text structure and patterns of cohesion in narrative texts written by adults with a history of...
TRANSCRIPT
Text structure and patterns of cohesion in narrativetexts written by adults with a history of languageimpairment
Lynne Mortensen Æ Karen Smith-Lock ÆLyndsey Nickels
Published online: 23 October 2008
� Springer Science+Business Media B.V. 2008
Abstract This paper examines text structure and patterns of cohesion in stories
written by a group of adults with a history of childhood language impairment. The
study aimed to extend our knowledge of writing difficulties in this group by building
upon a study that examined clause level phenomena (Smith-Lock, Nickels, &
Mortensen, this issue). Ten adults with a history of Language Impairment and 30
control participants were asked to write the story of Cinderella. Stories were ana-
lyzed for their generic structure and cohesion resources, both of which contribute to
the organization and coherence of a text. Results revealed that patterns of text
organization at the level of generic structure and measures of cohesion did not
distinguish the performance of writers with a history of Language Impairment from
the comparison group. A wide range of lexico-grammatical skills was evident within
the Language Impaired group, with few individuals with Language Impairment
falling outside the normal range of performance. It is suggested that generic
structure and cohesion are a relative strength in the writing of adults with Language
Impairment, within the constraints of their lexical and grammatical skills.
Keywords Adult language impairment � Written discourse � Text organization �Coherence � Generic structure � Cohesion
Introduction
The ability to produce written texts is a critical life skill in a literate society. Writing
is a complex and demanding form of communication that draws upon a number of
L. Mortensen (&) � K. Smith-Lock � L. Nickels
Macquarie Centre for Cognitive Science, Macquarie University, North Ryde, NSW 2109, Australia
e-mail: [email protected]
K. Smith-Lock
e-mail: [email protected]
123
Read Writ (2009) 22:735–752
DOI 10.1007/s11145-008-9150-x
underlying cognitive, linguistic, and social factors. It uses information-processing
abilities such as attention, perception, memory and reasoning, language skills such
as vocabulary, syntax and text organization, as well as phonological and
orthographic skills. In addition, a writer requires an awareness of the person(s)
for whom the written material is intended, i.e., ‘‘a sense of audience’’ (Gregg, 1991).
Impairment in the ability to write, be it a developmental or acquired deficit, clearly
has far-reaching implications for an individual’s literacy, learning, and employment
opportunities, as well as expression of self and identity (Ivanic, 1998).
In studying and analysing the discourse abilities of adults with writing
impairment, a number of different theoretical orientations are apparent. Discourse
production has been described in terms of contextual influence on the function of a
text (Halliday & Hasan, 1985), cognitive organization and process (e.g., Stein &
Glenn, 1979; van Dijk, 1981), linguistic structure and textual coherence (Bamberg,
1987; Halliday & Hasan, 1976), and sentence-level complexity (Bishop & Clarkson,
2003; Gillam & Johnston, 1992; Scott & Windsor, 2000). These perspectives reflect
the complex of social, cognitive, and linguistic factors that comprise written
discourse and provide different perspectives from which to examine discourse
production, written or spoken.
Written discourse has been found to be a sensitive indicator of residual or
persistent language impairment in adults. Investigations of adults and adolescents
with traumatic brain injury, for example, have revealed problems with text
organization and information management in written texts, while their lexicogram-
matical abilities remain relatively intact (Mortensen, 2004a; Wilson & Proctor,
2002). Adults with aphasia, on the other hand, have demonstrated competent
narrative text structure in spite of difficulties at the clause level (Freedman-Stern,
Ulatowska, Baker, & DeLacoste, 1984; Mortensen, 2004b).
Investigations have also addressed the written narratives of adults with a history
of specific language impairment and learning disorders. These studies have adopted
different linguistic foci for examination. For example, Smith-Lock & Nickels (2005,
2006, Smith-Lock, Nickels, & Mortensen, this issue) examined grammatical
complexity, semantic content and accuracy of spelling and punctuation in stories
written by adults with Language Impairment. They found that as a group, the adults
with Language Impairment showed no difference from the comparison group in the
length of their stories as measured by total words and total utterances (T-units).
However, there were group differences on measures that assessed grammatical
complexity, measured by mean length of T-unit, as well as accuracy of grammar,
spelling and punctuation, with Language Impaired subjects performing below the
comparison group. Individual profiles of the writers with Language Impairment
indicated substantial variability.
Narrative ability, both written and spoken has also been identified as problematic
in adolescents with Learning Difficulties (Gregg & Hoy, 1990; Reed, Patchell,
Coggins, & Hand, 2007; Wong, 2000). Wong (2000), in a paper that describes
strategies in expository essay instruction, reports on the different perspectives that
writing research has taken over time in relation to adolescents with Learning
Difficulties, including an initial focus on spelling and punctuation, followed by
attention to the cognitive processes involved in written text production. Wong
736 L. Mortensen et al.
123
discusses the importance of expository as well as narrative discourse in the
educational setting and, together with other researchers (e.g., Englert, Raphael,
Fear, & Anderson, 1988; Englert & Thomas, 1987; Wong, Wong, & Blenkisop,
1989), outlines a number of features that are typical of the learning disabled
adolescent writer. These include delay in expressing ideas, concern with spelling
errors when editing rather than clarity and organization of content, unproductive
word finding strategies, poor spelling and punctuation, poor knowledge of paragraph
structure and logical organization of ideas. These features imply that problems for
the adolescent with Learning Difficulties exist for both cognitive, and linguistic
aspects of discourse production. Gregg and Hoy (1990), however, report different
findings in a study that examined organizational and linguistic features in expository
essays written by 35 college students with a history of Learning Difficulties, when
compared to normally achieving writers. Essays were analyzed in terms of cohesion
and coherence and their relationship to text organization. The writers with Learning
Difficulties demonstrated a good understanding of principles of text structure,
similar to the normally achieving writers. Their cohesive referencing ability was
also similar to normal writers, making effective use of pronouns for referencing and
selecting a wide range of pronoun types. Gregg and Hoy (1990) report that these
strengths were evident in spite of spelling and punctuation errors.
