I continue to suggest that predication is an asymmetric relation between two that roles, in fact THE asymmetric theta relation. Any NP type can be a predicate, including definite NPs and proper names; ony quantified NPs are excluded (putting aside second order quantification, a quantifier must bind a variable in an argument position, which a predicate exactly isn't). The proper-name/definiteness/indefiniteness distinction overlays the asymmetrical predication relation to give the variety of sentence types; in particular, predication encompasses both of Higgin's predicational and the specificational structures. There are no equative sentences, strictly speaking, though the illusion that there are must be accounted for.
Binding theory is defined over theta relations, and so asymmetries of binding reduce to the asymmetry of predication.
The semantic content of predication is relative "epistemic priority": in "I consider Bill Sam" we know Bill directly (relative to Sam), whereas we know Sam only under some description, or less directly. The same applies even to generic indefinites, as in "I consider a stoat a weasel", which is compatible only with a situation in which I know relatively directly how to identify stoats, but know weasels only under some (perhaps technical) description. The "normal" order is for the theta role donor to follow the receiver, or for the epistemically prior NP to precede the other one. The overt copula permits a marked order, whose markedness gives rise to focussing effects that show up in extraction patterns.
The overall view is not new, but will be defended in new ways
against recent alternatives, especially Heycock and Kroch (1996).
