The Great Proto-Indo-European Thread's Sequel

[Pabappa]

Also, while Im not a believer in the glottalic theory myself, I dont think Armenian having /bʰ dʰ/ etc provides any new evidence against it, since they already answer that objection with their derivations of the aspirated stops in Greek and Indo-Iranian. Greek, Indo-Iranian, and Armenian may well form a clade. Their solution is that the aspiration was allophonic in PIE and survived in the east but not in the west. This seems to go back to at least 1973, with Gamkrelidze, Ivanov, and Hopper all proposing the same three series:

https://en.wikipedia.org/wiki/Glottalic ... c_proposal

This makes perfect sense to me. The reason I'm not convinced of the glottalic theory is that there is very little positive evidence for it, and possibly none at all. e.g. we dont see glottalic reflexes anywhere unless we can assume that the creaky voice found in Balto-Slavic is a remnant of that, which I think is a stretch.

[Kuchigakatai]

"Oh, by the way, Greek θεός 'god' is not related to Latin deus 'god', but it is related to fānum 'temple', fēriae 'sacred festival, holidays', fēstus 'joyous [holiday celebrations]; related to the holidays', fērālis 'of a funeral', all of which talk about sacred things. This is part of a larger pattern of Greek theta θ corresponding to Latin f and even English d: τίθημι 'to put' and faciō 'to do' and to do, θύρα 'door' and forās 'out through the doors' and door, θιγγάνω 'to touch sth' and fingō 'to touch sth, shape it' and dough."

Now there's a copypasta for the next time you see someone insinuate θεός is related to deus, as it often happens.


"While we're at it, did you know the Germanic goddess Frigg, Odin's wife, whose Proto-Germanic name was *Frijjō meaning something like 'The Beloved and Free', basically bears the name Priya, a perfectly normal name for women in India today?"

[alynnidalar]

"While we're at it, did you know the Germanic goddess Frigg, Odin's wife, whose Proto-Germanic name was *Frijjō meaning something like 'The Beloved and Free', basically bears the name Priya, a perfectly normal name for women in India today?"

That's beautiful.

[Vijay]

Priya is also a term of endearment in a lot of Indian languages (something along the lines of 'my love').

[mae]

-

[WeepingElf]

Let me say something about PIE migrations.

Maria Gimbutas believed to have found three waves of "Kurgan invasions":

1. Around 4400 BC, into the Danube valley.
2. Around 3500 BC, into the Balkan Peninsula and Anatolia.
3. Around 3000 BC, across most of Europe.

The waves are shown in this map. (The web site where I found the map is totally bonkers, though.)

I used to equate the third wave with non-Anatolian IE, the second wave with Anatolian, and the first wave with my hypothetical "Aquan" group which I explore fictionally in my Hesperic conlang family. But this genetic study has since shown that the first of these three waves did not exist. There was no incursion from the steppe at that time. Also, speaking of a Kurgan wave at that time is an anachronism because people AFAIK did not build kurgans that early, neither on the steppe nor in the Danube valley. They also didn't have domestic horses (the much-discussed horse skull with but wear found at Dereivka turned out to be an intrusion from the Iron Age), so how should they have invaded Europe?

One could perhaps say that even without an invasion, maybe their language spread into Europe back then? I think that is very unlikely. At that time, the people on the Balkan Peninsula were culturally much more sophisticated than the people on the steppe! They already had copper metallurgy, and were close to becoming an urban civilization and would surely have founded the first cities of Europe if those Yamnaya steppe barbarians hadn't spoiled the feast. (Some people even claim they had writing, but I doubt that the Vinča symbols were that.) The steppe was a barbarian backwater then. So why should those semi-civilized people adopt those barbarians' language?

So no IE outside the steppe before 3500 BC. Fair. But the "second" and the "third" waves appear to be real, so we can account for both Anatolian and non-Anatolian IE. And my hypothetical Aquan family may be a third prong on the Anatolian vs. non-Anatolian fork, descending from an extension of the "second" wave into Pannonia.

