Although concentrate on providing evidence from English for the shift in unaccusativity of verbs of manner of motion in the presence of a directional phrase, they do suggest that the pattern of auxiliary selection in languages such as Dutch and Italian supports a treatment of verbs of manner of motion as unaccusative on a directed motion use. The evidence they cite is shown in res13-res14 below (:185(13)). These data show that the shift of a manner of motion verb to a directed motion use is accompanied by a change in the selected auxiliary, from have to be. Levin and Rappaport Hovav assume that a verb's selection of the auxiliary be is an indication of its underlying unaccusativity.
There has, however, been much discussion on the topic of the
relationship between auxiliary selection and unaccusativity. For
example, Zaenen (1988, 1993) suggests that auxiliary selection in
Dutch has to do with semantic features such as
controllability and telicity, rather than being a
reflection of the underlying syntactic argument structure of the verb
phrase. Everaert (1992) proposes that auxiliary selection depends on
the theta- or case- assigning properties of a verb, again independent
of the argument structure of the verb phrase. He argues that both
auxiliary selection and unaccusativity can be defined in terms of
these properties, but that there is not necessarily a direct
correspondence between them. Furthermore, consider the Italian
sentences in res15.
In Italian, different types of directional phrases result in the selection of distinct auxiliaries (as in res15a), suggesting that the telicity of the directional phrase influences auxiliary selection. This is in contrast to the syntactic specification of the directional phrase as introduced by L&RH, as mentioned earlier. In addition, parallel sentences, containing verbs which are semantically highly similar and identical directional phrases, may differ in acceptability res15b (Di Tomaso, ). These data suggest that semantic factors interact with auxiliary selection, causing conclusions about syntactic encoding of unaccusativity drawn solely on the basis of auxiliary selection to be highly questionable.
Although there is no consensus on precisely which factors affect auxiliary selection, the analysis of the relevant data does suggest that while unaccusativity might be sufficient to determine the selection of the be auxiliary, it may not be a necessary condition for be selection (i.e. all unaccusative verbs may require a be auxiliary, while it may not be true that all verbs which select a be auxiliary are unaccusative). Thus on the basis of auxiliary selection alone, we should not assume that the uses of manner of motion and sound emission verbs in a directed motion sense are necessarily unaccusative.