Content uploaded by Eva Wittenberg
Author content
All content in this area was uploaded by Eva Wittenberg on Mar 04, 2015
Content may be subject to copyright.
1
Topicalization in German particle verb constructions:
The role of semantic transparency
Andreas Trotzkea, Stefano Quagliaa & Eva Wittenbergb
a University of Konstanz, b UC San Diego
Abstract
In this paper, we investigate topicalization patterns of German particle verbs by comparing the syntactic
behavior of semantically transparent and non-transparent particle verb constructions. We propose a classi-
fication that allows us to cover the whole transparency spectrum and to distinguish between fully transpar-
ent and fully non-transparent particle verbs. Given this classification, we report on a questionnaire study
that provides empirical evidence for the claim that information structural constraints in combination with
the degree of semantic transparency govern topicalization patterns in particle verbs. We conclude by
pointing out potential additional constraints on topicalization in particle verbs that go beyond information
structure.
Keywords: particle verbs; topicalization; semantic transparency; information structure; acceptability
judgments
1 Introduction
In this paper, we investigate topicalization patterns of German particle verbs by distinguishing
different classes of particle verbs in terms of semantic transparency. While the occurrence of the
whole particle verb in the prefield of the clause is a common option in German syntax, the topi-
calization of only the particle is classified differently in the literature. In particular, scholars
claiming that particle verbs are complex words rather than proper syntactic constructions often
doubt the acceptability of those configurations (Eisenberg 1999; Stiebels & Wunderlich 1994).
Fuhrhop (2007: 50) even goes so far as to claim that prepositional particles such as aus (‘out’) or
ein (‘in’) never occur in the prefield. In contrast, we follow work by Lüdeling (2001), Müller
(2002a), and Zeller (2001), among others, who provide a range of examples, partly based on cor-
pus evidence, demonstrating the option to prepose the particle to the left periphery. In section 2,
we discuss the claim that topicalization of particles is governed by information structural con-
straints. In section 3, we propose a classification of particle verbs that allows us to cover the
whole transparency spectrum and to distinguish between fully transparent and fully non-
transparent particle verbs. Given this classification, in section 4, we report on a questionnaire
study that investigates whether and to what extent the option of topicalizing the particle depends
on the grade of semantic transparency of the particle verb construction. Section 5 summarizes and
concludes the paper.
2
2 Topicalization in particle verb constructions and information structure
In particle verb constructions, the topicalization of only the particle is a phenomenon that has
been extensively discussed in the literature on present-day Germanic including English (cf. Dehé
to appear). As is the case for other Germanic languages, both semantic and structural factors have
been claimed to constrain particle topicalization in German (e.g. Lüdeling 2001; Müller 2002a;
Stiebels & Wunderlich 1994; Zeller 2001). Most researchers agree that one major condition on
particle topicalization consists in the possibility of attributing a contrastive interpretation to the
particle (e.g. McIntyre 2001: 44–45; Müller 2002: 275; Zeller 2001: 93). This explains why the
sentence in (1) is grammatical, whereas the one in (2) is not.
(1) Zu hat er die Tür gemacht (und nicht auf). (Zeller 2001: 89)
close.PART has he the door made and not open
‘He closed the door.’
(2) * Auf hat Peter mit dem Trinken gehört. (Zeller 2001: 90)
PART has Peter with the drinking heard
‘Peter stopped drinking.’
While the particle topicalized in (1) may enter a relation of paradigmatic opposition with the par-
ticle auf in auf-machen (lit. ‘open-make’, to open), the particle auf in auf-hören enters in no such
paradigmatic opposition (cf. #zu-hören, #ab-hören etc.). However, contrastiveness of the particle
does not hold for cases like the following, a corpus example by Müller (2002a).1
(3) VOR hat er das jedenfalls. (Müller 2002a: 276)
PART has he that anyway
‘He intends that in any case.’
A natural account in terms of information structure would be to analyze such configurations as
‘pars-pro-toto-constructions’.2 That is, elements that do not fulfill any discourse-semantic func-
1 A tentative corpus search via http://www.ids-mannheim.de/cosmas2 confirms Müller’s (2002a)
finding that cases such as (3) exist. Interestingly, in the case of vor-haben we found that 80%
of all occurrences of left peripheral vor contained modal licensers such as allerdings (lit. ‘in-
deed’), eigentlich (lit. ‘actually’), and schon (lit. ‘already’). We will come back to this issue in
section 4.5.
2 In the literature, these constructions are referred to as cases of so-called ‘pars-pro-toto-
movement’ (Fanselow 2003). In this paper, we abstract away from theoretical issues such as
the question if and how discourse features in general – including information structural aspects
3
tion in the left periphery alone can appear in the prefield ‘pars-pro-toto’, thereby highlighting the
whole predicate. This is a very common strategy, given that the category that appears in the left
periphery of the German clause may be smaller than the focus (4) or larger than the focus (5a),
and sometimes it coincides with the focus (5b).
