Download Pseudo-incorporation in Dutch Geert Booij

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Germanic weak verb wikipedia , lookup

English clause syntax wikipedia , lookup

Arabic grammar wikipedia , lookup

Udmurt grammar wikipedia , lookup

Germanic strong verb wikipedia , lookup

Ojibwe grammar wikipedia , lookup

Macedonian grammar wikipedia , lookup

Ukrainian grammar wikipedia , lookup

Modern Greek grammar wikipedia , lookup

Zulu grammar wikipedia , lookup

Portuguese grammar wikipedia , lookup

Old Norse morphology wikipedia , lookup

Kannada grammar wikipedia , lookup

Chinese grammar wikipedia , lookup

Japanese grammar wikipedia , lookup

Inflection wikipedia , lookup

Navajo grammar wikipedia , lookup

Old Irish grammar wikipedia , lookup

Malay grammar wikipedia , lookup

Spanish grammar wikipedia , lookup

Swedish grammar wikipedia , lookup

Georgian grammar wikipedia , lookup

French grammar wikipedia , lookup

Modern Hebrew grammar wikipedia , lookup

Esperanto grammar wikipedia , lookup

Russian grammar wikipedia , lookup

Compound (linguistics) wikipedia , lookup

Vietnamese grammar wikipedia , lookup

Icelandic grammar wikipedia , lookup

Old English grammar wikipedia , lookup

Italian grammar wikipedia , lookup

Latin syntax wikipedia , lookup

Ancient Greek grammar wikipedia , lookup

Scottish Gaelic grammar wikipedia , lookup

Polish grammar wikipedia , lookup

Dutch grammar wikipedia , lookup

Lexical semantics wikipedia , lookup

Serbo-Croatian grammar wikipedia , lookup

Yiddish grammar wikipedia , lookup

Pipil grammar wikipedia , lookup

Transcript
Pseudo-incorporation in Dutch
Geert Booij
University of Leiden
Abstract
This paper proposes an analysis of bare noun + verb combinations that behave as lexical units but are
phrasal in nature. Such combinations are interpreted as cases of pseudo-incorporation and denote a
habitual or nameworthy activity. The bare noun that functions as Theme is optionally incorporated
into a minimal verb phrase that behaves as an intransitive predicate. This pseudo-incorporation
structure is also available for adverbs, adjectives, and verbs. Non-argumental nouns that are not
licensed by regular syntax can also appear in this structure. The analysis proposed correctly predicts
when the different word combinations can be separated in syntactic structure. The pseudoincorporation structure is claimed to be a structure for lexical phrases. Thus, the lexicon encompasses
certain types of syntactic structure.
Keywords: pseudo-incorporation, particle verbs, lexical integrity, lexical phrases, bare nouns
1. Introduction.*
In many languages nouns can be combined with verbs into verbal compounds with the
structure [N V]V or [V N]V. This word formation process is referred to as incorporation, and
has been studied extensively (Carlson, 2006; Mithun, 1984; Mithun, 2000; Mithun and
Corbett, 1999; Riehl and Kilian-Hatz, 2005). In prototypical incorporating verbal compounds
the noun saturates a thematic role of the verb, often the role of Patient. Noun incorporation is
used primarily to form verbs that express an institutionalized (“nameworthy”) activity. These
incorporated nouns do not denote specific objects, and receive a generic interpretation.1
The difference between a VP with a Determiner Phrase (DP) and a VP with noun
incorporation is illustrated by the following examples from the Micronesian language
Ponapean:
(1) (a) I kanga-la wini-o
I eat-COMP medicine-that
‘I took all that medicine’
(b) I keng-winih-la
I eat-medicine-COMP
‘I completed my medicine-taking’
(source Mithun 1984: 850)
In (1b) we see a case of noun incorporation. It has a different meaning than (1a), which has a
syntactically independent object wini-o. In (1b), with an incorporated object winih-, the
sentence indicates completion of the action of medicine taking, while there may be medicine
left. Typically, incorporated nouns are unmarked for definiteness, number and case, and the
verbal compound behaves as an intransitive verb, whereas its verbal head is transitive. Thus,
noun incorporation often has the effect of creating verbs with reduced syntactic valency:
since the Patient-argument of the verb is expressed by the incorporated noun, this argument
will no longer receive an independent syntactic expression.
In a number of cases, such combinations of a noun and a verb have been argued not to
have the status of compounds stricto sensu, that is words, but rather that of lexical units with
phrasal status. For instance, in Hungarian we find the following possibilities for
incorporation (Farkas, 2006; Farkas and de Swart, 2003):
Groninger Arbeiten zur Germanistischen Linguistik 46 (2008), 3-26
Center for Language and Cognition Groningen
http://gagl.eldoc.ub.rug.nl
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
4
(2) (a) Mari olvas egy
verset
Mari read a
poem.ACC
‘Mari is reading a poem’
(b) Mari verset
olvas
Mari poem.ACC read
‘Mari is reading a poem/poems/poetry’
(c) Mari verseket
olvas
Mari poem.PL.ACC read
‘Mari is reading poems’
In sentence (2a), the object egy verset ‘a poem’ occurs after the verb, the regular word order
of Hungarian being SVO. In sentence (2b), a bare noun verset precedes the verb. In this
sentence, the NV combination denotes the act of reading one or more poems, that is, poetry.
So this sentence does not mean that Mari is reading one, unspecified poem. In sentence (2c),
the plural noun verseket is used, again without determiner, and with a generic interpretation
for this plural noun.
A clear indication of the phrasal status of the NV combinations in (2b) and (2c) is that the
