[Faculté du langage & cognition] Shanti Ulfsbjorninn (Lyon / UCL)

Universals of syllabification and the case of Arrernte
Lundi 23 Mai 2016 - 10:00 à 12:00
Lieu détaillé: 

UPS Pouchet, salle 124


Shanti Ulfsbjorninn (U. Lyon / UCL - University of London)

Universals of syllabification and the case of Arrernte

A narrative of exceptionality pervades the phonological analyses of Aranda. Most famously, Aranda (a.k.a. Arrernte) has been presented as the only language in the world with an exclusively VC syllabification (Breen and Pensalfini 1999). This would make it the sole example of a language without onsets, a condition explicitly forbidden by formal phonological frameworks (GP, CVCV and OT). It is therefore taken as a case study against the attempt to formalize linguistic universals at all (Evans and Levinson 2009). Other analyses, those that contest this conclusion, rely on the controversial ‘moraic onset’ (Topintzi 2010, Topintzi and Nevins 2015, Nevins 2016), a combination of phonological objects that was itself (until recently) universally excluded (Hayes 1989). Moreover, from a diachronic and ‘general linguistic’ perspective, Arandic languages also appear extraordinary because they are (probably) the only language group to have systematically lost all initial consonants, in apparent violation of formalization of positional strength (supposedly universal) (Harris 1997; Ségéral and Scheer 2001).

            In this talk, I will assess the competing accounts of syllabification in Aranda. Firstly, the ‘onsetless’ VC analysis of Aranda will be comprehensively destroyed along with its concomitant support for rejecting phonological universals. I will show that syllabification in Aranda follows an ordinary CV pattern, which can only be interpreted as VC by an aggressive misinterpretation of the language’s phonetic form, aided by an internally inconsistent fallacious logic.

            Secondly, the moraic onset account, will be shown to be more problematic for questions of phonological architecture that it initially appears. From our assessment, the only instance where the CV syllabification of Aranda is problematic is based on an instance of allomorphy selection. The allomorphy selection superficially looks to support a moraic-onset analysis, however, I will show that the assumptions required for its explanation are far from obvious.

On balance, we should assume that Aranda is an ‘ordinary’ language that can be insightfully syllabified in CVCV, whose superficial ‘weirdness’ lies in its underlying forms which have been shaped by language specific phonetic factors, some of which are already known to the literature (Dixon 2002).

As part of this explanation, I will conclude by showing a way in which Aranda actually is informative for ‘big thinking’ on phonological theory.  I will demonstrate that the surface ‘VC’ patterns of Aranda can be functionally unified with the process of ‘initial dropping’ attested in typically-developing first language acquisition (L1). The two processes occur for different phonetic reasons, but they both demonstrate the functioning of Attentional Domains; a concept that has wide ranging implications for our understanding of windows in stress systems.