• Dutch
  • Frisian
  • Afrikaans
Show full table of contents
5.1.2.4.Reported speech
quickinfo

The complement clauses discussed in the preceding sections all have the form of finite embedded clauses, that is, they are introduced by a complementizer ( dat'that' or of'whether') or a wh-phrase, and have the finite verb in clause-final position. Complement clauses of this kind are also found in sentences such as (123a), in which the speaker reports what someone else has said, thought, etc. The sentences in (123b&c) show, however, that there are also alternative ways.

Example 123
a. Jan zei/dacht dat hij ziek was.
indirect reported speech/quote
  Jan said/thought  that  he  ill  was
  'Jan said/thought that he was ill.'
b. Jan zei/dacht: "Ik ben ziek".
direct reported speech/quote
  Jan said/thought    I  am  ill
  'Jan said/thought: "Ik ben ziek".'
c. Jan zei/dacht hij was ziek.
semi-direct reported speech/quote
  Jan said/thought  he  was  ill

Although the examples in (123) show that constructions like these are not strictly limited to speech proper but may also pertain to thoughts, they are normally said to involve reported speech. We will therefore refer to the whole set of constructions as reported speech constructions, and to the parts in italics, which express the reported parts, as quotes. Although quotes are often analyzed as direct object clauses (see, e.g., Haeseryn et al. 1997), we will see that this is not entirely correct for all cases: see also Corver (1994), Corver & Thiersch (2003), and De Vries (2006). For this reason we will refer to the clauses headed by the verb of saying/thinking not as matrix clauses but, more neutrally, as say-clauses.
      The way of reporting speech in (123a) is normally referred to as indirect reported speech. An important property of this construction is that the embedded clause(s) does/do not necessarily correspond in a one-to-one fashion to the actual utterance(s) of the source indicated: for example, if Jan is a very talkative person, the embedded clause may simply summarize a story that took 30 minutes to tell, that is, example (123a) does not imply that Jan literally said: "Ik ben ziek". This distinguishes indirect from direct reported speech; example (123b) is only true if Jan pronounced the sentence Ik ben ziek, for which reason we repeated this sentence literally in the translation of (123b). Another difference, which is illustrated in (124), is that direct quotes can consist of a sequence of independent sentences, whereas in indirect reported speech constructions each assertion must be realized as a separate dependent clause.

Example 124
a. Jan zei/dacht [[dat hij ziek was] en [dat hij thuis bleef]].
indirect
  Jan said/thought    that he ill  was  and   that he  at.home  stayed
  'Jan said that he was ill and that he would stay at home.'
b. Jan zei/dacht: "Ik ben ziek. Ik blijf thuis".
direct
  Jan said/thought    I  am  ill  stay  at.home
  'Jan said: "Ik ben ziek. Ik blijf thuis".'

In example (123c), we are dealing with semi-direct reported speech (also known as erlebte rede), which constitutes a kind of in-between category. It differs from indirect reported speech in that the quote has the form of a main clause. This is clear from the position of the finite verb: if we are dealing with indirect reported speech, the finite verb should occupy the clause-final position, whereas it should be in second position in semi-direct reported speech. The placement of the finite verb is clearly related to the distribution of the complementizer: the examples in (125) show that the complementizer is obligatory in indirect reported speech constructions with declarative quotes, but that it cannot appear in semi-direct reported speech constructions. This also shows that semi-direct reported speech constructions such as (125b) cannot be derived from direct reported speech constructions such as (125a) by deletion of the complementizer dat, but that they constitute a construction type in their own right.

Example 125
a. Jan zei *(dat) hij ziek was.
indirect
  Jan said    that  he  ill  was
  'Jan said that he was ill.'
b. Jan zei (*dat) hij was ziek.
semi-direct
  Jan said     that  he  was  ill

Although semi-direct reported speech does not involve a literal quote, it differs from indirect reported speech in that the relation with what was actually said is much tighter. Example (123c), for instance, suggests that Jan said something like Ik ben ziek. Semi-direct quotes differ from direct quotes mainly in that first and second person pronouns are replaced by third person pronouns and that the present tense of the reported sentence is adapted to conform to the past tense of the verb zeggen'to say'; cf. Lodewick (1975:169-70). The semi-direct equivalent of the direct reported speech construction in (126a) would then be as in (126b).

