26 Aug 2009

Is an active-stative or subjective-objective system more appropriate for earliest Common Proto-IE

The neat thing about keeping a blog for any length of time is that one can see how old one's ideas are and how one can reuse past ideas to eventually crystallize a thought into something more detailed and polished. Almost two years ago, I had talked about stages of Pre-IE with subjective-objective contrast which was in turn inspired by Allan Bomhard's words on Pre-IE in his idea-packed book, Indo-European and the Nostratic Hypothesis (1996). While I will never be a fan of current Nostratic reconstruction for its lack of scholarly rigour and poor methodology, the idea of a subjective-objective contrast ultimately linking PIE to Uralic stuck in my mind for... well... more than two years!!!

It's sadly only now that I realize how much of an impact these ideas can have towards finally understanding PIE grammar completely. COMPLETELY! I'm going to now assert the following premise for the sake of discussion: PIE *itself* still retained a subjective-objective contrast and this is what lies behind the suppletive *mi- and *h₂e conjugations. To sum up, "objective" is a conjugational form that marks verbs as "object-focussed" while "subjective" implies "subject-focus", as in modern Uralic languages like Hungarian or Nenets. For example, any action effected through one's own body parts can be automatically considered "subjective" such as "to know" (via one's brain), "to see" (via one's eyes) or "to feel" (via one's physical skin or by one's internal emotions). Other subjective verbs might simply be subject-focussed arbitrarily such as "I hunt" (naturally implying that animals of an unspecified sort are being hunted, and hence that the action is unavoidably transitive despite a lack of expressed object).

It looks to me that what I've previously mused about the origin of the 1ps pronoun is yet one example of evidence in favour of a subjective-objective contrast in PIE itself. If *h₁ég-o-h₂ (or perhaps more accurately *h₁ég-o-h₂e) is originally a subjunctive verb form meaning "(as for) my being here" and if Jay Jasanoff is correct in deriving all thematic verbs from the subjunctive, then the unique subjunctive 1ps ending *-o-h₂e, which later becomes the "thematic 1ps" familiar to avid Hellenists and Indo-Iranianists and which so self-evidently borrows at some point from the *h₂e-conjugation (but curiously only in this 1ps), is indeed a relic of this subjective-objective contrast. Afterall, one's thoughts (and one's potential expressed so specifically by the 1ps subjunctive, eg. "I would go/I will go") simply imply subjectivity, making the replacement of an earlier Pre-IE 1ps subjunctive *-o-mi, the otherwise expected form, with this new subjective ending all the more natural.

Again too, the use of the specific *h₂e-conjugation as a basis for the middle endings (*-h₂ór < MIE *-xá-ra, *-t(h₂)ór < *-tá-ra, *-ór < *-á-ra, etc.) is simultaneously also explained best through this paradigm being used to express the subjective rather than stative or perfective. The result of the action isn't important, neither is any implied state, whether arrived at by an action or not. What's rather most important to the *h₂e-conjugation is that the action and/or subject is the true focus and not the object which remains comparatively less defined or even entirely undefined (ie. less important to topicality). Considering that the intransitive verb, and the subjective forms that may spring from it, can in turn develop into an aspectual perfective (aka punctive, punctual) is a delicious bonus too since it now helps to link this hypothetical "Indo-Anatolian" system with the later system of durative-aorist-perfective in the later "Core PIE" from which non-Anatolian and non-Tocharian dialects sprang such as Indo-Iranian and Hellenic (Greek).

22 comments:

  1. Hey Glen!

    Maybe it's just me, but it seems like maybe the IE "perfect" verb form is really just a specialized kind of middle. I say this because IE apparently had plenty of stative verbs which were conjugated the same way as active verbs. See e.g. stative verbs derived with the suffix *-eH1.

    On another note, I've long considered the 1sg active thematic ending *-o: to be somehow related to the perfect and/or middle endings. But I'm still perplexed as to why (assuming it is related) it would replace *-mi without the same thing happening to the other "primary" endings. Any ideas there?

