This section discusses the negative adverb nietnot and its affirmative counterpart wel in (61). Note that the adverb niet can also be used as constituent negation (cf. Section 13.3.2, sub IC), and that both niet and wel can also be used as a kind of degree modifier of adjectives; Jan is niet onaardig\`1Jan is wel aardig Jan is quite nice (cf. Section A25.3). These uses will not be discussed here.

Polarity adverbials are clearly not VP adverbials, as shown by the fact that the sentences in (61) do not pass the two VP-adverbial tests. The primeless examples in (62) first show that the pronoun doet dat + adverb paraphrase does not lead to a felicitous result: the left-right arrow with a slash (⇎) indicates that it leads to a contradiction in the case of nietnot and the left-right arrow without a slash (⇔) indicates that it leads to a tautology in the case of wel. The primed examples also show that the entailment test also fails: the entailment holds in neither direction in the case of niet and in both directions in the case of wel (at least as far as the meaning expressed by traditional predicate calculus is concerned).

Polarity adverbials take scope over the proposition expressed by the lexical domain of the clause. This is the standard assumption for negation in predicate calculus, which treats negation as an operator taking scope over a well-formed expression Ф: ¬Ф. It is also clear from the fact that both negative and affirmative clauses pass the scope test in (60a): the examples in (61) can easily be paraphrased by the examples in (63).

The polarity adverbials are very low in the functional domain of the clause, as shown by the fact that they must be preceded by all the clause adverbials that will be discussed in the following subsections. This immediately shows that these other adverbials are also part of the functional domain of the clause and thus cannot function as VP adverbials; cf. Section 8.1, sub II.

The negative adverbial niet is probably not in an adjoined position, but in the specifier of a functional projection (NegP): the reason for this assumption is that this position is not only accessible to niet but arguably also functions as a landing site for negative phrases. This is especially clear when the negative phrase is part of a PP-complement of a complementive adjective, as in (64): while there is good reason to assume that the PP is base-generated in a position following the adjective, it must occur in a position preceding the adjective if the nominal part of the PP is a negative phrase such as niemandnobody. This would follow if we assume that a negative phrase must be moved into the specifier of NegP, as indicated in (64c), in order to allow negation to be take scope over the whole proposition; cf. Section 13.3.1 for a detailed discussion of negation movement.

Finally, note that the semantic contributions of the two polarity adverbials are quite different: from a logical point of view, the negative adverbial niet is needed to express negation (unless it is expressed in some other way), whereas the affirmative marker is superfluous. This is demonstrated in (65): omitting niet results in an affirmative expression, whereas omitting wel results in a logically equivalent expression.

It is therefore not surprising that the use of the affirmative marker wel is mainly pragmatically motivated: it is used to indicate contrast, to deny an assertion or a presupposition held by the hearer, to make a concession, and so on. Illustrations are given in (66).

The affirmative marker wel thus plays a prominent role in signaling that the background (the shared information of the discourse participants) needs to be updated, and this heavy information load may be the reason why affirmative wel is always accented (unlike the downtoning degree modifier wel in Hij is wel aardigHe is rather nice, discussed in Section A25.3.2, which is never accented).

Sentence negation can be preceded by focus particles such as alleenjust/only, ookalso, and zelfseven. A number of typical examples are given in the primeless examples in (67). That these particles function as clause adverbials is clear from the fact that they pass the scope test in (60a), as shown in the primed examples.

As in the case of negation, there are reasons to assume that focus particles are not in an adjoined position, but in the specifier position of a functional projection (FocusP). To show this, first note that focus particles are not only used as independent adverbials, but can also be used as narrow focus markers, in which case they form a constituent with the focused phrase. This can be seen in the examples in (68); the fact that the particle and the focused phrase co-occur in clause-initial position shows that they must be a constituent (cf. constituency test).

The reason for assuming that the focus particles are in the specifier of FocusP is that this position is not only accessible to focus particles; it also functions as a landing site for narrowly focused phrases. This is especially clear when the focused phrase is a PP-complement of a complementive adjective, as in (69). It is uncontroversial that the PP is base-generated in a position following the adjective; however, it must precede the adjective if it is narrowly focused. This would follow if we assume that narrowly focused phrases are moved into the specifier of FocusP, as indicated in (69c), in order to be given scope over the backgrounded part of the clause.

