10 Jun 2008

The early Indo-European case system and definiteness

A simple structure can be seen in the Proto-Indo-European (PIE) case system between animate and inanimate nouns and pronouns:


animateinanimate
nominative*-szero/*-d
accusative*-mzero/*-d


We can see first of all a syncretism between the nominative and accusative cases in the inanimate declension which might be of some interest to people who have nightmares at night about object agreement and grammatical case like I do. While it's popular to derive the animate nominative from the genitive singular *-ós by way of ergative voodoo, I try to buck the trend in order to take advantage of a simpler explanation of the above pattern. It sometimes strikes me that these "ergative Pre-IE proponents" are simply intoxicated by the mere exoticness (or should we say apparent exoticness) of ergativity, seduced by a fashion that will some day pass (hopefully). In the global scheme of things, ergativity isn't exotic or rare; it's natural and common! So get used to it!

Anyways, back to PIE, the opposition between the animate subjects in *-s and inanimate pronouns in *-d can be quite satisfyingly compared to a similar opposition between the demonstrative stems *so-, likewise used strictly for animate subjects, and *to- used for inanimate subjects as well as for cases other than the nominative for either gender.

The fact that both *so and the nominative singular *-s are used only for animate gender is too coincidental to pass up. A deictic origin of this case ending seems painfully obvious to me and it surely is the simplest solution available by far. It however would then suggest that the nominative ending was originally optional for nominative subjects, being used more specifically to mark the definite subject as opposed to an originally endingless indefinite one. This solution works quite well considering that the pronominal inanimate ending *-d can likewise be sourced to the deictic *to- in somewhat symmetrical fashion. Adding to this, we should realize that the Indo-European accusative *-m is technically only the definite accusative case form since indefinite objects are often given other case forms (such as genitive, ablative, partitive, etc.) in many languages around the world.

If we factor in definiteness into the Pre-IE declensional system, we get the following structure that will hopefully inspire and enlighten. This is what I theorize for the Mid IE case system that preceded the PIE stage:


animate
(definite)
animate
(indefinite)
inanimate
nominative*-sazerozero
accusative*-(a)m*-átazero


As you can see, I propose that the PIE declensional system originally specified definiteness for only animate subjects and animate direct objects. For all other cases (genitive, locative, etc.) and for all inanimate nouns and pronouns, definiteness was not conveyed by the case system at all. This then may explain the later pattern in the PIE system and explains how the nominative came to be marked when the tendency in languages is for nominatives to be unmarked. By this solution, I'm also suggesting that the case system was governed by an underlying animacy hierarchy[1] of definite animate > indefinite animate > inanimate.


NOTES
[1] For related information, read Woolford, Animacy effects on Object Agreement (1999), University of Massachussetts (see pdf).

6 comments:

  1. It seems fairly obvious, that animacy was not the feature that defined gender in PIE. It was rather the capability of being an agens (agentivity? agentiveness? Agensfähigkeit). Mere animacy would not support the dichotomy of genera in Anatolian or any other IE language.
    Other than that, you may well reconstruct the core system as you did, though there is a fair share of arbitrariness in your conjectures (mostly in what is related to defeiniteness).

    ReplyDelete
  2. Interesting stuff!

    But, it is not immediately apparent to me why you say the *-d ending is the accusative of the indefinite animate.

    Is this due to the reasoning: We'd like to see an unmarked nominative and a marked accusative thus at least the indefinite nominative should be unmarked?

    I think I just answered my question. But please do confirm.

    the *t > *d shift is similar to the voicing of *s in that same environment, correct? (Also giving a nice expected *o later on)

    Just to nag though:

    You also bring back the 3sg ending *-t to the stem *ta- right. So how come the 3sg ending is *-t/*-e-t and not *-d/*-o-d?

    My gut tells me this probably has something to do when this ending was suffixed to the verb, as opposed to the time that the same thing was done to the noun.

    Also:

    Variant forms like we see with *-bʰi/*-bʰis could be a remnant forms of case endings with both a definite and indefinite marking.

    Ah there's a lot of stuff going on with this *s showing up in all sorts of places, I'll be blogging about it soon :D

    ReplyDelete
  3. Alexey Fuchs: "It seems fairly obvious, that animacy was not the feature that defined gender in PIE."

    And it also seems fairly obvious that I can't rationally be held responsible for the terms "animate" and "inanimate" which are widespread throughout PIE literature. Your point is correct, but your complaint is moot here because the nature of animacy is not the topic at hand. You may call PIE's two genders any names you wish, but most use the terms "animate" and "inanimate" (as you can read here and elsewhere).

    Alexey Fuchs: "...though there is a fair share of arbitrariness in your conjectures (mostly in what is related to defeiniteness)."

    Without the courtesy of being specific about your criticism, there's simply no way for me to address it. Could you by chance elaborate?

    ReplyDelete
  4. Phoenix: "But, it is not immediately apparent to me why you say the *-d ending is the accusative of the indefinite animate."

    This is incorrect. I show MIE *-áta in that field which becomes PIE *-ód, the ablative ending. In Finnish, for example, definite objects are marked with the accusative while indefinite objects are declined in the partitive case. Since PIE lacks a partitive case, it seems to me that the ablative or genitive would be the next best thing.

    Phoenix: "the *t > *d shift is similar to the voicing of *s in that same environment, correct?

    Yep.

    Phoenix: "You also bring back the 3sg ending *-t to the stem *ta- right. So how come the 3sg ending is *-t/*-e-t and not *-d/*-o-d?"

    I'm forced to appeal to analogy between 3ps presentive *-ti and word-final *-t used for pasts and aorists, whereas the inanimate *-d never occurs word-internally.

    Phoenix: "Variant forms like we see with *-bʰi/*-bʰis could be a remnant forms of case endings with both a definite and indefinite marking."

    I don't think so. One of things I notice in the structure of the Pre-IE case system is that if definiteness was a dimension of declension, it was only important in subjects and direct objects. This only makes sense since indirect objects are outside the immediate focus of discourse and so their associated details are less of a concern in comparison to those of the subject and direct object.

    Beware too however that *-s is sometimes the remnant of plural *-es, as in the accusative plural *-ns and *treis "3".

    ReplyDelete
  5. In Finnish, for example, definite objects are marked with the accusative while indefinite objects are declined in the partitive case.

    Except the usual analysis is that the distinction involves what's called the "telicity" of the action, that is, how complete the action on the object is. Definitness is independantly marked by pronouns (and is a recent development anyway).
    Löysin (ne) rahat "I found the money"
    Sain siitä hyvät rahat "I got some good money out of it"
    Söin tänään sitä keittoa "I had some of the soup today"
    Söin tänään keittoa "I had soup today"
    Just a note...

    ReplyDelete
  6. Tropylium: "Except the usual analysis is that the distinction involves what's called the "telicity" of the action, that is, how complete the action on the object is."

    Since I'm not a Finnish speaker, I'll allow you that possibility, however people do associate the partitive with indefiniteness (see for example Finnish partitive case as a determiner suffix (2006) [pdf] by Anna Asbury: "Finnish partitive case shows up on subject and object nouns, alternating with nominative and accusative respectively, where the interpretation involves indefiniteness or negation (Karlsson 1999)."). Consider also that there's a French parallel of the Finnish partitive in the phrase Je mange de la soupe "I eat some soup" where the preposition de, normally used to mean "from", is here used for indefiniteness. By contrast, Je mange la soupe means "I eat the soup."

    ReplyDelete