Indo-European language varieties

[Nortaneous]

The Anatolian word for wheel comes from a different root than any other branch, but the other branches show two different roots, so clearly there was variation within any stage of PIE.

for lexical isoglosses there's at least also *meyu- 'four'

[hwhatting]

and the only really important non-phonological innovation is the feminine gender. That's a big deal, but is that the result of a thousand years of separation? Consider:

That very much depends on your reconstruction of PIE grammar. Some points:
- Anatolian doesn't have a reduplicated perfect: actually, it doesn't have the IE perfect at all. What it has is a separate mi- and hi- conjugation. Perfect, mediopassive and hi-conjugation share an original basis, but Anatolian and Core IE developed different verbal systems out of that original basis, with Anatolian only going along with core IE in creating a mediopassive.
- Anatolian doesn't have the formal aorist-imperfect opposition that Core IE developed out of an original inherent verbal aspect.
- Anatolian doesn't have the standard thematic verb class with full grade in the root; IIRC, it only has -ye- and -sk'e- stems. It also doesn't have a subjunctive based on thematisation. The consensus nowadays seems to go in the direction that both only evolved in Core IE, and IIRC Tocharian shows a transitional stage, with relatively few full-grade thematic stems.
- Anatolian doesn't have the optative in (athem.) -yeh1- / (them.) -oih1-
- Anatolian seems to show that the 2nd sg. pronoun was originally *ti- in the nom. and *tu- in the oblique cases, while Core-IE has *tu- in all cases.
- Anatolian shows a much more rudimentary case system, especially in the plural; it looks like the plural case forms in -bh-/-m- are an innovation of Core IE.
People reconstructing a PIE closer to the traditional model explain most of these differences as Anatolian losing categories, but IMO (and in the opinion of many modern IE scholars) it is much likelier that PIE was quite different from the traditional model and the traditional model is actually only applicable to Core IE.
The Mycenaean Greek texts are only a couple of centuries younger than the average Hittite texts, and Vedic is only a couple of centuries younger than Mycenean. But if you compare the three, Greek and Vedic show much more similarities between each other; but even they wouldn't have been mutually comprehensible anymore; they're farther apart than the average Romance languages. And Anatolian is far more distant from both of them.

[Moose-tache]

Thank you for the informative reply! I presented this question as a discussion point, not something I will take to my grave, so I appreciate you adding some counter arguments.

Some of these I think are slightly unconvincing. For example, the fact that the Anatolian hi-conjugation only formally matches the PIE perfect without the same semantics/usage doesn't strike me as very damning, given how every branch of PIE has either simplified, merged, or abandoned large chunks of PIE morphology, both in terms of form and meaning (see my comment on the optative below). And the cup game of Greek, Aryan, and Anatolian is a bit of a sneaky argument, since it is an open question whether or not Hellenic and Indo-Iranian (and Armenian) shared a brief time as a separate sub-branch or at least heavily cross-pollinating separate branches. If we try this exercise with Vedic, Old Latin, and Hittite, suddenly none of them stands out too sharply as the odd one out. Also, the s-aorist exists in Anatolian, so I'm not sure what you mean by there not being such a distinction. Finally, the PIE optative is also absent in Celtic, and outside the "steppe" group mentioned above the optative is extremely unstable anyway. It started off as a derivational suffix (since it can only be applied to bare roots in some branches), and its use as a grammatical category may have spread during the dialectal period, given how un-optative it gets with distance from the third millennium Pontic steppe.

However, the rest of your arguments seem quite solid. I do think it's likely that Anatolian branched off early, but it's hard to demonstrate from the existing evidence that it was very early. This could be a Tocharian situation, where the split occurred less than half a millennium before the Corded Ware explosion.

[hwhatting]

Some of these I think are slightly unconvincing. For example, the fact that the Anatolian hi-conjugation only formally matches the PIE perfect without the same semantics/usage doesn't strike me as very damning, given how every branch of PIE has either simplified, merged, or abandoned large chunks of PIE morphology, both in terms of form and meaning (see my comment on the optative below).

Sure. But Anatolian comes in early enough (actually, Anatolian has the oldes texts of any IE languages) that it's more likely that it never had a full-blown perfect than that it lost it. And the other IE languages with early attestation all show the perfect or remnants of it, so to me, the perfect looks very much like a common development after Anatolian broke off.

