The Great Proto-Indo-European Thread's Sequel

[Zaarin]

I don't insist on *h3 being a labialized *h2. It hasn't escaped me that few scholars reconstruct it that way. My main reason to assume that *h3 was a labialized *h2 was the Greek triple reflex, which clearly shows that in that language, the o-colouring effect of *h3 was still productive at a time when PIE *o almost certainly already was rounded, but it may just be that whatever *h3 was, it was associated with rounding, or the process that added rounding to *o in Late PIE also added rounding to *h3. One possibility was that *h3 was pharyngeal (while *h2 was uvular), and the feature [+pharyngeal] changed into [+rounded], moving *h3 from [ħ] to [χʷ] and *o from [ɑ] to [ɒ] in one sweep. But how much phonetic sense does that make?

I know of several languages where the allophone of /a/ next to a pharyngeal is [ɒ] rather than the expected [ɑ], for what it's worth.

[mèþru]

Maybe *h₃ is actually labialised *h₁?

[WeepingElf]

What concerns the Greek triple reflex, analogy may have been involved, or a perceived connection between *h3 and *o; after all, Greek is the language in which the workings of the PIE ablaut system are preserved more faithfully than in any other IE language, so the connection between *h3 and *o was pretty transparent as long as the laryngeals had not been lost. Consider how just about every handbook on PIE uses the Greek word for 'father' and its derivatives to demonstrate the way ablaut worked in PIE - because it is all faithfully preserved in that language.

Also, while there are IE languages that segment out labiovelars as /kw/ etc., and ones that shift them to labials, there is none that does that sort of thing to *h3, which speaks against *h3 being labialized *h2.

@meþru: I don't think *h3 was labialized *h1, because these two do not seem to form a natural class anywhere in IE. For instance, *h1 is always lost in Hittite, while *h3 is preserved in more or less the same contexts as *h2.

[KathTheDragon]

Slight quibble - *h₃ is only retained word-initially before a vowel. But this does still speak for a fundamental difference to *h₁.

[WeepingElf]

Fair. If *h3 survives in a more limited set of contexts than *h2 in Hittite, that speaks for *h3 being "weaker" than *h2. Labialization does not do the trick, rather the assumption that *h2 was uvular and *h3 was pharyngeal (and *h1 glottal),

[KathTheDragon]

I'd prefer *h₂ being velar and *h₃ being uvular - the fact that *h₃ does show up as a velar/uvular fricative in Anatolian means it was more likely to be one of those two in PIE than pharyngeal, since it's more likely that an original uvular shifted to pharyngeal than the reverse (see this paper yet again). In terms of the pan-IE colouring, we don't strictly speaking need *h₂ to be a-colouring - if we assume that the pre-colouring quality of *e was a central *ɐ, then *h₂ need only inhibit fronting, which a velar *x probably can do.

[WeepingElf]

That's fine, but I feel that the facts that *h2 did not split into three and that velar stops do not (consistently) colour vowels (words with *a next to a plain velar stops, with no *h2 anywhere near are IMHO late loanwords) speak against *h2 being velar. So *h2 = uvular, *h3 = pharyngeal seems to make sense a bit better, but that are minutiae.

[Tropylium]

I feel that the fact (…) that velar stops do not (consistently) colour vowels (words with *a next to a plain velar stops, with no *h2 anywhere near are IMHO late loanwords) speak against *h2 being velar.

Velar fricatives, maybe particularly voiceless ones, can have stronger vowel coloring effects than the stops. In continental West Germanic we can see *ai *au monophthongizing to *ō *ē before *x but not *k *g; and in Old English, *i *e *æ breaking to /iu eo æa/ before /x/ but again not before velar stops.

[Salmoneus]

That's fine, but I feel that the facts that *h2 did not split into three

There's also no evidence that the velar stops split into three, other than that you'd like them to have done so in order to give you a route to your preferred, Uralic-friendly vowel inventory.

and that velar stops do not (consistently) colour vowels

As Tropylium says, it's commonplace for fricatives to have more effect on adjacent vowels than stops.

[does *ai only monopthongise before *x in 'continental west germanic'? C.f. Dutch 'steen'. I suppose you could say that 'continental west germanic' has a common monophthongisation change, and then Dutch has an extra one. But that requires us to believe in 'continental west germanic' as a concept. Given that *ai monopthongised in most contexts in Old English, in Old Frisian, in Old Dutch and in Old Saxon - in fact, everywhere other than in Old High German - I'd think it would make more sense to pin that conditioned change on Old High German specifically. But I suppose it's a matter of taste...]

(words with *a next to a plain velar stops, with no *h2 anywhere near are IMHO late loanwords) speak against *h2 being velar. So *h2 = uvular, *h3 = pharyngeal seems to make sense a bit better, but that are minutiae.

It's also worth pointing out that it seems a bit quixotic to try to pin down the POA of the laryngeals by comparison with the POA of the stops, when there's also debate about the POA of the stops...

[Frislander]

I would still agree with Weeping Elf that a uvular realisation of h2 is more likely than velar, and would actually be similar to the situation in modern IE, where having "velar fricatives" actually have a uvular realisation is commonplace. That's not to say that it could have patterned like a velar in the stop system (as in Modern IE), just that phonetically I doubt they could have been at the same POA.

Also I struggle to see the merit in theories of the PIE stop system that do a retraction of the dorsals back a place, simply because none of the daughter languages support uvulars in any way, whereas I think there is much more support for saying that palatovelars, if there even was a point in IE history where all three dorsal series were present, represent a relatively recent development from pre-PIE.

