Re: [tied] Timing of ablaut

>
> Specifically, if the root is indeed *styeH rather than *steyH,
> stya:yate: can be *styoH-ey-e-toi.

That is what I discarded to begin with on the basis of Goth. stains and
Gmc. *{th}ai-sman- 'leaven' (OHG deismo), which latter may form a paradigm
with Gk. sô:ma (*[s]táyH3-smn, *[s]tiH3-smén-s 'compact mass'), which then
formed the o-infix derivative *stóyH3-no- 'made of a compact mass'.

> There's the other s-stem anas- (Lat. onus), with *h3o- (*h3e-) before a
> voiced consonant.

I believe this word had a laryngeal also after the /n/.

> I agree with your previous statement that páti- is analogical (although
> for different reasons: I think the original nom.sg. of the word was
> *pótyo:n, as preserved in Toch.B. petso).

The Toch. form reflects transfer to the productive oy-stem type which had
nom. in *-o: just like the n-stems.

>
> káti, práti, katará are indeed problematical.  I don't believe in
> the "Kleinhans (or Pedersen) formulation": Brugmann's law
> applies to *o (but not *h3e) in _all_ open syllables (except in final
> position).  Perhaps the last parenthetical condition can be invoked
> here, and we have to analyze *kWo + *ti, *pro + *ti and *kWo + *teros,
> i.e. these words were formed _after_ final *-o had already been
> shortened.

That is not logical if the words are PIE, and Brugmann's law is
Indo-Iranian only. Are you denying one of these statements? The general
account says that káti is either a replacement of *cati (Avestan caiti)
or, like, prati, an antevocalic sandhi variant katy, praty (like Gk.
prós). For katara- one might think of influence from a lost variant with
syncope (Lith. katràs), or simply of lack of application of the law in
words of less than full stress.

> The treatment of Brugmann's law in Collinge ("The laws of IE")
> has six pages of this back-and-forth, ranging from Burrow's assertion
> that *o _always_ gives /a:/ in Sanskrit (even in closed syllables!),
> through Brugmann's own initial formulation (*o > a: in open
> syllables), the Kleinhans amendment (*o > a: in open syllables,
> provided a nasal or liquid [voiced segment?] follows), to Hirt's total
> dismissal of the law.
>
> I am with Brugmann (the Brugmann before he retracted his own law). 
> The rule is simply that *o gives *a: in open syllables, no matter what
> the following consonant.  As to the concrete examples and
> counterexamples given in Collinge's discussion, I have the following
> remarks to make:
>
> pá:dam (*pod-m.) -- this incorrect: Brugmann's law predates vocalic
> resonants, and thus *podm has a closed syllable.  The length here
> has been carried over from the nom.sg. *po:ds.

I would like to see some *real* evidence that there was no lengthening
before the reflexes of syllabic resonants.

> bharama:n.a- (*bher-o-mh1no-) -- Not a counterexample at all: *o is
> followed by three consonants (Brugmann's law predates the vocalization
> of laryngeals).

Under the present understanding¨of laryngeal phonetics this indeed had a
closed syllable if syllabic /H1/ was realized as a sequence of a short
spirant + an ultrashort central vowel.

> ápas- (*h3ep-os/es-) -- OK, *h3e- not *Ho-.
> ávi- (*h2owi- or *h3ewi-?) -- If *h2owi-, analogical after oblique
> *h2awy-. páti- (*poti-) -- Analogical (it's remarkable that outside of
> Sanskrit, the analogy has worked the other way, generalizing strong
> *pot-i- over weak *pet-y-).
> páda: (ins.sg.) -- Not a counterexample at all, because reflecting
> oblique *ped-.
> vá:cam (*wokWm.) -- See pá:dam: analogical after nom.sg. *wo:kWs.
> cakára (*kWe-kWor-h2a) 1sg.pf.
> caká:ra (*kWe-kWor-e) 2sg.pf. -- as explained by Kuryl~owicz: the 1sg.
> has a closed syllable (ending *-h2a), the 3sg. an open one  (ending
> *-e). ánas- (*h3enos) -- As ápas-
> rátha- (*rotHos) -- Brugmann's law predates the formation of voiceless
> aspirates: *rotH- makes a closed syllable.
>
> Interesting is the sentence "Viredaz (1983) links PIE /e, o/ with
> North West Caucasian /a, a:/, suggestively, as the basic minimal
> vocalism in each case", which predates my own insight by seventeen
> years or so.  This of course means that the nature of Brugmann's
> law is interpreted very differently by me.  It is not a lengthening
> of *o in Indo-Iranian under certain circumstances, but the retention of
> length (*o < **a:) in Indo-Iranian under the same circumstances (open
> syllable).  Other phenomena suggestive of the same thing are the
> facts that *o is not coloured by laryngeals (just like *e: isn't), and
> that in Tocharian, *o does not behave like a short vowel (the short
> vowels *e, *i and *u are all reduced to Pre-Tocharian *ä, while *o gives
> /e/, as does *e:).  There is to my knowledge no evidence in
> Tocharian for a distinction between *o and *h3e, as there is Skt., but
> then again *h2e/*a in Tocharian is also not reduced to *ä, but surfaces
> as /a:/, so that is consistent.

Are you applying Brugmann's law to a prestage of the IE protolanguage? Did
PIE *dór-u, *g^ón-u have a longer vowel than *H3éd-os 'smell', *H3ékW-iH1
'eyes'?

Jens