And the cup game of Greek, Aryan, and Anatolian is a bit of a sneaky argument, since it is an open question whether or not Hellenic and Indo-Iranian (and Armenian) shared a brief time as a separate sub-branch or at least heavily cross-pollinating separate branches.

That wasn't my point - my point was that even Greek and Indo-Aryan differ sufficiently that they must have started diverging at least a millennium before attestation, so the separation of Anatolian must have happened much earlier.

If we try this exercise with Vedic, Old Latin, and Hittite, suddenly none of them stands out too sharply as the odd one out.

Well, I don't want to get into the game of dating Vedic, but it's sure a couple of century older than the oldest Latin texts. And it's trivial to derive Latin from the traditional model; even for many things it has lost as categories, it still shows clear remnants. That's not true for Anatolian.

Also, the s-aorist exists in Anatolian, so I'm not sure what you mean by there not being such a distinction.

I know that people have tried to claim that, but never saw a convincing argumentation. Can you point me to any? I know that there are various cases where there are s-stems in Anatolian and -s as a 3rd Person sg. ending in the past in some cases, but there is no s-aorist (or any other aorist) as a category; I'm with those who think that these elements containing -s- are what Core IE built the s-aorist on, not remnants of a PIE category s-aorist existing before Anatolian split off.

Finally, the PIE optative is also absent in Celtic, and outside the "steppe" group mentioned above the optative is extremely unstable anyway. It started off as a derivational suffix (since it can only be applied to bare roots in some branches), and its use as a grammatical category may have spread during the dialectal period, given how un-optative it gets with distance from the third millennium Pontic steppe.

Well, Celtic comes late enough that the optative could have simply been lost, and besides Graeco-Aryan at least Latin, Slavic, Germanic and Tocharian have it, so it looks likely it was there in Core IE. Based on Tocharian, one could argue that the optativ was a past (or non-present) first that was then integrated as a mood in some of the Core IE languages after Tocharian split off.

[Nortaneous]

Thank you for the informative reply! I presented this question as a discussion point, not something I will take to my grave, so I appreciate you adding some counter arguments.

Some of these I think are slightly unconvincing. For example, the fact that the Anatolian hi-conjugation only formally matches the PIE perfect without the same semantics/usage doesn't strike me as very damning, given how every branch of PIE has either simplified, merged, or abandoned large chunks of PIE morphology, both in terms of form and meaning (see my comment on the optative below). And the cup game of Greek, Aryan, and Anatolian is a bit of a sneaky argument, since it is an open question whether or not Hellenic and Indo-Iranian (and Armenian) shared a brief time as a separate sub-branch or at least heavily cross-pollinating separate branches. If we try this exercise with Vedic, Old Latin, and Hittite, suddenly none of them stands out too sharply as the odd one out. Also, the s-aorist exists in Anatolian, so I'm not sure what you mean by there not being such a distinction. Finally, the PIE optative is also absent in Celtic, and outside the "steppe" group mentioned above the optative is extremely unstable anyway. It started off as a derivational suffix (since it can only be applied to bare roots in some branches), and its use as a grammatical category may have spread during the dialectal period, given how un-optative it gets with distance from the third millennium Pontic steppe.

However, the rest of your arguments seem quite solid. I do think it's likely that Anatolian branched off early, but it's hard to demonstrate from the existing evidence that it was very early. This could be a Tocharian situation, where the split occurred less than half a millennium before the Corded Ware explosion.

The Latin verbal system is also extremely innovative. For revising the place of Anatolian within IE I'd expect Germanic to be useful, but I haven't looked into it, and the common development of the feminine still points to to core IE being monophyietic even if some other arguments don't hold.

I don't think core IE minus Tocharian was monophyletic. The "you can't fuck into a building" argument seems weak (probably someone could look at Spanish for this?), Fellner has established that the morphological arguments are also weak, and it seems too convenient for the two most recently discovered branches to also be the two earliest.

[hwhatting]

The Latin verbal system is also extremely innovative.

Yes. There are some old things that don't fit with Graeco-Aryan, e.g. thr r-endings of the mediopassive, which may be retentions that Graeco-Aryan lost, and maybe the ā-/ē-preterites that exist in similar forms in e.g. Balto-Slavic and Celtic, but which still would be developments inside Core IE.

For revising the place of Anatolian within IE I'd expect Germanic to be useful, but I haven't looked into it, and the common development of the feminine still points to to core IE being monophyietic even if some other arguments don't hold.

