LLF – Bât. ODG – 5e étage – Salle du conseil (533)
Rob Mailhammer (UWS)
Morphophonemics and phonological processes in Iwaidjan languages: the significance of morpheme boundaries
The topic of this talk are morphophonemic alternations in Iwaidjan languages, traditionally spoken in Northwestern Arnhem Land. I will argue that the fact that these alternations only occur at morpheme boundaries is crucial to understanding how these alternations originated and possibly why they persist.
The Iwaidjan languages show allomorphy in stem and affixal morphemes in both inflection and word formation. This allomorphy shows up as alternations in inflectional paradigms and as variation of morphemes in composition and derivation. Previous work has claimed that this allomorphy is largely phonologically conditioned, and has analysed the relevant alternations as the outcome of phonological processes, both synchronically and diachronically (see e.g. Pym and Larrimore 1979, Evans 1998, Singer 2006). The main problem with this analysis is it is not actually the case that all of this allomorphy is phonologically conditioned. In several cases it there is no synchronic or diachronic evidence, and resulting alternations are completely morphologically conditioned. Going back to Pym and Larrimore (1979), previous work has assumed an underlying segment, labelled K, that acts as a phonological conditioning factor. However, the hypothetical phonological effects that K would have had to cause are too divergent that they could have been going back to one single segment. For example, in one type of alternation K supposedly causes a glide to be “preserved”, while in another it supposedly stops a stop from leniting.
a. K-wara vs. ng-ara
‘3sg goes’ ‘I go’
b. K-bani vs. nga-wani
‘3sg sits’ ‘I sit’
I will argue that the historical cause for these alternations are not phonological processes that show up as sound changes, but that they are the result of prosodic processes occurring during cliticization but which only morphologised independently in the daughter languages, possibly in combination with re-occurring processes like lenition, which is an ongoing but irregular process e.g. in Iwaidja. Thus, it is not the case that all alternations were created at once and are therefore inherited from Proto-Iwaidjan, but that they may also have arisen independently in the Iwaidjan languages. This view can better explain why the respective alternations only occur at morpheme boundaries and why they do not fit neatly into a coherent phonological pattern. This has also ramifications for the synchronic description of Iwaidjan languages in that a process view of these alternations is not useful in describing them.