Word order and control?

[jess tauber]

The language with which I work, Yahgan (isolate, Tierra del Fuego) exhibits extensive case marking for nonsubject NP's (Subject pronouns are coreferenced on the verb, with voice markers between). There is also quite variable NP ordering, with full subjects and objects on either side of the verb, and seemingly in no particular order between S and O.

But as I examine more deeply into usage I'm finding that ordering does in fact appear to be motivated, especially by animacy, control, and affectedness relations.

For instance hai skaia ha-ta:gu-de: (1.S 2.O 1-give-pst) 'I gave you' versus hai hat-aki-de: skaia (1.S 1-hit-pst 2.O) 'I hit you'.

In most instances I've seen so far when Obj appears before the verb the NP is less affected, and often the affect is beneficial, and welcome. But when it appears after it is usually more affected, and in malefactive senses (which I assume are usually unwelcome).

Conversely, when a Subj appears AFTER the verb, then the NP appears to have less than usual agentive control over the situation. Some constructions seem to have grammaticalized this ordering pattern, for instance in negative or unrealized situations, or where the speaker is asserting something about the subject.

I am not sure this type of patterning, for objects, extends to all persons, numbers, etc. In Yahgan, for singulars, lowered situational animacy requires the use, in 3rd person, of the case form -ima for higher animates (while lower animates as objects do not require an overt case mark), whereas dative -ikaia only is found affixed to 1st and 2nd person pronoun bases when these are objects. As number increases (duals, plurals) both -ima and -ikaia can be used.

The vast majority of the pronominal object forms with variable ordering seem to be the 1st and 2nd singulars (1.O haia (hakaia in another dialect), 2.O skaia) where there are no alternatives in -ima, though there is no shortage of 3rd person forms with variable ordering.

I've heard more than once that variable word order is used to disambiguate situations where case marking is insufficient to do the trick, even when a case system is extensive. Is this usually true in languages, and is it usual even when case marks are retained on the NP's so ordered, and if so, is reordering aimed at more pragmatic nuancing, as seems to be the situation with Yahgan?

Typologically, Yahgan veers between SOV and SVO orders for the most part, with the latter seemingly used more often for main clauses. One might expect, then, that SVO is the newer one. But then if this is so, does the change of ordering (which also often interacts with voice and TAM), combined with what was laid out above, imply a change in expected or default values for control/animacy or responsibility /affectedness for the entire system?

Much of the grammatical morphology of Yahgan appears to be spanking new. In addition the language makes extensive use of compound or serial verbs. I've been told that extensive case and extensive serialization don't usually go together. It would be interesting to know which system was older- my guess is that it is case, and that this was losing ground to a newer SVO system focusing on verb morphology. In addition, perhaps SVO was just a rest stop on the way to verb-initial ordering. Most of the languages that I've seen with relatively open bipartite systems are verb-initial, and Yahgan is like this, as a specialization of serialization, perhaps?

Anyway, I would love to hear from those in the know about similar ordering patterns in other languages, as well as anyone that might tell me what the diachronic typological motivation could be.

Hmmph- I guess I shoulda stayed with the program..... 

Thanks all,
Jess Tauber
phonosemantics at earthlink.net