Re: Existence of PIE

On 2008-01-26 23:13, Richard Wordingham wrote:

> I'll hazard a guess at what the reasonable doubts are based on, though
> I think they're largely nit-picking.
>
> 1. The reconstruction of Proto-Indo-European is comparable in
> principle to the reconstruction of Proto-Romance. However, some of
> the words reconstructed seem distinctly late, such as the word for
> 'bishop'.

Not the best example, I think. <episcopus> 'overeseer' is attested
already in Classical Latin, and why should not the Greek ecclesiastical
term have been adopted into Latin early enough to become a valid
Proto-Romance lexeme?

> 2. It may well be that Indo-European languages derive form
> recombinations of dialects, and that there is no single ancestral
> dialect that could be identified as 'PIE'.

That almost goes without saying. Proto-Romance was not a single dialect
either. If fact, any language with enough speakers to expand and split
up is bound to have dialects; that's part of the normal condition of
natural languages. It's also obvious that the comparative method gives
us an artificially smoothed-out reconstruction. At the level of
dialectal networks the family-tree model doesn't work. Still, the
comparative method takes us back to a time when IE was a single
LANGUAGE, if not a homogeneous dialect.

> 3. PIE may well have had more phonemes than we reconstruct, because
> some phonemes have merged identically in all dialects. Again, going
> back to Proto-Romance, is it historically correct to say that it
> lacked a contrast between /a/ and /a:/? The only trace left is accent
> placement, and that may be ascribed to Pre-Proto-Romance.

> 4. We don't actually know the sounds of PIE phonemes! The
> identification of the 'voiced aspirates' as such may actually be a
> hangover from the old view that Sanskrit was particularly close to PIE.

The fact that the reconstruction is possibly incomplete doesn't mean
it's false. As for phonetic identifications, they may be disputable even
for excellently attested languages like Old English, not to mention
Hittite. We know that *dH contrasted with *d and *t in terms of
phonation, and that allows us to account for the recurring, systematic
contrasts observed in the daughter languages. The details would be nice
to know, but are of little importance as regards the validity of the
comparative method.

> I don't buy Trubetskoy's concept of 'mutual Indo-Europeanisation'
> beyond the possibility that some Indo-European features may have
> arisen and spread after the initial divergence - especially in the
> crown clade of extant IE languages. Terms for chariot terminology
> (should that be wagon terminology?) are a particularly awkward issue,
> as the meanings may well not have existed until after the initial
> divergence.

I agree here, and it makes me wary of relying too much on linguistics
palaeontology, especially when the lexical items involved refer to
easily borrowable technology.

> I'll happily accept Hittite as Indo-Hittite. Whether or not it is
> Indo-European is a terminological convention. That brings us to a
> fifth niggle over PIE - many roots are identified as going back to PIE
> even though there is no evidence that they go back to Proto-Indo-Hittite.

This ambiguity is one of the really vexing problems of IE studies; we're
partly living in a Brugmannian Indoeuropia where most verbs roots have
*bHereti-type presents. Unfortunately, in terms of documentation,
Anatolian is a very poor relative of the other (usually unnamed)
first-order branch, so if there's no Hittite or Luwian evidence for
something, an accidental attestation gap is often likely. Anyway, this
problem, real as it is, does not invalidate either the reconstruction of
PIE in general, nor the recognition of the Anatolian and Tocharian as
related to the crown group.

> While there is indeed general agreement over what is Indo-European,
> there are disagreements of varying degrees of apparent crankiness.
> Etruscan is sometimes claimed as Indo-European, though that is a
> minority view that probably depends on the definition of Indo-European
> rather than just the possibly irretrievable facts of its ancestry.
> It's also claimed that one of the Ossetian languages is actually
> Turkic - I think that's just a chauvinistic claim.
>
> Some claims are just plain wrong - Estonian often gets listed as an
> Indo-European language!

If people remember anything about Estonia, it's the fact that it's one
of the Baltic States, from which they infer that its language must be
Baltic as well. However, no _linguists_ think so, while _some_ linguists
have claimed Etruscan as an oddball cousin of Luwian. They are almost
certainly wrong, but being wrong is not the same thing as being ignorant.

> For some languages only known from fragments, there is genuine doubt.
> Pictish immediately comes to mind. That, however, comes from the
> poorness of our knowledge of Pictish, rather than from any lack of
> understanding of Indo-European.
>
> However, are there not some genuinely doubtful members of the family?
> (You may fairly regard this as nit-picking.) Are Tok Pisin and
> Sranan Germanic languages? I presume Afrikaans is a pukka Germanic
> language.

True creole languages are chimaeras. This is another case when the
family-tree model is inadequate to describe the relationship of, say,
Tok Pisin to Germanic. TP does contain a significant Germanic component,
and if in the distant future TP happens to expand and break up into
daughter languages, the resulting family will inherit that component. I
doubt, however, if the historical linguists of the future will
reconstruct the grammar of Proto-Germanic differently from us because TP
will be taken into account. TP grammar is too obviously non-IE, the
etymology of function words (bilong, long, stap) and suffixes (-pela,
-im) is pretty transparent and betrays their pidgin origins.

Piotr