The present study examined narrative development in children and adolescents with Down syndrome longitudinally. Narratives were collected from 32 children and adolescents with Down syndrome three times over a 1-year period. Both micro- and macrolevel analyses were conducted. Significant growth over the 1-year period was seen in semantic complexity and narrative structure. However, there was no evidence of growth in syntactic complexity or narrative length. Mental age and comprehension skills at Time 1 predicted scores in all 4 areas at Time 3. Expressive language skills added further to the prediction of syntactic complexity and story length.
Down syndrome is the most common chromosomal cause of intellectual disability (Patterson & Lott, 2008). Although much is known about the vocabulary and syntactic skills of children and adolescents with Down syndrome, relatively little is known about their discourse-level skills. It is important to understand the development of discourse skills in order to have a more complete picture of language development in individuals with Down syndrome. A discourse genre that has received considerable attention in recent years is narrative. The study reported on here was a longitudinal investigation of narrative skill development in children and adolescents with Down syndrome. Micro- and macroanalyses were used to examine narrative development, and regression analyses were conducted to explore the contributions of cognitive and linguistic abilities to narrative skills.
Narratives, as described by Peterson (1990), are the recounting of past events that preserve chronological order. The production of narratives requires an integration of linguistic, cognitive, and social abilities (Boudreau & Chapman, 2000). Narratives play an integral role in successful communicative exchanges (Boudreau & Chapman, 2000), and the development of narrative skill is linked to literacy development and academic achievement in typically developing children (Dickinson & Tabors, 2001). They are an important aspect of development during the late preschool and school-age years.
Narratives provide a rich source of information about an individual's language skills. Narratives can be analyzed at two levels, microstructure and macrostructure. Microstructures refer to the syntactic or semantic complexity of narratives. Specifically, measures such as mean length of utterance (MLU), number of clauses per utterance, as well as total number of words and different words are used to evaluate narrative skill at this level. In contrast, macrostructure refers to the general coherence and organization of the narrative, which can be evaluated using measures of episodic (plot) structure or content organization of the narratives (Hughes, MacGillivray, & Schmidek, 1997).
Language Development in Individuals with Down Syndrome
Although much is known about the oral language development of individuals with Down syndrome, the majority of the research on children and adolescents with Down syndrome has focused on their language abilities in conversation (see Chapman & Kay-Raining Bird, 2011). This research has revealed a typical profile, although there is substantial individual variability. Children and adolescents with Down syndrome show a dissociation between linguistic and cognitive impairments such that their language difficulties, particularly deficits in expressive language, are more severe than their nonverbal cognitive skills would predict (Chapman, 2006). Children with Down syndrome display stronger receptive skills than expressive skills (e.g., Chapman, Schwartz, & Kay-Raining Bird, 1991). In addition, they display relative strengths in vocabulary development whereas morphosyntactic development is an area of particular weakness. Nevertheless, growth in morphosyntactic skills can be seen throughout adolescence and into early adulthood (Chapman, Seung, Schwartz, & Kay-Raining Bird, 1998; Elin Thordardottir, Chapman & Wagner, 2002).
Narrative Development in Individuals with Down Syndrome
There has been some research done on the narrative skills of children and adolescents with Down syndrome. Both micro- and macrolevel analyses have been used. Most studies have compared the performance of children and adolescents with Down syndrome to that of children with typical development matched on a variety of dimensions. At a microlevel, in comparison to peers matched for mental age (MA), children and adolescents with Down syndrome produced narratives with shorter MLUs, fewer words, and fewer different words (Boudreau & Chapman, 2000; Chapman et al., 1998; Miles, Chapman, & Sindberg, 2006). Similarly, Keller-Bell and Abbeduto (2007) reported that children and adolescents with Down syndrome produced shorter MLUs than an MA-matched group, but the difference did not reach statistical significance. However, these researchers did find evidence of language form weaknesses in that the individuals with Down syndrome produced a lower percentage of grammatically acceptable utterances. They also reported no difference in number of utterances or number of different words. Using reading-matched controls, Kay-Raining Bird et al. (2008) found that children and adolescents with Down syndrome produced oral narratives with more utterances but a comparable number of words. In addition, no significant differences were found in MLU (although the mean MLU was lower for the group of children and adolescents with Down syndrome) or number of different words (Kay-Raining Bird, Cleave, White, Pike, & Helmkay, 2008). In comparison to typically developing children with similar receptive language skills, children and adolescents with Down syndrome have been found to produce narratives with lower MLUs and poorer cohesion but a similar number of different words and total words (Boudreau & Chapman, 2000; Miles & Chapman, 2002). Comparisons have also been made with MLU-matched groups. These have revealed no differences in number of words or number of different words (Boudreau & Chapman, 2000; Miles & Chapman, 2002).
