Taalportaal - the digital language portal

Dutch
Frisian
Saterfrisian
Afrikaans

Show all

8.1.3.Object noun phrases in the middle field of the clause: Scrambling

quickinfo

Generally, nominal objects can occupy various positions in the so-called middle field of the clause, that is, that part of the clause bounded to the left by the C(omplementizer)-position, which is filled by the complementizer in embedded clauses and by the finite verb in main clauses, and bounded to the right by the verbs in clause-final position (if present). This variation in word order especially relates to the position of the nominal object relative to adverbial phrases of various sorts: for instance, the noun phrase zijn auto'his car' in (24) can either follow or precede the modal adverb waarschijnlijk'probably'.

a.	dat	Jan waarschijnlijk	zijn auto	verkoopt.
	that	Jan probably	his auto	sells
	'that Jan will probably sell his car.'

a'.

dat Jan zijn auto waarschijnlijk verkoopt.

b.	Jan heeft	waarschijnlijk	zijn auto	verkocht.
	Jan has	probably	his car	sold
	'Jan probably sold his car.'

b'.

Jan heeft zijn auto waarschijnlijk verkocht.

Since the direct object is generally assumed to be base-generated within the VP, it is expected to be adjacent to the main verb, as in the primeless examples in (24). In order to account for the word orders in the primed examples, it has been assumed that Dutch has a scrambling rule that may move the arguments of the verb from their VP-internal base-position into a position preceding the adverbs. The structures of the primed examples in (24) are therefore assumed to be as indicated in (25).

[ ... C ... DPi ... ADV ... [VP ... ti V]].

dat Jan zijn autoi waarschijnlijk [VPti verkoopt].

Jan heeft zijn autoi waarschijnlijk [VPti verkocht].

Actually, it can be argued that there are various types of scrambling (cf., e.g., Neeleman 1994b); for example, there is a rule of Focus movement, which optionally places emphatically or contrastively focused phrases, and a rule of Neg-movement (Haegeman 1995), which obligatorily places negative phrases into a more leftward position in the middle field. We will not discuss these two movement types here, given that they are not restricted to nominal objects, but restrict our attention to the type of scrambling in (25a), which is limited to nominal objects and can be recognized by the fact that the moved phrase is never accented. We will see that word order variations like those in (24) are typically related to the information structure of the clause: scrambled noun phrases normally belong to the presupposition (“old” information) whereas noun phrases that are not scrambled are instead part of the focus (“new” information) of the clause. Other effects of scrambling may be that the moved noun phrase is assigned a special (e.g., generic or partitive) meaning.

I. Clauses containing a clausal adverb
II. Clauses containing a VP adverb
III. Clauses containing an adverb to the left of a clausal adverb
IV. Clauses containing the negative adverb niet'not'
V. Conclusion

[+] I. Clauses containing a clausal adverb

This subsection discusses the distribution of nominal objects in clauses containing a clausal adverb such as modal adverbs or adverbs of frequency. It will be shown that the position of the noun phrase is intimately related to the information structure of the clause, especially the distinction between focus and presupposition, that is, “new” and “old” information. These notions will be explained in Subsection A. We start with a discussion of definite noun phrases and personal pronouns, which is followed by a discussion of indefinite and quantified noun phrases.

[+] A. Definite noun phrases

Definite nominal objects can occur both to the left and to the right of a clausal adverb. The placement of the noun phrase to the left or to the right of such an adverb is not free, however, but intimately related to the information structure of the clause. Consider the examples in (26). The direct object het boek in (26a) follows the modal adverb waarschijnlijk'probably' and is construed as part of the “new” information or focus of the clause; due to the prosodic structure of Dutch clauses, the sentence accent naturally falls on the direct object, which enhances a focus interpretation for this noun phrase. In (26b), on the other hand, the object has been scrambled to the left of the adverb, and as a result it no longer receives sentence accent; scrambled nominal objects cannot be interpreted as (part of the) focus, but are rather construed as topics of discussion, belonging to the “old” information or presupposition of the utterance (Verhagen 1986).