(4) Was hat er gemacht? (Jacobs 1991: 8)
‘What has he done?’
Ein BUCH hat er gelesen.
a book has he read
(5) a. Was hat er gelesen?
‘What did he read?’
Ein BUCH gelesen hat er.
a book read has he
b. Ein BUCH hat er gelesen.
a book has he read
At the level of information structure, preposing only a subpart of the focus (4) is equivalent to
fronting the whole focal constituent, as in (6):
(6) Was hat er gemacht?
‘What has he done?’
Ein BUCH gelesen hat er.
a book read has he
Accordingly, following Fanselow (2003), we can analyze (7a) as a pars-pro-toto-construction that
is equivalent to (7b) at the level of information structure.3
(7) a. VOR haben wir das schon gehabt. (Fanselow 2003: 35)
PART have we that well had
– should be represented in the syntax and trigger movement (for discussion, cf. Horvath 2010;
Rizzi 2014; Trotzke 2010; Trotzke & Zwart 2014). Accordingly, we use the more neutral term
‘pars-pro-toto-construction’.
3 As already indicated by example (2), the option of topicalizing the particle ‘pars-pro-toto’ is
not available in all cases involving non-contrastable particles. We will address this issue in sec-
tion 4.5.
4
b. VORgehabt haben wir das schon.
PART.had have we that well
‘We had intended that.’
The topicalization of non-contrastable elements is a regular option in German syntax. It also
shows up in phrasal idioms, like in (8a), which is equivalent to (8b); cf. similar cases in Müller
(2002b).
(8) den Löffel abgeben (‘to die’, lit. ‘the spoon pass’) (Trotzke & Zwart 2014: 138)
a. [Den LÖffel] hat er abgegeben.
the spoon has he passed
b. [Den LÖffel abgegeben] hat er.
‘He died.’
If we analyze particle verb constructions such as (7) as pars-pro-toto-constructions, then we also
make a prediction concerning the acceptability of non-adjacent vs. adjacent configurations of the
verb and the particle. In particular, Zeller (2001, 2003) claims that adjacency is strongly preferred
if the interpretation of predicate focus is the only available option due to the non-contrastability
of the particle (cf. also McIntyre 2001: 44 for similar remarks).4
(9) a. * Ab ist Nixon 1974 getreten. (Zeller 2001: 97)
PART is Nixon 1974 stepped
b. Abgetreten ist Nixon 1974.
PART.stepped is Nixon 1974
‘Nixon resigned in 1974.’
Recently, Heine, Jacobs & Külpmann (2010) objected to the claim that the particle verb receives
a focal interpretation when it occurs in the left periphery. They present a couple of examples,
4 The adjacency of a particle and a verb in the left periphery can also be achieved by topicalizing
the particle and preposing the verbal part of the particle verb into second position, as in (i), also
taken from (Zeller 2001: 97):
(i) ? Ab trat Nixon 1974.
PART stepped Nixon 1974
However, in the rest of the article, we focus on cases involving non-contrastable particles such
as (9b) where an information structural explanation in terms of predicate focus follows
straightforwardly. Accordingly, when we talk about adjacency of a non-contrastable particle
and a verb, we refer to configurations such as (9b).
5
mainly taken from Müller (2002a), and claim that in none of these cases it is very plausible that
they involve an interpretation in terms of predicate focus.
(10) Auftritt im blauen Anzug der König. (Müller 2002a: 273)
up.PART-steps in.the blue suit the king
‘The king appears in a blue suit.’
We leave aside the fact that we consider (10) as belonging to a poetic or a specific professional
register. The reason (in addition to what we said in footnote 4) why their discussion of configura-
tions such as (10) is not relevant for the purposes of our paper is that all their examples suggest
an interpretation of presentative focus in the sense of Hetzron (1975). That is, the occurrence of
the particle in the prefield (together with other reordering operations in the middle field of the
clause) yields a rhematization of the subject, thereby creating more tension (Zifonun, Hoffmann
& Strecker 1997: 1621).5
In what follows, we are only concerned with configurations such as (7) where either the
particle or the whole particle verb appears in front of the finite verb in V2 clauses. Furthermore,
focusing on these structures, our goal is not to contribute to the debate of what kind of infor-
mation structural interpretation particle topicalization exactly yields. In the literature on particle
topicalization, both topic and focus interpretations are attributed to these constructions. Therefore,
we can merely observe, in accordance with Wurmbrand (2000: 8), that “the claim that topicaliza-
tion […] has some effect on the interpretation seems to be a minimal assumption of all approach-
es” that are concerned with particle topicalization and its information structural impact. Conse-
quently, we adopt the broad notion of an information structural effect according to which focus
on an item α (in a pars-pro-toto-setting or not) indicates that alternatives to the denotation of α are
relevant for the interpretation of the utterance (Rooth 1992). This general notion underlies both
the concepts of contrastive topic and contrastive focus (Repp 2010), and thus it does not contra-
dict concrete proposals such as Zeller’s (2003) claim that the topicalization of a non-contrastable
particle such as in (7) is always interpreted as a contrastive topic and always resists a focus inter-
pretation. Since the goal of our study is to examine to what extent and how the semantic transpar-
ency of particle verbs interact with their topicalization patterns, we now turn to a classification of
5 In addition, it can be argued that these structures are actually V1-declarative structures, as Hei-
ne, Jacobs & Külpmann (2010: 41) point out themselves. However, this hypothesis is probably
unsound, given that in the string [auftrIt] it is the particle that bears primary stress (['auftrIt])
and not the verb stem (*[auf'trIt]), as in prefix verbs (e.g. Stiebels & Wunderlich 1994: 921). In
other words, the V1-explanation would only work if the particle verb could be analyzed as a
prefix verb, which is not the case in (10). Therefore, it seems that in examples like (10) the par-
ticle auf indeed occupies the prefield, while the inflected lexical verb appears in second posi-
tion.
