Noun phrases must be assigned case by some case-assigning head. In a transitive construction such as (581a), the subject and the direct object are assigned nominative and accusative case by what we have called tense and the verb, respectively. Example (581b) further shows that the direct object of the active construction becomes a derived theme-subject in a passive construction. This is usually explained by the assumption that passive participles are not able to assign accusative case to their internal argument, which must therefore be assigned nominative case by tense, which further implies that the subject of the active construction is suppressed; cf. Section 3.2.1 for more details.

Section 2.1 has argued that theme-subjects occur not only in passive constructions, but also with unaccusative verbs; such verbs cannot assign accusative case to their internal argument either, which must therefore be assigned nominative case by tense. This can be illustrated by the causative-inchoative alternation in (582): if the verb breken selects the auxiliary hebbento have, as in (582a), it is a transitive verb and thus able to assign accusative case to its internal argument, but if it selects the auxiliary zijnto be, as in (582b), it is an unaccusative verb, so that accusative case is no longer available and the internal argument of (582a) must appear as the subject of the construction (and the subject of the corresponding transitive construction cannot be expressed).

The examples in (583) show that a word-for-word translation of example (582b) into a language like French or Italian results in an ungrammatical construction; to obtain an acceptable result, the simplex reflexive se/si must be added. Burzio (1986: §1.5) claims that the simplex reflexive marks the subject of the construction as a derived theme-subject; cf. also Dobrovie-Sorin (2006) for a survey of the relevant literature.

Although the standard Dutch simplex reflexive cannot be used in the same way as the Romance reflexive markers se/si, it is worthwhile noting that Heerlen Dutch, a variety of Dutch spoken in Limburg, does employ the simplex reflexive in the same way as French and Italian; cf. Cornips (1994) and Cornips & Hulk (1996).

For completeness’ sake, note that Heerlen Dutch also has examples like Jan brak zich het glas, but in these examples the reflexive functions as a possessor (Jan broke his glass) or as a benefactive (Jan broke the glass for himself); cf. also Subsection III.

It is important to note that the Heerlen Dutch example in (584b) differs from standard Dutch (582b) in that the verb does not select the auxiliary zijn, but hebben, suggesting that the verb retains its ability to assign accusative case. If this is indeed the case, we need to explain why the internal argument cannot be assigned accusative case, i.e. why we are dealing with a theme-subject in (584). Burzio (1986: §1.5) and Everaert (1986) have argued that in inherently reflexive constructions like (583) and (584) the non-argument se/zich not only marks the presence of a theme-subject, but actually forces the suppression of the regular subject of the corresponding transitive construction. The argument goes as follows. Since simplex reflexives can be used as arguments, they are ordinary noun phrases which must be assigned (accusative) case. Since verbs can only assign accusative case to a single argument, this means that the internal arguments in (583) and (584) can no longer be assigned accusative case, and must therefore be assigned nominative case, which results in the suppression of the subject of the corresponding transitive construction (as in passive constructions).

Returning to inherent reflexivity, the case-absorption approach predicts that there are no inherently reflexive verbs that take a direct object, and it does seem to be the case that the vast majority of inherently reflexive verbs do not select a DP but a PP-complement (if any). A representative sample of inherently reflexive PO-verbs is given in (585).

There are a number of possible counterexamples to the claim that inherently reflexive verbs do not take a direct object. Examples are: zich iets aantrekken vanto care about sth, zich iets aanwennento get used to sth, zich iets afwennento cure oneself of sth, zich iets afvragento wonder about sth, zich iets herinnerento remember sth, zich iets permitterento afford sth, zich iets toeëigenento appropriate sth, zich iets verwervento acquire sth, zich iets voorstellento imagine sth, zich iets voor de geest roepento remember sth. However, it seems that in these cases we are not dealing with accusative but dative reflexives; we will return to this in Subsection III.

The inchoative constructions in (582b) and (584b) suggest that languages can in principle use two strategies to detransitivize causative verbs such as breken: either the verb is deprived of its ability to assign accusative case, in which case the verb selects the perfect auxiliary zijn, or accusative case is absorbed by a simplex reflexive. This subsection shows that standard Dutch uses both strategies.

Although standard Dutch does not use simplex reflexives to mark causative-inchoative alternations with verbs such as brekento break in (582) above, the examples in (586) and (587) show that there is a comparable alternation in which the simplex reflexive is used to obtain a detransitivizing effect; cf. Everaert (1984:52-3). Given our conclusion on the basis of the Heerlen Dutch examples in (584) that this effect is due to accusative case absorption by the simplex reflexive zich, it is not surprising that the reflexive inchoative construction takes the auxiliary hebben in the perfect tense.