IMO, Germanic's deviations from the traditional model aren't very old, certainly not turning it into a separate branch on a level with Anatolian or an early split-off like Tocharian (I see that you have your own opinions on this) - it clearly shows the reduplicated present, although not the regularly expected reduplication vowel, it has integrated the optative in the verbal system as mood, it seems to be on the vocalic ending side of the the mediopassive, and to me it looks more likely that it lost the aorist / imperfect distinction than that it never developed it. In my view, Germanic is more interesting for ordering developments inside Core IE than for re-assessing the relationship with Antolian. But I'd be interested in your findings.

[WeepingElf]

I concur with Hans-Werner here. There is good morphosyntactic evidence for Anatolian branching off significantly earlier than the rest, even if the phonology seems to suggest that not very much time has elapsed between Early and Late PIE - the phonological structure doesn't seem to have changed much, even if the phonetic values of the phonemes may have changed.

In my personal model, which I have already laid out elsewhere but shall lay out here once more for easy reference, there was a division between a northern and a southern dialect for some time before the Yamnaya expansions which started around 3000 BC. These dialects were phonologically similar but developed different morphological structures. There were, among other things, three genders and three verb aspects in the north, but only two genders - animate and inanimate - and a simple two-tense system - present and past - in the south. Northern IE spread into Central Europe north of the Carpathians and eastwards into Central Asia; its speakers were the Corded Ware people. Southern IE spread into the Lower Danube area and from there into Anatolia and into Central Europe south of the Carpathians. From the northern dialect, all the known non-Anatolian languages descend, while the southern dialect spawned Anatolian and perhaps a number of lost languages further west including that of the Bell Beaker culture, later to be clobbered by Northern IE languages that became the ancestors of Italic and Celtic (perhaps in the course of the Urnfield expansion). Germanic may be the language of a Corded Ware outlier in southern Scandinavia, cut off by the Bell Beaker expansion into Central Europe. Tocharian is a Northern IE language that lost contact to the rest of the Northern IE dialect continuum at an early date, but surely considerably later than the separation of Northern and Southern IE. The Afanasievo culture of southern Siberia is IMHO a good candidate.

As for Anatolian *meyu-, it is AFAIK not really certain that it actually meant '4'. I have seen suggestions that it meant something like 'a few', or a set of items belonging together like the four legs of an animal or the four wheels of a wagon. Perhaps, it originally had such a meaning and acquired the meaning '4' later (since '4' is a frequent cardinality of such sets). Alas, I am not a Hittitologist, and can't tell how much sense this makes.

[Talskubilos]

much less could anybody make much sense of the texts. There are some words which look like some Basque words (e.g. a set of what appear to be numerals whose number values are unknown, and the place name Iliberi which looks like Basque hiri berri 'new town'),

There're many lookalikes between Basque and Iberian, but much less genuine matches. As regarding town names, some of them have got the element ildiŕ, ildi- (whose Latinized form Ili- has been liked to Basque (h)iri 'town') which IMHO would mean 'tribe' or 'people'.

As a matter of fact, there were several Iliberi(s) in Roman times, but one of them (medieval Elvira in Granada region) was actually quoted as Ilibyrge by Hecataeus of Miletus in the 5th century BC. This reminds me of the Spanish slang pronounciation of Burger King as Burri Quin.

How about iturri though? Is it a wanderwort?

This is one of the few nouns with a fossilized i- prefix in Basque, namely idoi 'marsh', idulki 'plinth, pedestal' or ilendi/ilindi 'tinder'. Furthermore, there's a rarer prefix ti- in tilista, dilista 'lentil', although it's also found in Sardinian. My bet is these are probably related to Berber i- (m.)/ti- (f.) noun prefixes.

[Zju]

Hmm... I must have meant etorri. Not that those couple of examples show that there's anything going on with i- or ti-. At most there may be something going on with itu-/ido- of iturri 'source, spring' and idoi 'marsh', but even that is pure speculation.

[Talskubilos]

At most there may be something going on with itu-/ido- of iturri 'source, spring' and idoi 'marsh', but even that is pure speculation.

I don't think so, because *doi would be related to lohi 'mud', both from an earlier *dokki or something like that.

[keenir]

At most there may be something going on with itu-/ido- of iturri 'source, spring' and idoi 'marsh', but even that is pure speculation.

I don't think so, because *doi would be related to lohi 'mud', both from an earlier *dokki or something like that.

That seems pretty plausible to me...they're all geologically/spatially related, and it doesn't seem a leap to have "source" & "marsh" & "mud" & "spring" flowing from a common waterway.