6
the semantic transparency of particle verbs.
3 Particle verbs and semantic transparency
Apart from particle verbs in topicalization structures, it is generally observed that the discontinu-
ous appearance of a particle and its verb is strongly dispreferred if the particle semantically de-
pends on the verb to a certain extent (Hawkins 2011).6 In this context, it is very common to dis-
tinguish between two classes of particle verbs: ‘idiomatic/opaque’ and ‘literal/transparent’ con-
figurations (e.g. Chen 1986; but see Jackendoff 2002). In what follows, we want to operationalize
the notion of ‘semantic transparency’ by adopting a test proposed by Lohse, Hawkins & Wasow
(2004) that concerns the relation of dependency between the verb and the particle (i = independ-
ent; d = dependent).
(11) Particle entailment test (Lohse, Hawkins & Wasow 2004: 245)
If [X V NP Pt] entails [NP PredV Pt], then assign Pti. If not, assign Ptd.
PredV = predication verb (BE, BECOME, COME, GO, STAY)7
Given this diagnostics, we can distinguish transparent (independent) cases such as (12) from non-
transparent (dependent) cases like (13).
(12) a. die Tür zu-machen (‘to close the door’, lit. ‘the door close-make’)
b. Die Tür ist zu. (‘The door is closed.’) [+ predicative]
(13) a. etwas vor-haben (‘to intend something’, lit. ‘something before-have’)
b. * Etwas ist vor. [– predicative]
In section 2, we already saw that vor in vorhaben is also not contrastable (i.e. vor cannot be sin-
gled out from a set of alternatives). A possible formulation of this property is the following.
6 There is a systematic exception to this generalization in German syntax, though: separation of
verb and particle under V2. In this context, discontinuity seems not to affect acceptability in
any way.
7 This corresponds to Wurmbrand’s (2000) proposal to refer to a particle verb as transparent if
the particle can be a predicate in a copula construction. However, we cite the more broad ver-
sion of this test by Lohse, Hawkins & Wasow (2004), since it has been demonstrated that the
‘copula test’ is too restrictive in the context of particle verbs (McIntyre 2002: 97–98).
7
(14) Particle contrastability test
Assign a particle Prt (in a particle verb [Prt V]) the feature [+ contrast] iff Prt triggers a set of
alternatives different from the empty set.
When we take into account (14) and combine it with the test given in (11), we observe that the
semantic autonomy with predication verbs and the property of contrastability do not always co-
occur. Consider (15), where the particle has no autonomous denotation, but a contrast is neverthe-
less possible (cf. ab-nehmen vs. zu-nehmen).
(15) Nein, nicht ab muss er nehmen, sondern zu. (Müller 2002a: 265)
No not PART has.to he take but PART
‘He has to increase and not decrease in weight.’
We thus suggest that both contrastability and semantic autonomy are relevant in measuring the
semantic bond between verb and particle. By combining the two criteria of semantic transparency,
we arrive at the following four classes.
(16) [+ contrast, + predicative]
a. die Tür zu-machen vs. die Tür auf-machen
‘to close the door vs. to open the door’
b. Die Tür ist zu./Die Tür ist auf.
‘The door is closed./The door is open.’
(17) [– contrast, + predicative]
a. die Tür zu-knallen (*die Tür auf-knallen)
‘to slam the door’
b. Die Tür ist zu.
‘The door is shut.’
(18) [+ contrast, – predicative]
a. aus-ziehen vs. an-ziehen
‘to take off clothes vs. to put on clothes’
b. * Das Kleid ist/geht/wird/bleibt aus./ *Das Kleid ist/geht/wird/bleibt an.
‘The dress is/goes/becomes/stays off./ The weight is/goes/becomes/stays on.
