A serious flaw in Dowty's tests is his assumption that achievement verbs are all
punctual; this is clearly false, as sentences like The clothes are drying and My cof
fee is cooling attest. Hence where relevant, achievement verbs will be divided into
punctual (P) and durative (D) subclasses.
See Schwartz, this volume, for a discussion of the LS of attributive-identificational
constructions.
do' is a generalized unspecified activity predicate proposed in FVV and was not
included in Dowty (1979).
It should be noted that this contrast cannot be reduced to the presence or absence
of articles, because it occurs in languages which do not have articles, e.g. Geor
gian and Japanese (cf. Van Valin 1990b). This contrast revolves around whether
the direct object is a specified entity or quantity, in which case it delimits the
action and supplies a temporal boundary for it, or whether it is unspecified and
therefore does not serve to delimit the action. See Dowty's discussion (1979:60ff.)
for a detailed explication of the semantic basis of these contrasts. These examples
bring up an important point about the Dowty system. Even though this discussion
is phrased in terms of verb classes, it includes both the lexical meaning of verbs
and their use in the context of the whole clause. Hence in some languages, e.g.
English, the status of the direct object or other core argument is relevant to deter
mining the use of a verb in a particular clause but not its basic lexical classifica
tion, as illustrated in the pairs of sentences in (22). In German and Hungarian, on
the other hand, the system of verb prefixation signals the contrast.
This role covers a whole group of locative relations. "Goal" may be defined as the
locative argument in the following LS configuration: ... BECOME be-at/in/on' (x,y).
"Recipient" may be defined as the locative argument in the following configura
tion: ... BECOME have'(x,y). "Source" may be defined as the locative argument in
either of the configurations: ... BECOME NOT be-at'(x,y) or ... BECOME NOT have'(x,y).
Oblique core arguments were termed "inner peripheral arguments" in FVV; cf.
fn. 5.
The notion of "proto-roles" proposed in Dowty (1987, 1991) is similar but not
identical. See Van Valin (1992) for detailed discussion. Jackendoff (1987) posits
multiple tiers in his thematic relations theory, but the agent-patient action tier is
quite different from the RRG macrorole tier.
For a discussion of "inverse verbs", i.e. verbs like gefallen "like" in German and
piacere "like" in Italian which appear to have their "subject" in the dative case
and their "object" in the nominative, cf. section 5.2.1, Van Valin (1990a, 1991b).
Not all verbs of this class work this way, but the vast majority do in English. Cf.
FVV, pp. 58-9.
See Van Valin (1990a) for a discussion of the syntactic consequences of this fact.
