The Great Proto-Indo-European Thread's Sequel

[Travis B.]

Those cases present their own issues, namely the rare correspondence of Hittite š- with Luwic t-, which is wholly unexplained, and laryngeal explanations are quite unconvincing.

Presumably a laryngeal explanation would be /sh₃/ > /ʃ‌/ > /tʃ/ > /t/. My thought is inspired by Dutch <sch> and the deaffrication seen in a few languages, such as the development of Egyptian <t̲>.

I can attest to /ʃ‌/ > /tʃ/ in the English here, specifically in the word sure when pronounced with NURSE.

[KamareKitsune]

Those cases present their own issues, namely the rare correspondence of Hittite š- with Luwic t-, which is wholly unexplained, and laryngeal explanations are quite unconvincing.

Presumably a laryngeal explanation would be /sh₃/ > /ʃ‌/ > /tʃ/ > /t/. My thought is inspired by Dutch <sch> and the deaffrication seen in a few languages, such as the development of Egyptian <t̲>.

The first steps aren't implausible, but does /tʃ/ > /t/ actually happen? Afaik there's no solid evidence for Egyptian <ṯ> actually being an affricate (at least at the relevant time), rather than a straight palatal stop.

[Travis B.]

Those cases present their own issues, namely the rare correspondence of Hittite š- with Luwic t-, which is wholly unexplained, and laryngeal explanations are quite unconvincing.

Presumably a laryngeal explanation would be /sh₃/ > /ʃ‌/ > /tʃ/ > /t/. My thought is inspired by Dutch <sch> and the deaffrication seen in a few languages, such as the development of Egyptian <t̲>.

The first steps aren't implausible, but does /tʃ/ > /t/ actually happen? Afaik there's no solid evidence for Egyptian <ṯ> actually being an affricate (at least at the relevant time), rather than a straight palatal stop.

Yeah, when deaffrication occurs it crosslinguistically is far more likely to be to a fricative rather than to a stop.

[Richard W]

Yeah, when deaffrication occurs it crosslinguistically is far more likely to be to a fricative rather than to a stop.

Well, there's still a path - /tʃ/ > /ts/ > /θ/ > /t/, though some ducking and diving is needed to avoid a collision with Luwian /ts/ of various origins, including thorn clusters, such as in Cuneiform Luwian īnzagan- (sp?) 'inhumation'. Of course, the final step need not have happened - workable writing systems don't have to distinguish all sounds.

[Brilliant-Resource14]

Are you referring to išḫaḫru "tear"? I can't find any other words where (pre-)Hittite *sh₂- would correspond to IE *h₂-. And, in this case, the etymology connecting it to *h₂eḱru- isn't as strong as you'd like.

like "eye" and "fingernail" (the only examples i could find on swadesh)

But these suggest *sh₃ rather than *sh₂, and ša-ku-u̯a and ša-an-ku-wa-i show different treatments of the putative cluster.

sameshit in anatolian

[Travis B.]

Yeah, when deaffrication occurs it crosslinguistically is far more likely to be to a fricative rather than to a stop.

Well, there's still a path - /tʃ/ > /ts/ > /θ/ > /t/, though some ducking and diving is needed to avoid a collision with Luwian /ts/ of various origins, including thorn clusters, such as in Cuneiform Luwian īnzagan- (sp?) 'inhumation'. Of course, the final step need not have happened - workable writing systems don't have to distinguish all sounds.

As you allude to, the problem with these sound change chains is that if any other sounds in the chain are found elsewhere, one has to explain how they weren't also dragged along by them.

[KamareKitsune]

like "eye" and "fingernail" (the only examples i could find on swadesh)

But these suggest *sh₃ rather than *sh₂, and ša-ku-u̯a and ša-an-ku-wa-i show different treatments of the putative cluster.

sameshit in anatolian

Actually no - *h₂ and *h₃ are treated rather differently in Hittite, so Anatolian did in fact distinguish them.

[Nortaneous]

Those cases present their own issues, namely the rare correspondence of Hittite š- with Luwic t-, which is wholly unexplained, and laryngeal explanations are quite unconvincing.

Presumably a laryngeal explanation would be /sh₃/ > /ʃ‌/ > /tʃ/ > /t/. My thought is inspired by Dutch <sch> and the deaffrication seen in a few languages, such as the development of Egyptian <t̲>.

The first steps aren't implausible, but does /tʃ/ > /t/ actually happen?

*c > t in most of Samoyedic, *tɬ > t in Pipil (which apparently isn't a retention?)

There are also cases of sibilants turning into plosives. *s > t happens in Vietnamese and most of Micronesian.

