Locational adverbial phrases are prototypically PPs. The prepositions in these adverbial phrases function as two-place predicates, locating the eventuality in a specific place. For example, the adverbial phrase in de tuinin the garden in (149a) expresses that eventuality e of Jan playing takes place in the garden, while onder de boomunder the tree in (149b) expresses that it takes place under the tree; this can be expressed in logical notation by in(e,garden) and under(e,tree), respectively; cf. P32.3 for further discussion of the different spatial relations that can be expressed by prepositions. Note also that prepositional phrases like buiten/binnenoutside/inside and boven/benedenupstairs/downstairs can be (pseudo-)intransitive and therefore appear as particles; cf. Section P32.2.4.

Proforms associated with place adverbials are typically R-words: cf. Section P36.1. All the forms in (150) can be used to replace the adverbial PPs in (149). Note, however, that these forms are not specifically used as adverbials, but can also be used as complementives; calling these R-words adverbs would therefore not do justice to their actual use.

A special proform-like element is the somewhat bookish form elderselsewhere, which is not part of the set of R-words. The fact that this form can also be used as the complement of a directional preposition such as naar suggests that it is actually nominal; cf. example (151).

The R-words in (150) are also used to format pronominal PPs, which can also be used as locational adverbials, so that we can find the two examples in (152) side by side with virtually the same meaning. Again, it would be wrong to call the pronominal PPs adverbs, because they can also be used as complementives.

Time adverbials are probably the most versatile adverbials in terms of their categorial form. The examples in (153), repeated from the introduction to Section 8.3, show that they can be adjectival, prepositional, nominal, or clausal. We will not dwell on these forms here, as they are discussed in detail in Sections N21.3.1, A30.2.1.4, P32.3.2, and P33.4.1.

The examples in (154) show that the interrogative proform wanneerwhen is used in questions, while dat is used in relative clauses; in the latter case, it is often also possible to use a pronominal PP.

There are many specialized deictic forms which locate eventuality k expressed by the clause with respect to some syntactically specified or contextually determined time, which we will loosely call anchor time t: (155) shows that k can be (virtually) simultaneous with t, or anterior or posterior to it.

The discussion of the deictic forms in (155) takes as its starting point the claim from binary tense theory that present-tense interval i includes speech time n, while past-tense interval i includes a virtual speech-time-in-the-past n', where n' precedes n; cf. Section 1.5.1. Furthermore, the discussion includes the conclusion from Section 1.5.4 that the default interpretation of the present/past tense is that the so-called present j of eventuality k also includes n/n', and that k is located at n/n' in the simple present/past, while it precedes n/n' in the present/past perfect. That these default readings can be overridden by e.g. adverbial modification shows that we are dealing with pragmatics, not semantics. The default readings are most easily observed in the simple-present tense: in the absence of any indication to the contrary, (156a) is interpreted to mean that eventuality k of Jan reading the book occurs at n. This default reading is overridden by temporal adverbials such as morgentomorrow in (156b), which locate j in a position following n, as a result of which eventuality k is also located after n.

Similar effects can be observed in the present-perfect examples in (157). In the absence of any indication to the contrary, (157a) is interpreted as meaning that eventuality k of Jan reading the book was completed before n, so that the resulting state of Jan having read the book occurs at n. Again, the temporal clause adverbial morgentomorrow overrides this default reading and locates the present j of k after speech time n; as a result, (157b) cannot be used to express that eventuality k was completed before n, so that the resulting state can only occur after n. We refer the reader to Section 8.2.3 for a more detailed summary and further discussion.

The mechanisms that determine the non-default readings of the present/past tenses outlined above also play a role in the interpretation of the deictic adverbial forms in (155). We will illustrate this first for the forms in (155a), which express that eventuality k occurs more or less simultaneously with anchor time t. The default reading of the simple-present example in (158a) is that eventuality k occurs more or less simultaneously with speech time n. We therefore expect that its past-tense counterpart in (158b) to express that eventuality k occurs more or less simultaneously with virtual speech-time-in-the-past n', but this is only partially borne out: while directdirectly and onmiddellijkimmediately do indeed satisfy this expectation, nunow does not. This contrast suggests that we should distinguish between tense-sensitive and speaker-oriented adverbials: whereas tense-sensitive adverbials like direct and onmiddellijk locate k relative to n or n' depending on the tense of the clause, the speaker-oriented adverbial nu always locates k relative to n. href='S'>href='peaker-'>href='oriented'>href='S'>href='peaker-'>href='oriented'>

The readings of the examples in (158), according to which eventuality k expressed by the main clause is located approximately at n/n' are default readings, which again can be overridden by the use of temporal adverbials that shift anchor time t to a position on the time axis other than n/n'. This is illustrated by the examples in (159), where eventuality expressed by the matrix clause is at approximately the same position on the time axis as the eventuality expressed by the adverbial clause: the events of Jan leaving and Marie entering occur more or less simultaneously. The interpretive effect of this is clearest in the present-tense example (159a): since Marie's entry occurs after n, Jan’s departure will also occur after n. Consequently, the speaker-oriented adverbial nunow also leads to an infelicitous result in (159a), unless, perhaps, Marie's entry is also expected to occur approximately at speech time n. From the resulting contrast between (158a) and (159a), we conclude that the distribution of nu is not determined by the past/present-tense marking as such, but by the position of anchor time t on the time axis. Example (159b) illustrates essentially the same for the past tense.

