Sonority and the Unusual Behavior of /s/
Heather Goad [+]
A cross-linguistic examination of phonological behaviour generally supports the position that obstruents, except for /s/, form a single sonority class. For example, obstruents prefer to be in onset followed by segments of higher sonority. /s/, by contrast, is often less restricted. For example, /s/-initial clusters need not respect the sonority rise required of obstruent-initial branching onsets (in Indo-European); /s/ also commonly resists participating in processes that target or yield voiceless fricatives (e.g. Grimm’s Law in Proto-Germanic). Generalizations such as these support the view that obstruents are optimally located in positions that maximize their perceptibility: they prefer to ‘lean on’ adjacent sonorants. /s/, by contrast, contains robust cues for place and manner, which ensures its perceptibility regardless of where it appears (Wright 2004), and makes it resistant to participating in change. Most of the literature on the unusual behaviour of /s/ has focused on Indo-European languages. These languages typically contain sC clusters in word-initial position, so much of the literature has motivated analyses for /s/ in this context. Initial /s/ has been analysed as extraprosodic (unprosodified) (Steriade 1982); as directly organized by higher prosodic structure (syllable (van der Hulst 1984) or prosodic word (Goldsmith 1990)); or as a coda preceded by an empty nucleus (Kaye 1992). We are sympathetic to Kaye’s position, which we support through an examination of sC sonority profile across Indo-European (Goad 2011). However, sC clusters are sometimes referred to as an Indo-European ‘accident’, which leads us to the following questions: • Do non-Indo-European languages have sC clusters displaying properties similar to those seen in Indo-European? • More generally, does /s/ behave in unusual ways in non-Indo-European languages, consistent with its acoustic properties, or does it pattern with other members of its sonority class? This paper will examine the behaviour of /s/ in various non-Indo-European languages, minimally Blackfoot (Algonquian) (Frantz 2009), Ōgami (Ryukyuan) (Pellard 2009) and Acoma (Keres) (Miller 1965). In all three, /s/ functions unusually, but in ways different from what is found in Indo-European. For instance, all three languages have initial sC clusters. Acoma, though, otherwise does not permit clusters, and other properties of sC in this language suggest that /s/ is a singleton onset where an empty nucleus interrupts the putative cluster (Goad 2012). Blackfoot has initial ssC in addition to sC. In medial position, /s/ and /ss/ can be flanked by consonants. Contrasted with an otherwise simple syllabification system, Blackfoot /s/ appears to function as a vowel (Derrick 2006, Denzer-King 2008, Goad & Shimada to appear). A preliminary look at Ōgami suggests that /s/ requires the same analysis in this language (Goad & Shimada to appear). In short, /s/ patterns outside its sonority class in both Indo-European and non-IE languages, although the behaviour it displays is not necessarily the same in all languages/language families. We conclude that, although /s/ is unusual in diverse ways across typologically-related and unrelated languages, its behaviour can be analysed in terms of ordinary syllable constituents, without recourse to appendix-like machinery, if an abstract view of the syllable is adopted.