Serial verbs in Oceanic

[Joel Bradshaw]

Claire,

I would guess every serializing language has some verbs on the verge of mutation
to auxiliaries, prepositions/postpositions, complementizers, or adverbs, but
semantic bleaching alone is a very problematical criterion for defining a
cutover point. Are English 'do' and 'be' still verbs?

AN languages all along the north coast of the New Guinea mainland (at least on
the PNG side) have a class of manner-causative first verbs (with core meanings
like 'hit', 'hold', 'chop', 'say') in extant or erstwhile SVCs that have
semantics so bleached that they are often hard to distinguish from the inherited
AN causative prefix *pa-. In the SOV languages these verbs have been analyzed as
"classificatory prefixes." The inflectional pattern is S O s-VV-o, where the
lowercase s- and -o show the position of subject and object agreement affixes.
The same classificatory prefixes also function to verbalize nouns, in much the
same way that Japanese -suru and Korean -hada do.

But cognates of those same "classificatory prefixes" (with semantics every bit
as bleached) show up in main (first) verb position in the SVO languages of
Morobe Province (where I did my fieldwork). They even help create denominal
verbals. In the SVO languages, the manner verbs (Vm) have kept their verbal
status, while the result (Vr) verbs in S Vm O Vr serial constructions have
degenerated into a class of uninflected resultatives, yielding SVOR
constructions where the manner V and R correspond to the classificatory V and
second/main V of the SOVV languages. (This is covered in much more detail in my
never-published dissertation, and summarized in an equally obscure paper for the
15th Pacific Science Congress in NZ.) In short, I suspect forms lose their
verbal status for reasons other than bleached semantics.

Inflectional criteria are more clearcut, but still problematical. If
prepositions are never inflected for subject and tense in a particular language,
and verbs always are, then I'm not sure what we gain by creating a new class of
prepositions that are (sort of) inflected for subject and tense. Numbami, for
instance, has both a sometime inflected verb -su 'to descend (upon)' and a very
common preposition su 'onto, on, to', which overlap sometimes in actual usage.
In other cases, some of the inflections on the verbs may well be defective, with
either the subject or tense frozen (not showing "proper" agreement) to some
extent. But it's often hard to alter the context sufficiently to test whether
agreement works as expected. That's especially true of the "ambient" serial
verbs.

Jabêm, for instance, has an all-purpose locative/goal "prepositional" verb
-(n)dêng 'to reach, (be) at' that is always inflected to agree with the tense
("realis/irrealis" or nonfuture/future) of the other verbs in the SVC, but that
only takes '3s' subject prefixes. Time-of-day verbs ('to dawn', 'to dusk') that
function as "ambient" time adverbials behave similarly. It's hard to concoct
ways to test other person and number agreement possibilities on 'to dawn' and
'to dusk'.

I certainly don't have any ready answers (just longwinded ones), but do want to
call attention to a wealth of recalcitrant data on serialization (and
clause-chaining) in Oceania, much of it published in obscure places like Oceanic
Linguistics, Pacific Linguistics monographs, Language and Linguistics in
Melanesia, SIL workpapers, NUSA volumes, and German works of pre-WW2 vintage.

Joel
--
Joel Bradshaw <bradshaw at hawaii.edu>
Journals Manager, University of Hawai'i Press
2840 Kolowalu Street, Honolulu, HI 96822
Tel: 808-956-6790; Fax: 808-988-6052
http://www.uhpress.hawaii.edu/

Claire Bowern wrote:

> With Paul's permission I'd like to add a question to the thread.
>
> (c) What prevents us from analysing (at least in some languages) the type
> (2), *ambient* serials as prepositions? Especially when the "verb" seems to
> be highly bleached, such as in this Titan example:
>
> yo ku angani ala nat _kile kan_
> 1sg fut-1sg support 3pl child fut-"go to" food.
> "I'll support the children _with food_"
>          (where 'go to' is the meaning of the independent lexical verb
> 'le', and (k)ile is a deverbal 'preposition' (or ambient serial) meaning
> usually either 'towards' or 'with, by' (instrumental)
>
> What is the point at which such a 'thing' stops being a verb and starts
> being a (deverbal) preposition?
>
> - Claire
>
> _________________________
> Claire Bowern
>
> Department of Linguistics
> Harvard University
> 305 Boylston Hall
> Cambridge, MA  02138
> fax: 617-496-4447
> ph: 617-547-3521
> http://www.fas.harvard.edu/~bowern/