On Sat, 20 Jun 2009 15:36:54 -0400, John Cowan wrote:

> Jörg Rhiemeier scripsit:
> > Certainly, these languages have changed since the time when
> > Indo-European and Uralic displaced most of the other languages spoken
> > before their expansion, but if we assume that they are more or less
> > representative examples of pre-Indo-European languages, we can say a
> > bit about the typology of the pre-IE European languages.
> This assumption is radically unsound: languages change typology rather
> freely.

Indeed they do; the notion that large-scale typological trends (such
as clines encompassing entire continents) tend to be stable *has* been 
questioned.  But I said that "if we assume that they are more or less
represantative examples of pre-Indo-European languages, ...", and that
is of course quite a big "if".

> > Basque and the Caucasian languages are highly synthetic bordering on
> > the polysynthetic,
> In fact, Basque is rather transparently agglutinating: while certainly
> synthetic in the sense that it has lots of bound morphemes with
> grammatical functions, it is nowhere near polysynthetic.

I think that *no* European language is polysynthetic in the way many
North American languages are, but most Caucasian languages are more
synthetic than the "old-style" IE languages, as far as I know.  Basque
is also highly synthetic, though it tends towards analytic constructions
involving auxiliary verbs in the verb system.

> > Basque and most Caucasian languages, but not Etruscan, are ergative,
> > so there probably were many ergative languages in pre-IE Europe.
> European IE languages are not ergative, but lots of Indo-Iranian ones are
> at least split-ergative (these may even constitute an absolute majority
> of surviving IE languages, though ergativity is certainly a secondary
> feature that cannot be reconstructed for either PI or PII).

Sure.  I was speaking about languages that were spoken in Europe *before*
the spread of Indo-European, not at all about Indo-European languages!
Please read with more attention.  (And in Indo-Iranian, it is a recent
development.  Sanskrit and Avestan show no trace of ergativity.)

But I'd say that only *some* of the languages in question were ergative.

> > How many languages and families were there?  A good comparison for
> > Neolithic Europe probably is pre-colonial North America.  In the area
> > that is now the United States and Canada, there were at least 300
> > languages falling into about 50 families.
> Don Ringe's guesstimate for pre-IE Europe at
> is 60 languages
> representing 40 families grouped into 30 stocks.
> > Europe is about half the size of that area, so we get something like
> > 150 languages in 25 families.  Of course, there is a large margin of
> > error in this estimation.
> I think you underestimate the amount of isolates and microfamilies.

Do I?  I indeed think there were many isolates and microfamilies, but
also some larger families.  Anyway, it is difficult to say how many
languages a family has in absence of a written standard.  What is a
language, and what is a dialect?

> > A group of languages for whose "weirdness" a substratum is often held
> > resposible are the Insular Celtic languages with their VSO word order
> > and initial mutations.  Some scholars (including the aforementioned
> > Vennemann) assume a Semitic substratum here, but there is actually no
> > valid reason to assume that.  Most likely is a Hesperic language that
> > had developed a verb-initial order, possibly from contact with a pre-
> > Neolithic language group.
> The *most* likely hypothesis, in the sense of Occam's Razor,
> is that VSO developed spontaneously within the Celtic languages.

Sure.  Also, the initial mutations can be accounted for without
having to take recourse to a substratum: they are the results of
fairly typical sound changes that just happened to operate across
certain word boundaries.  It is just the fairly rapid restructuring
of the Insular Celtic languages that raised the idea of a substratum

> > The consonant inventories of modern European languages seem to follow
> > an east-west cline, with more copious inventories, especially regarding
> > sibilants and affricates, in the east.
> The Goedelic languages have lots of consonants, a trend which has much
> to do with episodes of palatalization (the more episodes, the more
> consonants, roughly speaking).  Tocharian and Anatolian, which died
> out early, had no such episodes and are not especially consonant-rich
> despite their easterly positions.

You are right.  The "east-west cline" thing is a simplification; there
*are* languages that do not follow that trend.

... brought to you by the Weeping Elf