L�deling, Anke (2001) On Particle Verbs and Similar
Constructions in German. Stanford: CSLI. (http://csli-
Reviewer: Andrew McIntyre, University of Leipzig.
Constructions like _work out, hang around, eat up_, and
their counterparts in other Germanic languages, known as
'phrasal verbs', 'separable verbs', 'verb-particle
combinations' and, in German-speaking circles, as 'particle
verbs', have inspired much literature, recent large-scale
studies being Deh� 2001, Deh� et al. 2001, den Dikken 1995,
McIntyre 1998, Olsen 1998, Stiebels 1996, Toivonen 2001,
Zeller 1999. It is unclear how we should describe the
syntactic, morphological, semantic and argument structural
properties of particle verbs (hereafter: pv's), which is a
pity, because a correct approach would teach us much about
the structure of the verb phrase and about the interfaces
between semantics, syntax and morphology.
Anke L�deling (hereafter: L) presents a well written
study of the German constructions. Her perspective is that
'there are no particle verbs' (p. ix). The objects usually
bearing that name have neither a unique nor a uniform
syntactic structure. Some are seen as V' structures
(analogous to one common treatment of resultative
constructions), while, in a novel move, L treats other
particles (along with object depictive predicates and
certain adverbials) as adjuncts to V'. L uses the term
'particle (verb)' as shorthand for 'an object usually
called a 'particle (verb)', and I will do so too.
CHAPTER 1 introduces the basic problems studied by L, viz.
the problem of the structure of pv's and that of how and
whether pv's should be distinguished from other
constructions. The latter aim involves clarifying the
nature of 'particles'. While other authors posit a
definition for 'particle', L argues that there is no
natural class corresponding to the term (Section 1.2.2).
It is argued that pv-like structures with nominal and
verbal nonheads are distinct from pv's by virtue of e.g.
their apparent unproductivity and inability to be input for
derivational processes. This is at variance with
the rest of the study, which downplays the importance of
these phenomena as criteria for distinguishing pv's from
other structures, but the inconsistency is not
detrimental to L's general analysis.
L (p. 15-18) dismisses the idea that particles are
intransitive prepositions, but I query some of the
arguments (without wishing to claim that all particles are
prepositional). L sees the intransitive preposition view of
particles as problematic because intransitive prepositions
'by definition only have one argument and thus cannot
introduce arguments in addition to the arguments of the
verb' (p.17). I disagree: an 'intransitive' preposition
is one which does not case-mark a complement, and this is
compatible with cases where it contributes an external argument
(=theme, trajector, figure), as in _vote a government in_
(cf. *_vote a government_), or even an internal argument
(=reference object, landmark, ground), as in _run someone
over, pour the bucket out_.
L (p.17) also argues that many core particles (e.g. the
German cognates of 'in, out, off, up, on') were 'originally
all adverbs and have changed their category over time' and
that their semantics (e.g. their assumed ability to act as
functors over the verb) might retain this adverbial
character. Here L. seems to be using 'adverb' in the
traditional sense, which lumps together adjective-related
elements like _schnell_ 'quick(ly)' with complementless
directional/locational elements, even though the latter are
for the most part formally related to complement-taking
prepositions and share many of their distributional
properties. One could query the usefulness of this
taxonomy. The traditional use of 'adverb' smacks strongly
of being a repository for unclassified elements, and cannot
legitimately be appealed to as a primitive category without
discussion. Anyway, the 'adverbs' which were the ancestors
of today's particles would still be called 'intransitive
prepositions' by most people using that label. We miss
generalizations if we deny the prepositional character of
the core particles in German. For instance, they pattern
with prepositional phrases in freely forming unaccusative
directional constructions (_einsteigen_ 'get in' vs. _in
den Zug steigen_ 'get in the train'), while, as L, p.148,
fn 114 notes, German does not easily form unaccusative
resultatives with AP predicates: German cannot literally
render _the toast burnt black_ or _the butter melted soft_.
While I queried L's arguments against the existence of a
natural class corresponding to any normal use of the term
'particle', familiarity with the full inventory of
particles in German (including some exotica mentioned at
the end of this review) leads to the conclusion that no
satisfactory definition and classification of German
particles has been arrived at, and that it is reasonable to
assume that 'particles' are not a grammatical primitive,
but rather a set of entities behaving uniformly with
respect to some, but not all, of a set of more primitive
parameters, such as their category, their ability to
project phrases, their position within VP, their ability to
predicate over a NP, etc.
