Linguistic PIE

I'll repost the same image I did yesterday, but in a slightly bigger format (although the image is the same; if you click on it, you get the same sized version), and then talk about it and the article I mentioned yesterday, with it's working theories.

Let's reiterate, shall we? There are several known branches of Indo-European that we know enough about to classify (as well as several other languages that we don't know enough about to classify because they were parahistorical; going extinct without ever being written down in more than a few words, phrases, names or glosses, but were mentioned by literate neighbors like the Greeks or the Romans.) These branches are:

1. Anatolian
2. Tocharian
3. Albanian (from Illyrian?)
4. Celto-Italic (later branching into Celtic and Italic specifically)
5. Greco-Armenian (later branching into Greek and Armenian. May have included Phrygian as well.)
6. Germanic
7. Balto-Slavic (later branching into Baltic and Slavic branches)
8. Indo-Iranian, later branching into Indic and Iranian branches)

The chronological branching off of groups from a unified Proto-Indo-European (PIE) has created a confusing mass of modified titles; archaic PIE, common PIE, Indo-Hittite, etc. Koch divides PIE simply by numbers, and it is the following, as shown on the graphic:

1. PIE 1 is the most archaic PIE, and includes all branches of the family.
2. PIE 2 has had Anatolian split off, but includes Tocharian, Albanian, Italo-Celtic, Greco-Armenian, Germanic, Balto-Slavic and Indo-Iranian. According to Koch, this is the classic Pontic Steppes Indo-European.
3. PIE 3 has Tocharian now split off as well, but still includes a Common Indo-European of Albanian, Italo-Celtic, Greco-Armenian, Germanic, Balto-Slavic, and Indo-Iranian. Archaeologically, this would be late Yamnaya or early Corded Ware, I presume?
4. PIE 4 has now lost Albanian (odd, considering that it is located right in the center, or at least only a little to the south, of the Indo-European development central at this point.) It still contains Italo-Celtic, Greco-Armenian, Germanic, Balto-Slavic, and Indo-Iranian. This is supposed to take place before the Bell Beaker phenomena still.
5. PIE 5 has now lost the Italo-Celtic branch, which is archaeologically visible as the rise of the Bell Beaker phenomena. Keep in mind, this isn't suggesting that the Bell Beakers spoke Italo-Celtic, merely that they started a process that introduced enough linguistic isolation from the rest of the Corded Ware and post-Corded Ware community that they would develop in the future into Italo-Celtic.
6. PIE 6 has now spun off the Greco-Armenian branch, and is the last rump of "unified" Indo-European, being left with the Balto-Slavic and Indo-Iranian branches (and probably Thracian too, for that matter, since it appears to be a satem language.) The big innovations that belong to this phase include satemization and the ruki rule. I will point out, however, that the ruki rule is not limited to the languages Koch puts in PIE 6; it does seem to include Albanian and Armenian (possibly as an areal feature?) and there are odd exceptions in Slavic and Indo-Iranian. 
7. You'll notice that Germanic floats above the branching structure and has dotted lines to PIE 4 and PIE 6. This is explained in the working hypotheses below.

Koch then offers a number of working hypotheses, which I'll summarize and discuss very briefly. There's at least one that I think is a poor syllogism, but since it's a faddish theory right now, it can't simply be dismissed without reviewing, I suppose.

