On 2/22/06, Jörg Rhiemeier <[log in to unmask]> wrote:

> So far, it isn't worked out.  It would of course be nice if the number of
> arguments could be predicted from the morphology.  Perhaps triliteral
> predicates have one argument, quadriliteral ones two, etc.  The example
> sentence I give contains several unary predicates which are all triliteral,
> and one binary predicate which is quadriliteral.

If this is the case, it'd be a shame not to use preposition-to-verb
"incorporation" for the production of applicatives, etc.  A popular
theory these days, from Mark Baker, is that at some level of
representation, applicatives are formed by incorporating prepositions
into the verb.  (Not necessarily the phonological realization of the
preposition itself, but things such as its case-assigning features.)

You could have a handful of uniliteral "prepositions" -- they need not
show up independently if you don't want them to -- that, by
incorporation into an existing root, add an additional argument with
the role their own argument would have had.

So you could have a predicate rbp "jump", and into that incorporate
"over", which is realized by prefixing "k-", giving us the new
predicate krbp "jump over".  (Which is already there in your example.)
 Since we need a predicate "jump over", it would be nice if it could
be related in some way to a predicate for "jump", and both of those to
"jump into", etc.  (Of course, any scheme like this where we derive
new predicates from old is going to be complicated by the
self-segregation rules.  Maybe it should be saved for X-2.)

-- Pat