Some distinctions in the German Nachfeld*
Hubert Truckenbrodt
ZAS Berlin
July 2013
Introduction
Section 1 of this paper addresses distinctions in prosody and information structural in German
between extraposition on the one hand and right dislocation (RD) and afterthought (AT) on the other
hand. In section 2, previous analyses of correlate constructions by Pütz (1986), Sudhoff (2003), Frey
(2010) and others are fit into the account. Section 3 presents an analysis of the distinction between
extraposition and RD/AT in terms of a deletion analysis of RD/AT and the special status of root
sentences in the sense of B. Downing (1970), which are tentatively identified with the non-integrated
clauses of Reis (1997) and Holler (2008). It is shown that these root sentence/non-integrated clauses
also allow for a representation of the distinction between RD and AT. The results are summed up in
section 4.
1. A prosodic distinction between extraposition and right dislocation
For developing the distinction between extraposition and RD/AT, we require some background.
Background on default stress is introduced in section 1.1. Background on focus and givenness
(information structure) is introduced in section 1.2. With this background section 1.3. shows the
prosodic stress distinction between extraposition and RD/AT, and section 1.4. shows the information
structure distinction between extraposition and RD/AT.
1.1. Default stress in German
Consider the sentence melody of the German sentence in (1). This German sentence carries four beats
of phrasal stress, highlighted by underlining in the rendition below the plot. Each non-final phrasal
stress corresponds to a dramatic rise in the sentence melody, the final phrasal stress corresponds to a
fall in the sentence melody. These tonal movements are analyzed in terms of pitch accents (here L*+H
for the non-final rises, H+L* for the final fall1) that are assigned to the positions of phrasal stress. In
Gussenhoven (1983, 1992), Selkirk (1984, 1995) this combination of phrasal stress and pitch accent
was called accent. I follow Gussenhoven and Selkirk in analyzing German sentence prosody in terms
of two prosodic levels: The lower level is that of accent, here also: phrasal stress. The higher level,
often called nuclear stress, is here called sentence stress. It is determined by strengthening the
rightmost phrasal stress in (1) and shown by double underlining in the sentence below the plot.
*
I want to thank Werner Frey for many helpful discussions and hints in connection with this paper. I
also thank Dennis Ott and members of the ZAS for comments on this material.
1
For the tonal analysis of the German sentence melody see Grice et al. (2005), for the particular tonal
pattern in (1) see Truckenbrodt (2007). These analyses are based on the analysis of the sentence
melody of English in Pierrehumbert (1980) and Beckman & Pierrehumbert (1986), with early
applications to German including Uhmann (1991) and Féry (1993).
1
(1)
Was gibt’s Neues?
Le-na
will dem Wer-ner im Ja-nuar ein La-ma malen.
Lena
wants the Werner
in.the January a
llama paint
‘Lena wants to paint a llama for Werner in January.’ Adapted from Truckenbrodt (2007).
Die
the
There are a number of principled proposals for deriving the positions of accents from the
syntactic structure: In terms of focus feature percolation along syntactic lines (Selkirk 1995), in terms
of XPs (Truckenbrodt 2006, 2012, to appear) and in terms of phases (Kratzer & Selkirk 2007). For the
purpose at hand, it is sufficient to work with a slightly simplified version of the descriptive Sentence
Accent Assignment Rule (SAAR) of Gussenhoven (1983, 1992), which was applied to German by
Uhmann (1991). I employ the formulation in (2).
(2)
Sentence Accent Assignment Rule (Gussenhoven 1992, minimally simplified):
Each argument and each adjunct receives accent. The verb is unaccented next to an accented
argument, otherwise accented.
In (1), then, each argument (die Lena, dem Werner, ein Lama) and the adjunct (im Januar) is accented.
Furthermore, the final verb malen is unaccented. This correctly follows from (2) because this verb
stands next to an accented argument (ein Lama). (2) also states that the verb is accented when not next
to an accented argument. Two instances of this are shown in (3): The verb is accented when it follows
an adjunct as in (3a) (see also Krifka 1984, Jacobs 1993, Uhmann 1991 for German). (3b) shows that
the verb is also accented when it follows an unaccented argument, here etwas (Kratzer & Selkirk
2007, Truckenbrodt in press). Throughout, we assume that definite and indefinite pronouns are exempt
from (2) by stipulation.
(3)
a.
Peter hat während der Pause geschlafen.
Peter has during the break slept
‘Peter has slept during the break’
b.
Lena hat im Januar etwas gemalt.
Lena has in January something painted
‘Lena has painted something in January.’
The assignment of sentence stress is then achieved by the additional rule of rightmost
strengthening in (4), here cast in terms of the notion sentence stress. Following Uhmann (1991:179)
the domain of strengthening is the intonation phrase, to which we will return.2 In simple sentences like
(1) the entire sentences is an intonation phrase.
(4)
2
Rightmost strengthening (Uhmann 1991):
Strengthen the rightmost accent in the intonation phrase to sentence stress.
See Nespor & Vogel (1989), Hayes (1995), and Selkirk (2008) for principled assumptions about an
integrated prosodic and metrical representation in which this is placed, see also the notion of nuclear
stress in Beckman & Pierrehumbert (1986). In German, a close connection between prosodic
constituents and corresponding stress-assignment was part of the perspective of Pheby (1981) and has
been similarly assumed in much work since.
2
The prosodic regularities described in this section hold in all-new sentences with no narrow
focus. The predicted accents were regularly found in extensive experiments in Truckenbrodt (2002,
2004, 2007), see also Féry & Kügler (2008).
Other contributions of interest in the description and account of these prosodic patterns in
German include Höhle (1982), Krifka (1984), von Stechow & Uhmann (1986), Jacobs (1993), Cinque
(1993) and Wagner (2005).
1.2. The prosodic effects of focus and of givenness
The prosodic regularities described in the preceding section may be seen as a syntactic default system
of stress assignment. The defaults can be overridden by the prosodic demands of information
structure. The most well-known of these is the attraction of sentence stress by a narrow focus, marked
F, as in (5). Here the constituent answering the question, die Lena, is focused (F-marked) and attracts
sentence stress.
(5)
Wer hat ein Eis gegessen?
‘Who has eaten an ice-cream?’
[F Die Lena] hat ein Eis gegessen
the Lena has an ice-cream eaten
‘Lena has eaten an ice-cream.’
The general regularity behind this was first formulated by Jackendoff (1972) and is given in (6). See
Truckenbrodt (1995) for a refinement.
(6)
Prosodic effect of F (Jackendoff 1972)
F attracts the strongest stress of the sentence.
Less well known, but also important for this paper, is the stress-rejecting effect of givenness (Ladd
1983, Selkirk 1995, Schwarzschild 1999). It is demonstrated in (7). Here the final part of the sentence,
ein Eis gegessen, is contextually given, and therefore rejects sentence stress, which then falls on the
last accented element preceding it.
(7)
Wurde ein Eis gegessen?
‘Was an ice-cream eaten?’
Ja, die Lena hat in der Pause [G ein Eis gegessen]
yes the Lena has in the break an ice-cream eaten
‘Yes, during the break Lena has eaten an ice-cream.’
I here follow Féry & Samek-Lodovici (2006) in implementing this in terms of G-marking on
contextually given constituents as in (7). Following Féry & Samek-Lodovici (2006), the prosodic
effect of G-marking in the terms of this paper is shown in (8).3
(8)
Prosodic effect of G (Féry & Samek-Lodovici 2006):
G rejects accent and sentence stress.
Both the prosodic effect of F and the prosodic effect of G override the default-rules in (2) and (4), as
the examples (5) and (7) show.
3
As refinement developed in joint ongoing work with Daniel Büring is that the rejection of sentence
stress on given constituents is categorical so long as there is a non-given element in the intonation
phrase to which the stress can shift. On the other hand, the rejection of accent by given constituents in
positions preceding the sentence stress is optional. This refinement is not crucial for this paper.
3
In sum, default prosody is assigned by (a) the SAAR that assigns accents to arguments,
adjuncts and sometimes verbs and (b) rightmost strengthening to sentence stress in the intonation
phrase. This default can be overridden by information structure: F attracts the strongest stress in the
sentence and G rejects accent and sentence stress. We now have the tools to turn to the prosody of
extraposition and of right dislocation.
1.3. A prosodic distinction between extraposition and right dislocation/afterthought
The cases of extraposition discussed here are extraposition of PPs from NPs as in (9), extraposition of
object clauses as in (10) and extraposition of heavy objects as in (11).
(9)
Was ist geschehen?
‘What happened?’
Die Maria hat ein Buch t gelesen von Chomsky
the Maria has a book read by Chomsky
‘Maria has read a book by Chomsky.’
(10)
Wie wird das Wetter?
‘What will the weather be like?’
Der Peter hat t gesagt dass es regnen wird
the Peter has said that it rain will
‘Peter said that it will rain.’
(11)
Was hat Peter gemacht?
‘What did Peter do?’
Er hat t gegessen ein Schnitzel Pommes und einen Salat
he has eaten a schnitzel, french-fries and a salad
‘He ate a schnitzel, french-fries and a salad.’
The case to be contrasted with these is right dislocation (RD) and afterthought (AT). To begin with,
both may involve the resumption of a personal pronoun as in (12).4
(12)
Ich habe sie gesehen, die Maria.
I have her seen, the Maria
‘I have seen her, the person Maria.’5
There is a clear syntactic distinction between extraposition and RD/AT. The extraposed
elements bind a trace in the preceding clause, and in that sense belong to the preceding clause. On the
other hand, a RD/AT constituent is, in the picture of Zifonun et al. (1997:1647), a kind of doubled
version of an element in the clause (i.e. of the pronoun) and in that sense, according to Zifonun et al.
