On 2008-07-23 03:22, gprosti wrote:
> A general methodological question: what criteria do you think should
> be used to distinguish a probable semantic change from an improbable
> one? Is it possible that comparative linguistics has not yet developed
> any such criteria?
If you want formal criteria that work infallibly, I don't think it's
possible to lay them down. Of course there is a gradient of
plausibility, from what is the _most_ probable (meaning no change at
all, e.g. 'wolf' > 'wolf'), to what may be likely under certain
circumstances (a slight shift of meaning, e.g. 'wolf' > 'jackal', in a
country where jackals predominate), to changes which, while not
impossible in principle, cannot even be understood if you don't know the
complex story behind them (like 'wolf' > 'harshness in some tones of a
stringed instrument'). We know why Polish <komórka> '(garden) shed'
acquired extended meanings like '(living) cell' and has now ended up
with 'mobile phone' as its most common meaning, beacuse the whole
process is recent and has been well documented; but we can only guess
why OIr. súil came to mean 'eye' rather than 'sun'. Semantic change may
involve the figurative use of words, fossilised metaphors, unexpected
associations, unpredictable misunderstandings etc. That's how the human
mind works. Still, the more complex change you posit, the more
justification should be given, preferably indicating the particular
circumstances responsible for it, parallel examples from other
languages, etc. The only thing that doesn't require an explanation is
semantic continuity.
Piotr