> > Semantic lexemes can be distinguished from grammatical morphemes:
> >
http://www.facstaff.bucknell.edu/rbeard/homepage.html#lexmorph
> >
> =20
> Charles Fillmore, Ronald Langacker, and George Lakoff (among many
> others) would strongly disagree; their work on "construction grammar"
> begins from the hypothesis that the "grammatical vs. semantic"
> distinction is bogus, and that *all* constructions are symbolic --
> e.g. they pair forms with meanings. In the case of what are
> traditionally called "grammatical" morphemes, these meanings tend to
> be fairly abstract and sometimes complex.
grammemes are another category of deictic referring to utterance itself. in=20
other words they are semes specialising in utterance. so i think both charle=
s=20
and you are right (or wrong) ;-)
mathias