Epenthetic prefixation in Alawa and Marra

ABSTRACT

Languages adopt a number of strategies to avoid dispreferred phonotactic structures. The general pattern is that such strategies involve minimum departure from the input form. One of these strategies is epenthesis. With epenthesis, the minimum departure from the input form is usually the addition of a singleton consonant or a singleton vowel. We show that Alawa and Marra have epenthetic CV syllables in prefix complexes, and that this epenthesis of a syllable is not motivated in phonological theory. We provide evidence that these prefixal structures did not originate in epenthesis, but rather were by-products of reduction processes targeting prefixed article paradigms. We propose that the synchronically epenthetic prefixes are remnants of old article roots.

KEYWORDS:

• Alawa
• Marra
• epenthesis
• grammaticalization
• noun class

Acknowledgements

The authors would like to thank the two referees for their detailed comments on this paper, which led to notable improvements.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Mark Harvey is a Conjoint Associate Professor in Linguistics at the University of Newcastle. He has worked with speakers of Australian languages in the Darwin region since 1980. The phonetics and phonology of Australian languages are one of his principal research interests. His other research interests include morphosyntax, historical linguistics, kinship, ethnobiology and indigenous spatial heritage.

Brett Baker is an Associate Professor in Linguistics at the University of Melbourne. His main interest is the processing of complex morphology. To this end, he specializes in the phonology and morphology of Australian languages and has been carrying out fieldwork with speakers of these languages since 1993. He has also published in the domains of discourse structure, semantics and syntax of Australian languages.

Notes

1 However, a component of noun class morphology which we reconstruct to an earlier article survives in a number of northern Australian languages: Ngalakgan, Wubuy, Burarra, Gurr-goni and Bininj Gun-wok, as well as the current article systems of Marra and Warndarrang; see Baker & Harvey (2020). So, these articles were not always elided in this position. Whether more languages had this kind of structure but lost the article component entirely is beyond the scope of this paper.

2 The following abbreviations are used in glossing: fem ‘Feminine’, masc ‘Masculine’, min ‘Minimal’, neut ‘Neuter’, poss ‘Possessive’.

3 As a reviewer points out, the paradigm of ‘poor fellow’ could equally be analyzed in synchronic terms as involving two root allomorphs, kaɭuku appearing in the 3FMin and aɭuku appearing elsewhere. In Table 6, we follow the analysis in Sharpe (1972).

4 There is an analysis of Kayardild, which is exceptional in having a number of coronal sonorant clusters within words, e.g. r + n, l + n, ɻ + n (Round 2009: 99). These clusters are not reported in the other principal analysis of Kayardild (Evans 1995: 69).

5 Our survey of grammars of Australian languages turned up two principal types of epenthesis: epenthesis of glides word-initially and between vowels across a morpheme boundary, e.g. Nhanda (Blevins 2001: 10), and (much rarer) epenthesis of a vowel to break up dispreferred consonant clusters again across morpheme boundaries, e.g. Bardi (Bowern 2012: 125–130), Wambaya (Nordlinger 1998: 39). We found no examples of CV epenthesis apart from the widespread use of the dummy syllable /pa/ to prevent words from being consonant final in Warlpiri, Nyangumarta and various Western Desert varieties.

6 The alternation is optional if the preceding [–continuant] segment is /k/ (Sharpe 1972: 49).

Additional information

Funding

This research was supported by ARC Discovery Project DP140100863 ‘Reconstructing Australia’s linguistic past: Are all Australian languages related to one another?’