[KamareKitsune]

Presumably a laryngeal explanation would be /sh₃/ > /ʃ‌/ > /tʃ/ > /t/. My thought is inspired by Dutch <sch> and the deaffrication seen in a few languages, such as the development of Egyptian <t̲>.

The first steps aren't implausible, but does /tʃ/ > /t/ actually happen?

*c > t in most of Samoyedic, *tɬ > t in Pipil (which apparently isn't a retention?)

There are also cases of sibilants turning into plosives. *s > t happens in Vietnamese and most of Micronesian.

Ok, so each step in the chain is possible. The timing probably works better if the first step /sh₃/ > /ʃ/ happened in Anatolian already, where Hittite merged it with existing /s/ and Luwic did /ʃ/ > /tʃ/ > /t/ before the breakup of that subgroup (since Lycian also has /t/ in "eye"). "Eye" also presents another issue - the labiovelar is lenited, and there's no clear reason for it (I disagree with Kloekhorst's lenition after *ó on other grounds). Another problem is that this is the only word showing Hittite /s/ = Luwic /t/ that even could plausibly be taken back to *sh₃ - neither Hittite šākan ~ Luwian tāin "oil" nor Hittite sēḫur ~ Luwian *dūr "urine" has a good etymology. And the only other word that could plausible reflect *sh₃, "nail", has no Luwic cognate. So this is speculative at best.

[Richard W]

The timing probably works better if the first step /sh₃/ > /ʃ/ happened in Anatolian already,

I was thinking of Hittite having /sh₃/ > /sah₃/ (or whatever), Perhaps that doesn't work for Hittite, and the hiatus is purely orthographic, as in the usual way of writing short-vowelled CVC.

[KamareKitsune]

The timing probably works better if the first step /sh₃/ > /ʃ/ happened in Anatolian already,

I was thinking of Hittite having /sh₃/ > /sah₃/ (or whatever), Perhaps that doesn't work for Hittite, and the hiatus is purely orthographic, as in the usual way of writing short-vowelled CVC.

What hiatus?

EDIT: Oh, you mean the spelling ša-a-? No, that's just how long vowels are written.

[Richard W]

The timing probably works better if the first step /sh₃/ > /ʃ/ happened in Anatolian already,

I was thinking of Hittite having /sh₃/ > /sah₃/ (or whatever), Perhaps that doesn't work for Hittite, and the hiatus is purely orthographic, as in the usual way of writing short-vowelled CVC.

What hiatus?

EDIT: Oh, you mean the spelling ša-a-? No, that's just how long vowels are written.

The spelling I have found quoted is ša-an-ku-wa-i. I was remembering that *h₃ can 'vanish', so I was looking for a syllable starting with a vowel, and misread an as such, forgetting the standard writing of CVC with a short vowel as CV-VC. I don't know what the rule is for word-initial *h₃ to vanish; I'd got it into my head that it always did, which is wrong, at least under the 3-laryngeal reconstruction.

[KamareKitsune]

I was thinking of Hittite having /sh₃/ > /sah₃/ (or whatever), Perhaps that doesn't work for Hittite, and the hiatus is purely orthographic, as in the usual way of writing short-vowelled CVC.

What hiatus?

EDIT: Oh, you mean the spelling ša-a-? No, that's just how long vowels are written.

The spelling I have found quoted is ša-an-ku-wa-i. I was remembering that *h₃ can 'vanish', so I was looking for a syllable starting with a vowel, and misread an as such, forgetting the standard writing of CVC with a short vowel as CV-VC. I don't know what the rule is for word-initial *h₃ to vanish; I'd got it into my head that it always did, which is wrong, at least under the 3-laryngeal reconstruction.

Word-initial *h₃ is retained before a vowel, and lost otherwise, cf. ḫāraš "eagle" < *h₃érons, arāi "rises" < *h₃ri- (by whichever scenario you prefer for deriving i-stem verbs). It's lost medially too, but I doubt /sah₃ongʷ-/ would end up spelled as anything other than ša-a-an-ku-. Not that the epenthesis you imagine is at all paralleled in Hittite, *sh₃- would almost certainly remain a cluster until *h₃ is lost.

[Richard W]

Word-initial *h₃ is retained before a vowel, and lost otherwise, cf. ḫāraš "eagle" < *h₃érons, arāi "rises" < *h₃ri- (by whichever scenario you prefer for deriving i-stem verbs). It's lost medially too, but I doubt /sah₃ongʷ-/ would end up spelled as anything other than ša-a-an-ku-. Not that the epenthesis you imagine is at all paralleled in Hittite, *sh₃- would almost certainly remain a cluster until *h₃ is lost.