Example 126
a. Jan dacht: "Ik haat je uit de grond van mijn hart".
direct
  Jan thought    I  hate  you  from the bottom of my heart
b. Jan dacht, hij haatte hem uit de grond van zijn hart.
semi-direct
  Jan said  he  hated  him  from the bottom of his heart

      Semi-direct reported speech is not often used in colloquial speech but is regularly found as a stylistic device in modern literature, especially for expressing the internal thoughts of the protagonist(s) of a story (the so-called interior monologue); Lodewick in fact claims that it is a characteristic feature of impressionistic writings from around 1900. The use of semi-direct reported speech constructions implies that, like direct quotes, semi-direct quotes may involve sequences of sentences; this expectation is borne out, as is illustrated in (127b) by means of the semi-direct counterpart of the direct reported speech construction in (124b), repeated here as (127b).

Example 127
a. Jan zei/dacht: "Ik ben ziek. Ik blijf thuis".
direct
  Jan said/thought    I  am  ill  I stay  at.home
  'Jan said: "Ik ben ziek. Ik blijf thuis".'
b. Jan zei/dacht, hij was ziek, hij bleef thuis.
semi-direct
  Jan said/thought  he  was  ill  he  stayed  at.home

      Embedded clauses in indirect reported speech constructions such as (123a) can be pronominalized (Jan zei het'Jan said it'), which suggests that they function as direct object clauses. It is often assumed without much argumentation that direct and semi-direct reported speech constructions like (123b&c) also involve direct object clauses; see Haeseryn et al. (1997:1100). This is, however, far from obvious: the quotes in the two examples in (127) consist of series of sentences, and this makes it is very unlikely that quotes have the function of direct object. In fact, it may even be the case that we are dealing with a relation of an entirely different sort given that the part Jan zei need only be used in examples like (123b&c) when the context leaves open what the source of the quote is; if the source is known, it can readily be omitted. This is illustrated in the little scene in (128), which might be used as the start of a story. See also the discussion of what Huddleston and Pullum (2002: 1029) call free indirect/direct speech.

Example 128
Jan kwam in zijn pyjama de kamer binnen.
  Jan came  in his pajamas  the room  inside
'Jan entered the room in his pajamas.'
a. (Hij dacht:) "Ik ben ziek. Ik blijf thuis".
direct
  he thought    I  am  ill  stay  at.home
  '(He thought:) "Ik ben ziek. Ik blijf thuis".'
b. (Hij dacht,) hij was ziek; hij bleef thuis.
semi-direct
  he thought  he  was  ill  he  stayed  at.home
  '(He thought,) he was ill. He would stay in.'

Matrix clauses in indirect reported speech constructions, on the other hand, can only be left unexpressed under very special circumstances. Sentence (129b), for example, cannot replace the continuations of the story in (128a&b), but is only acceptable as an answer to a question such as (129a)—we are dealing with some kind of ellipsis; the part of the answer that can be recovered from the original question (here: the matrix clause) is simply omitted.

Example 129
a. Wat zei Jan?
  what  said  Jan
  'What did Jan say?'
b. Dat hij ziek was en dat hij thuis bleef.
indirect
  that  he  ill  was and  that  he  at.home  stayed
  'That he was ill and that he would stay in.'

We will see in Subsection II that this difference is reflected in several other ways, and that there are reasons for assuming that in many cases direct and semi-direct reported speech constructions are not regular transitive constructions. Instead, the quotes function as full-fledged sentences with parenthetical say-clauses.
      We already mentioned that semi-direct reported speech is normally used in written language and cannot be found in colloquial speech so frequently, subsection III will show, however, that there is also a reported speech construction that is normally avoided in writing but which is highly frequent in speech; cf. Verkuyl (1977) and Romein (1999). This construction, which is illustrated in (130), involves the quotative preposition van followed by an intonation break, which may optionally be preceded by a hesitation marker like eh'er', and a quote. The quote can be either direct or, less frequently, indirect; cf. Verkuyl (1977).

