Peninsulas and Probes

[Carl Pollard]

>
Probe: Well, Liz, a 'probe' is a collection of (uninterpretable)
features that 'probe' down a tree structure within some specified local
domain until it finds some element with which the agreement relation
holds, Once this happens, the agreeing element is copied (via MERGE) to
the specifier position of the head that sent the probe. This may not be
a notational variant of SLASH, but it looks very much like it (or, for
that matter, to LFG Functional Uncertainty), and more so if you take
into account that extraction in the MP is now traceless and erstwhile
traces are now exact copies of the whole phrase being 'moved' (there is
not yet a notion of LOCAL features in the MP, but wait... we may soon
hear about Probe-able features [is that word formation indecent enough?]
that get copied). In fact, if I got the whole story correctly (which is
unlikely), in the MP there is no MOVE operation anymore, just AGREEMENT
(of features) and MERGE. So...
>>

Well, Sergio, this may sound a lot like things that were proposed in
LFG and GPSG 22 years ago, but there is an important scientific
difference.  They were not then proposed by Chomsky, and therefore
incapable in principle of being taken REALLY seriously, to say nothing
of being right. Now, from what you say, it sounds like the ideas may
have become right, and if so then surely Chomsky get the credit.
Surely Gerald Gazdar and Ron Kaplan shouldn't get any credit for
having these ideas when they were wrong! What sense would that make?

Remember the old phrase "it's not what you know, it's who you know"?
To that perhaps should be added: "it's not what you know, but who you
are when you know it."

C.