    ReplyDelete
  2. I found the following article to be very tasty - you might agree: http://www.artsci.wustl.edu/%7Ecwconrad/docs/NewObsAncGrkVc.pdf

    ReplyDelete
  3. Rob: "See e.g. stative verbs derived with the suffix *-eH1."

    Actually, it's precisely your example that we may better explain from my new perspective.

    Let's start again more concisely: I'm suggesting here to throw away the "active/stative" model and to replace the word "active" with "objective", the word "stative" with "subjective".

    Once we realize the connection on the one hand between objective verbs and durative aspect and the connection on the other hand between subjective verbs and punctual aspect, we can certainly explain your example of "stative-flavoured actives", so to speak, since they are to be reconceived as "stative-flavoured *objective verbs*" instead. Naturally, durative statives will gravitate to the objective conjugation (mi-class) while resultatives will gravitate to the subjective (h2e-class) due to their natural punctual aspect. In this revised model, neither the mi-conjugation nor the h2e-conjugation are actives or statives par excellence, a usage that isn't attested in Anatolian anyway.

    ReplyDelete
  4. Can you explain what you think the connection is between objective verbs and durative aspect, on the one hand, and between subjective verbs and punctual aspect, on the other?

    On another note, I think that article I linked to actually holds the key to the Hittite -mi/-hi distinction. Have you looked at it yet?

    ReplyDelete
  5. "Core PIE". That's a very convenient name! I usually call it "Classical PIE", but that's easily misunderstood.
    I think I'll adopt your terminology for "Post-Anatolian-and-Tocharian-PIE" It's sensible!

    "...modern Uralic languages like Hungarian or Nenets" I appreciate that you've included label "modern", perhaps to skirt around this question, but, well... do you think that PIE and Uralic go back to a common ancestor, PIE having been given an injection of North-West Caucasian words? Or is PIE a way ward North-West Caucasian language with simplified phonetics? (If you haven't guessed, I''ve been reading Colarusso again...)

    While I realize this may be getting WAY off topic (my apologies- this is why I'm no good at essays!), the more I read this blog, the more I'm convinced that you should collect some of your findings and send them to periodicals, like the Journal for Indo-European Studies, or Indogermanische Forschungen, or a some periodical that deals with (at least among other things) Etruscan (have you come across such a journal or periodical?). Even if it's just your corrections of Etruscan authors myriad screw-ups, it would be a welcome read in those publications!

    ReplyDelete
  6. Rob: "Can you explain what you think the connection is between objective verbs and durative aspect, on the one hand, and between subjective verbs and punctual aspect, on the other?"

    This question made me think long and hard. After verifying the logic behind my statement, an intriguing paradox is uncovered: intransitives/subjectives are directly linked with *non-punctual* aspect and transitives/objectives with *punctual* (seemingly contradictory to what I'm proposing for PIE).

    I have ideas but I'm working on it. Bear with me. I don't think all is lost with this idea just yet since a former subjective/objective contrast seems to be most readily observable in stative verbs: subjective *wóidh2e 'I know' versus objective *h1ésmi 'I am'. Verbs of action require a more involved explanation it seems.

    "On another note, I think that article I linked to actually holds the key to the Hittite -mi/-hi distinction. Have you looked at it yet?"

    The article is a little long to scan through and I'm not sure by what you're most intrigued but the article admits to the subjective usage of the middle. I'll still look at it some more when I get a chance and ponder.

    ReplyDelete
  7. Seadog Driftwood: '[...] do you think that PIE and Uralic go back to a common ancestor, PIE having been given an injection of North-West Caucasian words?"

    I'd be utterly shocked if PIE and Uralic weren't related at some point in the past. Too many coincidences. Allan Bomhard suggested NWC influence on early PIE in Indo-European and the Nostratic Hypothesis (1996) and I rather like the idea since it would only be expected if PIE ultimately came into the Pontic region from Central Asia by 7000 BCE.

    "Or is PIE a way ward North-West Caucasian language with simplified phonetics?

    Only in an alternative universe where unicorns exist.

    "Even if it's just your corrections of Etruscan authors myriad screw-ups, it would be a welcome read in those publications!"

