From: dgkilday57
Message: 63636
Date: 2009-03-23
>I made several errors in that posting. First, the presumed impersonal marker is simply */r/, not syllabic */r./. Italic *-tor and *-ntor were originally secondary suffixes (in Umbrian, they contrast with primary -ter and -nter) and the /o/ comes from the secondary marker, not a vocalized sonant.
> > > It's my contention that PIE originally had only an impersonal, by
> > > nature consisting of 3sg and 3pl, and that this was supplemented
> > > later (except in Sabellic) by constructing 1st and 2nd forms, to
> > > form a full medio-passive paradigm.
> >
> > My view is that Old PIE had an impersonal pronominal marker *r.
> > which combined with other pronominal markers to form a true passive
> > voice. This was likely before the origin of PIE ablaut, so I write
> > the impersonal as *bhr.-r. 'one bears', the 3sg. pass. as
> > *bhr.-t-r. 'one bears him/her' = 'he/she is borne', the 3pl. pass.
> > as *bhr.-n.t-r. 'one bears them' = 'they are borne'. A striking
> > morphological parallel is found in Middle Egyptian, where the
> > impersonal element <.tw> (regarded as an old indefinite pronoun
> > 'one, someone') is followed by the ordinary suffix-pronouns to
> > create a passive verbal paradigm. This is, of course, independent
> > of what I have hypothesised for Old PIE. The Egyptian marking
> > sequence is VSO (as is the neutral order in ordinary sentences)
> > while the PIE sequence is VOS, since the impersonal *r. is the
> > subject of the passive formation.
> A VOS contruction in a SOV language? Hm.Old PIE, not PIE just before the diaspora.
> > In my view Italic inherited both the /r./-impersonal and theAnother error on my part. All the old examples of <estur> come from 'eat' not 'be' and have nothing to do with double-marked passives. Since the root was *h1ed-, this <estur> cannot be a true impersonal.
> > passive based on it, but the passive fell together with the middle,
> > and only the 3sg. and 3pl. true passive forms survived. Latin has
> > a mixed bag of passive, middle, and double-marked forms in its
> > paradigm. Both Oscan and Umbrian have the old impersonal, but
> > Latin replaced it with the 3sg. pass. as noted, "in consilium itur"
> > and the like. I am not sure whether the archaic <estur> represents
> > a true impersonal (i.e. *h1es(t)r. with epenthetic /t/), but it
> > seems more likely that it originated in double-marked perfect
> > passives (for such double-marking cf. "res coepta est geri").
> I don't think a language would need an impersonal and a passive both.Rick and Brian have settled that, I think. Anyhow, in order to justify my theory, I must first explain all the P-Italic simple /r/-forms as active impersonals, not passive impersonals syntactically equivalent to ordinary passives (the usual view found in Buck, Poultney, etc.). Then I need to show that certain other /r/-forms in other branches of IE can be explained in a similar way, and finally to argue that taking the simple */r/ as impersonal in origin, not a 3pl. pass. marker which sometimes developed an impersonal sense (thus Buck) gives simpler results. That will take a week or two.
> Anyway, here's my version.I am not willing to postulate such violent phonology, both -i (primary 'here and now') and -r (impersonal) from *-en, even in Old PIE.
> PIE verb stems were originally also nominal (there might have been nominalizing now lost suffix). To nominal elements, thus also to verb stems, could be added the three deictic particles PPIE 'nu' "at me", 'sa' "at thee" and 'ta' "at him/her/it". The latter, in PIE -tó-, gave the impersonal 3sg preterite. PIE forms presents from that by adding either -i or -r, I suspect both are the postposition *en, so that present forms are originally participial, cf French 'en parlant ...', which by some creolizing stage became finite, cf. those sub-standard Englishes which leave out the copula in the progressive tenses, making -ing a finite suffix.