« Scirpture on Schwebeablaut and Consonant gradation | Main | Quick update on Scripture »



Glen Gordon

There's a lot to talk about here but I have just enough time right now to make a small objection regarding conceiving of the Indo-Uralic accusative case ending as a particle terminating in a vowel rather than a bona fide consonant-final ending.

In my theory which I summarized in a pdf, I note that paradigmatic accent shifts are the result of Syncope and a fixed penultimate accent before Syncope took place (see QAR in the pdf). Thus all Indo-European weak case endings originally ended in a vowel but since the accusative case, whether singular or plural, doesn't show an accent shift like these other case endings, QAR shows us that the ending was *always* consonant-final, even in Pre-IE stages.

So considering that Indo-European, through this internal reconstruction, shows a consonant-final ending, together with the Uralic evidence which likewise only shows a consonant-final ending, the conclusion must be that unlike the other "Indo-Uralic" case forms which were the result of agglutinized postpositions *after* the dissolution of "Indo-Uralic", the accusative ending is inherited from some pre-Indo-Uralic stage. In other words, the case system common to both protolanguages contained only the accusative *-m while other cases were endingless with optional accompaniment of postpositions like *di "in" or *si "to".



I still am also inclined to join you in your opinion on the Accent Shift. But I hope  that I'll find a way to disprove Kortlandts theory definitely; Otherwise it's just picking what you think is prettier, which is sort of a personal opinion.

But let's be Devil's Advocate for a second:

If you want to assume an accusative ending *-m though, how will you explain the Greek forms ἄμμε and ὔμμε. Though I must say 2 pronouns found in Homeric Greek might not be the best evidence for a particle *-me, they are pretty odd and would definitely need at least some kind of explanation. And Kortlandt's theory gives an explanation.

The last thing on both Kortlandt's and Gordon's Syncope (Can I call it that? :D) have not yet been said; Hopefully my research will shed some light on it. Who knows.

Glen Gordon

I don't think it's really just a matter of "personal opinion". A theory's comparative explanatory power counts for something and I for one try my darnedest to be thorough. Unless I'm missing something, Kortlandt appears to be just assuming that the accusative was a particle in an Indo-Uralic stage yet neither PIE nor Uralic show such a thing. Of course there's also Proto-Aegean demonstrative accusative in *-n attested by Etruscan cn "this (acc.)" (< *ka-n) and tn "that (acc.)" (< *ta-n). All these language families are showing *only* a consonant-final ending.

"If you want to assume an accusative ending *-m though, how will you explain the Greek forms ἄμμε and ὔμμε."

First off, *-m is not an assumption. It's an attested fact in PIE. Rather an accusative in **-me is an ad hoc assumption. The apparent *-me- affix in the 1pp paradigm is found in all the *oblique* cases of those pronouns, not just accusative. So this fact itself invalidates this a priori labeling of it as an "accusative" ending.

It's generally understood that *n̥s "us" is syncopated and related to the 1pp verbal ending *-mes. The verbal ending makes it clear that this plural pronoun originates from the application of the plural ending *-es to the first person singular *me. Yet, if *-me- were the accusative ending as you say, placing a case suffix directly on a plural ending is awkward and then must surely be a late innovation. The plural accusative ending, *-ns, attests to a reverse order of endings to boot (ie. *-ns < *-m [acc.sg.] + *-(e)s [plural]). So something is clearly wrong with this view.

The internal evidence shows that the pronominal case system was originally impoverished in comparison to nominal declension and I've already offered a table of the Old IE pronominal system in Paleoglot:The trouble with the PIE 1st & 2nd person plural endings (3) to illustrate what I mean. Since the etymology of *n̥s is apparent and shows that *n̥sme-, no matter how one slices it, must be a later innovation (since one expects the plural ending to be word-final unless corrupted by later changes), I only recognize two pronominal case forms in the first and second persons plural for Old IE (c.7000 BCE): 1pp *wəi (nom.) vs. *məs (obl.) & 2pp *təi (nom.) vs. *tə (obl.).



[this is good] Clear enough ;-)

The problem with Kortlandt is that sometimes he has these great idea's but his extremely condensed way of writing things; among just a generally bad writing style makes it so that it's really hard to figure out what he means.

Anyway, all these particles and assumptions aside, his points stands on the Indo-Uralic consonant gradation. With your view on Indo-European, Kortlandts Rule III is invalidated. And this is exactly the rule he needs to get the shift of accents that are so typical of Indo-European.

I couldn't agree more that when you see a system in Indo-european where adding a suffix results in a shift of accent towards the back clearly indicates a fixed accent, it doesn't (yet) properly explain the consonant gradation. Especially not in the schwebeablaut-like pairs.

I'm curious to hear what your stance on this gradation is. Because surely the stastical evidence of the relation of accent and gradation found by Lubotsky can't just be a statistical anomaly.

Glen Gordon

I don't get it still. Let me try the following argument now. If we can agree that at some point Pre-IE had a strong stress accent that caused Syncope and subsequent "zerograde" syllables., then surely at this stage, the relationship between a lack of stress accent and zerograding was 100%.