Example 130
a. Marie dacht van (eh) ... Hij komt straks wel weer terug.
  Marie thought  van  er  he comes  later  prt  again  back
  'Marie thought something like: "Heʼll probably return later again".'
b. Marie dacht van (eh) ... dat hij straks wel weer terug komt.
  Marie thought  van   er  that  he  later  prt  again  back  comes

The three types of reported speech constructions introduced above will be discussed in separate subsections, subsection I discusses indirect reported speech and shows that the indirect quote functions as a regular argument clause, subsection II continues with a discussion of (semi-)direct reported speech and argues that the say-clause in such constructions is often (but not always) parenthetical, subsection III concludes with a discussion of the colloquial quotative van-construction in (130).

readmore
[+]  I.  Indirect reported speech

Quotes in indirect reported speech constructions behave in many respects like other types of direct object clauses. The following subsections will show this for a number of properties of object clauses, which are discussed more extensively in Sections 5.1.2.1 to 5.1.2.3. We will also discuss some facts not mentioned there, which can be used to provide support for the claim that indirect quotes are regular object clauses.

[+]  A.  Selection restrictions on the embedded clause

The form of indirect quotes is determined to a large extent by the main verb: verbs like zeggen'to say' and denken'to think' select declarative clauses, whereas verbs like vragen'to ask' select interrogative clauses. See Section 5.1.2.1 for a more extensive discussion of the selection restrictions on declarative and interrogative object clauses.

Example 131
a. dat Peter zei/dacht [dat Jan ziek was].
  that  Peter said/thought   that  Jan ill  was
  'that Peter said/thought that Jan was ill.'
b. dat Marie vroeg [of Jan ziek was].
  that  Marie asked  whether  Jan ill  was
  'that Marie asked whether Jan was ill.'
[+]  B.  Position of the embedded clause

The examples in (131) show that indirect quotes normally follow the verb(s) in clause-final position—placing such quotes in the middle field is normally marked and triggers a factive reading; cf. Section 5.1.2.3. Topicalization of indirect quotes is possible, in which case they are optionally followed by the resumptive pronoun dat'that'; see Section 5.1.2.2 for a more extensive discussion of the placement of direct object clauses.

Example 132
a. [Dat Peter ziek is] (dat) zei/dacht Jan.
  that  Peter ill  is   that  said/thought  Jan
b. [Of Jan ziek was] (dat) vroeg Marie.
  whether  Jan ill  was   that  asked  Marie
[+]  C.  The use of an anticipatory pronoun

The use of an anticipatory pronoun seems possible but marked; the examples in (133) are more likely to be construed with a regular, discourse-related interpretation of the pronoun, which again favors a factive reading of the embedded clause; see Section 5.1.2.3, sub IIIB.

Example 133
a. dat Peter het zei/dacht [dat Jan ziek was].
  that  Peter it  said/thought   that  Jan ill  was
  'that Peter said/thought it that Jan was ill.'
b. dat Marie het vroeg [of Jan ziek was].
  that  Marie it  asked  whether  Jan ill  was
  'that Marie asked it whether Jan was ill.'
[+]  D.  Wh-extraction

Embedded declarative clauses are fully transparent for wh-extraction in the sense that both arguments and adjuncts can be extracted. See Section 5.1.1, sub III, for discussion of the fact that wh-extraction becomes unacceptable if an anticipatory or deictic pronoun is added.

Example 134
a. Wiei zei/dacht je [dat ti dat boek gekocht had]?
subject
  who  said/thought  you   that  that book  bought  has
  'Who did you say/think had bought that book.'
b. Wati zei/dacht je [dat Peter ti gekocht heeft]?
object
  what  said/thought  you   that  Peter  bought  has
  'What did you say/think that Peter has bought?'
c. Wanneeri zei/dacht je [dat Peter ti vertrokken was]?
adjunct
  when  said/thought  you   that  Peter  left  had
  'When did you say/think that Peter had left?'

