In "classical" OE the status of palatalised <sc> seems to have been like that of <s^c^> in modern Russian. Note that intervocalic <-sc-> (presumably = [-S.S-]) behaved like a real cluster in OE verse: a word like <bisceope> 'bishop [dat.]' could be metrified as "heavy primary-stressed + light accented + unstressed", so that <to bisceope> was a well-formed half-line. Even in Middle English medial "sh" had geminate-like properties: open-syllable lengthening did not apply before it (while even the syllabification of /-st-/ seems to have vacillated between /s.t/ and /.st/). I don't think initial "sh" was ever voiced in the southern dialects. The full phonemicisation of /S/ as a member of the voiceless fricative series was probably complete when intervocalic geminates became simplified.
----- Original Message -----
From: richardwordingham
Sent: Friday, August 23, 2002 7:50 PM
Subject: [phoNet] Re: Parallel Potentially Feeding Changes with Sporadic Exceptions

I feel cheated when Quirk and Wrenn's 'An Old English Grammar' says, 'By the time of Classical OE, the biliteral sc had come to represent the single consonant source [S] occurring initially in Mod. E. 'ship' and in the form OE 'scip'; in poetry, words beginning with sc- could alliterate only woth other wrds beginning with sc-.'  The statement seems to claim this rule as evidence that 'sc' was [S] rather than [sk], whereas the applicabilty of the rule to 'sp' and 'st' greatly weakens this evidence.