Coarticulation in VCV sequences in Arabic

Abstract

Vowel‐to‐vowel coarticulation in VCV utterances has been the subject of several studies. Ohman [J. Acoust. Soc. Am. 39, 151–168 (1966)] found that vowels in VCV utterances in English and Swedish have some transconsonantal effect on one another. He also found some evidence suggesting that secondary articulation features, like palatalization in Russian, block coarticulation. Action theorists, such as Fowler [J. Exp. Phon. Gen. 112, 386–412 (1983)], explain V‐to‐V coarticulation in terms of timing; that is, they claim that vowels in speech production are underlyingly overlapping and consonants ride on top of the vowels. This suggestion implies that intervocalic consonants, whether they have secondary features or not, do not block coarticulation. Keating [UCLA WPP 62, 1–13 (1985)], on the other hand, explains it in terms of autosegmental phonology. She places the features for vowels and consonants on two separate tiers, and leaves consonants unspecified for vowel features, so that V‐to‐V coarticulation affects an interpolation between vowel targets. Keating's model implies that consonants that have secondary articulation (i.e., vowel features) must block coarticulation. A preliminary analysis of VCV utterances in Arabic shows that pharyngealized consonants indeed tend to block coarticulation, a finding which supports Keating's prediction over Fowler's.