Macrolevel analyses have been reported in a number of studies. Kay-Raining Bird, Chapman, and Schwartz (2004) reported that children and adolescents with Down syndrome produced less story information than an MA-matched comparison group when minimal visual support was provided. In contrast, when more visual support was provided, this weakness in the amount of story information produced has not been found (Boudreau & Chapman, 2000; Keller-Bell & Abbeduto, 2007; Miles & Chapman, 2002). Similarly, children and adolescents with Down syndrome have been found to display similar macrolevel skills in comparison to children with similar receptive language skills (Boudreau & Chapman, 2000; Miles & Chapman, 2002). However, when compared with MLU-matched groups, children and adolescents with Down syndrome produced narratives with more story information (Boudreau & Chapman, 2000; Miles & Chapman, 2002).
In addition to group comparisons, some studies have explored predictors of narrative skills. The best predictors of MLU in narratives were age, auditory memory, and visual memory (Chapman, Hesketh, & Kistler, 2002). Mental age and expressive and receptive language skills have been found to be predictors of the amount of story information included (Boudreau & Chapman, 2000; Kay-Raining Bird et al., 2004; Miles & Chapman, 2002).
For the most part, studies of narrative skills in Down syndrome have not explored developmental change. The one area that has been explored developmentally is syntactic skills. Growth in these skills has been seen using cross-sectional (Chapman et al., 1998) and longitudinal (Chapman et al., 2002) data. Both of these studies involved a very large age range; in the cross-sectional study, the children and adolescents spanned an age range of 15 years; and in the longitudinal study, the children and adolescents were followed for 6 years.
In summary, a typical pattern has emerged from the group comparison studies. When narratives are elicited with visual support, individuals with Down syndrome in comparison to MA-matched controls show deficits in their narratives at the microlevel of analysis while they perform similarly to their peers matched on expressive language. At a macrolevel of analysis, children and adolescents with Down syndrome have similar skills when compared with MA- or receptive language–matched controls and better skills when compared with expressive language–matched controls (e.g., Boudreau & Chapman, 2000; Chapman et al., 1998; Miles & Chapman, 2002).
Although there are several studies that have examined narrative skills in children and adolescents with Down syndrome, none of these studies has taken a comprehensive look at how these narrative skills change over time. The purpose of the current study was to examine both micro- and macrolevel changes that occur in narrative development of school-age children and adolescents with Down syndrome over a 1-year period and to examine whether cognitive and linguistic abilities predict narrative skills.
Thirty-two children and adolescents with Down syndrome ranging in age from 5;10 (years;months) to 16;6 participated. The children and adolescents were recruited from public schools across Nova Scotia, Canada, where they were integrated to varying degrees into regular classrooms in their neighborhood schools. To be included in the study, the children and adolescents had to be native English speakers and display a nonverbal MA of at least 3 years. Table 1 presents the participants' age and cognitive and language skills at the onset of the study. The study was approved by the research ethics board of Dalhousie University in Halifax, Nova Scotia, and by the research ethics committees of each participating school board.
Data collection occurred over 3 years, but each participant was followed for 1 year. Children and adolescents received initial testing in November (T1). Additional testing occurred twice, once in May (T2) and once in November (T3) of the following year. The children and adolescents were individually tested in their school by a speech-language pathologist. T1 testing took approximately 90 min, and T2 and T3 testing took approximately 60 min.
Measures of cognitive and language skills were administered. Nonverbal cognitive skills were measured with the Pattern Analysis and Bead Memory subtests of the Stanford-Binet Intelligence Scale (4th edition; Thorndike, Hagen, & Sattler, 1986) or the Columbia Mental Maturity Scale (Burgemeister, Blum, & Lorge, 1972). The Oral and Written Language Scales (OWLS; Carrow-Woolfolk, 1995) were used to measure expressive (Oral Expression subtest; OE) and receptive (Listening Comprehension subtest; LC) oral language. The scores from the subtests were combined to obtain an overall language score.