Now consider the anterior adverbials in (155b), which express that eventuality k expressed by the main clause is prior to anchor time t: t is again taken by default as n/n' in simple-present/past tense clauses. The examples in (160) show that netonly just and pasonly just are clear cases of tense-sensitive adverbials: while k occurs immediately before n/n' in (160a), it is located before the eventuality expressed by the adverbial clauses in the (b)-examples. Again, the interpretive effect is clearest in the present tense: because the event of Jan arriving follows n in (160b), the state of Jan being absent follows n as well.

The examples in (161) show again that we should distinguish between tense-sensitive and speaker-oriented adverbials. Contrary to net in (160), the adverbials zojuist and zo-even in (161a) seem to be intrinsically anchored at speech time n: their interpretation is independent of the tense of the clause, since they simply locate eventuality k at a time just before n. That these adverbials cannot be used to locate k with respect to an anchor time other than n is clear from the fact, illustrated in the (b)-examples, that they cannot co-occur with adverbials introducing such an alternative anchor time t.

The fact that speaker-oriented adverbials cannot co-occur with adverbials introducing an anchor time other than n also explains why zojuist/zo-even cannot occur in clauses in the simple present, since for pragmatic reasons such clauses do not allow k to be located in the actualized part of the present-tense interval. These adverbials thus behave like nominal modifiers such as gisterenyesterday and verleden weeklast week; this is illustrated in (162) for the speaker-oriented adverbial toenthen.

The posterior adverbials in (155c) are even more restricted in that they all seem to be anchored by speech time n: it is very difficult to find or even construct examples in which they occur in past-tense clauses. These adverbials thus behave essentially the same as nominal modifiers such as morgentomorrow and volgende weeknext week.

Now that we have discussed the deictic adverbial forms in (155), we can continue with the discussion of the various specialized subordinators in (164) that are used to introduce temporal clauses. These subordinators can again be divided into three semantic groups according to the way they locate eventuality k expressed by the main clause with respect to some anchor time t, i.e. the time at which eventuality k' introduced by the adverbial clause occurs: k can be simultaneous with t/k', or it can precede or follow it. We refer the reader to Section P33.4 for more information on the form of these subordinators.

There are usually no restrictions on tense marking: the examples in (165a&b) show that all kinds of conjunctions can be used in present-tense and paste-tense clauses. All that matters is the chronological order of the eventualities expressed by the matrix and the embedded clause. This may also explain the phenomenon that main and adverbial clauses must agree in present/past tense marking (the so-called sequence-of-tense effect). The unacceptability of the (c)-examples in (165) may follow directly from binary tense theory; by using non-agreeing tenses, the eventualities expressed by the main clause and the embedded clause are part of different tense intervals (namely, one in the present-tense interval and one in the past-tense interval), and this may make it impossible to linearize them. Note that the adverbial clauses introduced by nadat sound somewhat marked, but are perfectly natural in the perfect tense (nadat hij gedanst heeft/hadafter he has/had danced).

The (a)-examples in (166) show that adverbial clauses introduced by toenwhen are exceptional in that they can only occur in past-tense sentences. Since the adverbial nunow can only be used in present-tense clauses, we might expect something similar for adverbial clauses introduced by nu, but the (b)-examples show that this is not borne out (although we should perhaps point out that examples such as (166b') are used especially in narratives).

A number of temporal subordinators are listed in (164) as both simultaneous and posterior. Haeseryn et al. (1997: §10.3.3) noted that the interpretation of these elements is determined by the temporal properties of the adverbial clause. The examples in (167) show that the simultaneous (k ≈ t/k') reading occurs when the adverbial clause is in simple present/past tense, while the posterior (k > t/k') reading occurs when the clause is in perfect tense.

Durational adverbials can be nominal, adjectival, or prepositional.

There are many forms expressing frequency: nooitnever, eensone time, somssometimes, vaakoften, meestal/doorgaans/telkensgenerally, altijd always. It is often difficult to determine the categorial status of these forms: we can only say with certainty that vaakoften and veela lot are adjectives, since they can also occur in comparative and superlative form: vaak - vaker - het vaakst; veel - meer - het meest). Note further that it is not obvious that these forms can actually be used as VP adverbials: because examples such as (169a) do not easily allow the pronoun doet dat + adverb paraphrase, but do allow the scope paraphrase, we are probably dealing with clause adverbials.

The examples in (170) show that there are clear cases in which nominal phrases are used as VP adverbials: these adverbials are usually formed with the noun keer/maaltime(s) preceded by a cardinal numeral. The examples in (170b&c) show that (170a) can be easily paraphrased by a pronoun doet dat + adverb clause, while it does not allow the scope paraphrase.