The Great Proto-Indo-European Thread

[Tropylium]

In other news, a new overview of the scope of Szemerényi's law.

[KathTheDragon]

Interesting article! I think that's more or less the position I've been taking (but R does not include *m, in order to protect the accusative plurals of vowel-stems).

[MysteryMan23]

So, I'm a little confused on PIE's secondary derivation and how it actually works.

So, there's specific suffixes for secondary derivation when it comes to making verbs. How about nouns? And can the affixes used for primary derivation also be used for secondary derivation? And is there a good resource on this? Thank you.

[KathTheDragon]

I admit that PIE *k^wekw^wlos 'wheel' uses a not very common way of derivation, but a fitting one - a wheel does not just turn, it turns and turns and turns and turns - and the pattern is not unique, either. I can think of at least one other word with zero grade in the root and e-grade in the reduplication syllable from the top of my head, namely *h₁meh₁msos 'meat'.

But let's take this discussion to the Great Proto-Indo-European Thread.

Is this really a reduplicated word? Source? And don't *CR- clusters just reduplicate as *C-, not *CR-?

[Richard W]

And don't *CR- clusters just reduplicate as *C-, not *CR-?

HC- had a tendency to reduplicate as HC-. The best-known example is Attic reduplication.

[KathTheDragon]

And don't *CR- clusters just reduplicate as *C-, not *CR-?

HC- had a tendency to reduplicate as HC-. The best-known example is Attic reduplication.

Attic reduplication can be generated from *He-HC- reduplication, however. See this article. And if the root *h₁neḱ-, for example, did reduplicate as *h₁ne-h₁nóḱ-e, why is this reflected as Vedic ānāśa instead of *nānāśa?

[hwhatting]

Is this really a reduplicated word? Source? And don't *CR- clusters just reduplicate as *C-, not *CR-?

NIL s.v. *mē(m)s- mentions reduplication as one possibility. It doesn't posit any laryngeal in this root, but says that there have been attempts to link it to *Hem- "raw". In any case, there originally seems to have been a consonant stem noun *mē(m)s- that is attested in Sanscrit and perhaps in Avestan (the problem in Avestan seems to be that the word is not unambiguously attested, but a conjecture.) The exact relationship of the o-stem *me(:)(m)so- found in other IE languages to this consonant stem is not clear - old (PIE) derivation or late & independent thematization of the original consonant stem in the daughter languages?
Neither NIL nor Pokorny (IEW 725) track this word back to a root, so it's not even clear what, if anything, has been reduplicated here. The length of /e:/ may go back to reduplication (something like Elf's *H1me-H1ms-, with all the problems you noted), but it also may be a regular development of the sequence *-ems- in at least some case forms of the consonant stem noun. The root could then as well be *mes-, with a regular reduplication *me-ms- or it coould be that the root is *mem- and the -s- is actually a suffix.
Short summary: It has been speculated that the "meat" word has been formed by reduplication, but that's only one among the possibilities, and the formation of this word has a lot of problems, plus we don't even know what the root is.

[Richard W]

And don't *CR- clusters just reduplicate as *C-, not *CR-?

HC- had a tendency to reduplicate as HC-. The best-known example is Attic reduplication.

Attic reduplication can be generated from *He-HC- reduplication, however. See this article.

Optimality arguments that aren't summarised in English leave me checking my wallet. I've seen some lovingly crafted examples that explain why certain well-attested Thai words don't exist. I know why words like them are infrequent - history pure and simple.

In this case, the 'Obligatory Contour Principle - Syllable' (OCP-SYLL) 'constraint' leaves me wondering why the perfect of Greek histēmi is hestēka rather than estēka. Is the aspiration supposed to have suddenly restored itself after the context-sensitive lenition s > h started?

And if the root *h₁neḱ-, for example, did reduplicate as *h₁ne-h₁nóḱ-e, why is this reflected as Vedic ānāśa instead of *nānāśa?

Why didn't the 'Obligatory Contour Principle - Syllable' apply to Sanskrit? The laryngeals are still there in at least the early part of the Rg Veda - Vedic tape recorders aren't all they're cracked up to be.

The answer is that each language chose how to handle the reduplication of clusters - and the rules varied over time, and indeed different reduplication processes had different rules. Compare Latin present sistō but perfect stetī with the Greek forms above. Are you going to tell me that OCP-SYLL acted in Latin perfect reduplication? If so, why not in present stem reduplication?

