The Great Proto-Indo-European Thread's Sequel

[Howl]

The thematic adjective in Common IE aligns with a neuter in a very strange way. It uses the accusative -om when referencing a neuter noun as subject or agent. For example take this Latin sentence:

Animal magnum templum sacrum destruxit.

Why would Common IE use an accusative marker for an agent? That makes no sense to me. And it indicates to me that the unmarked case of the neuter was originally not a direct case. But it is possible that the unmarked case of the neuter was an absolutive case, as it is in Hittite.

[WeepingElf]

According to my personal hypothesis, Proto-Indo-Uralic was an active-stative language, and PIE may have been the same as late as PIE1. There were two classes of nouns: animate and inanimate, and two classes of verbs: active and inactive. Active verbs were either transitive or intransitive, while inactive verbs were always intransitive.

An active verb required an animate subject, while inactive verbs allowed inanimate subjects and transitive verbs allowed inanimate objects. When an animate NP functioned as a subject of an inactive verb or as an object of a transitive verb - i.e., a function where inanimate nouns were admitted - it was marked with a suffix *-m, which was thus something like an "animacy-neutralizing" suffix. The nature of this suffix was more or less derivational, which explains while in the PIE accusative plural, the number marker follows the case marker rather than preceding it, as one would expect from an agglutinating language as PIU, PU and PIE1 would have been. There probably were no case markers otherwise in Proto-Indo-Uralic, only postpositions. (Other than the accusative *-m descending from this marker in both families, case markers do not seem to match up well between IE and Uralic.)

Another morphological difference between active and inactive verbs were the personal endings. Each of the two classes had its own set of endings. Active verbs used a set related to the personal pronouns, which is reflected in the active non-perfect endings in PIE3, the mi-conjugation in Hittite and the "normal" endings in Uralic. This may be called the "m-set" after the consonant *m characteristic of the 1st person. Inactive verbs used a different set, perhaps stemming from demonstratives of some sort, which is ancestral to the PIE3 perfect and middle endings, the Hittite hi-conjugation, and the indefinite/stative endings in "Southeast Uralic" (Hungarian and Selkup), This may be called the "q-set" after the consonant *q (> PIE *h₂, PU *k) in the 1st person.

Transitive verbs may have had a bipersonal conjugation, with a q-set ending referencing the object followed by an m-set ending referencing the subject. This is typical of active-stative language, though I have to admit that there is little evidence of this. The PIE thematic vowel may once have been a transitivity marker which in turn descended from a 3rd person singular object marker of the q-set. In Uralic, only Mordvin has a bipersonal conjugation, but this is conventionally considered an innovation. The "Northeast Uralic" (Khanty, Mansi and Northern Samoyed) languages have paradigms which mark the number, though not the person, of the object in the transitive verbs; very similar paradigms are found in Eskimo-Aleut languages, which may point at a Proto-MItian age of those paradigms.

The active-stative alignment of PIE1 shifted to a split-ergative alignment in PIE2 (this may have been at the same stage when the sigmatic nominative arose as I wrote here recently), in which nouns of the animate class had a nominative in *-s and an accusative in *-m, while nouns of the inanimate class had a zero-marked absolutive (but one marked with *m if thematic). The Hittite ergative probably is an Anatolian innovation, but I seem to remember reading somewhere that several ancient IE languages avoided neuter transitive subjects by passivizing the clause - alas, I don't remember where, can anyone help me? From this emerged the familiar accusative alignment of PIE3. Uralic may have shifted along a more direct path.

[Tropylium]

Another morphological difference between active and inactive verbs were the personal endings. Each of the two classes had its own set of endings. Active verbs used a set related to the personal pronouns, which is reflected in the active non-perfect endings in PIE3, the mi-conjugation in Hittite and the "normal" endings in Uralic. This may be called the "m-set" after the consonant *m characteristic of the 1st person. Inactive verbs used a different set, perhaps stemming from demonstratives of some sort, which is ancestral to the PIE3 perfect and middle endings, the Hittite hi-conjugation, and the indefinite/stative endings in "Southeast Uralic" (Hungarian and Selkup), This may be called the "q-set" after the consonant *q (> PIE *h₂, PU *k) in the 1st person.