4
The discourse conditions under which right dislocation is employed in English are investigated in
Ziv & Grosz (1994) and Averintseva-Klisch (2009). While right dislocation may be used to clarify the
reference of the preceding pronoun, Averintseva-Klisch argues that that need not be so. Right
dislocation also occurs where the reference of the pronoun is unambiguous:
(i)
Meine Chefin1 dreht manchmal völlig durch. Die1 ist halt so, die Frau1.
my boss.fem turns sometimes totally trough she is MP so the woman
‘My boss sometimes goes cracy. She is like that, that woman.’
5
There is a poorly understood condition on dislocated DPs that disallows names without articles (*Ich
habe sie gesehen, Maria, *I have seen her, Maria), bare plurals and bare mass nouns. In German,
names are also acceptable with an article, and can be dislocated when they stand with the article.
Many examples of this paper are of this kind. They do not directly translate into Enlish, where names
with article are not possible. I translate them as ‚the person ...’ here.
4
(1997), not part of the preceding clause. It is shown in the following that this syntactic and semantic
separation in RD/AT has clear reflexes in the prosody and in the information structure, and I develop
formulations for these reflexes in the following.
For comparison, consider first the prosodic structure of extraposition. According to Altmann
(1981:201), extraposed elements typically carry the sentence stress. This can be seen in (9) – (11). It
seems that the extraposed constituent together with the preceding sentence forms a stress-domain in
which the SAAR in (2) and rightmost strengthening in (4) apply.6 Let us compare this to right
dislocation. Altmann (1981:189ff) discusses two prosodic classes.7 In the first class, the dislocated
element is essentially deaccented. This occurs naturally when the referent of the pronoun and of the
dislocated element is contextually given, as in (13).
(13)
Was ist mit Maria?
‘What about Maria?’
Ich habe sie gesehen, die Maria.
‘I have seen her, the person Maria.’
Altmann’s second case involves, in current terms, a separate intonation phrase with a separate
sentence stress. There is a noticeably pause corresponding to the intonation phrase boundary preceding
the dislocated element. An example is shown in (14). The intonation phrases are annotated I. As
Altmann notes, such cases are often particularly natural with introductory words before the dislocated
element such as ich meine ‘I mean’ (here: Peter hugged her, I mean Maria).
(14)
Was ist geschehen, als Maria, Claudia und Peter ankamen?
‘What happened when Maria, Claudia and Peter arrived?’
(
x )I (
x )I
der Peter hat sie umarmt – die Maria
the Peter has her hugged the Maria
‘Peter has hugged her, the person Maria.’
This prosodic division is also reported by Ziv & Grosz (1994) and Averintseva-Klisch (2006). These
authors call only the integrated version in (13) right dislocation (RD) and refer to the separated version
in (14) as afterthought (AT). I adopt this terminology, though I will sometimes, across RD and AT,
refer to ‘the dislocated constituent’.
Importantly, next to the two prosodic options in (13) and (14), a third option is not available,
namely the option that we saw with extraposition in (9) – (11). It is not possible to take the entire
domain of [clause + dislocated element], assign accents in it by the SAAR, and to strengthen the
rightmost of these, as in (15).
(15)
a. *
b. *
Der Peter hat sie umarmt die Maria.
(* with this stress-pattern in any context)
the Peter has her hugged the Maria
Die Hans hat ihr die Zeitung vorgelesen, der Maria.
the Hans has her the newspaper read the Maria
This is quite generally true of right dislocation: The sentence that precedes the dislocated element
requires sentence stress of its own. It has sentence stress in (13) and in (14), but not in (15). Notice
that the same requirement does not obtain for extraposition in (9) – (11). In these examples, the
sentence that precedes the extraposed element is without sentence stress. Unlike in (15), this does not
lead to a problem in these cases. I capture this difference in the formulations in (16).
6
Movement sometimes interacts with stress-assignment, as discussed in Bresnan (1971, 1972) and, for
German, Truckenbrodt & Darcy (2010), where the interaction with extraposition is also addressed.
However, the point at hand seems not to be affected by this.
7
He sees them as extremes in a continuum, but I analyze them as discrete.
5
(16)
a. In an extraposed structure clause+XP, sentence stress is assigned and required in [clause +
XP].
b. In an RD or AT structure clause+XP, sentence stress is assigned and required in [clause]
(where AT but not RD establishes a following, separate domain of sentence stress [XP]).
So far, we observed this difference in patterns of default stress. In the following argument, I
will show that it also obtains under duress as we bring givenness into the picture. (17) is an example of
extraposition, in which everything is contextually given, except for the extraposed element. Sentence
stress is on the extraposed element, and the sentence is natural. This is expected by (16a): Sentence
stress can avoid the given first part of the clause and naturally fall on the extraposed element, which is
still part of the domain of sentence stress assignment by (16a).
(17)
Claudia sagt, dass Maria ein Buch gelesen hat.
‘Claudia said that Maria has read a book.’
______given_____________
Ja, Maria hat ein Buch t gelesen von Chomsky
yes Maria has a book read
by Chomsky
A similar example for RD/AT is shown in (18a). Unlike in (17), it is not possible to leave the sentence
to the left of the dislocated constituent without sentence stress, even though it is entirely given. Instead
sentence stress is required on it as in (18b,c). In (18), (16b) is in direct conflict with (8). (16b) requires
sentence stress in the clause on the left. (8) requires the absence of sentence stress on that clause
because it is contextually given. Empirically, sentence stress is assigned and required, so that we
conclude that (16b) overrides (8).
(18)
Claudias Mutter sagt, dass der Peter sie gesehen hat.
‘Claudia’s mother says that Peter has seen her.’
a. *
b.
c.
______given_________
Ja, der Peter hat sie gesehen, die Claudia.
Ja, der Peter hat sie gesehen, die Claudia.
Ja, der Peter hat sie gesehen – die Claudia.
yes the Peter has her seen
the Claudia
‘Yes, Peter has seen her, the person Claudia.’
In summary, we observed the distinction in (16) and we observed that it cannot be overridden by the
prosodic demands of contextual givenness. Extraposed constituents constitute a part of the stressdomain that requires sentence stress, while RD/AT are not, so that the preceding sentence requires
sentence stress of its own.
1.4. An information structure distinction between extraposition and right
dislocation/afterthought
Consider the placement of narrow focus on the constituent on the right. As shown in (19) – (21), it is
easily possible to put focus on an extraposed constituent. As expected, main stress falls on the focused,
extraposed, constituent.
(19)
Von wem hat Maria ein Buch gelesen?
‘By whom did Maria read a book?’
Sie hat ein Buch t gelesen [F von Chomsky]
she has a book read by Chomsky
‘She read a book by Chomsky.’
6
(20)
Was hat Peter gesagt?
‘What did Peter say?’
Er hat t gesagt [F dass es regnen wird ]
he has said that it rain will
‘He said that it will rain.’
(21)
Was hat Peter gegessen?
‘What did Peter eat?’
Er hat t gegessen [F ein Schnitzel, Pommes und einen Salat ]
he has eaten a schnitzel, french-fries and a salad
‘He hate a schnitzel, french-fries and a salad.’
Right-dislocation is seriously restricted in this regard, as shown in (22).
(22)
Wen hat Marias Vater gesehen?
‘Who did Maria’s father see?’
a.
b.
c.
d.
* Er hat sie gesehen [F die Maria]
* Er hat sie gesehen [F die Maria]
* Er hat sie gesehen [F die Maria]
[Er hat sieF gesehen] [F die Maria]
he has her seen the Maria
‘He saw her, the person Maria.’
Here (16b) can exclude (22a). The demand to stress the focus excludes (22b). However, (22c), which
meets (16b) and stresses the intended focus, is also impossible. It therefore seems that we are dealing
with an additional restriction. The structure in (22d) is possible. Here the sentence to the left of the
dislocated constituent contains a focus of its own that answers the question. It therefore seems that the
restriction at work in (22) is that the dislocated constituent cannot form a focus-background domain
with the preceding clause, unlike an extraposed constituent in (19) – (21). This is formulated in (23) in
terms of the scope of a focus in Rooth (1992).
(23)
a. In an extraposed structure clause+XP, the scope of a focus on XP can be ~[clause + XP].
b. In a right dislocated structure clause+XP, the scope of a focus on XP cannot be ~[clause +
XP].
The scope of a focus in the sense of Rooth (1992) includes the focus and the material that is
backgrounded relative to the focus. In typical question-answer sequences such as (19) – (22), the
scope of the focus is the entire answer. (23a) captures that the focus on the extraposed constituents in
(19) – (21) can take the entire clause as its scope. (23b) captures that this is not possible for a rightdislocated constituent in (22a-c). (22d) is not ruled out by (23b): Here the clause on the left is the
scope of the focus on the pronoun.8
8
The resumption of the focus in the AT-constituent in (22d) may suggest, on the surface, that this
second focus shares the background with the first focus, in possible violation of (23b). This
appearance is overcome with the deletion analysis of RD and AT adopted in section 3.1., in which the
second clause carries a background of its own. What goes across the two parts is deletion, not a focusbackground structure:
(i)
~[Er hat sieF gesehen], ~[er hat [die Maria]F gesehen]
he has her seen he has the Maria seen
7
A further application of (23) is with a construction to which I return in section 3, the
resumption of indefinite pronouns as in (24). As noted by Zwart (2001) this is only possible with a
stressed AT-resumption.
(24)
Was hat Peter gemacht?
‘What did Peter do?’
Er hat jemanden besucht – die Maria.
he has someone visited the Maria
‘He visited someone, the person Maria.’
