paradigms

[Martha McGinnis]

Dan,

>I am still bothered by a lack of understanding on my part as to how DM
>would handle non-compositional meaning in periphrastic morphology, when
>that meaning corresponds exactly to what would be expected for
>individual  paradigm cells. That is, the basic idea is that the
>meanings of the phrases are not composed bottom-up, but that their
>literal meanings are overridden  to meet the 'needs' of the paradigm.

Two possibilities come to mind: (1) the periphrastic alternative has
the same underlying syntactic structure as the synthetic alternative,
but is realized morphologically differently (cf. comparative more vs.
-er), and the non-compositionality is only apparent, arising because
of underspecification or even homophony; (2) the periphrastic
alternative has a different underlying syntactic structure from the
synthetic alternative, and this structure is co-opted to fill a
perceived "gap" in the synthetic system.  I don't know enough about
the facts you're talking about to hazard a guess as to which of these
two alternatives is more plausible; you could explore both options
and see what you find.

>Prima facie, this type of phenomenon would seem to offer strong support
>for a word and paradigm type of morphology, against a syntactic-tree
>approach.

Since paraphrases aren't words, why would a word-and-paradigm model
be particularly suited to accounting for such a phenomenon?  I'm
skeptical about the traditional claim that non-compositional meaning
indicates that a phrase has been reanalyzed as a word, since there's
evidence that idioms retain syntactically (though not
Encyclopedically) determined aspects of compositional meaning.

Such a phenomenon *might* be taken to suggest that morphology is not
purely interpretive, but influences the syntax in some way.  So far
I've been unconvinced by specific claims of this kind.

-Martha
--
mcginnis at ucalgary.ca