate populations from such data.
Whilst it is true that children in a dyad of mixed hearing abilities can have positive
modelling influences on each other (Duncan, 1999; Anita and Kriemeyer, 2003),
obstacles to social communication between HI and H children are well documented.
Vandell and George (1981) found that dyads of two deaf children exhibited the same
social skills as two H children but mixed dyads displayed difficulties in pragmatic
behaviours such as appropriate initiation. More recently, Nunes, Pretzlik and Olsson
(2001) found that whilst deaf children in mainstream schools were not rejected by
peers, they were often isolated because of communication difficulties. These issues
concerning integration suggest that research and intervention must focus on the
functionality of communication between the dyad as a whole rather than the isolated
abilities of individuals. Indeed, Stinson and Liu (1999) found that the way H children
communicated with HI children contributed to the success or failure of the conversation.
Taking into consideration these issues as well as the pragmatic difficulties of the HI child
creates a more sensitive and applicable research programme. The present research
therefore focuses on the pragmatic behaviours of children at the level of the dyad rather
than the individual and seeks to identify patterns that are co-constructed by both
conversational partners.
The difficulties identified by the above research into the integration of HI with H peers
should not be taken as problems with integration per se. The inclusive approach
introduced by the 1981 Education Act brought with it the potential for a rich and diverse
range of communicative experiences for all children. However, as Webster & Ellwood
(1985, p. 243) stress, “Integration, whatever that may entail, is not an end in itself.”
There is consensus that familiarity can mediate communication difficulties in mixed
dyads (Anita and Kriemeyer, 2003) but active inclusion must be based around
meaningful verbal and non-verbal exchanges and structured by specific activities
(Stinson and Liu, 1999).
6
27723
This is where the use of multi-user technology, designed for collaborative learning tasks,
could be useful in creating these experiences. Single Display Groupware (SDG) is the
name given to devices such as an interactive touch-tables or ipads that two or more
users can work on collaboratively. This is opposed to individual devices with only one
input, such as a computer mouse. Piper and Hollan (2008) have used an application for
a tabletop display to mediate communication between deaf adults and their hearing
doctors. They found that the visual representation of the focus of conversation and
shared ownership of the device facilitated discussion effectively. No research has yet
explored the way that SDG could be used with deaf or HI children. However, there is a
growing wealth of research into the way technology can be used to support conversation
and collaboration in an educational context. This will inform the exploratory use of SDG
with HI children and their hearing peers.
The benefits of using SDG with special needs populations have been highlighted by the
SIDES Project (Piper et al., 2006). A puzzle game for four co-located players was
designed for use on an interactive touch-table and implemented in group therapy
sessions for young adolescents with Asperger’s syndrome. The game was found to be
beneficial for collaborative discussion and encourage perspective taking behaviours.
Although this is a very specific user-group, the aforementioned parallels between
perspective taking skills in HI children and those on the autistic spectrum mean that
these findings are pertinent for the present research. For example, the finding that
human rather than computer enforced rules elicited quantitatively fewer nonresponsive behaviours, can inform the way SDG games could be used to facilitate
successful conversation in HI children.
A key issue in the literature on shareable technologies is how to maximize equitable
verbal participation. Consistent with the Gricean maxim of quantity (Grice, 1975), equity
is considered to be important in successful conversational exchange. Alterations of the
interface of certain applications for SDG have been shown to increase equity of nonverbal participation but have been unsuccessful in changing verbal participation rates
(Fleck et al., 2009). Marshall et al. (2008) were particularly concerned with reducing the
dominance of certain users and found that the multi-touch rather than single-touch
7
27723
condition increased only the equality of touch input. Status effects (something that
would be an important consideration in an integrated school setting) persisted in
shaping verbal participation. It seems that technology-based manipulations may not be
the most fruitful line of investigation for verbal participation. The use of more personbased (rather than technology-based) variations may be a more fruitful line of
investigation. For example, “guided participation” (Rogoff, 2008) focuses on the social
context of the task. The structure that is offered by a culturally meaningful activity is
said to facilitate successful collaboration between peers.
The Present Research
This study seeks to identify conversational patterns between peers of varying hearing
levels and to explore how a maths game called Digitile (used on ipads as SDG) can be
best used to mediate this communication. Given the evidence for difficulties in
perspective taking in HI children, and the speculation that members of a dyad will be
influenced by each other’s behaviour, it is predicted that dyads of two hearing peers will
display more contingency and less pragmatic difficulties than dyads of mixed hearing
ability. It is further predicted that for the mixed dyads, pragmatic behaviours and equity
of participation will be more successful when working on a structured task than on an
unstructured task on SDG.
Method
Full ethics approval was granted from the Psychology Research Ethics Committee in
December 2010. CRB checks were obtained copied for the school’s record and an ethics
checklist outlining all details of the study was signed by the head before the study
commenced. (See appendix 1)
Participants
20 participants took part in the study, 5 of whom had hearing impairments (HI) and 15
of whom were normally hearing with no other sensory impairments (H). HI children will
be labelled as HI 1- HI 5 to maintain anonymity. There were 2 HI females, 3 HI males, 6 H
females and 9 H males. Participants were from years 2, 3 and 6. Ages ranged from 6 to
8
27723
12 (M= 8.95, SD= 1.82). All participants attended a state Primary school in Brighton. 10
participants took part in the mixed dyad condition (where dyads consisted of one HI
child and one hearing child) and 10 participants took part in the matched dyad condition
(where both children were hearing). All participants took part in the structured and
unstructured task conditions.