We refer the reader to Section 13.3.2, sub IC, for a more detailed discussion of focus movement, which includes a wider range of focus particles.

Sentence negation can also be preceded by aspectual adverbs like habitual gewoonlijkusually, continuative nog (steeds)still, terminative niet meerno longer, iterative weeragain, and frequentative vaakoften. Other adverbials that may belong to this group are alalready and spoedigsoon but these do not easily co-occur with the clause adverbial niet. Some cases are given in the primeless examples in (70); the primed examples show that these adverbials pass the scope test in (60a).

Note that the frequency adverb vaakoften can also be used as a VP adverbial; cf. Section 8.2.1, sub IIIA. The primeless examples in (71) illustrate this by showing that it can either precede or follow the negative adverb nietnot. The two examples differ in the relative scope of the adverbials vaak and niet, as can be seen from the paraphrases in the primed examples.

The difference in scope becomes even clearer with frequency adverbials such as drie keerthree times. Suppose we have a series of four lectures; then example (72a) expresses that Jan attended less than three meetings while (72b) expresses that Jan attended only one lecture. Example (72c) shows that the two uses can occur in the same clause: if we are dealing with six series of four lectures, (72c) expresses that for two of these series Jan attended less than three lectures.

A class of adverbs that can perhaps also be considered aspectual are the adverbs helemaalcompletely and gedeeltelijkpartly in (73a), since they indicate whether the eventuality was completely finished or not. That these adverbs are not VP adverbials is clear from the fact that they do not restrict the denotation of the verbal predicate; this can be seen from the fact that the entailment test in (73b) fails in the case of gedeeltelijk. However, it is not obvious either that these adverbs function as clause adverbials, because the scope test in (73c) yields questionable results.

Nevertheless, there are good reasons to think that we are dealing with clause adverbials, since the adverb gedeeltelijkpartly can precede sentence negation; cf. (74). Note that the order niet gedeeltelijk is also possible when the adverb is accented; this case can be left aside because we are probably dealing with constituent negation in this case.

Examples similar to (74) are also difficult to construct for helemaal, because this adverb is then construed as a modifier of the negation in helemaal nietabsolutely not, i.e. it functions similarly to at all in the English translation of Jan heeft de film helemaal niet gezienJan has not seen the movie at all.

Adverbs like bijnaalmost, echtreally, and haastnearly are called clause-degree adverbs in Ernst (2002). These are clear cases of clause adverbials: they pass the scope test, as shown in the primed examples in (75).

It may be that (inherently negative) adverbs like amperhardly and nauwelijksscarcely in (76a) belong to the same class, although (76b) shows that they do not pass the scope paraphrase in a convincing way. We leave this as a problem for future research.

Propositional modality provides an evaluation of the factual status of propositions expressed by the lexical projection of the main verb. By uttering a sentence such as (77), the speaker asserts/commits to the truth of the proposition be home (Marie).

However, the speaker may also comment on the factual status of the proposition. Palmer (2001) claims that these judgments can be of two different kinds: there are epistemic and evidential judgments. Epistemic judgments are concerned with the likelihood of a particular eventuality actually occurring. Section 5.2.3.2, sub IIIA1, has shown that epistemic judgments can be expressed by modal verbs such as kunnenmay, moetenmust and zullenwill.

By uttering the sentences in (78a-c), the speaker makes three different epistemic judgments about (his commitment to the truth of) the proposition be home (Marie). The modal verb kunnenmay presents the proposition as a possible conclusion: the speaker is uncertain whether the proposition is true, but he cannot exclude it on the basis of the information available to him. The modal verb moetenmust presents the proposition as the only possible conclusion: on the basis of the information available, the speaker concludes that the proposition is true. The modal verb zullenwill presents the proposition as a reasonable conclusion on the basis of the available evidence. A wider range of epistemic judgments can be expressed by the adverbial phrases in (79a).

