PIE and PU heteroclitics

[Zju]

Has it occured to anyone that IE heteroclitics -r/-n on the one side and U heteroclitics -s/-nen on the other might be cognates? On the IE side, there could have been rhotacisation s > r, while on the U side, there could easily have been expansion of the ending n > nen. Has this resemblance been dealt with anywhere? How plausible it is? And what is the supposed origin of heteroclitics in both families?

[WeepingElf]

Has it occured to anyone that IE heteroclitics -r/-n on the one side and U heteroclitics -s/-nen on the other might be cognates? On the IE side, there could have been rhotacisation s > r, while on the U side, there could easily have been expansion of the ending n > nen. Has this resemblance been dealt with anywhere? How plausible it is? And what is the supposed origin of heteroclitics in both families?

Interesting; I did not know that. These may indeed be cognate.

[Genome]

It seems very plausible.

[Richard W]

For PIE, the sugeestion I've seen is that the ending was originally /n/, and final /n/ became /r/ but /n/ + suffix remained /n/. I've also seen the same phonetic process suggested for the 3pl ending in /r/ in the past tense of some verbs.

[WeepingElf]

For PIE, the sugeestion I've seen is that the ending was originally /n/, and final /n/ became /r/ but /n/ + suffix remained /n/. I've also seen the same phonetic process suggested for the 3pl ending in /r/ in the past tense of some verbs.

Yes; this is also what I think about them. There are a few neuter n-stems, but these are all (to my knowledge) formed with the suffix *-men where the /m/ may have prevented the denasalization of the /n/. Are there others?

The existence of broadly similar heteroclitics in Uralic may still be coincidence; it would be interesting to know whether there are IE heteroclitics that seem to correspond to Uralic heteroclitics. Otherwise, one can hardly rule out independent developments in the ancestor languages of the two families.

[Salmoneus]

Some thing there are l/n heteroclitics in PIE, though. On the other hand, those could be by analogy, especially since l-stem roots in general don't seem to have survived well as a class.

[Terra]

The existence of broadly similar heteroclitics in Uralic may still be coincidence; it would be interesting to know whether there are IE heteroclitics that seem to correspond to Uralic heteroclitics. Otherwise, one can hardly rule out independent developments in the ancestor languages of the two families.

Indeed. What PU words that have heteroclitics? (Is there a list somewhere?)

***

A list of PIE's: (That I can find searching my Leiden Latin and Celtic dictionaries.)
jek^w-r/n- (liver) (Latin jecur, jecinoris (and various other forms; This word seemed rather volatile.))
h_1ej-t-r/n- (way, journey) (Latin iter, itineris)
pet-r/n- (wing, feather) (Latin pinna)
swep-r/n- (sleep) (Latin sopor, soporis, (Leiden thinks that this isn't a rhoticized s-stem.) somnus)
wer-r/n- (spring) (Latin vér, véris)
kok^w-r/n- (excrement) (PCelt *kek^worá "swamp, mud")
wod-r/n- (water) (Latin unda)
?h_1owd^h-r/n- (udder) (Latin úber, úberis)
d^hen-w-r/n- (thigh) (Latin femur, feminis)
h_1esh_2-r/n- (blood) (Latin sanguis, sanguinis, (with a -ng^w- stem extension) assyr, aser)

??? (Vedic áhar, áhnas "day")
??? (Vedic vádhar "weapon")

Some more in Hittite: ( this paper: http://www.academia.edu/4319318/PIE_heteroclitics , a paper by a Lauffenburger, and a paper by Hoffner and Melchert)
pahhur (fire)
mehur (time)
sehur (urine)
pankur (family)
pir, parnas (house)
uttar (word, speech; matter)
lammar (hour, moment)

watar (water)
eshar (blood)

That's 17 different roots (not counting the Vedic ones). Here's hoping for some promising comparisons.

[gach]

Indeed. What PU words that have heteroclitics? (Is there a list somewhere?)