These varied findings confirm the knowledge that writing is a complex process
and that different levels of language (i.e., spelling, lexico-grammar and text
organization), as well as different types of texts may be usefully examined to
identify and explain patterns of difficulty that characterize written discourse by
adults with Language Impairment. The purpose of this study was to extend our
understanding of the nature of narrative writing difficulties of adults with a history
of Language Impairment by looking beyond the clause level phenomena examined
by Smith-Lock and colleagues (2005, 2006, this issue). The aim was to investigate
the integrity of text structure for a narrative task and to explore grammatical and
lexical abilities for the creation of a coherent text. The study took a functional
linguistic approach, which provides a number of perspectives from which to
examine language form in terms of various strata including word, clause and text
levels. This framework has been adopted for the teaching and evaluation of text
writing in primary education (Christie & Martin, 1997; Derewianka, 1992), in adult
literacy, both remedial and second language acquisition, (Hammond, Burns, Joyce,
Brosnan, & Gerot, 1992), and for the investigation of written communication
disorders in adults (Gregg & Hoy, 1990; Mortensen, 2004a, b).
Method
Participants
Ten adults with a history of childhood Language Impairment (3 women and 7 men)
took part in the study. Their ages ranged from 18 to 54, with a mean age of 27 years
(SD = 11.12). A comparison group included 30 university students, participating
for course credit. This group was a randomly selected subset of the comparison
Narrative texts in language impairment 737
123
subjects participating in the study of grammatical skills in narrative writing by
Smith-Lock et al. (this issue). These participants had no history of Language
Impairment or reading/writing difficulties, and their ages ranged from 18 to
55 years, with a mean age of 22 years (SD = 8.7). All participants were native
speakers of Australian English.
Language Impaired participants were recruited in two ways: (1) from past
clinical records or (2) from self-reported history of Language Impairment with
evidence of current oral language difficulty. The two different recruitment
procedures were used to generate sufficient numbers for the study, given the
paucity of clinical records available for follow-up. Self-reported history as a means
of identification has been used in pedigree studies of specific language impairment
(Plante, Shenkman, & Clark, 1996; Rice, Haney, & Wexler, 1998; Tallal et al.,
2001). These selection procedures consistently identify more family members with a
self-reported history of Language Impairment in the affected group than in a
comparison group with no Language Impairment. Interestingly, self-report has been
found to identify fewer affected family members than testing (Plante et al., 1996),
suggesting that self-report, if anything, might under identify rather than over
identify participants as affected. Nevertheless, on a case by case basis, individuals
identified by self-report do not always coincide with individuals identified by testing
(Rice et al., 1998). For this reason, it was decided to use both self-report and testing
in adulthood in those cases without documented clinical history. The legitimacy of
the ‘self-report ? current testing’ in the study was bolstered by the fact that in all
but one of the self-reported cases, the participant had a child with specific language
impairment for whom clinical records were available, suggesting a familial pattern
(the third participant did not have children). The combination of recruitment
methods, while not ideal, resulted in rigorous participant selection. Any error made
was likely to be in the omission of an appropriate participant due to potentially
‘resolved’ Language Impairment resulting in no oral deficits in the current battery in
spite of self-reported history and affected offspring.
Thus, participants were selected if they met one of the following two selection
criteria.
1. Documented evidence of childhood developmental language difficulty (n = 7).
Five of these participants were invited to participate after their childhood files
were reviewed at a local speech pathology clinic. One was referred by a
community speech pathologist. The final participant in this subgroup contacted
the researchers due to concern over his language skills.
2. Self-report of childhood developmental language difficulty, corroborated by
evidence of oral language difficulty as an adult (n = 3). Two of these
participants were parents of language-impaired children taking part in another
study and who reported histories of childhood language impairment themselves.
The remaining participant reported a history of language difficulties when he
was recruited for the comparison group.
All LI participants completed the following oral language tests:
1. Boston Naming Test (Kaplan, Goodglass, & Weintraub, 1983)
738 L. Mortensen et al.
123
2. Sentence Repetition subtest of Test of Adult and Adolescent Language-3(TOAL-3, Hammill, Brown, Larsen, & Wiederholt, 1993)
3. Non-word Repetition Test. This test consisted of a total of 30 words: 10 two
syllable, 10 three syllable and 10 four syllable words taken from Gathercole,
Willis, Baddeley, and Emslie (1994). Z-scores were calculated based on the
performance of the comparison group.
Individual participant details can be seen in Table 1. All Language Impaired
participants except one received a below normal score on two or more of the above
language measures, with 5 of the 10 scoring below normal on all 3 measures
(participants 3, 5, 6, 9 and 12). The one participant who scored below normal on the
Boston Naming Test only (participant 1) had a confirmed history of childhood language
impairment, including placement in a primary school language impairment class and
assessment as an adolescent by a community speech pathologist. Participants’ formal
education ranged from grade 7 to post high school (university or technical college).
Procedure
Participants were asked to handwrite the story of Cinderella, with no time constraint.
Their instructions were to ‘‘Write the story of Cinderella in as much detail as you can.’’
The stories were then transcribed into a computer file and analyzed for their text
structure and organization, and linguistic features of cohesion, both of which
contribute to the coherence of a text (Eggins, 1994). An analysis of the grammatical
features of the stories is reported in Smith-Lock et al. (this issue).