All HI children were well integrated with the rest of the school but a hearing unit
facilitated regular one-to-one teaching sessions for HI children with the school’s teacher
of the deaf. All were prelingually hearing impaired but oral in their communication
language. The degree of hearing loss varied significantly across participants (see
appendix 2 for profiles).
Pairing
Due to the applied nature of the research question, mixed pairs were matched according
to academic year rather than cognitive or verbal ability. This meant these pairs were of
mixed ability but approved by individual teachers. Matched pairs were also age matched
but they were also paired according to ability. Official SAT scores were not available but
teachers were asked to recommend pairs within approximately one SATs sub-level of
each other. The important consideration was that at least one of the hearing children in
the matched pair was of a similar ability to the hearing child in the mixed pair. This
ensured that there would be a similar average ability across mixed and matched pairs
and that the hearing pairs would not be of a significantly higher cognitive ability than
the ones with a HI child.
Children were all familiar with one another but each was matched with a child from a
different class (there were two classes per year). Care was taken not to pair hearing
impaired children with friends who were particularly used to assisting them and guiding
their communication. This controlled for the possibility that the pair had adapted to
each other’s style of communication over time and responded to a call in the literature
for “more research to be done with a wider range of conversation partners including less
familiar peers.” (Toe, Beattie and Barr, 2007, p.117).
9
27723
Parents of all the HI children were individually contacted by the school’s teacher of the
deaf and asked to complete a written consent form (see appendix 3). All but one of the
parents returned the consent form (83.3 % response rate). All hearing children in the
classes with HI children were given a written consent form. Those who returned them
(46.67% response rate), and met the criteria for pairing, took part. Participation was
voluntary on the day of testing.
Digitile software
An educational psychologist who specialises in deaf education was consulted about how
to use the multi touch technology appropriately with HI children. Issues concerning the
language impairments often experienced by HI children were brought to light,
prompting the researchers to use Digitile rather than one of the more language-based
applications such as WordCat (Kerawalla et al, 2008).
Digitile is a software programme adapted from a single-user desktop application into a
multi-touch application. (Rick and Rogers, 2008). It was originally developed for a
DiamondTouch tabletop but has recently been configured for use on ipads, which were
used in the present study. The interface is designed to support collaborative learning
about fractions by allowing two users to move virtual coloured shapes around the
screen to make patterns or complete challenges (See figure 1). The interface is designed
so both users are situated in front of the screen and have the same view of the screen.
Users drag and drop the shapes with their fingertips and can work simultaneously.
10
27723
4x4 grid to drag and drop
shapes on to.
Scrollable
bar where a
history of
patterns is
presented
Colour
palette
Figure 1: Image of Digitile interface (adapted from Rick and Rogers, 2008)
Design
‘Work snap’
tiles, where
shapes can be
rotated or
deleted.
The study was a 2 x 2 mixed design. Structured or unstructured task (within subjects)
and mixed or matched dyad (between subjects) comprised the independent variables.
The dependent variables were the frequency pragmatic behaviours and equity of verbal
participation displayed by each dyad.
Procedure
All participants took part in a familiarization session before they took part in the
experimental sessions. Each child had their own individual ipad and was given a print
out of a pattern to copy onto their grid (see appendix 4). This was informal and not
filmed. The familiarization controlled for any practice effects on the repeated measures
11
27723
conditions and gave children a chance to familiarize themselves at their own individual
pace.
The experimental sessions began running immediately after familiarization sessions.
Children were given 10 minutes, measured by a sand timer, to complete each task. The
structured task always followed the unstructured task. The experiment took place in a
naturalistic setting, however to control for noise interference the experiment took place
in a quiet room in the hearing unit in the school.
Tasks
The unstructured task was given with minimal instruction (see appendix 5). Children
were told to use Digitile to design a “good-looking” tile together.
This was the
familiarization task used by Rick, Rogers, Haig and Yuill (2009). Children worked on the
task until the sand timer ran out and the researcher praised the design to signal the end
of the task.
The structured task comprised of specific challenges to create patterns containing
certain amounts of each colour. It was introduced with a standardized instruction to
work together and to talk to each other as much as possible and agree on how best to
complete the challenge (see appendix 6). This aimed to create the conditions for ‘guided
participation’ (Rogoff, 2008). Tasks of increasing difficulty were presented visually on
an A4 instruction sheet and the experimenter gave the same information orally. Each
task was given after the preceding one had been completed to ensure that all abilities
had a chance to be challenged.
The structured tasks were designed to elicit conversation but to fit the needs and
abilities of the participants. Advice was taken from teachers of the relevant classes on
what the highest and lowest ability students would be able to understand and tasks
were ordered to get progressively harder and the highest ability children would reach
the hardest tasks within the 10 minutes. The tasks were adapted from Haig (2009). See
appendix 7.
12
27723
At the end of the session, children were thanked, debriefed (see appendix 8) and a
picture was taken of their first tile for them to keep.
Coding analysis
All of the sessions were recorded using a video camera and a microphone and coded
using Mangold Interact observational research software. The coding scheme was
derived from Toe, Beattie and Barr (2007). The scheme was altered in order to capture
specific elements of conversation relevant to Digitile. For example, “initiation turns” was
excluded because conversation initiation depended on the task rather than the
pragmatic skills of the children. No adequate definition for “smooth transfer between
turns” was given by Toe, Beattie and Barr (2007) so the criteria used by Ensor and
Hughes (2008) to measure connectedness of conversation was adopted to define a
contingent turn transfer: “The speaker’s utterance is semantically related to the other
interlocutor’s previous turn.” (see appendix 9 for full coding scheme).