Two narrative elicitation procedures, both with visual support, were used to assess narrative abilities. The Bus Story (Renfrew, 1991), a standardized story retelling task, was administered according to the test instructions. The examiner read a predetermined story to the participant as they both looked at 12 sequenced pictures depicting the story. The participant was then asked to retell that story with support from the pictures. The second narrative task involved a single episode from Frog on His Own by Mayer (1973). This single episode contained four picture plates. The participant was instructed to examine all the pictures and then to make up a story about them. To ensure that the participant looked at all the pictures before beginning, the examiner pointed to each picture in sequence. Nonspecific prompts were used for story continuation and completion. Both narratives were audiotaped for later transcription and analysis.
The narrative samples were transcribed using the Systematic Analysis of Language Transcripts (Miller & Chapman, 2002) conventions. All narratives were segmented into t-units, defined as an independent clause and all its modifiers. The Bus Story (Renfrew, 1991) was scored according to the test instructions. At the microlevel, this included two measures of syntactic complexity: the MLU of the five longest t-units (sentence score) and the average number of clauses per t-unit (clause score). At the macrolevel, this included the information score, which represents how many key content elements were included. In addition to the standard scoring, two other microlevel analyses were completed. As a measure of narrative length, the total number of words used in the Bus Story was calculated, and as a measure of semantic complexity, the total number of different words used was determined. The total number and number of different words were computed only for the Bus Story as it was of greater length.
In addition, both the Bus Story and the Frog Story were scored using a story structure analysis system developed by A. McKeough (personal communication, July 2007; see Appendix). This analysis was composed of questions designed primarily to evaluate several components of the macrostructure of the narrative. Specifically, the researchers answered 15 yes/no questions that queried such things as the identification of the characters, the type of connectives used, or the presence of a problem. The number of events and mental state verbs were also tallied. For all analyses, the researchers were blind as to the time of testing.
One trained research assistant transcribed the samples, and then a second checked the transcriptions. Consensus between the two transcribers was used to resolve any disagreements. Reliability was established for the measures that were not calculated by computer. Two researchers independently scored the narratives, and agreement on scoring was calculated for five narratives at all three times of testing (16% of the sample). For the story structure analysis, the percent agreement was found to be 97.2% for the Frog Story and 98.5% for the Bus Story. The percent agreement for the Bus Story information score was 97.8%.
Growth Over Time
Repeated measures analyses of variance with three levels of time were conducted for both micro- and macrolevel measures. Significant results were explored further using paired t-test planned comparisons between T1 and T2 and between T1 and T3. Significance levels were set at .025 for the follow-up tests, using a Bonferroni correction. As there were a priori predictions of growth, one-tailed tests were used. Effect sizes (partial η2) were determined for all significant results.
Descriptive statistics for dependent measures of syntactic complexity, narrative length, and semantic complexity obtained from the Bus Story are presented in Table 2. Two dependent measures of syntactic complexity were used: the sentence score (MLU for the 5 longest t-units) and the clause score (average number of clauses per t-unit). Analyses revealed no significant effects of time for the sentence score, F(2, 30) = 0.092, p = .46; or the clause score, F(2, 30) = 1.25, p = .11. There was also no change in the measure of narrative length, the total number of words, F(2, 30) = 0.973, p = .20. In contrast to the other analyses, a significant effect of time was found for the semantic measure, the number of different words, F(2, 30) = 4.20, p = .012; η2 = .218. Post hoc analyses showed a significant difference between T1 and T2, t(31) = 2.06, p = .023; η2 = .121; and between T1 and T3, t(31) = 2.93, p = .003; η2 = .218. Furthermore, 22 of 32 (69%) of the participants made a gain between T1 and T3 on the semantic measure.
