English /hw/ > /w-/

[Richard M Hogg]

The cluster vs single segment argument about <hw,hl,hr,hn> is less
open and shut than Mark Hale suggests. Fundamentally, he cannot
dismiss the evidence of alliteration so easily. That all vowels
apparently alliterate with one another is explicable as alliteration
of zero onsets (ie, vowels don't alliterate, initial consonants do,
even when they're zero). And the ability of, say,< hw> and <hr> to
alliterate with one another is in sharp contrast to the behaviour of
<sp, st, sc>.
Whether we analyse, say, <hw>, as /xw/ or voiceless /w/ may not be
decidable, and it seems best to acknowledge that both possibilities
have considerable merit (sorry to be so spineless).
On  Miguel Vidal's question about the spelling shift, I seem to
recall that Mosse explained the shift from <hw> to <wh> (etc.) as a
readjustment to conform with the standard Anglo-Norman digraphic
sequences of consonant + <h>, as in <sh>, <ch>. That seems very
probable, especially if at the time the special sounds represented by
<hl, hr, hn> were being lost. But if this is so, the spelling change
has no phonological or phonetic significance.
Finally I don't think that Mark Hale should be quite so dismissive of
spelling evidence as he is when he says OE spelling evidence is
"completely irrelevant". A large part of our task is to reconcile
plausibly reasonable phonological analyses with actual spellings and
we should never give up on that.