noun is case-marked. In compounds, an incorporated N constituent does not bear its own
marking for structural case. The noun and the verb are also separable, for instance by the
word nem ‘not’ (Kiefer, 1992). Hence, Farkas and de Swart correctly conclude that these
Hungarian NV combinations are instances of pseudo-incorporation rather than
incorporation. Note furthermore that the noun can be marked as a plural, hence carry a
specification for Number, as in (2c). This, however, is itself not a proof of phrasal status since
plural nouns can occur in the non-head position of compounds with an unambiguous word
status (Booij, 1993; Booij, 1996).
Pseudo-incorporation has been reported to exist for a number of languages: Danish
(Asudeh and Mikkelsen, 2000), Dutch (Booij, 1990; Kooij and Mous, 2002), and other
Germanic languages (Dahl, 2004; Ebert, 2000), some Eastern-Indonesian languages
(Klamer, 2001), Hindi (Dayal, 2007), and Niuean, an Oceanic language (Massam, 2001).
Mithun (2000) discusses a similar phenomenon based on Miner (1986), the combination of a
bare noun and a verb into a combination that does not have the status of one word. She refers
to this kind of construction as the ‘noun stripping’ construction since the noun is stripped of
its inflectional markings. However, pseudo-incorporation cannot be identified with noun
stripping. As shown by the Hungarian examples above, noun, case and number markings
may be preserved in pseudo-incorporation.2
In this article I will discuss pseudo-incorporation in Dutch, The basic claim to be made is
that certain types of word + verb combinations (in particular N + V combinations) can occur
in a phrasal construction of pseudo-incorporation. The word and the verb then form a
minimal phrasal V-projection: they are neither real verbal compounds (that is, words) nor
regular VPs. This structural interpretation of such word combinations paves the way for a
systematic account of their complicated syntactic behaviour. I will argue that a similar
structure must be assigned to particle verbs. The existence of such ‘phrasemes’ implies a
lexicon in which certain phrasal patterns are specified.
2. NV combinations
In Dutch we find NV combinations that are sometimes referred to as (a subclass of) separable
complex verbs of this language. A number of such combinations is listed in (3):
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
(3)
(a)
Noun Verb
bier brouwen
brood eten
koffie zetten
thee drinken
gloss
beer brew
brood eat
make coffee
tea drink
(b)
adem halen
les geven
piano spelen
ruzie maken
breath take ‘to take breath’
lesson give ‘to teach’
piano play
‘to play the piano’
quarrel make ‘to quarrel’
5
‘to brew beer’
‘to eat bread’
‘to make coffee’
‘to drink tea’
These NV combinations are given here in their citation form, with the infinitival form of the
verb with the ending -en. According to the rules of Dutch orthography, these NV
combinations have to be written as one word, without spacing when there are no intervening
words. However, in order not to prejudge the linguistic analysis of these NV combinations, I
will write them with spacing. The nouns are all used as bare nouns here, without a
determiner. Those in (3a) can be used both as mass nouns, hence without a determiner, and
as count nouns (with a determiner); those in (3b), can also occur with or without a
determiner.3
The Dutch word combinations in (3) can receive a standard syntactic analysis as VPs with
a direct object that has the form of a bare noun, that is, a DP without an overt determiner.
Most of these nouns can occur without a determiner anyway because they can receive a mass
noun interpretation (3a). Count nouns such as les and piano can also appear without
determiner in certain syntactic contexts, as in:
(4) (a) Morgen
heeft ze
weer les
Tomorrow
has
she
again lesson
‘Tomorrow she will be taught again’
(b) Ik houd veel van piano
I love much of
piano
‘I love piano music very much’
The absence of the determiner implies a generic, non-specific use of these singular nouns.
This is a more general phenomenon, as illustrated by the following Dutch examples
(Haeseryn et al., 1997: 196-220):
(5)
per trein
zonder bril
op school
‘by train’
‘without glasses’
‘at school’
In these prepositional phrases, the bare noun receives a general, non-specific reading. There
is a large set of such P + N expressions with generic interpretation in Dutch.
These N + V combinations must receive a phrasal interpretation in those syntactic contexts in
which they are separable (for instance, in root sentences and in verbal clusters); the form of
their past participles shows the same for contexts in which N and V are adjacent:
(6) (a) Jan speelt piano
John plays piano
‘John plays the piano’
(b) … dat Jan piano wilde
spelen
… that John piano wanted play
‘... that John wanted to play the piano’
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
6
(c) Jan heeft piano gespeeld
John has
piano played
‘John has played the piano’
In sentence (6a), the finite verb speelt occurs in second position, but the noun piano occurs
at the end of the sentence. In sentence (6b), the verb wilde ‘wanted’ forms a verb cluster with
the verb spelen, and thus splits the combination piano spelen. The past participle prefix
ge- in (6c) appears before the verbal stem, but after the noun piano. These facts show that
the grammar has to assign phrasal status to piano spelen.
On the other hand, there are indications of the lexical unity of piano spelen. One
indication is that spelen ‘to play’ cannot be used as a transitive verb with phrases like een/de