Wh-extraction is not possible from embedded interrogative clauses. The standard analysis in generative grammar is that this is due to the fact that wh-extraction cannot apply in one fell swoop but must proceed via the clause-initial position of the object clause; this position is available in declarative examples such as (134), but occupied by a wh-phrase in embedded wh-questions such as (135) or a phonetically empty question operator in embedded yes/no-questions.

Example 135
a. * Wiei vroeg je [watjtitj gekocht heeft]?
subject
  who  asked  you  what  bought  has
  Compare: '*Who did you ask what has bought?'
b. * Watj vroeg je [wieititj gekocht heeft]?
object
  what  asked  you  who  bought  has
  Compare: '*What did you ask who has bought?'
c. * Wanneerj vroeg je [wieititj vertrokken was]?
adjunct
  when  asked  you   who  left  had
  Compare: '*When did you ask who had left?'

Note in passing that, contrary to what has been reported for English, wh-extraction of the subject in (134a) is acceptable despite the presence of a complementizer and that most Dutch speakers find the three examples in (135) equally unacceptable. We will not digress on these issues here but refer the reader instead to Section 11.3.1 for an extensive discussion of the restrictions on wh-extraction. Note also that example (135c) is fully acceptable if wanneer'when' is construed as a modifier of the matrix clause but this is, of course, not the reading intended here (as is indicated by the placement of the trace tj within the embedded clause).

[+]  E.  Binding

Referential personal pronouns as part of an indirect quote can be bound by an antecedent in the say-clause; see Section N5.2.1.5, for an extensive discussion of binding of such pronouns. Since such pronouns can also co-refer with some referential expression as a result of accidental coreference, we have to appeal to examples in which the antecedent is a quantified expression like iedereen'everyone' or niemand'nobody' in order to show this. Example (136a) first shows that the pronoun hij cannot be used as referentially dependent on a universally/negatively quantified expression if the latter is part of some other sentence; in such cases, the pronoun must refer to some known entity in the domain of discourse. The fact that the pronoun can have a bound-variable reading, that is, can be interpreted as referentially dependent on the quantifiers in (136b) shows that we are not dealing with accidental coreferentiality but with binding. Italics indicate the intended binding relation.

Example 136
a. * Iedereen/ Niemand bleef thuis. Hij was ziek.
  everybody/nobody   stayed  at.home  he  was  ill
b. Iedereen/ Niemand zei [dat hij ziek was].
  everybody/nobody  said   that  he  ill  was
  'Everybody/Nobody said that he was ill.'

The acceptability of the bound variable reading in (136b) unambiguously shows that we are dealing with an object clause; if the indirect quote were not the object of the verb zeggen'to say', there would be no c-command relation between the subject of the say-clause and the pronoun and, consequently, binding would be wrongly predicted to be impossible, just as in (136a).

[+]  F.  Licensing of negative polarity items

That indirect quotes are object clauses is also shown by the fact that negative polarity items (NPIs) like ook maar iets'anything' as part of an indirect quote can be licensed by some negative element in the say-clause. The reason is that, like binding, NPI licensing requires c-command between the NPI and its licenser. NPI-licensing is excluded in (137a) since the NPI and its potential licenser niemand'nobody' are not in the same sentence and there is consequently no c-command relation between them; NPI-licensing is possible in (137b) since the subject of the matrix clause does c-command the NPI in the embedded object clause. Italics indicate the relation between the NPI and its intended licenser.

Example 137
a. Niemand bleef thuis. *Hij had daar ook maar iets te doen.
  nobody  stayed  at.home    he  had  there  anything  to do
b. Niemand dacht dat hij thuis ook maar iets te doen had.
  nobody  thought  that  he  at.home  anything  to do  had
  'Nobody thought that he had anything to do at home.'
[+]  G.  Conclusion

The previous subsections have shown that quotes in indirect reported speech constructions are direct object clauses. They exhibit the behavior of regular object clauses, which was discussed more extensively in Sections 5.1.2.1 to 5.1.2.3. Additionally, the discussion of binding and NPI-licensing has established that subjects of say-clauses c-command the constituents in indirect quotes, which lends credence to the claim that such quotes are regular direct object clauses.

