Proper names such as Jan are typically used for addressing people, but the same holds for nouns indicating kinship relations or professions; see Section N5.1.2.2 for details. The examples in (9) show that such vocatives do not have to precede the sentence, but can also follow it.

There is no immediate reason for assuming that vocatives are syntactically connected to the sentence, as they can also be used as independent utterances, e.g., when they are used to call someone: Jan! That vocatives are not syntactically connected to the sentence is also clear from the fact that they cannot bind reflexive pronouns, which must have an antecedent within their minimal clause.

In this respect they are similar to salutations (hoi'hi'), interjections (hé!'hey!') and hesitation markers (eh'er'), which can likewise be used as independent utterances. It should be noted, however, that there is a word order restriction regarding interjections and vocatives in that the former must precede the latter. It is not clear whether this shows that vocatives are more intimately related to the following sentence, as the independent uses of Hey, Jan! and Eh, Jan? exhibit a similar word order restriction.

It is not the case that vocatives are always closest to the following sentence; the examples in (12) show that they can be preceded or followed by phrases such as zeg/kijk (eens), which are likewise used for drawing attention.

It should be pointed out, however, that the structure of the examples in (12) is not very clear. First, note that zeg/kijk (eens) appears to be a regular imperative clause, which is clear from the fact that at least kijk (eens) can also be used as an independent utterance: this suggests that we might be dealing with two juxtaposed clauses. If so, the vocatives in the primed examples may belong to the imperative, as is in fact also suggested by the acceptability of Jan, kijk eens!. It is therefore not so obvious that the primed examples show that vocatives can be separated from their associate sentences by other extra-sentential elements. The primeless examples are potentially ambiguous for the same reason, as the vocatives can be construed either with the preceding imperative or with the following interrogative/declarative clause.
      The discussion above has shown that elements used for addressing and drawing attention can often be used independently. Despite the orthographic convention of using a comma, we might therefore be dealing with separate utterances or, alternatively, with some sort of juxtaposition.

One of the prototypical functions of sentence-external pragmatic markers is that they help in organizing the discourse by pointing to some earlier statement or some state of affairs. Diewald (2009) distinguishes two different kinds, which she refers to as text-connective and discourse markers, respectively.

The polar elements ja'yes' and nee'no' are prototypically used as answers to yes/no-questions. This is illustrated in the (a)-examples in (20): the polar elements indicate whether or not the open proposition expressed by the question is applicable to the domain of discourse. Such polar phrases can, however, also be used to affirm or to contradict propositions given earlier in the discourse: the (b)-examples provide an instance in which affirmation/contradiction is consciously elicitated by the first speaker by virtue of the discourse marker toch. The polar phrases ja/nee in (20) can be used as independent utterances but they can also be followed by a clause expressing the propositional content of the reply in full.

It is not easy to determine the precise meaning contribution of ja and nee in examples like those in (20). It seems too simple to say that ja expresses confirmation and nee expresses denial of some presupposition held by the speaker. This is clear from the fact that the negative counterparts of the examples in (20a&b), which suggest the opposite presuppositions, may trigger exactly the same answers. We ignore the fact that jawel may be preferred to ja in (21b'); we return to this in our discussion of (26).

The fact that the sentences in the primed examples above can be omitted without a clear change of meaning suggests that ja and nee are in a sense shorthand for, respectively, the positive and the negative sentences following them. One way of formally accounting for this is by saying that these sentences are syntactically present but elided if these polar elements are used independently; See Van Craenenbroeck (2010:ch.15) for related discussion. There are other reasons for assuming this. First, the examples in (22) show that polar ja and nee can easily be coordinated with full sentences; this would immediately follow if the polar elements are followed by phonetically empty sentences.

Secondly, polar ja and nee can also be combined with a non-main clause, as shown by the answers to the question in (23a); this again follows if they are followed by phonetically empty sentences.

Thirdly, if we are indeed dealing with ellipsis, we would expect it to be subject to the recoverability condition on deletion: elements that cannot be recovered from the context must be overtly expressed. The examples in (24) suggest that this is indeed the case; see Pope (1971) for a larger sample of English examples. Note in passing that there is little or no reason for assuming that waarschijnlijk morgen and natuurlijk niet are constituents, which gives the ellipsis analysis even greater credibility.

