Liquid phonology: A test case for the listener misperception hypothesis

Abstract:

The listener misperception hypothesis of sound change (Ohala 1981, 1993, 2003) has been a fruitful area of inquiry over the past several years, in part because it makes testable predictions. One prediction is that long-distance dissimilation such as liquid (lateral) dissimilation should be a result of listener hypercorrection. While a number of studies have found experimental evidence for the perceptual origins of assimilation, to date no work has shown empirically that the origins of dissimilation are perceptual. The present study focuses on understanding the origins of liquid dissimilation by testing listener categorization of liquids along an /r/-/l/ continuum to explore perceptual patterns of co-occurring liquids, which have been shown to have robust long-range coarticulatory effects (Tunley 1999, Heid & Hawkins 2000, West 1999, 2000).

Listeners have long been known to have perceptual access to the fine-grained acoustic details that accompany coarticulation, and to use these acoustic cues in phoneme discrimination (e.g. Whalen 1990, Kingston & Diehl 1994, Gaskell 1997, Beddor et al. 2007). A novel aspect of this study is that, while past studies have generally found that listeners perceptually “undo” the acoustic effects of coarticulation (e.g. Mann & Repp 1980 et seq.), the results here suggest that for liquids, listeners adjust their perception in the same direction as coarticulation, strengthening rather than undoing the effect.  Furthermore, since for liquids listeners are shown to compensate in an assimilatory direction, dissimilation would result from a failure to compensate normally, suggesting dissimilation may be a result of hypocorrection rather than hypercorrection.