Generic structure analysis (GSA) was selected to evaluate text structure and
organization. GSA describes the structural elements of a text according to the text’s
purpose. Different text types have uniquely defining structures, which include
obligatory (core) and optional (elaborative) elements that occur in a particular order
(Hasan, 1996). A text is perceived as well organised and complete if it contains all
of the obligatory elements in the appropriate order. Drawing on Hasan’s framework,
a generic structure statement for the Cinderella story was established which
encompassed core obligatory content, based upon the classic version of the story by
Charles Perrault (Ehrlich, 1986). Obligatory elements included Introduction, which
names Cinderella and describes how she lives with her stepmother and stepsisters
and their harsh treatment of her. This was followed by Complication which
embodies a sequence of events associated with a ball that Cinderella attends,
Resolution, which provides a logical outcome of the activities associated with the
ball, and Conclusion, which describes the overall outcome or conclusion to the
story. Optional elements included Placement which identifies the narrative as a fairy
tale as well as additional orienting information, such as events leading to Cinderella
living with her stepmother and stepsisters, and Coda which evaluates the whole
story and may provide a moral. Details of the generic structure and related content
are outlined in Appendix Table 4. Stories were analyzed with respect to:
(i) Structural completeness, i.e., presence of obligatory elements.
(ii) Elaboration of the text structure, i.e., presence of optional elements.
(iii) Logical development of the text, i.e., appropriate ordering of elements.
Narrative texts in language impairment 739
123
Tab
le1
Par
tici
pan
tch
arac
teri
stic
s
Num
ber
Age
Sex
Rec
ruit
men
tE
duca
tion
Occ
upat
ion
Bost
on
nam
ing
test
(z-s
core
)
Sen
ten
ce
rep
etit
ion
(SS)
No
n-w
ord
rep
etit
ion
ou
to
f4
0(z
-sco
re)
P1
21
FR
efer
red
by
SL
P2
yea
rspost
-sec
ondar
yN
ot
avai
lable
-2
.49
29
(0.2
7)
P2
19
MC
lin
icre
cord
sG
rad
e1
2N
ot
avai
lab
le-
2.4
31
8(-
3.4
5)
P3
22
FC
lin
icre
cord
sG
rad
e7
Ho
me
du
ties
-6
.94
11
7(-
3.7
9)
P4
22
FC
lin
icre
cord
sG
rad
e1
0N
ot
avai
lab
le-
4.5
21
5(-
4.4
7)
P5
18
MC
linic
reco
rds
Tes
ted
inG
r1
2S
tuden
t-
4.5
12
5(-
1.0
9)
P6
22
MC
lin
icre
cord
sG
rad
e1
2N
ot
avai
lab
le-
0.6
52
24
(-1
.43)
P7
29
MS
eek
ing
hel
p;
chil
dre
cord
sS
up
po
rted
hig
hsc
ho
ol
Kit
chen
han
d-
6.9
42
14
(-4
.81)
P8
28
MR
epo
rted
his
tory
wh
en
recr
uit
edfo
rco
mp
aris
on
gro
up
Fir
sty
ear
un
iver
sity
stud
ent
atti
me
of
test
ing
Un
iver
sity
stud
ent
1.1
62
0(-
2.7
8)
P9
38
MS
elf-
rep
ort
edh
isto
ry&
par
ent
of
SL
Ich
ild
Gra
de
10
(plu
s
app
ren
tice
ship
)
Car
pen
ter
-0
.23
32
4(-
1.4
2)
P1
05
4M
Sel
f-re
port
edh
isto
ry&
par
ent
of
SL
Ich
ild
Gra
de
10
(plu
s
app
ren
tice
ship
)
Cab
inet
mak
er(t
each
er
of
cab
inet
mak
ing)
-1
.45
52
6(-
0.7
5)
740 L. Mortensen et al.
123
A Cohesion analysis, based on the work of Halliday and Hasan (1976), was also
undertaken. Cohesion refers to the way in which a text holds together to form a
coherent whole. A coherent text demonstrates continuity of topic, reference, and
ideas. Cohesion involves a set of linguistic resources—lexical and grammatical, for
linking one part of a text to another, thus forming the foundation of coherence in a
text. Lexical cohesion refers to the cohesive effect achieved by selection of
vocabulary, established through relationships of reiteration or collocation. Reiter-
ation involves the use of repetition, synonyms, antonyms, hyponyms (i.e.,
taxonomic relations such as family: father mother, children), or meronyms (i.e.,
part-whole relationships, such as house: cellar). In this study meronyms and
hyponyms were merged into the broader category of super/subordinate relations.
Collocation refers to lexical items which regularly co-occur, or are causally related,
(such as magic spell; hum music). Grammatical cohesion includes, among other
resources, referencing through the use of pronouns, definite articles, demonstratives,
and comparatives. Stories were analyzed for:
i. The total number of lexical and grammatical tokens—known as Total tokens.A token is a content word used either specifically in the text or interpreted
through an implicit cohesive device.
ii. The number of grammatical and lexical tokens entering into cohesive chains—
known as Relevant Tokens.1
iii. The number of grammatical tokens forming reference chains (grammatical
relevant tokens)
iv. The number of lexical tokens entering repetition, synonym, antonym, super/
subordinate and collocation chains (lexical relevant tokens).2
Inter-rater reliability for 20% of the stories (5 Language Impaired texts and 3
comparison texts) was undertaken by two independent raters. Percentage agreement
for the Generic structure analysis was 96. Percentage agreement for the Cohesion
analysis was 98.1
Results
Generic structure analysis
Results of the generic structure analysis are presented in Table 2. The high mean
number of obligatory elements included indicates that, overall, both groups
produced structurally complete texts. On average, the groups included one (out of
two) optional elements. A mixed analysis of variance with one between groups
factor (group: Language Impaired, comparison) and one within groups factor
1 When calculating relevant tokens, the initial token and all related subsequent tokens (grammatical &
lexical) are included in the relevant token count.2 When calculating the lexical and grammatical tokens which form chains, the initial token is excluded
because there may be a number of chains associated with it, e.g. Cinderella- (Initial token) may have a
related repetition chain (Cinderella), a reference chain (she, her), a super/subordinate chain (young girl/
Cinderella; Cinderella/foot;). Therefore, including the initial token could artificially inflate the count.
Narrative texts in language impairment 741
123
(element: obligatory, optional) indicated no significant group 9 element interaction
(F(1,38) = 2.48, p [ .05) and no significant main effect for group (F(1,38) = 1.19,
p [ .05).
