Andreas van Cranenburgh 0440949
Core Logic Research report
Martin Stokhof - Hand or hammer? On formal and natural languages in semantics

	Summary of paper:

Stokhof argues that there are two important assumptions regarding logical and
grammatical form: that meanings are available to us independent from their
grammatical form, and that they are determinate before being expressed.  These
assumptions imply that semantics is actually psychologistic, contrary to
Frege's commonly accepted anti-psychologism, because semantics relies on
intuitions about meanings as its data. However, semantics aims to be an
empirical discipline, so its foundations cannot solely rely on conceptual
considerations such as in logic & philosophy. 
In the end the strict dichotomy between natural and formal languages should be
loosened. Both function organically to elucidate the role of the other. The
assumed inseparableness of language and thought should make way for the
realization that external, worldly matters co-determine meaning, which
semantics should embrace.

	Summary of discussion:

The background of this paper is a project to get a better idea about the status
of semantics, something which is neglected in linguistics.

Why does the assumption that meanings are available as intuitions lead to
problems in semantics? This is because these intuitions are what semantics
should explain, as well as being the data for doing so. There can be no crucial
experiment; when two people disagree about intuitions there is no way out.
Instead of relying solely on intuitions semantics should make use of corpora
and statistics, to calibrate such intuitions against other data.

Why does leaving the idea of an intimate relation between language and thought
help solve the problem of circularity? If language is primarily for expressing
thoughts then the aforementioned problem of intuitions occurs. Instead one
should also consider external aspects such as in language acquisition.

Formal semantics cannot make predictions because it doesn't have a meaning
except for the one we give to it. It is not a model but only a description.

Word count: 295