[Fair warning to list readers: this one is
theory-ish, and has no pedagogical utility]
Herb,
My experience with serial verb
constructions comes primarily from doing some work on Khmer years ago, and
since Khmer is isolating -- there are some fossilized affixes, but nothing
productive – establishing dependency relations and finiteness in pairs of
verbs can be difficult (I tried using differential modification, but was never sure what the results really meant). I’m thus using a
far looser definition of ‘serial verb’ than I should be (“a
construction involving seriated verbs used as if they’re a single unit”).
That said, I can’t help but wonder whether a (hypothetical) serial verb
construction that develops from a previous finite+dependent nonfinite pair might retain the
morphological markings of dependence on the second element without it actually
being dependent in terms of cognitive processing (and yes, that begs a giant question
of what “dependence” means in cognitive processing).
Now, immediately I want to object to my
own point, based on its empirical problems – I’ve just come up with
a reason to rationalize away any inconvenient counterevidence. It may be
possible to get some support for the idea from psycholinguistic research,
though. At a very, very informal level, I’ve noticed that when I ask
beginning linguistics students to “split” sentences into
constituents, they readily split some verb combinations but not others, and I
can’t help but wonder if their behavior represents psychological reality
(whatever that is) better than some of our models do.
Bill Spruiell
Dept. of English