- A Reevaluation of Proto-Polynesian *h1
Proto-Polynesian (PPN) is divided into two primary branches: Tongic and Nuclear Polynesian. Between the two subgroups, only the Tongic languages regularly retain what is traditionally reconstructed as PPN *h, which derives from Proto-Oceanic *s. However, residues of PPN *h are found in certain words in Nuclear Polynesian languages. Furthermore, in some of these languages, it shows up as s, which is a violation of theories of sound change for which h > s is considered implausible. This paper shows that the traditional reconstruction of PPN *h actually conflates PPN *h and a lenis reflex of PPN *s, and that the sporadic retentions are not borrowings, but rather reflect a direct inheritance from Proto-Polynesian. Sporadic and irregular patterns of retention are attributed to lexical diffusion.
1. Background: Pre-Polynesian and Proto-Oceanic *s.
Proto-Oceanic had palatal obstruents *c and *j and oral-nasal grade pair *s (oral grade) and *j (nasal grade).2 The Proto-Polynesian phonemes traditionally reconstructed as continuations of these palatal obstruents are *s and *h (Dempwolff 1928, Elbert 1953, Grace 1959). Dempwolff (1928) noted that Proto-Austronesian *s, *z/Z, *c, and *j (which he wrote *t´, *d´, *k´, and *g´) became s in Futunan and Samoan, but h in other Polynesian languages. He suggested that the corresponding nasal grades, *ns, *nz/nZ, *nc, and *nj merged with the oral grade series in Tongan and 'Uvean, but disappeared in the remaining members of the Polynesian group. From all this, he concluded that the oral palatal series merged into a single palatal, *s, and the nasal grade series merged into a single nasal grade, *j, in Proto-Oceanic. Proto-Polynesian continued to reflect these two palatals. Like Demp-wolff, Elbert (1953) and Grace (1959) reconstructed PPN3 *s as the indifferent reflex of the entire simple palatal series, and PPN *h as the corresponding reflex of the nasal grade. Not all scholars agreed with this analysis, some preferring to see *s as the nasal grade and *h as the oral grade reflex (Milner 1963, Hockett 1976). [End Page 20]
In 1976, Blust showed that neither s nor h, but rather t is the nasal-grade reflex of Proto-Oceanic *s in the Polynesian languages. However, this implied that *s and *h were both reflexes of oral grade POc *s. In 1988, Ross integrated these into a larger pattern of "fortis" and "lenis" reflexes of obstruents in Oceanic languages generally.
2. Introduction: The Problem.
The Polynesian languages are divided into two primary branches: Tongic and Nuclear Polynesian. In the standard hypothesis of sound change for Polynesian languages, the PPN *h, which derives from Proto-Oceanic *s, is regularly retained only in Tongic languages and lost in all Nuclear Polynesian languages. However, there are a few words in some of the Nuclear Polynesian languages that appear to retain PPN *h. Furthermore, some of these languages seem to retain this phoneme as s. This causes a problem for standard theories of sound change, in which h > s is considered implausible.
There is a possibility that this phoneme could have been borrowed from another source, either from one of the two Tongic languages, or from a non-Polynesian source. However, the evidence favors a hypothesis that the phoneme is native to these Nuclear Polynesian languages, and this can be justified using phonological and lexical criteria for borrowing.
In this paper, I argue that part of what is generally reconstructed as PPN *h is actually a lenis reflex of PPN *s, and that the sporadic retentions are not borrowings, but rather reflect a direct inheritance from Proto-Polynesian. The sporadic and irregular pattern of retention is explained by an appeal to lexical diffusion.
3. Problematic Reflexes
3.1 Irregular Retentions of Proto-Polynesian *h.
Until recently, Proto-Polynesian *h was thought to be lost in all Polynesian languages except Tongan and Niuean, the two languages that constitute the Tongic subgroup. However, a number of Nuclear Polynesian languages do retain PPN *h. This problem was first pointed out by Marck (2000:91), who remarks that "East 'Uvean, East...