Adpositional predicates are probably the most common in the absolute met-construction. When the predicate is prepositional, as in (76), the PP can express either a location or a direction.

Since the directional reading is possible, it will not be surprising that postpositional and circumpositional phrases can also be used in the construction; cf. (77).

Example (78a) shows that intransitive prepositions can also be used. Verbal particles like op in (78b) are marginally possible if they occur with the main verb hebbento have (such as Hij heeft een das omHe wears a tie) or if they can be used in a copular construction (cf. De drank is opThe booze is finished); they are excluded in all other cases.

In the examples in (79), we see that the predicative part of an absolute met-construction can also be an adjectival phrase.

In order to be able to figure in this construction, the adjective must denote a transitory property: typical stage-level adjectives like ziekill and dronkendrunk generally give rise to a felicitous result, while typical individual-level predicates like intelligentintelligent or klein van stuksmall of posture in (80) yield degraded results.

Nominal predicates cannot easily be used in the absolute met-construction. Instead, the nominal predicate is preceded by the element alsas; cf. the acceptability contrast between the two examples in (81).

The contrast between the two examples in (81) is not surprising, since the examples in (82) show that we find similar contrasts in the clausal domain with supplementive and complementive constructions.

It seems that participles and infinitives can only marginally function as predicates in absolute met-constructions. Example (83a) is a more or less acceptable example with a passive participle; the percentage sign expresses that the judgments on this example seem to vary from “perfect” to “marginal”. That we are dealing here with a (verbal) passive participle is supported by the fact that an agentive door-phrase is present. However, the presence of the perfect participle gedronkendrunk in example (83b) leads to an ungrammatical result; the particle op must stand alone. To appreciate the significance of (83b), it should be noted that many apparent cases of past/passive participles actually involve deverbal adjectives. Example (83c) illustrates this; that we are dealing with an adjective here is clear from the fact that geslotenclosed expresses a stative property.

The use of present participles, as in (84a), is generally judged to be marginal. However, when the present participle is suffixed with an -e ending, as in (84b), the result is well-formed; this is related to the fact that such forms can also be used in copular constructions such as (84b'); cf. Section A31.3.1, sub IIB.

Modal infinitives can also be used in this construction, which is to be expected since they can also appear as predicates of copular constructions. Note in passing that we are not dealing with dependent infinitival clauses; this is clear from the fact that the te + infinitive sequences must precede the finite verbs in the primed examples; clausal te-infinitives always follow the finite verb in clause-final position (cf. Chapter V7).

Example (86a) shows that regular (i.e. non-modal) (te-)infinitives are excluded in standard Dutch met-constructions. However, it can be noted that similar examples do occur in certain dialects spoken in Flanders and Brabant. The grammatical example in (86b) is from the Flemish variety spoken in Wambeek; the properties of this construction are discussed in Haslinger (2007: §3).

In (87a&b), the absolute met-construction is used attributively, as can be seen from the fact that the sequence consisting of the noun phrase and the absolute construction is placed in clause-initial position. In the attributive use of the absolute construction, there are several additional restrictions on the predicative part, as shown by the markedness of the examples in (87c&d).

The examples in (87) suggest that adjectives are not possible in attributively used absolute met-constructions. However, it has been argued that the difference between the two cases is related to the fact that the constructions in (87a&b) can be paraphrased by a relative clause containing the verb hebben, whereas the examples in (87c&d) cannot.

Note, however, that the correspondence between the absolute met-construction and the relative construction with hebben does not work in reverse: while example (89b) with a relative clause containing hebben is fully acceptable, the use of the absolute met-construction in (89a) leads to a highly degraded result.

We conclude that it is not yet clear whether there is a categorial restriction on the predicate of an attributively used absolute met-constructions, or whether some other restriction is at play.

Adverbially used absolute met-constructions express an accessory circumstance with respect to the eventuality expressed by the clause: they can express a cause, as in (90a), specify a condition under which the eventuality in the main clause occurs, as in (90b), describe a state or an event that occurs simultaneously, as in (90c), etc.