Evidential judgments are concerned with the source of information on which the judgment is based: cf. Section 5.2.3.2, sub IIIA2. For example, perception verbs such as ziento see are used in AcI-constructions such as Ik zag Peter vertrekkenI saw Peter leave to express that the evidential judgment is based on direct sensory evidence: the speaker was an eyewitness to the eventuality. And modal verbs such as blijkento turn out, lijkento appear, and schijnento seem indicate whether there is direct evidence for the truth of the proposition, whether there are identifiable persons who can be held responsible for the truth of the proposition, or whether we are dealing with hearsay/rumors; cf. Vliegen (2011).

Again, a wider range of evidential judgments can be expressed by the adverbial phrases in (81a):

The propositional modal adverbials in (79a) and (81a) pass the scope-test in (60a), as shown in (82) for the examples in (79b) and (81b). That epistemic modal adverbials allow the scope paraphrase is consistent with the fact that epistemic judgments are expressed in formal logic with the operators □ and ◊, which take scope over the proposition p: □p and ◊p.

Subject-oriented adverbials like slim genoegcleverly and wijselijkwisely in (83) provide the speaker’s subjective evaluation of the subject of the clause in relation to the predicate expressed by the lexical projection of the verb. Example (83a) expresses that the speaker considers Jan clever for not attending the performance and (83b) that he considers Marie wise for not contradicting Peter.

Example (83b) shows that subject-oriented adverbials can precede negation. The fact that the reverse order leads to a marginal result also suggests that they function as clause adverbials. Further support comes from the fact that they do not restrict the denotation of the predicate; (84) clearly shows that the examples in (83) cannot be paraphrased by a conjoined pronoun doet dat + adverb clause. Note in passing that the paraphrase Jan vertrok en hij deed dat slim genoeg voor de voorstelling is acceptable, but here the subject-oriented adverb slim genoeg still has scope over the time adverbial and may be applied to the subject pronoun of the second conjunct. The acceptability of this paraphrase is therefore not relevant here; cf. Section 8.1, sub III, for discussion.

The examples in (85) show that scope paraphrases are not possible either. This is not surprising, however, because the matrix clauses in these paraphrases do not contain a suitable subject to which the adverbial could be applied: the paraphrases are uninterpretable as a result.

Note that the examples in (83) can be paraphrased as in (86), suggesting that subject-oriented adverbials do have scope over the proposition expressed by the lexical domain of the clause. We will not pursue this idea further, but will tentatively assume that the infelicity of the scope paraphrases in (85) is indeed due to the fact that the matrix clauses do not have a suitable subject to which the adverbial could be applied.

Subjective adverbials indicate a particular mental attitude towards the state of affairs referred to by the clause. These adverbials are difficult to distinguish from epistemic adverbials because they also comment on the factual status of the proposition by expressing that the proposition is or is not necessarily/yet true.

However, the main informational load of these adverbials is a subjective evaluation of the eventuality. By uttering (88a) the speaker expresses that the proposition expressed by the clause is true, while the two adverbials gelukkig and helaas reveal that the speaker has either a positive or a negative attitude towards the eventuality of Jan having arrived on time. By uttering (88b) the speaker expresses that he does not know whether the proposition expressed by the clause is true, but that he would consider it a good thing if it were true.

Example (88b) is clearly not epistemic, since the speaker does not provide an evaluation of the factual status of the proposition. This is different with adverbial phrases such as naar ik hoop/vrees in (89): these adverbials are subjective in that they provide an evaluation of the proposition, but they are also epistemic in that the speaker expresses that the proposition is a reasonable conclusion on the basis of the evidence available to him. Since the epistemic verb vermoedento suspect can also be used in this phrase, it is not obvious that the adverbial phrase naar ik +V should be regarded as intrinsically subjective in nature.

That subjective adverbials are clause adverbials is clear from the fact that they pass the scope test; this is illustrated in (90a&b) for the examples in (88). For completeness’ sake, we have added the paraphrase in (90c) for the examples in (89).

Other examples of subjective adverbials are toch, maar, dan and nou. These particle-like items often occur in combination and can express different, often subtle, meaning modulations of the sentence; cf. Haeseryn et al. (1997;457/1349).