Some rain on your parade: I re-checked Mikola's Studien zur Geschichte der samojedischen Sprachen, and the 1PS indefinite ending is actually *-ŋ in Proto-Selkup, not **-k. (Final nasals generally become stops in Northern Selkup.)

Retention of *-k in Hungarian would be also weird, for that matter; there are no precedents for retention of final stops otherwise. (The closest precedent is ideg 'nerve, bowstring', among whose cognates Finnic *jändek 'sinew, bowstring' suggests original *jäntək, but some other cognates like Mansi *jäntəŋ suggest instead original *-ŋ, which would be indeed expected to give Hung. g.)

[Howl]

According to my personal hypothesis, Proto-Indo-Uralic was an active-stative language.

I don't know of any trace of an active/stative system in Uralic, which is one of the reasons why I don't think Proto-Indo-Uralic would have had one.

There probably were no case markers otherwise in Proto-Indo-Uralic, only postpositions. (Other than the accusative *-m descending from this marker in both families, case markers do not seem to match up well between IE and Uralic.)

I have more potential matches between the Uralic and Indo-European cases:

- PU absolutive ø ~ PIE absolutive ø
- PU accusative m ~ PIE accusative, illative -m
- PU genitive (ergative?) -n ~ PIE oblique -n; ergative -nt (with t from a pronominal origin?)
- PU locative -s ~ PIE locative plural -su, genetive -s ?
- PU ablative -ta ~ PIE ablative -t
- PU lative -k ~ PIE instrumental -h₁ ?

Some rain on your parade: I re-checked Mikola's Studien zur Geschichte der samojedischen Sprachen, and the 1PS indefinite ending is actually *-ŋ in Proto-Selkup, not **-k. (Final nasals generally become stops in Northern Selkup.)

Retention of *-k in Hungarian would be also weird, for that matter; there are no precedents for retention of final stops otherwise. (The closest precedent is ideg 'nerve, bowstring', among whose cognates Finnic *jändek 'sinew, bowstring' suggests original *jäntək, but some other cognates like Mansi *jäntəŋ suggest instead original *-ŋ, which would be indeed expected to give Hung. g.)

And yet, the 1st person subjective -k and plural -k of Hungarian must come from somewhere. And that somewhere is not *ŋ, since it regularly produces g in Hungarian, not k. What might be happening here is that Hungarian doesn't retain the medial -k- in nouns and verbs, but does retain the final -k in inflections. Also, the Samoyed 1st person *-ŋ could be from an earlier **-mx ~ *-mh₂

[Tropylium]

the 1st person subjective -k and plural -k of Hungarian must come from somewhere. And that somewhere is not *ŋ, since it regularly produces g in Hungarian, not k. What might be happening here is that Hungarian doesn't retain the medial -k- in nouns and verbs, but does retain the final -k in inflections. Also, the Samoyed 1st person *-ŋ could be from an earlier **-mx ~ *-mh₂

Sure, though the main problem is that we then have no supporting evidence for an original 1PS marker *-k anymore (reaching over to PIE is surely much too prone to false positives).

Even in more general, no original word-final single-C suffixes seem to survive in Hungarian at all; the accusative *-m is lost (and replaced by a new marker -t), genitive *-n is lost (for good), 2PS *-t is lost (and replaced by *-Ntə > d taken from the possessive suffic system), even 1PS -m probably gets routed thru the possessive suffix inventory (the Hungarian definite/-ik-verb endings are mostly identical to the possessive suffixes). This makes me think it is a bad idea to ground any new *-C suffixes primarily in Hungarian evidence (and all such cases could be also reconstructed as *-CV just as well).