The episodic structure of the participant's narratives was measured using the story structure analysis for both the Frog Story and the Bus Story. Specifically, the dependent measures included were the story structure score, the number of events reported, the number of characters identified, and the number of mental states. In addition, the Bus Story information score was analyzed (see Table 2). A significant effect of time was found for the story structure score calculated for the Bus Story, F(2, 30) = 4.19, p = .012; η2 = .218. Post hoc analyses showed that the difference between T1 and T2 was not significant, t(31) = .31, p = .378, but that the difference between T1 and T3 was t(31) = 2.44, p = .010; η2 = .162. A significant effect of time was also found for the Frog Story structure score, F(2, 30) = 3.52, p = .021; η2 = .190. Post hoc analyses showed significant differences between T1 and T2, t(31) = 1.96, p = .029; η2 = 1.10; and between T1 and T3, t(31) = 2.63, p = .006; η2 = .182. On the Bus Story 19 of 32 (59%) of participants made a gain between T1 and T3, whereas 18 of 32 (56%) did so on the Frog Story.
A significant effect of time also was found for the number of characters included for both narratives: Bus Story, F(2, 30) = 5.12, p = .006; η2 = .254; Frog Story, F(2, 30) = 2.45, p = .05; η2 = .141. Post hoc analysis of characters in the Bus Story showed that the difference between T1 and T2 was statistically significant, t(31) = 2.52, p = . 008; η2 = .170; as was the difference between T1 and T3, t(31) = 2.83, p = .004; η2 = .205. Between T1 and T3, 16 of 32 (50%) of the participants made a gain. For the Frog Story, only the difference between T1 and T2 was significant, t(31) = 2.25, p = . 016; η2 = .140. There were no significant changes in the number of events reported (ps > .115) or in the number of mental state words used (ps > .472).
A significant effect of time was found for the Bus Story information score, F(2, 30) = 6.56, p = .002; η2 = .304. Post hoc analysis showed that the difference between T1 and T2 did not reach significance, t(31) = 1.78, p = .042; but the difference between T1 and T3 was statistically significant, t(31) = 3.68, p < .001; η2 = .303. For this measure, 24 of 32 (75%) of the participants made a gain between T1 and T3.
Predictors of Narrative Skills
Hierarchical multiple regression analyses were used to determine the role of cognitive and linguistic measures from T1 in predicting narrative micro- and macrostructure 1 year later at T3. Predictor variables were MA and receptive and expressive language raw scores from the OWLS. Given the range of developmental levels in the sample, MA was entered first as a control for general development. As reviewed above, receptive language is an area of strength for children with Down syndrome relative to expressive language. Thus, receptive language (OWLS-LC) was entered second and expressive language (OWLS-OE) last to determine whether expressive skills made a unique contribution after the contribution of receptive skills was removed. As OWLS scores were not available for all participants, the number of participants for these analyses was 25 (see Table 3).
To limit the number of analyses, a single measure was chosen for the syntactic and macrolevel analyses. For the syntactic analysis, the MLU in the longest five utterances of the Bus Story was chosen because there was a floor effect on the clause measure. As all the macrolevel measures were highly correlated, one was chosen. The total score for the story structure analysis of the Bus Story was chosen as it included both story content and narrative structure information. Total number of words was used as the measure of narrative length and number of different words as the measure of semantic complexity.
For syntactic complexity, as indexed by the MLU in the five longest utterances, MA accounted for 14% of the variance, whereas OWLS-LC accounted for an additional 20% of the variance, and OWLS-OE accounted for an additional 19% of the variance. All three measures were also significant in predicting the total number of words. MA accounted for 17% of the variance, OWLS-LC accounted for an additional 33% of the variance, and OWLS-OE accounted for an additional 11% of the variance. For the semantic complexity measure, total number of different words, MA accounted for 16% of the variance, and OWLS-LC accounted for an additional 31% of the variance. Expressive language skills did not add significantly to the model. Finally, for the story structure measure, MA accounted for 19% of the variance, and OWLS-LC accounted for an additional 20% of the variance. Again, expressive language skills did not contribute significantly to the model after accounting for the contribution of mental age and receptive language skills.
The present study examined the narratives of children and adolescents with Down syndrome longitudinally, exploring both micro- and macrolevel changes over a 1-year period. In addition, the contribution of cognitive and language skills to narrative abilities was explored. To our knowledge, this is the first longitudinal study to examine both levels of narrative ability.
