Another fortis-lenis language: A reanalysis of Old English obstruents

Abstract

The theoretical framework adopted in the analysis of Old English obstruents is laryngeal realism, a framework using privative features in modelling laryngeal oppositions. Equipollent oppositions, although real in the phonetic sense, must clearly be delineated from phonology. Old English obstruents are either unmarked (lenis/neutral) or marked: 〈b〉 /b/, 〈d〉 /d/, 〈cg〉 /dʒ/ or /ɟ/, 〈g〉 /ɡ/ are not marked for [voice] (although they are passively voiced between sonorants) and as such cannot regressively voice obstruents, singleton 〈p〉 /p/, 〈t〉 /t/, 〈ć〉 /tʃ/ or /c/, 〈c〉 /k/ are marked for [spread] (aspiration, or GW ‘glottal width’), singleton 〈f〉, 〈þ/ð〉, 〈s〉, 〈g〉, 〈h〉 are unmarked, but are passively voiced in the FricV/Son environment. Fricatives in unstressed syllables (even when couched between sonorants) are not voiced. If there is a sonorant separating the fricative from the stressed vowel there is no voicing (SonFricV/Son). The only voiced fricatives after a stressed vowel+sonorant consonant are /f/ [v] and /x/ [ɣ] (but this is a historical coincidence). (Phonetically voiceless) Geminates, s+stop and f/h+stop clusters are special in that they constitute a sequence of a fortis followed by a lenis obstruent impervious to passive voicing.