[+]  II.  Direct and semi-direct reported speech

This subsection discusses the question as to whether (semi-)direct quotes should be considered direct object clauses, subsections A and B show that the evidence is rather varied, from which we will conclude that (semi-)direct reported speech constructions are often structurally ambiguous, subsection C provides some additional support for this conclusion, and Subsection D concludes with a brief note on the internal structure of the relevant constructions. Since (semi-)direct reported speech constructions have not yet been studied extensively from a syntactic point of view, much of what follows is tentative in nature and should therefore be taken with care.

[+]  A.  Direct reported speech

Direct reported speech constructions are often ambiguous. We will argue that such constructions allow not only an analysis as regular transitive constructions in which the quote functions as a direct object, but also an analysis in which the quote can function as a main clause with an embedded parenthetical say-clause; cf. De Vries (2006).

[+]  1.  Are direct quotes direct objects?

Example (138a) strongly suggests that the direct quote in (138b) functions as the direct object of the verb zeggen'to say'. The fact that the pronoun in (138a) cannot be omitted shows that zeggen is a transitive verb that cannot be used pseudo-intransitively. The fact that the direct quote is the only candidate that could function as direct object in (138b) therefore seems to leave us no other option than to conclude that it must have this syntactic function.

Example 138
a. Jan zei *(het).
  Jan said     it
b. Jan zei: "Ik ben ziek".
  Jan said    I  am  ill

Although this line of argumentation seems quite convincing, there are various reasons to reject the conclusion that direct quotes always function as object clauses. First, it seems that introducing the direct quote with an anticipatory/deictic pronoun het'it' is not normally possible. Although example (139a) is fully acceptable, the pronoun het does not seem to refer to the direct quote but to some other proposition. This is evident from the fact illustrated in (139b) that the pronoun can be replaced by an indirect quote such as the one in square brackets. Besides, example (139c) shows that we would rather use phrases like als volgt'as follows' or the manner adverb zo'thus' if we want to anticipate the direct quote.

Example 139
a. Jan vroeg het haar eindelijk: "Als ik je zie begint mijn hart te bonken: boem, boem, boem ... Ik kan niet langer zonder jou!"
  'Jan finally asked her it: "Whenever I see you my heart starts pounding boom, boom, boom ... I can no longer live without you!"'
b. Jan vroeg haar eindelijk [of ze met hem wilde trouwen]: "Als ik je zie begint mijn hart te bonken: boem, boem, boem ... Ik kan niet langer zonder jou!"
  'Jan finally asked her whether she would marry him: "Whenever I see you my heart start pounding boom, boom, boom ... I can no longer live without you!"'
c. Jan vroeg het haar als volgt/zo: "Als ik je zie begint mijn hart te bonken: boem, boem, boem .... Ik kan niet langer zonder jou!"
  'Jan finally asked her it as follows/thus: "Whenever I see you my heart starts pounding boom, boom, boom ... I can no longer live without you!"'

From the discussion of the examples in (139) we are forced to conclude that the direct quote does not function as a direct object in the examples in (139). Barbiers (2000:190) even suggests that postverbal direct quotes are not even part of the preceding say-clauses given that their intonation contour is entirely independent; they are always preceded by a distinct intonation break. He suggests that this makes it more likely that postverbal direct quotes function as some kind of afterthought since afterthoughts exhibit the same prosodic effect. Barbiers does not claim that direct quotes are never direct objects, but he asserts that they can only have this function if they occur in the middle field of the clause, as in (140), in which case they have the same distribution as nominal objects. Note in passing that examples such as (140) quickly degrade when the quote gets longer.

Example 140
a. Jan heeft "hallo" tegen de leraar gezegd.
  Jan has  hello  to the teacher  said
  'Jan has said "hallo" to the teacher.'
b. Jan heeft "ik ben ziek" tegen de leraar gezegd.
  Jan has    I  am  ill  to the teacher  said
  'Jan has said "Ik ben ziek" to the teacher.'