If I'm not mistaken, Lubotsky via Sanskrit evidence is suggesting paradoxically that there is an association between voiced stops and accentuation in PIE. Yet if we already accept Pre-IE Syncope in some form as do so many IEists, then any such association (if real, for the sake of argument) must surely postdate Syncope as a kind of later innovation. Unless you have a really good explanation preferably based on some strong facts, we can be reasonably sure then that such consonant gradation cannot be inherited (let alone existent at all in Proto-Uralic as per the view of most Uralicists).

If and only if consonant gradation had formed in PIE, it can only sensibly arise *after* Syncope and, if anything, caused by an accent shift by way of the tonogenic qualities of phonemic voice itself. That is, voiced phonemes have an inherently low frequency while voiceless phonemes exhibit a high frequency so that voiceless stops could conceivably attract a tonal accent that arose, again, a while *after* Syncope

However I still see no obvious pattern in this mess of roots and derivatives that Lubotsky presents. It looks like analytic overkill - impressive but empty.


[זה מעולה] I too felt it was a stretch to push the description of tonogenetic qualities found in Sanskrit back to Indo-European. But rather than concluding Kortlandt must be wrong I assumed that he was probably forgetting to say something more explicitly.

This is where my scripture comes in. I have found several fairly sound correspondences of roots that alternate between voiceless and voiced aspirates. If one were to believe that these pairs are indeed true alternation, and I for one do, it's going to at least need some kind of confirmation.

You make a strong point on saying the the tonogenesis is rather a feature of the voicing, than that the voicing is a result of the tonal shifts; Though both are possible.

Either way, I won't be withholding the verbal roots I've found that seem to display gradation for much longer. I'll soon discuss it with some of my teachers. And the obviously blog about it as well.

But yeah where you say there must have been Syncope and then accent shift due to voicing, Kortlandt says the exact opposite. Voicing due to accent shift and then Syncope.

If I were to present roots that seem to show some historical consonant gradation your Syncope system would need an explanation for it, while Kortlandt's system explains it right of the bat.

Well, obviously I don't know for sure whether it explains it right of the bat either. I would have to look and see if any paradigmatic shifting Kortlandt expects would yield the alternation as in the forms I found. I'm hoping that it turns out it is impossible to unify with Kortlandt's theory, because then we can safely discard it, and return to QAR.

Withing the system of QAR there would still need to be an explanation why there is an alternation between voiceless and voiced aspirates though. But of course there is no reason for us to think a-priori that this has anything to do with accent. It might just be some bizarre morphological derivation, or some, yet unexplained, phonetic variant.

Glen Gordon

Strictly put: Your pattern is not proven. Hysterokinetic (eg. *ph₂tér-m̥/*ph₂tr-és) and proterokinetic (eg. *séh₁ti-m/*sh₁téi-s) accentual paradigms however are very well substantiated and standard knowledge in IE studies. Between developing a theory to explain an *alleged* pattern and one grounded in strong facts, the choice is simple. QAR unites the hysterokinetic and proterokinetic paradigms and thus forms the basis for my theory.

A reasonable theorist is not guided by what she believes should be fact but by what is fact.


[this is good] Absolutely. 
Like I said, I give preference to QAR because of its strong explanatory power. I will be digging more into these consonant gradations later on. Since that is what I'll write about for my scripture. Soon I'll be presenting the root pairs that I've found on this blog; And you can see for yourself whether you find them convincing enough as evidence.

For now we wait ;-)

Glen Gordon

Alright then, I'll wait. ;-)


Hi, I was recently reading Kortlandt's article on Indo-Uralic consonant gradation and am still completely in the dark.
After reading your blog I understood the two examples he gave in his article: *nepot and *keim, but still don't understand at all his discussion on the i- and u-stem nouns. He gave 2 examples: *bheh2ghu- (arm) and *tenh2u-, according to him, the first one had a moble accentuation and the other one had a fixed accentuation, which I don't understand. He then gives his arguments to support Lubotsky's observation, that is, i/u-stem nouns containing T had oxytone and those containing Dh has barytone. I don't understand this part at all.
I am wondering maybe after all those years you have figured out what he meant in this article and could give me some insight?


Hi Yunran!

Wow... No I don't even understand my own post anymore. Last time I looked at Indo-European seriously was years ago. Let alone Pre-Proto-Indoeuropean.

It's a fascinating topic though, but I'm going to have to actually completely reread the article, which I might do at some point.

I ended up writing a BA Thesis on consonant gradation, however. I think that was after I wrote this blogpost?

I found a fairly large list of pretty clear cases of roots that occur in unaspirated/aspirated pairs. But never really got it to come together as a particularly round story.

Ever now and then I float my results with some people that are still actually working on Indo-European, in the hopes to do something of a collaborative article... but no-one has taken the bait yet :-(


Hi, PhoeniX
Thank you so much for your reply!
I enjoy so much your posts and understand you have switched your research direction. After I read Kortlandt's article, I searched it on google hoping to find some reviews online, and I was lucky enough to find your blog :).
I am so fascinated by this topic, since this provides such a clear evidence of the relation between the two linguistic families. I wish I had the honor and luck to read your BA thesis.

The comments to this entry are closed.