26 Oct 2008

The trouble with the PIE 1st & 2nd person plural endings

I'm still thinking about a problem that I've never completely solved to full satisfaction yet so it's a good topic to amble through right now. It seems to me that the singular personal endings are easy enough to work backwards into Pre-IE. In the most ancient stages of Pre-IE, there must have surely been two completely different sets of endings, one used for objective/active (i.e. the *mi-set) and the other for subjective/stative (i.e. the *h₂e-set). It's safe to say that the *-t- of the 3ps and 3pp active is merely an eroded form of deictic stem *to- which was attached before the Syncope rule. This ending must have then been fully entrenched in the language by late Mid IE. So far then, this leaves me in Old IE with objective singular *-em, *-es and *-e and subjective singular *-xa, *-ta and *-a. There may also have been a special stative set: *-x, *-t and *-0. The later so-called indicative *-i (or should it be renamed declarative?) must have been agglutinated to the existing objective endings in Mid IE at the time of QAR (Quasi-penultimate Accent Rule).

So what's so problematic? Well, look at how the 1pp and 2pp active endings are reconstructed for later Proto-Indo-European (PIE). The 1pp primary ending is apparently a slight embarrassment of numerous parentheses: *-mé[s/n](i). That is to say, it could be *-mé, *-més, *-mési, *-méni or all of the above for all we know. The secondary ending is supposed to be *-mé, or possibly *-mén, or maybe even *-més. Egad! The 2pp is also idiosyncratic because for some IE dialects, specifically the "internal IE" dialects, *-té must be prescribed for both the primary and secondary conjugation in the parent language (as well as for the 2pp imperative) while in other branches such as Anatolian, primary *-téni and secondary *-tén seem more in order. Despite the madness, we can thank our lucky stars that the athematic primary and secondary 1ps, 2ps, 3ps and 3pp endings are securely reconstructed as *-m(i), *-s(i), *-t(i) and *-ént(i) respectively, but the 1pp and 2pp endings are our bratty problem children.

Now, with that intellectual teaser, perhaps we should start trying to figure out what is going on with these two persons. Which variants of these endings are more archaic? Or are they all the same age? What should we reconstruct for the Old IE stage? What is the significance and origin of this *-n- plural marker in some dialects that replaces the far more productive plural ending *-(e)s? Why isn't the deictic *-i attached to the 1pp and 2pp primary endings in some dialects? And why isn't the plural marker *-(e)s attached to the 2pp as it is in the 1pp?

So many questions, so few answers. But don't worry. Glenny's been thinking very long and hard in the past week about this. I believe I have some solutions that I will share in subsequent blog entries because this is a large topic with lots of grammatical details to juggle. Stay tuned.

(Continue reading The trouble with the PIE 1st & 2nd person plural endings (2).)


  1. First of all: Fun that you have a picture up. It's always nice to put a face to someone's name on the wild west of internet anonymity.

    Might try and make a picture someday too.

    Second: I've been out of the picture for a bit on Indo-European. Currently finally doing an attempt to learn all those damn Kanji, for Japanese, and finally truly push into fluency and full literacy. This, as you can imagine, is quite time consuming. But I'm at 50% of the kanji now!

    This is also the reason I haven't gotten round to reading your latest update on the Semitic Loan file, nor have I gotten round to replying to your posts on my last entry. Haha. But it'll come!

    Now my personal thoughts on this matter for now:

    I like to think of *-més and *-th₁és being the least archaic. Simply because they can be explained by a post-syncope explanation of *-mé/th₁é + -es from the nominal plurals. While the bare form nor the -n allow such an easy 'late' explanation. Also the absence of the *s variant in Anatolian speaks for this.

    What to do with (i) and (n), I don't know though :D

    Also the alternation between primary th₁é and secondary té is not particularly wanted, but it's there. I'm basing this on the Sanskrit opposition -thá versus -tá, which really can't be explained in a different way, as far as I know. But let's hear it ;-)

    So much for my thoughts on the matter.

    I also think that the *-es plural suffix can help explain the weird alternations of érs/ér/r in the perfect 3pp.

  2. Phoenix: "Fun that you have a picture up. It's always nice to put a face to someone's name on the wild west of internet anonymity."

    Yes, I decided to make a stand against internet anonymity. However, I must say, it's a tricky balance between standing behind individual responsibility on the one hand and scaring the children on the other. Hahaha.

    Phoenix: "I like to think of *-més and *-th₁és being the least archaic."

    Hate to burst your bubble, Phoenix, but Burrow has something to say precisely about that in The Sanskrit Language, page 315: "Hittite treats the 2 pl. in the same way (ten/teni) and here again Sanskrit expresses the difference in quite a different way (-ta/tha). In all other languages the distinction does not exist. Apart from lack of support from other languages, the fact that the distinction between t and th is used in the related dual endings for a totally different purpose (2 -thas, 3 -tas) makes it altogether unlikely that the difference between the two forms of suffix was from the beginning connected with the distinction between primary and secondary ending."

    So... it appears that *-té is a more apt reconstruction than *-th₁és.

    Phoenix: "What to do with (i) and (n), I don't know though :D"

    Precisely. That conundrum sours things a little, doesn't it?

  3. Interesting. Beekes still reconstructs the laryngeal in the 2pp.

    The thing is, even if it would not be archaic we are left to wonder why Sanskrit ended up replacing *t with th.

    Assuming an analogy from the dual makes little sense to me. And just discarding the Sanskrit aspirate as an odd irregularity is too easy. Sanskrit doesn't just sprout voiceless aspirates out of nowhere.

  4. May I direct you to Mallory/Adams, The Oxford Introduction to Proto-Indo-European and the Proto-Indo-European World (2006), p.64, table 4.10 where they explicitly reconstruct PIE *-té despite the aspirated Sanskrit reflex -thá? Besides, aspiration can easily be supplied by analogy through the original 2ps perfect *-th₂e (which becomes Sanskrit -tha as well).