In this respect, the absolute met-constructions in (90) do not differ from the PPs in (91), the complements of which do not involve predication.

The examples in (92) and (93) show that the constructions in (90) and (91) are not only semantically, but also syntactically similar: they do not allow R-extraction, in contrast to what is usually the case with other types of met-PPs; cf. the discussion of example (401) in Section 32.3.3, sub IIA.

This similarity in meaning and syntactic behavior seems to justify the assumption that the two constructions are essentially the same, the only difference being that in (90) the preposition met takes a complex phrase expressing a predicative relation as its complement, whereas in (91) the preposition simply takes a nominal complement; cf. Beukema & Hoekstra (1984) for an alternative account of the unacceptability of the examples in (92), and Subsection VC for some putative counterexamples to the claim that R-extraction is excluded from absolute met-constructions.

For completeness, note that the absolute met-constructions in (90) can be paraphrased by adverbial clauses containing a copular construction with zijnto be.

The examples in the previous subsection are cases in which the absolute PP is used adverbially and refers to some accessory circumstance in which the eventuality denoted by the verb takes place. However, the absolute PP can also be used as a supplementive, conveying additional information about one of the arguments of the verb. This is illustrated in (95).

The absolute met-PP in (95a) modifies the direct object of the clause: it expresses that the robber, who is running away, has a revolver in his hand; that the object is modified is also clear from the fact that the possessive pronoun zijnhis must be construed as coreferential with the noun phrase de roverthe robber (which is indicated by italics). Like supplementive APs, supplementive absolute PPs can also modify the subject of the clause; the absolute met-PP in (95b) expresses that Marie, who is approaching the robber, has a revolver in her hand; that the subject is modified is also clear from the fact that the possessive pronoun haarher must be construed as coreferential with the noun phrase Marie. Supplementive absolute met-PP must modify some argument of the verb: since there is no suitable antecedent available in (95c), the sentence is ungrammatical.

Like attributively used absolute met-PPs, the supplementive absolute met-PPs in (95) can be paraphrased by an adverbial clause with the verb hebben; cf. (96).

The instances in (97) show that when the hebben-construction is excluded, the supplementive use of the absolute met-PP is not possible either.

The previous subsections have shown that in terms of paraphrasing there is a difference between the adverbial use of the absolute construction on the one hand and its attributive and supplementive use on the other. The fact that the latter must allow a paraphrase with hebbento have has led to the hypothesis in (98a), according to which the complement of absolute met has a clause-like structure with an empty abstract verb [V e] meaning “to have” and a PRO-subject corresponding to the subject of the paraphrase with hebben; cf. Klein (1983). This hypothesis has been refuted by bringing adverbially used absolute constructions into the question. These clearly do not involve the postulated empty verb or a PRO-subject, but favor the small clause structure in (98b); cf. Beukema & Hoekstra (1983) and also Van Riemsdijk (1978) for additional arguments against structures like (98a). However, a problem with the proposal in (98b), which has also been defended in a non-generative framework in Duinhoven (1985), is that it does not take into account the clause-like properties of the complement of met: Subsection V will show that the complement of met can contain all sorts of phrases that we would expect to occur in a clause rather than a small clause; these include adverbial phrases, supplementives, (moved) R-words, and so on. An attempt to reconcile the two approaches can be found in Smits & Vat (1985), which assumes that the complement of met is a verbal projection that is smaller than a full clause and therefore does not contain a PRO-subject, as in (98c).

To our knowledge, the discussion of the internal structure of absolute met-constructions has not been continued since Smits & Vat (1985). Since the issue is still unfinished, we will not discuss it further here. However, the reader will find a number of comments pertaining to it in the following subsections, and in Subsection VI we consider the various proposals again in the light of our findings.

The examples in (106) show that the noun phrases Jan and Marie en Jan can act as antecedents of an anaphor in the complement of the predicative adjectival phrases headed by geïnteresseerdinterested and verliefdin love (coreference is indicated by identity of indices). This supports the idea that the nominal part of the absolute construction acts as the logical subject of the predicative part; if the anaphors zichzelf and elkaar are complements of the adjectival head, they can only be bound by the logical subject of the predicative AP; cf. Section N22.3 for a detailed discussion.