The assumption that we are dealing with derived theme-subjects in the reflexive inchoative examples is supported by the fact that they are subject to the same selection restrictions as the direct objects of the corresponding transitive constructions. The object of the transitive verb verspreiden in (588a), for instance, cannot refer to a single concrete entity; it is usually plural or headed by a collective noun like menigtecrowd or a propositional noun like geruchtrumor. Example (588b) shows that the subject of the corresponding reflexive construction is restricted in exactly the same way.

The causative-inchoative alternations with and without a simplex reflexive should certainly not be seen as idiosyncratically constrained alternatives; they probably reflect a more principled difference. This is clear from the fact, illustrated by (589), that they are sometimes available simultaneously.

The two inchoative constructions also differ in meaning. In the story alluded to (Paulus en het levenswater by Jean Dulieu), the witch Eucalypta accidentally drinks her own transformation potion, which was originally intended for the goblin Paulus. The presence of the simplex reflexive in the inchoative constructions depends on the feature [±control] discussed in Section 1.2.3, sub IIIB: if the subject of the inchoative construction is (in principle) able to control the activity, as in (589b), the simplex reflexive is preferably present, but if the subject is not able to control the activity, as in (589b'), the reflexive must be absent.

The same condition may apply to the examples in (590): the use of the simplex reflexive in examples such as (590b) is preferred by many speakers, because the subject of the clause is taken to be the instigator of the event denoted by the verb, but disfavored in examples such as (590b'), because the subject is typically taken to be a patient. However, judgments are subtle, and there are speakers who consider example (590b') fully grammatical with zich; cf. Everaert (1986:84). Some of our informants share this judgment, but claim that the use of zich creates a “spooky” effect in the sense that it feels as if the curtain acts like an animate being, which would of course be consistent with our proposal above.

However, there are also reflexive inchoative constructions such as (591b) in which the proposed semantic effect is clearly absent; despite the fact that the referent of the subject is unable to control the event, the reflexive must be realized in this example. Thus it remains debatable whether the semantic contrast between the two (b)-examples in (589) and (590) is really due to the absence or presence of the reflexive.

Nevertheless, we will take the fact that the two (b)-examples in (589) and (590) differ in the way they do as evidence for the claim that the two types of inchoative constructions are different. Cross-linguistically, there also seem to be two strategies for obtaining a causative-inchoative alternation. Schäfer (2008:120) calls the first way anticausativization; it involves some detransitivization morpheme which is usually reflexive in nature. The second way is called causativization, which involves a morpheme that introduces a causer argument. The morpheme can be overt, but Pesetsky (1995) shows that it can also remain phonetically unexpressed.

The case absorption hypothesis proposed in Subsection A amounts to saying that the reflexive inchoative construction is derived by anticausativization: the reflexive absorbs the accusative case of the verb, as a result of which the theme argument must be promoted to subject and the external argument of the verb can no longer be expressed. Since Section 3.2.3 will show that non-reflexive inchoatives are always regular unaccusative verbs, we can assume that they can be the input of a derivational process with a phonologically empty morpheme, which adds an external causer to the argument structure of the input verb.

An advantage of assuming the two processes in (592) is that it makes it possible to allow for the fact that the two inchoative constructions can occur side by side without having to assume idiosyncratic constraints on the processes involved. For instance, the examples in (589) can be accounted for by assuming that the verb veranderen is stored in the lexicon as an unaccusative verb, as in (589b'). This verb can be the input of the causativization process that derives the transitive version of the verb in (589a). This derived transitive verb can then be used as input for the anticausativization process that derives the inherently reflexive verb in (589b). Of course, this proposal does not imply that reflexive and non-reflexive inchoative verbs always coexist; the fact that verspreidento spread in (586)/(588) cannot be used as a non-reflexive inchoative verb could be accounted for by assuming that the inherently reflexive verb is stored as a lexical verb in the lexicon.

For completeness’ sake, note that the difference between standard and Heerlen Dutch examples in (593) may now simply reflect a lexical difference between the two languages: if brekento break is stored as an unaccusative verb in standard Dutch, its causative transitive counterpart must be derived by the process of causativization in (592b), i.e. by adding a causative zero morpheme; if, on the other hand, breken is stored as a transitive verb in Heerlen Dutch, the anticausative counterparts must be derived by the process of anticausativization (592a), i.e. by using the case-absorbing morpheme zich.

French and Italian, which also use the reflexive form in such cases, can be considered similar to Heerlen Dutch. We should point out, however, that these languages pose a possible problem for the proposal outlined above; the French and Italian inchoative constructions in (583) take the auxiliary être/essereto be rather than avoir/avereto have in the perfect tense: cf. the French perfect-tense example Le verre s’est briséthe glass has broken. Although this is not the place to solve this problem, we would like to suggest that this difference may be related to the fact that se differs from (Heerlen) Dutch zich in that it cannot be used in argument positions or to the fact that it cliticizes onto the verb, which may eliminate the need for case assignment. This may be empirically supported by the existence of so-called clitic doubling constructions, in which a clitic doubles a syntactically present argument; cf. Anagnostopoulou (2006) for an overview. If so, the role of Romance si/se cannot be case absorption, but must be something else; cf. Dobrovie-Sorin (2006) for relevant discussion.