piano ‘a/the piano’; the transitive verb to be used in such contexts is bespelen, as in de piano
bespelen ‘to play on the piano’. It is with the verb spelen that the noun piano can and must be
used in its bare form, without an attributive adjective or a determiner:
(7) (a) Jan speelt *een piano/ *de piano/ *een mooie piano
‘John plays a piano / the piano / a nice piano’
(b) Jan bespeelt {*piano / een piano / de piano / een mooie piano}
‘John plays on *piano / a piano / the piano / a nice piano’
These facts show that piano spelen is a lexical collocation of Dutch, and in that sense a lexical
unit. That is, the verb spelen when combined with an object-NP denoting a musical
instrument requires an object-NP consisting of a bare noun. These facts do not imply,
however, that piano spelen is formed in the lexicon: the combination as such can be
generated in the syntax.
But this is not the whole story of lexical units like piano spelen. There are two syntactic
tests that show a special property of these NV combinations, their possible occurrence in the
aan het + INF-construction, and their position in verb clusters.
First, NV combinations such as piano spelen may occur as a constituent of the aan het
infinitive-construction of Dutch. This is a construction with a progressive meaning Booij
(2008) found in a number of West-Germanic languages (Ebert, 2000). Consider the use of
piano spelen in sentence (8):
(8)
Jan is piano {aan het
John is piano at
the
‘John is playing the piano’
spel-en / aan
play-INF at
het
the
piano spel-en}
piano play-INF
Normally, the object of a verbal infinitive in the aan het INF-construction has to appear
before aan het, as illustrated in (9). The parenthesized part is the part of the sentence for
which two alternatives are considered:
(9)
Jan is {de piano aan het bespelen / * aan het de piano bespelen}
In (9) piano is preceded by a definite determiner de ‘the’, and hence it is not used as a bare
noun. In contrast, in the case of NV combinations such as piano spelen the noun can appear
right before the verb since it is a bare noun.
The special nature of these NV combinations also manifests itself in the verb cluster
construction mentioned above. Direct objects of main verbs cannot be clustered in standard
Dutch together with their verb, but these bare nouns can form part of the verbal cluster:
7
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
(10)
(a) … dat Jan {piano wilde spelen / wilde piano spelen}
‘… that John piano wanted play / wanted piano play’
(b) … dat Jan {de piano wilde bespelen / * wilde de piano bespelen}
‘… that John the piano wanted to play / wanted the piano play
Hence we have to conclude that such NV combinations have a special status, and behave as a
very tight lexical (though phrasal) unit. This raises the question of the structure of such NV
combinations.
The default syntactic analysis of bare nouns, that is nouns that are not preceded by an
overt determiner, is that they have the status of DPs with a zero-determiner (Radford 2004:
143). This would mean that a NV combination like koffie zetten can receive the structural
interpretation of a normal syntactic phrase, [DP V]VP. This structural interpretation is
confirmed by the fact that the negative determiner geen must be used in the negative versions
of such sentences:
(11)
(a) Jan zet
geen koffie /
John makes no
coffee
‘John does not make coffee’
*Jan zet
niet
John makes not
(b) Morgen
geef ik geen
Tomorrow
give I no
‘Tomorrow, I will not teach’
les
/
lesson
*Morgen
Tomorrow
(c) Ik rijd
helaas geen
I ride alas no
‘Alas, I do not drive’
/
*Ik
I
rijd
ride
*Hij
He
haalt niet
takes not
auto
car
(d) Hij haalt geen adem meer /
He takes no
breath more
‘He does not breathe anymore’
koffie
coffee
geef
give
helaas niet
alas not
ik niet
I not
les
lesson
auto
car
adem meer
breath more
The use of the negative determiner geen presupposes the presence of a DP with a Determiner
position. Hence, these bare nouns have DP-status. The negative adverb niet is used, on the
other hand, if an intransitive verb has to be negated:
(12)
Jan speelt niet
John plays not
‘John does not play’
If koffie zetten were a single word, that is an intransitive compound verb, we would expect
that it could always be negated with niet, but this is not the case, as shown in (11). However,
negation with niet is possible besides negation with geen in syntactic contexts in which the N
and the V are adjacent, as shown in (13):
(13)
(a) … dat Jan {geen / niet} koffie zet
… that John no / not
coffee makes
‘John does not make coffee’
(b) Jan is geen koffie aan
John is no
coffee at
‘John is not making coffee’
het
the
zetten
make.INF
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
(c) Jan is niet aan het
John is not
at
the
‘John is not making coffee’
8
koffie zetten
coffee make.INF
(d) Jan is niet KOFFIE aan het
John is not
coffee at
the
‘John is not making coffee, but tea’
zetten,
maar THEE
make.INF but
tea
An analysis that does justice to this array of facts is one in which NV sequences such as koffie
zetten can receive two structural interpretations. They form regular VPs when geen is used,
and are cases of pseudo-incorporation (the formal structure of which will be discussed below)
when niet is used. In sentence (13d), the negative operator niet can be used for negating the
constituent koffie, but only in a contrastive context.
There is a difference in the use of these two constructions, which can be illustrated by the
following sentences both meaning ‘John cannot make coffee’:
(14)
(a) Jan kan geen koffie zetten
(b) Jan kan niet koffie zetten