Individual analysis
Each individual’s scores were compared to the comparison group’s scores using
z-score comparisons. Four of the 10 Language Impaired participants (40%) were
significantly different from the comparison participants (Modified t-test, p [ .05,
2-tailed, Crawford & Garthwaite, 2002). This reflects three Language Impaired
narratives (30%) which omitted one obligatory element (P2, P9, and P10) and one
Language Impaired narrative (10%) that omitted two obligatory elements (P6).
However, it should be noted that four of the comparison participants (13%) also
omitted one obligatory element (C12, C13, C23, and C26). Nevertheless, the
difference in the proportion of participants per group who omitted obligatory elements
showed a trend towards significance (Fisher Exact test, p = 0.089, 1 tailed).
With respect to number of optional elements, three Language Impaired participants
(30%) included significantly fewer than the comparison participants, including no
optional elements (P4, P7, P9, t = -2.249, p = 0.032 2-tailed). However, four
comparison group participants (13%) also wrote narratives including no optional
elements (C11, C22, C24, C26). The difference in the proportion of participants per
group who omitted optional elements was not significant (Fisher Exact test,
p = 0.228, 1 tailed).
Examples of texts analyzed for generic structure are presented in Appendix
Table 5–7. These include Language Impaired participant P7 who showed complete
generic structure, Language Impaired participant P6 who omitted two obligatory
elements, and comparison group participant C26 who omitted one obligatory
element.
Errors
As a reflection of logical development of the text, stories were examined for
appropriate ordering of the story elements. One subject from each group: P7
(Language Impaired group), and C26 (comparison group), demonstrated inappropri-
ate ordering of the story elements (see Appendix Table 5–7). P7 in his unelaborated,
yet structurally complete, text introduces the godmother (element 3) before
introducing or describing the ball (element 2). The following elements, however,
are in keeping with the established story sequence. In the case of C26, the introduction
is followed by an incomplete complicating event, which is out of sequence [One day
Table 2 Group results for
generic structure analysisGroup Number of obligatory
elements included (/9)
Number of optional
elements included (/2)
Mean Std Mean Std
LI 8.5 0.76 0.8 0.63
Comp 8.8 0.36 0.9 0.35
742 L. Mortensen et al.
123
she met a fairy God-mother who told her that she would be able to go to]. This
incomplete event comes too early in the story (i.e., in the original story, Cinderella is
invited to the ball and is told she cannot go before the fairy godmother appears). The
fact that the sentence was incomplete suggests that this was perhaps a false start, with
no explicit repair. This, together with a slightly different interpretation of the story,
renders the text initially confusing and somewhat disjointed.
In sum, group comparisons of generic structure found no significant differences
between the Language Impaired and comparison groups. Individual analyses
indicated that members of both groups omitted obligatory and optional elements.
Nevertheless, a greater proportion of Language Impaired group members than
comparison group members omitted elements, either obligatory or optional, from
their narratives.
Cohesion analysis
Results for the cohesion analysis can be seen in Table 3. The number of relevant
tokens (i.e., lexical or grammatical tokens entering chains subsequent to an initial
token) was divided by the total number of grammatical and lexical tokens in the
narrative in order to control for narrative length across participants in all analyses.
A mixed analysis of variance with one between groups factor (group: Language
Impaired, comparison) and one within groups factor (number of subsequent tokens
entering cohesive chains: grammatical, lexical) indicated no significant chain by group
interaction (F(1,38) = 0.75, p [ 0.05), no main effect for group (F(1,38) = 0.25,
p [ 0.05) but a significant main effect for chain type (F(1,38) = 51.46, p [ 0.001).
Thus, the groups did not differ significantly in the number of tokens in cohesive chains.
However, significantly more tokens were involved in lexical chains than grammatical
chains. This is to be expected, by the nature of the large range of lexical relationships
possible as opposed to the more restricted range of grammatical relationships that have
the specific function of reference.
In order to examine the type of lexical chains used, a mixed analysis of variance
was undertaken, with one between groups factor (group: Language Impaired,
comparison) and one within groups factor (number of subsequent tokens per chain
type: repetition, synonym, antonym, super/subordinate, collocation). Number of
tokens for each chain type was divided by the total number of lexical tokens in
cohesive chains in order to control for number of lexical chains established. There
was no significant group by chain type interaction (F(1,38) = 1.9, p [ 0.05), but a
significant main effect for chain type (F(1,38) = 909.16, p [ 0.001). Inspection of
the data indicates that the majority of lexical tokens in cohesive chains were
involved in repetitive chains.
A series of independent group t-tests compared the use of each lexical chain type
between the groups. Significance levels were adjusted using the Bonferroni correction
for repeated t-tests. No significant differences were found (repetition: t(38) = 1.1,
p [ .01; synonym: t(38) = 1.39, p [ .01, antonym: t(38) = 1.61, p [ .01; sup/
subordinate: t(38) = 0.67, p [ .01; or collocation: t(38) = 0..72, p [ .01). Thus, the
groups did not differ significantly in the number of tokens involved in the various chain
types.