Example (107a) shows that if the predicative part of the absolute met-construction is a locational PP headed by e.g. naastnext to, the simplex reflexive pronoun zich cannot be bound by the nominal part of the absolute met-construction but must be bound by an argument of the main clause. This again supports the idea that the noun phrase functions as the logical subject of the predicative part of the absolute construction, because subjects of predicative PPs never function as antecedents of a simplex reflexive embedded in the PP; cf. Section N22.4. Example (107b) shows that if the absolute met-construction is used attributively, the antecedent of zich must also be external to the absolute PP; it is bound by the head of the modified noun phrase.

Klein (1983) has given examples of this kind as evidence for hypothesis (98a) in Section 33.5.1, sub IIID, that absolute met-constructions with a PP-predicate contain a phonetically empty PRO-subject: PRO would then satisfy the condition that zich needs a local binder, i.e. an antecedent within a certain anaphoric domain (here: the absolute met-construction).

Examples (108a&b) show that the nominal part (i.e. the logical subject of the predicative part of) of the absolute met-construction can also appear in a reflexive form. A notable restriction is that the predicate must be adpositional or appear as an als-phrase; if the predicate is adjectival, as in (108c), the construction is excluded. This may be related to restrictions on computational complexity, because the supplementive ziek in (108c') expresses the intended meaning more economically than the absolute construction met zichzelf ziek; cf. Smits & Vat (1985) for an alternative proposal.

Since personal and reflexive pronouns are usually in complementary distribution, it is not surprising that zichzelf cannot be replaced by hem without changing the coreferentiality relations: the examples in (109a&b) are only possible if hem refers to a person other than Jan. Note that the fact that the pronoun hem cannot be bound by Jan in the acceptable example in (109c) supports the earlier suggestion that (108c) is blocked by the more economical expression in (108c'). Assuming that anaphors and pronouns are in complementary distribution, the claim that binding of zichzelf by Jan in (108c) is not allowed by the binding conditions would lead to the false prediction that binding of hem by Jan is possible in (109c).

The simplex reflexive zich cannot be substituted for any of the occurrences of zichzelf in the examples in (108). It is not clear a priori whether this is related to the binding restrictions on this reflexive form. Since zich is phonetically weak, the unacceptability of the examples in (110) may also follow from the fact that a pronoun functioning as the nominal part of the absolute met-construction must be strong; cf. example (103).

The complement of the preposition met can be modified by temporal, locational or other adverbial phrases. In (111), we provide some examples with the temporal phrases nog altijd/steedsstill and voortdurendcontinuously.

The temporal phrases in (111) are quantified and are normally used as clausal adverbials. However, the addition of temporal adverbials that refer to a fixed point or interval on the timeline yields marked results. This is shown in (112).

In (113) we give examples with locational phrases. Note that the use of locational phrases can lead to ambiguity; it is not clear whether we should consider the als-phrase or the locational phrase voorop as the predicate in (113b), since the primed examples show that both can be used in this function.

Modal adverbial phrases are used in the examples in (114).

Absolute met-PPs can also contain supplementive phrases. In (115a), the adjective verfrommeld preceding the locational PP functions as a supplementive predicated of the noun phrase zijn hoed and in (115a) the adjectival participle phrase vastgebonden is predicated of the noun phrase zijn handen.

Example (116) further shows that the predicative PP can be modified by modifiers of distance (e.g. vlakclose) or direction (schuindiagonally; cf. Section 34.1.4, sub I, for a discussion of these modifiers.

Finally, example (117) shows that the absolute met-construction itself can also be modified; this is only possible with focus particles like alleenonly, zelfseven, etc.

The (b)-examples in (118) show that the predicative PP voor de kast can undergo R-extraction. The fact that the R-word must follow the preposition met but precede the modifier vlak shows that the landing site is internal to the absolute met-phrase but external to the predicative PP.