a.		Hij	heeft	waarschijnlijk	het boek	gelezen.
		he	has	probably	the book	read

b.		Hij	heeft	het boek	waarschijnlijk	gelezen.
		he	has	the book	probably	read

At this point a remark on the terminology is in order. The notions “new” and “old” information may be confusing since the former suggests that the referent of the noun phrase het boek in (26) is not part of the domain of discourse (domain D), whereas the latter suggests that it is. This is clearly not the case, since in both cases the hearer is assumed to be able to uniquely identify this referent. The notions rather refer to the information structure of the clause; the “old” information refers to the entities currently under discussion, whereas the “new” information refers to entities that may be part of the background of the discourse (that is, part of domain D) but were so far not a topic of discussion. In order to avoid the misleading connotations of the notions of “new” and “old” information, we will generally use the notions “focus” and “presupposition” in this work (despite the fact that the former can be easily confused with the notion of contrastive or emphaticfocus).
The distinction between presupposition and focus is especially clear in question-answer contexts. A question like (27a) introduces the referent of het boek as a topic of discussion, and therefore the answer preferably has the noun phrase in front of the adverb, that is, presents the noun phrase as “old” information; in actual speech, this is made even clearer by replacing the noun phrase het boek by the personal pronoun het, which typically refers to “old” information (see Subsection B below).

a.		Wat	heeft	Jan	met	het boek	gedaan?	question
		what	has	Jan	with	the book	done

Hij heeft waarschijnlijk het boek gelezen.

answer = (26a)

b'.

Hij heeft het boek waarschijnlijk gelezen.

answer = (26b)

A question like (28a), on the other hand, clearly does not presuppose the referent of the noun phrase het boek to be a topic of discourse, and now the preferred answer has the noun phrase following the adverb. The answer in (28b') with the nominal object preceding the adverb is only possible if the context provides more information, e.g., if the participants in the discourse know that Jan had the choice between reading a set of articles or reading a certain book; in that case the nominal object preceding the adverb is likely to have contrastive accent.

a.		Wat	heeft	Jan gelezen?	question
		what	has	Jan read

Hij heeft waarschijnlijk het boek gelezen.

answer = (26a)

b'.

Hij heeft het boek waarschijnlijk gelezen.

answer = (26b)

That the noun phrase het boek refers to “new” information is also clear from the fact that replacing the noun phrase het boek by the personal pronoun het gives rise to an infelicitous result: using the pronoun makes the answer uninformative since it presupposes (contrary to fact) that the identity of the referent is already known to the person asking the question.
Note that in (28) the activity of reading is still presupposed as a topic. This is not the case in an example such as (29), but in this case also the utterance with the direct object following the adverb is strongly preferred. The answer with the nominal object preceding the adverb is only possible if the context provides more information, e.g., if the participants in the discourse know that Jan had the choice between reading the book or following a crash course in linguistics. Note that (29) shows that (26a) can be construed not only with the noun phrase het boek, but also with the complete verb phrase het boek gelezen, as “new” information.

a.		Wat	heeft	Jan gedaan?	question
		what	has	Jan done

Hij heeft waarschijnlijk het boek gelezen.

answer = (26a)

b'.

Hij heeft het boek waarschijnlijk gelezen.

answer = (26b)

To conclude we want to note that according to some research the informational-structural effect described above are tendencies not absolute rules.

[+] B. Referential personal pronouns

Referential personal pronouns are typically used to refer to active topics of discussion. Therefore, we correctly predict them to normally occur in a position preceding the clausal adverbs. This is clear from the fact that in an example such as (30a), the pronoun het must precede the adverb waarschijnlijk'probably'.

Jan heeft	<het>	waarschijnlijk <*het>	gelezen.
Jan has	it	probably	read
'Jan has probably read it.'

The requirement that personal pronouns precede the clausal adverbs can, however, be overruled in contrastive contexts by assigning contrastive focus accent to the pronoun. Given the fact that weak pronouns cannot be assigned accent, this is only possible with strong pronouns. Some illustrative examples are given in (31); the primed examples show that in these cases placement of the contrastively focused pronoun in front of the adverb is also possible, and even seems to be preferred by some. Note that the ungrammatical variant of (30) cannot be saved by assigning contrastive accent to the pronoun het, due to the fact that het normally cannot be assigned accent; cf. Section 5.2.1.1, sub V.

a.		Jan kiest	waarschijnlijk	(?)mij/*me	als begeleider,	niet jou.
		Jan chooses	probably	me/me	as supervisor,	not you

a'.

