Children’s Story Retelling and Comprehension Performance under Different Task Conditions Marleen Westerveld Massey University, New Zealand This poster presents three experiments investigating the impact of administration variations affecting the amount of linguistic or picture support before or during story retelling to school-age children’s story retelling and comprehension performance. The presence of pictures during retelling was the only variable to influence performance (on measures of verbal productivity and verbal fluency) with no effect of task conditions on any other measures.
Rationale: Oral narrative analysis is considered an ecologically valid way to describe children’s oral language skills. To improve the clinical utility of the narrative assessment process, groups of researchers have developed databases of samples elicited from typically developing speakers to use as a comparison for their clients with suspected language impairment. While the importance of adhering to language sampling protocols has been well established (e.g., Schneider & Dubé, 2005), in clinical practice we may vary our elicitation procedure depending on the purpose of the assessment, the availability of elicitation materials, or our personal preferences. The purpose of this series of experiments was to establish the relative impact of administration variations affecting three aspects of the elicitation process, i.e., 1) type of exposure to the model story, 2) timing of the comprehension questions, and 3) presence or absence of pictures during story retelling. It was anticipated that variations to the amount of support during the exposure to the model story or during retelling of the story would affect children’s performance.
Method: Children, aged 6.0 – 7;11, were recruited through teacher referral. These children were progressing normally at school, had no history of speech therapy intervention, and had no sensory or neurological impairments. Children were randomly divided into two groups. The performance of children who participated in all three experiments was considered for the current study (n = 59). There were no significant differences between the two groups on age (p = .42), or on the PPVT-III score (p = .35). The following three tasks were administered over three consecutive sessions in a two-week period. In all three tasks, the children listened to the novel story on tape, while looking at the pictures of the story. All tasks involved a story comprehension and a story retelling component. Model stories
Exposure
Ana gets lost (AGL)
Book vs PC
Frog goes to dinner (FGTD)
Book
Frog where are you? (FWAY)
Book
Comprehension questions After retelling
Retelling
Before retelling vs after retelling After retelling
No pictures
No pictures
Pictures vs No pictures
Analyses: All samples were transcribed and the following microstructure measures were generated automatically using Systematic Analysis of Language Transcripts, New Zealand version (SALT-NZ; Miller, Gillon, & Westerveld, 2008) : Total number of utterances (UTT), mean length of utterance in words (MLUw), number of different words (NDW), percentage of mazed words (%Mazewords), and number of words per minute (WPM). Grammatical accuracy (GA) was calculated as the percentage of utterances without grammatical errors. Comprehension was scored as the number of questions answered correctly.
Results: Results from a series of analysis of variance equations indicated that the variations in task conditions had no effect on children’s syntactic skills (MLU, GA), verbal fluency, story comprehension, or rate of speech. However, the children told significantly longer stories
(containing more different words), and demonstrated significantly fewer mazes when they were allowed to refer to the pictures during retell; results indicated that the presence of pictures during retelling accounted for approximately 30% of the variance in length and semantic diversity, and 6.7% of the variance on %Mazewords. All other measures were not significantly different and accounted for < 2% of the variability in performance (table 1).
Table 1. Eta squared values summarizing the amount of variance accounted for by 1) presentation mode, 2) timing of comprehension questions, and 3) presence of pictures during retelling. Task AGL
condition Book vs PC FGTD SC first / SC last FWAY pics vs no pics
UTT .019
MLUw .017
NDW .002
%Mzwds .000
WPM .001
GA .007
SC .007
.001
.001
.000
.002
.015
.004
.008
.290**
.067*
.009
.000
.001
.340** .000
** Group differences significant at p < .001; * Group differences significant at p < .05
Discussion: The results from this study suggest that variations in task conditions related to the type of exposure, the amount of exposure, or the presence/absence of picture support during story retelling will not affect children’s performance on measures of syntax (GA and MLUw) and rate, nor will it affect their story comprehension performance. These results were somewhat surprising. It was hypothesized that asking the child questions prior to the story retelling (and providing the correct answers when the child does not respond or when the response is clearly incorrect) would act as a second exposure to the story and affect the child’s MLUw and story length (Gummersall & Strong, 1999). The findings from the current study suggest that it may have been the second exposure to the linguistic model of the story in the Gummersall and Strong (1999) study that affected the children’s syntactic skills, rather than the exposure to the story per se. Consistent with this assumption, mean MLUw values varied by elicitation task, with the model story that contained the lowest linguistic complexity yielding the lowest MLUw (AGL) in a retelling condition. Variation to the level of support during retelling did affect the children’s story retelling skills on measures of verbal productivity (NDW and UTT) and mazing behavior. When allowed to refer to the pictures during retelling, children were able to provide longer stories containing more different words. Moreover, verbal fluency improved, possibly as a result of a decreased processing demand on the children’s working memory (Crystal, 1987). Conclusion: These findings highlight the importance of adhering to the sampling protocol with regard to picture support when evaluating children’s narrative production performance on measures of verbal productivity and verbal fluency. Further research should consider different discourse contexts and include children with differing language ability profiles. References: Crystal, D. (1987). Towards a "bucket" theory of language disability: Taking account of interaction between linguistic levels. Clinical Linguistics and Phonetics, 1, 7-22. Gummersall, D. M., & Strong, C. J. (1999). Assessment of complex sentence production in a narrative context. Language, Speech, and Hearing Services in Schools, 30, 152-164. Miller, J., Gillon, G., & Westerveld, M. (2008). SALT NZ: Systematic Analysis of Language Transcripts - New Zealand Version. Madison, WI: SALT Software. Schneider, P., & Dubé, R. V. (2005). Story presentation effects on children's retell content. American Journal of Speech-Language Pathology, 14(1), 52-60. Contact: Marleen Westerveld, PhD
[email protected]