The notions 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 (1991). The topic-comment division may coincide with the subject-predicate division, but this is not necessarily the case. In the question-answer pair in (34), for instance, the topicalized object pronoun die'that' refers to the topic. The term topic drop refers to the fact that such topics can be omitted if certain conditions are met. In (34), this results in a V1-structure; cf. Jansen (1981:ch.5). We will argue this to be a phonetic phenomenon: despite being not pronounced, the pronoun is syntactically present.

Topic drop requires that the reference of the topic can be reconstructed from the context (which is known as the recoverability condition); cf. Weerman (1989:53ff.). The examples in this section provide the reference in the preceding question, but it may also be expressed in other ways. It should further be noted that Thrift (2003: Section 2.3) found that acceptability judgments made by speakers depend on the person features of the omitted topic; omission of arguments referring to (sets of individuals including) the speaker and, especially, the hearer are often judged to be unacceptable. Thrift suggests that this is due to the fact that the reference of first and second person pronouns shifts in conversation due to turn-taking, which may also account for the fact that first person pronouns are easily dropped in ego-documents and monologues, in which turn-taking does not play a role. The data in Jansen (1981) and Thrift (2003) further suggest that in speech topic drop is more frequent and considered more common with objects than with subjects.
      Example (34b) has shown that topics can be topicalized and then occupy the sentence-initial position. The question-answer pairs in (35) show, however, that this is not obligatory: the topic may also occur in the middle field of the clause.

Pronouns in the middle field differ from those in sentence-initial position in that the former can be either referential (het'it') or demonstrative (dat'that'), whereas the latter are normally demonstrative. This is probably related to the fact that topicalized phrases must bear accent, while referential pronouns are normally unstressed. Note that in some cases, referential and demonstrative pronouns are even in complementary distribution as many speakers reject demonstratives in the middle field if their antecedent is +animate. This will become clear by comparing the question-answer pair in (36) to the one in (34).

The examples in (37) show, however, that topic drop is only possible in topicalization constructions: omission of the pronoun in the middle field of the clause, as in (37a), results in an inappropriate response to (35a).

The difference in appropiateness of the two discourse continuations in (37) is due to the fact that the verb lezen'to read' receives a pseudo-intransitive interpretation if the omitted pronoun is part of the middle field of the clause but not if it is topicalized. This strongly suggests that the pronoun is still syntactically present in the V1-construction in (37b). Additional support for the hypothesis that the initial position of the 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 (38). Given that the PP-complement of the verb horen obligatorily has a nominal complement, the omitted topic must be syntactically present in (38b').

      The discussion above leads to the conclusion that topic drop constructions have a syntactically realized, but phonetically empty constituent in sentence-initial position. This in turn suggests that topic drop constructions involve a (phonetically empty) topic operator, [CP OPTopic Vfinite [TP .......]], or elision of a topic in sentence-initial position: [CP Topic Vfinite [TP .......]]; see Jansen (1981:ch.5), Thrift (2003) and Barbiers (2007) for discussion. Analyses of this sort are supported by the fact that topic drop is excluded in questions; the pronoun dat cannot be dropped in the two (b)-examples in (39) because the sentence-initial positions are already occupied by, respectively, the wh-phrase wanneer and the phonetically empty question operator OPpolar discussed in Section 11.2.1.

      More evidence is provided by the fact that topic drop can be applied to a single constituent only. First consider the examples in (40), which show that topic drop may affect subjects and (in)direct objects alike; cf. Jansen (1981:ch.5).

The examples in (41) show that R-parts of pronominal PPs may also be dropped provided that the PP can be split; applying topic drop to the R-part of the pronominal PP in the reaction to the question in (41c') is unacceptable because R-extraction from temporal adverbial phrases is also excluded—it is instead the adverbial pro-form dan'then' in (41c) that is dropped; see Thrift (2003: Section 2.3) and the references cited there for more discussion.

Despite the fact that topic drop may 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 the same time; although the subject and the direct object in the two (b)-examples in (42) are both possible targets for topic drop individually, the unacceptability of (42c) shows that they cannot be dropped simultaneously. This follows immediately on the assumption that topic drop requires the topic to be in clause-initial position, and this position can only contain a single constituent.

      Topic drop is sensitive to a recoverability condition: the substantive content of the dropped topic must be reconstructible from the context. This is illustrated by means of the examples in (43), which show that topic drop of a subject does not affect subject-verb agreement. The fact that there is subject-verb agreement in examples such as (43) of course constitutes additional evidence for the hypothesis that the topic is syntactically present.

The same is shown by examples like (44a&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 further show that the form of the reflexive pronoun is determined by the person feature of the omitted topic.

Examples such as (44a) are sometimes considered to be cases of "diary drop" (which can also be found in English); see Haegeman (1990). Diary drop always involves the first person pronoun ik'I' and is typically found in ego-documents and letter 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. handjeklappen'to 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 overtly expressed (which is readily possible without affecting the meter by the use of the proclitic form k'I'); see liederenbank.nl for alternative versions.

Although we do not see any compelling reason for assigning subject drop a special syntactic status in Dutch, we added the examples in (46) to show that the number specification of the omitted topic is likewise relevant: reciprocals like elkaar'each other' normally have a plural antecedent.

A final illustration of the fact that the substantive content of an elided topic must be reconstructible from the context is given in (47). These examples show that supplementives and floating quantifiers can be used despite the fact that they are normally associated with a noun phrase in their own clause.

      This section has discussed V1-clauses with topic drop and has 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.