(19) [– contrast, – predicative ]
a. etwas vor-haben (*etwas hinter-haben)
‘to intend something’
8
b. * Etwas ist vor.
‘Something is before.’
In sum, the present classification not only distinguishes between fully transparent (16) and fully
non-transparent (19) particle verbs, but also identifies intermediate classes, which capture gradual
dependencies between the verb and the particle. In the next section, we report on a questionnaire
study on the acceptability of topicalization patterns in particle verbs. Crucially, in this study we
only used either fully transparent (16) or fully non-transparent (19) particle verbs in order to gain
clear results regarding the interaction between semantic transparency and syntactic flexibility in
the context of topicalization.
4 Topicalization in particle verb constructions and acceptability
4.1 Method
For reasons discussed in section 2, we wanted to avoid referring to contexts that trigger a specific
information structural interpretation (i.e. we did not want to specify either a focal or a topical
reading). Thus, our cover story to participants was that they would be exposed to fragments of a
dialogue between two elderly ladies at a café, which we as researchers had transcribed. We also
told the participants that since the location had a lot of background noise, the transcription might
not have been correct in all passages. The participants’ job would be to judge how likely the tran-
scription was correct (see http://tinyurl.com/PV-SupplementalMaterial for complete instructions
and materials). A filler example of how the questions were presented is given below. Preceding
each example, the participants saw a random made-up time (first line of (20), in double brackets)
that indicated when in the transcribed dialogue the utterance supposedly appeared. Since time
specifications were randomized over the questionnaire, we thus reinforced the impression that
there is no contextual coherence between the items following each other in the questionnaire.
Then, we specified which of the ladies was speaking; then, the utterance followed. After each
utterance, we asked participants how likely it is that the transcription is correct, and gave them
six options in 20% intervals.
(20) ((27:01))
Müller: „Damals hatte ich ja noch die Bild-Zeitung abmoniert.“
(‘Back then, I had subscribed [abmoniert, correct form: abonniert] to the Bild-Zeitung.’)
Mit welcher Wahrscheinlichkeit haben wir das richtig transkribiert?
(‘How likely is it that we transcribed the utterance correctly?’)
☐0% ☐20% ☐40% ☐60% ☐80% ☐100%
9
This methodology allowed us to a) avoid providing explicit contrast categories, yet making parti-
cle fronting in principle felicitous, b) avoid explicit judgments about grammaticality or accepta-
bility, which are sometimes problematic (see Myers 2009 for an overview and discussion), and c)
avoid binary judgments, instead providing a range of possibilities for finer-grained distinctions. A
follow-up question confirmed that most participants did not doubt the cover story and were not
aware of the true manipulation (only five participants noted that the focus of the questionnaire
was word order variation).
4.2 Materials
Our experimental sentences were manipulated at two levels: type, that is, whether the particle
verbs were transparent (21 a, c, e) or not (21 b, d, f), and site, that is, whether the particle was in
situ (21 a, b), fronted and adjacent to the verb (21 c, d), or fronted and non-adjacent to the verb
(21 e, f). The in-situ position items served as baseline to determine whether the lexical items, in
the general context we chose, sounded plausible.
(21) a. Heute morgen hat sie den Eingang zugeschlossen.
this morning has she the entrance close.PART-locked
‘This morning, she locked the entrance.’
b. Der Bankdirektor hat das nächstes Jahr eigentlich vorgehabt.
the bank.director has that next year actually PART.had
‘The bank director actually intended to do that next year.’
c. Ausgegangen ist die Musik diesmal schon früh.
out.PART-gone is the music this.time already early
‘The music went out early this time.’
d. Runtergemacht haben ihn seine gemeinen Schulkameraden jahrelang.
PART.made have him his mean classmates for.years
‘His mean classmates bullied him for years.’
e. Aus sind die Lampen erst vorhin gegangen.
out.PART are the lamps only a.short.while.ago went
‘The lamps went out only a short while ago.’
f. Auf ist der Betrüger erst letzte Woche geflogen.
PART is the cheater only last week flown
‘The cheater’s actions have been revealed only last week.’
For each combination, there were five examples. In addition, we constructed seven fillers we ex-
pected to get ‘very likely’ judgments (‘good’ fillers), seven fillers we expected to get mixed
judgments (‘medium’ fillers), and seven fillers we expected to receive ‘very unlikely’ judgments
10
(‘bad’ fillers). Thus, there were 51 sentences in total (for a full stimuli set, see
http://tinyurl.com/PV-SupplementalMaterial). Each participant saw all items; in order to avoid
repetition effects, we varied all lexical items except for the verb between the in-situ, the adjacent
and the non-adjacent condition. The utterances were presented in random order, starting with two
fillers.
4.3 Participants
We collected judgments from 37 native German speakers, all of them students at the University
of Konstanz. One participant routinely checked more than one option in her answers, thus data
obtained from her was excluded from the analysis, resulting in 36 participants (25 female, aver-
age age: 23.8).