Inter-rater reliability on 10% of the data produced a Kappa value of k= .75 (see appendix
10 for kappa table). This is described by Fleiss (1981, cited in Watkins and Pacheco,
2000) as good inter-rater agreement.
Results
Due to the relatively open-ended nature of the unstructured task, and the variation in
how long it took the pairs to complete all the sections of the structured task, not all pairs
provided the full 10 minutes of task time for each task. In order to standardize the data,
all analysis was carried out on the first 8 minutes of each task time.
Frequency of pragmatic behaviours:
A mixed design ANOVA was used with dyad type as the between subjects factor and task
type and pragmatic behaviour as the within subjects factors.
Initially, the assumption of normality was violated for “requests for clarification” in both
tasks, D (5) = .47 , p < .01 , D (5) = .47, p < .01. However, it was noted that ‘Requests for
13
27723
clarification’ was the only within subjects factor with a Z-score for kurtosis greater than
2.58 (the criterion for accepting a deviation from normality suggested by Field, 2005).
Considering the erratic and rare frequency of this behaviour occurring (mean frequency
of occurrence in structured task = .40, SD= .70; mean frequency of occurrence in
unstructured task= .90, SD= .99), it was considered beneficial to eliminate this from the
analysis and retain only 4 levels of pragmatic behaviour. After elimination, the
assumption of normality was met.
The assumption of homogeneity of variance was met for all pragmatic behaviours except
for “non-contingent transfers in unstructured task”, F (1, 8) = 16.64, p< .01. Outliers
were not omitted due to the small size of the sample; and after transformation, the
homogeneity of variance of the data was not improved. Notwithstanding this violation,
the mixed ANOVA was carried out on the original data but results regarding particular
level of pragmatic behaviour should be treated with caution.
Between subjects’ effects:
Mean frequency of behaviour
90
80
70
60
Dyad type mixed
50
Dyad type matched
40
30
20
10
0
contingent
transfers
non-contingent
transfer
non-verbal
response
no response
Pragmatic behaviour
Figure 2: Overall frequency of pragmatic behaviours for mixed and matched dyads
over both tasks.
14
27723
Levene’s test of equality of error variances was significant for all pragmatic behaviours,
indicating that the assumption of homogeneity of variance had been met.
Consistent with the hypothesis that matched dyads would display different
conversational patterns to mixed dyads, there was a significant main effect of dyad type,
F (1,8) = 8.80, p < .05, r = .72 Overall, matched dyads displayed more contingent transfers
(M = 74.60, SD= 25.88) than mixed dyads (M = 26.30, SD= 18.36). They also displayed
marginally more non-contingent responses (M= 7.4, SD = 8.26) and non-verbal
responses (M = 16.6, SD = 2.88) than mixed dyads (M= 5.00, SD = 2.34; M= 12.4, SD =
8.87). The only pragmatic behaviour that mixed dyads displayed more of was no
response (M= 8.2, SD = 4.77) compared to matched dyads (M= 4.2, SD = 2.77). These
observations support the hypothesis. In order to confirm this, the test was re-run with
contingent transfers as the only level of pragmatic behaviour. The effect remained
significant (F (1,8) = 11.63, p<. 01, r = .77), indicating that mixed dyads did display
significantly more contingency than matched dyads overall. Interestingly, there was a
significant interaction between task type and dyad type when measuring just contingent
transfers, (F (1,8) = 10.83 p< .01, r = .75). Figures 3 and 4 illustrate this interaction:
whilst matched dyads made more contingent transfers on the unstructured task than the
structured task, mixed dyads made more contingent transfers on the structured task
than the unstructured task. This further supports the hypothesis, as it suggests that only
the mixed dyads (who have a lower level of contingency than the matched ones) benefit
from the structured task.
15
27723
Within subjects’ effects:
Mean frequency of behaviour
60
50
40
Task type
unstructured
30
Task type
structured
20
10
0
contingent
transfer
non-contingent
transfer
non-verbal
response
no response
Pragmatic behaviour
Figure 3: Frequency of pragmatic behaviours for mixed dyads in structured and
unstructured tasks.
Mean frequency of behaviour
60
50
40
Task type
unstructured
30
Task type
structured
20
10
0
contingent
transfer
non-contingent
transfer
non-verbal
response
no response
Pragmatic behaviour
Figure 4: Frequency of pragmatic behaviours for matched dyads in structured and
unstructured tasks.
16
27723
Sphericity was violated for pragmatic behaviour, χ² (5)= .07, p < .001. GreenhouseGeisser estimates of sphericity were therefore used to correct degrees of freedom (ε =
.41). The remaining repeated measures effects met the assumption.
There was a significant main effect of pragmatic behaviour, F (1.22, 9.73) = 32.82, p <
.001, indicating that there was significant variation between the different conversational
turn transfers.
There were also significant interactions between pragmatic behaviour (contingent
transfers compared to all other behaviours) and dyad type (mixed or matched), F (3,24)
= 10.65, p < .001. Matched dyads made significantly more contingent than noncontingent transfers but mixed dyads did not, F (1,8) = 12.94, p < .01, r = .78 This is
consistent with the hypothesis that conversation would run more smoothly between
matched dyads. Matched dyads made more contingent transfers relative to noncontingent transfers than mixed dyads. Similarly, matched dyads displayed significantly
more non-verbal responses than no responses whereas mixed dyads displayed almost as
many no responses as non-verbal responses, F (1,8) = 6.11, p < .05, , r = .65.
