Reconstructions

Note

Also Kanakanabu namíci, Saaroa ɬamici, Rukai (Maga) amicu, Rukai (Tona) amici, Itbayaten humti ‘Solanum nigrum Linn.’, Ilokano amtík ~ agamtík ‘nightshade plant, Solanum nigrum’, Hanunóo untí ‘a small wild herbaceous plant (Solanum nigrum Linn.); the young leaves are used as a green vegetable by the Hanunóo’, Maranao moti ‘morel plant: Solanum nigrum L.’, Manobo (Western Bukidnon) muti ‘a plant: Solanum nigrum’, Javanese ranti ‘tomato’, Manggarai kenti ‘a wild vegetable: Solanum nigrum’. This word shows multiple irregular variations, making any reconstruction difficult. I have arbitrarily chosen to compare Paiwan sametsi with forms in other languages that are ambiguous for initial *S or zero.

Finally, I am indebted to R. David Zorc for drawing my attention to the Itbayaten and Manobo (Western Bukidnon) forms, and for reminding me that he proposed an alternative reconstruction, *SamŭCi, in Zorc (1982:118). Zorc’s motivation for positing an unstressed high back vowel in this form was to relate the otherwise unexplained high back vowels in Maga Rukai amicu (metathesis of *i and *u), Itbayaten humti (metathesis of *a and *u), Hanunóo untí (metathesis of *a and *u), Maranao moti, and Manobo (Western Bukidnon) muti.

The precarious nature of this reconstruction, however, should be obvious, since: it requires the assumption of 1) two different metathesis patterns in three different languages, 2) an irregular loss of *S- in Hanunóo and Manobo (Western Bukidnon) (this would be equally true if these forms were assigned to the present reconstruction), 3) syncope of an unstressed low vowel *a in Itbayaten and Hanunóo, and 4) an irregular loss of prepenultimate *a (rather than reduction to schwa) in Manobo (Western Bukidnon). Moreover, it forces us to assume that the well-known syncope of schwa /VC__CV also applied to other vowels, including both *u and *a. I know of no evidence that might justify this assumption, which should be apparent in any of these languages that undergo rightward stress shift through suffixation (none of them shows a syncope pattern when this happens). Add to this the fact that the initial consonants of the Kanakanabu, Javanese, and Manggarai forms, the medial vowel of the Kanakanabu, Saaroa and Rukai forms, and the final consonant of the Ilokano form are unexplained by either reconstruction, and it appears clearly preferable to settle on *SameCi and leave the rest of these apparently related but wildly irregular forms unexplained.