The Great Proto-Indo-European Thread

[WeepingElf]

But the entire reason for looking for IE-Uralic lexicon matches is the existence of suggestive morphological similarities between the two!

Exactly. The morphological resemblances between IE and Uralic are hard to explain otherwise, and are of a similar degree as those between, say, Semitic and Cushitic. Indeed, the Afrasian family is mainly based on morphology, and lexical comparison has not yet yielded many generally accepted results. The morphologies of IE and AA, in contrast, have very little in common and do not invite comparison.

Yet, the majority of lexical lookalikes between IE and U look more like loanwords as the sound correspondences are near-trivial and the Late PIE vowels are faithfully reflected in Uralic which is not what one would expect in actual cognates in two languages that had separated thousands of years ago, but there are other, less obvious ones which may be genuine cognates.

[Znex]

Yet, the majority of lexical lookalikes between IE and U look more like loanwords as the sound correspondences are near-trivial and the Late PIE vowels are faithfully reflected in Uralic which is not what one would expect in actual cognates in two languages that had separated thousands of years ago, but there are other, less obvious ones which may be genuine cognates.

Speaking of which, I have a fondness for a possible correspondence between Indo-Hittite *meh1w- {4} and Uralic *neljä {4}, though that might be a bit of a stretch. It just strikes me as interesting since this older(?) PIE word for {4} has a nasal onset, and the Uralic shows a tendency towards palatalisation of the onset by the lateral (maybe the onset comes from **mj > *n?), and the correspondence is less obvious than would be expected for a loanword pair.

[KathTheDragon]

The problem with that comparison is that the connections between Hittite me(y)u-, together with its Anatolian counterparts, and the rest of IE, are uncertain, so the semantics of the root are unclear. What is sure, though, is that it did not mean "4", since the usual root for "4", *kʷetur-, is attested in Hittite kutruwan- "witness", i.e. a fourth party (in plaintiff, defendent, judge, witness). As such, Kloekhorst suggests that me(y)u- is a derivative of *meh₁- "measure". All this fatally undermines your idea.

To not be a total nay-sayer to Indo-Uralic comparisons, an interesting idea is suggested by Kümmel, that if we assume original implosives, we can have comparisons such as PIE *yeg- "ice" ~ PU *jäŋə- "ice", and PIE *deḱ- "perceive" ~ PU *näkə- "see", which also are impossible to explain as loanwords.

[Neek]

@Znex: Someone (either Kloekhorst or Melchert I believe, though I don't recall exactly where I saw this) compares the -n(i) with the final -ν of the Greek 1pl -μεν. It's likely to be some particle with a meaning that made it possible to become fused to plural verbs.

The first person plural in -μεν looks more like an analogical development with other endings. Compare secondary -ον, -εν, -τον, -την, -ον for 1s, 3s+ν, 2d, 3d, and 3s; primary -τον, -τον, -ουσιν for 2d, 3d, 3p+ν. Secondary should give us -με, and I’m wondering if it would have received movable ν, but was ultimately retained with pressure of other forms.

Hittite 1p ending may be analogically formed from preterite (secondary?) 1s -un, 1p -wen (looks like 1s to be a virtual zero grade of *-un; 1p full grade. Isn’t, just looks like it.) Primary endings in Hittite are shamelessly preterite (secondary?) + hinc et nunc -i, with the exception of 3p (where we have primary -anzi, preterite -er.)

[Znex]

The problem with that comparison is that the connections between Hittite me(y)u-, together with its Anatolian counterparts, and the rest of IE, are uncertain, so the semantics of the root are unclear. What is sure, though, is that it did not mean "4", since the usual root for "4", *kʷetur-, is attested in Hittite kutruwan- "witness", i.e. a fourth party (in plaintiff, defendent, judge, witness). As such, Kloekhorst suggests that me(y)u- is a derivative of *meh₁- "measure". All this fatally undermines your idea.

Aw shucks, it was worth a shot. Thanks for the references!

[Nortaneous]

The problem with that comparison is that the connections between Hittite me(y)u-, together with its Anatolian counterparts, and the rest of IE, are uncertain, so the semantics of the root are unclear. What is sure, though, is that it did not mean "4", since the usual root for "4", *kʷetur-, is attested in Hittite kutruwan- "witness", i.e. a fourth party (in plaintiff, defendent, judge, witness). As such, Kloekhorst suggests that me(y)u- is a derivative of *meh₁- "measure". All this fatally undermines your idea.

If you buy Gasiorowski's *kʷet- 'group into pairs', kutruwan- could potentially be an earlier stage of what later gave 'four' -- sort of the inverse of a hypothetical development of "odd man out" -> "three".

Gasiorowski seems to think kutruwan- may not even be related to *kʷetur-, but I don't know why.

[Soap]

Well, if theyre direct cognates, it would have to involve either metathesis of /u/ or deletion of it, so either way it'd be irregular. If its just / kʷet-/ its likely to have just meant "pair(ed)" in both languages.
I'm not sure if /kʷe/>/ku/ is normal or not for Hittite.

