Académique Documents
Professionnel Documents
Culture Documents
Abstract
Narrative ability is one of the most interesting and ecologically valid ways in
which to measure communicative competence both in normal populations and
in clinical groups, since narratives form the basis of many childhood speech
acts. Narrative may also prove to be a good tool for distinguishing clinical
groups who show overlapping symptoms but who are thought to experience
subtly different impairments. This article gives an overview of some of the
theoretical reasons for using narrative to assess both linguistic and pragmatic
impairments. As part of a preliminary investigation examining possible
similarities and differences across groups, ve children with severe pragmatic
language impairments (PLI) and ve with more typical speci c language
impairments (SLI) completed short picture based narratives using the Bus
Story and the Frog Story. These illustrative data are included throughout the
paper to highlight features of use to clinicians, particularly with respect to
differences in the narratives of children with PLI compared to their peers with
SLI. Furthermore, when compared to Tager-Flusbergs (1995) data from
children with autism, SLI narratives seem to be more similar to those of the
group with autism than did PLI narratives. Narrative ability was found to
relate directly to pragmatic skill but in different ways according to clinical
subgroup. Implications for both theory and practice are discussed.
Introduction
Narrative ability is one of the most interesting and ecologically valid ways in
which to measure communicative competence both in normal populations and
in clinical groups. It is also one of the most abundant sources of information
Address for correspondence: Nicola Botting, Centre for the Study of Language Impairments, Human
Communication and Deafness, School of Education, University of Manchester, Oxford Road, M13 9PL,
UK. E-mail: nicola.botting@man.ac.uk
# Arnold 2002
10.1191=0265659002ct224oa
Downloaded from clt.sagepub.com at American University in Cairo on November 5, 2015
since narratives form the basis of many childhood speech acts. Spontaneous
speech samples have been used widely in speech and language therapy
practice, but completely unstructured samples are time consuming and are
often only used in descriptive ways by therapists. Narrative appears to be used
less often as a tool. One exception to this is The Bus Story (Renfrew, 1991)
narrative assessment. This story is regularly employed in primary language
units in the UK, and has strong relationships with future language and literacy
performance (Stothard et al., 1998) as well as genetic trends (Hohnen and
Stevenson, 1999). However, other forms of narrative may prove to be more
useful after the infant years, especially for children who have been regularly
tested using the Bus Story.
In addition to this, a group of children with pragmatic language impairments (PLI), who prove dif cult to assess using formal language tests (Botting
et al., 1998), seem to have increased in number within language units and
speechlanguage therapy services. These children are also dif cult to assess
using free conversational speech samples (e.g., Bishop and Adams, 1989; see
later discussion). Children with PLI are children often previously referred to in
the literature as having semantic-pragmatic disorder (Rapin and Allen, 1983;
Bishop and Rosenbloom, 1987; Bishop, 1989; Rapin, 1996). These children
have primary problems with the social use of language especially conversation
and some also have linguistic dif culties. However recent studies have
suggested these are not speci cally semantic in nature (Botting and ContiRamsden, 1999) and therefore this part of the terminology has been dropped
by some researchers (McTear and Conti-Ramsden, 1992; Bishop, 1998; ContiRamsden and Botting, 1999). It is still unclear, despite over a decade of
research in this area, whether these individuals are best described as having
language impairment, as part of the autistic spectrum, or neither (see Boucher,
1998; Botting, 1998;Brook and Bowler, 1992;Rapin and Allen, 1998;Gagnon et
al., 1997. Despite the belief of some that children with PLI are children with
an autistic spectrum disorder, a number of these individuals nd their way into
language units and by school year 2 (age 7), make up a signi cant proportion
(around 15%) of those receiving educational help for language impairments
(Conti-Ramsden and Botting, 1999; Conti-Ramsden et al., 1997; Botting and
Conti-Ramsden, 1999).
Narrative as a tool
2) Narrative and literacy skills. Oral narrative and literacy abilities are
closely interwoven skills. The latter develops later than the rst with written
narratives being estimated at around 60% the length of their oral counterparts
(Gillam and Johnston, 1992). Other recent studies also emphasize the
association between oral language and literacy skills and have pointed to
interesting mechanisms for the poor literacy levels often reported in children
with SLI (e.g., Conti-Ramsden et al., 2001). Kaderavek and Sulzby (2000) for
example, report that oral narratives using a familiar storybook as a prompt (a
task they describe as emergent story reading) given by typically developing
pre-school children were more complex and showed more of the devices used
in written language than the narratives given by peers with language impairment. Thus, a failure to gain early emergent reading skills may limit children
with language impairment. Cain and Oakhill (1996) found that reading
comprehension skill was also linked to story telling abilities in a group of
normally developing children with poor or strong comprehension skills. In
particular, those with poor comprehension skills also produced poor narratives
with less story structure. These authors conclude that poor story knowledge
(i.e., knowledge about how a complete story might look) is more likely a cause
than a result of poor reading comprehension. Oral narrative may also serve to
inform clinicians about the risk of literacy problems and give insights into the
ways in which therapy may aid development of this skill.