[Talskubilos]

At most there may be something going on with itu-/ido- of iturri 'source, spring' and idoi 'marsh', but even that is pure speculation.

I don't think so, because *doi would be related to lohi 'mud', both from an earlier *dokki or something like that.

That seems pretty plausible to me...they're all geologically/spatially related, and it doesn't seem a leap to have "source" & "marsh" & "mud" & "spring" flowing from a common waterway.

As a matter of fact, I don't think idoi and iturri are related at all. The former would be related tot lohi 'mud', zohi 'clod of earth' and sokitu 'to dirty; to corrupt', from a 'clay, earth' root found in several language families, including Indo-European. The latter would be in turn linked to Spanish chorro 'jet', with no obvious etymology.

[keenir]

I don't think so, because *doi would be related to lohi 'mud', both from an earlier *dokki or something like that.

That seems pretty plausible to me...they're all geologically/spatially related, and it doesn't seem a leap to have "source" & "marsh" & "mud" & "spring" flowing from a common waterway.

As a matter of fact, I don't think idoi and iturri are related at all. The former would be related tot lohi 'mud', zohi 'clod of earth' and sokitu 'to dirty; to corrupt', from a 'clay, earth' root found in several language families, including Indo-European.

still not seeing a problem, sorry.
and if i fall into a marsh, you wouldn't expect me to have dirtied my clothes? they all seem conceptually and chronologically/behaviorally related to one another.

The latter would be in turn linked to Spanish chorro 'jet', with no obvious etymology.

"jet" as in...?

[Talskubilos]

still not seeing a problem, sorry.
and if i fall into a marsh, you wouldn't expect me to have dirtied my clothes? they all seem conceptually and chronologically/behaviorally related to one another.

That's OK, but the thing is Basque iturri 'spring' doesn't belong here, unlike proposed by Zju.

[Zju]

still not seeing a problem, sorry.
and if i fall into a marsh, you wouldn't expect me to have dirtied my clothes? they all seem conceptually and chronologically/behaviorally related to one another.

That's OK, but the thing is Basque iturri 'spring' doesn't belong here, unlike proposed by Zju.

How do we know that any one of those four words is cognate? It's all speculation based on one or few phonemes.

[Moose-tache]

How do we know that any one of those four words is cognate? It's all speculation based on one or few phonemes.

Hi, welcome to this thread! I see you didn't receive your orientation pamphlet, so I'll bring you up to speed. In here, there is no scientific method, no null hypothesis, no evidentiary burden, and no productive discourse. Instead, we find words that are "similar," in some non-quantifiable sense, and delcare them to be derived form the same Basque monks or Pontic shepherds or whatever.

[keenir]

still not seeing a problem, sorry.
and if i fall into a marsh, you wouldn't expect me to have dirtied my clothes? they all seem conceptually and chronologically/behaviorally related to one another.

That's OK, but the thing is Basque iturri 'spring' doesn't belong here, unlike proposed by Zju.

wait, i thought you said most of those were Basque words to begin with.

How about iturri though? Is it a wanderwort?

This is one of the few nouns with a fossilized i- prefix in Basque, namely idoi 'marsh', idulki 'plinth, pedestal' or ilendi/ilindi 'tinder'. Furthermore, there's a rarer prefix ti- in tilista, dilista 'lentil', although it's also found in Sardinian.

At most there may be something going on with itu-/ido- of iturri 'source, spring' and idoi 'marsh', but even that is pure speculation.

I don't think so, because *doi would be related to lohi 'mud', both from an earlier *dokki or something like that.

yep, like i thought.

[Talskubilos]

How do we know that any one of those four words is cognate? It's all speculation based on one or few phonemes.

As pointed out by keenir, they've got all similar meaning.and the same vocalism, the only differences being consonants.

[Talskubilos]

wait, i thought you said most of those were Basque words to begin with.

That's right, but the only thing in common is they've got a Berber-like prefix *i-.

[WeepingElf]

How do we know that any one of those four words is cognate? It's all speculation based on one or few phonemes.

Hi, welcome to this thread! I see you didn't receive your orientation pamphlet, so I'll bring you up to speed. In here, there is no scientific method, no null hypothesis, no evidentiary burden, and no productive discourse. Instead, we find words that are "similar," in some non-quantifiable sense, and delcare them to be derived form the same Basque monks or Pontic shepherds or whatever.

Well put.