The main effect of task type was non-significant, F (1, 8) = 1.69, p = .23, r = .41. However,
the effect was qualified with a significant interaction between task type and dyad type, F
(1,8) = 8.92, p < .01, , r = .72. Whilst matched dyads displayed fewer pragmatic
behaviours on the structured than on the unstructured task, mixed dyads displayed
more pragmatic behaviours on the structured than the unstructured task. The fact that
this was significant for contingent transfers supports the hypothesis (see post hoc test
on between subjects effects).
There was a significant three way interaction between task type, pragmatic behaviour
and dyad type, F (3,24) = 8.64, p < .001. Again, Contrasts were made between contingent
transfers versus non-contingent transfers and no response versus non-verbal response,
but this time dyad type and task type were taken into consideration:
For the mixed dyads, there was a difference between contingent and non-contingent
transfers for the mixed dyads but only in the structured task. However, for matched
17
27723
dyads there was a significant difference between non-contingent and contingent
transfers, regardless of task type. F (1, 8)= 9.35, p < .05.
The graphs show a similar pattern of results for non-verbal responses versus no
responses, taking into account dyad type and task type. However, this was nonsignificant, F (1, 8)=3.66, p = .09, r = .56.
Durations of talk
Durations of utterances made by the experimenter or directed towards the
experimenter and durations of utterances undecipherable to the coder were measured
in order to identify them as confounding variables to differences between tasks if
necessary. Also, durations of each child’s utterance were measured in order to gauge
equity of verbal participation. This was done by logging the start and end time of each
utterance using Mangold Interact software.
Figure 4: Mean duration of experimenter talk and undecipherable talk for dyads
over 16 minutes of task time.
It is apparent from these results that whilst there was very little undecipherable talk in
either task, there was considerably more experimenter talk in the structured (M= 86.04,
18
27723
SD= 16.95) than the unstructured tasks (M= 28.83, SD= 15.27). This was not included in
the main analysis of duration data, as it did not directly affect equity of participation.
Condition
Mean Gini Coefficient
SD
Mixed
.22
.09
Matched
.15
.04
Unstructured
.21
.12
Structured
.20
.08
Note: the closer to 0, the more equity of verbal participation.
Table 1: Average equity of verbal participation from collapsed results of each
condition.
Gini coefficients were calculated to measure the equity of verbal participation for each
pair:
N
G=¾∑ │XI–½
i=1
(Adapted from Harris et al., 2009, who used the equation to calculate equity between 3
participants)
This was calculated for each pair separately, using the raw scores of seconds each child
talked over each condition. The expected proportion of talk time (½) was subtracted
from the absolute proportion and a normed gini was calculated from this figure. This
was the measure of conversational balance. A coefficient of 0 indicates perfect equity
and a coefficient of 1 indicates perfect inequity. A mixed ANOVA was conducted on the
gini coefficients. Assumptions of normality and homogeneity of variance were met.
There was no significant difference in equity of participation between task types, F (1, 8)
= .36, p = .57, r = .17 There was also no significant difference in equity of participation
between dyad types, F (1, 8) = 4.91, p = .06, r = .61. This refutes the hypothesis that
using a more structured task would increase equity of participation.
19
27723
Taken together, the results suggest that whilst the nature of conversational patterns
depends on the experimental condition, the amount of talk from individuals does not.
This visual representation of duration of talk from each child in a mixed dyad illustrates
this point well. The timelines represent the durations in seconds of child 1 and child 2’s
utterances over a sample of 6 minutes. Durations of talk appear as the shaded blocks.
The first timeline represents the unstructured task and the second represents the
structured task. There is little difference in equity in verbal participation between the
two tasks. However, the pattern of results does differ: in the unstructured task, the
periods of talk for both children are randomly dispersed over the timeline and do not
appear to relate to one another coherently. Conversely, in the structured task, the
periods of talk appear in blocks and utterances from child 1 and child 2 show up
alternately. Despite the fact there is often more talk from one child than the other in a
block of alternating talk times, there is almost always a response from the other and
turn taking is consistent throughout.
Discussion
In terms of the nature of conversations across conditions, the hypotheses were
supported. Most notably, a three-way interaction revealed that as predicted, matched
dyads made significantly more contingent than non-contingent turn transfers regardless
of task type. Mixed dyads on the other hand, only displayed more contingent transfers
than non-contingent transfers when they were engaged in the structured task. In terms
of equity of participation, the hypotheses were not supported: there were no significant
differences in equity of participation across either condition.
20
27723
Given that the analysis of conversations measured the behaviour of different dyads
rather than different individuals, it is essential to remain clear about what the present
research does and does not tell us about deaf and hearing populations. It does provide a
valuable insight into the way H and HI children interact with each other with different
conversation partners and the results are consistent with research that suggests HI
children struggle with certain pragmatic language skills (such as conversational
contingency) more than their hearing peers (Most, Shina-August and Meilijson, 2010).
However, as Stinson and Liu (1999) found, communication success between children of
mixed hearing ability depends on more than just the abilities of the HI partner. It is
likely that the H child will respond to the HI child differently than they would another
hearing child and that both children will be influenced by each other’s communicative
strengths and weaknesses. This means it does not tell us anything conclusive about the
pragmatic skills of the HI children at an individual level (something that Duncan (1999)
fails to acknowledge). The differences observed may have been down to poorer
perspective taking skills of the HI children (Woolfe, Want and Siegal, 2002) but may also
have been because of the difficulties that come with integrating different hearing
abilities (Vandell and George, 1981). The size of the sample prevented a further
comparison group of HI children matched with other HI children, however this would
provide insight into the underlying causes of the behaviours.
