Main clauses with an object and a clause-final verb can have different accents. We take the sentence accent to be located at the end of the clause and to involve a sudden lowering in pitch, which means that we adopt a more restrictive definition of sentence accent than some of the references given below. It seems relatively uncontroversial that the sentence accent (in our sense) is usually located within the lexical domain of the clause, in some phrase preceding the clause-final main verb; cf. Baart (1987), Gussenhoven (1992), Booij (1995) and the references cited there. This observation has found a syntactic explanation in Cinque’s (1993) hypothesis that stress prominence is a reflection of depth of embedding: the default location of the sentence accent is the most deeply embedded constituent that can carry a word accent in the syntactic surface structure of the clause or, as a possibly better alternative, a prosodic structure derived from it by the elimination of phonetically empty nodes, as proposed in Baart (1987). This means that the sentence accent must be placed on the object, provided that it is within the lexical domain. The examples in (21) show that the proviso is indeed necessary, because leftward movement of the object into the functional domain results in deaccenting the object; cf. Verhagen (1986). Note that sentence accent is indicated by small caps.

We can illustrate the same with the examples in (22) using the particle verb uitnodigento invite; we follow the hypothesis in Section 2.2 that the object and the verbal particle form a small clause. The default placement of the sentence accent in (22a) is on the noun zuster, because this is again the most deeply embedded element with word/phrase accent. Example (22b) shows that leftward movement of the object to a position external to the lexical domain causes the sentence accent to shift to the particle, because this particle is now the most deeply embedded constituent in the resulting structure.

Further support for Cinque’s hypothesis that the default placement of the sentence accent is on the most deeply embedded constituent in the clause is provided in (23): example (23a) shows that the sentence accent is realized on the most deeply embedded phrase within the object, and (23b) shows that the sentence accent must be realized on the complementive if it is complex, since the nominal complement of the preposition phrase in de vaas is more deeply embedded than the subject of the small clause, bloemenflowers.

A final piece of evidence for Cinque’s hypothesis is provided in (24), which shows that the location of sentence accent depends on the syntactic function of the phrase preceding the verb. The PP in (24a) functions as a prepositional object, i.e. a complement of the verb, and this correctly predicts that the default placement of sentence accent is on the nominal complement of the PP, since this is the most deeply embedded phrase. The PP in (24b) functions as an adverbial phrase and is thus external to the VP, i.e. higher in the structure than the clause-final verb; this correctly predicts that the sentence accent is on the participle. Since the complementive PP in (24c) is part of the VP, it is again correctly predicted that the sentence accent is realized on the nominal complement of the PP; cf. also Gussenhoven (1992).

The previous subsection has described Cinque’s rule for deriving neutral intonation patterns: the sentence accent is assigned to the most deeply embedded phrase within the lexical domain of the clause that can carry a word accent, which is prototypically an object. Clauses with a neutral intonation pattern are often ambiguous with respect to the focus-presupposition division: the new-information focus may be restricted to the clausal constituent to which the sentence accent is assigned, but it may also extend to include larger projections of the clause containing it. In the examples in (25), for instance, the new-information focus can be restricted to the direct object, but it can also be extended to include the (particle) verb; that this extension is possible is clear from the fact that these sentences can be used as answers to the question Wat heeft Jan gedaan?What has Jan done?. The alternatives in (25) thus differ in the scope of the new-information focus, which is indicated by the underlining.

Clauses that deviate from the prototypical assignment of the sentence accent can arise in two different ways, both of which have implications for the information structure of the clause. First, the element that would normally be assigned the sentence accent can be removed from the lexical domain of the clause, which results in the sentence accent being assigned to the next most deeply embedded element in accordance with Cinque’s rule. Example (26) shows that the information-structural effect of the leftward movement of the objects in (25) is that they can no longer be interpreted as part of the new-information focus, and must therefore be interpreted as part of the presupposition of the clause. cf. Section 13.2 for a detailed discussion.

Another way to derive non-neutral intonation patterns, which will be discussed in more detail in Section 13.3, is to simply ignore Cinque’s rule. The examples in (27) show that this also results in a more restricted focus domain. The primeless examples in (27) have a neutral intonation pattern with the sentence accent on the most deeply embedded phrase and they can be interpreted in such a way that all phrases within the lexical domain (VP) are part of the new-information focus of the clause. The primed examples, on the other hand, have a marked main accent on a phrase higher in the structure, and this triggers a so-called contrastive reading: the contrastively accented phrase (indicated in italics) is taken as the relevant discourse-new information, while the remainder of the lexical domain is construed as (familiar) background information. A contrastive intonation pattern is often used to correct information given earlier in the discourse or to exclude alternative possibilities; we have indicated this in the English rendering by adding the part in brackets.

The same can be observed in examples such as (28): the sentence accent in the primeless examples is assigned to the most deeply embedded phrase within the lexical domain, and this allows an interpretation according to which the entire lexical domain is part of the new-information focus of the clause. Placing the accent on another element within the noun phrase/small clause, as in the primed examples, again leads to the more restricted contrastive reading; cf. Cinque (1993: §6) and Booij (1995:159), among many others.

Note that we have used different typographical means to indicate the accents in (27) and (28): regular small caps for the default sentence accent and small caps in italics for the contrastive accent. The reason for this is that the two accents are not identical, as can be seen from the fact that the contrastive accent can also be assigned to phrases that would normally be assigned the default sentence accent. The result of using contrastive accent instead of regular sentence accent is again that the new-information focus is narrowed: while the verb can be part of the discourse-new information under a neutral intonation pattern, as in (29a), this is not possible when the contrastive accent is used, as in (29b). The two accents in (29) differ phonologically in that the contrastive accent has an additional high tone.

Finally, it should be noted that contrastively accented phrases are often moved: the examples in (30) show that the unmarked order of the direct and prepositional indirect object can be optionally reversed when the latter is given a contrastive accent. This will be the main topic of Section 13.3.

The previous subsections have shown that the default placement of sentence accent is on the most deeply embedded constituent that can carry a word accent in the surface structure of the clause (or, alternatively, in a prosodic structure derived from it by eliminating the phonetically empty nodes). The default sentence accent allows for an interpretation of the entire lexical domain of the clause as a new-information focus, while the alternative placements of the main accent lead to a more restricted focus interpretation. The discussion has been restricted to main clauses with at least one object because this allows us to leave aside some complex questions about subject accentuation that do not immediately concern us here. For example, subjects in clause-initial position typically function as an aboutness topic or a contrastive topic/focus, and are therefore also marked with a special accent (cf. Section 11.3.3, sub IV), giving rise to the so-called intonational hat contour found in many Dutch declarative main clauses. In question-answer pairs such as (31b), the selection of the new-information focus can be determined in the usual way from the position of the sentence accent.

It has been noted, however, that certain simple monadic constructions with a single accent on the subject can be interpreted as “all new-information” focus; this is illustrated by the question-answer pair in (32). This contradicts Cinque’s (1993) hypothesis that stress prominence is a reflection of depth of embedding, while it can be explained by Baart’s (1987) earlier proposal that new-information focus is always projected from one of the verb’s arguments.

We will not digress on cases such as (32b), because the accent in (32b) may be different from the default sentence accent, and the phenomenon is, for reasons not well understood, restricted to simple monadic constructions; cf. Verhagen (1986), Baart (1987), Gussenhoven (1992), Cinque (1993) and the references cited there for extensive, sometimes conflicting, discussions of such cases.