[WeepingElf]

Another morphological difference between active and inactive verbs were the personal endings. Each of the two classes had its own set of endings. Active verbs used a set related to the personal pronouns, which is reflected in the active non-perfect endings in PIE3, the mi-conjugation in Hittite and the "normal" endings in Uralic. This may be called the "m-set" after the consonant *m characteristic of the 1st person. Inactive verbs used a different set, perhaps stemming from demonstratives of some sort, which is ancestral to the PIE3 perfect and middle endings, the Hittite hi-conjugation, and the indefinite/stative endings in "Southeast Uralic" (Hungarian and Selkup), This may be called the "q-set" after the consonant *q (> PIE *h₂, PU *k) in the 1st person.

Some rain on your parade: I re-checked Mikola's Studien zur Geschichte der samojedischen Sprachen, and the 1PS indefinite ending is actually *-ŋ in Proto-Selkup, not **-k. (Final nasals generally become stops in Northern Selkup.)

Retention of *-k in Hungarian would be also weird, for that matter; there are no precedents for retention of final stops otherwise. (The closest precedent is ideg 'nerve, bowstring', among whose cognates Finnic *jändek 'sinew, bowstring' suggests original *jäntək, but some other cognates like Mansi *jäntəŋ suggest instead original *-ŋ, which would be indeed expected to give Hung. g.)

OK, so much on the "Indo-Uralic q-set". It was problematic anyway, as there seems to be no way to match the 2sg. forms PIE *-th₂e, Hung -sz and Selkup -ntï. And the active-stative alignment may be an innovation in early PIE after the split from Uralic, if it isn't a mirage at all. The Mordvin bipersonal paradigm also doesn't seem to be a "living fossil" of Indo-Uralic; I tried to reduce it to my hypothetical paradigm and failed. It is probably an innovation.

[WeepingElf]

I am now also growing increasingly skeptical of Seefloth's Paradigm. The paradigms compared by Seefloth may just be parallel developments of Northern Samoyedic and Eskimo - perhaps built from cognate Mitian morphemes, but not as such reconstructible for Proto-Mitian.

[Howl]

Sure, though the main problem is that we then have no supporting evidence for an original 1PS marker *-k anymore (reaching over to PIE is surely much too prone to false positives).

Even in more general, no original word-final single-C suffixes seem to survive in Hungarian at all; the accusative *-m is lost (and replaced by a new marker -t), genitive *-n is lost (for good), 2PS *-t is lost (and replaced by *-Ntə > d taken from the possessive suffic system), even 1PS -m probably gets routed thru the possessive suffix inventory (the Hungarian definite/-ik-verb endings are mostly identical to the possessive suffixes). This makes me think it is a bad idea to ground any new *-C suffixes primarily in Hungarian evidence (and all such cases could be also reconstructed as *-CV just as well).

I'm not ready to throw ALL single consonant Hungarian suffixes into the garbage bin of 'unexplained innovations of Hungarian' just because Hungarian lost so many old single-consonant suffixes and innovated so many new ones. Also, there's the Hungarian plural marker -k which can be compared to the neuter plural marker -h2 from PIE. And this comparison does have supporting evidence in the oblique plural -k in the personal pronouns of Mordvinic, especially Erzya (eg. mińek 'us'; tiŋk 'you').

[KathTheDragon]

While the comparison with PIE is attractive, it's probably a coincidence as the neuter plural is transparently an offshoot of the collective, which is itself a development of abstract derivatives.

[WeepingElf]

As we have no Indo-Uralic sound correspondences yet, we can't tell whether the "Southeast Uralic" indefinite conjugations are in any way related to the IE "h2-conjugation", but is it out of the question that the Hungarian and Selkup forms are cognate to each other, even if the PU antecedent would be something else than just *-k? Surely, those endings come from something - but from what?

[Tropylium]

I'm not ready to throw ALL single consonant Hungarian suffixes into the garbage bin of 'unexplained innovations of Hungarian' just because Hungarian lost so many old single-consonant suffixes and innovated so many new ones.

