The Great Proto-Indo-European Thread

[WeepingElf]

I used to think of the laryngeals as voiceless fricatives whose PoAs match the three velar series, but recently, doubts against that have been growing in me. The comparative data do not really point at something like /x χ χʷ/, rather at something like /h ħ ʕ/. But maybe the former changed into the latter at some point. If *h3 wasn't labialized, this may mean that *o perhaps originally wasn't rounded.

[KathTheDragon]

If *h3 wasn't labialized, this may mean that *o perhaps originally wasn't rounded.

I've been running under the assumption that h₃ coloured *e (originally only [ə], but later colouring [e] analogically) into [ʌ], and that by the time Szemerényi's Law occured, *o had gone from [ə:] to [ʌ(:)], with length retained only in IIr. open syllables and in Anatolian stressed syllables. Thus, prior to Szemerényi's Law, there was no *ō.

[2+3 clusivity]

Anyone know of a grammatical sketch of Mycenaean? I am particularly interested in the case structure and pronouns and demonstratives.

[hwhatting]

In my experience, one normally finds only general introductions, summary discussion of Mycenean features in the context of Greek historical grammars, or discussions of individual features.
But maybe you find something here:
Google search results
Websites with lots of links
Grammar in German that does not have the bits you want, except for short sketches, because those chapters have not yet been written.
But it has a literature list.

Among printed books, there is A Tentative Grammar of Mycenaean Greek by Ebbe Vilborg, but that's from 1960, shortly after the decipherment.


On the whole, I assume that you know all this and know how to Google. I just got interested myself and decided to share the results.

[KathTheDragon]

How do old IE langs (Latin, Greek, Sanskrit, Hittite, and others I've probably not thought of) handle verb serialisation?

[Zju]

Do you know online lexicons of PIE (full or partial) other than that Nostratic reconstruction thing by Dostoevsky? More precisely I'm interested in the roots begining in ks-, k̀s-, kʷs-, sk-, sk̀-, skʷ-. Until know I've been using wiktionary, but it is light years away from being exhaustive.

[CatDoom]

Do you know online lexicons of PIE (full or partial) other than that Nostratic reconstruction thing by Dostoevsky? More precisely I'm interested in the roots begining in ks-, k̀s-, kʷs-, sk-, sk̀-, skʷ-. Until know I've been using wiktionary, but it is light years away from being exhaustive.

Somebody posted a good lexicon of roots in German earlier in the thread. Here's a list of roots I made using that document and running the glosses through Google translate to get them in English.

Largely because of laziness, I didn't bother with the roots that the author put a question mark after (which I'm assuming are less solidly reconstructed), and I think I may have also excluded roots for which he gave reflexes in only one language. I also put the vowel *a in parentheses, just as a reminder to myself that it's not generally reconstructed as a phoneme in PIE, and may indicate that the root is a borrowing or has some other weird thing going on. I'm sure there are a number of errors (since I know essentially no German and was relying entirely on Google's software), but you may find it interesting if you don't read German. I added my own question marks where I found it particularly difficult to decide between various translations Google offered; in retrospect I probably should have just listed what Google gave me, but again I'm a bit lazy.

[Salmoneus]

Do you know online lexicons of PIE (full or partial) other than that Nostratic reconstruction thing by Dostoevsky? More precisely I'm interested in the roots begining in ks-, k̀s-, kʷs-, sk-, sk̀-, skʷ-. Until know I've been using wiktionary, but it is light years away from being exhaustive.

A user who used to be called Goatface had a big spreadsheet. I think the core of it was from Mallory and Adams, but I think there were other sources too. Actually there must be, because it clearly has sources that use different formatting. Anyway, it's not 100% accurate (I try to double-check it with etymological dictionaries) but it's the best thing out there in terms of breadth.

Iirc the user is now Morrigan - she's not here anymore but you may find her on other forums, or someone may have a link to her stuff online?

Anyway, looking at the sheet now, it only gives one for kws: kwsep (which I think the exact form of is debated, iirc). It also gives kswek's-os (sixty), ksuróm (razor), ksun (with), k'súlom (shaped wooden post), ksiHróm (whey), kseubh (shake), kseu (whet, rub), k'se:ros/k'seros/kseros (dry), ksneu (sharpen), and ksen (card wool). It has about 40 sk- (etc) roots.

[KathTheDragon]

Morrígan's quite active on Anthologica. It's the best place I know of to find her. She's also working on a brand new version of her lexicon, which is eagerly anticipated.

[R.Rusanov]

If *h3 wasn't labialized, this may mean that *o perhaps originally wasn't rounded.

