The Centre for Speech Technology Research, The university of Edinburgh

11 Jul 2006

Suzanne Boyce (University of Cincinnati)


It is well-known that American English /r/ may be produced with a number of different tongue shapes (Delattre & Freeman, 1968; Tiede et al. 2004), and that its most salient characteristic is a very low third formant. It is perhaps less well known that all attested shapes converge to identical patterns of the first three formants (Westbury et al. 1998). This fact suggests that speakers choose among several possible strategies for deriving the correct acoustics from vocal tract shape. Our work focuses on the following questions:

  1. what are the different articulatory strategies that allow speakers to achieve similar acoustics with very different vocal tract shapes,
  2. when do speakers switch between strategies, and
  3. are there traces in the acoustic signal that indicate which tongue shape was used.
Some implications of these multiple acoustic strategies for our understanding of the phonetics and phonology of liquids will be discussed.

[back to PWorkshop]

<owner-pworkshop@ling.ed.ac.uk>