4.4 Results
All data were analyzed using R, specifically, the R packages lme4 and languageR (Baayen 2008;
Bates, Maechler & Bolker 2012). Filler data were analyzed using ANOVAs. Since Levene’s test
did not reach significance, we report results of Welch’s test, which does not assume equal vari-
ances. The particle verb data were analyzed by fitting a linear mixed-effects model with random
intercepts for items and participants (cf. Baayen, Davidson & Bates 2008).
4.4.1 Fillers
Figure 1 shows that the fillers were judged as we had expected: Participants thought that we had
likely transcribed the ‘bad’ fillers wrong (26.5%, SD: 34.9); judgments for ‘medium’ fillers were
at chance (51.5%, SD: 36); and ‘good’ fillers were thought to have been transcribed correctly
most of the time (91.1%, SD: 16.7). The differences were significant between filler types:
Welch’s F(2, 64.5) = 299.7, p<.0001. This data on fillers shows that our participants not only
understood the task well, but that they also used the full range of options for their judgments.
11
Figure 1: Judgments of filler items. Whiskers represent standard errors.
4.4.2 Particle verbs
The judgments for the particle verbs are summarized in Table 1. Figure 2 shows that the judg-
ments did not differ between construction types in the in-situ condition. That shows that transpar-
ent or non-transparent particle verbs were judged as being correctly transcribed equally likely
(83% vs. 84%), and that there was no inherent lexical bias for one or the other.
Position
Type
Likelihood ratings
in situ
non-transparent
84.2% (21.9)
transparent
83.3% (23.6)
adjacent
non-transparent
73.3% (24.9)
transparent
56.3% (28.6)
non-adjacent
non-transparent
28.7% (28.4)
transparent
39.7% (29.4)
Table 1: Ratings of particle verbs in percent with standard deviation (in brackets).
0
25
50
75
bad medium good
Judgment (in percent)
Filler Type
bad
medium
good
12
Figure 2: Judgments of particle verbs in situ, adjacent, and non-adjacent. Whiskers represent
standard errors.
When both the verb and the particle were fronted (adjacent), people thought it less likely that we
had transcribed the utterances correctly. They thought it particularly unlikely that we correctly
transcribed the sentences with transparent particles (56%) than with non-transparent particles
(73%). Finally, when the particle alone was fronted, participants thought it most likely that we
mistranscribed the utterance. Here, as well, we can observe a difference between the type of par-
ticle verbs: Sentences with non-transparent particle verbs received a likelihood rating of only
29%, and sentences with transparent particle verbs a rating of 40%.
The statistical analysis confirms the visual impression (see Table 2). With non-transparent
verbs, there was no effect of transparency in the in-situ condition (which served as baseline for
the regression analysis), confirming the visual impression that the particle verb type did not in-
fluence ratings in situ. However, main effects of site were significant, as were the interactions
between site and particle verb type.
in situ adjacent non−adjacent
0
25
50
75
non−transparent transparent non−transparent transparent non−transparent transparent
Judgment (in percent)
Site
in situ
adjacent
non−adjacent
13
Estimate
SE
t value
Pr(>|t|)
(Intercept)
84.22
3.68
22.88
<.001
Type (transparent vs. non-transparent)
-0.89
4.62
-0.19
>0.85
Site (in situ vs. adjacent)
-10.89
2.49
-4.37
<.001
Site (in situ vs. non-adjacent)
-55.56
2.49
-22.29
<.001
Type(transparent)*Site(adjacent)
-16.11
3.53
-4.57
<.001
Type(transparent)*Site(non-adjacent)
11.89
3.53
3.37
<.001
Table 2: Regression results for critical items.
One additional result of our study that is potentially interesting is that in case of non-transparent
particle verbs occurring non-adjacently, the judgments vary considerably compared to the varia-
tion we see in the other five type and position combinations. In particular, looking at the mean
judgments of the individual non-transparent particle verbs in non-adjacent position, we see that
nach-geben (‘to give way under pressure’), auf-machen (‘to head off’), and auf-fliegen (‘to leak
out’) were all judged at or below 20%. In contrast, the two items vor-haben (‘to intend’) and
runter-machen (‘to bully sb./sth.’) both received a rating of above 40% (see Figure 3).
14
Figure 3: Mean likelihood ratings on non-transparent, non-adjacent particle verbs. Whiskers represent
standard errors.
4.5 Discussion
Our study supports the general observation that topicalization in German particle verbs seems to
be more expected, and less marked, when the particle and the verb occur adjacently in the pre-
field. Since we distinguished between transparent and non-transparent particle verb constructions,
we arrive at a more detailed picture, however. In particular, the difference between non-
transparent and transparent particle verbs regarding the adjacent and non-adjacent occurrence of
0
10
20
30
40
50
auffliegen aufmachen nachgeben runtermachen vorhaben
Judgment (in percent)
non−transparent verb,
in non adjacent position
auffliegen
aufmachen
nachgeben
runtermachen
vorhaben
15
the particle clearly shows that information structural constraints govern topicalization patterns in
particle verbs.
