Rhythmic conversion between speakers of different dialects

Abstract

Speakers' convergence to each other's production of segments has been well established [Sancier and Fowler, J. Phon. 25, 421–426 (1995); Nielsen, J. Phon. 39, 132–142 (2011)]. Less is known about prosodic convergence, but it has been identified for stress and pitch accent [Krivokapic, JASA, 127, 1851 (2010); Krivokapic, JASA 129, 2658 (2011)]. In an acoustic experiment, rhythmic conversion between speakers of American and Indian English is examined. The two languages differ both in the stress patterns of individual lexical items and in their global rhythmic properties, with Indian English being syllable-timed and American English stress-timed. Changes in speakers' productions are examined using a synchronous speech task [Cummins, ARLO 3, 7–11 (2002), Zvonik and Cummins, Proc. Eurospeech 2003, 777–780 (2003)], where two speakers read sentences at the same time. Eight subjects (four dyads), each consisting of one Indian and one American speaker, read a short story that contained 12 words in which the two dialects differ in stress pattern and four sentences in which the global rhythmic properties are tested. The data of the dyad examined to date indicate convergence to a more stress-timed pattern for the Indian speaker.