In general, the modifier zo must be combined with a complement; if the examples in (120) are pronounced with a neutral intonation pattern, the result is not very felicitous.

If these examples are pronounced with an accent on the element zo, the result improves considerably, but it is questionable whether zo acts as a degree modifier in such cases. Instead, it seems to function as a deictic element: sentences such as (120) are usually accompanied by a manual gesture indicating the size or length of the object under discussion, and stressed zo refers to this gesture. Note that this deictic element zo can also occur in isolation, as in Zijn computer is zò [gesture: thumbs up], which means “His computer is great”.

Another way to make the examples in (120) acceptable is to lengthen the vowel of the element zo; such cases are also characterized by an intonational “hat” contour, i.e. with a rising accent on zóóó and a falling accent on the following adjective. This use of zóóó is particularly common with an amplifying meaning: the speaker wants to emphasize something or express feelings of sympathy, affection, etc. Some typical examples are given in (122).

The element zo can also be used in isolation in a number of very specific syntactic contexts. Some examples of negative imperatives are given in (123); zo can modify complementive predicates in a copular construction with zijnto be, supplementives predicated of the object of the clause, and manner adverbials.

In these cases, the interpretation of zo is induced by the non-linguistic context: in (123a) the speaker expresses that the addressee should not be as stupid as he apparently is at the time of the utterance, in (123b) the addressee is advised not to eat the soup as hot as it is at that very moment, and in (123c) the addressee is requested not to walk as fast as he is at that moment.

When zo is preceded by the negative adverb nietnot in a declarative clause, as in (124a), there is a downtoning effect. This effect is lost as soon as a comparative als-phrase is added: example (124a) implies that the bag is not very heavy, but this implication is completely absent in (124b), which merely expresses that the bag is less heavy than the suitcase.

The element zo can also be used in contrastive or concessive constructions of the kind in (125a) and (125b). In these examples, there is an amplifying effect; it is implied that Jan is quite young/smart. For completeness, the primed examples show that the addition of a comparative als-phrase is impossible.

When deictic or emphatic zo is combined with an attributive adjective, it must precede the indefinite determiner een. The combination zo + een is usually phonetically reduced to zo’n. This is illustrated for (120a) and (122b) in (126).

Example (126a) (but not example (126b)) is actually ambiguous: the element zo can function either as a modifier of the adjective, in which case the sentence can be combined with a gesture indicating the size of the computer, or zo’n can function as a complex demonstrative, in which case the sentence can be combined with a gesture pointing to a computer of comparable size or type; the latter, demonstrative use of zo’n is discussed in Section N18.2.3. Note that the modified adjective must follow the indefinite article een; constructions of the English type that/so big a computer, in which the adjective precedes the article, are not acceptable in Dutch.

Finally, note that zo’n cannot be used when the head noun is plural; Dutch uses the determiner zulksuch in such cases. Like zo’n, the determiner zulke can be used both as a modifier of the adjective and as a demonstrative.

The modifier zo is usually accompanied by some other element, and we will argue that it can be considered as the head of a complex degree phrase. In (129) zo is combined with a finite clause, which we will call degree clauses from now on.

The string zo A dat ... forms a constituent, which is clear from the fact illustrated in (130) that it can be placed in clause-initial position; cf. the constituency test.

There is at least one compelling reason to assume that it is the element zo, and not the adjective, that selects the degree clause. The availability of the clause depends on the presence of the element zo; if the latter is omitted, the result is completely unacceptable. This is shown in (131).

Since the phrase zo A dat ... must be considered a constituent, and since the presence of the degree clause depends on the presence of zo, we can conclude that the degree clause is selected by (i.e. is a complement of) zo. If the clauses in (129) are embedded, as in (132), the degree clause must be in extraposed position; it must follow the finite verb in clause-final position. The fact that extraposition is obligatory is a feature of a larger set of dependent clauses.

Given that the adjective and its complement clause must occur discontinuously in (132), we should not really be surprised that the modified adjective can be topicalized by itself, as the primeless examples in (133) show. The primed examples, however, show that the degree clause cannot be topicalized; it cannot be placed in a position preceding the AP.

