[tied] Re: viridis.. just for Miguel, other please ignore

--- In cybalist@yahoogroups.com, "Richard Wordingham" wrote:

> --- In cybalist@yahoogroups.com, "alex_lycos" <altamix@...> wrote:
>
>> Miguel, just one question more please. Which is the phonetically
>> explanation of Romanian "verde" = green in this case?
>> We have for "varza" = Latin "vir(i)dia"
>> We have for "verde" = Latin "vir(i)dis"
>>
>> Why once "e" and once "a" , why once "d" and once "z" ?
>
> If you created green.lex with the two Latin words (minus the
> braketed 'i'), and remember to have an end of line after last word,
> and ran:
>
> sounds -p green romanian
>
> you would see the answer.

Well, there is room for improvement of that set of rules since
you didn't include the T (= /ts/) among the consonants, no rules
for derinving it, no rhotacization of -l-, etc. -- that is one
should not expect to get always the correct answer. On another
hand, you have done some good work. In this particular case, it
gives the correct derivation. And as you say:

> Effectively, the final /a/ makes all the difference. It does
> it by yodicising the last /i/

That is: /i/ becames a yod (/j/ in this notation) which is
responsible for alteration of /d/ in /dz/ and further in /z/
in Daco-Romanian. Without yodicising, /i/ > /e/.

> and then breaking what was the first /i/, which is then
> smoothed by the initial /v/.

Actually, the rule is like that: we have the first /i/ > /e/
accomplished somewhere around primitive common Romanian; then
stressed /e/ > /ea/ according to the late rule: before /a/ or /&/
in the next syllable (as proven by early Slavic loans: /mrena/ >
/mrean&/, /tSeta/ > /tSeat&/). Finally, /ea/ gets simplified in
/a/ (the tendency is attested since 1219: the toponym Fata instead
of the primitive Romanian "Feata", most monophtongations were
already accomplished in the XVI-th century).

Regards,
Marius Iacomi