Type identity

[Frederik Fouvry]

| > does anyone know of any examples where type-identity (or inequality), as
| > opposed to token-identity, plays a role in syntax?
|
| I am not sure about syntax, but Tilman Hoehle has a paper (in the
| "Slavic in HPSG" CSLI volume) in which he crucially relies on
| type-identity in his account of phonology in HPSG.
|
| We (the HPSG group in Warsaw) have an analysis of coordination relying
| on type identity (for checking 'sameness' of conjuncts in certain
| kinds of coordinate structures), but this analysis is rather
| preliminary and -- hence -- probably not a good argument for
| type-identity in syntax.

If I recall correctly, some work on ellipsis from a few years
back used type-identity, for similar reasons that Adam mentions:
parts of the structure had to have the same type, but not all of
it is the same, so token identity could not be used.
Unfortunately, I cannot recall any references.  Supposing that
there is a type-identity checking mechanism in place, this line
of thinking does not supply any arguments in favour of
token-identity though.

--
Frederik Fouvry     Universität des Saarlandes, Computerlinguistik
Gebäude 17/3.02 - Postfach 15 11 50 - D-66041 Saarbrücken, Germany
telephone: +49(0)681 302 4176        facsimile: +49(0)681 302 4700
fouvry at coli.uni-sb.de           http://www.coli.uni-sb.de/~fouvry/