Re: PLUG: SpecGram Current Issue
From: | Dirk Elzinga <dirk.elzinga@...> |
Date: | Saturday, March 3, 2007, 1:30 |
On 3/2/07, Eric Christopherson <rakko@...> wrote:
> I enjoyed it too, but I was wondering about the distinction between
> suppletion and use of separate lexemes. Are we to believe that
> English <am> and <were> "belong" to the same lexeme even though they
> have distinct roots?
Yes. A lexeme is the set of all inflectionally related forms. If you
plot the verbal categories on a grid with the columns headed singular
and plural, and the rows headed 1st person, 2nd person, 3rd person,
you get a 2x3 grid which you can fill in with verb forms. You can
extend this into three dimensions if you include tense (and there's no
reason not to include it since English verbs distinguish past from
non-past tenses). <am> fills the 1st person singular cell in the
present tense, and <were> fills everything but the 1st and 3rd person
singular in the past tense. Same paradigm, so same lexeme.
Some Uto-Aztecan languages show regular suppletion of verb forms based
on the number of the subject for intransitives or for the number of
the object for transitives; it is thus an ergative pattern. Some
examples from Shoshoni: nukki 'run (sg.subj)' ~ nuraa 'run (pl.subj)';
paikka 'kill (sg.obj)' ~ wase 'kill (pl.obj)' . The form alternations
themselves are unpredictable (else it wouldn't be suppletion), but it
is a regular feature of the language (at least for the several dozen
verbs it applies to).
Verb suppletion was probably a feature of Proto-Uto-Aztecan, though
many (if not most) of the Southern Uto-Aztecan langauges have lost it.
However, in Tepiman verbal number agreement is still ergatively
aligned. In Tohono O'odham, a Tepiman language spoken on the
Arizona/Mexico border, number agreement is marked by initial
reduplication. Again, for intransitive verbs verbal number agrees with
the subject, but for transitives it agrees with the number of the
object: cipkan 'work (sg.subj)' ~ cicpkan 'work (pl.subj)' ; ceposid
'brand (sg.obj)' ~ cecposid 'brand (pl.obj)'.
But that was more than you wanted to know.
Dirk
Replies