CHAPTER 2 examines syntactic, semantic and phonological
properties of German pv's, highlighting problems with
the morphological view of pv formation. Some supposedly
'morphological' effects of particles (idiomatic semantics,
fiddling with verbal selection restrictions) are found with
structures unilaterally treated as phrasal structures (e.g.
resultatives). Being separable, pv's have no place in any
normal lexicalist view of morphology. Particle
topicalization and modification are attestable and cases
where they are blocked do not bespeak deference to lexical
integrity but a lack of semantic independence of the
particle. L's arguments here are echoed in several other
studies (e.g. Kayne 1984:125, Booij 1990, McIntyre 1998,
Zeller 1999). Some may feel that arguing at length against
the traditional arguments for the morphological view of pv
formation is belaboring the obvious, but I support L on
this score because some linguists generating pv's in the
morphology have ignored the counterarguments in the
literature, failing to realize that they have the burden of
proof and must either (a) find new evidence for their
position (Olsen 2000, McIntyre 1998 attempted this) or (b)
supplement the traditional arguments with an expanded view
of morphology which can duplicate the same surface
structures as syntax (e.g. resultative constructions and
idioms like the capitalized material in _she COULD have
clearly DONE WITH some help_ or_THE CHIPS ARE DOWN_).
Option (b) is implausible, but is a logical consequence of
using some traditional arguments for the morphological view
Of course, it is possible that pv's are exceptional
morphological objects or that the normal signs of lexical
integrity, notably inseparability, have been overrated. But
whether pv's give sufficient evidence to warrant exploring
such possibilities is an open question.
CHAPTER 3 discusses another standard argument for the view
that pv's are morphological objects, based on the fact that
German pv's readily act as input to derivational
morphology, coupled with the No Phrase Constraint, which
bans phrases as input to morphology. L. adduces
empirical evidence against the No Phrase Constraint, and
argues that cases where an affix attaches to pv's but not
to resultative constructions do not reflect a structural
difference. Section 3.4 explores various German affixes and
their compatibility with pv's and resultative
constructions. 3.4.1 gives some convincing pragmatic
constraints on the use of agentive suffix _er_ , which
explain cases like the _anstreicher_ 'painter' (based on a
pv) vs. *_rotstreicher_ 'person who paints sth. red'. The
circumfix _Ge...e_ (p. 103ff) is discussed, but no
explanation is given for why it does not hapilly attach to
Cases where the affixes _-ung_ 'event nominaliser', _-bar_
'-able' and _un-_ 'un-' attach to pv's but not resultative
constructions are argued to be sensitive not to putative
structural differences between the constructions, but to
whether or not the affix' input is lexically listed. This
is potentially significant, but I use the hedge advisedly,
since L discusses only a very small data sample, and
readers cannot assess the accuracy of L's claim without
collecting data themselves. Assuming the proposal is
correct, it is interesting and one would have welcomed more
discussion of the questions it raises. I mention two.
Firstly, it is sometimes claimed that the listing of
fully regular constructions is possible if they are
frequent (e.g. Jackendoff 1997:122f). Perhaps L can appeal
to this to explain certain cases where apparently fully
compositional pv's can be suffixed with some of L's
putative listing-sensitive affixes (examples attested by
internet search are _Heraustrennung_ 'separating out',
_herausnehmbar_ 'take-out-able', _zusammenklappbar_ 'fold-
up-able'). If further work confirms L's hypothesis,
then the affixes involved would be a useful
test for cases where we are not sure if a compositional
structure is listed or not, although getting round the
potential circularity would need careful argumentation.
Secondly, why are there affixes which attach to complex
verbs only if they are listed? We know that affixes often
stipulate conditions on the semantics, phonology, category
or morphological structure of their bases. Is the
listedness of the base another requirement which affixes
can just stipulate? Or is there some deeper principle
behind L's observation? We might consider asking whether
the putatively listing-sensitive processes are constrained e.g.
in the following way:
(a) they operate presyntactically
(b) they can attach to a non-minimal verbal projection
This predicts that the affixes can only attach to phrasal
V-projections already existing when the affixation applies,
precisely those which are listed. (Affixes like _-er_ allowing
unlisted input would lack requirement (a).)