Sentence (14a) is ambiguous, unlike sentence (14b). In (14a) koffie zetten denotes either an
event or a habitual action, but in (14b), a case of pseudo-incorporation (as proven by the use
of niet as a negative adverb), koffie zetten can only denote a habitual action. Hence, a
sentence like Jan kan vandaag niet koffie zetten ‘John cannot make coffee today’ is
semantically odd. That is, the incorporating predicates function as intransitive predicates
that denote a habitual, nameworthy activity. Hence, sentence (14a) may be uttered in a
situation where there is no coffee available, whereas sentence (14b) may be used in a
situation in which John has not learnt how to make coffee.
The structure and corresponding meaning that I therefore propose to assign to the
structure of pseudo-incorporation is the following:
(15)
[[N][V]]V’ ‘to perform a habitual/conventional action V with respect to N’
This structure is a constructional schema that specifies the structure of pseudo-incorporation,
and (a first approximation of) the semantic correlate of its formal structure. This
incorporation structure licenses the use of a noun that is not a full-blown DP but just an N0.
Structure (15) represents intransitive complex predicates that can be negated with the adverb
niet ‘not’. In this structure, there is no DP that could accommodate a negative determiner
geen. In short, pseudo-incorporation can be used to explicitly construe the action involved as
a nameworthy or habitual action.
The structure in (15) requires that the incorporated noun be adjacent to the verb. This is
not in conflict with the fact discussed above that N and V can be non-adjacent, since
structures in which N and V are non-adjacent are the second structural option, licensed by
the regular syntax in which the N is part of a DP. The fact that this DP can consist of a bare
noun only is a property of the verb spelen that licenses such bare nouns because it invokes a
generic interpretation of the noun, just like the prepositions in (5). Yet, we cannot interpret
the non-split NV sequences as compound verbs, that is, as words. If this were the case, we
would expect that these NVs could appear in second position in Dutch root clauses, but this is
not the case:
(16)
(a) *Jan koffie-zet goed
(b) *Mijn vrouw auto-rijdt prima
(c) *Mijn dochter les-geeft met plezier
‘John coffee-makes well’
‘My wife car-rides excellently’
‘My daughter lesson-gives with pleasure’
This conclusion as to the non-compound status of these NVs is confirmed by the form of the
past participle: in the examples in (17), the participial prefix ge- is prefixed to the verbal stem,
9
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
not before the NV. This stands in contrast with the form of such participles in cases of
undisputed NV compounds with word status in (18). Past participles are formed by prefixing
ge- to the verbal stem, and simultaneously suffixing the verbal stem with either -t/d for
regular verbs or –en for ablauting verbs:
(17)
(a) Ik heb gisteren niet
I have yesterday not
‘I did not teach yesterday’
{les ge-geven / *ge-les-gev-en}
lesson giv-en
(b) Mijn vrouw heeft vandaag nog niet
My
wife has
today
not
yet
‘My wife did not yet drive a car today’
(18)
{auto ge-reden / *ge-auto-red-en}
car
driv-en
(a) Deze misdaad heeft hem {ge-brand-merk-t / *brand-ge-merk-t}
This crime
has
him brand-mark-ed
‘This crime has denounced him’
(b) Hij heeft altijd
{ge-beeld-houw-d / *beeld-ge-houw-d}
He has
always
statue-hewn
‘He always sculptured’
Note that the occurrence of the negative adverbial niet is proof of the incorporated status of
the N in the sentences (17).
In sum, a combination of N and V can occur in three different structural configurations:4
(19)
regular syntax:
[[N]DP V ]VP
pseudo-incorporation: [N V]V’
compounding
[N V]V
Pseudo-incorporation thus creates intransitive predicates, which license niet as negative
operator, can occur in the progressive aan het + INF-construction, and can cluster with
raising verbs such as willen ‘to want’, as in the clause … dat Jan wilde piano spelen (cf. 10a).
Incorporation is possible if the N V combination can be construed as a nameworthy activity.
Hence, not every verb with a bare noun lends itself to incorporation. For instance, the lexical
collocation honger hebben ‘lit. to hunger have, to starve’ does not denote an activity.
Therefore, it does not lend itself to incorporation, as proven by the ungrammaticality of
structures that presuppose incorporation:
(20)
(a) In Canada zul
je
{geen honger
In Canada will you no
hunger
‘In Canada you will not starve’
hebben / *niet
have
not
honger
hunger
hebben}
have
(b) ... omdat
je
daar {honger
because you there hunger
‘because you will starve there’
hebben / *zult honger
have
will hunger
hebben}
have
zult
will
Pseudo-incorporation is thus a process that is not blind to the semantics of the word
combination involved, and restricted to certain NV combinations. The lexically governed
nature of the process is confirmed by pairs like adem halen ‘lit. to take breath, to breathe’ /
adem krijgen ‘to get breath’. It is only the first N V combination that denotes the habitual
action of breathing with a certain duration, and that therefore exhibits the typical effects of
incorporation discussed above. Therefore, the pseudo-incorporation construction is to be
seen as a lexical construction even though it is phrasal in nature.
For a proper appreciation of this proposal as to Dutch pseudo-incorporation it is useful to
compare the cases of pseudo-incorporation with two other classes of complex predicates,
10