(25a,b) show the effect of (23b), regardless of the presence of sentence stress to the left of the
dislocated constituent. It is not possible to interpret a right-dislocated focus with the preceding clause
as its background.
(25)
Wen hat Peter besucht?
‘Who did Peter visit?’
a.
b.
* Er hat jemanden besucht, die Maria.
* Er hat jemanden besucht – die Maria.
he has someone visited the Maria
‘He visited someone, the person Maria.’
Thus, in the domain of prosody and information structure, we find two possibly related ways
in which an extraposed constituent is integrated with the preceding clause while a right dislocated
constituent is not: For the purposes of assignment of sentence stress regardless of narrow focus as in
(16), and for the purposes of assignment of narrow focus as in (23).
Observe also that we have now seen that neither F nor G can override the requirement of
sentence stress preceding a dislocated constituent in (16b). Thus (16b) holds absolutely, not only for
default stress. In the case of G this seems to be a strength-relation in which (16b) is stronger than the
effect of G in (8). In the case of F the reason may be (23b), i.e. that it is not possible to construct a
configuration in which F could be in conflict with (16b) and override it.
In section 2, which has its motivation in the theme of the current volume, observations about
correlates are integrated into the account in terms of (16) and (23). Section 3 presents an account of
(16) and (23).
2. On the prosody of correlate constructions in German
Pütz (1986) and Sudhoff (2003) argue that predicates that can stand with a combination of es ‘it’ and a
clause linked to the same argument slot fall into two classes: (a) those in which es ‘it’ is a regular
pronoun and the construction is RD/AT of an argument clause, and (b) those in which es ‘it’ is a
correlate, a pronominal filler with properties closer to those of a trace. A related case (c) discussed in
the literature involves PPs with the pronominal element da- ‘there’ incorporated into the preposition,
and a clause linked to the same argument slot. Frey (2010) provides a range of syntactic arguments
that class (a) is indeed right dislocation, and argues for dividing class (b) into two classes. In this
section, I fit these earlier findings into the account in section 1.9
2.1. PPs with daI begin with the case of da-PPs as in (26a), where the clause on the right is linked to the same
argument slot as da- in darüber. For this class it is fairly obvious, that the clause on the right is
extraposed via movement: There is always also a grammatical non-extraposed version in which the
9
See Schwabe (2011); Schwabe & Fittler (2011) for promising and formally very advanced
suggestions for a semantic classification of verbs in connection with their ability to take such proforms.
8
clause immediately follows the PP, as in (26b). Vikner (1995) and Müller (1995) suggest a movement
derivation in which the clause originates even closer to the da-, inside of the PP. This movement
analysis is adopted by Frey (2010).
(26) a. Sie hat sich darüber gewundert [dass es regnet]
she has herself there.about wondered that it rains
b. Sie hat sich darüber [dass es regnet] gewundert
she has herself there.about that it rains wondered
‘She was surprised that it was raining.’
Whatever the exact source position, the prosody and information structure confirms that we are
dealing with extraposition in (26a). Thus, we find that [clause+XP] forms a single stress-domain for
sentence stress in the default case in (27) as well as under duress by givenness in (28). This conforms
to the extraposition pattern described in (16a).
(27)
Warum schaut sie denn so?
‘Why is she looking this way?’
Sie hat sich darüber gewundert [dass es regnet]
she has herself there.about wondered that it rains
‘She was surprised that it was raining.’
(28)
Hat sich Maria gewundert?
‘Was she surprised?’
______given______________
Ja, sie hat sich darüber gewundert [dass es regnet]
yes she has herself there.about wondered that it rains
‘Yes, she was surprised that it was raining.’
Furthermore, the extraposed clause can be focused, with the preceding clause in the background, as in
(29). This conforms to the extraposed pattern formulated in (23a).
(29)
Worüber hat sich Maria gewundert?
‘What was she surprised about?’
Sie hat sich darüber gewundert [F dass es regnet ]
she has herself there.about wondered that it rains
‘She was surprised that it was raining.’
Thus, for PPs with da-, the syntactic extraposition analysis is confirmed by the facts of prosody and
information structure.
2.2. Right dislocation of clauses
Next, consider the opposite extreme, predicates that allow es ‘it’ with a clause only by way of RD/AT.
The verb behaupten ‘claim’ used by Pütz and Sudhoff illustrates the properties of this class with
particular tenacity. Other verbs in this class are denken ‘think’, glauben ‘believe’, sagen ‘say’ and
vermuten ‘suppose’. The prosodic properties of this class were already noted by Pütz (1986) and
Sudhoff (2003) and are part of their arguments for the right dislocation analysis. They discussed the
unaccented occurrence of the right-dislocated clause. The point here is merely to integrate their
argument into the prosodic analysis of this paper.
As noted by Sudhoff (2003:61), these verbs do not allow an all-new occurrence of a correlate
construction, as shown in (30):
9
(30)
Was war denn los?
‘What happened?’
* Fred hat es behauptet, dass Wilma wegfährt.
Fred has it claimed that Wilma leaves
Sudhoff’s description of the prosodic facts fits straightforwardly into the current discussion. He
notices that sentence stress must be in the clause that precedes the right dislocated clause. An example
in which the right dislocated clause is contextually given is shown in (31).
(31)
Fährt denn Wilma weg?
‘Is Wilma leaving?’
Fred hat es behauptet, dass Wilma wegfährt.
‘Fred has claimed it, that Wilma is leaving.’
In addition to the unstressed right dislocation case that Sudhoff describes, there is also the realization
as afterthought with two intonation phrases, as in (32).
(32)
Fährt Wilma weg? Kommt sie wieder?
‘Is Wilma leaving? Is she coming back?’
Fred hat es behauptet – dass Wilma wegfährt.
‘Fred has claimed it, that Wilma is leaving.’
When we try to eliminate sentence stress from the clause on the left by givenness, we do not succeed,
as shown in (33a). This is parallel to other cases of right dislocation. In (33) and in the following
examples, the a.-examples involving right dislocation are compared to b.-examples without es, which
involve regular extraposition and therefore do not show the restrictions of RD/AT.
(33)
Hat Fred irgendwas dazu behauptet?
‘Did Fred claim anything in this connection?’
a.
b.
_____given_____
* Er hat es behauptet, dass Wilma wegfährt.
Er hat behauptet, dass Wilma wegfährt.
he has (it) claimed that Wilma leaves
‘He claimed that Wilma was leaving.’
This is accounted for by (16b) if resumption under behaupten ‘claim’ is RD/AT, as also argued by
Sudhoff.
Focus assignment is also not possible on the clause on the right, as illustrated in (34a). This
confirms the right dislocation analysis via (23b).
(34)
Was hat Fred behauptet?
‘What did Fred claim?’
a.
* Er hat es behauptet [F dass Wilma wegfährt ]
b.
Er hat behauptet [F dass Wilma wegfährt ]
‘He claimed that Wilma was leaving.’
Notice that (23b) also predicts that a narrow focus inside of the argument clause cannot have scope
outside of the argument clause. This is correct, as shown in (35a).
10
(35)
Von wem hat Fred behauptet, dass er oder sie wegfährt?
‘Of whom did Fred claim that he or she is was leaving?’
a.
b.
* ~[Er hat es behauptet, dass [F Wilma ] wegfährt]
~[Er hat behauptet, dass [F Wilma ] wegfährt]
‘He claimed that Wilma was leaving.’
See Frey (2010) for syntactic arguments for the right-dislocation analysis of the class of cases
discussed in this section. As was seen, their prosody and information structure supports this in the
context of (16) and (23).
2.3. Genuine correlates
The class of verbs that allows genuine correlates include bedauern ‘regret’, aufgeben ‘give up’,
bewundern ‘admire’ and lieben ‘love’. Sudhoff (2003:61,70f) points out that correlate constructions
are possible in all-new contexts, as shown in (36), which contrasts with (30).
(36)
Was war denn los?
‘What was the matter?’
Fred hat es bedauert, dass Wilma wegfährt.
Fred has it regretted that Wilma leaves
‘Fred regretted that Wilma was leaving.’
As Sudhoff (p.70f) observes, sentence stress is typically found in the argument clause, as indicated in
(36). Note that the absence of sentence stress to the left of the argument clause conforms to the
extraposition case (16a) but not to (16b). I add to this that where the matrix clause is given, it can be
destressed as in (37). This is also in line with (16a) but not (16b).
(37)
Hat Fred irgendetwas bedauert?
‘Did Fred regret anything?’
_____given_____
Er hat es bedauert, dass Wilma wegfährt.
‘He regretted that Wilma was leaving.’
Sudhoff (p.70f) observes that the regular stress rules obtain in the correlate-construction. He also notes
that the clause on the right can be focused. This is shown in (38). In addition, focus inside of the
embedded clause, with matrix scope, is shown in (39).
(38)
Was hat Fred bedauert?
‘What did Fred regret?’
Er hat es bedauert [F dass Wilma wegfährt]
‘He regretted that Wilma was leaving.’
(39)
Bei wem hat Fred es bedauert, dass er oder sie wegfährt?
‘In regard to whom did Fred regret that he or she was leaving?’
Er hat es bedauert dass [F Wilma ] wegfährt
‘He regretted that Wilma was leaving.’
This patterns with extraposition according to (23). All observations in (36) – (39) contrast with the
observations in connection with right dislocation in the behaupten-class of verbs in the preceding
11
section. They therefore support an analysis of the correlate construction in terms of movement of the
argument clause to the right, despite the presence of the pronoun in the clause.
