Reductive Paraphrase again
Hi All,
I found an interesting paper that argues against the
sufficiency of the NSM primitive classes for explaining
natural language semantics through reductive paraphrase.
Its at
http://ling.ucsd.edu/~barker/Research/barker-nsm.pdf
He gives some examples that seem a bit stretched. The
list starts with names of people, places and things (which
are simply identifiers for logical individuals, and should be
part of the logic of composition/decomposition, IMHO).
Since Wierzbicka et al never address the questions of
composition/decomposition, I suppose they opened
themselves up for this kind of argument. But it seems
that W et al never really cared much about how the
compositions would be formally structured.
Backing into Levin's EVCA examples, it appears that
the composition rules do require an extensible syntax,
with lots of terminals (the primitive verbs) and nonterminals
(the composition verbs) and that each composition has
the possibility of a number of overloadings. So the paper's
point can be argued by ignoring the complexity of the
composition/decomposition structures.
Then he addresses modal issues (without calling them
that) of truth or belief, contradictory descriptions of
situations, and so on.
And finally, he ignores Pierce's Thirdness, or the subjectivity
of belief, perception and therefore of meaning.
He does identify some important prepositions (nonprime)
as being in need of semantic interpretation, and that part
seems convincing.
But it still is an interesting exercise in identifying some
issues with NSM's (nonexplicit) paraphrasing mechanics that
should be considered in detail before following that
approach.
Is anyone else more or less convinced by this argument?
Does anyone think the argument implies that effective
reductive paraphrase semantics is impossible or impractical?
Comments welcomed,
Rich Cooper