[Zaarin]

Well, if theyre direct cognates, it would have to involve either metathesis of /u/ or deletion of it, so either way it'd be irregular. If its just / kʷet-/ its likely to have just meant "pair(ed)" in both languages.
I'm not sure if /kʷe/>/ku/ is normal or not for Hittite.

Hittite <ku> is /kʷ/.

[KathTheDragon]

Yeah, the Hittite form is /kʷtru-/

[Soap]

There's also apparently teriwan "judge", which looks to me like it means there was a suffix -wan, to which were attached roots teri- and kutru-, meaning "judge" and "witness". These looks like obvious cognates to the mainline PIE words for three and four. Im not entirely convinced though.

Linking kutru- to kʷetwóres has a couple of issues ... one is that, apparently, Hittite ku can be read either as /ku/ or /kʷe/, even before a consonant cluster, so we can't know which was the intended reading. I find /kʷtr/- an unlikely word-initial cluster but perhaps Hittite is known to have clusters like that in other roots as well. But a greater problem is that the second /u/ is in the wrong place ... the PIE word has it before the /r/, not after, and with an additional vowel in between the two, making it unlikely that metathesis was responsible. This can be explained by saying that only the /kut/- stem is cognate, and that the /r/ is part of a different morpheme, and perhaps that the word for witness is a derivation of the word for judge. This makes perfect sense because the PIE word for four is reconstructed with a variant stem, kʷetisres, attested in some families in the feminine declension, paralleling the feminine declension of the numeral for three. Thus, the PIE stem itself was a compound, built on a root that was originally just kʷet-.

But this reduces the coincidence to just two letters: /kʷ/ and /t/, and it's not clear what the original meaning of the root was. Anatolian is half of the Indo-Hittite family, and if we weight the two branches equally, we can't assume that the wider PIE meaning was the ancestral one.

Getting back to the original goal, though, I don't think the evidence for a connection between meyu- and neljä is any firmer. Given the evidence above, I would say that it's likely that pre-PIE speakers either did not have a word for four, or had one that was derived from another root in the language, such that it was easily replaced. I'd expect that the ancestors even further back likely were in the same situation, and that the roots are unlikely to be related.

[Howl]

Speaking of which, I have a fondness for a possible correspondence between Indo-Hittite *meh1w- {4} and Uralic *neljä {4}, though that might be a bit of a stretch. It just strikes me as interesting since this older(?) PIE word for {4} has a nasal onset, and the Uralic shows a tendency towards palatalisation of the onset by the lateral (maybe the onset comes from **mj > *n?), and the correspondence is less obvious than would be expected for a loanword pair.

For this to work, both m <-> n(j) and w <-> l(j) would have to happen. Finno-Uralic *mu 'second, another' (UEW#554) might be a better candidate.

Given the evidence above, I would say that it's likely that pre-PIE speakers either did not have a word for four, or had one that was derived from another root in the language, such that it was easily replaced. I'd expect that the ancestors even further back likely were in the same situation, and that the roots are unlikely to be related.

The Samoyedic branch of Uralic has totally different roots for the numbers than the rest of Uralic. So the numbers were not part of the basic vocabulary back then, assuming they even had them.

[Soap]

Also: laryngeals are real!! Everyone, even scholars, keeps trying to make connections like *teh2r~dór (dummy example) as if the laryngeal was just a relic of vowel coloring. Sure,that's a theory,but most people think it was the other way aeound.. if anything, a laryngeal should get stronger the further back we go, and should have a protoform like /k/,/x/,/g/, etc who would be likely to be kept in the other language.

[Zaarin]

I find /kʷtr/- an unlikely word-initial cluster

Why so? Many languages allow labiovelars in clusters. Cf. the Alaskan town of Yakutat, from Tlingit Yaakwdáat /jaːkʷtáːtʰ/.

[KathTheDragon]

Linking kutru- to kʷetwóres has a couple of issues ... one is that, apparently, Hittite ku can be read either as /ku/ or /kʷe/, even before a consonant cluster, so we can't know which was the intended reading. I find /kʷtr/- an unlikely word-initial cluster but perhaps Hittite is known to have clusters like that in other roots as well. But a greater problem is that the second /u/ is in the wrong place ... the PIE word has it before the /r/, not after, and with an additional vowel in between the two, making it unlikely that metathesis was responsible. This can be explained by saying that only the /kut/- stem is cognate, and that the /r/ is part of a different morpheme, and perhaps that the word for witness is a derivation of the word for judge. This makes perfect sense because the PIE word for four is reconstructed with a variant stem, kʷetisres, attested in some families in the feminine declension, paralleling the feminine declension of the numeral for three. Thus, the PIE stem itself was a compound, built on a root that was originally just kʷet-.

