At 2:37:16 PM on Wednesday, May 14, 2008, fournet.arnaud
wrote:
> From: "Brian M. Scott" <BMScott@...>
>> It may not make sense, but it can happen: OE <ha:ligdo:m>
>> became ME <halido:m> as part of a regular set of changes in
>> trisyllabic words. In fact, non-northern varieties of ME
>> eliminated /a:/ altogether, partly by shortening and partly
>> by a change /a:/ > /O:/, but it still had length contrasts.
> What is this word halido:m ?
<
http://www.merriam-webster.com/dictionary/halidom>
> The change /a:/ > /O:/ is not at all the same as /a:/ >
> /a/
Of course not. I didn't say that it was. I said that the
total loss of /a:/ was the combined result of two different
changes, both regular.
> What you are describing is not a general change /a:/ >
> /a/.
It's not an unconditional change, but it *is* a general
change /a:/ > /a/ in certain environments. But this is
irrelevant, because Patrick's supposed law of phonological
entropy operates at the level of individual roots and isn't,
properly speaking, phonological at all. In particular, it
doesn't require a general change /a:/ > /a/; it merely says
that this change may occur in a particular root if the
resulting root does not already exist as a distinct root.
Moreover, you appear to have recognized this when you wrote:
> The idea that a long a: could become short while there
> still are long vowels in the system does not make sense.
Taken at face value, this is not a claim that /a:/ > /a/ is
impossible when there are still long vowels in the system;
it is a claim that no instance of /a:/ > /a/ whatsoever
should occur when there are still long vowels in the system.
Since it is the second (and in English normal)
interpretation that is actually relevant to Patrick's 'law',
that is the one that I used. If you actually meant the
first interpretation, your statement may perhaps be correct
-- I've not given it much thought -- but it has nothing to
do with Patrick's 'law'.
Please note that I am not in the least defending Patrick's
'law of phonological entropy', which is little more than a
license to to make arbitrary adjustments to adjust the
theoretically predicted forms to fit the actual data.
Brian