Jan kiest mij waarschijnlijk als begeleider, niet jou.

b.		Jan heeft	waarschijnlijk	(?)hem/*’m	uitgenodigd,	niet haar.
		Jan has	probably	him/him	prt.-invited	not her

b'.

Jan heeft hem waarschijnlijk uitgenodigd, niet haar.

If the negative adverb niet'not' is placed in the first conjunct, both orders are completely acceptable; this is illustrated in (32). The difference between the primeless and the primed examples is that in the former the negative adverb niet acts as constituent negation and in the latter as sentential negation; cf. Section 8.1.3, sub IV.

a.		Jan kiest	waarschijnlijk	niet	mij	als begeleider,	maar jou.
		Jan chooses	probably	not	me	as supervisor,	but you

a'.		Jan kiest	mij	waarschijnlijk	niet	als begeleider,	maar	wel	jou.
		Jan chooses	me	probably	not	as supervisor,	but	aff.	you

b.		Jan heeft	waarschijnlijk	niet	hem	uitgenodigd,	maar	haar.
		Jan has	probably	not	him	prt.-invited	but	her

b'.		Jan heeft	hem	waarschijnlijk	niet	uitgenodigd,	maar	wel	haar.
		Jan has	him	probably	not	prt.-invited	but	aff.	her

[+] C. Indefinite noun phrases

Scrambling of indefinite nominal objects across a clausal adverb is possible in some but not all constructions. If it occurs, scrambling has important semantic repercussions: it may change the scope relation between the indefinite noun phrase and some other quantified expression, or force a generic reading on the moved noun phrase.

[+] 1. Scope

The examples in (33) show that (both nonspecific and specific) indefinite nominal objects cannot readily appear to the left of a modal adverb like waarschijnlijk.

a.		dat	Jan waarschijnlijk	een vriend	zal	bezoeken.
		that	Jan probably	a friend	will	visit

a'.

dat Jan een vriend waarschijnlijk zal bezoeken.

b.		dat	Jan waarschijnlijk [NP ∅	vrienden]	zal	bezoeken.
		that	Jan probably	friends	will	visit

b'.

dat Jan [NP ∅ vrienden] waarschijnlijk zal bezoeken.

However, scrambling of indefinite nominal objects is often possible if the clausal adverb expresses frequency, and coincides with a difference in scope. First, consider example (34a), in which the indefinite noun phrase follows the adverbial phrase elke dag'every day'. This example asserts that Jan has the habit of watching (at least) one program a day on TV, where the program may change from day to day. However, if the speaker has a specific television program in mind that Jan watches every day (e.g., the eight oʼclock news), he is not likely to use example (34a); he would probably use an example such as (34b) instead, where één is stressed so that we cannot determine whether we are dealing with the indefinite article een'a' or the numeral één'one'. The fact that a nonspecific indefinite bare plural like programmaʼs in the primed examples cannot be placed in front of the adverb, however, suggests the latter.

a.		dat	Jan elke dag	een programma	op tv	bekijkt.
		that	Jan every day	a program	on TV	watches

a'.		dat	Jan elke dag	programmaʼs op tv	bekijkt.
		that	Jan every day	programs on TV	watches

b.		dat	Jan één programma	op tv	elke dag	bekijkt.
		that	Jan a/one program	on TV	every day	watches

b'.	??	dat	Jan programmaʼs op tv	elke dag	bekijkt.
		that	Jan programs on TV	every day	watches

This suggestion is also supported by the fact that a plural noun phrase preceded by a numeral show the same difference in reading as (34a&b): (35a) expresses that Jan watches two programs every day, where the programs may change from day to day, whereas (35b) expresses that Jan watches the same two programs every day.

a.		dat	Jan elke dag	twee programmaʼs op tv	bekijkt.
		that	Jan every day	two programs on TV	watches

b.		dat	Jan twee programmaʼs op tv	elke dag	bekijkt.
		that	Jan two programs on TV	every day	watches

From this we may conclude that the difference in scope between the indefinite noun phrase and the universally quantified adverbial phrase is reflected in the linear order of the two: in (35a) the universal operator expressed by the temporal adverbial phrase has scope over the existential operator implied by the indefinite noun phrase (∀t ∃x), and in (35b) the scope relation is inverted (∃x ∀t).