1. The Homeland of PIE 1 can't be the Pontic steppes, because Daamgaard's survey of the archaeogenetics of five supposed Hittite speakers have no trace of the Western Steppe Herder DNA component. Therefore, PIE 1 is to be located south of the Caucasus, and they somehow taught their language to the WSH's without transfering any of their DNA. As you can tell by my sarcastic tone, that is just as unlikely as the reasoning used to dismiss the kurgan theory for PIE 1 in the first place, if not actually a fair bit moreso. Not only that, as Davidski points out, there are good reasons for supposing that Daamgaard's "Hittites" were actually Hattians, who may well have spoken Hittite by this point, but weren't the people who brought the Anatolian branch to Anatolia. He also says that he detects a weak WSH signal in some samples from Bronze Age Anatolia, but admits that much more sampling is required to confirm this. In any case, this is exactly the wrong null hypothesis, because the exact same reasoning that suggests that PIE 1 can't have been spoken on the steppes means that it would be impossible to derive PIE 2 on the steppes from south of the Caucasus, since no recent Caucasian admixture is found in subsequent steppe populatons.
2. PIE 2 is the classic WSH cultures; Yamnaya and related. Again, as we suggested above, you can't derive this from anything south of the Caucasus, so the more likey story is that an early branch of this like Usatovo is the source of the Anatolian split-off, and those that remained on the steppes still continued linguistic innovation as PIE 2. This is related to the question of Tocharian, because the imminent splitting off of Tocharian is the defining line between PIE 2 and PIE 3. Three related questions need good answers of yes to make this hypothesis correct:
1. Are the Afansievo and Yamnaya populations genetically similar enough to derive one from the other? The answer here is a solid yes; Afanesievo seems to be genetically identical to Yamnaya.
2. Is the link between Afanasievo and Tocharian secure? The answer here is no, but on the other hand, no other credible alternatives have ever emerged, so it remains the only null hypothesis by default.
3. Was Tocharian the second branch to split off? While the mainstream view is certainly yes, there are a number of workers who dispute this and have introduced alternative explanations.
3. The archaeological Beaker package arose among probably non-Indo-Europeans on the far western coasts of Europe, but was spread by a genetically somewhat heterogenous population of Indo-Europeans with Y-DNA haplogroups derived from the Single Grave Culture. The SGC was awhich was a far western Corded Ware variant and probably early Indo-European speaking phenomena. Early Beakers from Iberia show no WSH genetics, but strong continuity with the Neolithic in the area, although later Iberia was subsequently over-run with male line WSH beakers. The British Isles in particular seem to have had a massive turnover from the Neolithic population to the Beaker population; over 90% of its genetics in the from the Neolithic was replaced by genetics that are largely identical to Single Grave genetics in the Bronze Age. This genetic replacement in both regions, (and in others for that matter) was heavily mediated by male intrusions; much of the female lines appear to remain in some cases; in Iberia, almost entirely (in the British Isles, the female lines seem to be replaced as well.) As the beaker phenomena spread eastward, it encountered more established Corded Ware variants, and although there is interaction, there isn't replacement as there was in the British Isles and Iberia, which were presumably non-Indo-European prior to the Beaker intrusions. This whole process probably initiated a dialect or language split between the far Western Corded Ware linguistic continuum and the rest of the Corded Ware cultures, and the Beakers can be likely credited with created a new branch of Indo-European that now differed from the Corded Ware mainstream. This new dialect probably eventually emerged as Italo-Celtic and Germanic (see below.)
4. This next one is complicated, and has two parts, but which work in harmony. The first is that the Old European hydronymy (river names), i.e., pre-Indo-European river names based on a recognizable root, is confined to the area outside of the Corded Ware culture distribution, suggesting that the Corded Ware is the spread of Indo-European into Europe, and that there was at one point a vast Corded Ware dialect continuum from the Low countries in the west to the Urals in the east, north of the Alps, which excluded what is today France, Iberia, the Italian area, the Balkans, Greece and the Aegean, etc. The Beaker phenomena later created a marked dialectical difference that led to different branches and separated this phenomena along a vaguely defined line running north-south through somewhere in Central Europe. The second part of this hypothesis is that the Germanic languages were initially more closely aligned with the Balto-Slavic group to their immediate East, but cultural conditions caused