[Pabappa]

i think PIE had a relatively humble phonology with just eight stops, /l m n r/, and the three laryngeals /h x xʷ/. aspiration and larynbgealization are too similar to be independent processes so i just unite them. vocalic allophones of laryngeals may have been a grammatical process like how /y w/ > /i u/. i read /i u/ as grammatically schwa + semivowel, and the laryngeals in words like /samxdas/ "sand" may have been pronounced as [əx]. if it can be shown that /xd/ > /dh/, the aspiration in this word is unnecessary as well. also i think there was no contreast between /a/ and /o/, so my vowel system is /e a i u/, but there might have to be longs as well if sequences like /ae/ cannot be explained away as coming from a pre- or post-PIE stage.

edit: obviously i left out /s/

[KathTheDragon]

if it can be shown that /xd/ > /dh/

It can't: *sweh₂dus > Skt. svādú-, Gk. ἡδύς, PGmc. *swōtuz, etc. Decomposing all voiced aspirates into voiced stop + laryngeal in whatever order is a) unjustifiable, as you won't find any positive evidence for most of these laryngeals you have to posit; and b) unsound, as everything indicates that the plain voiced stops were the more marked stop series, not the voiced aspirates as your theory would have it.

[mae]

-

[KathTheDragon]

It's used more as a theory-neutral convention more than anything. For example, I believe they were originally plain voiced stops - but I don't call them that because it'd be confusing. "Voiced aspirate" and "breathy-voiced stop" are generally understood as synonymous within PIE studies, as well.

[Richard W]

The term 'voiced aspirate' is incredibly misleading and I don't know why people still use it. Even if we assume that Indic, (dialectical) Armenian, etc. are conservative wrt the actual phonetic value of the series this would imply that the *dh series is breathy voiced rather than "voiced and aspirated" since breathy voice is what is actually attested in these branches (Greek has voiceless aspirates but this is a possible and common outcome of breathy voiced stops).

Doesn't Grassman's Law for Indo-Iranian fit better if aspiration is one of the features? Additionally, breathy voiced / murmured consonants are widely written as voiced consonant (stop or resonant) plus the letter representing /ɦ/. Thus, I don't think the Indian analysis as aspirated is so unnatural.

There's also the weak evidence for IIr-style Grassman's Law operating in Tocharian.

[WeepingElf]

AFAIK, the term "voiced aspirate" is a translation of what the old Indian grammarians called those stops. They used a combination of adjectives for them which for themselves meant "voiced" and "aspirated". And as Kath says, it is just a convention for naming a class of PIE phonemes whose true phonetic nature is unknown.

[Frislander]

The thing that annoys my most about the term "voiced aspirate" is that it leads to infelicitous comparisons being drawn with languages like Bario Kelabit, where the realisation is completely different from what we find in Indo-European. In these cases, the voiced-aspirate refers to a stop which transitions from being voiced to a voiceless aspirate during its release, which is quite different from a breathy voiced stop. Furthermore these voiced aspirates are highly marked segments with very limited distribution, exactly unlike the IE series.

[Zju]

IIRC Temematic and Thracian, like Germanic and Armenian, had consonant MoA shifts - if the former are taken into consideration, does this weaken or strengthen the case for some of the hypotheses of what the phonations of PIE stop series were?

[WeepingElf]

@Frislander: We don't know what the PIE "voiced aspirates" were like, so we cannot say whether they were like those of Hindi, or like those of Bario Kelabit, or whatever. All we know is that the ancient Indian grammarians ordered the four classes of Sanskrit stops in a 2x2 matrix with the features [voice] and [breath], but this may be (a) artificial and (b) not indicative of what those stops were like in PIE as they may have shifted on the way from PIE to Sanskrit (what we already know is that the Sanskrit voiceless aspirated stops are an innovation, emerging mostly from voiceless stop + laryngeal). All we can do is to estimate which kinds of sounds are the most likely to evolve in those kinds of stops we find in the known IE languages.

