Production of lexical stress by Spanish speakers of English as a second language and its effect on perceptual judgments of intelligibility and accentedness

Abstract

Acoustic analyses of recordings of read speech were used to compare the production of lexical stress in multisyllabic words by Spanish ESL (English as a Second Language) speakers to that of native English speakers. The participants read short sentences containing multi-syllabic target words for which syllable duration, intensity, and fundamental frequency in the pronunciations of stressed and unstressed syllables were compared. Target words were equally divided based on syllable count, the syllable that should receive lexical stress based on native English pronunciation, and Spanish/English cognate status. Both orthographically identical cognates (e.g., radio) and orthographically non-identical cognates (e.g., necessary/necesario) were included. Average values of syllable duration, intensity and fundamental frequency were recorded across the speaker groups, and effects of cognate status on production were noted. Target words pronounced by the native speakers (to avoid differing segmental factors) were then resynthesized using the PSOLA algorithm to simulate the acoustic patterns of the ESL speakers. Native English listeners were asked to rate a set of stimuli embedded in a carrier phrase for intelligibility and accentedness. Results of the listening tasks suggest that variations in the acoustic properties under investigation may impact a listeners perception of speech intelligibility and accentedness.