[+] 2. Genericity

Another possible effect of scrambling is that the indefinite noun phrase receives a generic interpretation. Consider the examples in (36). Example (36a) expresses that Jan is reading something which is probably a bestseller (or, alternatively, that Jan is doing something, which is probably reading a bestseller). Example (36a'), on the other hand, expresses that bestsellers are likely to be read by Jan. The same pattern is even clearer in (36b&b'): (36b) expresses that Jan generally reads some bestseller, whereas (36b') expresses that most bestsellers are read by Jan. The (c)-examples provide similar examples with plural noun phrases: (36c) expresses that Jan generally reads bestsellers, whereas (36c') expresses that most bestsellers are read by Jan.

a.		dat	Jan waarschijnlijk	een bestseller	leest.
		that	Jan probably	a bestseller	reads

a'.

dat Jan een bestseller waarschijnlijk leest.

b.		dat	Jan meestal	een bestseller	leest.
		that	Jan generally	a bestseller	reads

b'.

dat Jan een bestseller meestal leest.

c.		dat	Jan meestal	bestsellers leest.
		that	Jan generally	bestsellers reads

c'.

dat Jan bestsellers meestal leest.

Scrambling of indefinite nominal objects is also possible, and is perhaps even preferred, if the noun phrase contains an attributive adjective like volgende'next' or nieuwe'new' or an ordinal numeral, as in the examples in (37). The indefinite noun phrases in these examples seem comparable to English noun phrases containing free choice any: Jan will turn down any invitation that comes next; the Security Council will condemn any attack that comes next. Since we are not aware of any discussion of data like these in the literature, we will leave these for future research; see also example (64) for comparable examples with the negative adverb niet'not'.

a.	Jan zal	een volgende/nieuwe uitnodiging	waarschijnlijk	afslaan.
	Jan will	a next/new invitation	probably	turn.down
	'Jan will probably turn down any invitation that comes next/new invitation.'

a'.

Jan zal waarschijnlijk een volgende/nieuwe uitnodiging afslaan.

b.	De Veiligheidsraad	zal	een nieuwe/tweede aanval	waarschijnlijk	veroordelen.
	the Security Council	will	a new/second attack	probably	condemn
	'The Security Council will probably condemn a subsequent/second attack.'

De Veiligheidsraad zal waarschijnlijk een nieuwe/tweede aanval veroordelen.

[+] D. Quantified noun phrases and quantifiers

This subsection discusses scrambling of quantified nominal objects and quantifiers, and its semantic effects. Existentially, universally and negatively quantified noun phrases are discussed in separate subsections. Before we start we want to note that the felicitousness of a certain word order is often determined not only by the quantifier in question, but also by the meaning of the predicate; certain orders may be infelicitous because they give rise to an improbable reading with some predicates. In the following we will abstract away from these effects of the choice of the predicate but simply select predicates that give rise to felicitous results.

[+] 1. Existentially quantified noun phrases

The placement of an existentially quantified nominal object with respect to a modal adverb like waarschijnlijk'probably' seems to depend on the nature of the quantifier. If the quantifier normally triggers a nonspecific reading of the noun phrase, as does enkele'some' in (38a&a'), the nominal object is preferably placed after the adverb. If the quantifier allows both a nonspecific and a specific reading, as does veel'many' in (38b&b'), the nominal object can readily occur on either side of the adverb. If the quantifier normally triggers a specific reading, as does sommige'some' in (38c'), the nominal object is preferably placed in front of the adverb. In all cases, a nominal object in front of the adverb is construed as specific, wheras one following the adverb is construed as nonspecific (unless it is assigned emphatic focus).

a.		dat	Jan waarschijnlijk	enkele boeken	weggooit.
		that	Jan probably	some books	throws.away

a'.

dat Jan enkele boeken waarschijnlijk weggooit.

b.		dat	Jan waarschijnlijk	veel boeken	weggooit.
		that	Jan probably	many books	throws.away

b'.

dat Jan veel boeken waarschijnlijk weggooit.

c.	?	dat	Jan waarschijnlijk	sommige boeken	weggooit.
		that	Jan probably	some books	throws.away

c'.

dat Jan sommige boeken waarschijnlijk weggooit.