Narrative texts in language impairment 743
123
Ta
ble
3R
esult
sfo
rg
ram
mat
ical
and
lex
ical
coh
esio
n
To
tal
tok
ens
Rel
evan
t
tok
ensa
To
ken
sin
gra
mm
atic
al
chai
nsa
To
ken
sin
lex
ical
chai
nsa
Pro
po
rtio
no
fle
xic
alch
ains
for
each
chai
nty
pe
Rep
etit
ion
Synonym
Anto
nym
Super
/Subord
inat
eC
oll
oca
tion
Co
mpar
iso
ng
rou
pm
ean
(SD
)1
64
.03
(70
.6)
0.6
5(0
.09
)0
.21
(0.0
5)
0.3
2(0
.06
)0
.72
(.0
8)
0.0
6(0
.03
)0
.05
(0.3
)0
.06
(0.0
3)
0.1
0(0
.04
)
LI
gro
up
mea
n(S
D)
15
3.1
(10
9.8
)0
.66
(0.0
8)
0.1
8(0
.03
)0
.34
(0.0
6)
0.7
6(0
.1)
0.0
5(0
.03
)0
.04
(0.0
2)
0.0
7(0
.06
)0
.1(0
.05
)
Ind
ivid
ual
da
tafo
rL
an
gu
age
Impa
ired
gro
up
P1
42
70
.58
0.1
70
.32
0.7
40
.07
0.0
40
.02
0.1
3
P2
12
40
.72
0.1
80
.40
.86
0.0
40
.02
0.0
40
.04
P3
46
0.7
0.2
20
.29
0.5
40
.00
0.0
80
.23
0.1
5
P4
13
20
.70
.22
0.3
20
.86
0.0
20
.00
0.0
50
.07
P5
13
70
.69
0.1
90
.35
0.6
70
.06
0.0
40
.04
0.1
8
P6
78
0.6
50
.19
0.3
0.7
40
.04
0.0
40
.13
0.0
4
P7
60
0.6
30
.17
0.3
20
.84
0.0
00
.05
0.0
50
.05
P8
22
70
.48
0.1
10
.26
0.7
20
.07
0.0
50
.05
0.1
P9
13
50
.75
0.1
70
.46
0.7
80
.08
0.0
20
.03
0.0
8
P1
01
67
0.7
40
.21
0.3
60
.80
.07
0.0
20
.05
0.0
7
aA
sa
pro
po
rtio
no
fto
tal
token
s
744 L. Mortensen et al.
123
Individual analysis
Each individual’s scores were compared to the control group’s scores using
Crawford and Garthwaite’s (2002) modified t-test. One Language Impaired
participant had a significantly smaller proportion of relevant tokens than the
control group (P8, proportion of total tokens that are relevant tokens = 0.48; t =
- 2.07, p = 0.037, 2-tailed). However, two participants in the comparison group
had similarly small proportions (C1, 0.38; C6, 0.45). No Language Impaired
individual showed a significantly different distribution of cohesive chains across
lexical and grammatical chains compared to the comparison participants.
Analysis of the types of lexical chains used, indicated substantial individual
variation in both groups. This variability is due, in part, to the small overall numbers
of lexical cohesive chains of types other than repetition. There were very small
numbers of cohesive chains, particularly for synonomy, antonomy, and super/
subordinates. Hence, due to the small numbers of items in these categories
individual analyses are inappropriate. We will therefore focus on repetition and
collocation. In the use of repetition as a cohesive device, none of the Language
Impaired participants used significantly more repetition in lexical cohesion than the
comparison group. One Language Impaired participant (P2) was outside the range
of the comparison group although only significantly different from the mean on a
one-tailed test (t = 1.722, p = 0.048, 1-tailed; p = 0.096, 2-tailed). With respect to
use of collocation, one of the Language Impaired participants was significantly
above the comparison group mean (P5, t = 1.967, p = 0.029 1-tailed, p = 0.059,
2-tailed) although not outside the range (two comparison group participants used a
higher proportion of collocation).
Examples of the different lexical categories selected by the Language Impaired
and comparison groups are presented in Appendix Table 8.
Errors
Errors of reference were observed only in the Language Impaired group. Three
Language Impaired participants demonstrated one instance each of problematic
reference, which resulted in reduced clarity and coherence within the texts. The
example from P3 below shows difficulty establishing correct reference for the
pronoun they.
P3 and the prince went to her house and her sisters try the slipper on they didn’tfitthere feet. So Cinderella went to try it on and they lived very happily after
In the first instance, the use of the plural rather than the singular pronoun makes
they refer to the sisters when the full sentence they didn’t fit there feet indicates that
the intended reference was the slipper. The difficulty is compounded in the
following sentence by the use of they again, here without a clearly established
antecedent. Possible antecedents include (i) the previous they, already confused
between the sisters and the slipper, (ii) Cinderella, the subject of the previous
clause, which would require the singular pronoun she, and finally, (iii) the clearly
Narrative texts in language impairment 745
123
intended, but unstated antecedent, Cinderella and the Prince. The inexplicit use of
they requires knowledge of the story in order for it to make sense to the reader.
The example from P7 below also contains problematic pronoun reference.
P7 The three stepsister and cinderella and the godmother come out and made awhisper and she go to the dance
Here, a pronoun was used to refer to a portion of a noun phrase, rather than a
complete noun phrase. P7 has written she, clearly intending the reference to be to
cinderella. However, the structure as written forces she to refer to [the threestepsister and cinderella and the godmother]. Even if the segment [the threestepsister and cinderella] is ignored as an incomplete sentence, and the sentence re-
started with [and the godmother], she must then refer to the godmother, not
cinderella. As with the above text (P3), knowledge of the story is required to resolve
this reference.
Discussion
This study sought to characterize the written narrative skill of adults with a history
of Language Impairment. Ten adults with a history of Language Impairment and 30
participants with no history of Language Impairment were asked to write the story
of Cinderella. It had been established that this group of Language Impaired adults
demonstrated problems with syntax, spelling and punctuation (Smith-Lock &
Nickels, 2005, 2006; Smith-Lock et al., this issue), but features of textual structure,
organization, and coherence had yet to be explored.
Stories were analyzed with respect to generic structure and cohesion. Group
analyses found no significant differences between the groups in generic structure or
cohesiveness of the stories. Individual analyses of generic structure indicated that a
greater proportion of the Language Impaired participants omitted obligatory and/or
optional elements from the story than participants in the comparison group.
Individual analyses of cohesion indicated that Language Impaired participants were
no less likely to use cohesive chains in their narratives than participants in the
comparison group, nor to differ in the relative proportions of grammatical and
lexical chains, or lexical chain type. Errors of generic structure occurred in both
groups. Errors of grammatical reference, however, only occurred in the Language
Impaired group.
