Re: Family terms [was: Kluge's Law in Italic?]

W dniu 2012-02-13 22:03, dgkilday57 pisze:

> I do not see how the laryngeal in *p&2te'r- can be syllabic if the one
> in *st&2-to'- (I prefer *st&4-to'-) is non-syllabic. And if Olsen
> explains Italic *-flo-/*-fla:- on the basis of *-&1/2tlo-/*-&1/2tleh2-,
> she has another can of worms in Latin <status> and all the rest not
> reflected as *stafus etc.,

You mean *stadus < *staTos, don't you? The "ruBl" rule isn't applicable
here.

> since she effectively posits her *-tH- from
> *-&1/2t- falling together with *-dH- in Proto-Italic.

In Olsen's scheme, *st&2tós has a syllabic allophone of *h2; at any
rate, it doesn't follow a vowel or a syllabic sonorant, which I think is
the required condition. Therefore, Olsen does not predict *stadus
instead of <status>

> Olsen's whole scheme looks like a way to justify denying
> *-dHlo-/*-dHleh2- by lumping the reflexes together with *-tlo-/*-tleh2-,
> and while this may be ingenious per se, the ramifications are starting
> to look as complicated as a system of Ptolemaic epicycles. I think the
> Copernican thing to do is to accept *-dHlo-/*-dHleh2- (which need not be
> primary, since *-dH- was a root-extension and *-lo'- a suffix).

If you are interested in her full argument (which won me over), I have
to refer you to her own work:

Olsen, Birgit Anette. 1988. The Proto-Indo-European Instrument Noun
Suffix *-tlom and its Variants. Copenhagen: Royal Danish Academy of
Sciences and Letters.

-- 2008. "How many suffixes did Proto-Indo-European have?". In Jens
Elmegård Rasmussen and Thomas Olander (eds.), Internal Reconstruction in
Indo-European: Methods, Results, and Problems, Copenhagen: Museum
Tusculanum Press, 187-204.

One of my problems with *-dHlo- being really *-tHlo- was Slavic *-dlo-,
but I now accept that the voicing of *t (and, by extension, *tH) before
laterals and nasals is practically regular in Slavic (cf. *sedmI <
*setmi- < *septm-V-).

> MnE <sister> (and some of the pronouns) show Danish influence which was
> less pervasive in Chaucer's dialect, so this particular comparison is
> not compelling.

Scandinavian influence accounts for <sister> (rather than *swester), but
not for <father> and <mother> with the "wrong" intervocalic obstruent
(fricative rather than plosive).

> I do not deny the trickiness of kinship terms, of
> course, and I think recomposition did occur in the PIE terms. The basic
> suffix of *&2/4ner- 'man' was probably agential, with *p&2ter- derived
> from *peh2-t- 'watch for a long time, guard and feed, graze' etc.

Why not simply *p&2-ter-, with the normal agent suffix? It would
contrast with the occupational term *páh2-s-tor-. The extended stem
*pah2-s- is well attested (including Hitt. pahs-), which can't be said
of *pah2-t-.

> With
> the fading of the prolongative sense of *-t-, *dHug^H&2ter- 'milker,
> milkmaid' was then formed analogically (root *dHeug^H-, as in Skt. -duh
> 'milking'). Greek in this view absorbed the laryngeal, *-g^H&2- > -ga-,
> as in <me'ga>. The root of 'mother' was *meh2- or *meh4- 'suckle' (Lat.
> <mamma> is dialectal like <Juppiter>, but <ma:milla> shows the Roman
> Latin vocalism; if *ma:ma 'breast' had /o/-grade, it must be *meh4-).
> The double laryngeal in *m&2&2ter- or *m&4&2ter became *-a:- in the
> daughter languages. In my view, only *&4 can aspirate *t (and only in
> Indic; I believe the cluster became the unvoiced fricative *tT in East
> Augmentian (i.e. Armeno-Indo-Iranian), *t elsewhere).

I would be surprised if a globally occurring nursery term like <mama>
(Quechua mama 'mother', Xhosa umáːma 'my/our mother', etc.) had anything
to do with laryngeals. I'm inclined to think that the 'mother' word was
simply something like *ma:ter-, with the *ma:- part "borrowed" from baby
talk, and the *-ter- part analogical to 'father' (babbling lexicalised
and grammaticalised, if you prefer). The fact that *ma: does not conform
to the PIE root-structure constraints is immaterial. Nursery "roots" and
onomatopeias make their own laws.

Of course interaction between children and adults works both ways. In
some IE languages at least, forms like <tata> or <papa> could have
originated as imperfect imitations of *p&2ter-.