The Great Proto-Indo-European Thread's Sequel

[KathTheDragon]

Indeed, but what I meant is that we need a phenomenally better understanding of PIE proper before we can seriously start tackling these questions about what came before in a satisfying way. Just look at all the junk that gets thrown about!

[mèþru]

I recently was skimming Plooth's 2015 paper. Have any of you read it? If so, what do you think of it?

[KathTheDragon]

Two things: 1) Do you mean Roland Pooth? 2) "Pooth (2015)" is meaningless without a bibliography to look it up in. What's the paper, or better yet, what's the link?

[WeepingElf]

I can only guess that meþru means this paper. I don't know yet what to think about it, but it seems very adventurous to me.

[mae]

-

[Pabappa]

i dont believe it either, but i suspect one basis for the claim is the idea that there should be no primitives containing /o/, and that all such stems are derivatives, even if they were formed thousands of years befeore PIE proper. and i guess we cant put in a laryngeal because Hittite shows that there wasnt one ... although dont laryngeals sometimes disappear even there?

as for the voicing, well ... gʷt > kʷt seems fair and might even be known to have occurred elsewhere.

[KathTheDragon]

Judging 'night, evening' to be derived from an adjectival stem *negʷ- 'naked' through a metaphor 'bare of day' or something like that seems like an especially gratuitous example of unnecessary etymology

It does, and I'd rather take the Greek evidence for *gʷʰ at face value, and reconstruct a root *negʷʰ- "to become dark". Such a verb is supported by the existence of Hittite neku- "to become evening". Moreover, on the basis of Hittite nekuz mehur meaning "in the evening" (which, it should be pointed out, is a locative phrase, so nekuz could represent *nekʷti), the semantics of *nokʷts should rather be "evening".

as for the voicing, well ... gʷt > kʷt seems fair and might even be known to have occurred elsewhere.

Regressive voicing assimilation is well-known from across IE and is standardly reconstructed for the parent language.

[mae]

-

[KathTheDragon]

Many o-grade nouns have cognates showing a different vowel. PIE *pod- "foot" but Latin pēs, Sanskrit dāru ~ droḥ "tree", and so on. Most simply assign it to paradigmatic ablaut than mythical remnants of some earlier phonological system with no *o.

[Zju]

What was the public consensus on the Caland system again? Mainstream, or a side theory? How about the Narten present?

[KathTheDragon]

The Caland system is very real. Narten presents are a little controversial, the three main opinions being "they don't exist", "they're a primary present formation", and "they're a secondary present formation". I'm personally hesitating between the first and last, since the evidence is slim.

[Howl]

The problem with ablaut in PIE is that even in the oldest attested languages there was a lot of leveling and innovation. So we end up with lots of different patterns and we can't really tell which ones are archaic and which not. So everybody just reconstructs the ablaut patterns they want to reconstruct and explains away the rest.

[KathTheDragon]

Indeed. Case in point: what was the weak stem for *o-grade stems? *é? *e? *Ø? There's evidence for each of those.

[Howl]

Perhaps there were different vowels *o₁ (~ø), *o₂ (~e) and maybe even o₃ (~é) that merged into PIE *o but with different weak grades. Something like that might also be able to explain the exceptions to Brugmann's law with PIE *o₁= Indo-Iranian *ā but PIE *o₂= Indo-Iranian *a.

[KathTheDragon]

That's, uh... not what I was going for in the slightest. Positing new ad-hoc vowels is the worst idea.

[mae]

-

[Howl]

I have done some research and there is not much within PIE to support this idea. The exceptions to Brugmann's law in e/o ablauting stems can be explained by generalization of the e-grade forms. There is also Cowgill's law in Greek with a handful of examples and many more counterexamples. Miguel Carrasquer Vidal reconstructed a separate o vowel for that, but for me it's not worth it. So all I'm left with is a way to explain different ablaut patterns. There are already too many competing ideas out there for that.

[WeepingElf]

The problem with positing new proto-phonemes for every example of idiosyncratic correspondence is that we quickly get a huge behemoth of a phoneme inventory, when we don't need to--languages in the present are chock full of variation, inter-dialect borrowing, etc. and there's no reason to believe that languages spoken thousands of years ago were different in this regard. We of course should not propose an *a2 for Proto-Germanic on account of modern english [wʌn] < *ainaz vs. [own] < *aiganaz.

One sometimes meets gargantuan phoneme inventories in reconstructions. This page lists 166 consonants for Proto-NWC, and last time I checked Bomhard had 50 consonant phonemes for Proto-Nostratic. If you have exceedingly many phonemes in your protolanguage, this is either a sign that you have missed many conditional sound changes (as in the Proto-NWC example), or you are trying to compare unrelated languages (as probably in the case of Proto-Nostratic).

[Salmoneus]

166 is a lot, but 50 is hardly extreme. Many modern NWC languages have in the 45-70 consonant range.

[jal]

166 is a lot, but 50 is hardly extreme. Many modern NWC languages have in the 45-70 consonant range.

The thing being, imho, that the larger an inventory is, the greater the chance that it's reduced instead of expanded in decendents, especially over millenia. I don't buy complex, large inventories being preserved over such a swath of time.


JAL