In (140) it is not entirely clear whether we are really dealing in (140) with reported speech in the sense intended here. It may also be the case that we simply have to do with an autonomous use of the word/phrase in question. That this may be the case is strongly suggested by the fact that an utterance such as (140a) can quite naturally be followed by something such as (141a). De Vries (2006) provides a similar example and adds that the quote can also be in a language other than Dutch. This again suggests that quotes may involve the autonomous use of the word/phrase in question, and that this is the reason why they behave syntactically as nominal arguments of the verb. In the discussion below we will ignore the autonomous use of quotes in the middle field of the clause.

Example 141
a. Dat is onbeleefd: hij had "goedemorgen" moeten zeggen.
  that is rude he should    good.morning  have  said
  'That is rude: he should have said "goedemorgen".'
b. John heeft "I am ill" tegen de leraar gezegd.
  John has    I am ill  to the teacher  said
  'John has said "Iʼm ill" to the teacher.'

      Barbiers does not discuss direct quotes in the left periphery of the utterance, as in (142), but it seems that such constructions show that direct quotes have an ambiguous syntactic status. Although the construction in (142a) is the one commonly used, the examples in (142b&c) show that it is also possible to add the demonstrative pronoun dat or the manner adverb zo as a resumptive element.

Example 142
a. "Ik ben ziek", zei Jan.
  am  ill  said  Jan
  '"Ik ben ziek", Jan said.'
b. "Ik ben ziek", dat zei Jan.
  am  ill  that  said  Jan
c. "Ik ben ziek", zo zei Jan.
  am  ill  thus  said  Jan

Subsection B will show that example (142b) can be analyzed as a left-dislocation construction. This example would then receive a similar analysis as example (143a) in which the resumptive pronoun dat has a neuter singular antecedent functioning as the logical direct object of the sentence. Example (143b) is added to show that other resumptive pro-forms are used when the left-dislocated element has some other logical function: the resumptive pro-form dan, for example, is used when the left-dislocated element is the temporal adverb morgen'tomorrow'.

Example 143
a. Dat boek, dat heb ik al gelezen.
  that book  that  have  already  read
  'That book, Iʼve already read it.'
b. Morgen, dan ga ik naar Groningen.
  tomorrow  then  go  to Groningen
  'Tomorrow, Iʼll be going to Groningen then.'

Although Subsection B will argue that (142c) is not a left-dislocation construction, the fact that the manner adverb zo is used in a similar resumptive function immediately suggests that the direct quote does not function as the logical direct object of the say-clause. This conclusion receives further support from (144). Example (144a) shows that the left-dislocation construction with the resumptive pronoun dat does not allow the addition of the object pronoun het, which is to be expected given that the resumptive pronoun already performs this function. Example (144b), on the other hand, shows that, in the right context, the addition of the object pronoun het is admissible in the construction with zo, which proves that the direct quote does not function as the logical direct object of the say-clause in this case.

Example 144
a. "Ik ben ziek", dat zei Jan *(het).
  am  ill  that  said  Jan     it
b. "Ik ben ziek", zo zei Jan (het).
  am  ill  thus  said  Jan   it

The fact that direct quotes need not function as (logical) direct objects of the say-clause, established by the examples in (139) and (142) to (144), shows that our earlier conclusion on the basis of example (138a) that the verb zeggen'to say' may not occur without a direct object is wrong; if a direct quote is present with some other function than (logical) direct object of the say-clause, the direct object of the verb zeggen can apparently remain unexpressed.
      To sum up, this subsection has provided evidence that direct quotes preceded by a say-clause do not function as the (logical) direct object of this say-clause. The situation is different when the say-clause follows the quote; the quote may then have the function of (logical) direct object, in which case the resumptive pronoun dat can be inserted between the quote and the finite verb, or it may have an adverbial function, in which case the resumptive pro-form surfaces as the manner adverb zo. Observe that this conclusion raises the question as to how the selection restrictions imposed by the matrix verbs on the direct quote can be accounted for if the latter functions as an adjunct. Given that this cannot be accounted for by normally assumed syntactic means (that is, subcategorization), a pragmatic account seems to be called for. We leave this for future research.

