To appear in Proceedings of the 33 rd West Coast Conference on Formal Linguistics. Somerville, MA: Cascadilla Press. Phonotactic Evidence from Typology and Acquisition for a Coda+Onset Analysis of Initial sC Clusters Heather Goad September 2015 1. Introduction * Recent research in phonology has seen structural and perceptual approaches to syllable well- formedness pitted against each other. Concerning left-edge clusters, for example, different explanations are provided for the observation that obstruent-initial (play) and s-initial clusters (stay, slay) diverge in phonological behaviour. 1 Researchers who advocate a structural approach propose that this motivates an articulated view of the syllable: obstruent-initial and s-initial clusters have different representations (see Goad 2011 for a recent overview). Researchers who advocate a perceptual approach, by contrast, propose that these differences can be explained solely by perceptual considerations: segments are ordered to maximize their perceptibility and the acoustic properties of s versus other obstruents account for their different behaviour (e.g. Wright 2004). In this paper, I return to a structural approach to syllable well-formedness, but one which is informed by perceptual considerations. In view of the perspective I take, one question that arises is what type of structure holds for word- initial sC clusters, given the multitude of options that have been proposed. As will be shown, initial sC clusters mirror the phonotactic profile of ordinary coda+onset clusters; for instance, s+stop is always well-formed, while s+sonorant is not (Goad 2011, 2012). Because of this, I adopt the position of Government Phonology that sC clusters are analysed as coda+onset clusters (following Kaye 1992) and, critically, that this position holds for the left edge of words, as well as medially. In the latter half of the paper, I turn to explore the predictions that stem from a coda+onset analysis of sC clusters for second language (L2) acquisition. Specifically, it is predicted that in the acquisition of a subset grammar (i.e., where the L2 is contained within the L1 for the construction under focus), learners exposed only to the ill-formedness of s+stop should infer the ill-formedness of s+sonorant. A related question that emerges in this context concerns the kinds of evidence that learners can use to acquire subset grammars. I present some results from Schwartz & Goad (2015) who experimentally show that the acquisition of a subset grammar can be facilitated by exposure to a new type of evidence that has not, to our knowledge, been explored in L2 acquisition: Indirect Positive Evidence (IPE). IPE is evidence from errors in the learner’s L1 made by native speakers of the learner’s L2. 2. Phonotactics I begin by examining the role that sonority plays in syllable well-formedness. It has long been known that segments are optimally ordered to ensure that there is a (steep) rise in sonority toward the nucleus, a (shallow) fall in sonority away from the nucleus, and a fall in sonority across a syllable boundary. More concretely, the consequences for word-initial clusters are that: (i) they strive to be * Department of Linguistics and Centre for Research on Brain, Language and Music, McGill University; heather.goad@mcgill.ca. I would like to thank the WCCFL audience for questions and comments. Concerning the experiment reported on in section 10, I would like to thank my co-author on this work, Misha Schwartz, as well as Hye-Young Bang, Walcir Cardoso, Guilherme Garcia, Lauren Garfinkle, Natália Guzzo, Alexei Kochetov and Kris Onishi. This work was supported by grants from SSHRC and FRQSC. 1 Throughout the paper, ‘obstruent’ refers to stops and fricatives other than s, except where noted.