[Travis B.]

Stuff like this is why I tend to be suspicious of ideas that natural languages are supposed to conform to any particular kind of theory.

[KathTheDragon]

Optimality arguments that aren't summarised in English leave me checking my wallet. I've seen some lovingly crafted examples that explain why certain well-attested Thai words don't exist. I know why words like them are infrequent - history pure and simple.

In this case, the 'Obligatory Contour Principle - Syllable' (OCP-SYLL) 'constraint' leaves me wondering why the perfect of Greek histēmi is hestēka rather than estēka. Is the aspiration supposed to have suddenly restored itself after the context-sensitive lenition s > h started?

Would you like me to summarise it for you? With hestēka, the initial aspiration would be there for exactly the same reason that it's there in histēmi - it was never lost, because the initial syllable never contained two /s/. They are syllabified as /hi.stē.mi/ and /he.stē.ka/ respectively. The reason for this is that Greek allowed sT onsets.

Why didn't the 'Obligatory Contour Principle - Syllable' apply to Sanskrit? The laryngeals are still there in at least the early part of the Rg Veda - Vedic tape recorders aren't all they're cracked up to be.

Simple - Vedic didn't ban HC onsets, as Greek did. Thus the syllabification was Ha.Hnā.śa prior to laryngeal loss.

Compare Latin present sistō but perfect stetī with the Greek forms above. Are you going to tell me that OCP-SYLL acted in Latin perfect reduplication? If so, why not in present stem reduplication?

No, I'm going to tell you that Latin did not ban sT onsets, that the identification of sistō as a reduplicated present was lost early (likely since there were exceptionally few reduplicated presents in Latin) and stetī was remodelled according to a new productive way of forming reduplicated perfects.

In summary, you should've read the article properly. You clearly didn't, since you aren't even stating the assumptions properly.

[Richard W]

In this case, the 'Obligatory Contour Principle - Syllable' (OCP-SYLL) 'constraint' leaves me wondering why the perfect of Greek histēmi is hestēka rather than estēka. Is the aspiration supposed to have suddenly restored itself after the context-sensitive lenition s > h started?

Would you like me to summarise it for you? With hestēka, the initial aspiration would be there for exactly the same reason that it's there in histēmi - it was never lost, because the initial syllable never contained two /s/. They are syllabified as /hi.stē.mi/ and /he.stē.ka/ respectively. The reason for this is that Greek allowed sT onsets.

I do get the impression that the derivation is breaching the KISS constraint.

The paper uses curious syllabification rules - certainly not the ones evidenced by the makinɡ of position for metre. If we throw in SONORITY SEQUENCE, which appears to require *strictly* monotonic increasing sonority in onsets, we have to define the sonority sequence for Greek. We can't base it on word onsets, because the kt- is an allowed onset (kteinō 'to kill'); we clearly have to group voiceless stops as at the same level, because allegedly SONORITY SEQUENCE rules out *ke.kton.a as a possible perfect of the verb. If apply the normal hierarchy, we would have s >= t in sonority, and so attested he.stē.ka should be ruled out! We can't use other verbs to determine the Greek sonority hierarchy, because in fact unless the initial cluster is stop plus sonorant, the reduplicating syllable is simply e-. As I said before, check your wallet!

I have the feeling that the perfect hestēka owes its h- to analoɡy with the present stem histēmi.

[KathTheDragon]

I have the feeling that the perfect hestēka owes its h- to analoɡy with the present stem histēmi.

Then where did histēmi get it? By your own logic, the present shouldn't have the h either.

[Richard W]

I have the feeling that the perfect hestēka owes its h- to analoɡy with the present stem histēmi.

Then where did histēmi get it? By your own logic, the present shouldn't have the h either.

Attic reduplication is very rare in the present tense - arariskō 'to join' and ollumi 'to destroy' are the only examples I can find. Additionally, there is no 'repair mechanism' for the reduplicated present corresponding to the perfect prefix e-.

I have found another example - piptō 'to fall' with perfect peptōka. Either OCP-SYLL fails, or again SONORITY SEQUENCE has to yield to the rules for initial syllables.

There's also tiktō 'to bear, to beget' which seems to be derived by metathesis from *titkō. The root is ablautinɡ tek, whence innocent-lookinɡ perfect tetoka. Perhaps you'll suggest tweaking the rules to allow metathesis as a repair strategy.

