The definitions we have adopted predict that certain grammatical aspects will not be available for certain situation types. For example, the definition of imperfective aspect as indicating that TT ⊂ TSit implies that a situation expressed in the imperfective cannot be strictly punctiliar; the situation time must have some duration. A semelfactive event is construed as being instantaneous; it has no duration. For this reason, when a semelfactive event is described in the imperfective (e.g., he was tapping on the window), it cannot be interpreted as referring to a single instance, but must receive an iterative (= repetitive) interpretation. Similarly, an instantaneous change of state cannot be described in the imperfective (e.g. ⁇He was recognizing his old classmate) without some very unusual context. With other changes of state, the use of the imperfective (e.g., he was dying) may shift the reference from the change itself to the process leading up to the change. This kind of shift can be seen as a type of coercion.
The same constraint applies to semelfactives in Mandarin. In Chinese as in English, a semelfactive event described in the imperfective cannot be interpreted as referring to a single instance, but must receive an iterative interpretation (27). As we would predict, this is possible only with the progressive zai, and not with the continuous –zhe.
(27) Zhāngsān zài qiāo mén.
Zhangsan prog knock door
‘Zhangsan is knocking on the door.’ (C. Smith 1997: 272)
Similarly, the definition of perfective aspect as indicating that TSit ⊆ TT makes predictions about the kinds of situations that can appropriately be expressed in the perfective. When a state is described in the perfective aspect, what is asserted is that the state was true during the topic time, as discussed above. When an event is described in perfective aspect, what is being asserted is that the whole event took place within the topic time. For activities, which do not have an inherent endpoint, perfective descriptions in the past can be interpreted as bounded events, as in (28a): ‘I played tennis for a while.’ Alternatively, as illustrated in (28b), they can get a habitual interpretation, which has properties similar to a state.
(28) a. I played tennis yesterday.
b. I played tennis when I was in high school.
For telic events, and in particular for accomplishments, the end-point or culmination is an intrinsic part of the event; so a perfective description of that event should be false if the culmination is not in fact attained. This prediction holds true for English, as illustrated in example (15b) above, and for many other languages. However, a number of languages have been identified in which this culmination is only an implicature, rather than an entailment, for accomplishments expressed in the perfective. In Tagalog, for example, it is not a contradiction to say: ‘I removed the stain, but I ran out of soap, so I couldn’t remove it.’20 Other languages in which such “non-culminating accomplishments” are possible include Hindi, Mandarin, Thai, several Tibeto-Burman languages, various Philippine-type languages, and at least two Salish languages.21
The exact conditions under which “non-culminating accomplishments” can occur vary from one language to another, but the existence of such cases might suggest that we need to modify our definition of perfective in some way. Another alternative that we might consider starts with the recognition that accomplishments are composed of two “phases”: the first phase is a process or activity which leads to the second phase, a change of state.22 In building a house, for example, the first phase would be doing the work of building and the second phase would be the coming into existence of a completed house. We might account for the difference between languages like English vs. languages like Chinese or Tagalog by recognizing that for languages of the latter type, there are certain conditions under which a VP that normally describes an accomplishment can be used to refer to just the first phase of the event, i.e. a process or activity.
This two-phase analysis also gives us a way of thinking about a puzzling fact concerning accomplishment predicates in English, which Dowty (1979) refers to as the “imperfective paradox”. Building on Vendler’s (1957) discussion of these facts, Dowty points out that with state and activity predicates a statement in the imperfective (29a, 30a) entails the corresponding statement in the perfective (29b, 30b). With accomplishment predicates, however, this entailment does not hold (31–32).
(29) a. Arnold was wearing a wig.
b. Arnold wore a wig. [a entails b]
(30) a. George was speaking Etruscan.
b. George spoke Etruscan. [a entails b]
(31) a. Felix was writing a letter.
b. Felix wrote a letter. [a does not entail b]
(32) a. Sarah was running to the library.
b. Sarah ran to the library. [a does not entail b]
Dowty goes on to ask: Given the fact that accomplishments always have a natural end-point, how can the imperfective description of the event be considered true if that end-point was never achieved?23 It seems that English, like Chinese and Tagalog, allows a shift in meaning so that a VP which normally describes an accomplishment can be used to refer to just the first phase of the event. In English, however, this shift seems to be possible only in the imperfective.
20 Dell (1983: 186).
21 References: Hindi (Singh 1991; 1998), Mandarin (Soh & Kuo 2005; Koenig & Chief 2008), Thai (Koenig & Muansuwan 2000), Salish (Bar-el et al. 2005), Tibeto-Burman (Larin Adams, p.c.).
22 Klein et al. (2000).
23 Dowty’s solution was to propose that the progressive encodes not only aspect but also modality, that is, quantification over a certain class of possible worlds. He designated the relevant class of possible worlds inertia worlds, which he defined as follows: an inertia world is a possible world which is exactly like the actual world under discussion up to and including the topic time, “and in which the future course of events after this time develops in ways most compatible with the past course of events” (Dowty 1979: 148). In other words, inertia worlds are possible worlds in which the expected outcomes from a given situation are actually realized. Dowty then proposed a new definition of the progressive which says that John was X-ing will be true when asserted about a time interval I just in case (i) there is some longer time interval Iʹ which contains I and extends beyond the end-point of I; and (ii) John X-ed is true in all inertia worlds when asserted about time interval Iʹ.