I don't do so either; I only suggest that whatever they come from, it should be something with at least a full syllable's weight to it. Hungarian is pretty innovative morphologically, but not omni-innovative, and there does seem to be a clear bias towards the loss of final single consonants.

(Side note: we're probably ripe for a fork into a Great Indo-Uralic Thread about six posts ago.)

[mèþru]

Good, I don't like Uralic and Eskimo-Aleut in my Hittite reconstructions.

[WeepingElf]

I am thinking about the Late PIE ("PIE3") plural cases. It seems as if Early PIE ("PIE2" in my terminology, the ancestor of PIE3 and Anatolian) may have been a language where only the nominative and the accusative case distinguished number, perhaps from the loss of phonetically feeble allomorphs of the PIE1 number markers (somewhat like Uraiic *-j-) before oblique case endings. PIE3 would have remedied this by drawing other forms into services as plural cases.

This is fairly transparent with the genitive. Both *-os and *-om are found in Hittite, but there, their functions were not singular and plural, but specific and non-specific, it seems (I currently do not remember where I read that, though; could be one of Kloekhorst's papers) [EDIT: It's this one.]. So hassuwas meant 'of (a specific) king (or kings)' (e. g. the King of the Hittites), while hassuwan meant 'of a king/of kings (in general), kingly'. Only in PIE3, this opposition was transformed into one of genitive singular and genitive plural.

Similar things may have happened in other cases as well, and in some instances, the resulting plural case forms may be as late as the beginning disintegration of PIE3, as with the dative-ablative plural forms which vary between *-bhos, *-bhyos and *-mos.

[KathTheDragon]

On the dative endings, see here.

[WeepingElf]

An interesting article. Thank you.

[KathTheDragon]

No problem.

On the more general hypothesis that the oblique endings may have been originally unspecified for number, I'm... admittedly really unsure, since I don't know much about the diachronic progression of the developments involved.

[WeepingElf]

Well, the PIE3 plural case endings are a conundrum. They clearly did not originate from plural marker + case marker (there is, for instance, no way from something like *-es-ei to *-bhyos), so something else must have happened. But what? For the genitive, Kloekhorst has found, as I have written, a workable solution, so it seems plausible to assume that something similar was in play with the other plural cases. Apparently, in PIE2 there were true cases which, with the exception of the nominative and accusative, did not distinguish number (probably due to the phonological erosion of phonetically weak number markers before the case endings); and there were case-like derivative suffixes which, as derivative suffixes usually do, did not distinguish number either and had less specific meanings than the true cases. Does that make sense?

I don't know yet how to make sense of Jassanoff's paper in such a framework, though; I have to reread it a few times, it seems.

[KathTheDragon]

Jasanoff's article fits in pretty easily, in that it's just clarifying the original early PIE forms.

However, I'm skeptical of the base claim, that the Anatolian indifference to number should be projected back to PIE. Even before considering evidence, it smacks of the trendy new position to automatically assume that Anatolian is conservative when there's nothing to clearly show that it isn't, which is quite clearly what Kloekhorst rather ardently believes in. So, let's look at what we've got.

Let's assume that we can indeed reconstruct *-os and *-om as Proto-Anatolian genitives that are indifferent to number and rather distinguish specificity, since I don't really care to see if there are any holes to poke at this level of the argument. To these we compare the late-PIE genitive singular and genitive plural, *-s and *-om (leaving aside the question of the exact reconstruction of this ending). At first glance it seems quite reasonable that we can assume that an original specificity contrast was repurposed as a number contrast to fit these endings into a paradigm that otherwise wholly distinguished number. But this argument works the other way round too, since Anatolian presumably had quite a number of cases that failed to distinguish number, if we're allowed to project Hittite back, and indeed, Hittite eliminated -an in favour of -as in continuation of this pattern, so I don't see why we can't suppose that Anatolian, after innovating some cases that were indifferent to number, applied pressure on the genitive to similarly give up number. The first step was repurposing the contrast, leading to the reconstructed situation, but Hittite later took it a step further and completely eliminated the ending -an.

