Comments: To: lfg at list.stanford.edu

[Mark Johnson]

I'm leaving RXRC literally in a matter of minutes (well timed, Ron!),
so apart from saying goodbye to everyone and wishing all well, I can't
say much here ...

I'm not criticizing LFG as a linguistic theory when I suggest
reformulating in linear logic terms -- rather, I am suggesting that
the formalism of linear logic is a better way of expressing the
linguistic ideas in LFG.  As Bruce Mayo remarks, LFG has a lot to say
about functional valency, etc.; my point is that attribute-value
structures may not be the best way to say it.

In my view LFG is about grammatical relations, the mapping between
c-structure and f-structure, etc.  Attribute-value structures are a
tool for expressing these substantive linguistic ideas, but LFG is not
really about attribute-value structures or any other formal devices.

So it seems reasonable to me that when a new set of formal tools are
developed, we look at them to see if they help us to express our
linguistic ideas more clearly, and (we hope) gain a deeper
understanding of the _linguistic_ systems we are studying.

Mark