Narrative as a tool
for description). The groups were: ve children with the most common form
of SLI as found in language units (that is sentence level expressive and
receptive dif culties, see Conti-Ramsden et al., 1997) who had narrative data
available and Childrens Communication Checklist (CCC) scores above 132
(i.e., did not have pragmatic dif culties; see Bishop, 1998); and ve children
with relatively few expressive language dif culties but primary pragmatic
language impairments, all having CCC scores of 132 and below. These were
the only ve out of the 10 children from Botting and Conti-Ramsden (1999)
who were available to complete further narrative tasks. All childrens pragmatic impairment scores from the CCC were also used to explore the
relationship between narrative ability and pragmatic dif culty. This checklist
is completed by the childs teacher or therapist and consists of a number of
statements about communication and behaviour. The respondent must say
whether the behaviour de nitely applies, applies somewhat or does not
apply to that child. There are eight subscales, ve of which make up a scale of
pragmatic impairment. A score of 132 or less on this scale represents clinical
dif culty with social communication. The checklist is described more fully in
Botting and Conti-Ramsden (1999) and Bishop (1998).
The ages of the ve children with PLI (one of whom was female) ranged
from 7 years, 7 months to 8 years, 8 months. The ve comparison children
with SLI were matched for gender and age (range from 7 years, 7 months to
8 years, 1 month). In addition to the narrative stimuli which are described
shortly, children were assessed using a receptive language test (the Test for
Reception of Grammar, (TROG) Bishop, 1982); An expressive language
measure (ITPA Grammatic Closure subtest, Kirk et al., 1968); an expressive
vocabulary subtest (British Ability Scales (BAS) Naming Vocabulary, Elliot,
1983) and a non-verbal abilities test (Ravens Coloured Matrices; Raven,
1986). The BAS Single Word Reading subtest was also used. All children in
both groups had normal non-verbal intelligence scores as measured by Ravens
coloured progressive matrices (Raven, 1986) and borderline=low comprehension scores as measured by the Test for Reception of Grammar (Bishop, 1982).
There were no signi cant differences between the groups on these measures
although there was a tendency for the children with PLI to have higher nonverbal scores. A signi cant difference was seen on word reading ability (Mann
Whitney exact P = 0.03) with the group with PLI showing superior performance. Table 1 shows a summary of the childrens characteristics including
some other measures detailed later, and (for all but one) values are given in
centiles for age where <15.9 is equivalent to <1 SD and 50 is the theoretical
population mean.
Narrative as a tool
Table 1
7;7=m
7;8=f
7;9=m
8;7=m
8;8=m
7;11=m
8;0=f
7;8=m
7;10=m
8;1=m
16
25
92.5
97.5
92.5
37.5
82.5
82.5
62.5
62.5
3
17.5
37.5
25
1
25
10
17.5
5
50
9
77
48
48
13
73
7
30
9
48
38
50
46
83
10
3
28
1
4
21
32
29
37
38
11
32
28
30
24
44
Narrative measures
Retelling: The Bus Story (Renfrew, 1991). This is a standardized test used
frequently by speech-language therapists in the UK which consists of a story
book with pictures and no written words. Children are told the story orally by
the researcher whilst looking at each picture. The children are requested to
retell the story as close to the original as possible using the pictures as
prompts. The amount of original information included, the number of
subordinate clauses compared to mean for age and the mean sentence
length of the longest ve sentences compared to mean for age are recorded.
Generative: The Frog Story (Frog, where are you?; Mayer, 1969). This is
a 24-picture story book without words. In this study, the child was asked to
look through the pages of the book and then to tell the story spontaneously
from the beginning in whatever way he=she wants, using the pictures provided
as prompts. Berman and Slobin (1994) have reported a series of studies using
this assessment, and its episodic structure has been extensively examined in
normally developing childrens narratives (Bamberg, 1987; Bamberg and
Marchman, 1990; Bamberg and Damrad-Frye, 1991) and in clinical populations (Tager-Flusberg, 1995; Van der Lely, 1997).
