Although the primeless examples in (256) look structurally similar, they differ in that the latter allows the addition of a direct object. The primed examples show the same for the corresponding constructions with attributively used present participles.

The claim in Subsection I that adjunct middles do not allow the presence of a direct object suggests that the two constructions are different, and that example (256b) is not an adjunct middle construction. Another possibility, however, is to assume that this claim was wrong and investigate whether the contrast between the (a) and (b)-examples can be explained in some other way. One reason to go this way is that there is in fact no a priori reason to expect that a direct object cannot occur in this kind of middle construction.

Now consider the examples in (257) with the transitive verb snijden. These examples show that the realization of the direct object leads to a rather strange result when the adverbially used adjective lekker is present, but is easily possible when the adverb is goedwell. If lekker and goed differ in that the former, but not the latter, favors the pseudo-intransitive use of snijden, this may provide an alternative account of the contrast found in (256); the judgments in (256) simply reflect those found in (257).

If so, the two constructions in (256) can both be seen as instances of the adjunct-middle construction. The two examples in (256) both have an easy-to-please counterpart, as shown in (258); this can be seen as support for a unified analysis of the two constructions.

The discussion above suggests that the claim in Subsection I that adjunct middles do not allow the presence of a direct object may be incorrect, and that this restriction may be related to the choice of evaluative modifier. We leave it to future research to investigate whether this conjecture is on the right track.

This subsection discusses a second construction that looks quite similar to the adjunct middle, but may need to be analyzed differently. Consider the primeless examples in (259), which look structurally similar, but differ in whether they have an easy-to-please counterpart.

The impossibility of (259b') seems to be related to the inability of the adjectives snelfast and vlotsmooth to take an experiencer voor-PP: *snel/vlot voor mij. The explanation for this is that the easy-to-please construction requires that the implied subject PRO of the infinitival clause be controlled by the (implicit) experiencer of the evaluative adjective; if the experiencer is left implicit, as in (259a'), it receives an arbitrary interpretation, which results in the generic meaning of the whole sentence. The ungrammaticality of (259b') can now be explained by appealing to the fact that adjectives like snel/vlot do not select an experiencer, and that the implied subject PRO of the infinitival clause is therefore not controlled, so that it cannot be assigned an appropriate interpretation.

Since adjunct middle constructions usually also require an adjective that selects an experiencer voor-PP, it remains to be seen whether (259b) can really be analyzed as a middle construction. Adjunct middle-like constructions without an easy-to-please counterpart are quite common, but do not seem to have received much attention so far. Example (260) presents two more cases based on the pseudo-intransitive verb vervento paint and the adverb gelijkmatigevenly, which again lacks an implicit experiencer; the (a) and (b)-examples closely resemble the regular middle and the adjunct middle construction, respectively.

As noted earlier, it remains to be seen whether the constructions without an easy-to-please counterpart can be analyzed as regular adjunct-middle constructions. We leave this to future research, while noting one fact that favors a middle analysis, namely that these constructions have the typical middle semantic feature of referring to inherent properties of their subjects.

We have mentioned several times that adjunct middles like the primeless examples in (261) often have easy-to-please counterparts expressing more or less the same meanings, and in which the subjects of the matrix clause also correspond to the complement of some adverbial PP; the subjects in the (a)-examples correspond to the nominal part of the instrumental PP met dit meswith this knife and the subjects in the (b)-examples correspond to the nominal part of a locational PP. In fact, there is strong evidence that such PPs are syntactically present in the primed easy-to-please constructions, given the obligatory presence of the stranded prepositions meewith and opon in the infinitival clauses.

However, the generalization that adjunct middles have easy-to-please counterparts breaks down if the subject corresponds to the nominal part of an adverbial PP that does not allow R-extraction; we will argue below that the primed easy-to-please constructions in (262) are excluded because adverbial phrases of time and manner like op rustige middagenin quiet afternoons and in een hoog tempoat a fast pace usually do not allow R-extraction.

The contrast between the easy-to-please constructions in (261) and (262) can be directly related to a difference in R-extraction, since easy-to-please constructions arguably involve wh-movement of an empty operator into the initial position of the infinitival clause; cf. Section A28.5, sub IVA3, for a detailed discussion. This means that the structures of the easy-to-please constructions presented above are as given in (263); note that the prepositional form meewith in (263a) occurs only when R-extraction has been applied, which of course provides strong support for the proposed movement analysis.

The unacceptability of the easy-to-please constructions in (262) now follows straightforwardly from the fact that the adverbial PPs of time and manner in (263c&d) do not allow extraction in any form; the indicated movement of the operators is therefore ruled out. Note that this account of the contrast between the easy-to-please constructions in (261) and (262) implies that the adjunct middles cannot be derived from the corresponding non-middle constructions by syntactic movement, since we would then expect the adjunct middles in (262) to be excluded for the same reason as the corresponding easy-to-please constructions; this suggests that the formation of adjunct middles is not a syntactic process, but a morphological operation on the verb that affects its argument structure.

We see in (264) that adjunct middles and easy-to-please constructions also differ in that subjects of adjunct middles cannot correspond to the nominal parts of PP-complements, whereas subjects of easy-to-please constructions can. The acceptability of the easy-to-please construction in (264c) follows, of course, from the fact that R-extraction from PP-complements is allowed.

Subjects of easy-to-please constructions can also correspond to the nominal parts of predicative PPs, as shown in (265). Again, this is consistent with the R-extraction analysis.

Comparable adjunct middles are not to be expected, since we have seen in Subsection I that transitive and unaccusative verbs cannot normally be used as input for adjunct middle formation. That unaccusative verbs with a predicative PP cannot be the input of adjunct middle formation can be nicely illustrated by the examples in (266) and (267). The examples in (266) show that movement verbs such as springento jump have two uses: an intransitive use, in which case the verb selects hebben in the perfect tense and the PP functions as a regular adverbial phrase of place, and an unaccusative use, in which case the verb selects zijn in the perfect tense and the PP functions as a complementive indicating a change of location.

Semantically, the adjunct middle construction in (267) is clearly related to the intransitive construction in (266a), not to the unaccusative construction in (266b): it is the jumping on the trampoline that is said to be nice, not the jumping onto the trampoline.

The discussion in this subsection has shown that adjunct middles and easy-to-please constructions differ in that the subject of the latter may correspond to the nominal part of a PP that allows R-extraction; on the other hand, adjunct middles take subjects that correspond to the nominal part of a wide range of adverbial PPs, regardless of whether these PPs allow R-extraction. Moreover, adjunct middles live up to their name by never taking a subject corresponding to the nominal part of a PP-complement or a predicative PP.