In section 2, we saw that topicalization of particles of non-transparent particle verb con-
structions can only be analyzed as pars-pro-toto-constructions at the level of information struc-
ture. On the other hand, the preposing of particles of transparent particle verb constructions can
also be interpreted as contrasting only the denotation of the particle instead of the whole predicate.
We thus predict that adjacency is strongly preferred in the non-transparent cases, since the inter-
pretation of predicate focus is the only available option due to the non-contrastability of the parti-
cle. In the transparent cases, on the other hand, we predict that the preference for adjacency is not
as strong as in the non-transparent cases. Our results confirm this reasoning: the transparent cases
are judged as being less likely in the adjacent condition. The non-transparent cases, on the other
hand, received a higher rating in terms of likelihood in the adjacent condition because predicate
focus is the only available interpretive option. Apart from this difference, in both conditions,
transparent and non-transparent, the adjacent occurrence is preferred to the non-adjacent occur-
rence. The fact that adjacency is strongly preferred even in the transparent cases is an interesting
data point, and worth investigating further. We hypothesize that the narrow focus interpretation
of contrasting only the particle is harder to get in a setting where no specific context triggering
this interpretation is provided, as was the case in our study. In sum, the results of our study con-
firm that information structural constraints in combination with the degree of semantic transpar-
ency govern topicalization patterns in particle verbs. However, our findings go beyond hypothe-
ses formulated in the literature (e.g. by Zeller 2001). More specifically, in the literature, we find
no explicit predictions with respect to differences in acceptability of transparent vs. non-
transparent particle verb constructions when the verb and the particle occur adjacently in the pre-
field. In fact, the literature suggests that both classes behave alike in the adjacent configuration,
differently from what our findings show.
Let us now turn to the variation in judgments of the non-adjacent cases of non-transparent
particle verb constructions (see Figure 3). No such effect is predicted by the information structur-
al hypotheses that we discussed in section 2. That is, all non-transparent particles should behave
alike. As Figure 3 shows, however, in the cases of runtermachen (‘to bully sb./sth.’) and vorha-
ben (‘to intend’), participants thought it more likely that we had transcribed the utterances cor-
rectly than in the cases of the other three items. Let us point out right away that we do not have a
satisfying explanation for this finding. An obvious account in the case of runtermachen would be
to follow Stiebels & Wunderlich (1994) who argue that topicalization of resultative or directional
particles (like runter) clearly improves the acceptability of such configurations (cf. also
Webelhuth & Ackerman 1999: 44–51 for related observations). We hypothesize, however, that
this is, at least, not the whole story. Note that there are differences like the following between
particles with a clear directional semantics:
16
(22) a. (Regelrecht) Raus ist er geflogen!
downright out.PRT is he flown
‘He got kicked out yesterday.’
b. # (Regelrecht) Raus hat die Band ihr neues Album gebracht!
downright out.PRT has the band their new album brought
‘The band published their new album.’8
While rausfliegen entails that one has been dismissed in a harsh way, rausbringen does not refer
to any such intensity scale: either the band published or published not. The option of topicalizing
the particle here seems to depend on the lexical aspect of the verb and its aspectual composition
with degrees (e.g. Caudal & Nicolas 2005). This is also indicated by using the degree modifier
regelrecht in (22), thereby applying a common diagnostics to identify different types of scales in
an entry of a lexical item (Kennedy & McNally 2005).
However, this explanation, even if tenable, does not hold for vorhaben. In this case, we can
only speculate on the basis of what we found in a tentative corpus search via http://www.ids-
mannheim.de/cosmas2. In particular, in the case of vorhaben, we found that 80% of all occur-
rences of left peripheral vor contained modal licensers such as allerdings (lit. ‘indeed’), eigent-
lich (lit. ‘actually’), and schon (lit. ‘already’). Note that in our materials (see
http://tinyurl.com/PV-SupplementalMaterial), we also used the discourse particle schon. All these
particles reinforce a concessive reading related to utterances with left peripheral vor in the con-
struction with vorhaben, which could be paraphrased as ‘I really intended that, but then I did not
put this into practice.’ The interaction between the presence of such modal elements and the pos-
sibility of the occurrence of non-contrastable vor in the prefield could be explained by the speak-
er’s intention to add an extra touch of emphasis to his utterance, highlighting the sincerity of the
intention (for a suitable notion of the speaker's emphasis in this context, cf. Bayer & Trotzke to
appear; Trotzke & Turco subm.).