    Thank you very much, I'm very flattered and happy it's interesting for others and not just me ;) For now, my life only permits me the commitment of a casual blog but luckily knowledge is infinite so there will always *always* be something to explore, even five hundred years from now. Plenty of time, really.

    ReplyDelete
  8. Hey Glen,

    What intrigued me about that article was the author's explanation of the semantics behind the Ancient Greek middle and its differences from the active. Basically, the middle is the marked form (vis-a-vis the active), specifying that the grammatical subject was somehow affected himself by the action.

    In this light, it's very ironic that you brought up *wóidh2e vs. *h1ésmi. If my suspicions are correct, then the proper semantics for the former are "I have seen (it) for myself". As for the latter, how can one be or exist for himself? :P

    ReplyDelete
  9. Rob: "In this light, it's very ironic that you brought up *wóidh2e vs. *h1ésmi. If my suspicions are correct, then the proper semantics for the former are "I have seen (it) for myself". As for the latter, how can one be or exist for himself?"

    "Ironic"? But we seem to agree with the same thing. "Knowing" is subjective (the subject is affected by the state and also agent of it) while "being" is clearly non-subjective. Yet since "know" and "be" are equally stative, despite being from BOTH the mi- and h2e-conjugations, it seems senseless to use the opposing terms like "active/stative" to describe the two conjugations without encouraging deep confusion about their true functions.

    However, a "reflexive" can be seen as merely a subset of subjective. So we might seperate the notion of true reflexives where the subject and patient are perfectly identical (eg. "I soiled myself") from a seeming reflexive where the patient is only a subset of the subject (eg. "I know (through my brain/heart)" and "I feel (through my own skin)"). The former might once have been conveyed by the middle (derived from the subjective) and the latter by the simple subjective.

    ReplyDelete
  10. Wait a minute, I guess in the case of "knowing" I should have said that the patient-subject ("I") is a SUPERSET of the implicit agent ("brain") in "I know (through my brain)". Methinks I got mixed up. Oh my, speaking of brains, mine is being fried right now from all this grammar talk. Lol!

    ReplyDelete
  11. Although semantically "know" and "be" are both inherently stative, PIE did not treat them the same in a morphological sense. While *h1es- is a stative root, *weid- is not. My point was that the stative semantics of *wóidh2e were comparatively (very) late in contrast to those of h1ésmi. The earlier semantics of the former were decidedly not stative ("I have seen [it] for myself"), but only became so through semantic shift.

    What I find very interesting about our discussion is that we seem to be in the process of disproving the very thing that Sihler (et al.?) put forth - namely that the so-called "perfect" paradigm in PIE was properly a stative paradigm. My copy of Linguistic Semantics states that the perfect aspect (or perhaps "meta-aspect") refers to a past action with present relevance. Wikipedia says something similar (and perhaps even more accurate) - that the perfect aspect is used to denote the consequences of an action.

    Regardless of whether you associate the perfect aspect with relevance or consequences (or both!), it seems that the PIE middle inflection was the means to highlight either or both of these. After all, both relevance and consequence can be described as a kind of affectedness, right?

    ReplyDelete
  12. Rob: "My point was that the stative semantics of *wóidh2e were comparatively (very) late in contrast to those of h1ésmi."

    How are you sure of that? Rather it looks like *weid- is a genuinely ancient stative considering Tocharian ime 'consciousness, awareness, thought' from *widmén- 'knowledge'.

    Note also that if both these statives, *h₁es- 'be' and *weid- 'know', happen to have been borrowed during the Neolithic from Proto-Semitic statives *yiθ 'there is' and *{wdʕ} 'to know' respectively, then the idea of *weid- being once non-stative as you say would be unacceptable.

    Given that, a subjective/objective contrast then seems to have been best preserved in the statives as I mentioned above, and then this means that states and actions can freely occur in both *mi- and *h₂e-conjugations.

    ReplyDelete
  13. Of course, with science, we can't be sure of anything. :P

    Cited there with Tocharian ime is Greek idmo:n "skillful", both presumably from *widmén-. I can't help but notice the similarity between this noun ending and the middle participle ending in *-méno-. An inherently stative meaning for *weid- wouldn't make much sense if it can form a middle participle.

