The primary contrast between obstruents in English has been described in alternative ways: either /p, t, t∫, k, f, θ, s, ∫/ are voiceless and /b, d, d3, g, v, ð, z, 3/ voiced, or the former set is fortis and the latter lenis. This paper reviews the phonetic properties of fortis sounds and draws together both published and new arguments for a basic fortis/lenis contrast in English obstruents. Rules are presented for phonetic and phonological alternations involving English obstruents, and it is shown that stressed vowel shortening, final glottalization, and aspiration are all properties of fortis obstruents while certain devoicings and final cluster simplifications operate on lenis obstruents. Schwa devoicing is also related to consonant strength. It is argued that a fortis/lenis analysis provides a more general explanation of a wider range of phenomena than the voiced/voiceless analysis and that voicing is predictable from fortis and lenis properties. Two sound changes in English, the [θ/ð] split and the development of an initial implosive in Lower Great Plains English, are shown to require the fortis/lenis analysis, as is Notker's Law in Old High German, suggesting that the fortis/lenis contrast is quite old in Germanic. Finally it is argued that fortis and lenis obstruents form natural classes and that a feature [fortis] must be added to the root node of the feature tree.
Read full abstract