Russian voicing assimilation

[Peter Chew]

Can anyone help me with the following question:

My understanding of Russian voicing assimilation is that where you have a
consonant cluster as follows:

        C       C       ...     C       O       S
         1       2               n

(where C represents any consonant, O represents an obstruent consonant,
and S represents a sonorant), the "phonetic voice" of the entire cluster
is the same as the "phonemic voice" of O.

So, for example, in /stvol/ "trunk", /t/ is the obstruent consonant
represented in the schema above by O; it is phonemically unvoiced, so the
entire word-initial cluster is phonetically unvoiced, including /r/.

If I am wrong up to this point, please correct me. (I'm purposely ignoring
cases where the cluster contains no obstruent consonants, and word-final
clusters.)

I have my doubts as to whether the rules are the same when the cluster has
(a) a syllable boundary, or (b) a morpheme boundary intervening between
any of its constituent segments, as for example in /ot+vesti/ "to lead
away".

The question, therefore, is as follows:

Does anyone know of any empirical experimental evidence (based e.g. on
spectrographic analysis) which suggests either that the phonological
rules are, or are not, the same in both cases? If so, I'd be grateful if
respondents could provide me with the appropriate references.

Thank you in advance,

Peter Chew

_______________________________________________________________________________

Peter Chew

Oxford University Phonetics Laboratory
_______________________________________________________________________________

41 Wellington Square
Oxford
OX1 2JF
UNITED KINGDOM

Tel. +44 (1865) 270446
Fax  +44 (1865) 270445

Home page: http://www.phon.ox.ac.uk/~peter/homepage.html
_______________________________________________________________________________