- Syntactic and Lexical Factors Conditioning the Diffusion of Sound Change
A sound change may propagate through a language in different ways. Different studies attest sound changes spreading at different rates through different phonological and/or phonotactic environments, diffusing through the speaker population (or through different dialects) in different ways, or simply spreading differentially through the lexicon. In Palu'e there is evidence for a sound change applying at different rates for different grammatical categories, with the sound change advancing in the small set of bound grammatical morphemes perhaps more completely than in free lexemes. This is evidence that syntactic information on parts of speech can affect the diffusion of a sound change through a language, and that bound forms are not necessarily more conservative than free lexemes when it comes to phonological change.
1. Ways for Change to Diffuse.1
While there are many described instances of the exceptionless application of sound changes throughout a language, following the Neogrammarian model, there are many instances in which a sound change is found to apply to only some subset of the possible target words in a language. In these cases, the application of a sound change may be delimited by the phonological or phonotactic environment of the relevant sound, by different varieties of the language, or it may simply be that a sound change applies lexeme-by-lexeme—which is a confession of a sound change being irregular.
A combination of the first two factors can be seen in the *s > h sound change in the Kwerba languages of the Mamberamo and Apauwer rivers in Irian Jaya. Anggreso Kwerba, spoken on a tributary of the Apauwer river east of but close to the Mamber-amo, is an example of a Kwerba language that has not been affected by the sound change, and shows *s reflected as s in all environments and in all lexemes; it is a control for the rest of the data. The other languages, arranged from south to north along the river, show different degrees to which the *s > h rule has applied. The words in (1) show reflexes of *s in syllable-onsets in Anggreso, spoken between the Apauwer and the Mamberamo, and four other Kwerba languages spoken along the Mamberamo river, arranged from south to north. In (2) we can see reflexes of *s in syllable-codas. In both cases Anggreso shows no change, while the Mamberamo languages all show various [End Page 427] degrees of change of *s to h. This change has applied most thoroughly in onset positions, showing phonotactic conditioning of the sound change, with only one lexical item in this set (Bagusa 'earth, forest') resisting the change.
(1) Reflexes of Proto-Kwerba *s in syllable-onset position: Geographic (and lexical) variables controlling diffusion
'EARTH, FOREST' 'GOOD' 'FIRE' 'HOT' Anggreso ISIW esam ser papasam Kwerba Ih IW (anIIn) her (Ihpwam) Kasonaweja Ih IW ehama her papaham Trimuris Ih IW eham her papaham Bagusa ISIW eham h Ir papaham
In coda position, however, the sound change is much less advanced, being resisted completely in Kwerba and Bagusa at the opposite ends of the Kwerba language chain along the river, and applying most strongly to the variety spoken in Trimuris. In Kasonaweja there is free variation (both idiolectal, with some speakers favoring s and some h, and also simply random, with degrees of variation observed from the same speaker) in the pronunciation of 'cassowary', while the two different pronunciations of the reflexes of *IS 'woman' reflect its split into two lexemes in the modern Kasonaweja language, is 'woman' and ih 'wife', differentiated by the application of the *s > h change.
(2) Reflexes of Proto-Kwerba *s in syllable-coda position: Geographic and lexical variables controlling diffusion
'SNAKE' 'CUSCUS' 'CASSOWARY' 'WOMAN' Anggreso ISya qaraas qamas IS Kwerba ISya qaraas qamas IS Kasonaweja ISya qaraas qamas, qamah IS (also Ih) Trimuris ISya qaraah qama...