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
verbal compounds and particle verbs. As illustrated in (18), Dutch has some verbal
compounds. It is well known that the word formation process of verbal compounding is
unproductive in most Germanic languages, and certainly in Dutch. There are, however, some
specific sources of verbal compounds. One is the conversion of NN compounds into verbs.
The verb brandmerken ‘to fire-mark, to denounce’, for instance, is a conversion of the
nominal compound brandmerk ‘fire-mark’. A second source of verbal compounds is back
formation, as exemplified by the verb beeldhouwen ‘to sculpture’ with the structure [NV]V, a
back formation from beeldhouwer ‘sculptor’ that has the structure [[beeld]N[[houw]V -er]N]N.
A third class of verbal compounds in Dutch is the class of exocentric compounds such as
klapper-tanden ‘to rattle-teeth, to rattle with one’s teeth’ with the structure [VN]V
(Weggelaar, 1986).
As to particle verbs, it has been shown in many publications on Dutch and German (Blom,
2005; Blom and Booij, 2003; Booij, 1990; Booij, 2002b; Müller, 2002; Müller, 2003; Müller,
2006; Vikner, 2005; Zeller, 2001) that they differ from prefixed verbs in that the particle can
be separated from the verb. Hence, they must have phrasal status in the relevant syntactic
contexts, just like the NV combinations under discussion. The following sentences with the
particle verb op bellen ‘to phone someone’ illustrate the basic pattern of distribution of
particles:
(21)
(a) ... dat
Jan zijn moeder op belde /
that John his mother up phoned
‘John called his mother’
(b) ... dat
Jan zijn moeder {wilde
that John his mother wanted
‘that John wanted to call his mother’
(c) Jan is zijn moeder {aan
John is his mother at
‘John is calling his mother’
het
the
Jan belde
John phoned
op bellen /
up phone
op bell-en
/
up phone-INF
zijn moeder op
his mother up
op wilde
up wanted
op aan
up at
bellen}
phone
het bellen}
the phone-INF
A first option is assigning particle verbs the phrasal structure [Part V]V’ in the lexicon. This is
the kind of structure proposed in Booij (2002a; 2002b) and Blom (2005) for Dutch particle
verbs. This kind of lexical phrasal structure is assumed to be still abstract in that the Verb is
specified in its stem form. From this abstract form the specific inflectional forms will be
created by means of the regular rules of inflection, which will then appear in individual
sentences. This structure correctly predicts that the particle verb can behave as a unitary
predicate in verb clusters (cf. 21b) and in the progressive construction (21c). If such a
structure is assumed, we need additional mechanisms for accounting for sentences in which
the particle and the verb are separated (as in 21a, second option, and 21b, second option; cf.
also Spencer, 2005: 80). In a grammar with transformations, we can consider [Part V]V’ the
underlying structure of particle verbs in clauses with (underlying) SOV word order. Syntactic
transformations (such as Verb-Second and Verb Raising) take care of the appearance of the
verb in other syntactic contexts than embedded clauses. In non-transformational varieties of
grammar such as HPSG this distribution can be accounted for by having the verb co-indexed
with an empty position after the particle (Müller 2007). I will refer to both types of analysis
as ‘movement’ analyses since they share the assumption of a basic word order Particle Verb.
This analysis thus assigns the same structure to particle verbs and cases of nounincorporation. This raises the problem, however, of why the V cannot be ‘moved’ out of this
structure in the case of pseudo-incorporation of nouns, whereas it should not be blocked in
the case of particle verbs. Such a prohibition on ‘movement’ of the V in [NV]V’ structures
would be necessary even though at first sight Verb Second may be allowed to apply because N
and V can be separated. The crucial data is formed by sentences such as those in (11) with a
negative word. If ‘movement’ out of [NV]V’ structures were possible, we would predict the
following root clauses to be grammatical, but incorrectly so:
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
(22)
11
(a) *Jan haalti niet adem ti
John takes not
breath
‘John does not breathe’
(b) *Mijn moeder
bakti niet brood ti
My mother
bakes not
bread
‘My mother does not bake bread’
These sentences are grammatical only if the word niet is replaced with geen, the option that
is available if the noun is not incorporated, but forms a regular DP.
A second option for the analysis of particle verbs is to lexically represent them as verbs
that require a specific particle in their complement, where the distribution of the particle is
governed by the regular syntax. For instance, the particle verb op bellen then receives a
lexical representation as a transitive verb bellen with the meaning ‘to phone someone’ that
requires the presence of the particle op. The productivity of the class of particle verbs is
accounted for by lexical rules. This is the proposal for the representation of particle verbs in
the work of Müller (Müller, 2002; Müller, 2003; Müller, 2006), that is as verbs with the
specific collocational property that they select a particle. Müller also assumes that the lexical
entry for particle verbs requires the particle to be left-adjacent to the verb or a trace of that
verb. In Müller’s framework no disitinction is made between the nodes V and V’: both the
verb and the the particle verb are dominated by a V node.
Whatever the precise analysis of particle verbs may be, what we minimally need is a way