Note that the overtly expressed remnant need not be a clausal constituent. The second sentence in (25a) shows that it can also be a complementizer: as is shown by the near paraphrase in (25b), polar ja corresponds to the italicized part of the conditional clause introduced by the complementizer indien'if'.

Finally, the hypothesis of ellipsis may also account for the fact that the preferred answer to the negative question in (26a') is jawel: denying a negative proposition generally favors the presence of the affirmative marker wel. As contradicting a negative presupposition also favors the presence of wel, ellipsis would also account for the use of jawel in (26b').

      This discussion has shown that polar ja and nee are not primarily used to affirm or to deny a presupposition, but “agree” with a (possibly elided) positive or negative clause following it. This makes it understandable that these elements can also be used in contexts like (27), in which the hearer simply accepts the truth of the (negative/positive) propositions in the primeless examples.

Our brief (and incomplete) discussion of the use of the polar elements ja'yes' and nee'no' shows that they have the hallmark of pragmatic markers in that they do not carry truth-conditional content: they simply indicate that the (possible phonetically empty) clauses they are associated with are positive or negative. In this respect they differ from the polar elements welles and nietes in (28), which are typically used to contradict or refute a proposition in the immediate preceding discourse.

Like polar ja and nee, welles and nietes seem to have the status of a full clause: this is clear from the fact illustrated in the mini-dialogue in (28) that they can be in a coordinate structure with another clause. It is, however, not clear whether these elements are associated with an elided clause as this clause cannot be made explicit. Furthermore the examples in (29) show that it is not possible to combine these elements with embedded clauses. Note in passing that welles and nietes also differ in this respect from the affirmative marker wel and the negative adverb niet'not' in the two mini-dialogues in (29); this is not surprising as we are probably dealing with reduced clauses in that case: Jij komt wel/niet'you will/wonʼt'.

The discussion above has shown that the polar elements ja and nee are pragmatic markers: they do not have truth-conditional content. We also noted that this is less clear in the case of the polar elements welles and nietes.

Discourse particle are sometimes claimed to express the speaker’s emotional attitude towards some discourse aspect. Curses, for example, can be used to indicate what the speaker’s feelings are toward a specific state of affairs (Verdomme, wat een regen!'Damn! It's pouring!) or a specific behavior of the addressee (Jezus, man, dat pik ik niet van je!'Jesus, man, I wonʼt take that from you'). Many pragmatic markers have an additional emotional value. In answering a question such as Kom je morgen? in (30a), modal adverbs like natuurlijk'of course' and vanzelfsprekend'obviously' also have a certain emotional load in addition to expressing mere confirmation. However, it should be noted that the same emotional load is present in sentences such as Natuurlijk kom ik morgen'Of course I will come tomorrow', which can also be used as answers to this question. If the discourse particle natuurlijk is actually a reduced clause, its emotional load need not surprise us. A similar analysis may be given for speaker-oriented adverbs such as helaas in (30b).

This section has discussed the use of a set of sentence-external pragmatic markers, which have received a lot of attention since Schiffrin (1987) in, especially, the pragmatic literature. These markers are characterized by the fact that they often do not have clear truth-conditional content but instead perform a wide range of pragmatic functions; they are indexical in that they point to some aspect of the discourse: discourse participants, state of affairs holding at speech time, earlier statements, etc. Discourse markers may have various shapes: they can simply be sounds like mmm, mhm and ooo, lexical elements such as the interjection hé, phrasal expressions like mijn god'my god', or (reduced) conventional stock clauses like Ik bedoel'I mean'. Furthermore, we have seen that certain pragmatic markers like the polar element ja and nee may be inherently associated with a (potentially elided) sentence. Other pragmatic markers may be part of partially elided clauses, e.g. Dat is goed: ik kom morgen'O.K., Iʼll come tomorrow'. The examples above show that many pragmatic markers have arisen as a result of grammaticalization; this also seems to hold for a set of markers that can also be used as adverbial phrases such as helaas'unfortunately'. As a result of this, we find cases like dus'so' and toch'all the same' with a less clear status; see, e.g., Aijmer (2002) and Diewald (2009) on grammaticalization, and Evers-Vermeul (2005/2010) and Degand (2011) for specific case studies of Dutch. As pragmatic markers are characteristic of discourse and thus not part of syntax in the restricted sense defined in Section IV of the preface, we will not digress any further here.