[tied] Re: -s -> -i

From: richardwordingham
Message: 14473
Date: 2002-08-23

--- In cybalist@..., João Simões Lopes Filho <jodan99@...> wrote:
> But in Greek the rule is not exactly the same:
> -es- > -ei- (dialectal -e:-)
> -os- > -ou- (dialectal (-o:- or -oi-)

Nor is the rule uniform in Sanskrit.

While it works in the weak persons of the perfect, e.g. for
sad 'sit', we have 3s. sasa:da, 3pl. se:duh. < *sasdur, it does not
work in other cases.

1. For the past participle and infinitive of roots that appear once
to have ended in jH:

vah 'carry' (< PIE weg^H) has past participle u:d.Ha, infinitive
sah 'endure' has past participle so:d.Ha, infinitive so:d.Hum

Note 'o:d.H', not 'e:d.H'. (The past participle is formed from the
zero grade, the infinitive from the gun.a - i.e. IE e/o grade.)

The evolution of the cluster in these verb forms is g^Ht > jHt > jHdH
> z.d.H > d.H, with lengthening or diphthongisation of the preceding
vowel at the final stage. Any PIE assimilation and its undoing has
been omitted from the sequence for clarity. (For similar unvoiced
clusters, the evolution is k^t > s't > s.t. )

2. The external sandhi of final -as before voiced consonants (and
before case suffixes in bh-) is -o:, for which one might reasonably
reconstruct an intermediate sandhi form *-az.

There seems to be a lot of variation in what happened to vocalised
sibilants. Latin and some Greek simply made them a copy of the
previous vowel; Italian unconditional i; Sanskrit seems to have a
complexly conditioned high vowel.


> ----- Original Message -----
> From: Piotr Gasiorowski
> To: cybalist@...
> Sent: Thursday, August 22, 2002 4:39 AM
> Subject: Re: [tied] Re: -s -> -i
> ----- Original Message -----
> From: Miguel Carrasquer
> To: cybalist@...
> Sent: Thursday, August 22, 2002 12:56 AM
> Subject: Re: [tied] Re: -s -> -i
> > > /s/ > /y/ is seen in Occitan and Catalan (vassaletu "little
vassal" > vaslet >
> > Cat. vailet, Gasc. bailet "boy"), especially in Gascon: aine <
asne "donkey",
> > nai < nas "nose", nui < nus "we". There is of course also
Ancient Greek *esmi >
> > eimi (but <ei> can stand for /ey/ or for /e:/ with compensatory

> Compare also Skt. *z/*z. > *y as in *niz.d.a- > *niyd.a- > ni:d.a-
and *mazdHa- > *maydHa- > medHa-.