ACOUSTIC AND DISTRIBUTIONAL PATTERNS OF FILLED PAUSES IN URDU-ENGLISH BILINGUALS’ SPEECH

Abstract

This study investigates the production of filled pauses (FPs) in sequential bilinguals to assess whether hesitation behavior is primarily shaped by language-specific norms or stable speaker-specific habits. Eighteen female speakers of Urdu (L1) and English (L2) were recorded while performing identical spontaneous speech tasks in both languages, generating 216 minutes of data. FPs were manually annotated and analyzed according to phonetic type (vocalic-only, vocalic-nasal, nasal-only), segmental context (e.g., silence–word, word–word), utterance position (initial, medial, final, single), and acoustic properties (duration, pitch, formants, and intensity). Results revealed a higher frequency of FPs in English than in Urdu, reflecting increased planning demands in L2 speech. Both languages showed a strong preference for vocalic-only fillers, with Urdu FPs being longer and English FPs displaying lower pitch and more fronted vowel quality. Contextually, English exhibited more isolated (silence-bound) FPs, while Urdu favored embedded placements. Positional patterns showed that medial fillers dominated across both languages, though English had a higher proportion of initial and stand-alone tokens than Urdu. These findings suggest that sequential bilinguals adapt phonetic aspects of FPs to L2 norms while maintaining L1-derived structural preferences, highlighting the dual influence of language and speaker identity on disfluency production. The study contributes to bilingual fluency research and supports the development of language-specific models of disfluency.