The most notable finding was that, on average, both generic structure and
cohesiveness were relative strengths for the Language Impaired group. Individual
variation existed, but rarely fell outside the range of the comparison group’s
performance (only for one participant in the case of the use of repetition). This
strength of generic structure and cohesion was remarkable when taken in the context
of the sometimes substantial clause level grammatical difficulties demonstrated by
some Language Impaired participants. P7, for example, included all obligatory
elements and did not differ in the proportion of tokens in the narrative involved in
cohesive chains, in spite of severe grammatical difficulties.
746 L. Mortensen et al.
123
One point of difference did arise. Although errors of reference occurred
infrequently, all such errors were made by Language Impaired participants.
Interestingly, two of the three errors involved pronominal reference, a grammatical
technique linking phrases within and between clauses. Thus, the Language Impaired
participants’ errors of cohesion were of a grammatical nature, consistent with the
grammatical deficits shown in other areas (cf. Smith-Lock et al., this issue).
Nevertheless, for the most part, use of pronouns did not present consistent
difficulties within the Language Impaired group. In this case the findings are similar
to those reported by Gregg and Hoy (1990) with respect to the effective use of
pronouns for referencing in essays written by young adults with Learning
Difficulties.
The lack of group differences in generic structure and cohesion suggests that
narrative structure is a strength for the Language Impaired participants. Of course, it
is possible that the task itself i.e., the story of Cinderella, which was chosen because
of its cultural relevance and familiarity, may have been too simple or well rehearsed
to challenge the writers’ strengths and deficits. A number of researchers suggest that
the type of stimulus used may have a significant effect on the complexity of stories
produced by children (e.g., Liles, Duffy, Merritt, & Purcell, 1995; Pearce, 2003).
The same consideration also applies to adults. Furthermore, the framework
established to analyse text structure was relatively simple, and required minimum
information, in short, the ‘bare bones’ of the story. It is possible that different results
would be obtained if a more substantial degree of information were required. For
example, Reed et al. (2007) examined spoken narratives of younger and older
adolescents with specific language impairment for their informativeness, using the
categories of relevance, irrelevance and inaccuracy. Their findings suggested that
adolescent speakers with specific language impairment tended to give more vague
responses in their narratives in comparison to normally developing adolescents. In
the analysis adopted for this study, both explicit (accurate) and implicit (vague)
expression of structural elements were accepted. A more stringent requirement for
explicit information may have provided different outcomes.
Secondly, from a functional linguistic perspective, the choice of narrative over a
different style of text may have affected the results. Different functional text types
place different cognitive and linguistic demands on the writer (Halliday & Hasan,
1985). The purpose of a narrative is to entertain and to teach cultural values (Butt,
Fahey, Feez, Spinks, & Yallop, 2000). A narrative constructs a pattern of events
with a problematic and/or unexpected outcome, and is structured for the most part in
a temporally sequenced manner, using mostly additive, temporal and some causal
connectives to develop the story. Expositions, on the other hand, aim to argue and
persuade and are consequently organised in a rhetorical manner, relying more
heavily on causal and additive connectives to construct the argument (Eggins, 1994;
Martin, 1985). While narratives may contain verbs of action, expositions are likely
to employ verbs of thinking and feeling, together with attitudinal vocabulary and
modality to indicate the writer’s attitudes. Hence a more varied sampling of data to
include texts types that elicit varying cognitive and linguistic abilities would
provide further insights into the relative strengths and deficits of adults with
Language Impairment.
Narrative texts in language impairment 747
123
The apparently limited range of lexical variety demonstrated by both groups in
the small numbers of cohesive chains for synonomy, antonomy, and super/
subordinates could be a consequence of the choice of text, i.e., the Cinderella
narrative which may have dictated word selection and constrained lexical variety. A
different text type or more spontaneous narrative may provide a greater potential to
elicit a wide range of word choice and patterns of cohesion. The Cinderella story,
nevertheless, had the strength that it was able to achieve consistency across the
narratives. In so far as can be controlled, each individual was writing with the same
target story in mind, allowing direct comparisons across groups.
Future research should examine other text types to determine if Language
Impaired participants’ strength in text organization in the writing of a familiar fairy
tale carries over to more challenging situations such expository text. Comparison of
the written texts to parallel oral samples should also be carried out to see if similar
patterns of strength emerge in the oral narratives of Language Impaired individuals.
In conclusion, patterns of text organization at the level of generic structure and
cohesion did not distinguish the performance of Language Impaired writers from the
non-impaired participants. Both groups showed variability across individuals, with
the Language Impaired narratives more likely to contain errors or a restricted
variety of lexical cohesion. It is suggested generic structure and cohesion are a
relative strength in the writing of adults with Language Impairment.
Acknowledgements This research was supported by an Australian Research Council small grant and a
Macquarie University Research Development Grant. Lyndsey Nickels was supported by an NHMRC
Senior Research Fellowship. The authors wish to thank the participants for their time and cooperation.
Appendixes
Table 4 Obligatory and optional elements of the Cinderella tale drawn from the classic Perrault version
(Ehrlich, 1986)
Element Type # Detail
Placement ‘Person particularisation’ e.g. Once upon a time there was a girl, &/ornaming of Cinderella, together with additional orientation information
Introduction 1 Cinderella lives with a step-mother and 2 step-sisters who treat herbadly
Complication: 2 A (royal) ball is held
3 Cinderella attends with the magic help of her fairy godmother
4 The magic ends at midnight
5 Cinderella meets and dances with the prince
6 Cinderella leaves the ball in haste and leaves behind a glass slipper
Resolution: 7 The Prince searches for the person whose foot fits the slipper
8 The Prince (or Prince’s envoy) finds Cinderella who proves heridentity by fitting into the glass slipper
Conclusion 9 Prince and Cinderella marry/live happily ever after
Coda E.g. Stepmother & stepsisters are unhappy and must now serve Cinderella
Obligatory elements are in bold and numbered
748 L. Mortensen et al.
123
Table 5 Example 1 depicts Language Impaired group participant P7’s text, which included all obliga-
tory elements of the narrative despite significant problems with lexicogrammar
GS Element # Text
Placement
Introduction 1 The three stepsister and cinderella.