A pattern similar to that of the predicative adjectives in (132) emerges when the adjective is used attributively. As shown in (134), the degree clause cannot be adjacent to the prenominal adjective, but must be in postnominal position; cf. Section 28.4, sub II, for cases where the entire AP occurs postnominally.

That the finite clause cannot be adjacent to the adjective reflects a general property of attributive adjectives, expressed by the head-final filter on attributive adjectives in (135). This filter requires adjectives with the attributive -e/-∅ ending to be adjacent to the noun (phrase) they modify; cf. Section 27.3, sub IB, for a more thorough discussion of this filter.

This is not all, since the examples in (136) show that the degree clause is not only postnominal, but must also follow the finite verb in clause-final position in the embedded counterparts of the main clauses in (134); the degree clauses are in extraposed position, just as in the examples in (132).

In (134) and (136), the indefinite article precedes the element zo. Although this leads to an acceptable result, this order sounds a bit marked; the element zo is preferably placed in front of the article, which is illustrated in (137) for the examples in (134). Note that, unlike the cases discussed in Subsection A, these examples are not ambiguous; the reading in which zo’n acts as a complex demonstrative is not available.

The primeless examples in (138) show that zo can be replaced by the (somewhat formal) demonstrative dusdanigsuch when it follows the indefinite determiner eena, as in the primeless examples in (136). The primed examples in (138) show that this is not possible when zo precedes the indefinite article, as in (137).

The element zo can also be combined with an infinitival degree clause. The cases in (139) show that infinitival degree clauses differ from finite degree clauses in that they must precede the element zo. The number sign indicates that (139a') is fully acceptable with the irrelevant reading “Jan is so kind to kiss (someone)”.

The examples in (140) further show that the infinitival clause must be strictly adjacent to zo: for example, separating the degree clause from zo by the adverbial phrase weeragain yields an unacceptable result.

The fact that the infinitival and finite degree clauses have different distributions in the clause might lead us to expect that they can co-occur in a single clause. The examples in (141) show that this is not true: they are mutually exclusive. This suggests that, despite their different distribution, they have a similar or identical syntactic function.

The primeless examples in (142) show that the infinitival degree clause and the adjective can be placed together in clause-initial position, which shows that they form a constituent; cf. the constituency test. The primed examples show that the infinitival clause cannot be moved into the clause-initial position by itself, which is consistent with the previously established fact, illustrated in (140), that the infinitival clause must be strictly left-adjacent to the adjective.

For completeness, note that the primed examples in (142) become fully acceptable when the sequence zo + A functions as an afterthought, in which case it must be preceded by an intonation break, as in (143a&b). In these examples, the infinitival clause does not function as a degree clause, but as the complementive part of the copular construction. That it is not the AP that functions as the predicate is clear: it can be omitted in the primeless examples, and it must follow the clause-final verb in the primed examples; we can therefore ignore these cases here.

The infinitival degree clauses in (139) contain two interpretive gaps. The first is the implied subject PRO, which is usually found in infinitival clauses and which in this case must be interpreted as disjoint in reference to the logical subject of the AP, Jan. The second interpretive gap, on the other hand, is interpreted as coreferential with Jan. The second gap can perform several functions in the infinitival clause: it functions as the direct object of the verb zoenento kiss in (139a), and in (139b) it functions as the complement of the preposition bijat. The examples in (144) show that these implied arguments cannot be realized overtly; the italicized phrases cannot be present.

There are good reasons to think that the second gap is the result of movement. This is clear from the fact that the preposition met appears in its stranded form mee in (145); see the contrast between Jan pronkt [met zijn ring]Jan shows off his ring versus de ring waari Jan [mee ti] pronkt the ring that Jan shows off. Because there is no phonetically realized R-word in the degree clauses in (139) and (145), they probably involve an empty operator OP. Since this operator has the function of a relative pronoun, it can be assumed that it is moved into the clause-initial position and is construed as co-referential with the logical subject of the adjective; movement and coreferentiality are both indicated by coindexing.