A movement analysis for the genuine correlate construction has been suggested by Bennis
(1986), Vikner (1995), and Müller (1995). It is adopted and modified by Frey (2010). Frey suggests
that the argument clause originates in the specifier of the DP that is headed by es ‘it’. Previous authors
have pointed out that clause-internal positions are at best marked for the surface occurrence of the
argument clause. Frey points out in support of the movement analysis that they are nevertheless better
than in the behaupten-class.
(40)
a. ?
b. *
Fred hat es, dass es regnet, bedauert.
Fred hat es, dass es regnet, behauptet.
Fred has it that it rains claimed
See Frey (2010) for other syntactic arguments in favor of the movement analysis. As was seen in this
section, the (already known) facts of prosody and information structure support the movement analysis
of the correlate construction by (16) and (23).
2.4. Coindexed subject expletives
German has a number of verbs, including weather-verbs, that show subject expletives es. Examples
are shown in (41).
(41)
a.
Gestern hat es geregnet/geschneit/... .
yesterday has it rained/showed/...
‘It rained/snowed/... yesterday.’
b.
Gestern hat es mir gegraut.
yesterday has it me dreaded
‘I dreaded something yesterday.’
Frey (2010) argues that this phenomenon also occurs in cases that look like correlates, though in fact a
subject expletive is here lexically coindexed with an object clause in his analysis. His arguments
include that the link between es and the clause is more flexible with these cases, as in (42a), while
correlates as in (42b,c) and right dislocation as in (42d) show more narrow restrictions on this link.
(42) a.
Denkbar [dass es regnet]OBJ ist esSU schon.
conceivable that it rains is it indeed
‘It is indeed conceivable that it will rain.’
b.
??
[Bedauert [dass es regnet]] hat es Maria.
regretted that it rains has it Maria
‘Maria has regrettet it that it is raining.’
c.
*
[Gewundert [dass es regnet]] hat sich Maria darüber.
surprised that it rains has herself Maria there.about
d.
*
[Behauptet [dass es regnet]] hat es Maria.
claimed that it rains has it Maria
This class of cases then also has an extraposition analysis, albeit a different one. The clause is an
object clause and thus moved to the right by extraposition. We expect that it patterns in its prosody
and information structure with extraposition. This turns out to be the case. (43) and (44) show that the
matrix clause does not on its own require sentence stress and thus follows (16a). (45) shows focus
with wide scope, following (23a).
12
(43)
Für Peter ist es denkbar [F dass man ihn einlädt]
for Peter is it conceivable that one him invites
‘It is conceivable for Peter that he will be invited.’
(44)
Ist überhaupt etwas für Peter denkbar?
‘Is anything conceivable for Peter?’
_______given______
Für ihn ist es denkbar [F dass man ihn einlädt]
for him is it conceivable that one him invites
‘It is conceivable for him that he will be invited.’
(45)
Was ist für Peter denkbar?
‘What is for Peter conceivable?’
~[Für Peter ist es denkbar [F dass man ihn einlädt] ]
for Peter is it conceivable that one him invites
‘It is conceivable for Peter that he will be invited.’
Thus, the syntactic analysis of this class is also in line with our prosodic tests.
In sum, we are able to confirm the classification of extraposition vs. RD/AT for the verb
classes distinguished by Pütz (1986), Sudhoff (2003) and Frey (2010). This is particularly interesting
for genuine correlates and for the final class of subject expletives, where a movement analysis, though
not new, is not obvious, because of the presence of the pronominal es ‘it’.
3. Analysis of the distinctions in prosody and information structure between extraposition, right
dislocation, and afterthought
Section 3.1 introduces a syntactic deletion analysis of RD and AT. Section 3.2. introduces nonintegrated clauses (Reis 1997, Holler 2008) and identifies them with root sentences (B.Downing 1970)
and their connection to intonation phrases. Sections 3.3 and 3.4 show how the distinctions in stress
assignment and information structure between extraposition and RD/AT are naturally captured in
terms of these analytical devices. The distinction between RD and AT is also addressed.
3.1. A deletion account of right dislocation and afterthought
Kuno (1978) and Tanaka (2001) have argued for deletion accounts for right dislocation in Japanese.
For German, deletion analyses of right dislocation and afterthought were simultaneously developed by
Dennis Ott and in earlier versions of the current work. Ott’s suggestions are developed in Ott & de
Vries (2012), mine in the current paper. They nicely converge, despite some different analytical
preferences, and Dennis Ott and I do not see them as being in competition, but as reinforcing each
other.
As illustrated in (46), the accounts of Tanaka (2001) and Ott & de Vries (2012) postulate that
the right dislocated constituent originates in a second clause of its own, in which it is first fronted;
thereafter, the remainder of the second sentence is deleted by sluicing under identity with material
from the first clause.
(46)
Sluicing account of right dislocation (Tanaka 2001, Ott 2011)
ich habe sie gesehen [die Maria]i habe ich ti gesehen
I have her seen the Maria have I seen
The account pursued here is closer to the account of Kuno (1978) (written in Japanese and summed up
in Tanaka 2001): Deletion takes a repetition of the initial sentence (except for the right dislocated
constituent) as is input, without an initial step of syntactic fronting. This is illustrated in (47). I take the
deletion operation to be gapping, with the scope of gapping as worked out in Neijt (1979).
13
(47)
Gapping account of right dislocation
ich habe siei gesehen, ich habe [die Maria]i gesehen
I have her seen I have the Maria seen
As pointed out by Ott & de Vries (2012) the deletion account naturally provides a thematic
role in a larger structure as well as morphological case for the constituent dangling on the right. I add
to this the following extension. The infinitive subject PRO cannot serve as a pronoun in right
dislocation, as seen in (48).
(48) * Peter hat angeordnet [PROi die Straße zu fegen], die Arbeiteri/den Arbeiterni
Peter has ordered the street to sweep the workers
If the right dislocated element were simply base-generated on the right, on the condition that it
clarifies the reference of a referential element in the clause, we would expect this to be possible. Even
an analysis of right dislocation of movement, followed by insertion of an appropriate pronoun, might
not have a straightforward account for the impossibility of (48).
In the deletion account, the explanation is straightforward: There is no plausible source for the
deletion, as shown in (49). The infinitive in the deleted clause would not license an overt subject
instead of the PRO, as shown. (Notice that the German verb anordnen cannot take a DP object outside
of the infinitive; there is no English translation in which this restriction transfers to the translation.)
(49) * Peter hat angeordnet [PROi die Straße zu fegen],
Peter hat angeordnet [die Arbeiteri/den Arbeiterni die Straße zu fegen].
Peter has ordered (the workers) the street to sweep
An important argument against a base-generation account is provided in the form of the
connectedness effects of Ott & de Vries (2012): For many purposes the dislocated constituent behaves
as though it was in the position of the pronoun that it resumes. Consider for example (50). Though the
right dislocated element is in some sense external to the clause to its left, it shows Condition C effects
with regard to elements in that clause. This is unexpected if the right dislocated element can simply be
adjoined in a high position on the right, since it might be adjoined higher than the pronoun siei and
should then not trigger a Condition C violation.
(50)
A: Warum sieht Maria so traurig aus? Hat sie nicht einen neuen Freund?
‘Why does Mary look so sad? Doesn’t she just have a new boyfriend?’
B: *Gestern hat siei ihn mit einer Anderen gesehen, Mariasi neuen Freund.
Yesterday has she him with an other seen Marias new boyfriend
In a deletion analysis, the deleted copy of sie triggers the Condition C effect in the repetition, as
shown in (51).
(51) * Gestern hat siei ihn mit einer Anderen gesehen, gestern hat siei Mariasi neuen Freund mit einer
Anderen gesehen.
Yesterday has she him with an other seen, yesterday has she Marias new
boyfriend with an other seen
An argument against a movement analysis is a co-production with Dennis Ott (personal
communication). If right dislocation involved movement to a c-commanding position, followed by the
insertion of a pronoun in the position of the trace, the quantifier would c-command the inserted
pronoun. The possibility of a binding relation might be expected. However, binding is not possible in
right dislocation, as shown in (52).
(52)
*
Ich habe ihn gesehen, jeden Gast.
I have him seen every guest
14
This follows in a deletion analysis, where the quantifier does not c-command the pronoun:
(53)
*
Ich habe ihni gesehen, ich habe jeden Gasti gesehen.
I have him seen I have every guest seen
Is fronting before deletion empirically motivated? Ott & de Vries (2012) present Dutch
examples from Zwart (2001, 2011) that support a sluicing account with an initial step of fronting that
precedes deletion. The argument carries over to German and is shown in (54). In (54a,b), scope
between the subject and the adverb follows the surface order. The right dislocated adverb in (54c)
takes wide scope. This can be accounted for if it is fronted before deletion, as in (54d). (I return to
right dislocation of adjuncts in section 3.11.)
(54)
a.
Zwei Leute haben vermutlich Nauru gesehen
two people have presumably Nauru seen
‘Two people presumably saw Nauru.’
(2 > vermutlich)
b.
Vermutlich haben zwei Leute Nauru gesehen
presumably have two people Nauru seen
‘Presumably two people saw Nauru.’
(vermutlich > 2)
c.
Zwei Leute haben Nauru gesehen, vermutlich.
two people have Nauru seen presumably
‘Two people saw Nauru, presumably.’
(vermutlich > 2)
d.
Zwei Leute haben Nauru gesehen, vermutlich haben zwei Leute Nauru gesehen.
two people have Nauru seen, presumably have two people Nauru seen
However, there are also cases that point in the opposite direction. (55a,b) show two orders of oft
‘often’ and wem ‘to-someone’. The scope here also follows the surface order. (The indefinite wem is
used since it seems not to have a specific reading that could interfere with the scope judgments.)
