Reconstructions

Note

Also Itbayaten ma-yayuh ‘to run’, Itawis pa-ladyaw ‘run’, Bontok layaw ‘flee, run away, clear out’, and Tagalog lagyóʔ ‘spirit’ (the latter included under *laRiw ‘flee’ by Dempwolff (1934-38), but excluded here). As with the parallel discrepancy in PAn *balija or *baRija ‘weavers sword’ the Formosan languages indicate *R as the initial consonant of this item, while the MP languages indicate *l. I have posited several morphologically complex forms of *laRiw, although a distinctive meaning is not always apparent for these, as with *ma-laRiw. The form *pa-laRiw requires special comment, as its reflexes have survived in some languages when a reflex of *laRiw has not. In languages such as Tiruray, Wolio and Chamorro (WMP), Tetun (CMP), and Numfor (South Halmahera-West New Guinea), the morpheme boundary that is recognized is purely historical, and is marked by = to distinguish it from the similar synchronic morpheme boundary in forms such as Malay lari, pe-lari.

It is an interesting question why the morpheme boundary should have been lost more commonly in reflexes of *pa-laRiw than in reflexes of other morphologically complex forms of *laRiw. One might argue that affixes differed in how tightly they were bound to the stem. However, this should be generally true and not specific to particular lexical items. Moreover, just what tightness of binding might mean is obscure except in terms of text frequency. Perhaps *pa-laRiw occurred with particularly high text frequency because it not only signaled the abstract notion of flight or escape, but also the actor ("fugitive; one who has eloped"). In some descendants of PMP reflexes of *pa-laRiw may have occurred with higher text frequency than reflexes of *laRiw, thus favoring the survival of the bi-morphemic word even when the stem itself passed out of use.