Under the exonym ‘Araucanian’, Mapudungun (presumed isolate, Chile/ Argentina, ARN) has occupied a prominent place in word-prosodic literature for four decades (cf. among many others: Hyman 1977, Hayes 1995, Hyde 2016). Throughout this time it has been presented as a classic example of a perfect grid stress system. As its title implies, this chapter will argue against such claims, not- ably, those which make Mapudungun out to be a quantity insensitive, left-right iambic system. The problematic nature of the evidence on which these typological accounts are based, namely, the data in Echeverría and Contreras (1965), has already been highlighted in quite some detail by Paul de Lacy (2014). However, a more thorough review of the primary literature on Mapudungun stress — before and after Echeverrı́a & Contreras 1965 — is still unavailable. This article will provide such a review, alongside an analysis of new data gathered in Chile’s Araucanía Region.
My findings will show that Mapudungun does not fall squarely within traditional prosodic typology. For one thing, the position and hierarchy of prominences does not refer to the lexical level alone, but takes into account morphological structure as well, thus making it incompatible with a perfect-grid analysis. Furthermore, Mapudungun prominence has remarkably little effect upon the broader phonological structure of the language, a feature evidenced in the diachronic record (Molineaux 2017), as well as in the general lack of prominence-based synchronic alternations (cf, for instance, Sadowsky et al. 2013). In this sense, Mapudungun may be considered a Non-Dominant accent system, in the terms of van Coetsem (1996).
In a close examination of native speaker intuitions, as well as acoustic analyses, I will argue for a non-iterative, word-level, right-aligned, quantity-sensitive, trochaic parsing for Mapudungun. This said, I will also show how, in clash, an additional stem-final prominence regularly takes priority over the word-level trochee, elimin- ating it altogether, as in (1b).
(1) Stem (s) and word (ω) stress in Mapudungun (Molineaux 2014: 188)
a. [[ɨ.ˈʈʂɨf.]s -tu.-pu.-ke.ˈla-j.-m-i]ω
throw-rest-trloc-hab-neg-ind-2-s
‘You don’t usually throw x back here’
b. [[e.lu-ɲ.ˈma.]s -fi-j.-m-i]ω
give-appl-dir.3sp-ind-2-s
‘You give him/her/it x’
I conclude that Mapudungun prominence seems to have limited phonological activation (Clements 2001), but appears to be remarkably active at the morpho- logical level. Indeed, lack of prominence-based segmental asymmetries, absence of culminativity in words with multiple prominences (see 1a), as well as the unambiguous marking of the stem-edge, conspire to create transparency in the morphological structure. This is particularly relevant for a language within the polysynthetic spectrum. While languages of the analytic/fusional spectrum (such as English), can afford for the morphology to be subordinate to prosodic well-formedness, Mapu- dungun appears to avoid this, leading to clearer parsing of the morphology.
The overall argument of the paper has important implications for the roles of the phonological and morphological word in Mapudungun, as well as for the precise nature of ‘rhythm’, ‘demarcation’ and ‘culminativity’ as definitional traits of word prominence across the world’s languages..