Complement clauses differ from relative clauses in that only the latter contain an interpretive gap that is “filled” by the head of the nominal construction. This gap is the result of the movement of the relative pronoun into the initial position of the clause, and will be referred to as trace (t). The relative pronoun takes the NP (the head noun and its optional modifiers) as its antecedent, which is therefore taken to fill the interpretive gap in the clause; this allows the relative clause to provide additional information about the denotation of the NP. To clarify, in example (663) the relative pronoun datthat originates as the direct object of the relative clause and is moved into the initial position of the relative clause, while leaving the trace ti in its original position. The relative pronoun takes the NP feit as its antecedent, which is expressed by co-indexing. Consequently, feit is interpreted as the direct object of the relative clause, and as a result the modified NP denotes a subset of the set of facts, namely those accepted by everyone.

Complement clauses, on the other hand, simply specify the content referred to by the noun phrase headed by a proposition or speech-act noun. There is no interpretive gap in the complement clause: no part of the complement clause is coreferential with the nominal head, which therefore plays no role in the interpretation of the complement clause.

For completeness, note that the generalization that complement clauses do not contain an interpretive gap holds only for declarative clauses. Interrogative complement clauses introduced by a wh-word do of course contain a trace of the wh-word moved into the initial position of the dependent clause, but they differ crucially from relative clauses in that the wh-word does not take the head noun as its antecedent.

Both complement and relative clauses contain an element linking them to the nominal head of the noun phrase: this linker takes the form of a complementizer in complement clauses, whereas in relative clauses the linker is a relative pronoun taking the NP (the head noun and its optional modifiers) as its antecedent. In many cases the form of the linker will reveal the status of the subordinate clause, but in some cases the complementizer and the relative element may have the same form.

There are also distributional differences between complement clauses and relative clauses. These differences arise from the fact that relative clauses contain a relative pronoun which requires an antecedent, whereas complement clauses do not depend on the noun in the same way. As a result, complement clauses are freer in their distribution: for example, they can function as the subject or the object of a verb, as in (676b).

They can even be used as the predicate in a copular construction, in which case they are predicated of a noun phrase headed by a proposition or speech-act noun, as in (677). This need not surprise us, since the nominal head denotes the same abstract entity as the clause.

Relative clauses, on the other hand, never occur independently: the clause contains a relative pronoun which needs an antecedent and consequently the relative clause in (678a) can neither be used as an argument nor as a predicate of a copular construction. This is demonstrated by (678b&c) and (678d), respectively.

Another difference between complement clauses and relative clauses stems from the different communicative functions they fulfill. Finite complement clauses express the content of a proposition or speech-act noun. As such, they can be said to be uniquely determining; there is only one fact, assumption, question, request, etc. with the particular content specified in the complement clause. This can be supported by the fact that the adjectives interessant and triviaal in (679) can only be used on a non-restrictive (purely property-assigning) reading.

Another piece of evidence supporting this assumption is that NPs with a complement clause cannot be modified by a superlative, since these presuppose a non-singleton set. Note in passing that Dutch lacks the non-superlative interpretation available for the English translations in (680), which amount to “extremely interesting fact/trivial assumption” or “the most interesting fact/trivial assumption possible”, and which do not involve a selection from a presupposed set, but a non-restrictive assignment of a property.

The use of the comparative form is possible, but only if the comparison involves some other fact. In (681a), for instance, the fact referred to is compared with some other, possibly contextually evoked, fact, which is claimed to be less interesting; likewise, the sentence in (681b) is acceptable only in relation to some other, less trivial assumption.

Relative clauses have no such restrictions. Both the superlative and the comparative forms of the adjective can be used in their selective/comparative function, while adjectives can be used both restrictively and non-restrictively. Example (682a), for instance, implies that there is a larger set of facts, the most interesting of which is that there is life on Mars, and (682b) makes a comparison between two facts, the more interesting of which is the one mentioned. In (682c) the adjective interessant is used contrastively: a set of two facts is implied, one interesting, the other uninteresting. Note that a non-contrastive, non-restrictive reading of the adjective is also possible; in this case, the fact in question is simply assigned the property of being interesting.

A final difference between the two types of clause can be explained along the same lines as the previous one: due to the fact that the content of complement clauses serves to uniquely determine the entity referred to by the noun phrase, they can only be used in combination with the definite article (provided that the complement clause is the only modifying element); the use of the indefinite article yields an unacceptable result. This is demonstrated in (683).

Relative clauses, on the other hand, easily accept both the definite and indefinite article, as in (684).

Proposition nouns followed by infinitival complement clauses introduced by om differ from those followed by a finite complement clause in that they can be modified by an adjective and do accept the indefinite article. Some examples are given in (685).

One possible explanation for this contrast with finite complement clauses is that om-clauses, strictly speaking, do not specify the content of the head noun, but rather the purpose or cause of the action or emotion expressed by the proposition noun. As a result, the relation between the proposition noun and the complement clause need not be uniquely determined: there may be different ways of requesting to be admitted or ordering a person to leave. In (685a), for instance, a particular kind of request is referred to: the kind that is intended to obtain admission. However, the exact form of the request is not specified.

An alternative explanation might be that om-clauses are always non-factual, specifying requests, orders, wishes, possible situations and the like (cf. Section 16.3.2). Again, this means that although the complement clause is certainly used to specify the proposition noun, noun phrase and clause do not share their reference. That something like this might be on the right track also follows from the fact that the use of infinitival complements as the predicate of a copular clause is marked compared to the perfectly acceptable examples in (677) discussed in C, which involve finite complement clauses.

Finally, note that the infinitival clause in (686a) can also be interpreted as a purpose clause. This is related to the fact that the om-clause in (687) can also be interpreted either as the complement of the noun or as an adverbial phrase indicating purpose. The fact that om-clauses in sentence-final position are typically interpreted as purpose clauses may well affect the judgments on (686). This concludes our discussion of complementation of nouns.