The Great Proto-Indo-European Thread's Sequel

[Travis B.]

I wonder why Kümmel doesn't reconstruct *h₃ as labialized in his vision of PIE/PIH...

I don't know. In fact, I am not sure at all that *h₃ was a labialized *h₂. The only thing that suggests that is that it adds the feature [+round] (in addition to [+back]) to a preceding or following *e - but the feature may originally have been something different, maybe pharyngealization or whatever (a shift [ɑˤ] > [ɒ] is IMHO not out of the question). There are apparently good reasons to assume that *h₃ was voiced while *h₂ wasn't, so that may have been the original difference between the two; the developments in Hittite seem to suggest that *h₃ was a "weaker" sound than *h₂.

Kümmel reconstructs *h₂ as [χ] and *h₃ as [ʁ], and I don't get how [ʁ] would end up conditioning rounding but [χ] wouldn't, considering that they differ only in voicing.

[WeepingElf]

I wonder why Kümmel doesn't reconstruct *h₃ as labialized in his vision of PIE/PIH...

I don't know. In fact, I am not sure at all that *h₃ was a labialized *h₂. The only thing that suggests that is that it adds the feature [+round] (in addition to [+back]) to a preceding or following *e - but the feature may originally have been something different, maybe pharyngealization or whatever (a shift [ɑˤ] > [ɒ] is IMHO not out of the question). There are apparently good reasons to assume that *h₃ was voiced while *h₂ wasn't, so that may have been the original difference between the two; the developments in Hittite seem to suggest that *h₃ was a "weaker" sound than *h₂.

Kümmel reconstructs *h₂ as [χ] and *h₃ as [ʁ], and I don't get how [ʁ] would end up conditioning rounding but [χ] wouldn't, considering that they differ only in voicing.

Yes. As I said, *h₃ may not have rounded the vowels, but done something else such as pharyngealization to them that later developed into rounding. Do you remember KathTheDragon (whom I haven't seen here, or elsewhere in the ZBB, for a long time)? She maintained a similar notion, namely that PIE *o originally hadn't been rounded. She may be right about that after all. In fact, in a system where *e was [æ] and *a was [ɑ], an [ɒ] contrasting with the latter would rather be a bag on the side. Of course, an [ɑˤ] in addition to them would also be a bit odd, but it seems just a bit less bizarre to me, and when [æ] crept up the front edge of the vowel space, this could be easily fixed by rounding and raising (and depharyngealizing) [ɑˤ], giving the nice /i e a o u/ system we see in Late PIE.

So far, I have *h₃ develop into "*3w" ("3" being a symbol for what may have been [ʕ]) in Proto-Hesperic, but I think I will change that and have it become just *3, just like *h₂.

[Travis B.]

I don't know. In fact, I am not sure at all that *h₃ was a labialized *h₂. The only thing that suggests that is that it adds the feature [+round] (in addition to [+back]) to a preceding or following *e - but the feature may originally have been something different, maybe pharyngealization or whatever (a shift [ɑˤ] > [ɒ] is IMHO not out of the question). There are apparently good reasons to assume that *h₃ was voiced while *h₂ wasn't, so that may have been the original difference between the two; the developments in Hittite seem to suggest that *h₃ was a "weaker" sound than *h₂.

Kümmel reconstructs *h₂ as [χ] and *h₃ as [ʁ], and I don't get how [ʁ] would end up conditioning rounding but [χ] wouldn't, considering that they differ only in voicing.

Yes. As I said, *h₃ may not have rounded the vowels, but done something else such as pharyngealization to them that later developed into rounding. Do you remember KathTheDragon (whom I haven't seen here, or elsewhere in the ZBB, for a long time)? She maintained a similar notion, namely that PIE *o originally hadn't been rounded. She may be right about that after all. In fact, in a system where *e was [æ] and *a was [ɑ], an [ɒ] contrasting with the latter would rather be a bag on the side. Of course, an [ɑˤ] in addition to them would also be a bit odd, but it seems just a bit less bizarre to me, and when [æ] crept up the front edge of the vowel space, this could be easily fixed by rounding and raising (and depharyngealizing) [ɑˤ], giving the nice /i e a o u/ system we see in Late PIE.

