Effects of cognitive status on disfluency: A cross-linguistic study of Chinese- and English-speaking older adults

Disfluency in spontaneous speech has been reported to be a potential marker for cognitive impairment. However, consistent working definitions are lacking, and it remains unknown whether disfluency measures operate similarly across languages. This study investigates how cognitive status shapes disfluency in Chinese and English, aiming to delineate both language-general patterns and cross-linguistic differences. A total of 100 speech samples elicited via the Cookie Theft task (50 Chinese, 50 English) were selected from the MCGD corpus (Chinese) and DementiaBank (English), equally stratified into cognitively normal and impaired groups based on MMSE scores. Following manual annotation, speech samples were analysed using ten disfluency measures across three dimensions: temporal, pausal, and repetition-repair. Results showed that six measures were sensitive to cognitive status across both language groups, and language did not significantly moderate the effect of cognitive impairment on disfluency. Temporal measures declined consistently with cognitive impairment across both languages, with cognitively impaired individuals showing significantly reduced speech rate, phonation-time ratio, and mean length of runs. Pausal measures generally increased with cognitive impairment, with cognitively impaired individuals showing significantly longer silent pauses and more frequent filled pauses. For repetition-repair measures, repetitions increased with cognitive impairment in both languages but showed a higher baseline frequency in Chinese than in English, whereas repairs displayed less consistent patterns across the two language groups. Collectively, these findings support the value of disfluency measures as cross-linguistically relevant markers of cognitive impairment, while underscoring the importance of language-sensitive interpretation for specific disfluency types.