Levin and Rappaport Hovav (1995) aim for a unified treatment of resultative constructions, on the basis of the distinction between unaccusative and unergative verbs. Their proposals with respect to resultatives are intended to support their main argument that unaccusativity is syntactically encoded while being semantically motivated.
According to the approach to resultatives, variations in syntactic behaviour in the resultative construction stem only from underlying differences in the unaccusativity of the verbs participating in the construction and the existence of a Direct Object Restriction, summarised in DOR.
Direct Object Restriction (DOR): a resultative phrase may be predicated only of an immediately postverbal NP, not of a subject or of an oblique complement.
According to this constraint, unergative (intransitive) verbs with no underlying object and hence no immediately postverbal NP require a ``fake reflexive'' (an NP co-indexed with the subject) or some other NP to be inserted into postverbal position, to introduce something which the resultative can predicate res35d. Transitive verbs, on the other hand, have an overt postverbal NP and unaccusatives have an underlying object (which moves to subject position due to syntactic constraints) and so these verbs have (underlying) direct objects which the resultative can predicate res35a. The implication of this treatment of resultatives is that any verb which appears in what term the unaccusative resultative construction, in which an intransitive verb is directly followed by a resultative phrase, must be treated as having unaccusative argument structure in order to account for the felicity of the verb in a resultative construction without a postverbal NP and their infelicity in this construction with a postverbal NP. The verbs in res1-res3 must therefore all be given an unaccusative analysis.
They slowly swam apart. [L&RH (5.15b)] *They slowly themselves swam apart.
The refrigerator door clicked open. [L&RH (5.27a)] *The refrigerator door clicked itself open.
The prisoners froze to death. [L&RH (2.19b)] *The prisoners froze themselves to death.
The intransitive freeze is generally accepted to be unaccusative
and therefore this behaviour is explained via the Direct Object
Restriction, while swim and click are not. In the
context of change-of-location resultatives, however, the manner of
motion and sound emission verbs must be viewed as unaccusative in
order for the DOR to hold. This is accomplished in the account
via a lexical rule applied to verbs of manner of motion (e.g.
swim) and verbs of sound emission (e.g. click), to
the exclusion of other verb classes, which adds a directional phrase
requirement and shifts the lexical classification of the verb to be
unaccusative. There will therefore be two lexical entries for
swim: 
and
. Note
that the added directional phrase need not be one that makes the
denoted eventuality telic, and that thus its addition is a syntactic
rather than semantic specification.
The evidence which present in favor of this unaccusative treatment of verbs of manner of motion and verbs of sound emission on a directed motion use comes in part from auxiliary selection data in Dutch and Italian, but mainly from the English resultative construction itself and from the causative alternation. I will discuss each of these pieces of evidence in turn below.