Fwd: Fed up

[Arnold M. Zwicky]

[accidentally sent only to sean fitzpatrick]

Begin forwarded message:

> From: Arnold M. Zwicky <zwicky at csli.stanford.edu>
> Date: December 31, 2003 10:53:54 AM PST
> To: "Seán Fitzpatrick" <grendel.jjf at verizon.net>
> Subject: Re: Fed up
>
>
> On Dec 24, 2003, at 8:30 AM, Seán Fitzpatrick wrote:
>
>> << (b) and if we found adjectives that shifted from one marked
>> preposition to a different marked preposition (rather than to the
>> unmarked preposition "of"), then we'd have evidence against the
>> defaulting proposal.>>
>> Would examples be the UK/US variations like "different to" vs.
>> "different from (than)" and "respect to" (as in the Declaration of
>> Independence) vs. "respect for"?
>
> the larger issue is where properties of lexical items -- like which
> preposition an oblique-object -verb goes with -- "come from".  some
> properties of some lexical items (for instance, the semantics and
> phonological content of DOG) are quite clearly just idiosyncratic and
> have to be stipulated in a description of that item, while other
> properties (for instance, the fact that DOG has a /z/ plural) are just
> as clearly predictable, motivated.  sean's question brings up the
> issue of just what sorts of motivation there can be.
>
> default properties are one type of predictable property.
>
> semantics is another source of prediction, as when nouns with a
> particular semantics mostly belong to a particular grammatical gender
> (nouns referring to female human beings in the feminine gender, for
> instance) or declension class.  in the preposition cases above, it
> *might* be that verbs whose meaning invokes a standard of comparison
> can get the preposition "to" appropriate to goals; the question is
> whether this is a currently live relationship for speakers, or just a
> residue of history.  (i haven't examined the question at all.)
> certainly it is plausible to claim that the comparative semantics of
> "different" predicts (comparative) "than" as a possibility.
>
> still other properties can be sources of prediction: grammatical
> gender can predict declension class, and vice versa, for example.
>
> and morphological properties can be sources of prediction, in two
> different ways: (a) the "products" of a particular derivational rule
> can generally have some property, as when german derivative diminutive
> nouns in "-chen" and "-lein" are neuter gender; and (b) properties of
> the derivative item are "inherited" from the source item, which is
> presumably what's going on in the selection of "from" in "different
> from": an english adjective derived from a verb tends to select the
> same preposition as that verb  ("differ from" predicts "different
> from").
>
> all these implicational relationships are defeasible; they're merely
> defaults, and they can conflict with one another.
>
> arnold (zwicky at csli.stanford.edu), compressing a lot of stuff
>