[KathTheDragon]

There's also tiktō 'to bear, to beget' which seems to be derived by metathesis from *titkō. The root is ablautinɡ tek, whence innocent-lookinɡ perfect tetoka. Perhaps you'll suggest tweaking the rules to allow metathesis as a repair strategy.

Actually, this is an example of a general metathesis of the "thorn clusters" *TK > Gk KT, eg. *h₂rtḱo- > Gk arktos (cf. Hitt. hartagga-

[Richard W]

Which pushes things back! I was just quoting the 1871 abridgement of Liddell & Scott. I had the impression that the resolution of thorny clusters was more recent.

There's no such word as Hittite hartagga- any more. The spelling is something like ha-ar-tag-ga- (too lazy to check on accents and subscripts, so I won't even try to write it properly), which yields modern Hittite hartka-. There may be Hittite-specific rules on how to transliterate U+12056 CUNEIFORM SIGN DAG (if that's how the word is spelt - I can't find a dictionary); perhaps ták?

[KathTheDragon]

Which pushes things back! I was just quoting the 1871 abridgement of Liddell & Scott. I had the impression that the resolution of thorny clusters was more recent.

There's no such word as Hittite hartagga- any more. The spelling is something like ha-ar-tag-ga- (too lazy to check on accents and subscripts, so I won't even try to write it properly), which yields modern Hittite hartka-. There may be Hittite-specific rules on how to transliterate U+12056 CUNEIFORM SIGN DAG (if that's how the word is spelt - I can't find a dictionary); perhaps ták?

I was under the impression that citing words in bound transcription was the usual thing to do. Certainly all my references do it. Such as Kloekhorst:

... Hitt. ḫartakka- /Hrtka-/ ...

I used -gg- since this is the more usual spelling, cf. all attested spellings of this word: nom.sg. ḫar-tág-ga-aš, acc.sg. ḫar-ták-kán (OS), ḫar-tág-ga-an, gen.sg. ḫar-tág-ga-aš.

Also, 1871? Isn't that a bit dated now?

[Richard W]

Also, 1871? Isn't that a bit dated now?

Has Classical Greek changed much since then? It's only an abridgement, after all.

[KathTheDragon]

Also, 1871? Isn't that a bit dated now?

Has Classical Greek changed much since then? It's only an abridgement, after all.

No, but major advances in IE linguistics have been made since then.

[Tropylium]

On some PIE allomorphies induced by s-mobile

An interesting PIE-ish soundlaw / morphophonological rule: *s mobile + labialized voiced velar yields *sw-. This seems to resolve several apparent doublets, e.g. *swel- ~ *gʷel- 'swallow', *swenH- ~ *ǵʰwen- 'sound'.

[WeepingElf]

On some PIE allomorphies induced by s-mobile

An interesting PIE-ish soundlaw / morphophonological rule: *s mobile + labialized voiced velar yields *sw-. This seems to resolve several apparent doublets, e.g. *swel- ~ *gʷel- 'swallow', *swenH- ~ *ǵʰwen- 'sound'.

Thank you, that looks interesting.

[hwhatting]

Thanks for porting that. Intriguing, also the quoted proposal by Oettinger that *sH2V- > *dV.

[KathTheDragon]

Thanks for porting that. Intriguing, also the quoted proposal by Oettinger that *sH2V- > *dV.

I'm really curious what the justification is for that, but I didn't spot it. It seems to be [sχ] > [d] or [ɗ].

Edit: There also seem to be a number of words in various languages with the initial sequence *sh₂V- which do not turn up with *d, where the *s is part of the root.

[Zju]

Thanks for porting that. Intriguing, also the quoted proposal by Oettinger that *sH2V- > *dV.

I'm really curious what the justification is for that, but I didn't spot it. It seems to be [sχ] > [d] or [ɗ].

Edit: There also seem to be a number of words in various languages with the initial sequence *sh₂V- which do not turn up with *d, where the *s is part of the root.

Ditto. The furthest I can imagine is tχ < stχ < sχ. But I'm more concerned with the counterexamples of *sh₂; after all there were those SCs b > c / #_V nc / V_V and gr > kʰ.

It seems to me that if this is true it's not a mere sound change, but maybe a relict of an older alternation, an apophony, something along the lines of:
ɗ > ? > χ / some condition
sɗ > zɗ > ɗ
Not the most likely development in the world, I know, but just to give an example.

