Testing OCP-labial effect on Japanese rendaku
Gakuji Kumagai
January 2017
direct link: http://ling.auf.net/lingbuzz/003290
In Japanese rendaku, there are a number of factors that inhibit rendaku. One of them is that, although /h/ usually becomes labial /b/ when rendaku applies (e.g., hako ‘box’ + hune ‘ship’ ? hakobune ‘ark’), the rendaku application of /h/ is blocked if the following consonant is labial /m/ (e.g., suna ‘sand’ + hama ‘beach’ ? sunahama ‘sand beach’/*sunabama) (Kawahara et al. 2006). One contributing factor to this rendaku blocking is that, if /h/ became labial /b/, it would beget a sequence of homorganic consonants /b…m…/, which would violate the OCP-labial effect. The current paper is the first report of an experiment that examined whether this restriction applies productively to nonce words that contain labial consonants. The results show that 1) the OCP-labial effect can be generalized in rendaku; 2) it works locally rather than non-locally; and 3) the applicability of rendaku is gradient according to the following labial consonant: The more similar two consonants are, the more strongly they are disfavored. To account for this gradient effect, I argue that the process involves two OCP-labial constraints: OCP (labial) and OCP (labial, -continuant).