Prefixal ‘suffixes’ in Skou

Abstract

Skou, a language of North-Central New Guinea, shows agreement on verbs by what appear to be monoconsonantal prefixes. A detailed analysis shows that the morphemes must be considered to be aligned to the right edge of the word, but are realized further left due to strict phonotactic constraints. Furthermore, the phonotactic constraint that dictates against the right-aligned morphemes appearing as suffixes, NoCoda, is the same one that is famously responsible for the infixation of left-aligned morphemes in Austronesian languages, showing that not only is right-edge aligned infixation perhaps not quite as rare as has been thought, but that it is governed by the same phonotactic constraints that determine left-aligned infixation. A historical motivation for this unusual alignment change is discussed, with a cautionary note on the use of under-analysed data in typological work.

Keywords:

• Skou
• Papuan
• New Guinea
• Agreement
• Alignment
• Affix

* Discussions with Tara Mohanan and Dan Kaufman have played a very great part in shaping this paper, and two anonymous referees for AJL have helped to make it more complete. Dave Peterson helped me to make more sense of the data. None of these people can be held responsible for the final analysis.

* Discussions with Tara Mohanan and Dan Kaufman have played a very great part in shaping this paper, and two anonymous referees for AJL have helped to make it more complete. Dave Peterson helped me to make more sense of the data. None of these people can be held responsible for the final analysis.

Notes

* Discussions with Tara Mohanan and Dan Kaufman have played a very great part in shaping this paper, and two anonymous referees for AJL have helped to make it more complete. Dave Peterson helped me to make more sense of the data. None of these people can be held responsible for the final analysis.

²I have not used the more frequently cited Tagalog data here since the -um- and -in- infixes in that language do not show any variation in surface alignment based on the prosodic shape of the root, since all roots are C-initial. The oft-cited um-abot ‘reach for’, purportedly showing the prefixal use of -um-, is in fact [], with a glottal stop that is, as Matsuda French (1988) and Schachter & Otanes (1972) make clear, as underlying as any consonant in the language (the root is ). An alternative analysis of Tagalog infixation as being driven by a requirement that syllables have onsets does not apply to Tukang Besi, which permits V-initial words and roots contrasting with C-initial ones (including ). See Donohue & Maclachlan (1999) for additional arguments against the viability of an analysis using a high-ranked NoCoda for Tagalog.

³The clitics are identical in form to the singular and plural pronouns, the only difference being that clitic pronouns other than the 1sg or 2sg, but not the free pronouns, may be realized with a schwa rather than e, and that the ‘plural’ clitics may be used with dual reference. There are no separate dual clitic pronouns, and the dual free pronouns show more distinctions than either the singular or plural pronouns, marking gender for all persons.

⁴The ultimate source of the conjugational differences might be a different set of onsets (voiceless palatal stop and voiceless palatal fricative), and the unusual reconstruction of *hy clusters appears to provide some support for at least the latter segment. By Proto-Skou times, however, this putative contrast had been lost, and only the conjugational differences remained to be inherited by the daughter languages.

⁵The unnaturalness of the class of inflecting onsets is somewhat mitigated by appeal to a diachronic explanation. Modern Skou w <*κ^w (Donohue 2002), which along with k and h represent all of the [+back] consonants. This leaves only l and one verb root with r, the non-nasal sonorants, to account for. We have thus a disjunctive class of ([+back] or [+sonorant, -nasal]).

⁶Some verbs are phonologically suited for inflection, in that their onsets are either null (the vocalic paradigm) or else w l k or h, and yet do not show inflection. Disyllabic members of this class are not common, but do exist: ‘be surprised, be amazed’ is one such example of a verb with two syllables, both of which are in the ‘inflecting’ category (based on phonological constraints—that is, Segmental Structure Constraints), but which do not inflect for lexical reasons. These verbs will not be discussed further, being governed by stipulative, irregular conditions they are outside the regular rules and models of the grammar described here.

⁷Orthographically, vowel sequences are written with a glide between the two vowel symbols, thus /ia/ appearing as <iya> (Sikale et al. 2002). This practice has not been followed here, in order to avoid obscuring the phonotactic patterns being discussed. In terms of the map in Figure 1, One lies approximately as far north as Manem, and approximately the width of the map again (50km) to the east of the edge of the map.

Additional information

Notes on contributors

Mark Donohue1

Mark Donohue, Department of Linguistics, Research School of Pacific and Asian Studies, Australian National University, ACT 0200, Australia. E-mail [email protected]