Conditions on the variable interpretation of |U| in Japanese

This paper examines the historical and phonological properties of h in Japanese. It shows that, by analysing a specific case of segmental variation, we not only deepen our understanding of the sound which varies but also shed light on some general characteristics of the sound system as a whole. Using an Element Theory approach (Anderson and Jones 1974; Kaye et al. 1985; Harris and Lindsey 1995; Scheer 1999; Nasukawa 2005; Cyran 2010; Backley 2011), the discussion focuses on the distribution of the element |U|, arguing that |U| is naturally weak in Japanese. This helps explain two idiosyncrasies in Japanese phonology – the restricted distribution of labial consonants and rounded vowels, and the patterning of h with labials. In modern Japanese, labiality is phonologically and phonetically weak. In vowels, u and w are produced without lip rounding, as unrounded [ɯ] and [ɰ]. And in consonants, the labial stop p is banned from certain contexts. These facts point to the inherent weakness of |U| in Japanese, where weakness refers to structural headedness; following Backley and Nasukawa (2009), it is assumed that labials are represented by headed |U| (cf. non-headed |U| in velars). To account for the restricted distribution of labials, it is argued that labiality (headed |U|) is only realised in Japanese if a specific structural condition is met: |U| must co-occur with (i.e. be supported by) another element from the same sub-group of ‘dark’ elements. Thus, the paper exploits the natural division between dark elements {|A|, |U|, |L|} and light elements {|I|, |H|, |Ɂ|}.