Of course, (a) and (b) are, as they stand, stipulations from hell,
but if a case can be made
for their empirical necessity elsewhere, then we would be
closer to a more explanatory account of L's observations.
Whether or not this is a good approach, it seems worthwhile
to ask whether the listedness requirement of the affixes
under consideration can be made to follow from something
else. Attempts at answering this question might benefit
from asking whether there are affixes which require complex
bases which are NOT listed. If there are no such affixes,
we must explain why words and listed phrases behave like a
natural class as far as some (but not all) affixes are
concerned. If there are such affixes, then listedness
sensitivity is probably just stipulated by the affixes. (I
do not know whether there are such affixes. The nearest
thing to a relevant case I can think of is the suggestion
of Raffelsiefen 1992:157-160 that compositional/ productive
_un-_ prefixation of English adjectives is possible only
when the bases are productively derived adjectives
(_unfriendly_ but *_unnice_), but this is not the same
thing as requiring unlisted input.)
In sum, Chapter 3 presents a potentially significant
proposal (perhaps the most significant in the book) which
needs further empirical support and theoretical
clarification, aims which the author is pursuing, cf.
L�deling/de Jong 2001). I close discussion of the chapter
by noting a less central issue. The analysis of _un-_
prefixation (p. 109-113) either has interesting
consequences or is problematic. The acceptability
difference in (2a-b) is claimed to follow from the
listedness of the pv 'abschick-' and the nonlistedness of
'wegschick-'. However, even if we agree that these pv's
differ in terms of listedness, there is the additional
complication that _un-_ has to attach to an adjectival
participle formed from the pv, and the listedness of
_abschick-_ does not automatically entail the listing of
the adjectival participle formed from it. (Note that both
pv's can form adjectival participles (cf. (2c)), although
informants find _abgeschickt_ better than _weggeschickt_ in
(2) a. der Brief ist unabgeschickt 'the letter is un-sent-
b. *der Brief ist unweggeschickt 'the letter is un-
c. der Brief ist abgeschickt/weggeschickt 'the letter
is sent off/away'
To solve this puzzle convincingly, L must do either of the
(a) show that the adjectival participle of the pv
_abschick-_ 'send off' is listed. (I see no evidence for
(b) assess the implications of assuming that _un-_
prefixation is constrained such that, if it attaches to an
adjective zero-derived from a past participle which itself
is derived from a complex verb, the complex verb must be
listed. It might be objected that this flies in the face of
generalisations to the effect that morphological processes
only refer to information present in the immediately
preceding derivational step (e.g. the Adjacency Condition,
Bracketing Erasure), for L's analysis of (2b) suggests that
_un-_ prefixation is sensitive to the listing of a
structure which has undergone two subsequent morphological
operations whose output is not listed. Even though the
Adjacency Condition has rightly been criticised
by various authors (e.g. Raffelsiefen 1992), one would have
welcomed some discussion of the issue. Here L is too
modest: she underestimates the potential significance of
her own suggestion. Apart from its consequences for the
question as to what types of information morphological
processes can refer to, perhaps L could have commented on
the implications of her analysis for Kratzer's quite
influential (1994) theory, which bases certain arguments on
the contrast in (2).
CHAPTER 4 surveys some earlier proposals on pv structure.
CHAPTER 5 presents L's syntactic analyses for pv's and
related constructions. Some pv's (let us call them
'resultative pv's') are given the same structure as
resultative constructions, and are analyzed as V'
structures, while in other cases, the particles are seen as
V' adjuncts (like object depictive predicates and certain
adverbials). Various proposals exist for differing
structures for pv's (Kratzer 1994, Aarts 1989, Wurmbrand
2000, Harley/Noyer 1997), though none resembles L's
proposal for V' adjunction for some particles. The idea
that (most) pv's are structurally identical to resultative
constructions has many precedents in the literature (e.g.
Hoekstra 1988, Haider 1997, Winkler 1994:388), and L's
choice of a V' structure for these is a standard strategy
if one rejects the small clause approach and the
morphological one (cf. Haider, Winkler, Zeller 1999). I
have little to say on this part of the analysis, except
that L (p.148-55) provides some convincing remarks
explaining certain syntactic differences between the pv's
and resultatives for which the uniform V' analysis is
Very few instances of the putative type of particle
adjoined to V' are discussed (p. 156 exhausts the data
sample), so it is hard to know which particles are V'
adverbials in L's scheme of things. Most are so-called
'aspectual' particle uses. I will try to show that
aspectual particles are bad candidates for such an
analysis, although there are particles for which L's
adjunction analysis might be valid.
