p. 106
The distinction between introducing clause type semantics by lexical
rule or in a phrasal construction has nothing to do with
compositionality. In its simplest formulation (sufficient for the case
at hand), a semantics is compositional if for each syntactic mode of
composition, the semantics of the phrase is derived as a function of
the semantics of the semantics of the daughters. That is the case for
both analysis considered here: The fact, that a function is "contributed"
by the construction has no influence on this as long as it is systematically
contributed for a particular syntactic mode of composition, which is
absolutely the case.
Yes, but this is what construction grammarians mean by non-compositionallity.
To do: clarify.
The same kind of composition, deemed non-compositional here, is
already found in Montague's seminal fragment, deemed the ultimate
example of a compositional grammar fragment in linguistics. To
rephrase: What kind of semantic function composes the semantics of the
daughters (and what exactly it "contributes" to the meaning of the
phrase) has no bearing on calling a rule compositional, there is no
restriction on the nature of those functions in Montague's universal
algebra. Even a non-computable function would be perfectly fine, for
that matter. I would therefore strongly urge to not confuse the
distinction under discussion with matters of
compositionality. Compositionality in the sense of formal semantics
has nothing to do with it, the merits and disadvantages of the two
analyses have to be discussed and decided upon on independent grounds.
Yes, but this is what construction grammarians mean by non-compositionallity.
To do: clarify.
Yes. To do.