Development of the Computerized Revised Token Test-Cantonese for the assessment of language comprehension in Cantonese-speaking people with aphasia

• ¹ Hong Kong Polytechnic University, Hong Kong, SAR China
• ² University of Pittsburgh, United States

This preliminary study reports the concurrent validity and test-retest reliability of a newly adapted Cantonese Computerized Revised Token Test (CRTT-Cantonese) to evaluate sentence comprehension ability in people with aphasia (PWA) versus healthy, neuro-typical controls (HC). Method. The English CRTT (McNeil et al., 2015) was translated into Cantonese by a panel of professional translators and speech therapy professionals into standard Chinese for the reading version of CRTT (CRTT-R-WF-Cantonese) and into formal Cantonese for the listening version of CRTT (CRTT-L-Cantonese). Fifteen native Cantonese PWA and 15 HC (age range 49-69 years old) were initially recruited for this study. All the participants had more than six years of formal education with no premorbid history of speech, language, hearing and learning disabilities. All the subjects completed the Cantonese Aphasia Battery (CAB), Hong Kong Oxford Cognitive Screen (HK-OCS), Snellen visual test, pure tone hearing screening test, and CRTT pretests in the first session. Then they were assigned randomly to perform the CRTT-L-Cantonese and CRTT-R-WF-Cantonese across 2 separate sessions with at least one-week intersession interval. All of the HC were retested on both versions of the CRTT, whereas seven PWA were randomly retested using CRTT-L-Cantonese, and eight PWA were randomly retested using CRTT-R-WF-Cantonese. Results. Two-way repeated measure ANOVA (group by modality design) revealed a significant main effect of the group, i.e., PWA group showed significantly lower performance than HC group on the overall (F (1,28) = 24.8, p < .001, η2 = 0.47) and efficiency (F (1,28)= 29.5, p < .001, η2 = 0.51) scores on both CRTT-Cantonese tests, while the other effects were not significant (see Fig 1). Pearson correlation analysis revealed 1) adequate concurrent validity in terms of the correlations between overall CAB score and CRTT-L-Cantonese (r = 0.74, p < .01) and CRTT-R-WF-Cantonese scores (r = 0.72, p < .01) in PWA, and 2) test-retest reliability both for CRTT-L-Cantonese (HC: r =.65, p <.01, PWA: r = .985, p < .001) and CRTT-R-WF-Cantonese (HC: r = .72, p < .01, PWA: r = .96, p < .001). Lastly, there were significant and high correlations between the CRTT-L-Cantonese and CRTT-R-WF-Cantonese based on the overall (r = 0.83, p < .001) and efficiency scores (r = 0.839, p < .001). Conclusion. The preliminary results suggest that CRTT-L-Cantonese and CRTT-R-WF-Cantonese significantly differentiate the reading and listening comprehension of PWA versus HC. The preliminary finding provided acceptably high concurrent validity and test-retest reliability for both CRTT-Cantonese tests. High correlations between the CRTT-R-WF-Cantonese and CRTT-L-Cantonese in PWA might suggest that each version can be used interchangeably for assessment of language comprehension, especially among the patients with modality specific impairment such as auditory agnosia (Eberwein et al., 2007). As data collection is ongoing, it is expected that the sample size will be increased to 30 PWA and 30 HC, which will be helpful to report other psychometric properties of the CRTT-R-WF-Cantonese and CRTT-L-Cantonese including the cut-off point scores, internal validity and consistency as well (Hageman et al., 1982).