    On the other hand, "awareness" could be considered semantically equivalent to "looking at something and being (somehow) affected by doing so", given that vision is our primary sense. So I think the Tocharian word is a closer semantic derivation from the root than the Greek word, if the root means "see" (or more accurately "look [at]").

    Yet another thing - why would *weid- form a characterized stative in *wid-éh1- if it's already inherently stative? Furthermore, why would éidon < *wéidom be an aorist (i.e. a non-stative) in Greek?

    ReplyDelete
  14. Oops, I forgot to address your second and third paragraphs! Here goes...

    With all due respect, I remain skeptical of your claim that *h1es- and *weid- were borrowed from a Semitic language. As your third paragraph depends on this hypothesis, I can't really address it.

    On the other hand, it does seem like we're not quite aligned when it comes to subjective vs. objective (your terms) and middle vs. active (mine). For example, as far as I can tell, a verbal root with stative force would never distinguish between active and middle. That's because there's no actual activity for the agent to be affected by! But it seems that what you mean by subjective vs. objective has little (if anything) to do with affectedness on the part of the agent. Please correct me if you think I'm mistaken here.

    ReplyDelete
  15. Rob: "Yet another thing - why would *weid- form a characterized stative in *wid-éh1- if it's already inherently stative?"

    We find *wid-eh1- in Latin, Greek, Balto-Slavic and Germanic (Penney/Davies, Indo-European Perspectives: Studies in Honour of Anna Morpurgo Davies (2004), p.54), dialects which just happen to be contiguous and in a localized region within the larger PIE dialect area. You assume that this stative construct is ancient but the form is apparently not found throughout IE, which makes sense if *weid- had began as a stative before a semantic shift took place towards an active meaning "see". After this change, a stative marker would become far less redundant. Remember that the *h₂e-set of endings are now reemployed for the tense/aspect of perfect and this means that *wóide can no longer mean what it used to mean ("I know" (subjective stative) → "I have known" (perfect)).

    "An inherently stative meaning for *weid- wouldn't make much sense if it can form a middle participle."

    Proposal: a semantic shift of stative "know/be aware" (*wóid-e) to active "see" (*wid-ór 'he sees (by himself)').

    This middle suggested by Tocharian then expands to derived thematic presents in Core IE (*wéid-e-ti 'he sees') and aorist forms (*wéid-e-t 'he saw') that we know pretty darn well never existed in PIE proper anyway (ie. the subjunctive origin of thematics). Good?

    "For example, as far as I can tell, a verbal root with stative force would never distinguish between active and middle. That's because there's no actual activity for the agent to be affected by!"

    Quite simply, statives like *wóide 'knows' and middles like h₁ḗsor 'sits' are united under the subjective category. The key commonality between the two then is that the patient is indistinct from the subject, despite "know" being stative and "sit" being active.

    ReplyDelete
  16. By the way, for added perspective (or possibly added confusion), let's look at this. The Oxford introduction to Proto-Indo-European and the Proto-Indo-European world (2006), on pages 321 & 322: "The second root, *weid-, indicates 'seeing' or 'knowing as a fact' rather than recognizing a person. It was essentially a perfect, *wóide 'have seen', that was reinterpreted as a present 'know' [...]". Yet this is old-school nonsense and we aren't told clearly what verbal system finally unites the Anatolian dialect with the others. As much as some IEists have tried and tried, the idea that the *h₂e-endings originally signified the perfective as they do in Core IE is surely dead. The perfective can't explain the usage of the Hittite hi-class at all.

    So this means that the interpretation of *wóide here as 'has seen' → 'knows', while semantically plausible in general, is etymologically very unlikely in the specific case of PIE.

    ReplyDelete
  17. I consider the form *wid-éh1- to be ancient due to its morphophonemics, namely the zero-grade ablaut of the root. While Latin, Greek, Balto-Slavic and Germanic may be in a (more or less) contiguous area today, that does not mean they were always so. Greek, at least, appears closer to Indo-Iranian than any other branch of IE.

