Phonological Opacity and Icelandic Preaspiration
This paper explores the relationship between constraints on syllable contact and the emergence of so-called preaspirated stops in Icelandic. It is a well-known fact of Icelandic phonology that, when followed by a sonorant, a stop loses its aspiration. However, there are two patterns. When followed by /l, n, m/, a stop surfaces with `preaspiration', i.e. as a sequence of [h] + plain stop. When followed by /j, v, r/, the result is a plain stop preceded by a long vowel. The most promising approaches to this problem have attributed the dierence between the two patterns to dierences in syllabication, due to language-specic constraints on syllable contact. The argument is that the dierence between the two patterns lies in the amount of sonority rise within the dierent clusters, i.e. that consonant clusters that rise too much in sonority (apirated stop + /j, v, r/) cannot cross a syllable boundary and will therefore emerge as complex onsets preceded by a long vowel. Preaspirated stops, on the other hand, have nothing to do with syllable contact and emerge in the surface structure for independent reasons. I propose that laws of syllable contact are in fact also the main motivation behind the emergence of preaspirated stops in Icelandic, along with constraints on glottal activity in stressed syllables. The two patterns arise due to dierent rankings of specic syllable contact constraints relative to other phonological constraints in the system. I furthermore demonstrate that preaspirated stops are the result of an opaque interaction of phonological constraints and can therefore not be derived within the framework of classic OT.