Micro- and macrolevel analyses revealed growth over a 1-year period. At the microlevel, there was an increase in the semantic diversity (number of different words produced) of the narratives over time, in the absence of an increase in their syntactic complexity (MLU in the longest five t-units; average number of clauses per t-unit) or length (total number of words produced). The finding of improvement in semantic diversity in the absence of improvement in syntactic complexity has been seen in past research. Semantic ability is generally thought to be an area of strength in individuals with Down syndrome, whereas syntax has been found to be particularly problematic (e.g., Chapman & Kay-Raining Bird, 2011). Studies specifically comparing the narrative skills of individuals with Down syndrome with those of typically developing children have revealed this pattern. Semantic skills, as measured by number of different words, have been found to be an area of strength, as shown by the fact that Down syndrome groups do not differ from MA-matched groups (Keller-Bell & Abbeduto, 2007; Miles & Chapman, 2002). However, this has not been found in all studies. In studies by Chapman et al. (1998) and Boudreau and Chapman (2000), the Down syndrome group displayed less semantic diversity than the MA-matched group. In these studies, the method of elicitation was different than in studies that have found comparable skills. Chapman et al. (1998) used a variety of methods to elicit the narratives, including description of favorite movies, personal stories, and story stems, and Boudreau and Chapman (2000) had their participants tell a story after watching a wordless video. Thus, in both studies, there was less visual support provided during narration than in studies in which children and adolescents were able to view the pictures while telling their narrative, such as in the current study. Miles et al. (2006) reported a benefit of visual support for children and adolescents with Down syndrome in their study, which examined the effect of elicitation techniques on the MLU of narratives. The current study's finding of progress in expressive semantic diversity in the absence of progress in narrative length or syntactic complexity suggests that individuals with Down syndrome bring a level of conceptual abilities to narrative production that is higher than their skill for formulating utterances. Furthermore, it appears that visual support is necessary for children and adolescents with Down syndrome to display this semantic strength.
At the macrolevel, an increase was seen in several aspects of episodic structure. First, the number of characters listed in the Bus Story but not the Frog Story showed significant growth over time. The difference in findings across the Bus and Frog Stories is likely due to the fact that there was more room for growth in the Bus Story. There were five characters in the Bus Story and only three in the Frog Story. There were also improvements seen in the story structure analysis for both the Frog Story and the Bus Story. Recall that this measure evaluates several aspects of narrative macrostructure, such as whether the story shows temporal or causal organization and whether mental states are used. Improvement was also seen in the information score from the Bus Story. Changes in the macrostructure measures reflect an improvement in the structure and content of the information conveyed by the children and adolescents; they told more organized, complete, and coherent stories after 1 year.
The majority of studies on narrative skills of children and adolescents with Down syndrome have involved a single testing period; however, within these studies, some authors separated the participants into different age groups to explore development cross-sectionally (Chapman et al., 1998; Miles & Chapman, 2002). These studies found that there was a change in development from the youngest to the oldest participants with Down syndrome in the content conveyed in a narrative (Boudreau & Chapman, 2000; Miles & Chapman, 2002). This increase has been hypothesized to occur as a result of a greater level of world knowledge that developmentally older children and adolescents incorporate into the task (Boudreau & Chapman, 2000). Our longitudinal study revealed a similar change in narrative abilities over a 1-year period.
As micro- and macrostructural changes were apparent after 1 year, it can be assumed that with a greater length of study time, further progress would be made. One could also expect changes in measures that reflected no growth over the single year of this study, such as MLU and story length. Previous cross-sectional studies comparing groups of younger and older children and adolescents with Down syndrome have observed growth in expressive syntactic skills (Chapman et al., 1998; Elin Thordardottir et al., 2002).