Complication 3 And the godmother come out and made a whisper
2 and she go to the dance
4 and the godmother said you get at home 12.00 and not home then you
would dot in manget made a dress and cart for cinderella.
5 And she go to dance and saw a prince and have dance with him. And
the clock curn on 12.00
6 and Cinderella ran out side ane down stair. And cinderella shoe come
off and the prince got it the next day.
Resolution 7 He walk around street find someone fit in the shoe. Him have
Cinderella
8 he feet fit the shoe.
Conclusion 9 And Cinderella and prince them get marrege.
Coda
Table 6 Example 2 depicts Language Impaired group participant P6’s text which omitted 2 obligatory
elements (3 & 8), but included one optional element—Placement
GS Element # Text
Placement ONCE UPON A TIME THERE LIVED A GIRL CALLED
CINDERELA.
Introduction 1 SHE LIVED WITH HER TWO SISTERS IN A QUIT VILLAGE.
THE SISTERS WERE HARD ON CINDERALA, THEY MADE
HER DO ALL THE HOUSE WORK, COOKING AND NEVER
LET HER OUT.
Complication 2 THERE WAS A BALL COMING UP, CIND-ELA HAD ONLY ONE
WISH, THAT WAS TOO GO.
4 SHE GOT HER WISH AS LONG AS SHE WAS HOME BY
12.00AM.
5 AT THE BALL SHE WAS HAVING SO MUCH FUN, DANCING
WITH A PRINCE SHE LOST TRACK OF TIME,
6 AND RAN OUT SO SHE COULD BE HOME ON TIME. IN THE
RUSH SHE LOST HER SHE.
Resolution 7 THE PRINCE FOUND THIS SHOE, SEARCHED EVERY WHERE
FOR CINDERELA, RUNNING ALROUND THE VILLAGE.
Conclusion 9 EVENTUALLY HE FOUND HER, TAKING HER AWAY FROM
HER SISTERS AND LIVING HAPPILY EVERYAFTER.
Spelling punctuation and capitalisation as written
Narrative texts in language impairment 749
123
Table 7 Example 3 presents comparison group participant C26’s text where one obligatory structural
element is omitted (4), and one element (3*) is both incomplete and out of sequence
GS Element # Text
Placement
Introduction 1 There was a young girl with three evil sisters. She was constantly made aware of
her inferiority, and led a very unhappy life.
Complication 3* One day she met a fairy God-mother who told her that she would be able to go to
2 The prince caught a glimpse of Cinderella as he was walking down the street and
invited her to the grand ball.
Cinderella knew her sisters would not allow her to attend, but tried nethertheless
to go.
3 She met the fairy godmother who granted her her wish that she may once again be
able to meet the prince. She waved her wand and Cinderella was given a
beautiful dress with glass slippers. She attended the ball in secret, but her sisters
found out and proceeded directly to stop her.
5 In the middle of her dance with the prince the sisters proclaimed that she did not
belong (illegible) made the entire attendance aware of her status.
6 Overcome by emotion, Cinderella ran out of the room, leaving behind one Golden
slipper—she did not know the prince didn’t care about her status.
Resolution 7 Having only the slipper to work with the prince embarked on a journey to find his
princess.
8 After many trials (illegible) the way, he eventually found the princess and asked her to
marry him, and the 3 sisters were punished.
Conclusion 9 Cinderella ? the prince lived happily ever after.
Coda
Table 8 Examples of the different lexical categories selected by the Language Impaired and comparison
groups
Language Impaired participants Comparison participants
Synonyms: Synonyms:
P1: excited/thrilled C2: create/transform
P2: home/house C3: pretty/beautiful; announce/tell
P5: ball/dance C4: cruel/horrible
P6: notice /saw; tell/say C6: symbolic/represent; chores/work;rush/hurry
P10 shoe/slipper
Antonyms: Antonyms:
P1: born/die; miserable/thrilled C3: beautiful/ugly
P2: clean/dirty C4: live/die; ask/tell
P6: lose/find C6: enable/prevent
P10 come/go C11: young/old
Super/subordinates: Super/subordinates:
P5 women/ugly sisters; sisters/hair C6: parents: mother/father; menial tasks/clean,
cook, wash, mend, sew; animals/birds, mice
P6 housework/cook clean C9: house/cellar
750 L. Mortensen et al.
123
References
Bamberg, M. (1987). The acquisition of narratives. New York: Mouton de Gruyter.
Bishop, D. V. M., & Clarkson, B. (2003). Written language as a window into residual language deficits: A
study of children with persistent and residual speech and language impairments. Cortex, 39, 215–
237.
Butt, D., Fahey, R., Feez, S., Spinks, S., & Yallop, C. (2000). Using functional grammar: An explorer’sguide (2nd ed.). Sydney: National Centre for English Language Teaching and Research, Macquarie
University.
Christie, F., & Martin, J. R. (Eds.). (1997). Genre and institutions: social processes in the workplace andschool. York: Cassell.
Crawford, J. R., & Garthwaite, P. H. (2002). Investigation of the single case in neuropsychology:
Confidence limits on the abnormality of test scores and test score differences. Neuropsychologia, 40,
1196–1208.
Derewianka, B. (Ed.). (1992). Language assessment in primary classrooms. Sydney: Harcourt Brace
Jovanovich.
Eggins, S. (1994). An introduction to systemic functional linguistics. London: Pinter.
Ehrlich, A. (1986). Cinderella/by Charles Perrault; retold by Amy Ehrlich. London: Hamish Hamilton.
Englert, C., Raphael, T., Fear, K., & Anderson, L. (1988). Students’ metacognitive knowledge about how
to write informational texts. Learning Disabilities Quarterly, 1I(1), 18–46.
Englert, C., & Thomas, C. (1987). Sensitivity to text structure in reading and writing: A comparison
between learning disabled and non-learning disabled students. Learning Disabilities Quarterly,10(2), 93–105.