    Again, I don't see how statives can be subject-focused at all (in terms of the article I linked to). Or rather, I don't see how statives can be any more subject-focused than they already are, since they denote a state or condition on the subject. Statives thus stand in opposition to actives, which may affect the subject as well as the object (if there is one).

    Regardless of when you think *weid- had active semantics, to equate it to English "see" is a mistake IMO. "See" has stative semantics, so a better meaning to attribute to *weid- is "look at" or "look upon".

    With that said, what do you think would motivate a semantic shift from "know" to "look upon"? This is a shift from more abstract to less abstract (i.e. more concrete), which seems semantically unlikely to me. The reverse direction seems much more likely.

    From my own point of view, the evidence is secure that *weid- had perfective semantics (Greek eidon "I saw" < PIE *wéid-o-m) - what you would call "aorist" or "punctual". Hence there would have to be a derived imperfective form, which (AFAIK) were not formed by adding the apparently hic-et-nunc particle *-i. It would have to be something like *widyéti or *widskéti.

    More to follow!

    ReplyDelete
  18. Rob: "From my own point of view, the evidence is secure that *weid- had perfective semantics"

    But the perfective semantics are caused by it being sucked into the perfective category. That perfective category however isn't original since it can't explain the hi-class in Anatolian which use the exact same endings. People have tried, people have failed. So any evidence of perfective semantics in *woid-e are an illusion because it can't possibly have been originally perfective.

    ReplyDelete
  19. As for the idea that "know" could come to mean "see", check out stranger things in Vanhove, From polysemy to semantic change: Towards a typology of lexical semantic associations (2008), p.295: "In Russian, the link between the lexemes meaning 'taste' and 'know' is inverted: the derivative by prefixation of the verb vedat' 'knowing', developed the meaning 'taste'. It is otvedat' 'know','taste,sample' [...] whose systems of values are similar to the Latin verb degustare."

    ReplyDelete
  20. By "perfective" I mean the same thing as Wikipedia -- viewing the action expressed by the verb root or stem as a whole. This has nothing to do with the PIE verbal category traditionally called "the perfect conjugation".

    While I understand that (for some odd reason) it's traditional PIE parlance to distinguish between root-present/imperfects and root-aorists, the consensus linguistic terms would be "lexical imperfectives" and "lexical perfectives", respectively.

    Now I think the evidence is clear that the Hittite mi/hi distinction has nothing to do with verbal aspect. So you're absolutely right that the Hittite hi-conjugation is not restricted to perfective verbs. But I never said otherwise. :P It seems clear to me that the mi/hi distinction is one of voice, as you yourself have stated. The article I linked to way up above presents an even stronger case for this, I think. Basically it seems like Hittite verbs are "frozen" in terms of voice. Which voice a particular verb became "frozen" in must have been arbitrary, through customary usage or preference.

    ReplyDelete
  21. Yes, in hindsight I think I now see to what you were referring in your above-cited article (Carl Conrad, New Observations on Voice in the Ancient Greek Verb (19 Nov 2002) [pdf]), particularly where it says "What the teacher of ancient Greek must understand and the student must learn is that the fundamental
    polarity in the Greek voice system is not active-passive but active-middle."

    Similarly, we appear to be in accord that a like opposition existed in the earliest common PIE itself. As per my latest article (Paleoglot: Interesting quirks of a PIE subjective-objective model [13 Sep 2009]), "middle voice" would be summed up as a verb form expressing subjectivity, and subsequently it seems to me that rather than using the terminology of voice and a non-descriptive term like "middle" (middle of what?), we can simply call the contrast what it is: objective/subjective (a more descriptive terminology used in describing languages like Nenets).

    ReplyDelete
  22. Ooops, just something more to add...

    An excellent primer on Nenets grammar provided by Tapani Salminen explains that Nenets is, like my conception of PIE, a language opposing separate conjugations for subjective and objective. Unlike PIE however, it seems that the reflexive is separate from both subjective and objective paradigms whereas in PIE it seems that the "subjective progressive" would have taken on that role.

    ReplyDelete