Interesting set of Wiktionary errors then:

Page	PIE source	Cuneiform	Translit	Reading	Gloss
h₃nṓgʰs	h₃nógʰ-u-s ~ *h₃n̥gʰ-éw-s	𒊭𒀭𒆪𒉿𒄿	ša-an-ku-wa-i		nail (PIE)
𒌑𒉡𒊻𒍣	h₃u-néw-ti	𒌑𒉡𒊻𒍣	ú-nu-uz-zi	únuzzi	to adorn
𒅈𒋫𒊑	*h₃r̥tó	𒅈𒋫𒊑	ar-ta-ri	artari	to be standing, to occur

Or do I need to get the meaning of 'vowel' clarified?

[KamareKitsune]

Word-initial *h₃ is retained before a vowel, and lost otherwise, cf. ḫāraš "eagle" < *h₃érons, arāi "rises" < *h₃ri- (by whichever scenario you prefer for deriving i-stem verbs). It's lost medially too, but I doubt /sah₃ongʷ-/ would end up spelled as anything other than ša-a-an-ku-. Not that the epenthesis you imagine is at all paralleled in Hittite, *sh₃- would almost certainly remain a cluster until *h₃ is lost.

Interesting set of Wiktionary errors then:

Page	PIE source	Cuneiform	Translit	Reading	Gloss
h₃nṓgʰs	h₃nógʰ-u-s ~ *h₃n̥gʰ-éw-s	𒊭𒀭𒆪𒉿𒄿	ša-an-ku-wa-i		nail (PIE)
𒌑𒉡𒊻𒍣	h₃u-néw-ti	𒌑𒉡𒊻𒍣	ú-nu-uz-zi	únuzzi	to adorn
𒅈𒋫𒊑	*h₃r̥tó	𒅈𒋫𒊑	ar-ta-ri	artari	to be standing, to occur

Or do I need to get the meaning of 'vowel' clarified?

As we've been talking about, šankuwai is not directly reflecting word-initial *h₃ so I won't comment on it further. artari is interesting because yes, *h₃r̩tó ought to give ḫarta(ri) (though Kloekhorst does reckon with loss of *h₃ before syllabic consonants too) but in principle the lack of ḫ could be generalised out of the 3pl *h₃rentó > aranta(ri) where loss is undoubtedly regular. únuzzi I have no explanation for since this is the first I'm looking at it. In principle we could say that the root is actually *h₁ew-, the attested o-grades are indeed o-grades and not due to vowel colouring, and the h in Armenian is secondary (as many h's do seem to be) but this is speculative and I would want a far more detailed understanding of both secondary h and the loss of primary h in Armenian before committing to it.

[KamareKitsune]

No, Kloekhorst never proposed a relationship between IE and Japanese. He proposed that the PIE word for 'name' is derived from a root *h₃neh₃- 'to name, to call' reflected in the Hittite verb ḫannai- 'to call to court'. And that seems to make sense to me.

I actually want to push back a little on Kloekhorst's reconstruction of "name" here and advocate for *h₃nómn̩.

First of all, we have Celtic *anman "name", which must reflect *(H)n̩mn̩-, and cannot be derived from a form with a medial laryngeal.

Second of all, we have Germanic *nōmijaną "to give a name", which cannot be derived directly from *namô "name" due to lacking *n (a direct derivative from the noun would have to be *namnijaną, as is indeed created later). Hence it must be inherited, or derived from some inherited formation also lacking *n, which either suggests some unknown derivative in *-mo- from the root, or else the root actually includes the *m (and I would suggest it's the latter, a causative to a lost class 6 verb *namaną "to be named", with the *ō coming from the well-attested vrddhi in this formation).

To explain the ō in Latin, we can refer to the old explanation of contamination with cognōscere in cognōmen.

To be honest, I'm also unconvinced by his derivation of Hittite ḫanna- from a root *h₃neh₃- in the first place. In particular, the phonological developments he requires are essentially unsupported (*-HR- > -RR- is only evidenced by this exact word, and *-HRH- > -RR- has no parallels I can find, and in fact there's an example walḫ- "to hit" < *welh₃- that suggests the second laryngeal should be retained in this environment) and I suspect are invented purely to allow a connection with "name". Rather, a root *h₃enh₃- would work perfectly for the Hittite and Greek verbs, as Hittite had an early transfer of simple stems ending in a geminate into the tarna-class, a possibility that Kloekhorst conveniently ignores.