I'm pretty far from the home now so the best thing I can do is to give a short account on the kinds of words in Finnish that have the -nen ~ -s(e)- material in them:

* Diminutives are typically formed with the affix -nen, such as poikanen ("young animal, cub, fledgeling etc.") < poika ("boy") and lautanen ("plate") < lauta ("board").
* Many adjectives and in -nen (or rather -inen): valkoinen ("white"), toinen ("other"). The ending -inen is a common way to derive quality adjectives such as puinen ("wooden") < puu ("tree, wood") and ilmainen ("free [of charge]") < ilma ("air").
* -nen also appears in other adjective/noun derivations such as nationalities taking -lAinen, hämäläinen ("a person/thing from Häme"), similarities taking -lainen, tällainen ("like this"), and qualities taking -llinen, tavallinen ("ordinary"), onnellinen ("happy").
* There's also nainen ("woman") where the initial part nai- reflects the PU root *näxi and the -nen ending is a much later addition.

So if anything, you can see that here -nen is much more associated with derivational material than being an inseparable part of lexical roots.

[Salmoneus]

Hittite is full of r/n roots. In fact, there's two different classes of them - both static and proterodynamic. And apparently there's one possible hysterodynamic as well.
Since the r/n words are usually proterodynamic elsewhere, I'd suspect that the static ones (like me:hur) may be later forms by analogy - although they may instead by simplifications where the accent change has made the alternation too opaque?
Hittite also includes a general abstract suffix -a:ta:r/-a:nn, believed to be from -ótr/-ótn.

Others include pattar/pattan, 'basket', harshar/harshn- (head), lammar/lamn- (moment), mukar/mukn-, 'rattle', and so on.

Apparently sanskrit shíras/shi:rshnás (forgive spelling) is from k'er2sr/k'r2snos.

[WeepingElf]

Some thing there are l/n heteroclitics in PIE, though. On the other hand, those could be by analogy, especially since l-stem roots in general don't seem to have survived well as a class.

AFAIK, the only candidate is *sah2wel/n- 'sun', but the final *n shows up only in West Germanic; all other branches (including North Germanic) of IE have only *l. I think a better solution is to assume a substratum loanword or other quirk in West Germanic, rather than positing a new stem class with only one member in PIE to explain this West Germanic phenomenon.

[Hallow XIII]

Some thing there are l/n heteroclitics in PIE, though. On the other hand, those could be by analogy, especially since l-stem roots in general don't seem to have survived well as a class.

AFAIK, the only candidate is *sah2wel/n- 'sun', but the final *n shows up only in West Germanic; all other branches (including North Germanic) of IE have only *l. I think a better solution is to assume a substratum loanword or other quirk in West Germanic, rather than positing a new stem class with only one member in PIE to explain this West Germanic phenomenon.

Slavic has the n. Also, is "Sonne" and so on not from the alternate stem (*sh₂uén-)? Or what did you mean?

[Pole, the]

Slavic has the n. Also, is "Sonne" and so on not from the alternate stem (*sh₂uén-)? Or what did you mean?

Slavic has both the l and the n. Actually, *sŭlnĭ- is reconstructed, so it could be just l with some suffix containing n.
Other Baltic languages have only l.

[hwhatting]

Some thing there are l/n heteroclitics in PIE, though. On the other hand, those could be by analogy, especially since l-stem roots in general don't seem to have survived well as a class.

AFAIK, the only candidate is *sah2wel/n- 'sun', but the final *n shows up only in West Germanic; all other branches (including North Germanic) of IE have only *l. I think a better solution is to assume a substratum loanword or other quirk in West Germanic, rather than positing a new stem class with only one member in PIE to explain this West Germanic phenomenon.

The assumption of a heteroclitic declination is not just based on stem variations with /l/ or /n/ among daughter languages, but directly observable in Iranian (Avestan), where it is reflected as r/n heteroclisis, with the usual sound change PIE /l/ > Iranian /r/. It may well be possible that this heteroclisis is the result of a mixing of different stems, but at least IMO, the fact that it heteroclisis is actually attested for this word makes it more likely that the existence of stems in both /l/ and /n/ in the various daughter languages (with variation even inside of families like Germanic) is due to the generalization of stem variants of a heteroclitic paradigm - after all, this mirrors the situation we have with other heteroclitica, like "fire" or "water".

[Terra]

AFAIK, the only candidate is *sah2wel/n- 'sun'

There's one more: *h_2orol/n- (eagle) (Compare Greek "ornis, ornithos" (bird), Hittite "haras, haranas" (gull?))