At both the micro- and macrolevels of analysis, the scores from the OWLS Oral Expression and Listening Comprehension subtests at Time 1 were predictive of the narrative abilities of these participants with Down syndrome 1 year later. After removing the impact of mental age, as a general developmental measure, language comprehension skills contributed significantly to the prediction for all measures, including syntactic and semantic complexity, story length, and story structure. When expressive language skills were added as the third predictive variable, it was only the prediction of syntactic complexity and story length that were improved. The narrative skills used in these analyses all came from the Bus Story. Recall that in this test, the examiner told the story to the participant and the participant then retold the story. A child's ability to comprehend the examiner's story would be an important factor here. Thus, it is not surprising that language comprehension skills are predictive of performance on this task. A number of cross-sectional studies have found that language comprehension skills concurrently predict both micro- (Chapman et al., 2002; Chapman, Seung, Schwartz, & Kay-Raining Bird, 2000) and macrolevel skills (Kay-Raining Bird et al., 2004; Miles & Chapman, 2002). In the current study, expressive language skills added to the predictive model only for the measures of syntactic complexity (i.e., MLU) and story length (i.e., total words). Both these measures would be constrained by a child's expressive language skills. Taken together, these findings suggest that the macrostructure and semantic diversity of the narratives produced were less constrained by the difficulties that children and adolescents with Down syndrome have with expressive language than were the number of words and sentence structure used.
Educational and Clinical Implications
Understanding the development of narrative skills in children with Down syndrome has important educational and clinical implications given their link to social, literacy, and academic skills (Boudreau & Chapman, 2000; Dickinson & Tabors, 2001). Narratives are used extensively in schools in literature-based literacy instruction (Morrow & Gambrell, 2002). Narrative skills are particularly important for reading comprehension. For example, the National Reading Panel identified teaching story structure as a method used extensively to improve reading comprehension (National Reading Panel, 2000). Narratives also are an important part of programs designed to improve the language skills of children and adolescents with Down syndrome. They offer a rich context in which to address word-, sentence-, and text-level skills. Therefore, understanding the nature of narrative development, including areas of relative strength and weakness and factors that impact narrative development, is important for speech-language pathologists and educators.
The findings from this study provide evidence that semantic (i.e., vocabulary) diversity and narrative structure skills progress over a 1-year period in children and adolescents with Down syndrome. Conversely, there was no significant growth in morphosyntactic skills or story length. In assessment and intervention, it is important to recognize that for children and adolescents with Down syndrome, there are often dissociations among skills. In narratives, their morphology and syntax challenges may mask relative strengths in narrative content and structure. In addition, the cognitive skills of these individuals are stronger than their linguistic skills, and comprehension typically is better than production. As a result, it is inappropriate to assume that the narrative production is a good indicator of narrative comprehension abilities or overall cognitive level in individuals with Down syndrome. Therefore, it is important to include both narrative production and comprehension in assessment and intervention for the Down syndrome population. Assessments should include measures of syntactic and semantic complexity of the narrative as well as measures of the overall coherence and organization of the narrative text. One level of analysis without the other may provide an incomplete picture of narrative production. Furthermore, elicitation methods should include picture support if the individuals with Down syndrome are to demonstrate their best skills.
Limitations of this study must be recognized. Although this study focused on the longitudinal changes in individuals with Down syndrome over time, it took place only over 1 year. To more fully understand the development of narrative skills in children and adolescents with Down syndrome, following the participants for more than 1 year is needed. In addition, the narratives were quite short, particularly the Frog Story narratives. Longer narrative samples would allow more detailed analyses, which may reveal additional areas of growth. Replication of the current findings with other elicitation procedures, narrative analysis measures, and participant samples is critical to getting a more complete picture of narrative development in individuals with Down syndrome. Knowledge of their narrative development as well as how cognitive and linguistic skills play a role in that development will help in providing services to children and adolescents with Down syndrome to maximize their language and literacy skills.
We would like to thank the children and adolescents and their parents as well as the school boards and school staff for their participation in this project. The project was supported by research grants to Patricia Cleave from the Canadian Language and Literacy Network and the Nova Scotia Health Research Foundation. Portions of these data were presented at the annual convention of the American Speech-Language-Hearing Association, Philadelphia, PA, in November 2010.
Editor-in-Charge: Stephanie Dean
Patricia Cleave (e-mail: email@example.com), School of Human Communication Disorders, Dalhousie University, Halifax, Nova Scotia B3H 4R2, Canada; Elizabeth Kay-Raining Bird, Dalhousie University, Halifax, Nova Scotia; Rachael Czutrin, Zareinu Educational Centre of Metropolitan Toronto, Ontario, Canada; Lindsey Smith, Calgary Board of Education, Calgary, Alberta, Canada.