Note that we have avoided the use of the terms weak and strong quantifier (cf. Section 6.2.1, sub II) in the description of the data in (38): since we will see in the next subsection that the (strong) universal quantifier alle is preferably placed after the clausal adverbs, we cannot say that strong quantifiers are preferably scrambled, whereas weak quantifiers are preferably left in their position to the right of the clausal adverbs. Nevertheless, this seems to provide an apt description of the behavior of the strong/weak existential quantifiers.
In (35), we have observed that scrambling of indefinite nominal objects affects the scope relations in the clause. If we are dealing with a noun phrase containing an existential quantifier, the same effect can be observed. Consider the examples in (39). In (39a) the frequency adverb has scope over the quantified noun phrase veel boeken'many books': as a result the sentence expresses that it is often the case that Jan is reading many books. In (39b), on the other had, it is the noun phrase that has scope over the adverb: as a result the sentence expresses that there are many books that Jan often reads.

a.		dat	Jan	vaak	veel boeken	leest.
		that	Jan	often	many books	reads

dat Jan veel boeken vaak leest.

This difference in interpretation can also be held responsible for the fact that an adverb like meestal'usually' cannot follow a quantified nominal object: whereas it makes perfect sense to claim that Jan usually reads many books, it seems weird to say that many books are usually read by Jan. Similarly, it may account for the fact that a strong noun phrase like sommige boeken'some books', which presupposes a certain set of books and is therefore specific, cannot readily be used in the position following the adverb.

a.		dat	Jan	meestal	veel boeken	leest.
		that	Jan	usually	many books	reads

a'.

dat Jan veel boeken meestal leest.

b.	*?	dat	Jan	vaak	sommige boeken	leest.
		that	Jan	often	some books	reads

b'.

dat Jan sommige boeken vaak leest.

The existential personal pronouns iemand'someone' and iets'something' also allow both a nonspecific and a specific interpretation. As in the quantified noun phrases discussed above, the availability of these readings depends on whether the noun phrase occurs to the right or to the left of the adverb. Note that the specific readings in the primed examples are not completely natural.

a.		dat	Jan	waarschijnlijk	iemand	uitnodigt.
		that	Jan	probably	someone	prt.-invites

a'.

dat Jan iemand waarschijnlijk uitnodigt.

b.		dat	Jan	waarschijnlijk	iets	aan Peter	wil	geven.
		that	Jan	probably	something	to Peter	wants	give

b'.

dat Jan iets waarschijnlijk aan Peter wil geven.

Again, the position of the nominal object affects the scope readings: whereas the frequency adverb has scope over the existential pronouns in the primeless examples of (42), the pronouns have scope over the adverb in the primed examples. As a result, (42a) asserts that it has often been the case that Jan insulted some person or other, whereas (42a') expresses that there is a certain person who has often been insulted by Jan. Similarly, (42b) asserts that it has often been the case that Jan dropped something, whereas (42b') expresses that there is a certain thing that has often been dropped by Jan. Observe that, in contrast to the primed examples in (41), the primed examples in (42) are impeccable.

a.		dat	Jan vaak iemand	heeft	uitgescholden.
		that	Jan often someone	has	prt.-insulted

a'.

dat Jan iemand vaak heeft

uitgescholden.

b.		dat Jan	vaak	iets	laat vallen.
		that Jan	often	something	drops

b'.

dat Jan iets vaak laat vallen.

[+] 2. Universally quantified noun phrases

The examples in (43) suggest that universally quantified phrases have some preference for the position following the modal adverb waarschijnlijk'probably', but both orders seem to be grammatical. The difference between the two examples seems to be related to the information structure in the clause. In (43b), it is only the action of inviting that is part of the focus of the clause. Example (43a) is compatible with various information structures: the focus of the clause can be formed by the full VP alle studenten uitnodigen, the noun phrase alle studenten, or the quantifier alle — in the first two cases sentence stress falls on the noun studenten, and in the third case on the quantifier alle.

a.	Marie	zal	waarschijnlijk	alle studenten	uitnodigen.
	Marie	will	probably	all students	prt.-invite
	'Marie will probably invite all students.'

(?)

Marie zal alle studenten waarschijnlijk uitnodigen.