Analysis
Length and narrative devices. Narratives were rst analysed for overall
linguistic differences. Variables for length using both number of words and
number of propositions (that is complete phrase structures with at least a noun
and verb present) were examined. The number of tense marking errors was
Downloaded from clt.sagepub.com at American University in Cairo on November 5, 2015
noted for each childs narrative as this is also a good indicator of linguistic
limitation. The stories were also analysed in terms of the number of narrative
devices used based on Bamberg and Damrad-Fryes (1991) ve category
system: Frames of Mind (referred to here as mentalizing terms), Negatives,
Causatives, Hedges (such as might, perhaps) and character speech (number
of actual words spoken by a character).
Through other narrative devices, story telling may be able to provide
insights into a childs understanding of others and their ability to use
appropriate mentalizing terms. Therefore, in our analysis two further categories were used: affective enhancers and socio-cognitive enhancers, following
Tager-Flusberg (1995). The rst of these refers to emotional terms such as felt
scared, lonely, angry; character speech; sound effects; audience hookers
such as suddenly; and emphatic markers such as really, really slowly.
Socio-cognitive enhancers refers to mentalizing terms such as thought,
remembered, knew; negatives such as the boy wasnt asleep; causal
statements; and inferences where the narrative gives information not directly
given in the pictures. Finally, the number of time references were also recorded
for this preliminary data.
Story structure. The overall story-telling style was assessed using TagerFlusbergs (1995) scheme. This assessed whether each child had included a
formal opening, orientation to characters=setting, explicit mention of the
theme (i.e., looking for the frog), a resolution and a formal ending. These
were also summed to give an overall story structure score for each child.
The analyses described thus far form the data base of the illustrative data. In
addition, data from Tager-Flusbergs (1995) study using the Frog Story is
discussed in relation to the SLI and PLI ndings. Her study involved children
with autism as well as a normally developing group and children with global
learning dif culties.
Narrative as a tool
Word
length
Bus
Word
length
Frog
Tense
errors
Bus
Tense
errors
Frog
Bus
story
info
Bus
story
sub.cl.
Sociocognitive
devices
Affective
devices
SLI
151
221
28
28th
centile
7.6
22
PLI
137
276
3.6
20
35th
centile
2.6
(from
pop.mean)
3.6
(from
pop.mean)
13.4
51
10
tense marking is another valuable area of assessment from narrative, and may
prove to be a useful alternative to formal testing, especially if the child is
regularly assessed over a long period of time, to avoid testretest issues. In
addition, some older children with SLI manage to recall correct answers in
testing situations, but given the task of generating narrative, are unable to
apply this knowledge. It is worth noting here that children with PLI were far
from competent in their linguistic knowledge in both types of narrative, but
that their tense marking errors were not related to language test results.
14
wider sample, and in what ways it relates to every-day conversation with these
children. It may be that children with pragmatic language impairment actually
pretend to be other characters whilst they are speaking to others. This
speculation would explain why roles are often confused and why these
children sometimes seem to assign themselves incorrectly in conversation
(for example talking to the headteacher as another teacher might). Given the
limited generative ability of children with autism, and the dif culty often
experienced with pretend play and role taking generally, it would intuitively
seem unlikely that this result would be found in a group with ASD. Indeed
Tager-Flusberg (1995) reported a mean use of character speech of 4.2 in her
group with autism much lower than the two groups described here.
Relationship of narrative to pragmatic ability. An interesting possible
feature of narrative for clinical use may be its relationship to pragmatic ability.
The analysis of this illustrative sample revealed that the two factors may not
only be related, but that this relationship may differ across clinical groups.
Table 4 shows raw data for pragmatic scores at age 8 for children with PLI,
alongside various narrative factors. Spearman rank correlations are shown,
which because of the small data set only reach statistical signi cance when
near perfect. Nevertheless, several features show a clear relationship with
pragmatic skill including mentalizing terms, story length and social-cognitive
enhancers. The less pragmatic impairment, the greater the use of such devices.
Surprisingly, this relationship is reversed when examining the group with
SLI. Here pragmatic score is inversely related to factors such as frames of
mind (correlations also shown in Table 4). Thus the better the pragmatic skill,
the fewer frames of mind are included. This may be due to a dual mechanism
effect between pragmatic skills and linguistic skills. That is, in children with
PLI it may be that linguistic devices are not used largely because of sociocognitive limitations despite suf cient linguistic skill, whilst for the children
with SLI mentalizing terms may not be present because of linguistic impairment, despite suf cient understanding of mind and of mentalizing terms.