* Diminutives are typically formed with the affix -nen, such as poikanen ("young animal, cub, fledgeling etc.") < poika ("boy") and lautanen ("plate") < lauta ("board").
* Many adjectives and in -nen (or rather -inen): valkoinen ("white"), toinen ("other"). The ending -inen is a common way to derive quality adjectives such as puinen ("wooden") < puu ("tree, wood") and ilmainen ("free [of charge]") < ilma ("air").
* -nen also appears in other adjective/noun derivations such as nationalities taking -lAinen, hämäläinen ("a person/thing from Häme"), similarities taking -lainen, tällainen ("like this"), and qualities taking -llinen, tavallinen ("ordinary"), onnellinen ("happy").
* There's also nainen ("woman") where the initial part nai- reflects the PU root *näxi and the -nen ending is a much later addition.

So if anything, you can see that here -nen is much more associated with derivational material than being an inseparable part of lexical roots.

This looks similar to where Latin has "-ánus" and "-ínus" (deus > divínus, equus > equínus, etc). (Cognate to English "-en" in "wood-en, gold-en, etc".) Anyways, where's the alteration in the Finnish forms?

[Zju]

AFAIK, the only candidate is *sah2wel/n- 'sun'

There's one more: *h_2orol/n- (eagle) (Compare Greek "ornis, ornithos" (bird), Hittite "haras, haranas" (gull?))

* Diminutives are typically formed with the affix -nen, such as poikanen ("young animal, cub, fledgeling etc.") < poika ("boy") and lautanen ("plate") < lauta ("board").
* Many adjectives and in -nen (or rather -inen): valkoinen ("white"), toinen ("other"). The ending -inen is a common way to derive quality adjectives such as puinen ("wooden") < puu ("tree, wood") and ilmainen ("free [of charge]") < ilma ("air").
* -nen also appears in other adjective/noun derivations such as nationalities taking -lAinen, hämäläinen ("a person/thing from Häme"), similarities taking -lainen, tällainen ("like this"), and qualities taking -llinen, tavallinen ("ordinary"), onnellinen ("happy").
* There's also nainen ("woman") where the initial part nai- reflects the PU root *näxi and the -nen ending is a much later addition.

So if anything, you can see that here -nen is much more associated with derivational material than being an inseparable part of lexical roots.

This looks similar to where Latin has "-ánus" and "-ínus" (deus > divínus, equus > equínus, etc). (Cognate to English "-en" in "wood-en, gold-en, etc".) Anyways, where's the alteration in the Finnish forms?

Every Finnish noun or adjective ending in -nen has an oblique stem in -s, almost always followd by -e-, so that s doesn't form a cluster with the case ending, except in the partitive singular, where -ta/ä is added directly to -s. Look here.

[hwhatting]

Every Finnish noun or adjective ending in -nen has an oblique stem in -s, almost always followd by -e-, so that s doesn't form a cluster with the case ending, except in the partitive singular, where -ta/ä is added directly to -s. Look here.

Well, that's different from the PIE situation (/n/ in the oblique, /r/ in the nom-acc).

[Tropylium]

Indeed. What PU words that have heteroclitics? (Is there a list somewhere?)

I'm pretty far from the home now so the best thing I can do is to give a short account on the kinds of words in Finnish that have the -nen ~ -s(e)- material in them:

* Diminutives are typically formed with the affix -nen, such as poikanen ("young animal, cub, fledgeling etc.") < poika ("boy") and lautanen ("plate") < lauta ("board").
* Many adjectives and in -nen (or rather -inen): valkoinen ("white"), toinen ("other"). The ending -inen is a common way to derive quality adjectives such as puinen ("wooden") < puu ("tree, wood") and ilmainen ("free [of charge]") < ilma ("air").
* -nen also appears in other adjective/noun derivations such as nationalities taking -lAinen, hämäläinen ("a person/thing from Häme"), similarities taking -lainen, tällainen ("like this"), and qualities taking -llinen, tavallinen ("ordinary"), onnellinen ("happy").
* There's also nainen ("woman") where the initial part nai- reflects the PU root *näxi and the -nen ending is a much later addition.

So if anything, you can see that here -nen is much more associated with derivational material than being an inseparable part of lexical roots.

Yes. There is no such thing as "PU heteroclitics" known. This alternation between -inen/-ise- is a feature of Finnic only. All explanations of the alternation I have seen start from (and do not progress much beyond) assuming that it is suppletion.