[hwhatting]

I haven't read the Oettinger article, so I don't know how much supprting examples are there besides the "tear" word. I'd need more than one or two examples to believe in that rule. And yes, if the rule would be valid, it probably would mean that PIE *d was something else than a voiced stop when the rule was active (maybe a voiceless dental fricative).
The "tear" word is one of the cases where Latin has /l/ for PIE /d/. Another case is lingua "tongue", where Vedic also has an irregular reflex. Can someone help my memory, are there more cases of Latin /l/ for PIE /d/ than those two?

[Valdeut]

I wonder how this relates to some other proposed correspondences of Hittite s, Luwian t and Non-Anatolian 0. From this article:
https://www.academia.edu/18952423/Proto ... nt_of_view

(p. 315)
"An additional IE fricative *θ should probably be reconstructed for the correspondence Hittite s / Luwian t / Narrow IE 0, as proposed in Ivanov 2001: 133; 2009: 5 and (independently) in Kassian & Yakubovich 2013: 22. In our list, this affects two items: #13 ‘fingernail’ & #25 ‘eye’. In both cases, we double the reconstructed forms with and without such a fricative, e.g., ‘eye’ is reconstructed as two synonyms *okʷ- ~ *okʷ- (i.e., TK ~ HK in the simplified transcription)."

(p. 327)
θmogʰ ~ θn̥gʰ ‘fingernail’ (traditional *h₃negʰ-)
Luwian tam:uga
Hittite sank(-)uwai-

(p. 392)
θokʷ- ‘eye’ (traditional *h₃ekʷ-)
Luwian tawi-
Hittite saguwa-

I have not been able to find the original articles mentioned above, and only one of them are in English, the other two in Russian.

I haven't read the Oettinger article, so I don't know how much supprting examples are there besides the "tear" word. I'd need more than one or two examples to believe in that rule. And yes, if the rule would be valid, it probably would mean that PIE *d was something else than a voiced stop when the rule was active (maybe a voiceless dental fricative).
The "tear" word is one of the cases where Latin has /l/ for PIE /d/. Another case is lingua "tongue", where Vedic also has an irregular reflex. Can someone help my memory, are there more cases of Latin /l/ for PIE /d/ than those two?

Oettinger's original article was apparently about the word *dai̯u̯ḗr (*sh₂ai̯u̯ḗr) meaning ‘brother-in-law’, based on the title (there is a small list of sources in the Academia page linked above).
OETTINGER, Norbert. 2009. Die Derivationsbasis von idg. *dai̯ué̯r- (*sh2ai̯u̯ér-) ʻBruder des Ehemannesʼ. Sprache 48, 127-131.
I haven't found the article, all I could find was a table-of-content for the Festschrift were the article was included: http://www.harrassowitz-verlag.de/dzo/a ... 1285676078

I don't know what IE languages have words that could go back to *h₂ai̯u̯ḗr without the s-mobile. But interestingly, the latin cognate of *dai̯u̯ḗr is lēvir so this is another example of the change *d > l.
https://en.wiktionary.org/wiki/levir#Latin

For other examples of the latin change *d > l, there is the verb oleō ’to smell’ from PIE *h₃ed-. For the related noun, both olor and odor are found. There is perhaps also littera from Greek διφθέρα via Etruscan, although I'm not sure how certain that etymology is.

I used to think that Ulysses for Odysseus was another example, but the forms with l are apparently found already in Greek (actually, there are a lot of forms of this name).

Edit:

Oettinger talked about this on the Anatolian Spring conference in Copenhagen 2011
http://rootsofeurope.ku.dk/english/cale ... an_spring/

From the lecture notes:
"It has been suggested that in Luvian dental stop in word initial position partly stems from Proto-Anatolian *sH-, e.g. dūr ‘urine’ from *sh2ur, cf. Hittite sēhur. This development will be treated in detail, discussing ways of possible phonological explanations.

Besides, an attempt will be made to show a phonologically parallel development in PIE itself. Three samples will be used for this purpose: late PIE *dák̂ru- (and *ák̂ru-) ‘tear’; *daiwér- ‘brother of husband’; and *dou-s- ‘upper arm’."

The lecture (in German) is available here:
http://podcast.hum.ku.dk/mediaviewer/?objectId=325
(click See also/Flere optagelser and scroll to lektion 6)

Your German is probably better than mine, I would imagine.