It has been claimed that locational erthere can also be used in the absolute met-construction. We disagree; the use of the strong form daar is much preferred, which may be related to the stress properties of the construction; cf. the discussion of (102) to (105). The expletive er is never possible; insofar as the use of er in (119) is acceptable, it must be interpreted as an adverbial phrase of place.

The examples in (120b-d) show that absolute met-PPs can contain not only prepositional er but also quantitative er, and that er can even perform these two functions simultaneously; this conflation of functions, which also occurs in the clausal domain, will be discussed in more detail in Section 36.5.3. The slightly marked status of the examples in (120c-d) may again be due to the intonational properties of the construction.

Example (121) illustrates that absolute met-constructions differ from regular met-PPs in that they cannot undergo °R-pronominalization: while the PP met het boekwith the book can be pronominalized as daar ... meewith that, the pronominalized counterpart of the absolute met construction in (121a), given as (121b), is impossible. There are, however, some putative counterexamples, which will be discussed in the remainder of this subsection.

The singly-primed examples in (122) illustrate that the bracketed sequences met + NP + PP in the primeless examples can be placed in clause-initial position; this shows that these sequences form a constituent and can therefore be analyzed as absolute met-PPs. However, the relative constructions in the doubly-primed examples show that R-extraction is possible.

The fact that the bracketed sequences can be analyzed as absolute met-PPs does not, of course, mean that they must be. There are several reasons for thinking that the examples in (122a&b) are ambiguous between a reading with an absolute met-PP and a reading with a met-PP and an independent locational PP op zijn arm/aan de lijn. First, the examples in (123) show that the met + NP + PP do not have to occur together, i.e. they can be separated by wh-movement in relative and interrogative clauses and also by A'-scrambling; Subsection I has shown that this is not possible in the case of absolute met-constructions.

Second, the examples in (124) show that the noun phrases following met can take the form of a weak pronoun; Subsection IV has shown that this is not normally possible in the absolute met-construction.

The examples in (123) and (124) provide evidence that the doubly-primed examples in (122) should be analyzed as cases of R-extraction, not from an absolute met-construction, but from a regular met-PP that happens to be followed by some independent locational PP. We therefore conclude that R-extraction from absolute met-constructions is indeed blocked.

Example (125b) shows that adverbial phrases contained in absolute met-constructions can be placed after the predicative PP, i.e. we find a process similar to PP-over-V in clauses. However, the adverbial PP must remain within the absolute phrase; moving this PP across the clause-final verb(s) leads to unacceptability, as shown in (125c).

The examples in (126) and (127) show that the same is true for relative clauses and attributive PPs. The (b)-examples show that extraposition from the nominal part of the construction is possible; the resulting structures are somewhat marked, but this is probably due to their complexity. The (c)-examples show that the moved clause/PP must remain within the absolute phrase; it cannot be placed to the right of a verb in clause-final position.

The complement of met can be a coordinate structure. The predicates of the two conjuncts may or may not be of the same category: in (128a) the two predicates are both of the category PP, while in (128b) we have a predicative PP in the first conjunct and a predicative AP in the second.

Since the conjuncts of a coordination are usually taken to be constituents, the acceptability of the examples (128) shows that the nominal part and the predicative part of an absolute met-PP form a constituent; these examples thus provide support for all three analyses in (98); cf. Section 33.5.1, sub IIID.

If the nominal part of the construction is heavy, e.g. a modified noun phrase, the absolute met-construction allows inversion of the noun phrase and the predicate, provided that the predicate is adpositional or an als-phrase, as in (129).

The examples in (130) show that inversion is not possible if the predicate is an intransitive adposition, a particle, or an adjectival phrase.

Inversion of the noun phrase and a predicative PP is also allowed in contrastive contexts such as (131a), where contrastive accent is indicated by small capitals, and in constructions such as (131b), in which both the noun phrase and the nominal complement of the predicative PP are quantified; cf. example (25) in Section 35.2.1.1, sub II, for further discussion related to the word order in (131b).