Fronting of oft disambiguates in favor of wide scope of oft in (55c). This disambiguating effect is not
found with dislocated oft in (55d), where instead both readings are possible. This points towards an
analysis without fronting before movement. The two readings can be derived by gapping as shown in
(55e,f).
(55)
a.
Sie hat oft wem etwas mitgebracht.
she has often someone something brought
‘She often brought someone something.’
often > to-someone
b.
Sie hat wem oft etwas mitgebracht
she has someone often something brought
to-someone > often
c.
Oft hat sie wem etwas mitgebracht
often has she someone something brought
often > to-someone
d.
Sie hat wem etwas mitgebracht oft
she has someone something brought often
to-someone > often
or often > to-someone
e.
Sie hat wem etwas mitgebracht, sie hat wem oft etwas mitgebracht.
she has someone something brought she has someone often something brought
f.
Sie hat wem etwas mitgebracht, sie hat oft wem etwas mitgebracht.
she has someone something brought she has often someone something brought
15
Given this, it is possible that the judgment in (54d) reflects unmarked order, rather than an initial step
of fronting. However, the issue requires further study and will not be resolved here. The main point of
the current section is that there is a range of syntactic arguments in favor of a deletion analysis of right
dislocation and afterthought.
3.2. Root sentences, sentence stress and focus
Reis (1997) and Holler (2008) separate clauses that depend on another clause into those that are
structurally embedded in the higher clause (‘integrated’) and those that are not embedded in the higher
clause (‘non-integrated’). Integration and structural embedding are synonyms for these authors.
Dependency is a separate notion. A clause (such as an appositive relative) is dependent on another one
if it does not normally occur on its own for whatever reason, though this dependency crucially does
not entail embedding for these authors. As discussed below, this separation of dependency and
structural embedding goes back to earlier literature that includes McCawley (1982) for parentheticals
and Safir (1986) for appositive relatives.
Here I suggest to unify the notion of non-integrated, i.e. non-embedded clauses with the notion
of the root sentence of B. Downing (1970). Downing shows a correlation of root sentences with
‘obligatory pause’, or what we call intonation phrase boundaries today (Nespor and Vogel 1986).
B. Downing (1970)10 is a 215-page thesis that defends a single generalization in English in
considerable detail. Certain main clauses, root sentences (in the sense of the root S node of Emonds
1970), syntactically exclude certain other elements in the structure. While the exclusion is syntactic in
nature, it has the consequence of prosodic separation by an obligatory pause. Some core cases are: (a)
coordinated sentences are separate root sentences; (b) appositive relative clauses are not part of the
root sentence that is the main clause; (c) parentheticals are not part of the root sentence that is the main
clause; (d) fronted adverbial clauses, left dislocated elements and topicalized themes in English are not
within the root sentence that is the main clause; (e) vocatives are not part of the root sentence. On the
other hand, these elements are syntactically included in the root sentence: (i) fronted wh-phrases,
fronted constituents in V2-constructions and fronted foci; (ii) extraposed constituents on the right
(‘extraposed’ in today’s sense as in section 1); (iii) a case of what we might call subject correlate
construction in English (‘correlate’ in the sense of section 2.3); (iv) at least some adverbial clauses at
the right periphery.
Reis (1997) takes argument and many adjunct clauses to be integrated, while coordinated
sentences, appositive relatives, and certain high adverbial clauses are non-integrated and nonembedded in the relevant sense. For high adverbial clauses on the right, there is no theoretical
discrepancy between Downing and Reis. Central adverbial clauses are embedded in the clause to
which they attach on the right, peripheral adverbial clauses are not (Haegeman 2004, for German Frey
2012). The high adverbial clauses in Reis (1997) are arguably of the latter kind.
Holler (2008) classifies German V2-relative clauses (see also Gärtner 2000), weil-V2-clauses
and continuative relative clauses (see also Holler 2005: a species of non-restrictive relative that spells
out events following those in the main clause) as non-integrated. Both authors also establish additional
distinctions at the divide between integrated and non-integrated clauses, though these are not directly
relevant to the current paper.
B. Downing (1970) provides a syntactic definition of root sentences. For coordinated
sentences this cannot simply be ‘the highest node in a phrase-marker’ because the node representing
the conjunction is higher. Downing therefore first defines the notion of a predicative sentence as one
in which the S node immediately dominates a VP node. In terms of this, he defines a root sentence as
one that is not dominated by a predicative sentence. Thus, being dominated by a higher node of
conjunction does not constitute embedding, while being dominated by a sentential node that represents
a ‘regular’ sentence with a predicate constitutes embedding and embedded clauses in that sense are not
root sentences. Reis (1997) does not provide an explicit syntactic definition of non-integrated clauses
but pursues a similar intuition: She mentions sentence coordination as a case of bona fide absence of
embedding and points out that appositive relatives might not have any more of a structural connection
10
Downing’s thesis seems not to be available in most university libraries. A pdf can be obtained upon
request from the author of this paper.
16
with their host clause than coreference of the relative pronoun, where coreference is a process that also
obtains in discourse and does not require structural attachment.
It is plain that the notion of embedding vs. its absence in the host clause are the crucial
defining characteristics both of Downing’s root sentences and of the divide between embedded
(integrated) and non-embedded (non-integrated) clauses of Reis and Holler. In the terminology, we
want a term that generalizes across a simple main clause on its own and a dependent non-integrated
clause. The notion root sentence generalizes across these two. The notion ‘non-integrated’ does not
generalize as readily, since it would be odd to call a simple main clause standing on its own a ‘nonintegrated clause’: There is nothing that it could be integrated with. I will therefore adopt the terms
root sentence from Downing. For concreteness, I work with the following definition, which is in
essence that of Downing (1970):
(56)
A root sentence is a CP node not dominated by higher nodes apart from CP nodes created by
coordination and adjunction.
We may assume that root sentences have the following properties, which I address in the
following.
(57)
Properties of root sentences (RS):
a.
Syntax: No element in a RS c-commands elements outside of the RS.
b.
Prosody: Each UC is delimited by intonation phrase boundaries (B. Downing 1970)
and carries sentence stress.
d.
Information structure: A focus-background structure is limited to being established
within a UC and may not include elements outside of the UC.
c.
Semantics: The computation of compositional at-issue content stops at the level of
UCs, and their connection to elements outside of the URCs is at the level of discourse
connections that is also related to speech acts (Potts 2005).
In a sense, the intuition across modules is that the root sentence is the level at which regular clauseand sentence-internal connections end. I illustrate with appositive relative clauses, a central case for
Reis and Downing and, via the continuative relative clauses, also for Holler. It may usefully be
compared with restrictive relative clauses. Consider first (58). It is not possible to c-command from a
main clause into an appositive relative clause as shown in (58b) (Safir 1986). This is surprising
because the appositive appears to be genuinely contained in the root clause on the surface. Yet this is
not really so in the phrase structure, as the c-command test shows.
(58)
a.
Keiner1 hat einen Klassenkameraden, den er1 aus der Volksschule kennt,
wiedererkannt.
‘Noone1 recognized a classmate that he1 knows from elementary school.’
b. *
Keiner1 hat den Peter, den er1 ja aus der Volksschule kennt, wiedererkannt.
‘Noone1 recognized Peter, whom he1 knows, after all, from elementary school.’
This led Safir to suggest a surface representation of appositive relatives in which they are not part of
the structure of the main clause as in (59).
(59)
CP
Keiner hat den Peter
wiedererkannt
‘noone recognized Peter’
CP
den er ja aus der Volksschule kennt
17
‘whom he knows, after all, from
elementary school’
McCawley (1982) made a related suggestion for parentheticals, by which they are attached to the root
node of the main clause. Applied to our example this proposal is represented as in (60).
(60)
CP
Keiner hat den Peter
wiedererkannt
CP
den er ja aus der Volksschule kennt
These observations and suggestions give a sense of the high degree of syntactic separation of elements
that are not part of the root sentence, even where this is not apparent from their surface order. Under
either representation (59) or (60) both the main clause and the appositive relative are separate root
sentences by (56). Either representation, as well as (56), predicts the absence of c-command relations
across main clause and elements excluded from it.
It is important to distinguish root sentences or Downing’s root sentences from root clauses. As
mentioned Downing (1970) took the notion of the root from Emonds (1970), who distinguished root
transformations that could only apply in root S (sentence) nodes from structure-preserving
transformations that could apply everywhere. It was later discovered by Hooper & Thompson (1973)
that root transformation could also apply in a limited range of (assertive-like) embedded
environments. Today the notion of the root clause is more familiar. It characterizes both Downing’s
root sentences and certain clauses in these assertive-like environments of embedding (see Heycock
2006). Notice that the appositive relative in (60) is not such an embedded root clause. Notice also that
today root clauses are sometimes identified with ForcePs (Haegeman 2004).
Semantically it seems that the computation of at issue content in the sense of Potts (2005)
stops at the level of the root sentence, and discourse connections enter into the picture, to which the
coreference relation commented on by Reis (1997) but also speech acts belong (Reis 2006).
Turning to the prosody, B. Downing (1970) suggested that obligatory pauses are inserted at
the edges of root sentences. He saw obligatory pauses as edges of prosodic constituents. Nespor &
Vogel (1986) adopted Downing’s suggestion and identified his prosodic domain of obligatory pauses
as intonation phrases. Compatible suggestions are made in terms of the comma feature of Potts (2005)
in Selkirk (2005), and in terms of the notion non-selected clauses in L. Downing (2011).
(61)
Each root sentence is identified with an intonation phrase.
I assume that this intonation phrase can be further divided into multiple intonation phrases for other
reasons, so that the crucial effect of (61) is the presence of at least one intonation phrase in each root
sentence.