So far, I have *h₃ develop into "*3w" ("3" being a symbol for what may have been [ʕ]) in Proto-Hesperic, but I think I will change that and have it become just *3, just like *h₂.

The problem here is why do *h₂ and *h₃ behave differently in their effect on early PIE *e then unless *h₃ was also pharyngealized or something (which is not how Kümmel reconstructed it)? To me, unless we have a very good reason to believe otherwise, the most parsimonious hypothesis is that PIE *h₃ was labialized. It could have been something different in Pre-PIE, of course.

[Zju]

Really, my big problem with Indo-Uralic is that too many of its posited forms just really feel like loans -- they're too close, and when they're not very close to the PIE forms themselves they feel like they either went through a loaning filter (e.g. systematically stripping out laryngeals while keeping the other sounds as close as possible) or were borrowed from a PIE daughter, and furthermore they're often best linked not to PIE but to Pre-PII.

Sure. Most of the posited cognates are probably just loanwords. The PIE phonemes are usually represented by the closest PU phonemes, even such things as ablaut grades or the (still younger) vowel-colouring effects of laryngeals, which are hardly of Indo-Uralic age. Such trivial sound correspondences between two languages that can't be as close to each other as, say, English and German, are indicative of borrowing. Yet, the words Kümmel lists in the presentation I cited yesterday, show less trivial sound correspondences such as PIE *d: PU *n, which are more likely due to Urverwandtschaft than borrowing. And how does one language borrow an entire pronoun paradigm from another? Not impossible, but not particularly likely, either. (EDIT: This is more a matter of definition than anything else: What makes the "core" of a language it is classified by? Morphology, or basic vocabulary? If we classify languages by morphology, which seems reasonable to me, IE and Uralic are indeed quite close together, and probably have a common source for their morphology and thus form a family. If you insist on a substantial shared vocabulary, there is not enough reason to classify them together. As I wrote earlier, I am into the theory that PIE was a kind of mixed language of a source related to Uralic, which provided most of the morphology, and one related to NW Caucasian, which strongly influenced the phonology, such as the ablaut system and the palatalized and the labialized velars, reflecting the mixed genetic composition of the Yamnaya people who probably spoke it.)

At first glance the pronoun paradigms seem convincing, but apparently there are issues with regard to things like number (e.g. /m/ in PIE is singular while /m/ in PU is plural).

PIE has *m in some 1PL verbal suffixes

[abahot]

PIE has *m in some 1PL verbal suffixes

While we're speculating on proto-languages I have seen it suggested that 1PL suffixes like *-mos are from earlier -m (1st person) + -s (plural).

[WeepingElf]

The problem here is why do *h₂ and *h₃ behave differently in their effect on early PIE *e then unless *h₃ was also pharyngealized or something (which is not how Kümmel reconstructed it)? To me, unless we have a very good reason to believe otherwise, the most parsimonious hypothesis is that PIE *h₃ was labialized. It could have been something different in Pre-PIE, of course.

Perhaps they differed also in place of articulation: *h₂ was uvular [χ], and *h₃ was pharyngeal [ʕ]. But [ʁʷ] or even [ʕʷ] for the latter is not out of the question, either. (What points at *h₃ actually being labialized in Late PIE is the Greek triple reflex where a syllabic *h₃ manifests as /o/.) So I'd now guess that *h₂ was [χ] and *h₃ was [ʕʷ]. The latter was different from *h₂ʷ=[χʷ] for which there apparently is some (though controversial) evidence from Hittite. Perhaps Mallory & Adams's *h₄ is a thing after all then, being unlabialized [ʕ], giving a nice 2x2 system [χ χʷ ʕ ʕʷ]. (This means that I can keep *h₃ labialized in Proto-Hesperic.) *h₁ would still be just [h].

PIE has *m in some 1PL verbal suffixes

While we're speculating on proto-languages I have seen it suggested that 1PL suffixes like *-mos are from earlier -m (1st person) + -s (plural).

Yes - this is indeed quite obvious.

[abahot]

Yes - this is indeed quite obvious.

