When we look at the pair *-mén(i)/*-tén(i), the so-called plural *-n- appears to be only explainable as the product of analogy with early MIE 3pp ending *-éna before the time when *ta “that” (> PIE *to-) was employed to extend the 3rd person singular and plural endings. This termination may once have spread from the third person plural active to the rest of the plural terminations at an early date. QAR predicts accent on *e and the etymology of these endings is transparent indicating that *-mén(i) and *-tén(i) date to at least the Mid IE period before Syncope had yet to take effect.
Now, if we know that *-mén(i) and *-tén(i) are quite ancient, it follows that the accentuation of *-més and *-té can be explained as analogy with older *-mén and *-tén. However, the lack of *-i in primary *-més and *-té still begs an answer. Logically, whatever the source of these unextended endings, they must have once had no need for the indicative *-i. This may indicate a particular usage outside of the primary conjugation. I believe that a possible reason for this is that these latter pair of endings were taken directly from the independent oblique plural pronouns of the time: *mes and *te. MIE enclitic *mas regularly becomes *n̥s via Syncope, and was then later extended analogically as *nos by the time of PIE proper. MIE 2pp oblique *te however (*tei in the nominative case) was replaced in the meantime by an inanimate noun *yáuas “(the) group”[1] (> early Late IE *yaus), thereby obscuring the ultimate source of later 2pp ending *-té.
If this is all correct, it's then probable to me that these alternative endings were first coined as early as the late Mid IE period and that dialectal replacement of *-méni and *-téni by *i-less, pronoun-derived alternatives *-més and *-té began to spread during the Late IE period.
Thus I think we now have a sensible solution to the reconstruction of the Old IE objective endings preceding the agglutination of “indicative” postclitic demonstrative *əi (> PIE *-i):
| singular | plural | |
| 1st person | *-əm | *-mənə |
| 2nd person | *-əs | *-tənə |
| 3rd person | *-ə | *-ənə |
Furthermore, we may possibly reconstruct both the singular and plural independent pronoun forms for the first and second persons with greater depth:
| singular | plural | |
| 1st person | *məi (nom.) | *wəi (nom.) |
| *mə (enc.) | *məs (enc./obl.) | |
| *mə́nə (obl.) | ||
| 2nd person | *tau (nom.) | *təi (nom.) |
| *tʷə (enc.) | *tə (enc./obl.) | |
| *tə́nə (obl.) |
And now everything in the 3000 years prior to PIE is explained...
... Or is it?! Alas, my work is never done. Happy Halloween, everyone!
NOTES
[1] The basic root *yeu- "to bind, join together" is acknowledged in Mallory/Adams, The Oxford Introduction to Proto-Indo-European and the Proto-Indo-European World (2006), p.522 (see link). It is also deemed the underlying root of extended *yeu-g- with identical meaning which is the source of inanimate thematic *yugóm "a yoke" as expounded upon by Szemerényi, Introduction to Indo-European Linguistics (1999), p.272, fn.10 (see link).
Hey Glen!
ReplyDeleteI had an idea about why both the 3sg and 3pl active endings gained a -t "extension". The idea is that it came from the middle conjugation, which (another hypothesis) used oblique pronominal forms. As we know the oblique stem of the 3rd-person pronoun was *to-, this seems to make some sense. Over time, the endings of the middle came to be reanalyzed as personal ending plus middle ending (*-t-or/*-t-o(i)), hence spurring the import of the *-t into the active conjugation.
Rob: "The idea is that it came from the middle conjugation, which (another hypothesis) used oblique pronominal forms. As we know the oblique stem of the 3rd-person pronoun was *to-, this seems to make some sense."
ReplyDeleteOkay, but then we have to wonder what possessed early IE speakers A) to start using *to- only in the middle voice, and B) to model the active voice (which would be the default voice) on a more marked voice.
However, my idea works much better because it's intertwined with a few other nifty facts. For example, if we ponder on the suppletive declension of *so-/*to- "that", we notice that *so never receives case endings and is only ever used in the animate nominative while *to- is used everwhere else (i.e. for the inanimate and all other cases other than nominative for all genders). This tells me that originally *so was just an undeclined general deictic signifying "the", while *to- was originally used for all genders and cases, opposing the proximal demonstrative *ḱo- "this".
The significance of that revelation is that it explains why then that this *-t extension is likewise used for all genders of the third person singular and plural. The middle 3ps *-tor is rather modeled on the active *mi-endings. The middle secondary endings without the characteristic *-r are a post-IE innovation and not attested in the Anatolian branch. Note Jay Jasanoff in Hittite and the Indo-European Verb (2005), page 4 (see link) who reconstructs the earliest middle endings as extensions of the perfect system. Thus *-or would be the earlier 3ps middle ending (i.e. perfect 3ps *-e + middle marker *-(o)r).
Glen: "Okay, but then we have to wonder what possessed early IE speakers A) to start using *to- only in the middle voice, and B) to model the active voice (which would be the default voice) on a more marked voice."
ReplyDeleteA) I think this depends on whether the middle voice was marked inflectionally via oblique pronominal case-forms. If so, *to- was the logical option for third person.
B) This is a question of morphophonological reanalysis.
However, what bothers me is that this approach does not resolve the 3pl active and middle endings, which both have the *-t element appended to the *-n element.
Because of this, I have a question: do you think the 3pl endings have an etymological connection with the active participle ending, or do you think the similarity is a coincidence?
Rob: "A) I think this depends on whether the middle voice was marked inflectionally via oblique pronominal case-forms."
ReplyDeleteWell, it's not. I've already pointed you to Jasanoff's views (again, please see link) but I'm not sure what your views are based on other than hopeful wishing. You're hanging onto *-to whilst ignoring the fact that this is historically derived from the primary ending *-tor and that the vocalism *o was introduced via this final *-r, the characteristic ending of the mediopassive. As we can all see, neither the set *-mo(r/i)/*-so(r/i)/*-to(r/i) nor the set *-xo(r/i)/*-txo(r/i)/*-o(r/i) look much like your oblique pronouns *me, *twe and *to-. It's time to face the music.
Rob: "B) This is a question of morphophonological reanalysis."
That's just vague linguistic mumbo jumbo. Be more specific.
Rob: "[...] do you think the 3pl endings have an etymological connection with the active participle ending, or do you think the similarity is a coincidence?
Coincidence, since there is no plausible semantic connection possible between a third person plural and a participial ending. The *-t- component of the active participle *-ont- is no doubt the same as the simple *t placed at the end of some action nouns.
D'oh!
ReplyDeleteI completely forgot about the t-less Vedic forms, such as śáye "he lies" < *kéyoi. So I retract the theory that I advanced here.
Rob: "I completely forgot about the t-less Vedic forms, such as śáye "he lies" < *kéyoi. So I retract the theory that I advanced here."
ReplyDeleteNo problem. Just to add clarity, I think that the 3ps middle ending *-oi would have originally been *-or at the PIE stage (ie. the stage including Anatolian) since the marker *-i spread from the active present to the middle by analogy.
Pushing further into Pre-IE, we should reconstruct *-ər prior to Schwa Diffusion as I've defined it. I doubt that the mediopassive originally had a distinction between primary and secondary until the dissolution of PIE. It is the voiced *r that caused the preceding schwa to eventually lengthen slightly and merge with *o, the later characteristic vocalism of the middle endings. This vocalism was only later expanded upon by analogy, producing middle secondary endings *-mo, *-so, *-to and the like but these endings are purely analogical in origin and are not ancient.