Although effect sizes were generally large (above .50), the sample size did just fall short
of that recommended by Cohen (1992) to detect such effect sizes (n= 28 to detect sizes
of r = .5) Nevertheless, the small scale of the study facilitated the highly detailed,
bottom-up approach needed to reveal the subtle patterns that emerge during this kind
of interaction. Fleck et al. (2009) emphasised the importance of this after finding that
manipulations in the way SDG is used did not seem to affect overall equity of
participation but did affect what children said and the way that they collaborated. This
was reflected in the present results, where equity of participation was relatively good
throughout despite the fact that it was predicted that a structured task would improve
this for the mixed dyads. DiMocco, Pandolfo and Bender (2004) have used technology to
aid communication but have focused their efforts heavily on trying to encourage equal
verbal participation with histogram displays of individual contributions. This was only
partially successful and the present research suggests that this may not be the most
21
27723
insightful measure of communication success. Obstacles to verbal collaboration are
apparently more likely to manifest themselves in a lack of contingency or nonresponsiveness. These are aspects of communication that can be aided by sensitive use
of technology.
The fine level of analysis also justifies the relatively short task times. The labourintensive conversational coding of each turn transfer and utterance meant that in fact, a
large amount of information was gleaned from each session. However, there was one
pragmatic behaviour- requests for clarification- that may have been underrepresented
because of the short amount of time. This was deleted from the analysis because of its
rare occurrence but it was possible that there simply wasn’t enough time to allow for
possible breakdowns in conversation that would require such a request. This would
explain the discrepancy with Jeanes, Neinhuys and Rickards (2000), who found that deaf
children had particular difficulty with requesting and responding to misunderstandings
in conversation. Equally however, it is possible that the joint focus of attention offered
by the tasks on Digitile (in comparison to Jeanes et al.’s pen and paper task) meant that
there was less chance of conversational breakdown and therefore less need to request
clarification. There was certainly anecdotal evidence of the interactive visual display
offering a tool for joint understanding. Often, children would exclaim, “I see what you’re
doing” as they watched their partner try a possible solution to the task or hover their
finger above a particular shape. Indeed, Rick and Rogers (2008) intended for this when
they designed the interface for Digitile on SDG, emphasising the reflective nature of the
application (being interactive and giving instant visual feedback to both users). Longer
time periods of conversation would need to be analysed and compared to other
conversational
contexts
to
establish
whether
the
technology
did
mediate
communication in this way.
In terms of non-verbal responses in conversation, mixed dyads tended to fail to respond
to each other almost as much as they made physical responses. Conversely, matched
dyads were more likely to make a physical response than not respond at all, supporting
the hypothesis that they would have less pragmatic difficulties. A possible confounding
factor to this would be that failures to respond in the mixed dyads would simply be
down to the fact that the HI child did not hear rather than a deeper lack of pragmatic
22
27723
understanding. However, the use of the hearing unit in the school, where there was no
background noise controlled for this to some extent. More importantly, the data
screening of this behaviour showed a normal distribution of frequency, indicating that
the hearing pairs also displayed this behaviour. This suggests non-responsiveness was
not merely down to surface hearing difficulties. Nevertheless, this should be taken into
consideration when interpreting this particular finding.
Although there were clear themes that emerged from the conversational patterns of the
ten dyads that took part, it is essential to be mindful of the differences between dyads in
each group. Rick, Marshall and Yuill (2011) take an idiographic approach to analysing
how pairs of children collaborate using Digitile, demonstrating that even typically
developing dyads have very variable interaction patterns. This is even more crucial to
the present study, considering the range of different hearing and language abilities
across the HI children (see appendix 1). Interestingly, one of the HI children (HI 3) with
the most obvious developmental delays in language and social development elicited
some of the most sensitive and cooperative verbal interaction from his hearing partner.
He displayed good maintenance skills and kept up the conversation so that there were
fewer failures to respond than if HI 3 had been paired with a less patient partner.
However, the difficulty that this pair had in sharing ideas (HI 3 seemed unaware of his
partner’s input, as he kept exclaiming, “I done it!”) was consistent with the theory that
impaired early conversation impacts on later social understanding (Harris, 1996). This
pattern differed radically to HI 2’s dyad, which displayed very little difficulty in making
joint decisions and made the most contingent transfers out of all the mixed dyads on
both tasks. This also has interesting theoretical relevance, as HI 2 was the only child
whose parents were deaf and fluent signers, making her a ‘native signer.’ She was also
fluent in oral English, meaning that she would not have missed out on any early
conversational experience and so one would expect her social understanding (and
pragmatic language skills) to be at a high level (Woolf, Want and Siegel, 2002).
The interaction effect between task type and dyad type on contingent transfers indicates
the importance of varying the way SDG technology is used to fit the needs of the user
group. Whilst mixed dyads made more contingent conversation on the structured task
than the unstructured task, matched dyads made far more contingent conversation on
23
27723
the unstructured task than on the structured. This pattern seemed to reflect different
concentration levels across the tasks; matched dyads tended to be more engaged on the
unstructured task and lost concentration by the structured one, whereas mixed dyads
were not fully focused until they had been given the common ground of a structured
task. This supports the hypothesis concerning task type and is consistent with Rogoff’s
principle of guided participation. She stresses that guided participation is not “an
operational definition that one might use to identify some and not other interactions or
arrangements,” (Rogoff, 2008, p.60). Rather, it depends on whether the activity is
meaningful to those involved. Matched dyads seemed to be able to share an
understanding of more abstract ideas (one pair talked about creating a “wildlife” out of
the shapes in the unstructured task) but the mixed dyads did not tend find the same
shared meaning without more structure. One of the conclusions of the SIDES project
(Piper et al. 2006) was that games on SDG to promote social interaction worked best
when there was an adult moderator. Because there was much more researcher input in
the structured task, the present results suggest that this conclusion is true for user
groups with special needs (such as those on the SIDES project or in the mixed dyad
condition of this study). However, for other user groups, it may be more beneficial to let
them be more exploratory and independent.