@Zju: "Temematic" is a hypothetical entity whose existence is somewhere between highly doubtful and imaginary. Thracian is a language so poorly known that we cannot say anything about its sound developments. This means that nothing can be deduced from these shadowy entities.

[Moose-tache]

All we can say for certain about Thracian stops is that they only used two rows of Greek letters, ignoring the aspirated series. The voiced consonants seem to be very common, with b appearing several times. This suggests that the voiced series in Thracian is the result of a merger between the second and third rows of stops in PIE. Or the voiced and voiced aspirated distinction in Thracian became something that the Greek alphabet could not logically represent. Either way it tells us nothing about the exact phonetic quality of the third row of stops.

As far as I'm aware, the only IE languages that completely merged two and only two rows of PIE stops merged the second and third rows. Other examples include Celtic and Balto-Slavic, with the latter preserving evidence of the distinction in its phonation and length system. The other branches either kept the three rows separate to some extent (albeit sometimes with major alterations), or merged them all into one. So to my intuition, it makes sense that the third row would be something only slightly distinguishable from row number two. If you had something like p, p', b instead of p, b, bh, I would expect more languages to merge rows one and three, or one and two. Rows two and three would be the most phonetically distinguishable.

Just for fun, there are two other suspected sound changes for Thracian. The ratio of sibilants to velars is high compared to a random sampling of PIE roots, suggesting that it may be a satem language. Also, o is less common than e or a, but not completely absent, suggesting it may have partially merged with another vowel. All of these are sound changes shared by Baltic, which has led some people to go full-Nyland and insist that Thracian is a decipherable Baltic language. In reality, there is little chance of deciphering the existing texts without more data or a bilingual.

[mae]

-

[WeepingElf]

If Thracian merged the *D and *Dh sets, this is a BIG nail in the coffin of Georg Holzer's idea that the Cimmerians spoke Temematic. If the Cimmerians spoke anything else than Iranian at all, they would probably have spoken something intermediate between Iranian, Slavic and Thracian - and all three have the *D/*Dh merger. IE languages that have a different development of the stops than T D Dh > T D D are found in peripheral positions in the IE world, not in the centre like Cimmerian. That said, I think that Holzer's etymologies are bogus and "Temematic" never existed.

[Frislander]

Also Tocharian keeps the 2nd and 3rd series separate in the dentals only, though of course that's somewhat complicated by the fact that all three series collapsed.

Also it looks like there were no POA gaps when Iranian merged the two series.

[Moose-tache]

Celtic kept the two voiced series separate in the labialized velars. This is really strange to me and I can't come up with a phonetic explanation for it. Albanian also does this but with the palatals instead. According to Matasović when you exclude secondary lenition and analogical developments the regular reflex of *ǵh is d in Albanian and the regular reflex of *ǵ is ð, but the two series aren't distinguishable for any other POA. My best guess for Celtic is that maybe something similar happened, where either *gw or *gwh was actually a fricative in 'pre-Celtic' and then the lenition got reversed later, but that's pretty speculative.

It doesn't seem like much of a mystery to me. Late PIE had almost no instances of *b, so there was a hole in the phonology, and the only labio-velar that did not have a corresponding labial to be distinguish against, became labial. There would have been minimal pairs distinguishing kw/p and gwh/bh, but none or almost none to distinguish b/gw, so they could have been in free variation without consequence. A pre-Celtic speaker may even have wondered if b was an allophone of gw or gw was merely an allophone of b. This is much like how kw later became p in Gallic/Welsh, but gw did not merge with the already extant b. Then later rows two and three collapsed as normal, probably for the same reasons they merged in other branches. So gw becomes b through normal labialization, then bh/dh/gh/gwh become b/d/g/gw through normal deaspiration or something.