Example 145
a. Jan zei/*vroeg: "Els wil vast wel een ijsje".
  Jan  said/asked    Els  wants  prt  prt  an ice.cream
  'Jan said: "Iʼm sure Els would like to have an ice cream".'
b. Jan vroeg/*zei: "Wie wil er een ijsje?".
  Jan asked/said    who  wants  there  an ice.cream
  'Jan asked: "Who would like to have an ice cream?".'
[+]  2.  Direct quotes and parenthetical clauses

The previous subsection has shown that direct quotes can but need not function as direct object clauses of verbs of saying/thinking when they precede the say-clause. The following question now arises: what is the structure of those constructions in which the quote does not function as direct object? This subsection argues that direct quotes are regular main clauses in such cases, which contain a parenthetical say-clause. A first step in the argument involves the possible word orders in the three constructions in (146).

Example 146
a. "Peter zal het boek morgen brengen", zei Marie.
  Peter will  the book  tomorrow  bring  said Marie
  '"Peter zal het boek morgen brengen", Marie said.'
b. "Peter zal het boek morgen brengen", dat zei Marie.
  Peter will  the book  tomorrow  bring  that  said  Marie
c. "Peter zal het boek morgen brengen", zo zei Marie.
  Peter will  the book  tomorrow  bring  thus  said  Marie

We begin our discussion with example (146b), which we analyze as a left-dislocation construction. Example (147a) shows that the direct quote need not precede the say-clause but can also be right-dislocated, in which case the resumptive pronoun dat will be replaced by the proximate demonstrative pronoun dit'this'. The example which is crucial for our discussion is (147b), which shows that the direct quote cannot be split by the say-clause.

Example 147
a. Marie zei dit: "Peter zal het boek morgen brengen".
  Marie said  this:    Peter will  the book  tomorrow  bring
  'Marie said the following: "Peter zal het boek morgen brengen".'
b. * "Peter", dat/dit zei Marie, "zal het boek morgen brengen".
  Peter  that/this  said  Marie    will  the book  tomorrow  bring

We should keep in mind, however, that reliable judgments on examples such as (147b) are sometimes hampered by the fact that the same string is acceptable with a non-quote interpretation: the speaker then simply provides a statement of his own and uses a parenthetical clause to point at Marie as his source of information. This is brought out in example (148a), in which the adverb tenminste'at least' forces the intended non-quote reading. Example (148b) shows that the parenthetical clause cannot appear in a position preceding the constituent in sentence-initial position (here: Peter).

Example 148
a. "Peter", dat zei Marie tenminste, "zal het boek morgen brengen".
  Peter  that  said  Marie  at.least   will  the book  tomorrow  bring
  'According to Marie at any rate, Peter will bring the book tomorrow.'
b. * Marie zei dat/dit tenminste, "Peter zal het boek morgen brengen".
  Marie  said  that/this  at.least    Peter will  the book  tomorrow  bring

Putting aside the non-quote reading, we are forced to conclude that the construction in (146b) with resumptive dat differs sharply from the construction in (146a) without a resumptive pronoun. The examples in (149) bear out that in the latter case the direct quote can be split in various places by the say-clause. The examples in (150) show that the same thing holds for construction (146c) with zo.

Example 149
a. "Peter zal het boek morgen brengen", zei Marie.
  Peter will  the book  tomorrow  bring  said Marie
  'Peter will bring the book tomorrow, Marie said.'
b. "Peter", zei Marie, "zal het boek morgen brengen".
c. "Peter zal", zei Marie, "het boek morgen brengen".
d. "Peter zal het boek ", zei Marie, "morgen brengen".
e. ? "Peter zal het boek morgen", zei Marie, "brengen".
Example 150
a. "Peter zal het boek morgen brengen", zo zei Marie.
  Peter will  the book  tomorrow  bring  thus  said Marie
b. "Peter", zo zei Marie, "zal het boek morgen brengen".
c. "Peter zal", zo zei Marie, "het boek morgen brengen".
d. "Peter zal het boek ", zo zei Marie, "morgen brengen".
e. ? "Peter zal het boek morgen", zo zei Marie, "brengen".

