Though there are interesting features both in the consonant and vowel repertories of Icelandic, this paper will focus primarily upon Icelandic consonant phenomena, primarily those of aspiration, and where related, gemination. This paper will contain a brief introduction to the Icelandic consonant inventory and some cursory information on the diachronic progression of Icelandic consonants (which is helpful in that it enables one to predict Modern Icelandic pronunciation from the orthography, as I have done for this paper). Also introduced are some of the previous and current theories related to Icelandic consonant preaspiration. This will be followed by a survey of some of the data I have compiled from grammars and dictionaries, an analysis of the data as well as a discussion of possible derivations and rules governing preaspiration, and finally a discussion of some problems inherent in a purely syllable-based analysis.
Icelandic stops, instead of contrasting between voiced and unvoiced, use aspiration contrastively. Historically, consonants all contrasted based on voicing, and the orthography represents the earlier voiced contrast rather than the modern, aspirated one: though there are no longer any voiced stops in Modern Icelandic, the orthography includes b, d, gj, and g. Underlyingly, however, these sounds have changed to unaspirated [p, t, c, k], which are in contrastive distribution with [ph, th, ch, kh] (p, t, kj, and k, respectively, as seen in the orthography).
A feature of Modern Icelandic that makes it an interesting study is the presence of preaspiration in certain environments. Several arguments have been suggested to account for preaspiration, which Hansson (2003:24) deems a 'diachronically highly unstable' phenomenon. It is important to consider these arguments before examining the data, in order to assess their appropriateness. Preaspiration is not reflected in the orthography of Icelandic (though it can be accurately predicted from it, either from the orthographically represented gemination or based on environment), but it does not appear to be an allophonic phenomenon. There are several minimal and many near-minimal pairs that illustrate its contrastive status, and Hansson (2003), based on durational studies, suggests that, rather than a 'preaspiration' feature of the consonant, there is a full [h] segment, as part of a surface cluster. Implementing Hansson's suggestion, the Icelandic word epli ('apple') is transcribed as [ ɛhplɪ ] rather than as [ɛhplɪ], the method which had often previously been used. Because of the extreme vowel coloring that almost always occurs in a situation of preaspiration, Hansson also reflects the devoiced vowel in his analyses, and an allophonic equivalent to the previous example includes a devoiced [ɛ]. Though I accept Hansson's suggestion that there is no actual [preaspiration] feature, I continue to use the word to describe the phenomenon of h + deaspirated stop clusters, due to convenience.