In the linguistic literature on English since Jackendoff (1972), epistemic adverbials have been classified as speaker-oriented. The epistemic judgments of the proposition are usually taken to be the speaker’s; by uttering the sentence in (91), the speaker takes responsibility for the truth of the assertion that Jan will visit us.href='S'>href='peaker-'>href='oriented'>

Although the speaker-oriented reading of epistemic adverbials is certainly their default interpretation, it is not semantically determined, but the result of a pragmatic implicature. This is evident from the fact that the speaker’s responsibility for the truth of the assertion can be made explicit or canceled by adding an adverbial phrase indicating the person responsible for the truth of the relevant information: some more or less fixed expressions for emphasizing or cancelling the speaker’s responsibility are given in (92).

A common productive way of expressing a point of view is to use a PP headed by the preposition volgensaccording to: by using volgens mijaccording to me in (93a) the speaker makes his responsibility for the truth of the assertion explicit, while he shifts this responsibility to Els by using volgens Els in (93b). Example (93c) shows that point-of-view adverbials pass the scope test.

Subjective adverbials like gelukkigfortunately and helaasunfortunately are also usually considered to be speaker-oriented. This may seem justified in (94): the assessment of Jan’s dismissal as a fortunate event can be attributed to the speaker despite the presence of the point-of-view PP volgens Elsaccording to Els. However, it is also possible to attribute this assessment to Els, as can be seen from the fact that the part in brackets with the pronoun ikI referring to the speaker can be used without creating a contradiction. This means that the speaker-oriented reading of evaluation adverbials is probably a pragmatic effect; we leave this to future research.

Spatiotemporal adverbials are used not only as VP adverbials (cf. Section 8.2.1, sub III), but also as clause adverbials. That temporal adverbials can be ambiguous in this way is shown in (95); the primeless examples show that these adverbials can either precede or follow a modal adverb such as waarschijnlijkprobably, and the primed examples show that they can pass both the clause-adverbial test and the VP-adverbial test, depending on their position in the clause.

The examples in (96) show that the two time adverbials morgen and om drie uur in (95) can co-occur, but that they obey certain ordering restrictions: the time interval referred to by the clause adverbial includes the time (interval) referred to by the VP adverbial. Since (96b) becomes perfectly acceptable if one of the two time adverbials is omitted, it is unlikely that we are dealing with a syntactic restriction; Section 8.2.3 will argue that this restriction is semantic in nature, which is why we have marked the deviating order in (96b) with a dollar sign.

For locational adverbials we can make more or less the same observations. The examples in (97) first show that locational adverbials can either precede or follow a modal adverb waarschijnlijk, and that they pass both the clause-adverbial test in (97a') and the VP-adverbial test in (97b'), again depending on their position in the clause.

The examples in (98) show that the two locational adverbials in (97) can co-occur, but that they obey certain ordering restrictions: the location referred to by the clause adverbial includes the location referred to by the VP adverbial. Since (98b) becomes perfectly acceptable if one of the two locational adverbials is omitted, it is again unlikely that we are dealing with a syntactic restriction, which is why we have marked the deviating order in (98b) with a dollar sign.

Section 8.2.1, sub IV, has shown that adverbials indicating cause and reason can be used as VP adverbials. However, the fact that these adverbials can occur on either side of the modal waarschijnlijkprobably, as shown in (99), suggests that they can also be used as clause adverbials.

That the adverbials indicating cause or reason in the primed examples in (99) are clause adverbials is further supported by the fact that these examples can easily be paraphrased by the examples in (100).

The semantic difference between the primeless and primed examples in (99) is actually a matter of relative scope: in the primeless examples, the adverbials indicating cause and reason are in the scope of the modal adverb waarschijnlijk, while they are not in the primed examples. This leads to the following differences in meaning: example (99a) expresses that the pot has probably cracked because of frost, while (99a') expresses that the frost is a good reason to assume that the pot has cracked; example (99b) expresses that the shop is probably closed because of Easter, while (99b') expresses that Easter is a good reason to assume that the shop is closed.

The concessive counterparts of the cause/reason adverbials can also be used as clause adverbials; the examples in (101) illustrate this by showing that these adverbials can easily precede the modal waarschijnlijkprobably.

