Aspectual Shifting in the Perfect and Progressive

			Laura Wagner
		(University of Pennsylvania)

 The question that this paper begins with is this: Why can you have a
 perfect of a progressive (as in 1) but not a progressive of a perfect (as
 in 2)?  This paper takes a semantic-pragmatic approach and examines two
 kinds of aspect-shifting operations associated with the application of the
 PROG (progressive) and PERF (perfect) operators.
 
 (1)	Anne has been walking to school.
 (2)	* Anne is having walked to school.
 
 PROG and PERF are aspectual operators that shift the aspectual type of
 their predicate (cf. Dowty 1979, Moens 1987, Smith 1991, Klein 1994 among
 others for various versions of this claim).  Applying PROG to a sentence
 yields an aspectually stative sentence (Vlach 1981, Moens 1987).  Applying
 PERF to a sentence also yields a state, but a state of a particular sort:
 namely an Individual Level Predicate (ILP), in the sense of, for example,
 Carlson 1977 and Kratzer 1989 (cf Iatridou 1996 for relevant tests).
 
 In addition to the aspectual outcome of applying PROG or PERF, these
 operators initiate in some cases a second kind of aspectual shifting.  Both
 operators constrain the aspectual type of the sentence they operate over;
 they presuppose a sentence of a particular aspectual type.  If a sentence
 does not meet the presupposition of the operator, then it shifts types.
 This kind of aspectual shifting can be analyzed as presupposition
 accommodation (cf Karttunen 1974, Heim 1988).  Moens 1987 uses an analysis
 similar to this one to account for the imperfective paradox: PROG
 presupposes that it will apply to a durative activity; when the input is
 bounded (as in (3)) it shifts to an activity before PROG applies.  It is
 this shifting before PROG applies that accounts for the lack of a
 completion entailment in (4).  The presupposition of PERF is more general:
 it felicitously applies to anything but an ILP.  Thus the PERF creates
 shifted (and sometimes anomolous) readings for sentences only when they
 were ILP's before PERF applied (as in (5)).
 
 (3)	Mary ate a donut.
 (4)	Mary was eating a donut (... but she never finished it).
 (5)	Sue has been tall.
 
 The aspectual shifting done before the application of PROG or PERF and that
 done by applying them is different.  Presupposition accommodating shifting
 happens freely and leaves no semantic trace: it causes a re-construal of
 the meaning of the predicate.  The application of PROG and PERF, however
 retain part of the aspectual meaning of the pre-application sentence (how
 else could we identify the presupposition accommodation in the first
 place?).  Moreover, sentences that contain PROG or PERF cannot undergo the
 presupposition accommodation sort of shifting, but must remain as states or
 ILP's.  This last fact accounts for why you cannot have a PROG of a PERF:
 PERF application creates an ILP which cannot shifted to accommodate PROG's
 presupposition of a durative activity.  (PERF of a PROG is ok, though,
 since PROG's state is not an ILP and thus the PERF presupposition is
 satisfied.)