The naturalistic approach of the present research meant that it yielded results with high
practical applicability to integrated educational settings. It demonstrates the value of
identifying differences in the ways children interact with each other and of
acknowledging the impact of hearing impairment on these conversations. This can
inform the way that SDG can be used sensitively to encourage pragmatic language skills
and social cognitive development.
24
27723
References
Adams, C. (2002) Practitioner review: The assessment of language pragmatics. Journal of
Child Psychology and Psychiatry, 43, 973-987.
Anita, S. D. & Kriemeyer, K. H. (2003) Peer interactions of deaf and hard of hearing
children. In M. Marschark & P. E. Spencer (Eds.), Oxford handbook of Deaf Studies,
Language and Education (pp. 164-177). New York: Oxford University Press.
Astington, J. W. & Baird, J. A. (Eds) (2005). Why Language Matters for Theory of Mind.
New York: Oxford University Press.
Cohen, J. (1992) A power primer. Psychological Bulletin, 112, 115-159.
DiMicco, J. M., Pandolfo, A. & Bender, W. (2004) Influencing group participation with a
shared display, In Proceedings of the ACM Conference on Computer Supported Cooperative
Work CSCW 2004 Chicago.
Duncan, J. (1999) Conversational skills of children with hearing loss and children with
normal hearing in an integrated setting. Volta Review, 4, 193-211.
Ensor, R. & Hughes, C. (2008) Content or Connectedness? Mother-child talk and early
social understanding. Child Development, 79, 201-216.
Field, A. P. (2005). Discovering statistics using SPSS: and sex and drugs and rock 'n' roll
(2nd edition). London: Sage.
Fleck, R., Rogers, Y., Yuill, N., Marshall, P., Carr, A., Rick, J. and Bonnett, V. (2009) Actions
speak loudly with words: unpacking collaboration around the tabletop. In Proceedings of
Interactive Tabletops and Surfaces ITS 2009, 189-196.
Garrison, W., Long, G. & Stinson, M. (1994) The classroom communication ease scale:
The development of a self-report questionnaire for mainstreamed deaf students.
American Annals of the Deaf, 139, 132-140.
25
27723
Grice, H. P. (1975) Logic and Conversation. In P. Cole and J. L. Morgan (eds) Syntax and
Semantics: Vol. 3, Speech acts: 41-58, New York: Academic Press.
Harris, P. (1996) Desires, belief and language. In P. Carruthers and P.K. Smith (Eds.),
Theories of Theories of Mind. Cambridge: Cambridge University Press.
Harris, P. L., deRonsay & Pons, F. (2005) Language and Children’s understanding of
mental states. Current Directions in Psychological Science, 14, 69-73.
Harris, A., Rick, J., Bonnett, V., Yuill, N., Fleck, R., Marshall, P. and Rogers, Y. (2009).
Around the table: Are multiple-touch surfaces better than single-touch for children's
collaborative interactions? In Proceedings of CSCL 2009, 335–344.
Haig, C. (2009) Interactions with Digitile: An investigation into the collaborative
processes associated with successful learning via a novel interface resource. Unpublished
MSc Thesis.
Jeanes, Neinhuys and Rickards (2000) The pragmatic skills of profoundly deaf children.
Journal of Deaf Studies and Deaf Education, 5, 237-247.
Kerawalla, L., Pearce, D., Yuill, N., Luckin, R., & Harris, A. (2008) “I’m keeping those there,
are you?” The role of a new user interface paradigm – Separate Control of Shared Space
(SCOSS) – in the collaborative decision-making process. Computers and Education, 50,
193-206.
Marshall, P. Hornecker, E., Morris, R., Dalton, N. & Rogers, Y. (2008) When fingers do the
talking: A study of group participation with varying constraints to a tabletop interface.
Proceedings of IEEE International Workshop on Horizontal Human-Computer System, 3744.
Most, Shina-August and Meilijson (2010) Pragmatic Abilities of Children With Hearing Loss
Using Cochlear Implants or Hearing Aids Compared to Hearing Children. Journal of Deaf
Studies and Deaf Education, 15, 422-437.
26
27723
Nunes, T., Pretzlik, U. & Olsson, J. (2001) Deaf children's social relationships in
mainstream schools. Journal of Deaf Education International, 3, 123-136.
Peterson, C.C. & Siegal, M. (1995) Deafness, conversation and theory of mind. Journal of
Child Psychology and Psychiatry and Allied Disciplines, 36, 459-474.
Peterson, C. C. (2004) Theory of mind development of oral deaf children with cochlear
implants or conventional hearing aids. Journal of Child Psychology and Psychiatry, 45,
1096-1106.
Piper, A. M. & Hollan, J. D. (2008) Supporting medical conversations between deaf and
hearing individuals with tabletop displays. Proceedings of Computer-Supported
Cooperative Work (CSCW) 2008, 147-156.
Piper, A. M., O’Brian, E., Morris, M. R. & Winograd, T. (2006) SIDES: a cooperative
tabletop computer game for social skills development. In Proc. CSCW 2006, ACM, 1-10.
Rick, J., Marshall, P., and Yuill, N. (2011). Beyond one-size-fits-all: How interactive
tabletops support collaborative learning. In Proceedings of IDC 2011, to appear. ACM Press.
Rick, J. and Rogers, Y. (2008) From DigiQuilt to DigiTile: Adapting educational
technology to a multi-touch table. In Proceedings of TABLETOP '08, 79-86.
