One of the most important constraints on morphosyntactic change is the unidirectionality of changes seen in grammaticalization (Haspelmath, 2004). The overwhelming tendency for unidirectionality of change from more lexical to more grammatical along the grammaticality cline is undeniable. The claim that grammaticalization is irreversible holds if we understand it as a 'token reversal', that is once an item grammaticalized it is not going to reverse back (e.g. an adposition that has grammaticalized into an affix is not going to reverse back into an adposition). But if we understand degrammaticalization (antigrammaticalization in Haspelmath, 2004)as a type of change characterised by proceeding in the opposite direction to grammaticalization, we find a number of examples. Degrammaticalization occurs much less frequently and is less regular than grammaticalization (Norde, 2009), but its instantiations cannot be dismissed as insignificant because of this.

In my talk I am going to focus on what I believe are degrammaticalized reflexes of POc *pa[ka]‑ in one branch of Southeast Solomonic languages (Oceanic). Whilst *pa[ka]‑ has been reconstructed as a causative prefix, it had other functions such as deriving multiplicatives and functioning as a verbal modifier, e.g. POc *patur pa[ka]-qitik (weave caus small) 'weave small' (Evans, 2003:266). In some languages there are reflexes of *pa[ka]‑ which have associative (or similative) function, e.g. Boumaa Fijian gone 'child', va'a-gone 'childish' or Samoan Samoan mauga 'mountain', fa'a-mauga 'like a mountain' (Evans 2003:248). This function is found in languages from several primary subgroups of Oceanic: Meso-Melanesian (Teop), Central Pacific (the Fijian languages, Samoan), and Evans (2003:254)concludes that it is possible that these functions may be reconstructable for POc *pa[ka]-. The antiquity of this function is supported by the data from the Guadalcanal‑Gelic branch of Southeast Solomonic, where the reflexes of *pa[ka]‑ with the associative function also occur.

Reflexes of both *pa‑ and *paka‑ with causative function are reconstructable for Proto Southeast Solomonic, but in the Guadalcanal-Gelic branch the causative function has been for the most part lost. However, reflexes of *paka‑ with the associative function are productive. This is uncommon, as usually if a language has a reflex of *pa[ka]‑ it usually has at least the causative function, and may have other functions (Evans, 2003:242). But even more unusual is the fact that in these languages reflexes of *paka‑ occur not as a prefix, but as an unbound form, usually a preposition.

Tolo

1. Hia re~rei vaha tama-na.

3sg rdp~look like father-3sg.poss

'He looks like his father.'

(Crowley, 1986:39)

Furthermore, the data suggest that in some of the Guadalcanal-Gelic languages this morpheme functions not as a preposition but as a verb. Following the framework developed for classification of degrammaticalization by Norde (2009), I will discuss the degrammaticalized forms and the processes of change they have undergone.

References:

Crowley, Susan Smith. (1986). Tolo dictionary. Canberra: Pacific Linguistics.

Evans, Bethwyn. (2003). A study of valency-changing devices in Proto Oceanic. Canberra: Pacific Linguistics.

Haspelmath, M. (2004). On directionality in language change. In Olga Fischer, Muriel Norde, & Harry Perridon (Eds.), Up and down the Cline - The Nature of Grammaticalization (pp. 17-44). Amsterdam/Philadelphia: John Benjamins Publishing Company.

Norde, Muriel. (2009). Degrammaticalization. Oxford: Oxford University Press.