Andrew Jarrette wrote:
> -- Can't say anything in my defense, you really got me there, I, a
> native English speaker, really couldn't manage to remember "gone"
> and "shone" and "cloth" and "holiday" as other examples of *a: >
> "short o" besides "hot". But the more examples there are, the more
> I wonder: why?
They reflect sporadic shortenings, occurring for various reasons at
different times. The normal development of OE /a:/ was into ME /O:/
(except in the North, where an unrounded vowel was retained), and then
into Mod.E /oU/ (with well-known dialectal variants).
Any /a:/ that was shortened already in Old English yielded ME /a/ and
Mod.E /æ/ (as in compound placenames and surnames with <Stan-> 'stone'
as the first element).
Any /a:/ whose reflex /O:/ was shortened a little later, in Middle
English, yielded Late ME short /o/ and eventually the modern vowel of <hot>.
What I personally find more puzzling is the occasional Mod.E. /O:/, as
in <broad> and in the variant pronunciation of <gone> as "gawn"
(formerly also in a few other words). It reflects some kind of irregular
lowering in Early ModE, as if the vowel in these words had been exempted
from the normal raising and diphthongisation producing /oU/. The
dialectal pronunciation of <cloth> with the same vowel ("clawth") is the
result of lengthening before voiceless fricatives (the same accents
typically have a long vowel in <lost>, <cough>, <moss>, etc.).
Of course in accents that don't distinguish between <cot> and <caught>
(Canada, much of US English), there's also a secondary merger of the
vowels in <broad> and <hot>, both being low, tense and unrounded.
Piotr