I furthermore adopt an integrated account of prosodic and metrical phonology like that of
Nespor & Vogel (1986, 1989), in which each prosodic constituent is also a stress-domain. The
intonation phrase, in particular, is the domain of what I call sentence stress. This conception also
underlies the suggestion of rightmost strengthening in (4) by Uhmann (1991), which is strengthening
of the rightmost phrasal stress in the intonation phrase. It will be useful to separate an aspect of
Uhmann’s rightmost strengthening as in (62). It is an aspect of the Faithfulness Condition of Hayes
(1995) that prosodic domains and their stress-peaks are in a one-to-one relation.
(62)
Each intonation phrase must contain sentence stress.
A root sentence is then mapped to one (or more) intonation phrases by (61). The resulting intonation
phrase must contain sentence stress by (62). Taken together, then, (61) and (62) have the effect in (63).
(63)
Each root sentence must contain sentence stress.
18
A main clause and an appositive relative are then separate root sentences. By (63) they must
each contain sentence stress. This is as desired, and is illustrated in (64). Here the modal particle ja
enforces the appositive status of the relative clause.
(64)
Ich prüfe den Peter, der ja spät dran ist.
I examine the Peter who MP late running is
‘I am examining Peter who is, after all, running late.’
The examples in (65) bear on whether this also obtains under duress by contextual givenness. (65a)
shows a comparison case with a restrictive relative clause. Restrictive relative clauses are syntactically
and prosodically embedded or integrated for both Downing (1970) and Reis (1997). As shown in (65a)
contextual givenness of the main clause allows for a main clause without sentence stress, so long as
the relative clause carries sentence stress. Importantly, the same is not possible in (65b) with an
appositive relative clause. Here the main clause must carry sentence stress as in (65c) even when it is
contextually given.
(65)
Prüfst du jemanden?
‘Are you examining someone?’
a.
Ja, [ich prüfe jemanden]G, der spät dran ist.
[UC
]
‘Yes, I am examining someone who is running late.’
Prüfst du den Peter?
‘Are you examining Peter?’
b. *
c.
Ja, [ich prüfe den Peter]G, der ja spät dran ist.
Ja, [ich prüfe den Peter]G, der ja spät dran ist.
[UC
][UC
]
‘Yes, I am examining Peter, who is, after all, running late.’
The analysis of (65a) is that the main clause and the restrictive relative clause form a root sentence
together by (56). The demand that the root clause carries sentence stress is satisfied, since this root
sentence contains sentence stress in the relative clause. Within this large root sentence, givenness
shifts the stress away from the main clause. This is not in conflict with the stress-demand on root
sentences.
In (65b), on the other hand, the main clause is a root sentence of its own, to which the
appositive does not belong. The appositive constitutes a separate root sentence. Removing the stress
from the main clause leaves a root sentence without sentence stress, in violation of (63). Since this is
ruled out, we conclude that (63) is stronger than the stress-rejecting effect of givenness that is
formulated in (8). This is highlighted in (66).
(66)
(63) is stronger than the stress-rejecting effect of givenness in (8).
Notice that this effect also obtains in the absence of material outside of the root sentence. This is
implicit in an observation by Höhle (1992), who notes that (next to verum focus) (67b), in the context
of (67a), has a rendition as shown. In this case the content of (67b) is entirely given in the preceding
discourse. Yet sentence stress must be assigned to the clause. Here, as in (65c), (63) overrides the
stress-rejecting effect of givenness.
19
(67)
a.
Ich habe Hanna gefragt, was Karl gerade macht, und sie hat die alberne Behauptung
aufgestellt, dass er ein Drehbuch schreibt.
‘I asked Hanna what Karl is currently doing, and she made the silly claim that he is
writing a script.’
b.
(das stimmt) Karl schreibt ein Drehbuch.
‘(that’s right) Karl is writing a script.’
Let us turn to the effects of focus. In the terms of Reis (1997:128), integrated, embedded
clause can share a single focus with the main clause, while separate root sentences each require a
focus. Reis (1997) thus correlates separate non-integrated clauses with separate focus-background
structures (in her terms). In our current terms, this translates into two observations. For one thing, in
(65a) the main clause and the embedded clause can share a single sentence stress, while in (65b,c)
each root sentence requires a separate sentence stress. For another it is not possible to construct a
single focus-background structure (in the terms of Reis and the current paper) across two root
sentences. In (68) the focus-background structure is again shown in terms of the operator ~[...] from
Rooth (1992). A focus-background structure is possible across main clause and restrictive relative
clause in (68a). It is not possible across main clause and appositive relative in (68b,c). While (68b) is
also ruled out by (63) because the main clause lacks sentence stress, (68c) does not violate that
demand and thus shows in addition that a focus-background structure is not possible across the two
root clauses.
(68)
Welchen neuen Studenten hast du geprüft?
‘Which new student did you examine?’
a.
b. *
c. #
~[Ich habe den neuen Studenten geprüft, [den du kennst]F].
~[Ich habe den neuen Studenten geprüft, [den du ja kennst]F].
~[Ich habe den neuen Studenten geprüft, [den du ja kennst]F].
I have the new student examined that you MP know
We cast this in terms of the prohibition in (69).
(69)
A focus-background structure is confined to within a root sentence.
Thus (68a) does not violate (69) because the main clause is not a separate root clause here. On the
other hand (68b,c) are ruled out by (69).
In summary, root sentences require sentence stress and this requirement cannot be undone by
information structure. Further, root sentences delimit focus-background structures.
3.3. Analysis of extraposition, RD and AT
This analysis is now straightforwardly extended to the distinction in prosody and information structure
between extraposition on the one hand and RD/AT on the other. For an extraposed structure, we may
adopt the classical assumption that it involves movement and adjunction on the right as in (70), though
the outcome is the same under a range of alternative analysis in the syntax or at PF such as different
linearization with retention of the underlying structure. Downing (1970) also classified extraposed
structures are remaining within the root sentence. According to (56) there is only a single root
sentence in (70), namely the entire CP. Neither the extraposed element, nor a group of elements to its
left taken together, constitute a separate root sentence. (Here because they are not clauses, but if the
extraposed constituent were a clause, it would still not be a separate root sentence by (56), since it
would be embedded in the matrix clause.)
20
(70)
CP
Maria hat
VP
VP
PP
ein Buch t gelesen von Chomsky
(
x
)INTONATION PHRASE
Maria has a book read by Chomsky
The observations on extraposed constituents from section 1 now correctly follow. (63) requires that
sentence stress is found somewhere in this all-embracing intonation phrase. There is therefore no
principled obstacle to the extraposed constituent carrying the only sentence stress. Sentence stress may
fall on the extraposed constituent and the preceding elements may be stressless, since the preceding
elements and the extraposed constituent together constitute the root sentence. Similarly, a focusbackground structure may freely be build across the extraposed constituent and the preceding elements
without conflicting with (69). Taken together, the effects captured above in the formulations in (16a)
and (23a) are now accounted for in terms of (56), (63) and (69).
RD and AT are not discussed by Downing (1970). I adopt the deletion analysis as shown for
AT in (71).
(71)
Afterthought (AT)
CP
CP
Ich habe jemanden gesehen
ich habe die Maria gesehen
(
I have someone seen
x
)I (
x
)I
I have the Maria seen
The second CP is independent of the first CP in what may be a parallel structure (see also Ott & de
Vries 2012). By the definition in (56) both CPs constitute root sentences. By (56) both form intonation
phrases as shown in the last line of (71). By (62) each of them requires sentence stress. This is the
correct result for AT, as in the example (71). As in the discussion of relative clauses, the demand to
carry sentence stress can not be overridden by information structure. It correctly follows that the
preceding clause must carry sentence stress (i.e. we derive (16b)). It furthermore correctly follows
from (69) that no focus-background structure may be built across the two root clauses (i.e. we also
derive (23b)). Without further ado we also derive that the dislocated constituent carries sentence
stress, which correctly characterizes AT, as in the example (71).
What, then, about the case of RD? Here the dislocated constituent does not carry sentence
stress. I tentatively suggest that the distinction between AT and RD is in terms of root sentences:
While the dislocated constituent in both constructions are outside of the root sentence that precedes it,
it must count as the remnant of a root sentence in the case of AT but not in RD. Thus, if (71) is the
correct representation of AT, (72) must be the corresponding representation of RD. The second
unembedded CP is crossed out. We must assume that it is deleted together with the overtly deleted
material.
21
(72)
Right dislocation (RD)
CP
CP
Ich habe sie gesehen
ich habe die Maria gesehen
(
I have her seen
x
)I
I have the Maria seen
This is a reasonable way of structuring the terrain for the purpose of this paper. First, the RD
constituent does not carry sentence stress (by the descriptive definition of RD). This is represented
here in terms of the absence of a second root clause. Correspondingly (63) does not lead to a
requirement of sentence stress on the RD constituent. Second, the preceding clause requires sentence
stress under any information structure and regardless of the RD constituent, because it is a root
sentence to the exclusion of the RD constituent. Third, no focus-background structure may be built
across the RD constituent and the preceding clause by (69) because the preceding clause is a root
sentence that does not include the RD constituent.
In summary, the separation of RD and AT in terms of stress and focus, not shared by
extraposition, can plausibly be represented in terms of the dislocated constituent in RD/AT being
outside of the root sentence which they follow. This follows correctly from the deletion analysis of RD
and AT. The preceding clause must carry sentence stress because it is a root sentence to the exclusion
of the dislocated constituent. A focus-background structure cannot be built across the dislocated
constituent and the preceding clause because the dislocated constituent is not part of the preceding root
sentence.