a "realignment" towards the western group, which gives them simultaneously notable correspondances with Italo-Celtic but also with Balto-Slavic. The genetics seem to bear this out too; the Germanic area shows a high degree of the classic Corded Ware R1a clade of Y-haplogroups (relatively speaking) but also shows a high degree of the classic Beaker clade of R1b. The Germanic area is somewhat unique in sporting both in relatively high pluralities, whereas throughout most of the rest of Europe one or the other dominates much more markedly. Missing from this working hypothesis, however, is any explanation of where the Balkan and post-Balkan languages fit in. While it describes Italic, Celtic, Germanic, Baltic Slavic (and leaves room for Indo-Iranian farther to the east, where it shows more signs of a continuum with Balto-Slavic), where does Albanian come from in this scenario? Or the Greco-Armenian group? To be fair, this is still a problem, and few people address it, prefering to see in the Corded Ware not necessarily the expansion of PIE 3 out of the steppes, but PIE 4 or 6. (This is nonsense that it would be so strictly limited, especially in the early Corded Ware period, since the Beaker phenomena is genetically an outgrowth of the westernmost Corded Ware group, the Single Grave culture, and nobody is saying that the Beaker folks are the ancestors exclusively of the Balto-Slavic and Germanic languages.) The origins of the paleo-Balkan languages are a mess, and most specialists seem to prefer to ignore it and focus on the more northern sphere, probably because it's easier to do, and probably because their own ancestry and linguistic heritage is to be found there. (I mean, even the Balkans today is mostly Slavic speaking, amirite?) As working models by Davidski have suggested, it seems much more likely from a genetic perspective that the Corded Ware is the mother of PIE 3 at least, and that all branches except Anatolian and probably Tocharian can be derived from it.
5. Indo-Iranian comes from Abashevo and other easterly Corded Ware variants, and it's even more easterly later extension to the other side of the Urals, Sintashta, which later led to Andronovo. This is confirmed because Sintashta and Andronovo contain EEF and WHG ancestry which Yamnaya lacked, therefore, the Indo-Iranians can't be derived directly from the Yamnaya steppes; they have to be derived from the Corded Ware. Corded Ware itself can probably be derived from the western portion of the steppes, which spread northwest into the forest steppes and the European plain before spreading eastward again over the steppes. Exactly the relationship between Yamnaya and Corded Ware is unclear; they were genetically clearly very close, but Corded Ware had picked up some Neolithic European admixture, whereas Yamnaya were more "pure" WSH in their genetics. However, the later Indo-Europeans all show signs of this admixture, suggesting that ultimately the Yamnaya were perhaps mostly a dead end after they separated genetically from the Corded Ware, and that the roots of the rest of the Indo-European family from PIE 3 onwards are to be found in the Corded Ware, not the Yamnaya. Granted, I'm adding a bit more than what Koch himself said; he only said that in regards to Indo-Iranian, but as I've got from reading Eurogenes for some time now, I think it's a bit broader than Koch is admitting here. See the map below for early Corded Ware and Yamnaya spheres before Corded Ware expanded.
6. The sixth working hypothesis is a multi-part one, and is in many ways supportive of the Celtic from the West argument. It suggests that since 1) the Beaker cultural package seems to have had some kind of genesis with Atlantic non-Indo-European natives, 2) it was adopted by westerly Indo-Europeans of the Single Grave Corded Ware variant, and 3) the spread of the Beaker culture along with significant genetic replacement by Single Grave male lineages (but in many cases retaining native female lineages) that Celtic in particular emerged out of the interaction of Indo-European with a non-Indo-European substrate. A lot of correspondances can be made between Celtic and Basque/Aquaitanian and Iberian. Celtic from the West is a bigger hypothesis than what Koch is suggesting here, which he knows quite well, but this is just the executive summary portion of a part of the arguments that support it. Personally, I'm finding myself more and more in favor of it over time. Given that Koch is also one of the proponents of this theory, its no wonder that he put a working hypothesis in that supports it.