to ensure that in [N V]V’ structures, the V head that governs the incorporated bare noun
cannot be a trace. This can be achieved by assuming a particular licensing condition for bare
nouns. Bare nouns can be licensed in two ways, by being interpreted as regular DPs, or by
having the status of non-projecting noun (N0), which is only licensed by the pseudoincorporation structure. What we therefore need is the condition that an N0 that is not
headed by a DP node cannot be licensed by a trace, but only by a verbal head with a
phonological form. In this respect bare nouns are different from particles, which do not
belong to one of the lexical categories. This condition reminds us of the principle of lexical
integrity that forbids movement of parts of words by syntactic rules. Lexical integrity implies,
for instance, that in Dutch the verbal stem of a prefixed verb cannot be separated from that
prefix by ocurring in the second position of the root clause. In other words, the prefix can
only be licensed by an adjacent element with a phonological form. Incorporated nouns
appear to be subject to the same condition even though they form not parts of complex words
stricto sensu, but of tight lexical phrases.
A second case of pseudo-incorporation of nouns in Dutch is one in which the noun
exhibits plural morphology.5 Again, the NV combination denotes an institutionalized or
nameworthy action:
(23)
(a) aardappels schill-en
potatoes
peel-INF
‘to peel potatoes’
(b) appels
plukk-en
apples
pick-INF
‘to pick apples’
(c) brieven schrijv-en
letters
write-INF
‘to write letters’
12
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
(d) kous-en stopp-en
stockings mend-INF
‘to mend stockings’
Unlike the cases of singular N incorporation, these combinations are not written as one word,
but as two in Dutch orthography. The criteria for considering these word combinations
lexical units, and cases of pseudo-incorporation are the same as for the cases of incorporation
of singular nouns: the noun can occur right before the infinitive in the aan het + INFconstruction and in verb clusters, and the negative adverb niet can be used. This indicates
that these NV combinations function as intransitive predicates in such environments, and
hence the object-argument must be an incorporated one. As was the case for the adem halen
class, these NV combinations may also function as full-fledged transitive VPs in which the
noun is a DP :
(24)
full DP status of plural noun / pseudo-incorporation of plural noun
(a)
Jan is {aardappels aan het schill-en /
John is potatoes
at the peel-INF
‘John is peeling potatoes’
(b)
Jan is {de aardappels aan het schill-en /
John is the potatoes
at the peel-INF
‘John is peeling the potatoes’
(c)
Jan is {nieuwe aardappels aan het schill-en /
John is new potatoes
at the peel-INF
schill-en}
peel-INF
‘John is peeling new potatoes’
*aan het nieuwe aardappels
at the
new potatoes
(d)
... dat Jan {geen aardappels
schilt /
... that John no potatoes
peels
‘that John does not peel potatoes’
aardappels
potatoes
(e)
Jan {schilt geen aardappels /
John peels no potatoes
‘John does not peel potatoes’
(f)
Jan is {geen aardappels aan het schill-en /
John is no potatoes
at the peel-INF
schill-en
peel-INF
‘John is not peeling the potatoes’
niet
not
(g)
... dat Jan {aardappels wilde
schillen
... that John potatoes
wanted peel
‘... that John wanted to peel potatoes’
/
aan het
at the
schill-en}
peel-INF
*aan het de aardappels
at the
the potatoes
niet
not
*schilt niet
peels not
aardappels
potatoes
schill-en}
peel-INF
schilt}
peels
aardappels}
potatoes
aan het
at the
wilde
wanted
aardappels
potatoes
aardappels
potatoes
schillen}
peel
Again, we assume the plural nouns in the pseudo-incorporation to have the syntactic status of
N0, and their occurrence is licensed by the adjacent verb in the [N V]V’ structure.
As pointed out above, the pseudo-incorporation structure typically evokes the
interpretation of these NV combinations as a nameworthy activity. The effect is the creation
of intransitive predicates. The incorporation cannot be interpreted in terms of a derivation
from a clause with a transitive VP because it is only the incorporation structure that creates
the obligatory interpretation as a nameworthy acticvity. Therefore, this construction has a
13
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
lexical status. Additional evidence for this point is that the pseudo-incorporation structure
must also be available in syntactic contexts where a full DP structure is impossible as an
underlying structure because the syntactic context requires intransitive predicates. That is
the case for the aan het+ INF- complements of causative verbs such as brengen ‘to bring’,
krijgen ‘ to get’ and maken ‘to make’ (Haeseryn et al., 1997: 1052-1053). Note that twijfelen
‘to doubt’ is an intransitive verb, but vertellen ‘to tell’ a transitive one that becomes
intransitive when the object argument is pseudo-incorporated):
(25)
(a) Hij brengt ons
aan
He brings us
at
‘He makes us doubt’
het
the
twijfel-en
doubt-INF
(b) *Hij brengt ons
sprookjes aan
He
brings us
fairy tales at
‘He makes us tell fairy tales’
(c) Hij brengt ons
aan het
He brings us
at
the
‘He makes us tell fairy tales’
het
the
(intransitive predicate)
vertell-en
tell-INF
sprookjes vertell-en
fairy tales tell-INF
(transitive predicate)
(intransitive predicate)
The plural nouns in the pseudo-incorporation construction receive a generic interpretation,
similar to that in certain syntactic constructions. Bare plurals in Germanic languages allow
for generic readings in specific contexts. For instance, the plural noun sprookjes ‘fairy tales’
