7 Reading “is a book” for “B” and “Zachary-wants-it-to-be-true-that . . .” for “Wz “. It is clear that in (1) “wants” is a dyadic predicate, and in (2) is a semantically undetachable fragment of the sentence-operator “Wz’ It is far from obvious how to interpret (even partially) model theoretically the operator “Wz”. We might say that Wz expresses a function θ from persons, to sentences (or the propositions they express), with “wants-to-be-true” semantically primitive. What would be the full conditions on [θ (s) = δ ] (“s” ranging over persons and “⋔” over sentences) is hard to say; but at least it would be necessary to honour the differences between the various optative operators, “hopes-to-be- true”, “wishes-were-true”, etc. For example, wanting differs from wishing and hoping by virtue of being future-directed. I can hope that you have already done so-and-so, and wish that you had done so-and-so (though past wishings seem always to be counteractual).
It might be possible to defend a “getting” operator (“getting what you want”) I such that I (θ(s)) = Ω where Ω is a set whose members are those sentences defining the conditions on g-wanting that happen to have been met; otherwise such sentences would be omitted from Ω.
However, one wonders about the rationale of the semantic primitive “wants-to-be-true”. For any predicate, F, there is a semantic primitive “wants-to-be-F” expressing, say, a function from persons to things that are F. Given that our language contains a countable infinity of predicates, F, there will be countably many semantically primitive want-operators, “wants-to-be-F”. But it is problematic whether such a language is learnable (Donald Davidson, “Theories of Meaning and Learnable Languages”) so, it becomes necessary to motivate the one primitive “want”-predicate, without in so doing committing ourselves to a kindred position on the others. Perhaps this could begin by observing that the current model seems to obliterate most of the distinction between p-wantings and g-wantings.
No doubt, a more promising partial interpretation of “Wz” would be to associate with it a function θ• from persons to ordered pairs of sentences and the truth-value T. Thus θ* (s) = < ϕ T >;, and “Wz” might be read “Z wants—to be true.” However these are (important) complications that we cannot here go into.