3.4. On the order of stress-assignment and deletion in RD and AT
Consider first a case observed by Zwart (2001) in Dutch: It is possible to resume an indefinite pronoun
with a stressed constituent as in (73). As Zwart points out, stressless resumption as in (74) is not
possible.
(73)
(74)
a.
Ich habe jemandem etwas mitgebracht – einem Kollegen.
I have someone something brought a colleague
‘I brought something to someone, to a colleague.’
b.
Ich habe jemandem etwas mitgebracht – dem Peter.
I have someone something brought the Peter
‘I brought something to someone, to Peter.’
a. *
b. *
Ich habe jemandem etwas mitgebracht, einem Kollegen.
Ich habe jemandem etwas mitgebracht, dem Peter.
The deletion analysis is plausible for these cases in that it has an overt, independently acceptable
source, illustrated in (75a). Deletion from this source is illustrated in (75b).
(75)
a.
b.
Ich habe jemandem etwas mitgebracht – ich habe einen Kollegen etwas mitgebracht.
Ich habe jemandem etwas mitgebracht – ich habe einen Kollegen etwas mitgebracht.
I have someone something brought I have a colleague something brought
The stress is obligatory in the undeleted version, cf. (76), as it is in the elliptical version, cf. (73)/(74).
(76) * Ich habe jemandem etwas mitgebracht – ich habe einem Kollegen etwas mitgebracht.
I have someone seen I have a colleague seen
22
Since stress falls on the same ellipsis remnant whether or not deletion applies in (75a,b), this
cases is compatible with ordering deletion before stress-assignment and with ordering stressassignment before deletion.
Consider then another case of obligatory afterthought. It involves “resumption” of negation as
in (77).
(77)
a.
Ich habe ihn nicht gesehen – an keinem Tag / * an keinem Tag.
I have him not seen on no day on no day
lit.: ‘I didn’t see him, on no day.’
b.
Ich habe sie nicht gefunden – an keinem Ort / * an keinem Ort.
I have her not found in no place in no place
lit.: I didn’t find her, in no place.’
The deletion analysis is also plausible here. It is illustrated in (78).
(78)
a.
Ich habe ihn nicht gesehen – ich habe ihn an keinem Tag gesehen.
I have her not seen I have her on no day seen
b.
Ich habe sie nicht gefunden – ich habe sie an keinem Ort gefunden.
I have her not found I have her in no place found
As far as stress-assignment is concerned, this case is not identical to resumption of indefinites. On the
one hand, the sentential continuation must carry sentence stress somewhere, as can be seen in (79d).
On the other hand, the continuation prefers a stress-pattern in which the final verb is stressed in (79a)
to one in which the negation-resuming element is stressed in (79b,c).
(79)
a.
b. ?
c. ?
d. *
Ich habe ihn nicht gesehen – ich habe ihn an keinem Tag gesehen.
Ich habe ihn nicht gesehen – ich habe ihn an keinem Tag gesehen.
Ich habe ihn nicht gesehen – ich habe ihn an keinem Tag gesehen.
Ich habe ihn nicht gesehen – ich habe ihn an keinem Tag gesehen.
I have him not seen I have him on no day seen
These facts are similar for the source of (77b):
(80)
a.
b. ?
c. ?
d. *
Ich habe sie nicht gefunden – ich habe sie an keinem Ort gefunden.
Ich habe sie nicht gefunden – ich habe sie an keinem Ort gefunden.
Ich habe sie nicht gefunden – ich habe sie an keinem Ort gefunden.
Ich habe sie nicht gefunden – ich habe sie an keinem Ort gefunden.
I have her not found I have her in no place found
Similar effects can be seen for resumption of an adverb as in (81) and (82).
(81)
(82)
a.
Ich habe sie oft gesehen – jeden Tag /* jeden Tag.
I have her often seen every day every day
b.
Ich habe sie oft gesehen – ich habe sie jeden Tag gesehen.
I have her often seen I have her every day seen
a.
b. ?
c. ?
d. *
Ich habe sie oft gesehen – ich habe sie jeden Tag gesehen.
Ich habe sie oft gesehen – ich habe sie jeden Tag gesehen.
Ich habe sie oft gesehen – ich habe sie jeden Tag gesehen.
Ich habe sie oft gesehen – ich habe sie jeden Tag gesehen.
I have her often seen I have her every day seen
23
The preceding cases provide us with motivation for a particular order of stress-assignment and
deletion. This is because the location of the stress on the ellipsis remnants in (78) and (81a) is not
matched by the preferred location of sentence stress in the non-elided versions in (79a), (80a) and
(82a). The analysis therefore orders deletion before stress-assignment, as shown in (83): Deletion
applies first as in (83a). After deletion, sentence stress in the second part of the utterance is assigned to
the ellipsis remnant for lack of a different option.
(83)
a.
Ich habe ihn nicht gesehen – ich habe ihn an keinem Tag gesehen.
I have him not seen I have him on no day seen
b.
Ich habe ihn nicht gesehen – an keinem Tag.
I have him not seen on no day
See Truckenbrodt (2013:139f) for arguments that the order ‘deletion before stress-assignment’ is
independently motivated for gapping and right-node-raising in coordinated structures. See Tancredi
(1992) and Hartmann (2000) for a different view in which only backgrounded, unstressed elements
can be deleted.
The order ‘deletion before stress-assignment’ is compatible with the suggestion made for RD
in connection with (72): If deletion applies first, the CP node is first deleted in the derivation of RD.
When stress is assigned after the deletion, the CP node is no longer there and so (63) does not require
sentence stress on the dislocated constituent. The order we postulated is highlighted in (84).
(84)
In a deletion analysis of RD and AT, deletion precedes stress-assignment.
The distinction between AT and RD is further pursued in [Truckenbrodt/... in preparation].
4. Summary
Two important distinctions between extraposition and RD/AT were shown. For both distinctions
extraposition is part of the clause to which it attaches, while RD/AT is not. One distinction is prosodic,
the other concerns the information structure. In the prosody, the syntactic domain [...] for which a
sentence requires sentence stress includes the extraposed XP, as in [clause + XP]. With RD and AT,
the dislocated constituent is outside of this domain of sentence stress, so that the preceding clause
requires sentence stress regardless of the dislocated constituent: [clause]+XP. In the information
structure, the same domain [...] blocks the presence of a focus-background structure across its edges.
A focus background structure can therefore be built across the clause and an extraposed XP in [clause
+ XP]. but not across the clause and a dislocated XP in [clause]+XP.
We then placed earlier findings of Pütz (1986), Sudhoff (1993), Frey (2010) and others about
correlate constructions relative to this analysis. We confirmed the existence of genuine correlates:
Under certain verbs like bedauern ‘regret’ a clause on the right may be represented by a pronominal
correlate in the clause, yet it acts as though extraposed to the right, rather than right-dislocated.
The distinction between extraposition (including genuine correlates) on the one hand and
RD/AT on the other hand was then given an analysis in terms of root sentences [...], which are
identified with the non-integrated clauses of Reis (1997) and Holler (2008): They are the domains that
require sentence stress and they do not allow a focus-background structure across them. Extraposition
involves dislocation within the same root sentence: [clause + XP]. RD and AT involve syntactic
deletion in a separate root sentence: [clause] + [... RD/AT ...]. This correctly separates extraposition on
the one hand from RD/AT on the other. It was also suggested that the distinction between RD and AT
can be cast in terms of the retention of the second root clause after deletion: [clause]+RD but
[clause]+[AT]. The second root clause would be deleted in RD but not in [AT], so that its requirement
for sentence stress is retained in the case of AT but not in RD.
24
References
Altmann, Hans. 1981. Formen der "Herausstellung" im Deutschen. Rechtsversetzung,
Linksversetzung, Freies Thema und verwandte Konstruktionen. Tübingen: Niemeyer.
Averintseva-Klisch, Maria. 2006. The 'separate performative' account of the German right dislocation.
In Proceedings of Sinn und Bedeutung 10, C. Ebert & C. Endriss (Hg), Berlin: ZASPiL 44,
15-28.
—. 2009. Rechte Satzperipherie im Diskurs. NP-Rechtsversetzung im Deutschen. Tübingen:
Stauffenburg.
Beckman, Mary E. & Janet B. Pierrehumbert. 1986. Intonational Structure in Japanese and English.
Phonology Yearbook 3, 255-309.
Bennis, Hans. 1986. Gaps and dummies. Dordrecht: Foris.
Bresnan, Joan. 1971. Sentence stress and syntactic transformations. Language 47, 257-81.
—. 1972. Stress and syntax: a reply. Language 48, 326-42.
Cinque, Guglielmo. 1993. A null theory of phrase and compound stress. Linguistic Inquiry 24, 239-97.
Downing, Laura. 2011. The prosody of 'dislocation' in selected Bantu langauges. Lingua 121, 772-86.
Emonds, Joseph. 1970. Root and structure-preserving transformations. Bloomington, Ind.: IULC.
Féry, Caroline. 1993. German intonational patterns. Tübingen: Niemeyer.
Féry, Caroline & Frank Kügler. 2008. Pitch accent scaling on given, new and focused constituents in
German. Journal of Phonetics 36, 680-703.
Féry, Caroline & Vieri Samek-Lodovici. 2006. Focus projection and prosodic prominence in nested
foci. Language 82, 131-50.
Frey, Werner. 2010. On the 'nominal' character of clauses associated with a pronominal. Paper
presented to the DGfS Jahrestagung 2011, Workshop 'Inner-sentential propositional proforms: syntactic properties and interpretative effects', 2010.
—. 2012. On two types of adverbial clauses allowing root-phenomena. In Main Clause Phenomena:
New Horizons, L. Aelbrecht, L. Haegeman & R. Nye (Hg), Amsterdam: Benjamins, 405-29.