[Howl]

Jasanoff's article fits in pretty easily, in that it's just clarifying the original early PIE forms.

However, I'm skeptical of the base claim, that the Anatolian indifference to number should be projected back to PIE.

Unpacking morphemes is always a bit speculative. I could for example say that the -bhi/-mi case forms probably came from an earlier pattern of accusative -m or oblique -n + postposition bhi + case ending. And this postposition -bhi is probably the same root as the one that gave the Germanic preposition 'by'' (often compared to h2epi with "Verner's law", but I am skeptic of that.). But there is no evidence that could prove or falsify this.

When it comes to the plural there is something strange going on in PIE. The plural in PIE does not have a simple pattern that agglutinates a case marker with a plural marker. It is more like every plural case ending was formed in a different way. And when we see PIE doing something strange, it is only natural to look at the first subfamily that split off (Anatolian) for a clue to what happened.

[WeepingElf]

Jasanoff's article fits in pretty easily, in that it's just clarifying the original early PIE forms.

However, I'm skeptical of the base claim, that the Anatolian indifference to number should be projected back to PIE. Even before considering evidence, it smacks of the trendy new position to automatically assume that Anatolian is conservative when there's nothing to clearly show that it isn't, which is quite clearly what Kloekhorst rather ardently believes in.

Fair. There is no a priori reason to assume that whenever Hittite and Late PIE disagree, Late PIE must have innovated. There probably also were innovations in Anatolian. However, many of the disagreements involve Late PIE having some category - the feminine gender, some of the case forms, the three aspectually different past tenses - which Anatolian lacks and shows no traces of. Of course, there are plenty of non-Anatolian IE languages which have lost something, too, but usually the lost category leaves traces behind, such as the Latin words duo and ambo showing an idiosyncratic declension which can be explained as a residue of the PIE dual.

It would help if we had an outgroup - a language or language family more diastantly related to Anatolian and Late PIE - to compare things against, but we don't, though Uralic is IMHO (and also in the opinion of some scholars) a likely candidate. For instance, the fact that Uralic has a dual, with a marker that could be cognate to the Late PIE one, makes me think that Early PIE also had a dual, which Anatolian lost.

So, let's look at what we've got.

Let's assume that we can indeed reconstruct *-os and *-om as Proto-Anatolian genitives that are indifferent to number and rather distinguish specificity, since I don't really care to see if there are any holes to poke at this level of the argument. To these we compare the late-PIE genitive singular and genitive plural, *-s and *-om (leaving aside the question of the exact reconstruction of this ending). At first glance it seems quite reasonable that we can assume that an original specificity contrast was repurposed as a number contrast to fit these endings into a paradigm that otherwise wholly distinguished number. But this argument works the other way round too, since Anatolian presumably had quite a number of cases that failed to distinguish number, if we're allowed to project Hittite back, and indeed, Hittite eliminated -an in favour of -as in continuation of this pattern, so I don't see why we can't suppose that Anatolian, after innovating some cases that were indifferent to number, applied pressure on the genitive to similarly give up number. The first step was repurposing the contrast, leading to the reconstructed situation, but Hittite later took it a step further and completely eliminated the ending -an.

I can't say that your argumentation is faulty; it is certainly a possibility to be considered. Also, Hittite was later than Late PIE, so it had more time to evolve since both parted ways. Hence, one would expect Late PIE to be more conservative than Hittite - but we also know that some languages are more conservative than other, even earlier ones, and these "stragglers" are often outliers, while the centre is more innovative. And of course, Hittite is an outlier, and Late PIE "the centre" in the early IE world. (But Tocharian was an outlier, too, and was highly innovative, at any rate more so than, say, Lithuanian, which is closer to the centre and a thousand years later.) I have also seen the idea that Hittite innovated radically in a short period of time under the influence of the Hattic substratum (but AFAIK, so little is known about Hattic that one cannot really say much on that). This is a tough nut to crack, and I don't expect to be able to solve a riddle on which dozens of academic scholars have been gnarling for decades.