Advances in Role and Reference Grammar

PREPOSITION ASSIGNMENT IN ENGLISH 289

which occur in the semantic structures of prepositions analyzed in sections
2.5, 2.6 and 2.7.
The sub-class of predicative functions labeled basic functions (see sec­
tion 2.3) specifies two primitives, or in Dowty's terminology, "logical con­
stants": be-at' and be-via'. The first occurs in simple locational contexts, as
in (37).

(37) a. John is at school.

b. [be-at' (school, John)]

In this sentence, be-at' can have either a momentary (right now) or stage-
level (every day) interpretation. This primitive is also, by definition, part of
the semantic structure (CAUSE [BECOME be-at'] or CAUSE [BECOME
NOT be-at']) of all motion accomplishment verbs, as illustrated thus far in
numerous contexts such as (29), where be-at' specifies a locative-goal and
NOT be-at' specifies a locative-source. The second primitive, be-via',
defines the third semantic argument, whether or not syntactically realized,
in a motion accomplishment, as in (29) {Rita walked from the school
through the park to the store). Like locative prepositions, path prepositions
occur predicatively in contexts where the path is not derived from the LS of
the verb, as in (38).
(38) a. Rita walked through the park.
b. [be-via' (park, [walk'(Rita)])]
Nothing in the LS of the activity verb, walk — walk' (x) — specifies a path
argument of the verb. Rather, through specifies the location of the entire
situation described in the activity structure.

2.4.1 Be-at'

Dowty's decompositional model (210-211) specifies be-at', be-on', and be-

in' as primitives "though they are not literally the constants translating Eng­

lish at, on and in." (210) In Dowty's analysis, locative-goals in accomplish­

ment structures are listed as: [BECOME be-at'(x,y) = to; BECOME be-in'

(x,y) = into; and BECOME be-on' (x,y) = onto. Thus, be-at' is adjacency;

be-in', containment; and be-on', adjacency above a horizontal surface.

Although all three function as primitives in the lexicon, expressing reg­

ularities evidenced by their negative counterparts —from, out of and off—

we will consider the distinctions between the three to be irrelevant at this

level of analysis; i.e., be-at' is singly taken as representing all three. In