Nonword Repetition and Word/Nonword Discrimination: Comparing Dyslexia and Language-Learning Disability Christine Gray & Stacy Wagovich, The University of Missouri BACKGROUND Children with SLI and with dyslexia have distinguishing features but often the two disorders co-occur (Catts, Adlof, Hogan & Weismer, 2005). Children struggle with reading for different reasons. Children who exhibit poor reading comprehension also tend to exhibit poor language comprehension with normal phonological processing, while children with poor decoding skills alone show the opposite profile (Cain, Oakhill & Bryant, 2000; Catts, Adlof & Weismer, 2006). who obtained scores above the 25th percentile on the Letter-Word Identification and the Word Attack subtests of the Woodcock-Johnson, but below a standard score of 85 on either the CELF receptive or CELF expressive composite score were included in the “SLI” group. Those Based on prior research (e.g., Catts et al., 2005) we predicted that the DYS/COMBO group would perform more poorly on nonword repetition than the SLI group. Additionally, we expected that the DYS/ COMBO group would have greater difficulty on the checklist word/nonword discrimination task because of its emphasis on decoding. were audio recorded and later analyzed by syllable. For those syllables with no errors, participants received one point. For those syllables with one or more errors, no credit was given. Inter-rater Independent sample t-tests were used to compare the performance of the SLI group with the performance of the DYS/COMBO group on the checklist and NWR tasks. No statistically significant differences were found between the groups for identifying common words or nonwords on the checklist. 100 Participants Twenty participants, ages 12;8 to 18;10, were selected from a previously obtained sample of 30 children who had been referred by teachers or parents because of reported difficulties in reading and/or language in school. of the participants passed a hearing screening of 1000, 2000 and 4000 Hz at 20 dB HL. In addition, each scored within or above 1.0 SD of the mean on a nonverbal reasoning task (Symbolic Relations subtest of the Detroit Test of Learning Aptitude-4, Hammill, 1998), with one exception.* t = 1.98 p = .071 75 t = 1.87 p = .086 t = 1.74 p = .105 50 2 Syllables one participant in the SLI group, one NWR 4 syllable item was not recorded due to examiner error. Percentages of accuracy were computed accordingly. completed a checklist designed by the second author which contained common words, rare words, and nonwords (in one of three randomized sets). First, participants wrote definitions or sentences for words they knew. Next, they reviewed the list again, circling any words they knew were real, but could not define or use in a sentence. This procedure is an adaptation based on Durso and Shore (1991; Shore & Durso, 1990; Schwanenflugel, Stahl, & McFalls, 1997). These data were analyzed by merging all circled words and words for which the participant attempted a definition or sentence, since both reflected the participant’s impression that the words were real lexical entries within the English language. A “checklist percentage difference” score was found by subtracting the percentage of nonwords mistakenly identified as real words from the percentage of real words correctly identified as real words on this task. Ceiling effects were observed to impact NWR 2 syllable scores for many children. It appears that the small sample size of the study resulted in low statistical power. Effect size analysis suggested that, had the samples been larger, significant group differences on the NWR task would have emerged. Since the sample was collected retrospectively, more lenient criteria for the groupings (25th percentile and below on subtests of decoding versus above the 25th percentile on subtests of decoding) were used in order to establish comparison groups and retain the largest sample size possible. The 25th percentile criterion has been used previously in the literature (Catts et al., 2006). The finding of trends toward group differences with large effect sizes for NWR total, 6 syllable and 4 syllable scores is important because these trends exist despite the fact that the DYS/COMBO group’s mean CELF expressive score (89.0) did not differ significantly from that of the SLI group’s mean CELF expressive score (90.54). It appears that the trends seen resulted from differences other than global expressive language abilities. Cain, K., Oakhill, J., & Bryant, P. (2000). Phonological skills and comprehension failure: A test of the phonological processing deficit hypothesis. Reading and Writing: An Interdisciplinary Journal, 13, 31-56. Catts, H.W., Adlof, S. M., Hogan, T. P., & Ellis, Weismer, S. (2005). Are specific language impairment and dyslexia distinct disorders? Journal of Speech, Language, and Hearing Research, 48, 1378-1396. Catts, H.W., Adlof, S. M. & Weismer, S. E. (2006). Language deficits in poor comprehenders: A case for the simple view of reading. Journal of Speech, Language and Hearing Research, 49, 278-293. Nation, K. & Snowling, M.J. (1998). Individual differences in contextual facilitation: Evidence from dyslexia and poor reading comprehension. Child Development, 69, 996-1011. 0 4 Syllables 6 Syllables Total NWR Score Nonword Repetition Task Participants SELECTED REFERENCES 25 reliability for this task was 90.6%. Each * One participant scored -1.3 SD on Symbolic Relations SLI DYS/COMBO t = 1.48 p = .157 Figure 1. Mean Nonword Repetition Accuracy by Group Nonword Repetition Total Score in Percent The finding that the Letter-Word Identification subtest of the Woodcock-Johnson is correlated with the NWR task supports previous findings in the literature that children who have difficulty with decoding also perform less well on NWR tasks (Nation & Snowling, 1998). differences were not significant, effect sizes were large for NWR total score (d=.85), NWR 4 syllable (d=.89), and NWR 6 syllable (d=.78). For METHOD Although the SLI group, n=10, and for the DYS/COMBO group, n=13. However, data on NWR were missing for 3 participants of the DYS/COMBO group resulting in an n=10 for those analyses. Responses Pearson Product Moment Correlations revealed the following significant relationships: The percentage of common words identified as real words correlated significantly with the CELF expressive score (r = .509; p < .05). “checklist percentage difference” score correlated significantly with the CELF expressive score (r = .560; p < .01). 25th Experimental Tasks A 90-item nonword repetition task of 2, 4, and 6 syllable nonwords (a modification of Gupta’s 2003 task) was administered to all participants. Participants repeated digitally recorded productions of these nonwords. DISCUSSION The For who scored at or below the percentile on the Letter-Word Identification or the Word Attack subtest of the Woodcock-Johnson were included in the “DYS/COMBO” group. CELF scores were allowed to vary within this group. To compare a group of children with SLI only (SLI) to a group of children with decoding deficits either alone or in combination with SLI (DYS/COMBO) on two tasks: nonword repetition (NWR) and a word/nonword discrimination task. HYPOTHESES Groups Participants were administered the Reading portion of the Woodcock-Johnson Achievement Tests (Woodcock, McGrew & Mather, 2001) and the Clinical Evaluation of Language Fundamentals, 4th edition (CELF; Semel, Wiig & Secord, 2003). Participants PURPOSE RESULTS Accuracy in Percent METHOD (cont’d) 100 90 80 70 60 50 50 60 70 80 90 Letter-Word Identification Standard Scores Figure 2. The Relationship Between Letter-Word Identification Subtest Standard Scores and Nonword Repetition Total Scores, r = .549; p < .05 100 The presenters gratefully acknowledge Summer Windels for her help with data analysis. This study was funded by an NIH/NIDCD research grant awarded to the second author (R03 DC006827-01).
© Copyright 2026 Paperzz