Conditional adverbials seem to differ from adverbials indicating cause, reason, or concession in that they always function as clause adverbials. Although conditionals are usually expressed by clauses, there are also a number of more or less idiomatic prepositional phrases headed by inin and bijwith; cf. Haeseryn et al. (1997:1209). Two typical cases are given in (102a&b). These conditional adverbials do not restrict the denotation of the verbal predicate, as can be seen from the fact that they do not pass the entailment test in the singly-primed examples. Moreover, the scope paraphrase in the doubly-primed examples seems perfectly adequate. This means that conditional adverbials differ from adverbials indicating cause, reason, or concession in that they function only as clause adverbials, regardless their form.

That conditional adverbials cannot be used as VP adverbials is due to the fact that they function as the antecedent P of a material implication P → Q, not as a restrictor of Q: we can only conclude that proposition Q is true if proposition P is also true. For the same reason, we can conclude that the conditional clause in (103a) must function as a clause adverbial. Although this cannot be proved on the basis of the entailment test, we should probably conclude the same for its concessive counterpart in (103b). That these conditional and concessive clauses can be used as clause adverbials is evident from the fact that they pass the scope test in the primed examples.

Section 8.2.1, sub I, has shown that domain adverbials such as juridischlegally in (104a) can be used as VP adverbials. The fact, illustrated in (104b), that such adverbials sometimes precede the modal adverbial waarschijnlijkprobably suggests that they can also be used as clause adverbials. The primed examples show that this is supported by the application of the entailment and scope tests.

The two uses of domain adverbials involve different scopes. VP adverbials restrict the denotation of the verbal projection; consequently, the particular choice of one of the domain adverbials in (105) will have far-reaching consequences for the goal, means, and method used in performing the action of investigating adverbs.

The clause adverbials, on the other hand, have scope over the whole proposition expressed by the lexical domain of the clause and can affect the truth value of the clause: as shown by the invalidity of the entailment in (104b''), the fact that Jan is right from a legal point of view does not entail that he is right, since he might be wrong from e.g. a moral point of view. Related to this difference is the fact that the clause (but not the VP) adverbials appear prototypically in the form of a phrase headed by the participle gezienseen, which embeds a domain adverbial functioning as a modifier of the participle; this is illustrated in (106).

Conjunctive adverbials relate the clause they modify to a state of affairs mentioned earlier in the discourse. Although conjunctive adverbials are syntactically distinct from conjunctions in that they are clausal constituents, Haeseryn et al. (1997: §8.5) notes that they perform a similar semantic function in that both specify various relations between utterances. Conjunctive adverbials may simply function as linkers, or they may indicate contrast and various contingency relations between utterances: we have omitted from the lists in (107) various obsolete forms provided by Haeseryn et al., as well as particles such as ookalso, zelfseven, which were discussed in Subsection II as focus particles.

That the adverbials in (107) are clause adverbials is clear from the fact that they pass the scope test in (60b); this is shown in (108).

Some of the conjunctive adverbials in (107) easily occur external to the main clause; the adverb daarentegen in (109a') is at least parenthetical, as can be seen from the fact that it can be preceded and followed by an intonation break, while the adverb trouwens in (109b') is clearly outside the main clause, as it precedes the main-clause initial position. We discuss such cases in more detail in Section 11.1, sub VIII.

Finally, the examples in (110) show that conjunctive adverbials can also be phrasal if they contain a deictic (here: demonstrative) element referring to a proposition expressed by an earlier clause in the discourse.

In fact, several of the adverbs mentioned in (107) are diachronically derived from phrases (P + case-marked demonstrative pronoun); cf. desondanksdespite of that and dientengevolgebecause of that.

Speech-act adverbials such as eerlijk gezegdhonestly speaking are usually phrasal and consist of a participle preceded by a manner adverb. They are always speaker-oriented and provide information about the performance of the speech act; for example, by using the adverbial eerlijk gezegd in (111a), the speaker expresses that he gives his opinion straightforwardly, although he is aware of the fact that the addressee might feel uncomfortable about it. That speech-act adverbials are clause adverbials is clear from the fact that they easily pass the scope test, as shown for eerlijk gezegd in (111b). href='S'>href='peaker-'>href='oriented'>

Speech-act adverbials are placed high in the functional domain of the clause; they are often the first adverbial in the clause. They also occur, and often feel more comfortable, in clause-external position; cf. also Section 11.1, sub VIII.