The fact that the direct quotes can be split in (149) and (150) suggests that we are dealing with parenthetical constructions. A potential problem is that example (151a) shows that the presumed parenthetical say-clause in (149) may also precede the quote; this is unexpected as example (148b) has shown that parenthetical clauses cannot do so. However, there seems to be more to this than meets the eye given that the say-clause in (150) behaves as expected and is indeed unable to precede the quote: example (151b) is only acceptable if the sentence contains an object pronoun like het.

Example 151
a. Marie zei: "Peter zal het boek morgen komen brengen".
  Marie said  Peter will  the book  tomorrow  come  bring
b. Marie zei *(het) zo: "Peter zal het boek morgen komen brengen".
  Marie said   it  thus    Peter will  the book  tomorrow  come  bring

The fact that the addition of het to the examples in (150) is unusual, to say the least, suggests that (149) and (150) involve constructions entirely different from (151); whereas the former involve parenthetical say-clauses, the say-clauses in the latter may be regular transitive main clauses.
      If we are indeed concerned with parenthetical clauses in (149) and (150), we expect to find a wider range of examples that do not involve verbs of saying/thinking. This expectation is borne out; in fact, writers have created an infinite number of variations on this theme. A number of rather conventional examples are given in (152). Note that the quotes cannot be analyzed as arguments of the verbs beginnen'to start', vervolgen'to continue', and besluiten'to conclude' in these examples: these verbs already have a direct object, zijn verhaal'his story'; see De Vries (2006) for a number of less conventional examples.

Example 152
a. "De wind", (zo) begon hij zijn verhaal, "was stormachtig".
  the wind  thus  started  he  his story    was tempestuous
  '"The wind", (thus) he started his story, "was tempestuous".'
b. "De boot" (zo) vervolgde hij zijn verhaal, "was in gevaar".
  the boat  thus  continued  he  his story    was in danger
  '"The boat", (thus) he continued his story, "was in danger".'
c. "De schipper", (zo) besloot hij zijn verhaal, "spoelde dood aan".
  the skipper  thus  concluded  he  his story    washed  dead  ashore
  '"The skipper", (thus) he concluded his story, "washed ashore dead".'

In order to give an impression of the semantic verb types that can be used in parenthetical say-clauses, we provide a small sample in (153), adapted from De Vries (2006). Note that this list includes a number of intransitive verbs like schreeuwen'to shout', which provides further support for the claim that a direct quote does not function as an argument of the main verb in parenthetical say-clauses.

Example 153
a. Saying, thinking and writing: antwoorden'to answer', denken'to think', prediken'to preach', schrijven'to write', vertellen'to tell', vragen'to ask', zeggen'to say'
b. Manner of speech and sound emission, schreeuwen'to shout', vloeken'to curse', zuchten'to sigh', giechelen'to giggle', schateren'to roar', trompetteren'to trumpet', sissen'to hiss', zingen'to sing'
c. Thinking, observation and explanation: concluderen'to conclude', denken'to think', fantaseren'to fantasize', opmerken'to observe', peinzen'to contemplate', verduidelijken'to clarify'
[+]  B.  Semi-direct reported speech

Semi-direct reported speech constructions exhibit more or less the same syntactic behavior as their direct counterparts. The direct reported speech constructions in (139), for instance, can easily be transformed into the semi-direct reported speech constructions in (154). It shows that, like direct quotes, semi-direct quotes need not function as direct objects of the verb zeggen'to say'. It may therefore be the case that in examples such as (154) the direct quote is actually not part of the first sentence, but consists of a series of independent sentences.

Example 154
a. Jan vroeg het haar eindelijk. Als hij haar zag, begon zijn hart te bonken: boem, boem, boem ... Hij kon niet langer zonder haar!
  'Jan finally asked her it. Whenever he saw her his heart started pounding boom, boom, boom ... He could no longer live without her!'
b. Jan vroeg haar eindelijk of ze met hem wilde trouwen. Als hij haar zag, begon zijn hart te bonken: boem, boem, boem ... Hij kon niet langer zonder haar!
  'Jan finally asked her whether she would marry him. Whenever he saw her his heart started pounding boom, boom, boom ... He could no longer live without her!'
c. Jan vroeg het haar als volgt/zo. Als hij haar zag, begon zijn hart te bonken: boem, boem, boem ... Hij kon niet langer zonder haar!