Gärtner, Hans-Martin. 2000. Are there V2 relative clauses in German? Journal of Comparative
Germanic Linguistics 3, 97-141.
Grice, Martine, Stefan Baumann & Ralf Benzmuller. 2005. German intonation in autosegmentalmetrical phonology. In Prosodic typology: the phonology of intonation and phrasing, S.-A.
Jun (Hg), Oxford: Oxford University Press, 55-83.
Gussenhoven, Carlos. 1983. On the grammar and semantics of sentence accents. Dordrecht: Foris.
—. 1992. Sentence accents and argument structure. In Thematic structure, its role in grammar, I. Roca
(Hg), Berlin, New York: Foris, 79-106.
Haegeman, Liliane. 2004. Topicalization, CLLD and the left periphery. In ZAS Papers in Linguistics
35: Proceedings of the Dislocated Elements Workshop, C. Maienborn, W. Frey & B. Shaer:
ZAS Berlin, 157-92.
Hartmann, Katharina. 2000. Right node raising and gapping. Interface conditions on prosodic
deletion. Amsterdam: John Benjamins.
Hayes, Bruce. 1995. Metrical stress theory: principles and case studies. Chicago: The University of
Chicago Press.
Heycock, Caroline. 2006. Embedded Root Phenomena. In The Blackwell companion to syntax, Vol. II,
M. Everaert & H. van Riemsdijk (Hg), Oxford: Blackwell, 174-209.
Höhle. 1982. Explikation für "normale Betonung" und "normale Wortstellung". In Satzglieder des
Deutschen, W. Abraham (Hg), Tübingen: Narr, 75-153.
Höhle, Tilman N. 1992. Über Verum-Fokus im Deutschen. In Informationsstruktur und Grammatik, J.
Jacobs (Hg), Opladen: Westdeutscher Verlag, 112-41.
Holler, Anke. 2005. Weiterführende Relativsätze. Empirisiche und theoretische Aspekte. Berlin:
Akademie-Verlag.
—. 2008. German dependent clauses from a constraint-based perspective. In "Subordination" vs.
"coordination" in sentence and text, C. Fabricius-Hansen & W. Ramm (Hg), Amsterdam:
John Benjamins, 187-216.
Hooper, Joan B. & Sandra A. Thompson. 1973. On the applicability of root transformations.
Linguistic Inquiry 4, 465-97.
25
Jackendoff, Ray S. 1972. Semantic interpretation in generative grammar. Cambridge, Mass.: MIT
Press.
Jacobs, Joachim. 1993. Integration. In Wortstellung und Informationsstruktur, M. Reis (Hg),
Tübingen: Max Niemeyer Verlag.
Kratzer, Angelika & Elisabeth Selkirk. 2007. Phase theory and prosodic spellout: the case of verbs.
The Linguistic Review 24, 93-135.
Krifka, Manfred. 1984. Focus, Topic, syntaktische Struktur und semantische Interpretation.
Universität München.
Kuno, Susumo. 1978. Danwa-no bunpoo [Grammar of discourse]. Tokyo: Taishuukan-shoten.
Ladd, D. Robert. 1983. Even, focus, and normal stress. Journal of Semantics 2, 257-70.
McCawley, James D. 1982. Parentheticals and discontinuous constituent structure. Linguistic Inquiry
13, 91-106.
Müller, Gereon. 1995. On extraposition and successive cyclicity. In Extraction and Extraposition in
German, U. Lutz & J. Pafel (Hg), Amsterdam: Benjamins, 213-43.
Neijt, Anneke. 1979. Gapping: a contribution to sentence grammar. Dordrecht: Foris.
Nespor, Marina & Irene Vogel. 1986. Prosodic Phonology. Dordrecht: Foris.
—. 1989. On clashes and lapses. Phonology 6, 69-116.
Ott, Dennis. 2011. Eliminating constructions: the case of dislocation. Paper presented at the The
minimalist program: quo vadis? – Newborn, reborn, or stillborn?
Ott, Dennis & Mark de Vries. 2012. Right-dislocation as deletion.
Pheby, J. 1981. Phonologie: Intonation. In Grundzüge einer deutschen Grammatik, K.E. Heidolph et
al. (Hg), Berlin: Akademie-Verlag, 839-97.
Pierrehumbert, Janet Breckenridge. 1980. The Phonology and Phonetics of English Intonation.
Massachusetts Institute of Technology. Ph.D. thesis.
Potts, Christopher. 2005. The logic of conventional implicatures. Oxford, UK; New York, USA:
Oxford University Press.
Pütz, H. 1986. Über die Syntax der Pronominalform >es< im modernen Deutsch. Tübingen: Narr.
Reis, Marga. 2006. Is German V-to-C movement really semantically motivated? Some empirical
problems. Theoretical Linguistics 32, 369-80.
Rooth, Mats. 1992. A theory of focus interpretation. Natural Language Semantics 1, 75-116.
Safir, Ken. 1986. Relative clauses in a theory of binding and levels. Linguistic Inquiry 17, 663-89.
Schwabe, Kerstin. 2011. The German sentential proform es in all-focs sentences. ZAS Berlin.
Schwabe, Kerstin & Robert Fittler. 2011. Syntax and logical consistency conditions for German
matrix predicates. ZAS Berlin.
Schwarzschild, Roger. 1999. Givenness, AvoidF and other constraints on the placement of accent.
Natural Language Semantics 7, 141-77.
Selkirk, Elisabeth. 1984. Phonology and syntax: the relationship between sound and structure.
Cambridge, Mass: MIT Press.
—. 1995. Sentence prosody: intonation, stress, and phrasing. In The handbook of phonological theory,
J. Goldsmith (Hg), Cambridge, Mass: Blackwell, 550-69.
—. 2005. Comments on intonational phrasing in English. In Prosodies. With special reference to
Iberian languages, S. Frota, M. Vigário & M.J. Freitas (Hg), Berlin, New York: Mouton de
Gruyter, 11-58.
—. 2008. Contrastive focus, givenness, and the unmarked status of "discourse new". Acta Liguistica
Hungarica 55, 331-46.
Sudhoff, Stefan. 2003. Argumentsätze und es-Korrelate. Wissenschaftlicher Verlag Berlin.
Tanaka, Hidekazu. 2001. Right-dislocation as scrambling. Journal of Linguistics 37, 551-79.
Tancredi, Christopher Damian. 1992. Deletion, deaccenting, and presupposition. Massachusetts
Institute of Technology. Doctoral thesis.
Truckenbrodt, Hubert. 1995. Phonological phrases: their relation to syntax, focus, and prominence.
MIT. Doctoral thesis.
—. 2002. Upstep and embedded register levels. Phonology 19, 77-120.
—. 2004. Final lowering in non-final position. Journal of Phonetics 32, 313-48.
—. 2006. Phrasal Stress. In The Encyclopedia of Languages and Linguistics, 2nd edition, Vol. 9, K.
Brown (Hg), Oxford: Elsevier, 572-79.
26
—. 2007. Upstep of edge tones and of nuclear accents. In Tones and tunes. Volume 2: Experimental
studies in word and sentence prosody, C. Gussenhoven & T. Riad (Hg), Berlin: Mouton, 34986.
—. 2012. On the prosody of German wh-questions. In Prosody and meaning, G. Elordieta & P. Prieto
(Hg), Berlin: Mouton, 73-118.
—. 2013. An analysis of prosodic F-effects in interrogatives: prosody, syntax and semantics. Lingua
124, 131-75.
—. in press. Effects of indefinite pronouns and traces on verb stress in German. In Prosody Matters:
Essays in Honor of Lisa Selkirk, T.J. Borowsky, S. Kawahara & T. Shinya (Hg), London:
Equinox Publishing.
—. to appear. Notes on stress reconstruction and syntactic reconstruction. In Reconstruction effects in
relative clauses, M. Krifka & M. Schenner (Hg), Berlin: Akademie-Verlag.
Truckenbrodt, Hubert & Isabelle Darcy. 2010. Object clauses, movement, and phrasal stress. In The
sound patterns of syntax, N. Shir & L. Rochman (Hg), Oxford: Oxford University Press, 189216.
Uhmann, Susanne. 1991. Fokusphonologie. Eine Analyse deutscher Intonationskonturen im Rahmen
der nicht-linearen Phonologie. Tübingen: Niemeyer.
Vikner, Sten. 1995. Verb Movement and Expletive Subjects in the Germanic Languages. Oxford:
Oxford University Press.
von Stechow, Arnim & Susanne Uhmann. 1986. Some remarks on focus projection. In Topic, focus,
and configurationality, W. Abraham & S.d. Meij (Hg), Amsterdam and Philadelphia: John
Benjamins, 295-320.
Wagner, Michael 2005. Prosody and recursion. Massachusetts Institute of Technology. Doctoral
thesis.
Zifonun, Gisela, Ludger Hoffmann & Bruno Strecker. 1997. Grammatik der deutschen Sprache, Band
2. Berlin/New York: Walter de Gruyter.
Ziv, Yael & Barbara Grosz. 1994. Right dislocation and attentional state. In The Israel Association for
Theoretical Linguistics. Proceeding of the 9th Annual Conference and workshop on
Discourse, A. Mittwoch & R. Buchalla (Hg), Jerusalem: Akademon, 184-99.
Zwart, Jan-Wouter. 2001. Backgrounding ('right-dislocation') in Dutch. University of Groningen
(available at http://www.let.rug.nl/~zwart/docs/backgr.pdf).
—. 2011. The syntax of Dutch. Cambridge: Cambridge University Press.
27
© Copyright 2026 Paperzz