Post by DKleinecke Post by Daud Deden
Words vs rules: 2 PIE trees
The article seems to me as good as any of the others. There
is nothing new. An Italo-Celtic branch has been long suspected.
Even Greek-Armenian isn't surprising any more. And everything
except Tocharian makes geographic sense.
A decent research project could be based on this tree. The
first step would be reconstructing Greek + Indo-Iranian with
the help of Armenian. This (not a surprise) is what the
early 19th century PIE scholars did. Then reconstruct the
German + Balto-Slavic node which seems to be generally
neglected and thirdly bring these nodes together. I suspect
more work is needed on Italo-Celtic and suspect Tocharian
will add little to it.
So I would have 5 PIE's
PIE1: Greek + Indo-Iranian
PIE2: German + Balto-Slavic
PIE3: PIE1 + PIE2
PIE4: PIE3 + Italo-Celtic + Tocharian
PIE5: PIE4 + Anatolian
I think Arnaud would add
PIE6: PIE5 + Hurrian
I think the best name for PIE2 would be Proto-Baltic.
A rather simple and objective way of classifying IEan languages is to sort them according to the consonantal contrasts they exhibit.
A number of languages reached a four-way system, with the addition of a fourth series based on phonotactic *C-H.
This group includes Indo-Iranian, Greek, Armenian and Germanic.
There is no indication any of the other subgroups took part in that innovation.
This group is the Proto-Sanskrit-Improved core, on which most Indo-Europeanistic works are based, until now.
But for the famous centum vs satem issue, Armenian and Germanic are basically an evolved form of Indo-Iranian, with rearrangements of voiced and aspirated features.
This means that Germanic has little to do with either Balto-Slavic and IMO even less with Italic. But I agree that Germanic must have been in contact with Balto-Slavic, as these two groups have a number of shared features, for example as regards case-markers.
The rest of the IEan languages did not undergo the creation of the fourth series, so they share a preserved conservative feature.
Besides, one issue is the number of contrast Pre-PIE really had. I tend to think that Pre-PIE had only two series: undistinct *t/*d vs so-called voiced aspirate *dh.
All IEan languages agree on the way *t/*d split into two series, though we cannot affirm with certainty that the contrast existed in Anatolian, word-initially, because Cuneiform is somewhat clumsy and ambiguous.
Celtic and Albanian fused the three series into only two, but in a way that is not the original two series of Pre-PIE. There's no indication Celtic, Albanian and Balto-Slavic ever had aspiration as a feature.
Only Balto-Slavic and Italic preserve traces of the three series, though in a different manner. In all cases, Italic does not behave like Celtic here, rather like Greek.
So, to summarize, there are four innovations:
1. one is the creation of a fourth series *C+H > voiceless *Ch
this characterize Indo-Iranian, Germanic, Armenian, to a lesser extent Greek. This innovation is the most recent one.
2. one is the introduction of aspiration
this innovation prepares Innovation1
It also involves Italic
3. the fusion of three series into only two (voiceless vs voiced)
It involves Celtic, Albanian, and Anatolian.
In a way, this group went back to the two series of Pre-PIE, but with the different distribution of the features voiceless vs voiced among consonants.
4. the split of voiceless into two series (possibly voiceless vs glottalized), the other series (so-called aspirated voiced) being maybe voiced or the like.
This split created the phonological matrix of PIE, with three series.
This feature is shared by Hurro-Urartian.
It can be noted that Uralic does *not* exhibit any sign of that split, and there is no doubt that Hurro-Urartian is closer to core PIE than Uralic is.
My PoV is that Nostratic had only two series.
I'm not familiar enough with Tocharian, but one issue would be to determine if Tocharian even knew a three-way series. In all cases, every feature (2 or 3) collapsed into one in that language.