Many causative psych-verbs have inherently reflexive counterparts, as shown in the sentence pairs in (594). Section 2.5.1.3, sub IV, has shown that the object experiencers of the primeless examples appear as the subjects of the inherently reflexive constructions in the primed examples.

However, we cannot simply assume that the primed examples are derived from the primeless examples as a result of case absorption by the simplex reflexives. The reason is that promotion to subject is normally restricted to cases in which an external argument is present. However, we have reasons to assume that the subjects in the primeless examples in (594) are not external arguments; the most important is that external arguments seem to occur only with psych-verbs in a third alternant illustrated in (595); cf. Bennis (1986).

The case-absorption approach will lead to a more or less correct result if we assume that the inherently reflexive constructions in the primed examples of (594) are derived from the transitive examples in (595): zich will absorb the accusative case of the verb and consequently the experiencer object must be promoted to subject (for which reason the causer subject in (595) must be suppressed). Note, however, that this derivation cannot be syntactic, since it would leave unexplained why the met-PPs from the examples in (595) cannot be used in the inherently reflexive constructions in (594); we must therefore be dealing with a lexical process. Finally, note that the case-absorption approach correctly predicts that the simplex reflexive cannot be used in the causative psych-constructions in the primeless examples in (594); since accusative case must be assigned to the experiencer object, there is no case available for the simplex reflexive, so the requirement that each noun phrase be assigned case would be violated if it were present.

This subsection discusses reflexive middle constructions, which seem to be another case in which the simplex reflexive acts as an accusative case absorber.

The previous subsections have shown that in at least a subset of inherently reflexive constructions, the simplex reflexive can be analyzed as a non-argument that has the capacity to absorb case. Therefore, the main verb in these constructions will not be able to assign accusative case to its internal argument, which must therefore be promoted to subject. The discussion has further shown that the case-absorption approach applies not only to single verbs, but also to the more complex reflexive middle construction involving the exceptional case-marking verbs latento let and ziento see. We have also considered the question as to whether case absorption is restricted to the accusative, or whether it can also affect the dative. As this question has not been studied much, the best we can do at this time is make a number of remarks and hope for future research to arrive at a more conclusive answer.

Example (606a) shows that dative noun phrases cannot normally occur as simplex reflexives in simple clauses, whereas (606b) shows that they can do so as part of clauses embedded under the permissive verb latento let (provided that the subject of the embedded clause is not overtly present); italics indicate co-reference.

This contrast is what we would expect on the basis of the no co-argument restriction on the binding of simplex reflexives. The simplex reflexive in (606a) is selected by the same verb as its antecedent, and is therefore bound by a co-argument. In example (606b), on the other hand, the matrix subject is not an argument of the embedded infinitive, but of the matrix verb, so the reflexive is not bound by a co-argument. With this in mind, consider the examples in (607).

The examples in (607) are inherently reflexive constructions, as is clear from the fact that the simplex reflexive does not function as an argument of the verb: it cannot be replaced by a lexical noun phrase such as Marie, and in its English rendering a reflexive need not be used. If this line of reasoning is correct, and if inherently reflexive constructions do indeed involve case absorption, we may conclude that simplex reflexives are capable of absorbing not only accusative but also dative case; cf. example (604) in Subsection IID, for other potentially relevant cases.

There are at least three productive (semi-)morphological processes that can derive inherently reflexive verbs from regular simplex verbs: prefixation with over-, prefixation with ver-, and extension of the verb with the particle in. The semantic effects of these processes can be inferred from the glosses.

The examples in (609) show that if the input verb is transitive, the original direct object appears as a PP-complement (Everaert 1986); this suggests, again, that simplex reflexives absorb accusative case.

There are a number of idiomatic constructions such as (610), which are resultative in nature. These constructions differ from the inherently reflexive resultative constructions discussed in Subsection I in that the agent of the activity does not acquire a certain property as a result of the activity denoted by the verb, but rather becomes the “possessor” of the entity denoted by the noun: Jan will have a lump, an accident, or a delirium as a result of the activity of laughing, eating, or drinking, respectively. The constructions in (610) would be consistent with the no co-argument restriction on the binding of simplex reflexives if the strings in square brackets were phrases headed by some empty element comparable to the verb hebbento have; however, at present we have no evidence to support such a claim, and so we leave these examples as a problem for the current proposal.