Dowty’s aspectual tests: standing the test of time but failing the test of aspect

Abstract

To test situation aspect (lexical aspect, aktionsart), especially telicity, and/or to classify situations into Vendlerian classes (states, activities, accomplishments, and achievements), aspectual literature has been standardly relying on Dowty's (1979) diagnostic tests. These are of three kinds -lexical co-occurrence, grammatical co-occurrence and logical entailment tests. This paper points out problems associated with implementing the tests. It argues that Dowty's tests can give false results, mainly due to two factors. First, some of the tests are based on volition rather than aspectual features. Second, upon reviewing the various workings of the phenomenon of aspect shift and coercion, I show how it can affect the results of Dowty's tests. Coupled with the fact that some of the tests are not applicable to some Vendlerian classes and with the problem of varying judgments among native speakers, Dowty's tests for Vendler classes are far from being watertight. Awareness of these problems is imperative when testing situation aspect.

Notes on actionality in two Nguni languages of South Africa

“…Dowty 1979: 61), as long as a bounding object or goal can be construed, even "activity"-like verbs are felicitous with the nga-temporal adverbial forms (93). This phenomenon has been observed in other languages, as well (see Walková 2012: 509, Smith 1997Binnick 1991: 176 The "imperfective paradox" (Dowty 1977: 45) test in (91c) is even more difficult to meaningfully apply. For one thing, there is no dedicated present progressive construction in Southern Ndebele or Xhosa, so the test cannot be applied at all using untensed imperfective frames.…”

“…27 In summary, none of the tests in ( 91) can distinguish straightforwardly between accomplishment-like and activity-like predicates in Southern Ndebele (or, quite probably, in Xhosa), even if factors such as quantization and multiple-event readings are controlled for (cf. Walková 2012). Prototypically bounded and prototypically unbounded states-of-affairs do not appear to have significantly different interactions with markers of grammatical aspect.…”

“…The distinction between verbs in (69) and verbs like phola 'cool down' in (70) cannot be explained in terms of the presence vs. absence of a lexically encoded coming-to-be phase: hlangana 'come together', hlakanipha 'be(come) clever', and phola 'cool down' all seem to lexically encode coming-to-be phases (see Section 5.1). As shown in ( 70), availability of a coming-to-be reading with bare temporal measure adverbials is not necessarily related to subject agency or volition, properties known to affect the outcomes of tests for actional classes (see Walková 2012). Rather, it may be that the availability of this reading depends on the inherentness of the phasal boundary itself (as well as that marking the beginning of the coming-to-be phase), making the coming-to-be phase measurable in time.…”

“…Testing for telicity (defined here as having an inherent endpoint) traditionally builds upon tests introduced in Dowty (1979); see Binnick (1991: 173-197) for an overview. Specifically, tests that target the difference between activities and accomplishments in English have been characterized as follows, following Walková (2012: 501-502 Walková (2012), these tests are somewhat unreliable even in English, the language for which they were designed. We found their application to Nguni languages even more challenging (see also Crane & Fleisch 2019).…”