One particle treated as an adjunct is a use of _an_ in
forming verbs expressing partial affectedness like those in
(3) _andiskutieren_ 'discuss partly', _anlesen_ 'read
L (p. 144, 156) assumes, I think rightly, that adjuncts do
not change the argument structure of the verb. However,
pv's exhibiting the use of _an_ seen in (3) are always
obligatorilly transitive (Stiebels 1996:78f, Zeller 1997,
McIntyre 2001a:152ff), even if the simplex verb is
optionally transitive (as are those in (3)). Moreover, the
particle can introduce objects of types not allowed by the
(4a) _einen Plan andenken_ 'think about a plan in a
(4b) _*einen Plan denken_
Facts like these lead Zeller 1997 and McIntyre 2001a to
argue that this use of _an_ is not (as L assumes) an
inchoative marker behaving like a functor over the verb,
but a secondary predicate signalling partial affectedness
of the direct object. If so, the intuitition that _an_
conveys the beginning of an event is epiphenomenal. This
initially surprising assumption gains plausibility when we
consider another German particle also traditionally
analysed as an inchoative marker, _los_. Unlike _an_, it
forbids direct objects:
(4'a) _ich sang Lieder_ 'I sang songs'
(4'b) _ich sang *(Lieder) AN__ 'I sang parts of songs'
(4'c) _ich sang (*Lieder) LOS_ 'I started singing (songs)'
There are at least a dozen Germanic particle uses which
likewise block direct objects (e.g. _fight (*one's enemies)
on_, _hammer (*the metal) around_, _sing (*jazz) along_),
cf. McIntyre (2001a,b) and Stiebels (1996:64f) for attempts
at an explanation. These are the particles where L's
functor analysis is arguably most apposite, but they affect
the argument structure radically. The other group of
'aspectual' particles includes _an_ and all 'perfective'
particle uses. These convey information about the lexical
aspect of an event indirectly by specifying how the event
affects an incremental theme, which appears as a direct
object (often obligatory, often flouting verbal selection
restrictions). In either case, aspectual particles ride
roughshod over a verb's argument structure in a way that
adverbials and depictive predicates do not. This undermines
one of L's reasons for assuming the adjunction analysis in
the first place.
L must also explain some syntactic differences between
depictives/adverbials and the particles treated as V'
adjuncts. One is that the putative adjoined particles
behave like resultative particles and differently from
depictives in insisting on verb-adjacency in verb cluster
formation in standard German:
(5a) weil sie das Fleisch wird [KALT/AUF]ESSEN wollen
because she the meat will [COLD/UP]EAT want
'because she will want to eat the meat cold/up'
(5b) weil sie das Fleisch [KALT/*AUF] wird ESSEN wollen
because she the meat [cold/up]will EAT want
Further evidence bearing on the particles-qua-adjuncts
hypothesis might come from prosody (cf. Winkler 1994 on
depictives vs. resultatives), but the prosodic facts
concerning particles and other secondary predicates are too
subtle to judge without pitch extraction data.
There are some particles which are more likely to be
adjuncts than are aspectual particles. I mention them,
along with some problems they raise for L's theory (and
every other pv theory I know):
1. _vor_ and _nach_ in the readings 'beforehand' and
'afterwards, re-'. These are always called particles and
preserve the verb's argument structure. One might want to
treat them as adjuncts on a par with the synonymous adverbs
_vorher_ 'beforehand' and _nachher_ 'afterwards'. But pairs
like _vorbestellen_/_vorher bestellen_ 'order beforehand'
do not show a uniform syntactic behaviour e.g. in tests
parallel to (5).
2. _wieder_ 'again' behaves syntactically and
prosodically like a resultative particle when it has narrow
scope, except that it can co-occur with other particles
(_wieder aufbauen_ 'reconstruct'). Stechow (1996) adjoins
this use of _wieder_ to a small clause.
