Response to Ellen Prince

[George Lakoff]

If this were taking place on cogling, the discussion would be rather different.
Given advances in cognitive semantics, a cognitive grammar (according to
lancker, myself, and others, though not all cognitive linguists) uses no
formal syntax at all.
Rather is contains constructions that are direct pairings between aspects
of  cognitive semantics and their phonological expression in a given
language. The old syntactic categories are semantic categories (in some
cases radial categories). The old fashioned syntactic hierarchical
structure is semantic hierarchical structure, and so on,
as discussed in the various cognitive grammar literature. Since not
everything has been analyzed in these terms yet, one cannot claim thatwe
know this can be done, and some folks are more optimistic than others. But
the question would be how to handle recalcitrant cases. Most people would
agree that a "parse" yields an embodied cognitive semantic characterization
of meaning as an output.

In the evolving field of neural cognitive semantics (coming out of ICSI),
we would have more stringent criteria for a "parser" -- that it be neurally
realistic, done in structured connectionism, obey the hundred step rule, be
performable in real time, be learnable, have an embodied semantics, be able
to deal with blends (a la Fauconnier and Turner) and with metaphor, use
plausible neural binding techniques, be able to deal with garden path
sentences, be able to derive correct contextually appropriate inferences,
and on and on.

The general point, of course, is that what a "parse" is depends on the
field you're in
and on what assumptions you're making about what linguistics is. For this
reason, a parser challenge doesn't make much sense, unless everybody agrees
upon their theoretical assumptions, which doesn't seem to be the case in
the funknet group.

This is not to denigrate anybody's "parsing" efforts given whatever
assumptions about linguistics they happen to like. There are few enough
linguists and few enough people working on serious parsing efforts from
whatever theoretical perspective that we ought to welcome all efforts,
however diverse in theoretical perspective they seem to be.

However, since this is not happening on cogling, I'll yield my two cents
and leave the discussion to core funknetters.

Happy New Year To All!

George