... complex.¹

In this paper we use verb cluster as a descriptive term for the sequence of verbs which is typically found in clause-final position in German and Dutch subordinate clauses. The verbal complex is a theoretical notion, and refers to the phrasal category used in some grammars to analyze verb clusters.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... below:²

denotes the list which is obtained by appending the lists

and

.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree:³

In the examples below, VC is a verbal complex, VP is a (partial) verb phrase, and S is a verb phrase including a subject.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... infinitive).⁴

I.e. as an infinitive which occurs as the complement of an auxiliary selecting a past participle verbal complement.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... statements.⁵

This aspect of the analysis is inspired by [5].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...hpsg2: ⁶

As well as a number of other principles introduced in [13], such as the HEAD FEATURE and NONLOCAL FEATURE principle. For simplicity, we are silent about other valence features, since subjects or specifiers do not play any significant role in the following examples.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... derived.⁷

Note, however, that we do impose the requirement that the selected complements must form a suffix of COMPS on the head daughter. This implies that the order of complements on COMPS (i.e. obliqueness) does constrain the type of partial phrases that can be derived. In this respect, our proposal differs from that of [1].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... however.⁸

See [16] for an approach to the ordering of constituents in the Mittelfeld. In such an approach, obliqueness could be one of the several competing constraints determining word order.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

....⁹

Kathol adopts a somewhat different definition. Since he assumes a valence feature VCOMPL, on which verbal complements are represented and on which there is no inheritance, and since only the governors of verbal complements need to be identified, it is possible to define the governor of a verbal complement as the sign on whose VCOMPL-list this complement appears.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...]:¹⁰

This specification may seem redundant for cases where the modal governs a main verb, but is crucial for cases in which a modal governs another modal or auxiliary verb (i.e. this constraint will rule out ^* kann singen müssen).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... itself:¹¹

Note that this analysis predicts that example (41)b is ungrammatical. A reviewer claims that this example in fact is grammatical. That would constitute a problem for the analysis of [4], and also for our own analysis.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... grammar.¹²

One might object that the feature ZONE is encodes a distinction between two topological fields that is relevant to VPs only. This is not necessarily an objection, for if non-verbal heads only select [ZONE outer] complements, the zone LP constraints are satisfied trivially, while the zone principle will require that the head daughter must be lexical. On the other hand, it is not impossible that the distinction between I-ZONE and O-ZONE complements might play a role for non-verbal heads as well (i.e. to distinguish clitic-like elements from other complements).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... fronting.¹³

See [10] for an alternative monotonic proposal.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... follows:¹⁴

We assume that finite verbs include a subject as least oblique element on their COMPS-list.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... follows:¹⁵

The adjunct schon is taken to be a modifier of the verb können and is added to its COMPS-list by means of a lexical rule (see [17]). In order to account for verb-second, we could follow [12], by assuming that main clause word order is obtained by a rule schema which lets a finite verb combine with a clause with an empty verbal head daughter. The COMPS feature of the finite verb (as well as some other features) must unify with that of the empty verb. Alternatively, it is also possible to analyse initial verb-placement by extending the linear precedence component. Notice that in such an approach we should be careful to limit the scope of the linear precedence statements given here.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... verbs.¹⁶

Infinite verbs in Dutch are often preceded by te:

It may be worthwile pointing out that we treat te as an inflectional marker, and not as a separate word: unlike to in English, no material (not even separable prefixes) can ever intervene between te and the infinitive. Therefore, te haten is of type WORD.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.