What I'm trying to get at is that, if 1PL is derived from 1SG, then we cannot argue that Indo-European and Uralic share a 1PL marker -- at best, that Uralic 1PL corresponds to IE 1st person, which is slightly weaker.

[WeepingElf]

Perhaps they differed also in place of articulation: *h₂ was uvular [χ], and *h₃ was pharyngeal [ʕ]. But [ʁʷ] or even [ʕʷ] for the latter is not out of the question, either. (What points at *h₃ actually being labialized in Late PIE is the Greek triple reflex where a syllabic *h₃ manifests as /o/.) So I'd now guess that *h₂ was [χ] and *h₃ was [ʕʷ]. The latter was different from *h₂ʷ=[χʷ] for which there apparently is some (though controversial) evidence from Hittite. Perhaps Mallory & Adams's *h₄ is a thing after all then, being unlabialized [ʕ], giving a nice 2x2 system [χ χʷ ʕ ʕʷ]. (This means that I can keep *h₃ labialized in Proto-Hesperic.) *h₁ would still be just [h].

I am not sold to this theory, though! So I'd say that *h₂ = [χ] (or perhaps [ħ]) and *h₃ = [ʕʷ], whether the other two are a thing is questionable, the evidence is too tenuous (also, uvulars and pharyngeals tend to change into each other). Phoneme inventories often have gaps like that.

[WeepingElf]

As for the idea that PIE was related to Afroasiatic, I did indeed entertain for a while the notion that the "Caucasian" substratum in PIE could be related to Semitic, with the hi-conjugation descending from the latter's suffixal conjugation. But that is all that at first glance seems to match here, and it does not really do so all too well, and I now think the substratum contributed by the Caucasian ancestors of the Yamnaya people was probably related to NW Caucasian. It is of course not out of the question that IE and Semitic share Neolithic Wanderwörter, such as the much-discussed 'bull'-word.

Yet earlier, I had the idea that Proto-Mitian had a bipersonal conjugation, with the agent suffixes reflected in the PIE "normal" active endings, and the patient suffixes in the PIE hi-conjugation, perfect and middle endings, and also in the Hungarian and Selkup indefinite conjugation. But the Hungarian and Selkup indefinite conjugations do not match each other well, let alone the PIE endings in question (see also this for some of the thoughts I had). Also, I thought that the PIE thematic conjugation contained a fossilized 3rd person singular patient suffix in the thematic vowel, but that no longer makes sense to me. I now think both sets of endings are related to the personal pronouns, whether Mitian is a thing or not, the "normal" active endings quite obviously, and in the hi-conjugation, perfect and middle endings, a /m/ may have been lost before the *h₂ in the 1st person singular. Perhaps the laryngeal is the reflex of an old pronominal case marker for patients here?

[WeepingElf]

Some thoughts about the history of the Celtic languages

In this post, I shall share some thoughts about the history of the Celtic languages, not all new or original.

1. The shibboleth of La Tène

The first idea is that the shift *kw > p that characterizes "P-Celtic" (i. e., Gaulish and Brythonic) spread as a shibboleth of the more sophisticated La Tène culture that arose around the year 450 BC, that failed to reach the outliers in Ireland and the Iberian Peninsula, which, as is well known, remained "Q-Celtic", keeping the old labiovelar intact (in Irish, the *w was later lost). But there are some Gaulish names, such as the river name Sequana (> Seine), which show retained labiovelars. These names can be explained either by remaining pockets in which the labiovelar was retained, or, IMHO more likely, by assuming by a diglossia between an archaic "druidic" language (see below) and the more innovative vernacular dialects. It is also not certain that this change occurred only once and spread across Gaul and Britain from a single centre; it is a rather natural gap-filling change in a language that lacked /p/, and may have happened more than once, perhaps independently in Gaul and Britain.