Although we have seen that an empty operator can be present, the construction does not require it; if the infinitival degree clause is in the passive voice or contains an unaccusative verb, only the implied subject PRO is present. In the passive construction in (146a), PRO is interpreted as co-referential with the logical subject of the adjective (the infinitival is a fixed expression meaning “to look as neat as a new pin”). If it is an unaccusative verb, such as bevriezento freeze in (146b), PRO is interpreted arbitrarily.

The infinitival degree clauses in (139a&b) are similar to the infinitival clauses we find in the so-called easy-to-please construction in (147); cf. Section 28.5, sub IVA. However, they also differ from them in several ways. First, the presence of an empty operator, i.e. the second interpretive gap, is obligatory in the easy-to-please construction. Second, the infinitival clause of the easy-to-please construction must follow the adjective, as shown by the (a)-examples in (147). Finally, example (147b) shows that when the adjective is preceded by zo in the easy-to-please construction, the AP must contain an additional degree clause.

Since the adjective is the rightmost part of the complex phrases in (139), the head-final filter on attributive adjectives in (135) leads us to expect that these complex phrases can also be used attributively. As can be seen in (148), this expectation is indeed confirmed. The primed examples show that zo never precedes the indefinite article in such cases.

The infinitival clause can also be preceded by the negative element nietnot. Despite the fact that niet is external to the infinitival clause (it precedes the complementizer om), it must be assumed to be part of the AP, since it can be pied-piped by topicalization, as in (149b).

Note that the examples in (149) can easily be confused with the near-synonymous constructions without the complementizer om in (150).

However, there are several differences between these constructions. A first difference, illustrated in (151), concerns the placement of the AP zo zout: if the complementizer om is present, it must precede the clause-final verbs, while it is possible (and perhaps even preferable) to place it after these verbs if om is absent.

The second difference concerns whether the presence of the AP zo zout is obligatory: only in the absence of the complementizer om can zo zout be omitted. Since the resulting construction in (152b) clearly involves a modal infinitive (cf. Chapter 31), we would like to suggest that the same is true for the constructions in (150), but we realize that this may require further research in the future.

A special case of degree modification with zo is the discontinuous degree phrase zo A mogelijkas A as possible. As with the degree clauses in (129), the presence of the element zo is required if the element mogelijkpossible is to occur. The phrase zo A mogelijk does not easily occur in copula and vinden-constructions for semantic reasons (hence the percentage sign in (153a&b)), but it is possible in resultative constructions and adverbial phrases like (153c&d). The unacceptability of the primed examples in (153) shows that the element mogelijk must be right-adjacent to the adjective and thus cannot undergo extraposition.

If the adjective takes a PP-complement, the element mogelijk must be placed between the adjective and the complement, which is unexpected, given the general rule that a selecting head is usually closer to its complement than to its modifiers. We will not discuss this problem here, but postpone it to Section 26.3.1, where it will be argued that the word order results from leftward movement of the adjective across mogelijk: zo bangi mogelijk ti voor honden.

Since mogelijk must be right-adjacent to the adjective, the head-final filter on attributive adjectives in (135) would lead to the prediction that the string zo A mogelijk cannot be used in attributive position. As example (155a) shows, this prediction is clearly false. The acceptability of this example is probably related to the fact that the attributive -e ending can be added to the element mogelijk itself; cf. Section 27.3, sub IIB, for further discussion. Finally, the markedness of (155b) shows that the modifier zo preferably follows the indefinite article when combined with mogelijk.

The percentage sign indicates that the sequence zo’n A mogelijk(e) N can be found on the internet, but that its frequency is much lower than the competing sequence een zo A mogelijk(e) N. That speakers may have difficulty in deciding between the two options is suggested by the fact that sequences with doubling of the indefinite article (i.e. een zo’n A mogelijk(e) N) occur frequently on the internet.

This subsection concludes the discussion of the degree modifier zo by discussing its occurrence in the discontinuous phrase zo A als ...as A as .... The als-part of the phrase can have two kinds of relations to the modified noun (phrase), which we will call metaphoric and deictic, respectively.