Bipartite stems

Scott DeLancey delancey at DARKWING.UOREGON.EDU
Mon Oct 6 21:52:08 UTC 2003

On Fri, 3 Oct 2003, Bernhard Waelchli wrote:

> I would like to ask you how you define bipartite stems.

As far as I can tell, I seem to be the one who unleashed this terminology,
and I'm coming to think that was a mistake.  I owe the term, and the
concept, to William Jacobsen, who coined it to describe the facts of
Washo, a language unrelated but geographically relatively close to
Klamath.  To me, the notion, and the term, are really only a means to try
and communicate some of the peculiarities of these languages in terms of a
particular Americanist descriptive tradition.  North American languages
being generally polysynthetic, there has, since the early days, been
a constant concern with sorting out the morphemes of their complex verbs
into stems and affixes.  The problem in Washo and Klamath (and apparently
some of their neighbors) is that there is a substantial set of verbs
composed of two distinguishable parts, where it is very difficult to
clearly establish that one part is more stem-like than the other.  Hence
Jacobsen's solution, which is to NOT arbitrarily pick one or the other
to call the stem.

> The phenomenon as such is undisputed.

If you mean by that that there is no question that Klamath and Washo
are weird, you're certainly correct.  Whether there is any "phenomenon"
which they share with some significant set of other languages elsewhere
remains an open question.

> My question is about how it should be
> defined in a general manner to make it a cross-linguistically useable
> term so that we can decide for any language X whether it has bipartite
> stems or not (if this is possible at all).

I'm not sure if this is either possible or desirable; for me (and I think
for Jacobsen) the term is simply a descriptive convenience for dealing
with an odd feature of a handful of languages.  Any cross-linguistic,
operationalizable definition would have to be in the realm of
morphological theory, I think.

> I have some problems with the definition of bipartite stems consisting of
> "bound stem elements" (DeLancey 1999: 60). I wonder whether the notion
> of bound can be applied to stem elements, since it is usually
> associated with morphemes in general (including affixes). Thus, defining
> bipartite stems as consisting of at least two bound stem elements
> suggests that we already know beforehand what a stem element is.

Fair enough, but, again, the problem that the term was coined to deal with
is just a descriptive one--is there some element in the Klamath verb that
we can call the stem?  Or, assuming that, by definition, an inflected form
must have a stem, where is it?  J's solution is to call the pairing of
the two the stem.  But I don't want to call the two components "affixes",
since for me an affix is something which attaches to a stem.  So I ended
up calling them both stem categories.

> Couldn't there be a definition which doesn't contain "stem"?

No doubt there could, but it has to contain *something*.  If we're not
going to call the first elements of the Klamath verbs in question
something like "classifying stems", and the second elements something like
"locative-directive stems", then what are we going to call them?
The whole business is all about how to avoid calling them
prefixes and suffixes.

> Moreover, I wonder whether bipartite stems is always about verbal stems
> (are there any bipartite noun stems?)

Well, you can derive nouns from any verb stem.  Other than that, the
answer is, not in Klamath.  In some of the neighboring languages, e.g.
Northern Paiute (Uto-Aztecan), there are nouns which contain an
instrumental prefix attached to some other element, but it's not clear
that we have the same descriptive problem in Northern Paiute--the
"instrumental prefixes" aren't necessarily very instrumental, but it
doesn't do any great violence to the facts of the language to call them
prefixes.  This is quite different from Klamath, where if you call the
equivalent category "prefixes", you immediately have a great deal of
very un-affix-like behavior to try and explain away.

The Adyghe examples you present certainly are *semantically* reminiscent
of Klamath and Washo, but for me the question of whether they represent
*bipartite stems*--as opposed to just morphologically complex motion
verbs, which occur all around the world--is a question about the details
of the morphological status of the two categories involved.  When you say:

> In Adyghe, the final position is more stem-like, the initial one more prefix
> -like.

That says to me that they are not bipartite stems, in the sense in which
I use the term--unless the evidence is ambiguous enough that another
linguist could come along and claim that really, if you look at all the
facts, the initial elements are just as stem-like as the finals.

> Nevertheless, some "stems" like the illative and elative verbs,
> require a prefix and are bound in this respect. Are these bipartite
> stems?

No, not as long as it's still possible to decide that one element is the
stem and the other a prefix.  And I'd say the same about your Car-Nicobarese
examples.  The occasional lexicalized bound stem is a familiar enough
phenomenon.  What forces a novel analysis in Klamath and Washo is the
existence of a whole *category*, indeed of *two* whole categories, of
such bound stems--so that their behavior is not the peculiarity of a
handful of idiosyncratic forms, but an important fact about the grammar
of verbs.

> What Adyghe and Nicobarese share with Klamath is that one of the two
> morphemes encodes some information about the ground (the difference

It seems to me a bad idea to be trying to define the "bipartite stem"
notion in any kind of semantic terms.  There are hundreds, if not
thousands, of languages in which motion verbs have to separately
specify path and manner, or path and ground.  But in almost all cases
that I know of, it's a fairly straightforward matter to determine that
one piece of a motion verb belongs to a lexical set of verb stems, and
the other to a morphological set of verbal affixes.

Scott DeLancey
Department of Linguistics
1290 University of Oregon
Eugene, OR 97403-1290, USA

delancey at

More information about the Lingtyp mailing list