Reconstructions

Note

Reconstructed as *bali[jzZ]i ‘grass’ in Blust (1970). With regard to the last consonant Simalur balixi, falixi unambiguously indicates *j, but Pak penit, Nauna pelic point to POc *ns, PMP *z, all other known reflexes being indeterminate. Although this item can be securely reconstructed, the supporting cognate set is problematic in this and in certain other respects.
First, northern Philippine forms such Kankanaey balíli ‘a rush: Juncus effusus L.’ and Ifugaw balíli ‘kind of tough grass with long leaves’ are either similar to the above forms by chance, or exhibit an unparalleled development of *j, and *z. It is thus possible that all Philippine forms reflect an etymon *balili which is distinct from *baliji or *balizi.

Second, the two intervocalic consonants in POc *palisi appear to have metathesized independently in at least three Oceanic languages: Proto-Willaumez, Selau (and perhaps other languages of Buka and Bougainville), and Proto-Micronesian. In at least the latter case the metathesis in this item evidently is part of a more general change (cp. POc *paluja > Proto-Micronesian *fasula ‘to paddle’).

Nonetheless the possibility remains that POc had doublets *palisi and *pasili which show a complementary pattern of retention. Finally, it is unclear whether PMP *baliji, *balizi referred to a particular type of grass, or whether it functioned as a generic term. If the latter was the case this reconstruction runs counter to the claim of Brown (1984) that the generic category ‘tree’ is always lexically encoded before the generic category ‘grass’.