Piotr Gasiorowski wrote:
> Oops, sorry! I was writing in haste myself and it slipped my mind
> (sorry, Richard) that /æ(:)/ and /e(:)/ are prone to (at least
> orthographic) post-palatal breaking in WS (/æ(:)/ also in Northumbrian)
> in contexts where normal breaking doesn't apply: <sceap>, <ceaf>,
> <giest>, etc. My own inclination here is to accept the diacritic
> interpretation. Note the difference between <giest> and EWS giefu > LWS
> gyfu, gifu (with "real" <ie> due to back umlaut).
On second thoughts, I have to correct myself again and revoke my
declared preference. LWS i ~ y (<gyst> etc.) is common in words with
"broken" *e, contrary to what I thought at first, which suggests that
the pronunciation corresponding to <ie> from various sources ended up in
the same way -- ergo, the WS diphthong was likely real. This kind of
evidence does not allow us to decide what came first, i-umlaut or
palatal breaking, since the result would be the same either way. For
long /æ:/, however, we have the celebrated test-case of the "cheese"
word, LWS cy:se (Mercian ce:se). Here the development must have been:
PWGmc. *ka:- [with *i in the next syllable]
PIngv. *kæ:-
pre-OE *c^æ:-
pre-WS *c^iæa- (or some such kind of "Saxon drawl")
EWS c^i:y- (<ci:ese>, with i-umlaut as if of <e:a>)
LWS c^y:- (<cy:se>)
-- as otherwise i-umlaut would have left the /æ:/ unaffected and palatal
breaking would have produced *<ce:ase>. As far as I remember, Lass and
Anderson have no good answer to that, except for their weak ad-hoc claim
that <cy:se> is somehow exceptional.
Sorry about the confusion, but the topic is difficult and I haven't
looked into it for a while.
Piotr