Sections 39.1 and 39.2 have discussed two types of reduction processes that typically apply to coordinate structures: conjunction reduction and gapping. Some examples of these processes are repeated in (172). For convenience, we will use boldface to indicate material that is (presumably) deleted by conjunction reduction.

The structures in (172a&b) are based on the assumption that there are two types of conjunction reduction, the first applying forward and the second backward. In Section 39.1, however, we followed Neijt (1979) in arguing that backward conjunction reduction is the only genuine form of conjunction reduction; putative cases of forward conjunction reduction such as (172a) should be reanalyzed as cases of coordination of phrases smaller than clauses, as in (173).

Section 39.1, sub V, has shown that (supposed) forward and backward conjunction reduction can easily co-occur; cf. (174a). This is also to be expected in the proposed reanalysis of forward conjunction reduction. We have seen in Section 39.1, sub V, that backward conjunction reduction is not restricted to clausal coordinands, but can apply to a wider range of coordinate structures: the analysis in (174a) can thus be replaced by that in (174b).

Combinations of forward conjunction reduction and gapping are less likely to occur. So far we have assumed without discussion that examples such as Jan heeft haar een boek gegeven en hem een CDJan has given him a book and her a CD are derived by gapping, as in (175a); cf. Section 39.2, sub IB. If forward conjunction reduction could be seen as a reduction rule, we could in principle also derive this sentence by a combination of (i) gapping of heeft and the participle phrase gegeven (after scrambling of the direct object a CD) and (ii) forward conjunction reduction of the subject Jan, as in (175b).

On the other hand, the hypothesis that forward conjunction reduction involves coordination of phrases smaller than clauses rules out the derivation in (175b). The alternative structure in (176) is also ungrammatical, since gapping applies only to clausal coordinate structures; cf. Section 39.2. We conclude, therefore, that forward conjunction reduction and gapping do not co-occur, and that the structure in (175a) is the only one available. This is also desirable, because the target sentence is not ambiguous in meaning.

This leaves just one more case: the co-occurrence of backward conjunction reduction and gapping. Simple examples of gapping and backward conjunction reduction are given in (177a) and (177b), respectively, while example (177c) shows that the two reduction processes can also co-occur.

Although gapping and backward conjunction reduction can co-occur in examples such as (177c), various factors may conspire to block their co-occurrence in other cases. First, consider the cases of gapping in (178); the acceptability contrast between these two examples follows from the maximization requirement on gapping (cf. Section 39.2, sub IE), which prohibits gapping remnants identical to constituents in the antecedent clause (which, of course, can be derived from the A'-movement hypothesis proposed in Section 39.2¸ sub IID, according to which gapping remnants are prototypical contrastive topics/foci).

The maximization requirement does not hold in the case of backward conjunction reduction, as shown by the fact that both examples in (179) are acceptable.

It seems that gapping, as in (178a), can bleed backward conjunction reduction; because gapping obligatorily elides the participle gelezen, the right periphery of the first coordinand is no longer phonologically identical to the right periphery of the second coordinand, and this blocks backward conjunction reduction. This correctly predicts that the examples in (180) are unacceptable in the intended reading. The examples are also unacceptable because of a violation of recoverability, since all occurrences of gelezen are elided.

At the same time, it is not a priori predicted that backward conjunction reduction will block gapping. If we were to apply gapping to the examples in (179), we would derive the examples in (181). Although these examples are more intelligible than those in (180), which is obviously related to the fact that in (180) the main verb gelezen is omitted in both coordinands and thus not recoverable from the context, they are clearly degraded compared to those in (179). We assign two question marks to the examples in (181) to do justice to their relative acceptability. Again, boldface is used to indicate conjunction reduction.

The unexpectedly degraded status of examples like those in (181), brought to our attention by Anneke Neijt (p.c.), can be used to support our earlier conclusion that gapping and backward conjunction reduction are different in that the former is a regular syntactic rule (cf. Section 39.2, sub IIA), while the latter is a post-syntactic rule (cf. Section 39.1, sub IV). This implies that gapping must precede backward conjunction reduction, so that the examples in (181) cannot be derived for the simple reason that the conjunction reduction structures in (179) cannot be the input for gapping. It should be noted, however, that judgments seem to vary from speaker to speaker; Van Oirsouw (1987:120), for example, judges the (a)-examples in (182) to be acceptable without any reservations, and Haeseryn et al. (1997:1595) does the same with the somewhat more complex (b)-examples.

Although we consider the examples in (182) to be like those in (181) in that they are marked in comparison to the corresponding forms without gapping of the finite verb, the fact that Van Oirsouw and Haeseryn et al. consider them acceptable shows that determining their exact status is not an easy task. This means that the acceptability judgments on examples like (181) and (182) are simply not clear enough at this stage to be used to evaluate the claim that backward conjunction reduction is a post-syntactic rule that cannot precede gapping. We hope that future research will be able to shed more light on this issue.