7. Working hypothesis 6, along with other details, suggests that Basque and probably related languages like Aquitanian and maybe Iberian, which don't exist anymore today and only barely registered briefly in the Roman and Greek historical records, was there already when the first Indo-Europeanization of the Atlantic region started. While the paleohispanic linguistic situation, prior to the Latinization of the peninsula, is not as complicated as the paleobalkan linguistic situation, it's still got its share of mysteries. Some languages are obviously Celtic, some are Vasconic (Basque, Aquitanian, and possibly Iberian, although this is disputed), some may be para-Celtic, i.e., a separate branching of the Italo-Celtic branch, but linguistic dead-ends and unknown to us today, or perhaps some other type of ambiguous Indo-European, and some may be non-Indo-European, but also non-Vasconic. Some may even have a North African antecedent; we know from historical sources that the Carthaginians were very active in pre-Roman Hispania, after all. While Koch himself is the proponent of the idea that the Tartessian language is a Celtic one and its meager carved inscriptions can even be translated, this isn't widely accepted, and the mainstream opinion would be more likely to see ties to Iberian and Aquitanian. If true, and that's a big if because little is known about any of those languages really, then it would imply a larger Vasconic linguistic presence in Iberia prior to Indo-Europeanization. This begs the question, as Koch notes, that maybe Vasconic was much more widely spread than even that. While Colin Renfrew's Anatolian hypothesis never really worked from a time-depth and linguistic period, and genetics has pretty much killed it stone dead (it is an EX-hypothesis!) he was right in pointing out that the Neolithic expansion of Anatolian farmers who became the Early European Farmer (EEF) genetic clade and which spread into Europe on two fronts (along the northern Mediterranean coast in the Cardial Ware horizon, and up through the Balkans to Central Europe and beyond in the "Old European" Balkan cultures like Vinča and Starčevo clusters and the vast LBK horizon) probably brought a single language, or at least tightly related language family with them which had originated in the Mesolithic/Neolithic boundary in Anatolia. Is Basque the sole survivor of this linguistic spread ~5500 BC? While certainly plausible, the fact the WHG DNA survives in appreciable numbers into the Bronze Age in Iberia and elsewhere in Europe (even having a resurgence in the Funnelbeaker and Globular Amphora cultures, among possibly others) means that we can't rule out a Mesolithic ancestry for this language group. Koch suggests that more work be done (although where more material to work with is unknown) in trying to find correspondences between known languages like the non-Indo-European paleohispanic languages, Etruscan, Minoan and Eteo-Cretan, and non-Indo-European languages of Anatolia like Hattic and Hurrian to see if anything can be untangled here. But he's clearly suggesting a working null hypothesis of a Vasconic family being associated with the Neolithic farmer spread, which survives today only in the form of Basque in the Pyrenees. While this is fine, few people suggest an Etruscan and Vasconic link, and Etruscan is a reasonably well-known non-Indo-European language of Europe, and may be part of a more wide-spread Tyrsenian language family. How do Tyrenian and Vasconic compare? Are there connections? According to most linguists, no, none. Both are isolates. But this is known to be speculative, and Koch is calling for more work and research. Honestly, the tapestry of Neolithic cultures of Europe and their relationship to each other still needs some work period. It is possible that the two waves of EEF expansion brought different language families; the Cardial Ware being more Tyrsenian, for example, and the Starčevo to LBK being Vasconic, or something like that, or one or the other may be an older WHG holdout, or something else entirely may be happening too. Maybe Vasconic's origins are to be found in North Africa across from Iberia. Maybe the Etruscans are actually intrusive to Italy and come from somewhere else more recently than the Neolithic spread, as Greek writers seem to suggest, linking them with the Pelasgians and other far westerly Anatolian and Aegean peoples.

It's difficult to say, since genetically as far as we know, the EEF group forms a pretty solid cluster, not a widely spread group, although some admixture is known to have happened, and the Mediterranean was a conduit for travel for a long time in both the prehistoric and historic eras, bringing people from Anatolia, Southern Europe, North Africa and the Levant into association. But one people can't be expected reasonably to be responsible for the spread of two completely unrelated language families, as well as to originate in an area known for two other completely unrelated isolate languages—Hattic and Hurrian. How many language groups were lurking about in Anatolia anyway, all spoken by genetically indistinguishable people in a small geographical area?

As Koch says, there's plenty of tantalizing hints of stuff going on, but very little real knowledge. More work, if something to work with could be found, would be most welcome in untangling this mess. What would be especially helpful is if enough new Linear A (presumably Minoan) and Vinča script work could be found that those languages could be partially deciphered and studied with regards to their potential relationship to other known languages. But don't hold your breath for that to happen.