Rhodes linguist Will Bennett and his colleague Aaron Braver from Texas Tech 老虎机游戏_pt老虎机-平台*官网 have discovered an interesting pattern in isiXhosa which could be the result of a phonological process (a sound change), or a result of historical change in isiXhosa morphemes. Will Bennett presented this pattern at a research seminar in the Department of English Language and Linguistics on Tuesday 2 September, 2014.
IsiXhosa has a pattern of dissimilatory labial palatalization in which the passive suffix /-w/ causes stem-final bilabial consonants to alternate with palatal consonants, seen in alternations like iyahlamba ‘it is washing’ ~ iyahlanjwa ‘it is being washed’. The verb ukuhlamba has a bilabial consonant [mb], but when the passive suffix /-w/ is added, the /mb/ surfaces as a palatal consonant nj.
Much previous work has treated this alternation as a phonological process of assimilation and/or dissimilation (Louw 1975, Khumalo 1987, Chen & Malambe 1998). Some others, however, have argued that the pattern is fundamentally morphological or historical in nature, with palatalization encoded in the lexicon (Herbert 1977, Ohala 1978, Anderson 1992). No previous work has tried to directly assess the status of the pattern experimentally, until now.
This talk presented an experiment designed to probe the status of bilabial palatalization using a ‘wug test’ paradigm. 10 speakers were presented with 40 nonsense verbs, and were asked to produce passive forms of them. If palatalization is truly a phonological process, speakers would apply palatalization when making passive forms of the new nonce words. If, on the other hand, palatalized forms are simply variant forms stored in the lexicon, then speakers would not palatalize in nonce words.
Bennett and Braver's findings show that some speakers systematically extend palatalization to nonce words, while others do not. This suggests that the palatalization pattern is a genuine phonological rule for some speakers, but is a morphological pattern for others.