The terms topic and comment are used in the semantic description of sentences: the topic of a sentence is the entity about which something is said, while the further statement made about this entity is the comment; cf. Crystal (2008). The topic-comment division may coincide with the subject-predicate division, but this is not necessarily the case. In the question-answer pair in (62), for example, the object pronoun diethat refers to the topic. The term topic drop refers to the fact that such pronouns can be omitted if certain conditions are met. In (62) this results in a V1-structure; cf. Jansen (1981: §5). We will argue that this is a phonetic phenomenon: although the pronoun is not pronounced, it is syntactically present.

We will show later in this section that topic drop requires that the reference of the topic can be established from the context (the so-called recoverability condition); cf. Weerman (1989:53ff). In the examples, omitted topics are therefore introduced in a preceding question, as in (62a), but they can also be introduced in other ways. The data in Jansen (1981) and Thrift (2003) show that topic drop is more frequent in speech than in writing, and more frequent with objects than with subjects. Thrift (2003: §2.3) also found that acceptability judgments are sensitive to the person features of the omitted topic; the omission of arguments that refer to (sets of persons including) the speaker and especially the addressee are often judged to be unacceptable. It is suggested that this is due to the fact that the reference of first-person and second-person pronouns can shift in conversation due to turn-taking, which may also explain why first-person pronouns are easily omitted in ego documents and monologues, where turn-taking does not play a role, but see the discussion of (72a) below.

Example (62b) shows that topics can be can be moved into the main-clause initial position. However, the question-answer pairs in (63) show that this is not obligatory: the topic may also appear in the middle field of the clause.

Pronouns in the middle field differ from those in main-clause initial position in that the former can be either referential (hetit) or demonstrative (datthat), whereas the latter are usually demonstrative. This may have to do with the fact that topicalized phrases must be contrastively accented, whereas referential pronouns are typically unstressed. Note that in some cases referential and demonstrative pronouns are even in complementary distribution, as many speakers reject demonstratives in the middle field when their antecedent is [+animate]. This can be seen by comparing the question-answer pair in (64) with the one in (62).

The examples in (65) show that topic drop is only possible in clause-initial position (i.e. clauses with subject-verb inversion when the omitted topic is not itself a subject): omitting the pronoun in the middle field of the clause, as in (65a), results in an inappropriate response to (63a).

The dollar sign in (65a) indicates that the example is acceptable, but infelicitous as a response to (63a). The difference in the appropriateness of the two discourse continuations in (65) is due to the fact that the verb lezento read receives a pseudo-intransitive interpretation when the omitted pronoun is part of the middle field of the clause, but not when it is topicalized. This difference in interpretation strongly suggests that the pronoun is still syntactically present in (65b). Further support for the hypothesis that the initial position of V1-constructions of this type is syntactically filled is provided by the fact that topic drop is also allowed with R-pronouns extracted from pronominal PPs like er/daar ... van in the (b)-examples in (66). Since the PP-complement of the verb horen must have a nominal complement, the omitted topic must be syntactically present in (66b').

Van Kampen (2020a/2020b) argues that the topic drop is also sensitive to intonation in that it is only allowed with clause-initial unstressed pronouns; this includes topic-shifting pronouns such as die in (62b) and subject pronouns (which are by default used to refer to the topic of the previous clause), but not topicalized deictic pronouns given that these are always contrastively accented. Setting aside subject pronouns, this means that topic drop is restricted to third-person pronouns that trigger topic shift, i.e. that take an element from the new-information focus of the previous sentence and turn it into a new topic for the ongoing discourse; cf. Section N18.2.3.2, sub IIA2, for a detailed discussion of topic shift.

The discussion so far leads to the conclusion that topic-drop constructions have a syntactically realized but phonetically empty constituent in main-clause initial position. This suggests that topic-drop constructions involve either a (phonetically empty) topic operator, [CP OPTopic Vfinite [TP .......]], or the elision of a topic in main-clause initial position: [CP Topic Vfinite [TP .......]]; cf. Jansen (1981: §5), Van Oirsouw (1987: §2.2.3), Thrift (2003), Barbiers (2007) and Van Kampen (2020a/2020b) for further discussion. Such analyses are also supported by the fact, illustrated in (67), that topic drop is excluded in questions because their sentence-initial position is already occupied: in the wh-question this position is taken by the wh-phrase wanneer, and in the polar question by the phonetically empty question operator OPpolar (cf. Section 11.2.1).

Further evidence is provided by the fact that topic drop can be applied to only a single constituent. First consider the examples in (68), which show that topic drop can be applied equally to subjects and (in)direct objects; cf. Jansen (1981: §5).

The examples in (69) show that R-parts of pronominal PPs can also be omitted. The (c)-examples show that this requires that the PP can be split; applying topic drop to the R-part of the pronominal temporal adverbial PP in (69c') is unacceptable, because R-extraction from this PP is also ruled out; topic drop would rather affect the adverbial proform danthen in (69c), which replaces the whole PP; cf. Thrift (2003: §2.3) and the references cited there for further discussion.

Despite the fact that topic drop can apply to a large set of clausal constituents, it is impossible to construct cases in which topic drop applies to more than one constituent at a time; although the subject and the direct object in the two (b)-examples in (70) are both individually possible targets for topic drop, the unacceptability of (70c) shows that they cannot be omitted simultaneously. This follows directly from the assumption that topic drop requires the topic to be in the clause-initial position, since this position can only contain a single constituent.

We have already mentioned that topic drop is sensitive to a recoverability condition: the content of the omitted topic must be ascertainable from the context. This is illustrated by the examples in (71), which show that topic drop of a subject does not affect subject-verb agreement. Of course, the mere fact that there is subject-verb agreement in examples such as (71) is additional evidence for the hypothesis that the topic is syntactically present.

The same is shown by examples like (72a&b); since reflexive pronouns must have a syntactically present antecedent in their clause, we have to assume that it is present even after topic drop. The examples also show that the form of the reflexive pronoun is determined by the person feature of the omitted topic.

Examples such as (72a) are sometimes considered to be cases of “diary drop” (which can also be found in English); cf. Haegeman (1990). Diary drop always involves the first-person pronoun ikI and is typically found in ego-documents and letters, but it also occurs in speech and folk songs. A typical example of the latter is found in the onset of the following traditional “clapping” song: cf. handjeklappento strike a bargain by clapping hands. That we are dealing with diary drop is clear from the fact that there are several versions of this song in which the subject pronoun is expressed overtly (which can be done without affecting the meter by the use of the proclitic form ’kI); cf. liederenbank.nl for alternative versions.

Although we see no compelling reason to assign a special syntactic status to subject drop in Dutch, we want to add the examples in (74) here to show that the number specification of the omitted subject may also be relevant: reciprocals such as elkaareach other usually have a plural antecedent.

A final illustration of the fact that the content of an omitted topic is recoverable from the context is given in (75) with the supplementive naaktnude and floating quantifier allemaalall; since they both need a nominal associate in their clause, we can again conclude that the omitted topic must be syntactically present.

This section has discussed V1-clauses with topic drop and shown that there is ample evidence that the initial position of such clauses is syntactically filled by some phonetically empty constituent. Topic-drop constructions thus confirm the claim that the V1 order is merely a superficial phonetic phenomenon.