Freedman-Stern, R. B., Ulatowska, H. K., Baker, T., & DeLacoste, C. (1984). Disruption of written
language in aphasia: A case study. Brain and Language, 22, 181–205.
Gathercole, S. E., Willis, C., Baddeley, A. D., & Emslie, H. (1994). The children’s test of nonword
repetition: A test of phonological working memory. Memory, 2, 103–127.
Gillam, R. B., & Johnston, J. R. (1992). Spoken and written language relationships in language/learning-
impaired and normally achieving school-age children. Journal of Speech and Hearing Research, 35,
1303–1315.
Gregg, N. (1991). Disorders of written expression. In A. M. Bain, L. L. Bailet & L. Moats (Eds.), Writtenlanguage disorders: Theory into practice (pp. 65–97). Austin Texas: Pro-ed.
Gregg, N., & Hoy, C. (1990). Referencing: The cohesive use of pronouns in the written narratives of
college underprepared writers, nondisabled writers and writers with learning disabilities. Journal ofLearning Disabilities, 23, 557–563.
Halliday, M. A. K., & Hasan, R. (1976). Cohesion in English. London: Longmans.
Halliday, M. A. K., & Hasan, R. (1985). Language context and text: Aspects of language in a social-semiotic perspective. Geelong, Australia: Deakin University Press.
Hammill, D., Brown, G. D. A., Larsen, S. C., & Wiederholt, J. L. (1993). Test of adolescent and adultlanguage-3. Austin Texas: Pro-Ed.
Hammond, J., Burns, A., Joyce, H., Brosnan, D., & Gerot, L. (1992). English for social purposes. Sydney:
National Centre for English Language Teaching and Research, Macquarie University.
Hasan, R. (1996). The nursery tale as genre. In D. Butt, C. Cloran & G. Williams (Eds.), Ways of saying:Ways of meaning. Selected papers of Ruqaiya Hasan (pp. 51–72). London: Cassell.
Table 8 continued
Language Impaired participants Comparison participants
Collocation: Collocation:
P1: ashes/fireplace; wave/wand; broken spell C2 pair/slippers
P2: wash/clean C3 win/prize; compete/win
P3: clothes/wear C4: celebrate/birthday
P5: handsome/prince C6: hum/music
P7: shoe/fit
Narrative texts in language impairment 751
123
Ivanic, R. (1998). Writing and identity. Amsterdam: John Benjamins.
Kaplan, E., Goodglass, H., & Weintraub, S. (1983). Boston naming test. Philadelphia, PA: Lea & Febiger.
Liles, B., Duffy, R., Merritt, D., & Purcell, S. (1995). Measurement of narrative discourse in children with
language disorders. Journal of Speech and Hearing Research, 38, 415–425.
Martin, J. (1985). Factual writing. Geelong, Australia: Deakin University Press.
Mortensen, L. (2004a). Perspectives on functional writing following acquired brain Impairment.
Advances in Speech Language Pathology, 6(1), 15–22.
Mortensen, L. (2004b). Assessing the functional writing abilities of people with aphasia. ACQuiringknowledge in speech, language and hearing, 6(2), 75–79.
Pearce, W. M. (2003). Does the choice of stimulus affect the complexity of children’s oral narratives?
Advances in Speech Language Pathology, 5(2), 95–103.
Plante, E., Shenkman, K., & Clark, M. (1996). Classification of adults for family studies of developmental
language disorders. Journal of Speech, Language, and Hearing Research, 39, 661–667.
Reed, V. A., Patchell, F. C., Coggins, T. E., & Hand, L. S. (2007). Informativeness of the spoken
narratives of younger and older adolescents with specific language impairment and their
counterparts with normal language. Clinical Linguistics and Phonetics, 21, 953–960.
Rice, M. L., Haney, K. R., & Wexler, K. (1998). Family histories of children with specific language
impairment who show extended optimal infinitives. Journal of Speech, Language, and HearingResearch, 41, 419–432.
Scott, C. M., & Windsor, J. (2000). General language performance measures in spoken and written
narrative and expository discourse of school-age children with language learning disabilities.
Journal of Speech Language and Hearing Research, 43, 324–339.
Smith-Lock, K. M., & Nickels, L. A. (2005). Writing skills of adults with a history of specific-Language-
Impairment. Acquiring Knowledge in Speech Language and Hearing, 7, 78–81.
Smith-Lock, K. M., & Nickels, L. (2006). Does SLI in childhood lead to writing difficulties in adulthood?
Speech Pathology Australia 2006 National Conference, Fremantle.Smith-Lock, K. M., & Nickels, L., & Mortensen, L. (this issue). Story writing skills of adults with a
history of Language-Impairment. Reading and Writing. An Interdisciplinary Journal. doi:
10.1007/s11145-008-9138-6.
Stein, N. L., & Glenn, C. G. (1979). An analysis of story comprehension in elementary school children. In
R. O. Freedie (Ed.), New direction in discourse processing (pp. 53–119). Norwood, NJ: Ablex.
Tallal, P., Hirsch, L. S., Realpe-Bonilla, T., Miller, S., Brzustowicz, L. M., Bartlett, C., et al. (2001).
Familial aggregation in specific language impairment. Journal of Speech, Language, and HearingResearch, 44, 1172–1182.
van Dijk, T. A. (1981). Episodes as units of discourse analysis. In D. Tannen (Ed.), Analyzing discourse:Text and Talk (pp. 177–195). Washington, PC: Georgetown University Press.
Wilson, B. M., & Proctor, A. (2002). Written discourse of adolescents with closed head injury. BrainInjury, 16, 1011–1024.
Wong, B. (2000). Writing strategies instruction for expository essays for adolescents with and without
leaning disabilities. Topics in Language Disorders, 20(4), 29–44.
Wong, B., Wong, R., & Blenkisop, J. (1989). Cognitive and metacognitive aspects of composing
problems in learning-disabled adolescents. Learning Disabilities Quarterly, 12, 300–322.
752 L. Mortensen et al.
123