Re: Word-initial sound changes

>On Thu, 27 Jan 2000, John Cowan wrote:
>
>> raccoon@ELKNET.NET wrote:
>>
>> > I know that in languages I'm familiar with that use some form of lenition
>> > (voicing or fricativization, or both), lenition doesn't occur
>> > word-initially. My question is, how unreasonable would it be for
>>lenition to
>> > occur word (and phrase)-initially too?
>>
>> Totally reasonable, and what the Celtic languages (and their conlang
>> relatives) do.
>
>Reasonable, yes, but not as lenition. Lenition has a phonetic/
>phonological trigger; what Celtic has is mutation, which is
>triggered by the morphology and has very little to do with
>phonetics/phonology at this point in their histories. Not
>knowing how consonant alternations play out in Eric's language,
>I can't say if it is lenition or not.
>
>Perhaps it is just splitting hairs, but I have always preferred
>to use the term 'lenition' to refer to consonant alternations
>which are  triggered by phonetics/phonology, and reserve the
>term 'mutation' for consonant alternations which mark
>morphological categories. It's a useful distinction to make.
Even taking Dirk's distinction into account, there do appear
to be cases of word-initial lenition out there - at least in
the realm of diachronic sound change.  There are numerous
examples from Austronesian (e.g. */b/ became /v/ word-
initially in Malagasy).  Here's an example from the Australian
language Jajgir, culled from Terry Crowley's "An Introduction
to Historical Linguistics".  In Jajgir, voiced stops became
glides word-initially:

*/dja:lanj/     >       /ja:lanj/
*/ganja:mbil/   >       /janja:mbil/

I don't know what triggered this change in Jajgir, but I
suspect that most of the Austronesian cases were
extensions of intervocalic lenition:  */b/ became /v/
intervocalically in Malagasy, and this was extended to
cases where the /b/ was word-initial and the preceding
word ended in a vowel.  (It's my understanding that that's
sort of how mutation in Celtic got started...)

Matt.