Example (187) above has already shown that weak object pronouns cannot occur in main-clause initial position, but must occupy a position in the middle field of the clause. The examples in (189) further show that they immediately follow the subject when the latter is not in main-clause initial position; cf. Huybregts (1991). This is true not only when the subject is in the regular subject position, as in the primeless examples, but also when it is contrastively focused and can be assumed to be located in the specifier of FocP lower in the clause, as in the primed examples; cf. Section 13.3.2.

In subject-initial main clauses, weak object pronouns immediately follow the finite verb in second position. This is illustrated in (190) by showing that modal adverbials cannot precede the object pronoun, but this holds for other constituents as well.

Note that this correctly implies that the weak object pronoun cannot occur between the modal adverbial waarschijnlijk and the contrastively focused subject in the primed examples in (189) either.

The previous subsection has shown that weak proform shift cannot affect the unmarked order of the subject and the objects. This is different when it comes to the relative order of direct and indirect objects: while direct objects usually follow nominal indirect objects under a neutral (i.e. non-contrastive) intonation pattern, weak pronominal direct objects usually precede them. The examples in (191) show that this is true regardless of whether the indirect object is non-pronominal or pronominal. Note that it also holds when the two object pronouns have the same form: the first object pronoun in dat Peter ʼm ʼm aanboodthat Peter offered it to him is interpreted as the direct object.

Weak direct and indirect object pronouns cluster together, which may be related to the fact, illustrated in the previous subsection, that they must both be adjacent to the finite verb, or to the subject if it is not in clause-initial position; the only novelty is that this restriction applies not only to individual pronouns, but to the whole cluster. Haegeman (1993a) has further observed for West Flemish that the inversion of indirect and direct object requires scrambling of the indirect object. Example (192a) shows that the same is true for Dutch under a neutral intonation pattern; the degraded order improves when the indirect object is given a contrastive accent. For completeness’ sake, it should be noted that weak pronoun shift of the object is possible with a prepositional indirect object, as in (192b); this way of expressing the intended meaning may in fact be the preferred way for some speakers.

Reversal of direct and indirect objects is only possible with reduced direct objects. However, it is not easy to demonstrate reversal for strong referential personal pronouns, because they are not normally used to refer to inanimate entities. The examples in (193) therefore illustrate this restriction on object reversal with the demonstrative pronoun diethat one; again, the marked order in (193a) improves when the indirect object is given a contrastive accent.

The fact that object pronouns can be inverted while non-pronominal nominal arguments cannot has led to the hypothesis that they do not occupy the same position in the middle field of the clause, i.e. that only weak pronouns undergo weak proform shift; cf. Zwart (1996). Furthermore, if we assume that weak proform shift is similar to clitic movement in languages like French, this hypothesis can be supported by the fact that third-person direct and indirect object clitics appear in the same order in French as in Dutch: Jean leDO luiIO donnera Jean will give it to him/her. The fact discussed earlier that weak object pronouns cluster together provides additional support for the hypothesis that they are clitic-like.

Subjects and direct objects of infinitival complement clauses in AcI-constructions are morphologically indistinguishable: this holds not only for referential noun phrases, but also for their pronominalized counterparts, both of which appear as object pronouns. Nevertheless, the examples in (194a&b) show that weak proform shift of an embedded object can optionally cross the subject of the infinitival clause; cf. Zwart (1996). Example (194c) further shows that this is in fact the preferred option when the subject of the infinitival clause is also realized as a weak pronoun. The acceptability of the inversion shows that the restriction established above, namely that weak proform shift of objects cannot affect the unmarked order of subjects and objects, only holds when the subject is assigned nominative case.

The examples in (195) show that weak proform shift of an embedded direct object can also cross the subject when the infinitival clause is ditransitive: the direct object pronoun must cross the indirect object and can optionally cross the embedded subject.

However, it seems that embedded weak indirect object pronouns cannot cross the subject of the infinitival clause: according to us, (196b) can only be interpreted with the pronoun ʼm as the agent and Els as the goal. It seems plausible that the fact that (196b) cannot express the intended reading is related to the fact that the agent and the goal are both [+human].

Something similar holds for cases in which both the direct and the indirect object appear as weak pronouns: examples such as (197b), which are given as perfectly acceptable in Zwart (1993/1996), are acceptable to us only if the pronoun ʼm is interpreted as the agent and Els as the goal. The unacceptability of (197c) deserves special mention, as it is unexpected in view of the fact that (194c) is perfectly acceptable; that the weak object pronouns apparently have to be adjacent may follow from the fact that they are clitic-like in the sense that they are obligatorily clustered.