I've been running under the assumption that h₃ coloured *e (originally only [ə], but later colouring [e] analogically) into [ʌ], and that by the time Szemerényi's Law occured, *o had gone from [ə:] to [ʌ(:)], with length retained only in IIr. open syllables and in Anatolian stressed syllables. Thus, prior to Szemerényi's Law, there was no *ō.

What about the abl. plural ending -ōms ?

[Shm Jay]

Dostoevsky? I think his name is Dolgopolsky.

[Sleinad Flar]

What about the abl. plural ending -ōms ?

What abl. plural ending -ōms ?

[KathTheDragon]

If *h3 wasn't labialized, this may mean that *o perhaps originally wasn't rounded.

I've been running under the assumption that h₃ coloured *e (originally only [ə], but later colouring [e] analogically) into [ʌ], and that by the time Szemerényi's Law occured, *o had gone from [ə:] to [ʌ(:)], with length retained only in IIr. open syllables and in Anatolian stressed syllables. Thus, prior to Szemerényi's Law, there was no *ō.

What about the abl. plural ending -ōms ?

Never heard of one. Do you have a source? But I'd say the *ō would arise be Szemerényi's Law, then the *s would be restored analogically.

[R.Rusanov]

Ringe, A History of English, Volume I: From Proto-Indo-European to Proto-Germanic (Oxford University Press, 2006)

[Sleinad Flar]

...which doesn't mention abl. plural ending -ōms.

It does mention thematic ins. pl. ending *-ōys. Maybe that's what you meant?

That ending is indeed a hard to explain, but at least the element *-oy- appears in other plural case forms as well, especially in the pronominal declension (gen. *-oy-som, dat/abl. *-oy-bhos/mos). The second element could be *-is or something like that, but has no parallels elsewhere. Whatever its origin, it does look like there is some kind compensatory lengthening going on.

[Neek]

How do old IE langs (Latin, Greek, Sanskrit, Hittite, and others I've probably not thought of) handle verb serialisation?

If you're referring to verb serialization as in, sequencing roots or inflected stems of verbs in order to express a more complex answer, then I think I can offer a pretty good answer, but I can't speak for Sanskrit or Hittite. For the starters, verb serialisation may be used to explain away Latin future endings for 1st and 2nd conjugation and imperfect endings for all conjugations, which were ultimately from PIE root *bʰuH in some variation or another. In fact, the whole *bʰuH in Latin is a scholarship in itself: It gives the perfect forms of esse and the passive of facio. Allen and Greenough's New Latin Grammar state, "This was added to a complete word, originally a case of a verbal noun, as in I was a-seeing; hence vidē-bam[/i]. The form probably began in the Second or Third Conjugation and it was extended to others" (168b., note).

I don't agree with their analysis, because it supposes a verbal noun that doesn't exist, or at least I can't find any reason for it to exist outside of their documentation. Moreover, a verb like *bʰuH would take a predicate nominative. The simpler explanation is that they suffixed *-bam to the stem and called it a day. There's a reason Allen and Greenough took this approach, though: IE languages do not serialize verbs, rather they analytically form new tenses and constructions with infinitive derivations, such as participles, supines, and gerunds, plus another verb.

As a good digression: In Ancient Greek, perfects are frequently formed with perfect participles plus a conjugated form of "to be." This is for active, passive and middle, through the indicative, subjunctive, optative and imperative. So λελυκώς εἰμι is equivalent to λέλυκα (which is present active.) Hell, reading Smyth, there's a periphrastic perfect that is ἔχω + aorist participle <ekʰō>, which is used to create a transitive verb in the perfect from an intransitive one: ἔστηκα, I stood (intransitive); στήσας ἔχω, I have placed (it).

I'm guessing that Hittite and Sanskrit weren't that different in favoring periphrastic forms. A quick googling of Hittite grammar produces this shabby website. So here's my point: Verb serialization in the form of conjugated verb + conjugated verb, without subordination or coordination doesn't seem grammatical through ancient IE languages. Serialization relies on the flexibility of the verbal noun.

[Neek]

The trouble is that e̯ a̯ o̯ would imply that they are semivowels like i̯ u̯, standing in the same allophonic relationship to e a o as the latter to i u. This is most certainly not the case: they are seperate phonemes, and in the case of h₂ their consonantness is pretty certain (see above) and far from aimless guesswork.