Etymologically, the -(i)se- formant in oblique cases goes back to *-ńće-. (This is traditionally reconstructed as *-ŋće-, in which a vocalization *ŋ → *j / _Ć would have occurred, but I think a better choice is to reconstruct a homorganic cluster, the usual denasalization *ńć → *ć, and an unpacking *ć → *jc.) This is also reflected in at least the similar Samic adjectival/deminutiv suffix *-ńćë. In Northern Sami this is reflected as -š/-žža-, which is fairly regularly used to translate corresponding Finnish words, e.g. guđeš = F kuutonen "numeral six", vašálaš = F vihollinen "enemy".

In Permic there is also a fairly similar-looking suffix *-eś. E.g. Komi /vɨj/ "butter" → /vɨjɤś/ "buttery, greasy" could be compared with Finnish voi "butter" → voinen (voise-) "buttery". The sources I have do not say anything on its etymology though…

Also relevant here is the existence of an etymologically related suffix variant -isA. E.g. starting from the noun valo "light", there are both valoisa and valoinen, both meaning "bright, light, well-lit".

At any rate this does not look very similar to IE *-r/-n- at all.

---

I have not come across any good explanation for the origin of the -inen formant, or why it would have replaced the nom.sg. of -ise-. L. Hakulinen in the handbook Suomen kielen rakenne ja kehitys suggests that this could be in some unclear way related to the largely unproductiv suffix -na ~ -n found in words like joutsen "swan", räppänä "smoke latch", and that the -i- would be by analogy to -ise-.

I figure another option might be a relationship to the Mordvinic deminutiv suffix *-ńə. E.g. in Mokša, /pal/ "bit" → /palńä/ "piece of meat" seems to be quite nicely parallel to Finnish pala "bit" → palanen "piece, fragment" (originally: *pala : *palańə, and *ń → *jn in Finnic to yield *palaine?) This is a Wild Guess though, and also doesn't explain why -(i)nen would be used in suppletion to -(i)se-. (If I had to take an even wilder guess though: what if, in plural case forms, *-ńə-j- → *-ńj- → *ńjə- → *ńćə-, followed by generalization to everywhere except the nom. sg.?)

Neither of these approaches explains the final -n either. This feels somehow related to the extension of the nom. singulars *seiccemä *kahdeksa *ühdeksä *kümmen "7 8 9 10" to seitsemän kahdeksan yhdeksän kymmenen (but in other cases or in the plural e.g. seitsemä-ksi, kahdeksa-ssa, yhdeksä-ltä, kymmene-t). Hakulinen suggests the -n in these to be the old nom. dual marker, which would be original in "8" and would have spread by analogy to the others.

[WeepingElf]

This pretty much settles the issue.

[Tropylium]

Also the only PU roots that seem likely to be related to corresponding PIE heteroclitics are *wetə "water" and *päjə "sun" (~ "fire"? reflected in Komi as /bi/ "fire"), which behave as perfectly ordinary CVCV roots.

[Salmoneus]

AFAIK, the only candidate is *sah2wel/n- 'sun'

There's one more: *h_2orol/n- (eagle) (Compare Greek "ornis, ornithos" (bird), Hittite "haras, haranas" (gull?))

There's a bunch of others that have been suggested, too. For instance, *1edwol/n, 'pain, or evil' (Hittite ida:lu, Greek odýne:).

[Zju]

And what is the origin of PIE heteroclitics then?

[Terra]

And what is the origin of PIE heteroclitics then?

For PIE, the sugeestion I've seen is that the ending was originally /n/, and final /n/ became /r/ but /n/ + suffix remained /n/. I've also seen the same phonetic process suggested for the 3pl ending in /r/ in the past tense of some verbs.

I've never heard of the sound change [n] -> [r] though.

There's a bunch of others that have been suggested, too. For instance, *1edwol/n, 'pain, or evil' (Hittite ida:lu, Greek odýne:).

Good to know.

[Salmoneus]

To clarify: although a bunch of other l/n words have been suggested, none are actually agreed on by the consensus, so far as I know, other than the word for 'sun'. It's more "it looks like there was at least one l/n word, so there were probably others... hey, this could be an l/n word!".

[Tropylium]

I've never heard of the sound change [n] -> [r] though.

The Paradebeispiel is Tosk Albanian. Also see Rhotacism for a couple other ones. It appears that this is particularly likely syllable-finally.

n → r is also found as a conditional sound change in Ibero-Romance: *mn → *mr (→ mbr, as in hominem, feminam → hombre, hembra).

[Richard W]

Redundant post deleted - I took too long looking for a good /n/ > /r/ and Tropylium beat me to the reply.