Rick, J., Rogers, Y., Haig, C., & Yuill, N. (2009). Learning by doing with shareable
interfaces. Children, Youth and Environments, 19, 321–342.
Rogoff, B. (2008) Pedagogy and Practice: Culture and Identities Observing sociocultural
activity on three planes: Participatory appropriation, guided participation, and
apprenticeship. In J. Goodnow, P. Miller, & F. Kessel (Eds.), Cultural practices as contexts
for development (pp. 45-65). San Francisco: Jossey-Bass.
Silvestre, N., Ramspott, A. & Pareto, I. (2007) Conversational Skills in a semi-structured
interview and self-concept in deaf students. Journal of Deaf Studies and Deaf Education,
12, 38-54.
27
27723
Stinson, M. & Liu, Y. (1999) Participation of deaf and hard-of-hearing students in classes
with hearing students. Journal of Deaf Studies and Deaf Education, 4, 191-202.
Surian, L., Tedolsi, M. & Siegal, M. (2010) Sensitivity to conversational maxims in deaf
and hearing children. Journal of Child Language, 37, 929-943.
Toe, D., Beattie, R. & Barr, M. (2007) The development of pragmatic skills in children
who are severely and profoundly deaf. Deafness and Education International, 9, 101-117.
Vandel, D. L. & George, D. L. (1981) Social interaction in hearing and deaf preschoolers:
Successes and failures in initiations. Child Development, 58, 627- 635.
Watkins, M. W., & Pacheco, M. (2000). Inter-observer agreement in behavioural
research: Importance and calculation. Journal of behavioural Education, 10, 205-212.
Webster, A. & Ellwood, J. (1985) The Hearing Impaired Child in the Ordinary School.
London: Croom Helm.
Woolfe, T., Want, S. & Siegal, M. (2002) Signposts to development: Theory of mind in
deaf children. Child Development, 73, 768-778.
28
27723
Appendices
1. Ethics documents:
Life Sciences & Psychology Cluster based Research Ethics Committee
CERTIFICATE OF APPROVAL
Reference Number:
Title of Project:
Principal Investigator:
Student:
NY1110
Shareable interfaces for learning
N Yuill
Sophie Sharpe, Natassia Ford-Brenman, Abigail Higgleton, Lauren
McKay
Collaborators:
Duration of Approval
(not greater than 4 years)
9 months
Expected Start Date:*
December 2010
This project has been given ethical approval by the Life Sciences and Psychology
Cluster based Research Ethics Committee (C-REC).
*NB. If the actual project start date is delayed beyond 12 months of the expected start date, this
Certificate of Approval will lapse and the project will need to be reviewed again to take account of
changed circumstances such as legislation, sponsor requirements and University procedures.
Please note and follow the requirements for approved submissions:
Amendments to protocol.
 Any changes or amendments to approved protocols must be submitted to the CREC for authorisation prior to implementation.
Feedback regarding the status and conduct of approved projects
 Any incidents with ethical implications that occur during the implementation of the
project must be reported immediately to the Chair of the C-REC.
The principal investigator is required to provide a brief annual written statement to the committee,
indicating the status and conduct of the approved project. These reports will be reviewed at the
annual meeting of the committee. A statement by the Principal Investigator to the C-REC
indicating the status and conduct of the approved project will be required on the following date(s):
December 2011……….……………………………………………………………………………….
Authorised Signature
Name of Authorised Signatory
(C-REC Chair or nominated deputy)
Date
Jennifer Rusted
Jennifer Rusted
10.12.10
29
27723
Psychology Department, University of Sussex
Ethical checklist for students working in schools
All students are required to abide by ethical guidelines provided by the British Psychological Society
when working in local schools. These cover in particular keeping data confidential, permitting
children to withdraw and gaining appropriate consent. All work with children also has to gain
approval from the Psychology Ethics committee. We would be grateful if you could check and sign
the following sheet to show that you approve of the procedures chosen.
Names of researchers visiting the school: Natassia Ford Brenman, Sophie Sharpe
Supervisor and contact number/email: Nicola Yuill 01273 678630 nicolay@sussex.ac.uk
Period of visits: Spring term
Year: 2011
Classes/year groups visited: 2,3 and 6
Brief description of procedure
Students in years 2,3 and 6
Initial familiarisation with Digitile game: 25 minute group session.
Approximately 20 minute sessions, playing a game on an interactive Ipad. This
involves completing two tasks making patterns with coloured shapes to learn about
fractions. Children move virtual pictures around a touch sensitive screen and work
collaboratively to complete the task.
Students will work in pairs with age-matched peers. Half of the pairs will consist of
one hearing child and one hearing impaired child and the other half of the pairs will
be two hearing children. Information on academic performance will be requested
and kept confidentially.
Parents will be fully informed of the details of the study and will need to consent to
the students’ participation. Children can withdraw at any point. Sessions will be
videotaped but children will not be individually identified in reports and their data
will be treated confidentially.
Consent procedure used:
Opt-in (parent/carer signature required)
Police check required: yes
Name of school Bevendean Primary
Head’s signature………………………………….
Date ……………………….
30
27723
2. Profiles for HI children (as provided by the school’s teacher of the deaf):

HI 1 (year 6) had profound hearing loss. He had a unilateral cochlear implant
fitted on the right side in 2005, wore a hearing aid in the left and used a FM
system. Word retrieval difficulty and specific learning difficulties affecting overall
cognitive function.

HI 2 (year 6) had moderate hearing impairment and wore bi-lateral hearing aids.
Both parents were deaf and BSL was her first language. However, she was fluent
in oral English and had no diagnosed language difficulties.