Is it, though? We can have them coughing vague phlegm up through their epiglottis, but we can't imagine a vague consonant sound originating from a vowel? That system of representing them as semivowels at least shows a bit of honesty. They're vowel-coloring pencils that have a bunch of weird rules that do no interact well with other phonological processes in the language. They're fricatives, because fricatives can appear anywhere in a consonant cluster. But they also have resonants, and only sonorants get those. But Greek, Armenian and Hittite attest those sonorants, but they're not like the other sonorants at all! For instance, Latin has that sound change rule where m̥ n̥ > em en and l̥ r̥ > ol or, but H̥ > a, so it doesn't match up to that pattern, and Hittite R̥ > aR, but h̥₂, h̥₃ > ha, so that pattern's out as well.

[KathTheDragon]

I'm not convinced that the laryngeals could be vocalic in the same way that the sonorants could be. It's much easier IMO to have epenthetic vowels appear adjacent to laryngeals in certain envrionments (potentially varying between daughters), then have the laryngeals vanish as normal.

[CatDoom]

I tend to agree with Kath; since the laryngeals behave almost nothing at all like *w and *y (except in Greek), representing them as e̯ a̯ o̯ strikes me as a lot more dishonest than depicting them as h-sub-whatever. Using "h" at least makes sense based on the fact that stop+laryngeal clusters turned into aspirated stops in Indo-Aryan, implying that they were sounds at least somewhat similar to or prone to turn into [h]. Using diacritics might be more aesthetically pleasing and in keeping with PIE orthography in general, but they would have to be different from the "velar" diacritics, since it wouldn't be accurate to imply that the laryngeal places of articulation are necessarily the same as the velars.

[Neek]

I'm not convinced that the laryngeals could be vocalic in the same way that the sonorants could be. It's much easier IMO to have epenthetic vowels appear adjacent to laryngeals in certain envrionments (potentially varying between daughters), then have the laryngeals vanish as normal.

I was rather implying how strange they are, and how confirming their "consonantness" is a kind of a meaningless claim. Moreover, they do not behave like sonorants which generated their own epenthetic vowels.

I tend to agree with Kath; since the laryngeals behave almost nothing at all like *w and *y (except in Greek), representing them as e̯ a̯ o̯ strikes me as a lot more dishonest than depicting them as h-sub-whatever.

How is it less honest? In nearly all the IE languages, eh₁ > ē, eh₂ > ā, eh₃ > ō. Having them as e̯ a̯ o̯ is transparent about their function. Maybe I need to throw more diactrics in there, to really get the busienss going? Also, if I stop representing the "palatal velars" with ḱ ǵ ǵʰ, and instead use c j jʰ, I wonder how much harder I can present a trollworthy PIE.

[Sleinad Flar]

The "troll-worthy" PIE version already exist. At dnghu.org http://dnghu.org/Indo-European-Language-Europe/.

[Valdeut]

I tend to agree with Kath; since the laryngeals behave almost nothing at all like *w and *y (except in Greek), representing them as e̯ a̯ o̯ strikes me as a lot more dishonest than depicting them as h-sub-whatever.

How is it less honest? In nearly all the IE languages, eh₁ > ē, eh₂ > ā, eh₃ > ō. Having them as e̯ a̯ o̯ is transparent about their function. Maybe I need to throw more diactrics in there, to really get the busienss going? Also, if I stop representing the "palatal velars" with ḱ ǵ ǵʰ, and instead use c j jʰ, I wonder how much harder I can present a trollworthy PIE.

But it doesn’t transparently represent the function of the laryngeals. The inverted breve below is already used for something else in PIE linguistics, namely representing non-syllabic allophones of a vowel. Your notation suggests that the laryngeals relate to the full vowels the same way that *u̯ and *i̯ relate to *u and *i. That is misleading. The semivowels *u̯ and *i̯ don’t colour preceding vowels, they form diphthongs. The point of writing *u̯ and *i̯ rather that *w and *y is that the syllabic allophones are written *u and *i, thus making the relationship clear. Even if you think the laryngeals were some kind of semivowel, they are not non-syllabic allophones of *e, *a and *o (although they developed sort of that way in Greek).

If you want to represent the laryngeals with a diacritic over a vowel letter, just pick a diacritic mark that isn’t already used to mark something else. Write *ả *ẻ *ỏ or whatever (that’s a hook above that look a bit like the symbol for the glottal stop).

[WeepingElf]

I have seen the laryngeals represented as h with subscript e, a, o instead of 1, 2, 3. Quite neat, if you ask me.

[jal]

I have seen the laryngeals represented as h with subscript e, a, o instead of 1, 2, 3. Quite neat, if you ask me.

Given the discussion above, that indeed seems neat.


JAL

[KathTheDragon]

I have seen the laryngeals represented as h with subscript e, a, o instead of 1, 2, 3. Quite neat, if you ask me.

It's probably as transparent as you're going to get in regards to their function.