used in (25) receives the same generic interpretation in a sentence such as
(26)
Mijn kinderen houden van sprookjes
‘My children love fairy tales’
In this sentence with the verb houden van ‘to love’, the noun sprookjes denotes a property
rather than a discourse referent. The interpretation of this sentence is something like
(27)
My children love things that are fairy tales
The same interpretation holds for the plural nouns in these pseudo-compounds. Thus, we get
different interpretations for the following two sentences.
(28)
(a) Jan schilt aardappelen
‘John peels potatoes’
(There are potatoes for which holds that John peels them)
(b) Jan is aan het aardappelen schillen
‘John is peeling potatoes
(John peels things that are potatoes)
This specific semantic effect of pseudo-incorporation of invoking a generic interpretation for
bare plurals is therefore a semantic property of this construction (that it shares with other
constructions). Note that the plural marking of the nouns is obligatory since the singular
form is impossible:
(29)
*Jan {schilt aardappel /
Jan peels potato.SG /
‘John is peeling potatoes’
is aan
is at
het
the
aardappel schill-en}
potato.SG peel-INF
The same holds for the conventional activity of collecting stamps:
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
(30)
postzegel-s verzamelen
stamp-PL
collect
‘ stamp collecting’
/
*postzegel
stamp.SG
14
verzamelen
collect
The verb verzamelen ‘to collect’ is one of the verbs that require a plural object, and example
(30) shows that the feature [plural] is semantically active in pseudo-incorporation.6 This can
be contrasted with real compounding in which noun stems receive a generic interpretation
without plural marking. For instance, in the compound postzegelverzamelaar ‘stamp
collector’, the absence of plural marking on the noun postzegel ‘stamp’ does not block a
generic reading: a postzegelverzamelaar certainly collects more than one stamp. That is, it is
only in cases of real compounding that number neutrality is at stake.
3. Immobile verbs
There is another class of NV combinations in Dutch that differs from the N V combinations
discussed above in that the noun does not function as the Patient, and has no argument role.
Examples are the following (more Dutch examples, and similar examples for German can be
found in Vikner 2005).
(31)
NV
buik spreken
koord dansen
mast klimmen
steen grillen
stijl dansen
vinger verven
zak lopen
zee zeilen
‘to stomach speak, ventriloquizing’
‘to rope dance, walking a tightrope’’
‘to pole climb, climbing the greasy pole’
‘to stone grill, stone-grilling’
‘to style dance, ballroom-dancing’
‘to finger paint’
‘to bag walk, running a sack-race’
‘to see sail, ocean-sailing’’
Again, in order not to prejudge the linguistic analysis, I write these word combinations as two
words, although Dutch orthography requires them to be written as one word.
These word combinations cannot be compounds because they do not occur in root
sentences; in such sentences, they can only be used with the periphrastic progressive
construction mentioned above; in non-root-sentences, however, they can be used in both
their non-finite and their finite forms (Booij, 2002a, Van Marle, 2002; Vikner, 2005):
(32)
(a) *Mijn vader zee zeilt vaak
My
father sea sails often
‘My father often sails at sea’
(b) Mijn vader is vaak aan
My
father is often at
‘My father often sails at sea’
het zee zeil-en
the sea sail-INF
(c) ... dat mijn vader vaak zee zeilt
... that my
father often sea sails
‘that my father often sails at sea’
(33)
(a) *Mijn zuster stijl danst goed
My
sister style dances well
‘My sister is a good ballroom dancer’
(b) Mijn zuster is vaak aan het stijl
My
sister is often at
the style
‘My sister does ballroom dancing often’
dans-en
dance-INF
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
15
(c) ... dat mijn zuster goed stijl danst
... that my
sister well style dances
‘that my sister is good at ballroom dancing’
As Vikner (2005) pointed out, the same array of facts holds for German. Vikner (2005) refers
to these NVs as immobile verbs because the finite verb cannot be moved into other syntactic
positions such as the first position (questions) or second position (root clauses).
The non-occurrence of these pseudo-compounds in root sentences follows from assigning
them phrasal, that is, V’ status. A pseudo-verbal compound V’ cannot occur in second
position in root clauses, because this second position is for a single finite V only, not for a
phrasal predicate. Recall that in this respect, these NVs differ from real verbal compounds
such as voetbal ‘to play soccer’ (a verbal compound created through conversion of the
nominal compound voetbal ‘football’):
(34)
Mijn vader voetbal-t elke zaterdag
My
father football-s every Saturday
‘My father plays football every Saturday’
In the case of particle verbs, the finite form of the verb can appear in second position, with
the particle somewhere else in the clause. This is, however, no option in case the ‘stranded’
word is a noun that cannot be interpreted as an object-DP. This explains the difference with
N + V combinations such as piano spelen for which two structural interpretations are
available, one as a regular VP, and one as a case of pseudo-incorporation:
(35)
(a) Jan belde zijn moeder op
‘John phoned his mother’
(b) *Mijn vader zeilt vaak zee
‘My father often sails at sea’
(c) Mijn vader speelt soms piano ‘My father sometimes plays the piano’
This difference in behaviour between particles and argumental nouns on the one hand, and