HI 3 (year 3) had profound hearing loss, had a unilateral cochlear implant fitted in
2003 on the left side and used an FM system. General developmental delays
causing co-ordination problems and oral and verbal dyspraxia.

HI 4 (year 3) had severe/profound hearing loss with no usable hearing on the left
but wore a unilateral hearing aid on the right. No diagnosed language
impairments but had English as an additional language.

HI 5 (year 3) had mild hearing loss and wore bilateral hearing aids. General
developmental delay and some visual impairment.
31
27723
3. Written consent form:
Does your child like learning with others?
The ShareIT project, based at the University of Sussex and the Open University, is
looking at new and exciting ways to use technology to help people work together.
Researchers from Sussex are visiting your child’s class in the next few weeks to try out
a new learning game called Digitile, which involves groups of children working together
around a special digital tabletop.
To do this, we need to videotape how the children use the technology. If you would like
your child to have the chance to take part, please complete and return the form below
to the class teacher in the next week. Children will work in pairs and move shapes
around a touch sensitive screen to complete a task about fractions and patterns.
We expect the children to enjoy it, but they only take part with your permission, and if
they want to on the day: they are free to withdraw at any point in the research. If you
would like more information, you can contact the researchers at the school or Nicola
Yuill on 01273 678630; or email nicolay@sussex.ac.uk. You can also see more about
Digitile on the website: http://www.shareitproject.org/8
Thank you
Please complete all parts and return to the class teacher as soon as possible:
I am happy for my child to take part in the group learning game being carried out by the
University of Sussex.
My child’s name is
My child’s class is
I am happy for my child to be video-taped during this study,
for use only by the research team
yes/no
I am happy that video footage or stills of my child carrying out the
above task may be used to describe the research to other
researchers at meetings, conferences and in published work.
yes/no
Signed
(parent/carer)
32
27723
4. Familiarization task:
Copy the pattern onto the
square on the table…
5.Instructions for unstructured task:
“I want you to work together to design a really good-looking tile using lots of colours and
shapes.”
33
27723
6. General instructions for structured tasks:
“Now I am going to give you some special challenges to complete together… This time it is
really important that you talk to each other about how to do the task.”
7. Specific structured tasks:
Make a pattern with the
same amounts of green
and blue...
34
27723
Make a pattern with
3 DIFFERENT colours
using the same
amounts...
35
27723
Make a pattern which is
half red, a quarter blue
and a quarter orange...
36
27723
Make the square
3/8 orange and
3/8 brown...
Some of the squares can
be left empty...
37
27723
7. Debrief:
“Thank you both very much, you have done really well. The last task was tricky so don’t
worry if you didn’t quite get there in the time. This is the end of the session but I’ll take a
picture of your patterns and next time I come in I will bring them in so we can stick them on
the wall.”
8. Coding scheme:

Equity of verbal participation- Duration of each utterance made by each child.
This is measured for each child separately. If the proportion of total duration is
equal for both children, they have displayed equity.
Duration of an utterance definition: Time between start and end of an utterance. An
utterance can be a non-word.

Contingent conversational turn transfers- Number of smooth transfers
between one speaker’s utterance and the other’s.
Contingent transfer definition: One speaker makes an utterance which gives the other
speaker the opportunity to talk and the other speaker replies within approximately 3
seconds with a semantically related utterance. The response can be a non-word but it must
be an appropriate response to a verbal cue.

Non-contingent conversational turn transfers- Number of transfers where the
response is inappropriate or unrelated to the previous utterance.
Non-contingent transfer definition: The second utterance must be within approximately 3
seconds of the verbal cue of the first speaker and semantically unrelated or inappropriate.
For non-words, it is classed as non-contingent if the cue required a fuller response or if the
utterance is inappropriate to the cue.
Eg: “What colour shall we use?” –“mmm”

Non-verbal response- Number of times a child responds to verbal cue with a
nonverbal action.
38
27723
E.g. “put that there” followed by the appropriate action. This should be used instead of ‘no
response’ when a non-verbal reaction is more appropriate. However, if there is a verbal cue
to speak that requires a verbal response and the child simply continues to work, this is not
a non-verbal response. This should be classed as ‘no response’ (see below).

No response– number of times an utterance from one partner is ignored by the
other child.
No response definition: when a verbal cue for the other to speak elicits no verbal or nonverbal response from the talk partner.
NB if there is a non verbal reaction to a verbal cue (e.g. “hmmm”/ “errrr”) this is not
classed as ignoring, it must be judged as a contingent or non-contingent response.

Requests for clarification- number of appropriate verbal responses to
misunderstanding/ mishearing (as opposed to no response)
Request for clarification definition: General verbal requests:
E.g. “huh?” or “what?” (can be non-words),
Or specific verbal requests for clarification or repetition:
E.g. “Did you say four days?”
This has been adapted from Toe et al because they coded requests for clarification as
‘conversational breakdowns’. It seemed that ‘no response’ denoted more of a
conversational breakdown and that requests for clarification in fact signalled an
appropriate was to avoid conversational breakdown.
39
27723
9. Kappa table:
Natass
ia
Victoria
duration
contingent
non
contingent
no response
duration
214
contingent
noncontingent
no response
rfc
experimenter
undecipher.
55
16
1
rfc
experimenter.
undecipher.
No code
non-verbal
response
EF =
4
1
7
8
1
221
141.8452722
63
12.63610315
17
0.779369628
1
0.008595989
0
0
4
0.045845272
Note: Non-verbal response was added at a later stage of coding so reliability coding was
carried out with a different coder.
40
1
0.00286533
Download