Velars are typically more susceptible to palatalisation than any other stops, since the body of the tongue, which acts as the active organ in the articulation of velars, is almost inevitably pushed forward in anticipation of a front vowel or glide. If the speaker's control of the articulatory movement is relaxed, the area of contact may be large, involving the hard palate and the middle part of the tongue. In addition to being fronted the stop may be affricated as well, since a "pure" palatal stop is quite difficult to make.
There is a natural trajectory of palatalisation: [k] (velar stop) > [k^] (slightly palatalised velar stop) > [c] (palatal stop) > [ts'] or [tS] (alveopalatal or palatoalveolar affricate). The further development may consist in depalatalisation (e.g. [ts'] > [ts]) or in the weakening and loss of the stop element (e.g. [tS] > [S]). The various end products of such processes can be seen in various languages, e.g. [k > k^ > ts' > ts > s] before originally front vowels in the passage from Latin to French ([kentu-] > [ts...] > [sa~]). There is no single deterministic path of development, but the general tendency is for palatalised velars to evolve into coronal ("front-of-the-tongue") affricates or fricatives (e.g. French [s] : Italian [tS] : Castilian [รพ]).
Velars that did not undergo palatalisation at a certain historical stage may be affected by a similar change later, often with different results. For example, *k(W) stops were palatalised before front vowels in Proto-Slavic, producing *c^ [tS] (palatoalveolar affricates). In late Common Slavic there was another wave of palatalisation affecting velars before _new_ front vowels, especially *e^ from older *ai (< PIE *oi or *ai). This time the outcome was different: *kai > *ke^ > *ce [tse]. Several hundred years later, in early Polish, the unrounded back vowel *y became fronted after velar stops, yielding new *ki from earlier *ky. The velar was palatalised again, though rather slightly this time; the Modern Polish pronunciation is [ci] (rather like Romanian <chi>) in such cases. Incidentally, none of these changes is directly connected with the Satem shift, which was older than the palatalisation of *k(W), not conditioned by the following vowel, and produced still different sounds.
Romanian, too, has been affected by successive palatalisations, the more recent ones targeting velars in new palatalising environments. The sounds produced by the old palatalisation had undergone development into palatoalveolars in the meantime. If, for example, Latin /l/ had changed into a palatal glide in the clusters <cl-> and <gl-> at the time when velars were being palatalised before front vowels, we'd get palatoalveolar affricates from these combinations as well. But the change was late enough for the _new_ /kj/ and /gj/ to escape affrication, and the sequences in question have ended up as palatal stops.
The two palatalisations were independent of each other, but both followed a "natural" course of phonetic evolution.
Piotr
----- Original Message -----
From: "alex_lycos" <altamix@...>
To: <cybalist@yahoogroups.com>
Sent: Tuesday, April 15, 2003 8:15 PM
Subject: Re: [tied] Re: Albanian-Romanian Concordances
> Ammm.... I see I did not made up well the question. I do not keep now in
> sight the phenomenon for all languages , just the Rom. Lang is the one
> which i was thinking about.
> We have the alteration of Latin /k/ & /g/ in Rom. Lang having here the
> /c^/ and /g^/. An another kind of palatalisation of these /k/ & /g/ are
> the groups "chi(k'i)" and "ghi (g'i).
> I wanted to know if these are independents changes or id they are in a
> relationship since from /k, g/ we have 2 way of alterations, once /c^,
> g^/ and once /k'i, g'i/. The Rom. scholars mean that the /k/ and /g/ in
> words like "chin", "ghion" should be seen as palatalized /k/ and /g/ and
> not "clean" velars.
> So is there any chain as k/g > c^/g^ > k'i/g'i or k/g > k'i/g'i > c^/g^
> or there is nothing in a chain here but two different alterations with
> no intermediary stops like:
> a) k > k'i ; g > g'i
> b) k > c^ ; g > g^
> and the changes of "a" has nothing to do with changes of "b".
>
>