You have a nice plausible argument, but there are little holes that you probably aren't aware of. Firstly, initial *kʷtr- is paralleled by Latin quadr-, with an a that can only be explained by anaptyxis from such a preform. Metathesis is actually unavoidable, unless you want to deny that this Latin form is cognate, since tmk the *u couldn't have been syncopated early enough for the *t to have been voiced before the *r. Besides, metathesis of *ur to *ru is known from elsewhere, e.g. Latin nervus < *snéh₁ur. The presence of the *o is also a non-issue, because it isn't present in the weak stem *kʷtur-. Also, your comparison of the feminine stems of 3 and 4 is weak, since they don't line up: *tisres vs. *kʷetesres, so only the *-sr- suffix is common, and your assumption of 4 being a compound is unsupported.

[Howl]

Also: laryngeals are real!! Everyone, even scholars, keeps trying to make connections like *teh2r~dór (dummy example) as if the laryngeal was just a relic of vowel coloring. Sure,that's a theory,but most people think it was the other way aeound.. if anything, a laryngeal should get stronger the further back we go, and should have a protoform like /k/,/x/,/g/, etc who would be likely to be kept in the other language.

Is there any specific article you are referring to? And if so, do you have a link or reference?

For one thing, teh2r must be a dummy example, since laryngeals between a vowel and a resonant are really rare. Resonants are more sonorous than laryngeals in PIE. So something like terh2 would be more common. And that also illustrates that laryngeals were not really lightweight in PIE.

But let's assume that part of the laryngeals were actually more light in pre-PIE. Hittite already doesn't reflect all h2's with a h where a h is expected. So it is not a very unreasonable assumption. Then something like teh2r might be possible in pre-PIE. Later, as these light laryngeals got heavier (h->x) and merged with more heavy laryngeals, such a root would undergo metathesis and become terh2. Would something like this leave any traces? Well, we have the Saussure effect, where some of these terh2 roots become tor (without laryngeal) in O-grade. To me, this points to an original situation of teh2r ~ tōr.

[Nortaneous]

You have a nice plausible argument, but there are little holes that you probably aren't aware of. Firstly, initial *kʷtr- is paralleled by Latin quadr-, with an a that can only be explained by anaptyxis from such a preform.

Couldn't it just be vowel reduction to *kʷət(u)r-? (*ə > a, cf. pater)

Then again, there are attempts to connect Oscan (ordinal, I think) forms like trutas to *kʷtru-.

[KathTheDragon]

Certainly it's possible that the vowel was never fully lost in whatever shenanigans happened to create ablaut, though this requires further work in treating the Hittite reflex, namely you have to assume something like *kʷət- > /kʷot-/ kut- since that's a possible reading, similar to e.g. *kʷr̩- > /kʷor-/ kur-.

[Znex]

I'm trying to find in PIE where there are roots containing *kw as opposed to *kʷ in reference to a remark where Celtic and Italic languages treat them differently diachronically; namely, Celtic merges them but Italic does not. In a cursory search, I've only found *kwep- which yields "cupio" in Latin (also "kupyati" in Sanskrit and other reflexes in Balto-Slavic), though previously I was not even aware of the existence of *kw in PIE; I mistook the reference to refer to *ḱw initially.

Are there any other particular roots that would be helpful but hard to find?

[KathTheDragon]

The problem is that *kw with plain *k is obscenely rare - I don't even know of any such roots offhand. LIV doesn't list any roots starting *kw- that have reflexes in Latin (it doesn't have cupio) so that doesn't help.

[Nortaneous]

Stupid question: does loss of syllabic resonants apply before or after loss of laryngeals? If you have a *VHRC sequence, the R would be reconstructed as syllabic for PIE, but what happens with the reflexes?

[Howl]

I'm trying to find in PIE where there are roots containing *kw as opposed to *kʷ in reference to a remark where Celtic and Italic languages treat them differently diachronically; namely, Celtic merges them but Italic does not. In a cursory search, I've only found *kwep- which yields "cupio" in Latin (also "kupyati" in Sanskrit and other reflexes in Balto-Slavic), though previously I was not even aware of the existence of *kw in PIE; I mistook the reference to refer to *ḱw initially.

All forms seem to be derived from a zero-grade *kup. But I don't understand how one can link the semantics of 'to boil' (Lit. kūpėti, OCS kypeti, Skt. kupyati) with 'to desire' (Lat. cupiō, Eng to hope).

[KathTheDragon]

Stupid question: does loss of syllabic resonants apply before or after loss of laryngeals? If you have a *VHRC sequence, the R would be reconstructed as syllabic for PIE, but what happens with the reflexes?

As far as I know the development is specific to each branch, since neither laryngeals nor syllabic resonants were lost before the breakup.

[Astraios]

But I don't understand how one can link the semantics of 'to boil' (Lit. kūpėti, OCS kypeti, Skt. kupyati) with 'to desire' (Lat. cupiō, Eng to hope).

“I’m boiling (hot)” > “I’m hot for”.

[jal]

“I’m boiling (hot)” > “I’m hot for”.

Stranger things have happened...


JAL

[Vijay]

I'm warming up to the idea that Howl just might see how humans might associate increases in temperature with concepts like desire and hope.