2. A shibboleth of Hallstatt?

Maybe even the loss of PIE *p was a shibboleth of this kind, spreading through an earlier (Pre-)Celtic dialect continuum which still had *p - the shibboleth of the Hallstatt culture, that emerged about 800 BC from the Late Bronze Age Urnfield culture. Under this hypothesis, the Celtic (and Italic) languages were first spread by the "Urnfield migration" around 1300 BC, but the loss (or rather weakening; the full loss came later) of *p happened later, and was associated with the more modern, iron-introducing Hallstatt culture. Like the "shibboleth of La Tène", it may have missed some outliers, namely Lusitanian in the western Iberian Peninsula, Ligurian in northwestern Italy and southeastern France, and perhaps also the hypothetical "Belgian" or "Nordwestblock" language around the Rhine delta, postulated by some scholars on the ground of some geographical names that appear to have preserved PIE *p.

3. A Druidic Vedic?

Some ancient historians speak of a "secret language" of the druids. It is known that the druids tradited their sacred knowledge in oral form. They never wrote it down, and thus the corpus is lost. Perhaps the druidic knowledge was tradited in verses in an archaic Celtic language, similar to the Vedas in ancient India. This druidic language may have still been "Q-Celtic" even in Gaul and Britain, which may explain occasional preserved labiovelars in Gaulish names such as Sequana.

[þeprussianfrog]

Some thoughts about the history of the Celtic languages

In this post, I shall share some thoughts about the history of the Celtic languages, not all new or original.

1. The shibboleth of La Tène

The first idea is that the shift *kw > p that characterizes "P-Celtic" (i. e., Gaulish and Brythonic) spread as a shibboleth of the more sophisticated La Tène culture that arose around the year 450 BC, that failed to reach the outliers in Ireland and the Iberian Peninsula, which, as is well known, remained "Q-Celtic", keeping the old labiovelar intact (in Irish, the *w was later lost). But there are some Gaulish names, such as the river name Sequana (> Seine), which show retained labiovelars. These names can be explained either by remaining pockets in which the labiovelar was retained, or, IMHO more likely, by assuming by a diglossia between an archaic "druidic" language (see below) and the more innovative vernacular dialects. It is also not certain that this change occurred only once and spread across Gaul and Britain from a single centre; it is a rather natural gap-filling change in a language that lacked /p/, and may have happened more than once, perhaps independently in Gaul and Britain.

2. A shibboleth of Hallstatt?

Maybe even the loss of PIE *p was a shibboleth of this kind, spreading through an earlier (Pre-)Celtic dialect continuum which still had *p - the shibboleth of the Hallstatt culture, that emerged about 800 BC from the Late Bronze Age Urnfield culture. Under this hypothesis, the Celtic (and Italic) languages were first spread by the "Urnfield migration" around 1300 BC, but the loss (or rather weakening; the full loss came later) of *p happened later, and was associated with the more modern, iron-introducing Hallstatt culture. Like the "shibboleth of La Tène", it may have missed some outliers, namely Lusitanian in the western Iberian Peninsula, Ligurian in northwestern Italy and southeastern France, and perhaps also the hypothetical "Belgian" or "Nordwestblock" language around the Rhine delta, postulated by some scholars on the ground of some geographical names that appear to have preserved PIE *p.

3. A Druidic Vedic?

Some ancient historians speak of a "secret language" of the druids. It is known that the druids tradited their sacred knowledge in oral form. They never wrote it down, and thus the corpus is lost. Perhaps the druidic knowledge was tradited in verses in an archaic Celtic language, similar to the Vedas in ancient India. This druidic language may have still been "Q-Celtic" even in Gaul and Britain, which may explain occasional preserved labiovelars in Gaulish names such as Sequana.

As someone who currently goes onto a deep dive into the Italic and Celtic rabbithole, these are some interesting thoughts.

The kʷ/p branch split notably also did occur in Italic.
Latino-Faliscan retained the labiovelar and Sabellic shifted it into p. Neither did lose the original p, though.
Lusitanian is the same story, having both the original p and the kʷ -> p.
Venetic retained kʷ as well as the original p.

[WeepingElf]

As someone who currently goes onto a deep dive into the Italic and Celtic rabbithole, these are some interesting thoughts.

The kʷ/p branch split notably also did occur in Italic.
Latino-Faliscan retained the labiovelar and Sabellic shifted it into p. Neither did lose the original p, though.
Lusitanian is the same story, having both the original p and the kʷ -> p.
Venetic retained kʷ as well as the original p.

Yep.