Modeling prosodic rhythm: Evidence from second language speech.

The global prosodic structure of languages has been described using the typological dichotomy of stress-timed versus syllable-timed. Various indices have been successfully employed in literature for quantifying these classifications, one of which is the duration ratio between the total voiceless and total voiced stretches in the signal. It has been further shown that various language-specific characteristics, such as syllable-structure phonotactics and stress-sensitive lengthening and shortening, can contribute to this difference. To reveal the interaction of these components, acoustic data from running speech of L1 Spanish/L2 English and native English speakers were analyzed. Total ratio of voiceless-to-voiced durations discriminated L1 Spanish (lower) and L1 English (higher); L2 speakers showed ratios in between the two, with higher proficiency L2 speakers showing ratios closer to L1 English. A task dynamic application, a speech planning and production model [Nam et al., J. Acoust. Soc. Am. 115, 2430 (2004)] was used to model the performance of the L1 and L2 speech of L2 speakers, allowing determination of which contributors to global rhythm are more readily acquired by L2 speakers. [Work supported by NIH and NSF.]