The two examples also seem to differ in interpretation. Example (43a) can be interpreted either as referring to a single event of inviting all the students or as referring to several separate events of inviting a student or subgroup of students, whereas (43b) strongly favors the latter interpretation. This meaning difference is probably related to the scope of the modal adverb waarschijnlijk'probably'. In (43b) the universally quantified phrase is outside the scope of the modal adverb, and, as a result, it is claimed for each individual student that he will probably be invited. In (43a), on the other hand, the universally quantified phrase is within the scope of the modal adverb, and, as a result, it is claimed that it is probably the case that all students will be invited, where it is immaterial whether they are invited individually or as a group. The universal quantifiers iedereen'everyone' and alles'everything' also seem to prefer the position to the right of the modal adverb, but again both orders seem to be acceptable.

a.	dat	Jan	waarschijnlijk	iedereen/alles	meeneemt.
	that	Jan	probably	everyone/everything	prt.-takes
	'that Jan probably takes everyone/everything with him.'

dat Jan iedereen/alles waarschijnlijk meeneemt.

Universally quantified nominal objects can readily occur on both sides of the adverbs of frequency. This gives rise to a difference in interpretation, which can again be expressed in terms of scope: in (45a'), the universally quantified noun phrase is outside the scope of the adverb, and as a result it is claimed for each individual book in the relevant domain of discourse that John often takes it with him; in (45a), on the other hand, the universally quantified noun phrase is within the scope of the frequency adverb, and as a result it is claimed that it is often the case that John takes all books with him. Examples (45b&b') show that the same thing holds for universal quantifiers such as alles'everything'.

a.	dat	Jan vaak	alle boeken	meeneemt.
	that	Jan often	all books	prt.-takes
	'that Jan often takes all books with him.'

a'.

dat Jan alle boeken vaak meeneemt.

b.	dat	Jan vaak	alles	meeneemt.
	that	Jan often	everything	prt.-takes
	'that Jan often takes everything with him.'

b'.

dat Jan alles vaak meeneemt.

[+] 3. Negative quantifiers

Given that the negative quantifiers niemand'nobody' and niets'nothing' do not allow a specific interpretation, it does not come as a surprise that such noun phrases must follow the modal adverbs, as is illustrated by (46a&b). Another factor that may play a role here is that, in general, negative phrases tend to follow the modal adverbs. This even holds for subjects, as is shown in (46c).

a.		dat	Jan	<*niemand>	waarschijnlijk <niemand>	uitnodigt.
		that	Jan	nobody	probably	invites

b.		dat	Jan	<*niets>	waarschijnlijk <niets>	aan Peter	wil	geven.
		that	Jan	nothing	probably	to Peter	wants	give

c.		dat	<??niemand>	waarschijnlijk <niemand>	dat boek	gelezen	heeft.
		that	nobody	probably	that book	read	has

However, unlike modal adverbs, the negative quantifiers can precede the frequency adverbs. The two examples in (47), which are the negative counterparts of the primed examples in (42), respectively express that there is not a certain person who has often been insulted by Jan and that there is not a certain thing that has often been dropped by Jan.

a.		dat	Jan niemand	vaak	heeft	uitgescholden.
		that	Jan nobody	often	has	prt.-insulted

b.		dat	Jan niets	vaak	laat vallen.
		that	Jan nothing	often	drops

The examples in (48) show that the negative quantifiers can also follow the adverbs of frequency. In these examples the quantifier is in the scope of the adverb: (48a) expresses that it is often the case that Jan does not want to see anyone and (48b) that it is often the case that Jan does not want to eat anything.

a.	dat	Jan vaak	niemand	wil	zien.
	that	Jan often	nobody	wants	see
	'that Jan often doesnʼt want to see anyone.'

b.	dat	Jan vaak	niets	wil	eten.
	that	Jan often	nothing	wants	eat
	'that Jan often doesnʼt want to eat anything.'

[+] E. Interplay of indirect and direct objects

In the previous subsections, we have seen that scrambling is related to several meaning aspects of the clause: scrambling affects the information structure of the clause, it affects the scope relations between quantifiers, and it may trigger a partitive or generic reading of the moved nominal object. This subsection will show that there are also syntactic constraints on this movement.
So far, we have mainly considered scrambling of the direct object in the clause, but indirect objects behave in more or less the same way. This implies that in double object constructions such as (49), there are various word order possibilities. In (49a), neither of the objects is scrambled, which leads to an interpretation according to which both