3. Some uses of _mit_ 'with'. It can form idioms with
verbs (_mitteilen_ "with-divide", 'inform', _mitkriegen_
"with-get", 'hear about') which pattern like resultative
pv's with respect to stress and syntactic behaviour,
although there is no hint of a resultative meaning. If
_mit_ relates to the verb compositionally, it can behave
either like _mit_ in idiomatic combinations, or else like a
defocussed adverbial and may be separated from the verb in
contexts where particles cannot:
(6a) MIT im Team SPIELEN
WITH in.the team PLAY
'play with the others in the team'
(6b) im Team MITSPIELEN
in.the team WITH-PLAY
1. 'play with the others in the team'
2. 'cooperate in the team (not necessarily a sporting team)'
Since none of these, except _wieder_, has received a
theoretically informed and empirically representative
treatment, discussing them would have been worthwhile, if
only to expand the book's empirical focus, which
concentrates on types of pv's which have been discussed
elsewhere (e.g. Stiebels 1996, Zeller 1999). Introducing
new and difficult data into theoretical arenas can only be
Aarts, Bas (1989), "Verb-preposition constructions and
small clauses in English", in: Journal of Linguistics
Booij, G., 1990. The boundary between morphology and
syntax. Yearbook of Morphology 3, 45-63.
Deh�, N, Jackendoff, R., McIntyre, A., and Urban S., (eds.)
2001. Verb particle explorations. Mouton de Gruyter. (to
Deh�, N. 2001. Syntactic, phonological and information
structural aspects of English particle verbs. Diss.
den Dikken, M., 1995. Particles. Oxford University Press.
Haider, H. 1997: "Precedence among predicates", Journal of
Comparative Germanic Linguistics 1, 3-41.
Harley, H., & Noyer, R., 1997. Mixed nominalisations, short
verb movement and object shift. NELS 28.
Hoekstra, T, 1988. Small clause results. Lingua 74:101-39.
Kayne, R. 1984, "Principles of Particle Constructions", in:
Gu�ron, Jacqueline, H.-G. Obenauer (eds.), Grammatical
Representations, Dordrecht: Foris, S. 101-140.
Kratzer, A., 1994. The event argument and the semantics of
voice. Ms. Amherst.
L�deling, A. and de Jong, N. 2001. Particle verbs and word
formation. To appear in Deh� et al. (eds.)
McIntyre, A. 1998: German double particles as preverbs.
Diss. Sydney. To appear with Stauffenburg (T�bingen).
McIntyre, A, 2001a: "Argument blockages induced by verb
particles". In Nicole Deh� and Anja Wanner (eds.)
Structural Aspects of Semantically Complex Verbs. Berlin:
Peter Lang. 131-164.
McIntyre, A. 2001b. Preverbs, argument linking and verb
semantics. Ms Leipzig
Neeleman, A., & Weerman, F. 1993. The balance between
syntax and morphology. Natural Language and Linguistic
Theory, 11, 433-475.
Olsen, S., 1998 (ed.) Semantische und konzeptuelle Aspekte
der Partikelverbbildung mit ein-. T�bingen: Stauffenburg.
Olsen, Susan. 2000. "Against Incorporation". In:
Linguistische Arbeitsberichte 74.
Raffelsiefen, R., 1992. A Nonconfigurational Approach to
Morphology. In: Aronoff, M., (ed.) Morphology Now. Albany:
State University of New York Press.
Stechow, A. von (1996). The different readings of wieder
"again". Journal of Semantics, 13, 87-138.
Stiebels, B., 1996. Lexikalische Argumente und Adjunkte.
Toivonen, I., 2001. The phrase structure of non-projecting
words. Dissertation: Stanford.
Winkler, S, 1994. Secondary predication in English. Diss.
T�bingen. (reworked as Focus and Secondary Predication.
Berlin/ NY: de Gruyter, 1997)
Wurmbrand, S, 2000: "The structure(s) of particle verbs".
Zeller, J., 1997. How syntax restricts the lexicon:
particles as thematic predicates. Ms. Frankfurt. To appear
in Linguistische Berichte.
Zeller, J. 1999: Particle verbs, local domains, and a
theory of lexical licensing. Dissertation, Frankfurt. (to
appear as Particle Verbs and Local Domains. Amsterdam: John
Andrew McIntyre has a postdoctoral research and teaching
position in the English department at the University of
Leipzig, Germany. He has studied complex verb formation in
German and English, and is interested in the semantics and
argument structures of verbs.