Type identity

[pollard at ling.ohio-state.edu]

Hi everyone,

I believe the term "type" is being used in two ways. One of the senses
(sometimes called sorts instead, with species being the most specific),
are exemplified by the type _verb_, whose appropriate features are
(say) VFORM, AUX, and INV with appropriate value types (sorts) _vform_,
_bool_, and _bool_ respectively. On the other hand we also speak
of the type of a (totally well-typed, etc.) feature structure to mean
either the proper class of feature structures structurally isomorphic
to it, or to a canonical representative of that class. Surely RSRL is
expressive enough to define the relations that hold between two objects
just in case (1) they have the same species, and (2) they are structurally
isomorphic.


Carl