Gemination in West Germanic
Craig Joseph Callender
This dissertation offers a reanalysis of the role of consonant strength in gemination processes in West Germanic, arguing that they most readily affected the strongest consonants. The phonetic properties of different realizations of /r/ are considered with an eye toward explaining why it is generally the poorest candidate for gemination. Cross-linguistic data corroborate the claim that strong consonants geminate most readily (Taylor 1985, Podesva 2000, Kirchner 2001, Kawahara 2005).
Based on the behavior of geminates within Germanic, and a survey of instrumental phonetic studies, I propose three classes of geminating consonants to account for West Germanic Gemination (WGG), as well as gemination processes in general within West Germanic. Class 1 is composed of the best candidates for gemination, voiceless stops; they are the only consonants to geminate before liquids and word-initially in Swiss sandhi. Class two is composed of other obstruents, and the sonorants /n/, /m/ and /l/. Under normal circumstances, they geminate medially. Class three is composed of the most sonorous consonants, /r/, /j/ and /w/, which geminate rarely, and only in South Upper German. The dialects of Upper German also confirm the proposed analysis; where degemination is occurring, it affects sonorants first, then fricatives, and finally stops. By observing degemination in progress, we can trace our way backwards through the phonetic mechanism of WGG.
Previous analyses have attempted to explain WGG in terms of palatalization (Prokosch 1939), circumflex accent (Sievers 1878), vocalization (Bennett 1946), and syllable restructuring (Kaufmann 1887, Murray and Vennemann 1983, Lass 1994). Murray and Vennemann (1983) correctly emphasized the role of sonority and syllable weight in the implementation of WGG, although their stressed syllable law could more properly be termed a tendency. Denton (1998 and 1999) argued for lengthening via increased voice onset time, and Goblirsch (1999) argued for two different types of lengthening: compensatory before /j/ and non-compensatory before liquids. I follow an analysis similar to Goblirsch, but argue that neither type of WGG was compensatory; remnants of /j/ in Old Saxon settian 'sit' and Old Norse synnia 'refuse' attest that geminates arose before /j/ was lost.