8 In these cases, the particle verbs are not fully non-transparent according to our classification in
section 3. In particular, one can replace the directional particle (he)raus with a stative prefix
(and the required suffix), resulting in a predicative construction such as Er ist draußen (‘He is
out’, intended: ‘He is out of his job’). However, this is not the case in the even more metaphor-
ical case runtermachen (for the complex metaphorical meaning involved, cf. McIntyre 2001:
156).
17
5 Conclusion
In this paper, we proposed a classification of German particle verbs that allows us to cover the
whole transparency spectrum and to distinguish between fully transparent and fully non-
transparent particle verb constructions. Based on this classification, we reported on a question-
naire study that provides empirical evidence for the claim that information structural constraints
in combination with the degree of semantic transparency govern topicalization patterns in particle
verbs. Moreover, we pointed to potential additional constraints on topicalization in particle verbs
that go beyond information structure.
Our paper provides a first step towards an empirical foundation of an aspect of particle verb
behavior that is hallmarked by disagreement in the literature because most studies are based on
varying introspective judgments of the authors (for discussion in this context, cf. Meurers &
Müller 2009; Müller 2007). By providing such evidence, our paper may prove useful not only for
theoretical discussion, but also for a growing literature in psycholinguistics, where the processing
and representation of both semantically transparent and non-transparent particle verbs has been
gaining attention (Konopka & Bock 2009; Piai et al. 2013; Smolka, Preller & Eulitz 2014). In
addition, we have determined that topicalization options and semantic transparency are closely
linked, which might not only advance the discussion limited to particle fronting, but might also
prove useful for the analysis and psycholinguistic investigations of other constructions, such as
verbal idioms (Rommers, Dijkstra & Bastiaansen 2013) or light verb constructions (Wittenberg et
al. 2014; Wittenberg & Snedeker 2014), where the notion of semantic transparency is an interest-
ing factor in processing.
18
References
Baayen, R. Harald (2008): Analyzing linguistic data: A practical introduction to statistics using R.
Cambridge: Cambridge University Press.
—, Doug Davidson & Douglas Bates (2008): Mixed-effects modeling with crossed random
effects for subjects and items. In: Journal of Memory and Language 59, 390–412.
Bates, Douglas, Martin Maechler & Ben Bolker (2012): lme4: Linear mixed-effects models using
S4 classes. Retrieved from http://CRAN.R-project.org/package=lme4 (R package version
0.999999-0).
Bayer, Josef & Andreas Trotzke (to appear): The derivation and interpretation of left peripheral
particles. In: Josef Bayer, Roland Hinterhölzl & Andreas Trotzke (eds.), Discourse-
oriented syntax. Amsterdam: Benjamins.
Caudal, Patrick & David Nicolas (2005): Types of degrees and types of event structures. In:
Claudia Maienborn & Angelika Wöllstein (eds.), Event arguments: Foundations and
applications. Berlin: Mouton de Gruyter, 277–300.
Chen, Ping (1986): Discourse and particle movement in English. In: Studies in Language 10, 79–
95.
Dehé, Nicole (to appear): Particle verbs in Germanic. In: Peter O. Müller et al. (eds.), Word
formation: An international handbook of the languages of Europe. Berlin: Mouton.
Eisenberg, Peter (1999): Grundriss der deutschen Grammatik: Der Satz. Stuttgart: Metzler.
Fanselow, Gisbert (2003): Surprising specifiers and cyclic spellout. In: Generative Linguistics in
Poland 5, 29–46.
Fuhrhop, Nanna (2007): Zwischen Wort und Syntagma: Zur grammatischen Fundierung der
Getrennt-/Zusammenschreibung. Tübingen: Niemeyer.
Hawkins, John A. (2011): Discontinuous dependencies in corpus selections: Particle verbs and
their relevance for current issues in language processing. In: Emily M. Bender & Jennifer
E. Arnold (eds.), Language from a cognitive perspective: Grammar, usage, and
processing. Stanford: CSLI Publications, 243–265.
Heine, Antje, Joachim Jacobs & Robert Külpmann (2010): Quer zu den Feldern: Zur Topologie
von Partikelverben. In: Linguistische Berichte 221, 37–60.
Hetzron, Robert (1975): The presentative movement or why the ideal word order is V.S.O.P. In:
Charles N. Li & Sandra Thompson (eds.), Word order and word order change. Austin:
University of Texas Press, 346–388.
Horvath, Julia (2010): 'Discourse features', syntactic displacement and the status of contrast. In:
Lingua 120, 1346–1369.
Jackendoff, Ray (2002): English particle constructions, the lexicon, and the autonomy of syntax.
In: Nicole Dehé et al. (eds.), Verb-particle explorations. Berlin: Mouton de Gruyter, 67–
94.
Jacobs, Joachim (1991): Focus ambiguities. In: Journal of Semantics 8, 1–36.
Kennedy, Christopher & Louise Mcnally (2005): Scale structure, degree modification, and the
semantics of gradable predicates. In: Language 81, 345–381.
