(42) Joan will/may buy a car...
a. *She parked it on the street.
b. and park it on the street.
c. Fred already has bought one.
Within certain subordinate clauses wefind similar behavior. For instance, the adjectivepossibleis close in meaning to
themodalmay, and itis nosurprise thatitscomplementclauseis freeofreferentialcommitment: ifwesay something is
possible, we are obviously not committing ourselves to its having happened. Similarly, the verbwantcarries a strong
implication that the desired situation does not yet obtain.
(43) a.It is possible that Joan bought a car. [No implication that she did]
b. Joan wants to buy a car. [No implication that she has or will]Moreover, theanaphorictests applytoa carin thesame way as they do in (41) and (42)(It is possible that Joan bought a car
...and parked it on the street/*She parked it on the street, etc.).So as in the earlier examples,a caris referentiallylicensed by
the event of buying.
However, the main clauses in (43) do make existential claims: theyclaim that something is possibleand that Joan does
want something. Hence NPs in these main clauses can serve as antecedents for definite pronouns in a succeeding
sentence (italics indicate antecedent and pronoun).
(44) a.It seems possible toa friend of minethat Joan bought a car.Hethinks she's rich.
b. A friend of minewants to buy a car.Hedoesn't have much money, though.In order to encode the referential tier for such cases, we cut the graph of referential dependency above the index for
the subordinate clause; the cut-off portion lacks assertive force. It is notationally convenient to enclose the cut-off
portion in a frame, labeled with an“operator”that notates the alteration of existential claims. (45) shows the resulting
structure for (44b); connoisseurs of Discourse Representation Theory should begin to see the resemblance at this
point.^213
402 SEMANTIC AND CONCEPTUAL FOUNDATIONS
(^213) In this example, the subject ofbuy is notated in the descriptive and referential tiers as if it were a definite pronoun called PRO. Whether this appears also in syntactic
structure depends on one's school of thought: Standard generative grammar givesbuy a syntactic subject, most constraint-based theories do not.