prefixing verbal template and OV

[Guillaume Jacques]

Dear Andrej,



> I would be interested to know why exactly you believe that the prefixing
> character or Athabaskan is counter to expectations, given the V-final word
> order. I have some guesses why you say that, but please clarify. For
> example, Indo-Europeanists apply special efforts to explain why IE has
> suffixal personal inflection, going back to subject pronouns, in spite of
> the SV order in the known IE languages.
>

I refer here to the debate concerning cross-category harmony (see the
references in my article), basically in this particular case the notion
that OV word order and suffixes are connected because both are
modifier-head; NB: the idea behind this is that (at least some of the)
affixes are heads. To quote Hawkins (1988): "The affixal head of a word is
or-dered on the same side of its subcategorized modifier(s) as P isordered
relative to NP within PP, and as V is ordered relative to adirect object
NP".

This idea is this very much present especially in generative circles, and I
believe deserves to be examined seriously. I had to kinds of reactions; a
fieldwork linguist/typologist told me that "it is so obvious that this
principle is wrong that it is not interesting to argue against it", while a
generative linguist said that the data in my article "show that Rgyalrong
used to be SVO".

Here, we must distinguish two things. There are prefixes whose placement in
the verbal template is the same as the corresponding free word in the
sentence. Clearly, it is not a mystery why postpositions were incomporated
(like bi- in bi-hoosh'aah "I learn") or why person markers are prefixes.
These are only marginally interesting - it is still an interesting fact
that in most verb final languages, at least in Eurasia, person markers are
grammaticalized as suffixes rather than prefixes.

What is more interesting is how complex predicates are grammaticalized
(since in this case, arguably, we originate from a structure where both
elements were verbs, independent head by themselves). In strict verb-final
languages, as far as I know, it is generally the case that all verb
complements are preverbal (actually here Dixon's distinction between
complement and complementation strategy is useful). I haven't studied
Navajo far enough to be sure that this is always the case too in
Athabaskan, but it seems that it is the case too. Now, if you have a
construction with an auxiliary verb such as V Aux, you expect that it
grammaticalizes as a suffix, exactly following the order in the original
sentence. The kind of phenomena that interests me is when auxiliaries (or
simply "what would correspond to the main verb in European languages") are
grammaticalized as prefixes in these languages, because this precisely goes
counter to the Head Ordering principle. Of course, the idea is that they
are grammaticalized in this case from parataxis or serial verb
constructions (thus coming from a non-embedded structure). In these cases,
there was a choice between the harmonic embedded construction and the
disharmonic embedded one, and the diharmonic construction was favoured.

This scenario has been proposed for Athabaskan by Givon (2000), who
suggests that the "perfective prefixes" ghʉ-, nʉ- and sʉ- in Tolowa (the
last two ones corresponding to the ni- and the perfective si- conjugation
in Navajo, I presume) originate from verbs (in serial verb constructions).
I would like to know what Athabaskianists think of that, to what extent
this is a workable hypothesis or whether it is too speculative. Also, are
there other cases of prefixes in the verbal template of Athabaskan
languages that could be shown to have grammaticalized from auxiliaries?

The conclusion is that the head ordering principle is not an active
constraint on diachronic change, and that the fact that most SOV languages
are mainly suffixing could be a historical accident (or due to typological
convergence in areas such as Eurasia and the Andes), not the result of a
universal principle.


> Also, modern Athabaskan have variable verb stem finals that is mostly
> reconstructed as going back to suffixes and, actually, series of suffixes.
> Most of the Athabaskan prefixes are relatively recent, and there is a
> reconstructable stage at which prefixes (conjunct) were combined with a
> number of suffixes.
>

 Actually, the focus here is not so much how the Ursprache looked like in
its earliest recoverable stage as to how the verbal template came to be the
way it is. It is clear that many prefixes in the verbal template of
Athabaskan are recent, especially the first position which is basically
incorporated postposition with their possessive prefix.
Closer to the stem, the person prefixes are also transparent, though they
do present strange behaviour (like the 2sg ni- becoming high tone í- when
preceded by a conjunct prefix - I someone has an explanation for that, I am
interested to hear).

Sorry for this long message - I still feel it is not completely clear
enough,


Guillaume

-- 
Guillaume Jacques
CNRS (CRLAO) - INALCO
http://cnrs.academia.edu/GuillaumeJacques
http://himalco.hypotheses.org/
http://panchr.hypotheses.org/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/athapbasckan-l/attachments/20130916/3ec55930/attachment.htm>