Example (198a) shows that if all the arguments of a ditransitive infinitival clause appear as weak pronouns, they must occur in the order agent > theme > goal. Note, however, that some speakers find a sequence of three weak pronouns difficult to pronounce and therefore prefer the version in (198b) with a prepositional indirect object; as in (194c), the theme then preferably precedes the agent.

Finally, example (199) may suggest that weak proform shift can feed binding: while non-pronominal direct objects cannot bind a reciprocal indirect object, shifted direct object pronouns can. Since feeding of binding is generally considered to be a hallmark of A-movement, this may lead to the conclusion that weak proform shift is A-movement.

However, since weak proform shift can be preceded by A-scrambling (i.e. nominal argument shift), it is more likely that this preceding movement is responsible for feeding binding; cf. Haegeman (1993a/1993b). We conclude that weak proform shift of arguments is A'-movement, which Subsection III will support by showing that weak proforms that do not function as arguments can undergo a similar shift.

This subsection has shown that under a non-contrastive intonation pattern, weak object pronouns cannot be moved across nominative subjects. At first glance, this would suggest that weak proform shift cannot affect the unmarked order of nominal arguments (agent > goal > theme), but this turns out not to be true, as can be seen from the fact that weak direct object pronouns preferably precede nominal indirect objects, and that they can also be moved across an embedded subject in an AcI-construction. The fact that weak proform shift can affect the unmarked order of nominal arguments shows that weak pronouns can occupy positions in the clause that are inaccessible to their non-pronominal counterparts, which in turn supports the hypothesis that they are clitic-like. This hypothesis is also supported by the fact that weak object pronouns obligatorily cluster together.

Weak subject and object pronouns exhibit several properties often ascribed to clitics, e.g. the fact that they obligatorily cluster together. The examples in (202) also show that they cannot be used as independent utterances and cannot be topicalized or coordinated; cf. Haegeman (1993b) for a relevant discussion. On the other hand, Zwart (1996) notes that these properties also apply to the reduced pronouns in English, which behave syntactically like regular pronouns, and he concludes that they are not defining characteristics of clitics, but simply follow from the fact that weak pronouns cannot be accented.

One problem with the claim that Dutch weak pronouns are clitics is that they differ from undisputed clitics (as in e.g. French) in that they are not hosted by a verb. A related problem is that they can occur in PPs: bij ʼmwith him; cf. Haegeman (1993b). The hypothesis that Dutch weak pronouns are clitics thus requires that there be some (possibly phonetically empty) functional head to which they can cliticize. At present there seems to be no generally accepted analysis, but the tentative proposals in Haegeman (1993a/1993b) and Zwart (1993/1996) do agree that the prospective functional head(s) have nominal (i.e. case or agreement) features. We leave this claim for future research.

If the associate of quantitative er is a subject, a weak direct object pronoun must follow the associate, and even then the result is somewhat marked, which we indicate here by a question mark. This is shown in (221b) on the basis of the clause dat vier studenten het boek gelezen hebbenthat four students have read the book.

Example (222b) shows that the same applies to weak indirect object pronouns, on the basis of the clause dat twee studenten Peter het boek aangeboden hebbenthat two students have offered Peter the book. For completeness’ sake, the (c)-examples show that both the direct and the indirect object pronoun must follow the associate; although the primeless (c)-example itself is somewhat marked, the contrast with the primed ones is quite sharp.

Example (223a) shows that if the associate of quantitative er is a direct object, quantitative er can either precede or follow the indirect object. This is not entirely optional, however, as the (b)-examples show that the choice is partly determined by the surface position of the indirect object. Example (223b) shows that if the indirect object appears after the modal verb, then the shift of quantitative er is indeed optional, although it should be noted that the shift must cross the modal adverb. Example (223c) shows that if the indirect object has undergone nominal argument shift, then weak proform shift must also apply, although it can end either before or after the indirect object.

While the examples in (223) show that quantitative er can either precede or follow a non-pronominal indirect object, there may be a preference for it to follow weak indirect object pronouns, although Haeseryn et al. (1997:1321) considers both orders perfectly acceptable; note that the /d/ in (224b) is a linking sound inserted to break the sequence of two schwa’s. If the order preference is indeed significant, we should conclude that quantitative er behaves in this respect similarly to the French quantitative clitic en: cf. Je lui en ai donné uneI have given him one.

We conclude with cases in which the associate of quantitative er is an indirect object: constructing acceptable examples with a weak direct object pronoun is difficult, but it seems that the pronoun prefers to precede quantitative er. We illustrate this in (225) for the sentence dat ik twee studenten het boek heb aangebodenthat I have offered the book to two students.