[WeepingElf]

That's fine, but I feel that the facts that *h2 did not split into three

There's also no evidence that the velar stops split into three, other than that you'd like them to have done so in order to give you a route to your preferred, Uralic-friendly vowel inventory.

Oops, you caught me. We indeed have no real evidence for such a split; I should have said, "if *h2 was velar, I'd expect a triad of velar fricatives matching the velar stops, which we do not observe". The "split into three" of the velar stops is just a hypothesis of mine which may be utterly wrong. Indeed, the whole "Great Vowel Collaps"/"Velar Split" business is merely an attempt to nudge "PIE0" (a hypothetical earlier stage of PIE, even before the already speculative "PIE1", the stage before ablaut) typologically closer to Uralic, but there are different solutions (such as a merger of three velar series into one in Proto-Uralic), and as we don't know whether IE and Uralic are related at all (though this is IMHO the best hypothesis to account for their morphological resemblances, but still just a hypothesis!), such an argumentation is actually invalid.

and that velar stops do not (consistently) colour vowels

As Tropylium says, it's commonplace for fricatives to have more effect on adjacent vowels than stops.

Point taken.

(words with *a next to a plain velar stops, with no *h2 anywhere near are IMHO late loanwords) speak against *h2 being velar. So *h2 = uvular, *h3 = pharyngeal seems to make sense a bit better, but that are minutiae.

It's also worth pointing out that it seems a bit quixotic to try to pin down the POA of the laryngeals by comparison with the POA of the stops, when there's also debate about the POA of the stops...

Again a valid objection. We don't really know the POAs of the three "velar" stop series. Perhaps all this musing about places of articulation of back stops and fricatives we are doing here is utterly mistaken, and the reality may have been utterly different. For instance, the idea once flashed up in my head that *h3 may have been */f/! But there is no good reason assuming that, so I abandoned it immediately. If academic linguists haven't found the solution after decades of study and discussion, we can't really expect to find it by ourselves. (See the Middletown Astronomical Society simile I posted in the Great Macrofamilies Thread a few days ago. If you don't follow that thread: I compared our discussion to a group of amateur astronomers discussing theories of dark matter - who just can't expect finding out about its nature because they simply do not have the means to find out.)

[KathTheDragon]

This is of course why I don't make any reference to the stops in my reasoning for positing velar and uvular fricatives for *h₂ and *h₃, since reasoning based on actual facts is vastly stronger.

[Elizabeth K.]

Hi, I’m a newbie. Two questions:

1. Why do the different verbal aspects have different sets of person/number endings when they usually already have differentiation in the stems, reduplication, or augment - and where did these different sets of endings come from (particularly the 1sg)?

2. Where did the old Slavic aorist and imperfect come from? Because they sure as fuck aren’t holdovers from PIE.

[Zju]

As for two - they come from PIE sigmatic perfective and a blending of -ē deverbal nouns with the non-suppletive aorist of the copula respectively.

[Frislander]

When it comes to 1 there's several questions. Firstly why do the aspects have different sets of endings - why not? It's just redundancy and as we all know languages like redundancy, particularly in inflection. As for where they came from... that's basically the same as "where did PIE come from?" caus that's the system as far back as we can reconstruct it in PIE.

[Elizabeth K.]

As for two - they come from PIE sigmatic perfective and a blending of -ē deverbal nouns with the non-suppletive aorist of the copula respectively.

Okay I obviously wasn’t thinking straight when I wrote that last night, as I’d neglected to think about the sigmatic aorist. Are there any papers or other material that explain the changes more in-depth (or more explicitly)? The phonological shifts seem complex at least in the case of the imperfect.

[WeepingElf]

One hypothesis to explain the two different sets of personal endings in PIE is that the *-m, *-s, *-t endings were used when the subject was an agent (as in "The boy runs") and the *-h2e, *-th2e, *-e endings were used when it was not (as in "The stone fell to the ground"). There are languages that do such a thing, though not within the IE family, and this hypothesis is controversial.

[Pabappa]

Wikipedia has h2e, th2e, o (sic) as mediopassive, so that's make sense. Yet the one difference in the 3rd person suggests it's not that simple.

https://en.m.wikipedia.org/wiki/H₂e-conjugation_theory

Mediopassive is almost but not quite identical with preterite .... presumably context helped in 1st and 2nd person, and there may have been stem changes too.

[WeepingElf]

And old idea of mine is that the thematic conjugation originally was a transitive conjugation, with the thematic vowel being a fossilized 3rd person patient marker identical to the 3rd person stative *-e; before it was fossilized, it would have been a bipersonal conjugation, as in *leubh-th₂e-m-i 'I love you'. (I use this hypothesis in my conlang Old Albic.) Alas, this is a bold speculation unsupported by evidence.

[WeepingElf]

Also, while there are IE languages that segment out labiovelars as /kw/ etc., and ones that shift them to labials, there is none that does that sort of thing to *h3, which speaks against *h3 being labialized *h2.

This can easily be countered: those changes to labiovelars happened at times when the laryngeals already were lost. They happened after about 500 BC in Gaulish and Brittonic, probably not before 1000 BC in Sabellic, and after 1 AD in Germanic. Mycenean shows labiovelars still intact but laryngeals already gone by 1200 BC. So this is not really a reason to doubt that *h₃ was labialized. There may be other reasons, though.