these families of functions a nonstandard version of type logic.^189 Instead of the usual primitive types e and t
(individuals and truth values, respectively—the twopossibletypes to whichexpressions can refer, according to Frege),
this approach has a much broader range of primitive types: the major ontological categories Object, Event, and so
forth.Usingthestandard notation<a,b>todenotea functionfro mse manticobjectsoftypeaintosemanticobjects of
typeb, we can encode the types of these functions as follows:
(22) a.BE:<(X,Y), State>, X and Y an ordered pair, where the types of X and Y depend on semanticfield^190
b. STAY:<(X,Y), Event> (same stipulations as 22a)
c. GO: <(Object, Path), Event>
d. EXT, ORIENT: <(Object, Path), State>
e. TO, FROM: <x, Path>, where the type of X depends on semanticfield
f. INCH: <State, Event>
g. PERF: <Event, State>
h. CAUSE, HELP, LET(three-argument version):
<(Object/Event, Object, Event), Event>
i. CAUSE, LET(two-argument version):
<(Object/Event, Event), Event>(22) may seem likea major departure from type logic. However, I haveencountered no empirical arguments thateand
tare thecorrect choiceofprimitivetypes—only methodologicalargumentsbased on thedesiretominimize primitives.
In the end, the parsimony of a logic containing only two primitive types must be weighed against the potential
explanatory power of a richersystem; itis an empiricalissue, notjust a methodologicalone.Anotherdifferencefrom a
more standard typelogicis thatthecategory correspondingtoa sentence is an Eventor Staterather than a truth value,
as argued in Chapter 10. A truth value can be seen as the evaluation of the Event or State expressed by the sentence
with respect to world as conceptualized. Again, this choice is to be weighed on empirical grounds.
11.8.2 Building verb meanings
These functions can be used to build up revealing skeletons of verb and preposition
364 SEMANTIC AND CONCEPTUAL FOUNDATIONS
(^189) A different interpretation of this notation as a type logic is worked out in Zwarts and Verkuyl (1994).
(^190) There is no proble min principle reducing the functions of two and three variables in (22) to the more standard successively e mbedded functions of a single variable;
however, I have found no particularly compelling reason to do s o at the present level of detail of the theory.