Examples

Our account of the various word order possibilities within the German verb cluster rests on two assumptions. First, all argument-inheriting verbs select a verbal complement that must be in the izone. Second, the relative order of elements in the izone is determined by specifying the position of a complement relative to its governor. The second part of our account follows the proposals made in [7].

We start by presenting lexical entries for main verbs, nonfinite modals, and auxiliaries which occur in `flipped' position.

As in German it is the governed verb, rather than the governor, which determines whether `flipped' word orders are possible, we must assume that (main) verbs come with a specification for GVOR. Thus it is the complement, rather than the head or governor, that determines word order. German main verbs, for instance, do not `trigger' flipped word order, and thus are specified as [GVOR $\rightarrow$]:

Note also that transitive verbs select an NP-complement that is part of the ozone.

As a modal verb may trigger auxiliary flip, it is unspecified for GVOR. At the same time, modals require that their verbal izone complement must precede it. Note that this requirement is not redundant: if a modal governs another modal (where the governed modal will be unspecified for GVOR), we need this constraint to block `flipped' word orders such as ^* kann singen müssen:

A finite auxiliary such as wird places no constraint on the GVOR value of its verbal izone complement, and thus allows for both normal and flipped word order:

Given these lexical entries, and the HEAD-COMPLEMENT schema given in (22), we can derive exactly the three word orders given below:

The examples above illustrate that our account predicts the possibility of auxiliary flip and Zwischenstellung, if the auxiliary governs a modal verb. If the auxiliary governs a main verb, flipped word order is ruled out.

Two more issues need to be mentioned. First, for those speakers that do not allow the Zwischenstellung word order in (32), we assume, following Kathol, that the GVOR specification on main verbs is [GVOR $\hookrightarrow$] instead of [GVOR $\rightarrow$]. The restriction that the governor of a main verb must be right-adjacent rules out the derivation in (32), but not the derivations in (30) and (31).

Second, if a verb cluster contains two auxiliaries, the ordering possibilities are as follows:

The conclusions to be drawn from these data are that (i) the status of haben as a trigger for flipped word order is determined by the verbal complement that is governed by haben (i.e. if the governee is a main verb (33), haben is not a trigger, whereas if it is a modal (34), haben may trigger flipped word order) and (ii) that haben may itself occur in flipped position if and only if its governing auxiliary is too (34b,c,d). These observations lead to a lexical entry for nonfinite auxiliaries in which there is a reentrancy between the value of GVOR of the auxiliary and the value of GVOR on the verbal complement:¹²¹³