Re: [tied] Re: Daci

From: Piotr Gasiorowski
Message: 12775
Date: 2002-03-20

From:  "tgpedersen" <tgpedersen@...>
Date:  Wed Mar 20, 2002  12:10 pm
Subject:  Re: Daci
[Torsten:] > You are wrong, and I have corrected you on that point before. The only thing in Danish history I could possibly identify with your invented "puristic fashion", is the speech habits of the German-speaking upper layer of Denmark at the time when the Danish monarchy comprised also German-speaking Holstein and mixed Danish-German Scleswig and when one third of the inhabitants of Copenhagen were German-speaking. But such a mixed situation is common in the world, and therefore the "retrograde" changes that took place in Danish are mainstream, not an exception.
[Piotr:] No, Torsten. The mainstream of sound change is the sort of thing that occurs essentially unnoticed, has little or no sociolinguistic relevance and does not turn back. As for your Danish <sk>ibbolethisation (sorry for not remembering the socio-political details the first time): if something of this kind happens, the target of _deliberate_ replacement is an arbitrarily selected feature that attracts public awareness and social stigma -- a shibboleth. People have no sort of global control over their language; they focus on a detail at a time.
[Piotr:] >> Whatever motivated it, it was not the desire of the man in the street to prevent the corruption of grammatical paradigms (the replacement was not morphologically conditioned).
[Torsten:] > It was.
cf Swedish skæra <s^æra> "cut", skar <skar>, skurit <skyrit> with
   Danish  skære <skær&> "cut", skar <skar>, skåret <skor&D>
[Piotr:] Was it? It levelled out the paradigms to be sure, but I gather it took place elsewhere as well, and the fact that the process depalatalised the initial of <sky> 'aspic' "by mistake" but did not affect words like <chocolade> shows it to be essentially a spelling pronunciation.
[Torsten:] > Standard Danish has both grammatical gender and uses it at the same time to mark countable/uncountable. In other languages you see something similar. Let's say someone spilled water on the floor. Would you, in Polish, then refer to the spilled beer as "ta" or "to" (I assume it is?)?
[Piotr:] In which other languages, namely? Not in Polish, at any rate, or anywhere else in Slavic, as far as I know. 'A (pint of) beer' is <piwo> (neuter), and beer spilt on the floor (what a shame!) is still <piwo> (neuter). With a demonstrative, it's always <to piwo>, just as 'water' is always <(ta) woda> (feminine, unless it is miraculously transformed into beer), and 'juice' is always <(ten) sok> (masculine).
[Piotr:] >> Phonological naturalness is in a great part a matter of the physics of articulation. The tendency to maximise the "ease of articulation" is a manifestation of growing entropy.
[Torsten:] > I see. Using this metaphor backwards, perhaps you could argue that there were several kinds of physical entropy, and we could never foretell which one prevailed?
[Piotr:] No, there is only one kind, which means that any articulatorily grounded phonological process has a preferred ("natural") direction. For example, [-mt] > [-nt] is far more common than [-nt] > [-mt], the intervocalic voicing of [-s-] > [-z-] is likelier than [-z-] > [-s-] in the same context, etc. The Celtic erosion series of *p- > *f- > *h- > Ø (zero) is also known from other languages (e.g., all its stages are attested in the various Oceanic languages, and PIE *p- > Arm. h-), but I haven't yet seen a language in which Ø (or even [h-]) > [p-] by regular sound change.
Regular fortitions, which are statistically rarer than lenitions and so are likely to get swamped by the latter in the long run, are perception-, not articulation-oriented -- we inject some extra information, as it were, into selected points of the segmental string. Such reinforcement is costly in terms of muscular effort and neuromotor control, so we resort to it sparingly, whatever the gain.