Konopka, Agnieszka E. & Kathryn Bock (2009): Lexical or syntactic control of sentence
formulation? Structural generalizations from idiom production. In: Cognitive Psychology
58, 68–101.
Lohse, Barbara, John A. Hawkins & Thomas Wasow (2004): Domain minimization in English
verb-particle constructions. In: Language 80, 238–261.
19
Lüdeling, Anke (2001): On particle verbs and similar constructions in German. Stanford: CSLI
Publications.
Mcintyre, Andrew (2001): German double particles as preverbs: Morphology and conceptual
semantics. Tübingen: Stauffenburg.
— (2002): Idiosyncrasy in particle verbs. In: Nicole Dehé et al. (eds.), Verb-particle explorations.
Berlin: Mouton de Gruyter, 95–118.
Meurers, W. Detmar & Stefan Müller (2009): Corpora and syntax. In: Anke Lüdeling & Merja
Kytö (eds.), Corpus linguistics: An international handbook (vol. 2). Berlin: Mouton de
Gruyter, 920–933.
Müller, Stefan (2002a): Complex predicates: Verbal complexes, resultative constructions and
particle verbs in German. Stanford: CSLI.
— (2002b): Syntax or morphology: German particle verbs revisited. In: Nicole Dehé et al. (eds.),
Verb-particle explorations. Berlin: Mouton de Gruyter, 119–140.
— (2007): Qualitative Korpusanalyse für die Grammatiktheorie: Introspektion vs. Korpus. In:
Gisela Zifonun & Werner Kallmeyer (eds.), Sprachkorpora: Datenmengen und
Erkenntnisfortschritt. Berlin: Mouton de Gruyter, 70–90.
Myers, James (2009): Syntactic judgment experiments. In: Language and Linguistics Compass 3,
406–423.
Piai, Vitória et al. (2013): Sit down and read on: Working memory and long-term memory in
particle-verb processing. In: Brain and Language 127, 296–306.
Repp, Sophie (2010): Defining 'contrast' as an information-structural notion in grammar. In:
Lingua 120, 1333–1345.
Rizzi, Luigi (2014): Syntactic cartography and the syntacticisation of scope-discourse semantics.
In: Anne Reboul (ed.), Mind, values, and metaphysics. Dordrecht: Springer, 517–533.
Rommers, Joost, Ton Dijkstra & Marcel C. M. Bastiaansen (2013): Context-dependent semantic
processing in the human brain: Evidence from idiom comprehension. In: Journal of
Cognitive Neuroscience 25, 762–776.
Rooth, Mats (1992): A theory of focus interpretation. In: Natural Language Semantics 1, 75–116.
Smolka, Eva, Katrin H. Preller & Carsten Eulitz (2014): 'Verstehen' ('understand') primes 'stehen'
('stand'): Morphological structure overrides semantic compositionality in the lexical
representation of German complex verbs. In: Journal of Memory and Language 72, 16–36.
Stiebels, Barbara & Dieter Wunderlich (1994): Morphology feeds syntax: The case of particle
verbs. In: Linguistics 32, 913–968.
Trotzke, Andreas (2010): Rethinking syntactocentrism: Lessons from recent generative
approaches to pragmatic properties of left-periphery-movement in German. PhD Thesis,
University of Freiburg.
— & Giuseppina Turco (subm.): The grammatical reflexes of emphasis: New evidence from
German wh-questions. Submitted for publication.
— & Jan-Wouter Zwart (2014): The complexity of narrow syntax: Minimalism, representational
economy, and simplest Merge. In: Frederick J. Newmeyer & Laurel B. Preston (eds.),
Measuring grammatical complexity. Oxford: Oxford University Press, 128–147.
Webelhuth, Gert & Farrell Ackerman (1999): A lexical-functional analysis of predicate
topicalization in German. In: American Journal of Germanic Linguistics and Literature 1,
1–61.
20
Wittenberg, Eva et al. (2014): The difference between "giving a rose" and "giving a kiss":
Sustained neural activity to the light verb construction. In: Journal of Memory and
Language 73C, 31–42.
— & Jesse Snedeker (2014): It takes two to kiss, but does it take three to give a kiss?
Categorization based on thematic roles. In: Language, Cognition and Neuroscience 29,
635–641.
Wurmbrand, Susi (2000): The structure(s) of particle verbs. Ms., McGill University.
Zeller, Jochen (2001): Particle verbs and local domains. Amsterdam: Benjamins.
— (2003): Moved preverbs in German: Displaced or misplaced? In: Geert Booij & Ans van
Kemenade (eds.), The yearbook of morphology. Dordrecht: Kluwer, 179–212.
Zifonun, Gisela, Ludger Hoffmann & Bruno Strecker (1997): Grammatik der deutschen Sprache.
3 Bände. Berlin: Mouton de Gruyter.