These differences may prove important guiding features in the assessment and
description of children with communication dif culties. This suggestion is
supported by Tager-Flusbergs (1995) similar preliminary conclusion from
data showing that two autistic children with better narratives also passed false
belief tasks and by Capps et al., (2000) who also found an association between
theory of mind, social competence and narrative in children with autism. This
latter study also failed to nd the same associations in a group with general
developmental delay. Thus for children with PLI, poor socio-cognitive skill
may prove the limiting mechanism. For children with SLI, linguistic skill may
Downloaded from clt.sagepub.com at American University in Cairo on November 5, 2015
132
130
119
111
107
559
1
261
1
207
0
228
0
125
0
0.9*
0.87
0.6 0.22
66
98
11
23
18
0.6
0.8
8
7
3
1
2
0.90*
0.95*
2
2
0
0
0
0.87
0.58
1
2
3
4
5
Corr.
SLI r
4
3
0
0
0
0.89*
0.11
4
3
0
0
1
0.67
0.89*
3
0
0
0
0
0.71
0.35
35
13
7
7
5
0.95*
0.6
85
111
14
24
21
0.6
0.9*
5
3
4
5
3
0.32
0.74
Tense
Audience Emphatic SocialAffective
Story
marking hookers
markers cognitive enhancers structure
errors
enhancers
score
5
5
2
10
4
3
3
1
1
1
0.7
0.87
0.58 0.00
Child
CCC Words Hedges Character Frames Causatives Sound Time
with PLI
speech
Of
effects refs.
mind
Table 4
Narrative as a tool
15
16
be the central handicap to narrative and for those with autism, both mechanisms may be affected thus producing the poorest narrative style.
It is acknowledged that the numbers used in this illustrative analysis are too
small to draw rm conclusions, but the striking correlations in each clinical
group seem to suggest that the notion of a dissociation between pragmatic and
linguistic skill warrants further study. It is a widely held belief in the United
States, for example, that pragmatic impairment is very likely to be the result of
poor linguistic ability (Leonard, 1998). In addition there is much confusion
about the level of language impairment allowed within a diagnosis of
Aspergers disorder (Attwood, 1998). It may be that this latter problem can
be untangled by considering the different mechanisms behind communicative
dif culties. For example, a hypothetical ideal assessment of pragmatic
ability, considering the relationship with both linguistic and meta-cognitive
limitations, may be able to reveal that individuals with Aspergers Syndrome
have pragmatic language dif culties associated with social cognition. Conversely, peers with other types of language impairment might show dif culties
relating to linguistic knowledge. In theory, this would allow much more
satisfactory description and remediation of young children with a number of
different communication problems and also warrants deeper investigation.
In general, the illustrative and exploratory analysis used as an example here
seems to support previous work suggesting that children with PLI are not best
described as having speci c language impairment but also suggests that they
exhibit important differences in narrative style in relation to peers with autism
(Botting and Conti-Ramsden, 1999; Conti-Ramsden and Botting, 1999;
Bishop, 1998).
Concluding remarks
In work with clinical populations, the study of spontaneous communication in
the form of conversational analysis has not always proven as useful as one
might have hoped (e.g., Bishop and Adams, 1989). This is partly because it is
dif cult to control for the competent conversational partner (the researcher)
who often unwittingly eases the pragmatic burden for the child being studied.
Furthermore judgements about what constitutes an appropriate or inappropriate contribution are more problematic in naturalistic conversational settings.
On the other hand, in this article it is argued that the generation of stories from
pictures may provide an ideal mixture of structured but imaginative communication. Thus, one implication from this study is that narrative needs to be
used more regularly as an assessment tool in clinical practice. The Frog Story
used here would be a good choice for clinicians wishing to use it in this way
Downloaded from clt.sagepub.com at American University in Cairo on November 5, 2015
Narrative as a tool
17
Acknowledgements
This article was undertaken as part ful lment of a post-doctoral Simon
Fellowship (part-time) awarded to the author by the University of Manchester.
Many thanks also to the Nuf eld Foundation whose generous funding also
made this research possible (grant DIR=28), to Gina Conti-Ramsden for
looking at earlier drafts of this article, and to the families and children who
participated in the study.
References
Attwood, T. 1998: Aspergers syndrome. London: Jessica Kingsley.
Bamberg, M. 1987: The acquisition of narratives. Berlin: Mouton de Gruyter.
Bamberg, M. and Damrad-Frye, R. 1991: On the ability to provide evaluative
comments: further explorations of childrens narrative competencies.
Journal of Child Language 18, 689710.
Bamberg, M. and Marchman, V. 1990: What holds a narrative together? The
linguistic encoding of episodic boundaries. Papers in Pragmatics 4, 58
121.
Downloaded from clt.sagepub.com at American University in Cairo on November 5, 2015
18
Narrative as a tool
19
20
Narrative as a tool
21