non-argumental nouns, on the other hand, can be accounted for as follows. As to Ns, they
must normally occur in DP-expressions. As Radford (2004) points out, argument nominals
are always D-expressions (i.e. projections of D). However, N-expressions can be licensed in
specific structural configurations. A straightforward one is that of morphological structure: in
compounds, for instance, the occurrence of an N is licensed by morphological structure. That
is, N-expressions can be licensed in specific contexts. What I therefore propose is that single
Ns (N0) that do not form a DP with a non-overt determiner are licensed by the pseudoincorporation construction. In other words, we allow for nouns to be non-projecting and
appear as N0 in pseudo-incorporation structures.
The analysis presented here answers the question why NV combinations such as adem
halen behave different from NV combinations such as zee zeilen in which the noun does not
function as an argument of the verb. The NV sequence adem halen can be interpreted as a
regular VPs with a bare noun that forms a D-expression, and hence a noun like adem can
function as a DP. Such structures are only possible with nouns that can function as an
argument of the verb. In this respect, they are therefore different from NV combinations such
as zee zeilen that only occur in the pseudo-incorporation construction. In his discussion of
the different types of NV compounds of Dutch, Ackema (1999: 139) proposed the following
generalization:
(36)
“If N is an argument of V, N-V is separable.”
This generalization is meant to express the distributional differences between the NV
combinations of the adem halen type and those of the zee zeilen type. It follows directly from
the analysis presented here because the separable NV combinations are in fact regular
phrases, and not compound words. Yet, even the NV combinations that Ackema qualifies as
16
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
inseparable complex verbs have to be considered as phrasal in nature, in order to explain why
they do not occur in second position in root clauses and why the participial prefix appears
after the noun, before the verbal stem. In other words, none of the NV combinations has
word status.
The structure of pseudo-incorporation [N V]V’ is meant to require the two words to be
adjacent, and thus predict the class of zee zeilen combinations to be non-separable, unlike the
koffie zetten class for which another syntactic interpretation is available. The contrast is also
found in sentences with contrastive focus such as:
(37)
(a) KOFFIE kan hij niet
Coffee
can
he not
‘Coffee he cannot make’
zetten
make
(b) *ZEE kan hij niet zeilen
Sea
can
he not
sail
‘He cannot do sea-sailing’
This shows once more that a habitual interpretation is a necessary condition for N + V
combination to appear in incorporation structures, but that this structure is only obligatory
for those cases in which there is no other syntactic structure available (as in the case of zee
zeilen).
Individual cases of pseudo-incorporation may develop into real compounds, at least for
some speakers of Dutch. In that case, they can occur in second position in root clauses.
Examples from a Google search (15 January, 2008) are stijldansen en koekhappen. Both of
these NV combinations are used in root clauses:
(38)
(a) De koning zelf koek-hapt
‘The king himself cake-eats’
(b) ... en stijldanst hij met zijn nichtje
‘... and style-dances he with his niece’
There is variation among speakers in this respect, and this is to be expected given the fact
that NV sequences have potentially three different structural interpretations. My Google
search data confirm that generally we do not find these NVs in second position in clauses, but
it comes as no surprise that language users find it not always easy to assign the proper
structure to such word combinations. Hence, some language users impose a compound
interpretation on NV sequences. For instance, I found the following numbers of tokens for
the competing participle forms of NV combinations (Google search 15 January 2008):
(39)
pseudo-incorporation
buik-ge-sprok-en 23
steen-ge-grild
9
stijl-ge-dans-t
35
vinger-ge-verf-d
3
wad-ge-lop-en 176
zak-ge-lop-en
54
compounding
ge-buik-spreek-t
3
ge-buik-sprok-en 0
ge-steen-gril-d
363
ge-stijl-dans-t
355
ge-vinger-verf-d
257
ge-wad-loop-t
40
ge-wad-lop-en
0
ge-zak-loop-t
3
ge-zak-lop-en
0
gloss
ventriloquized
stonegrilled
ballroom-danced
finger-painted
walked across the shallows
run a sack-race
The forms on the left are to be expected if a pseudo-incorporation interpretation is imposed
on these NV sequences. The forms on the right are expected if these NVs are compounds.
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
17
Reinterpretation as compounds can also be deduced from the fact that NVs with ablauting
verbs (in these examples spreken ‘to speak’ and lopen ‘to walk’) tend to have the past
participle form of regular verbs: gebuikspreekt, gewadloopt, gezakloopt (the past participle
of spreken is gesproken and that of lopen is gelopen). These NV compounds, in particular
those in which the verb is regularly inflected, may therefore have been interpreted as back
formations from compounds of the type N + Infinitive. Recall that NV compounding is
unproductive in Germanic languages. However, since infinitives have also nominal status,
they can productively form the head of NN compounds. If we want to create inflected forms
for such compounds, they must be reinterpreted as NV compounds. The feature [ablaut] of
these verbal stems will not be accessible in the back-formed NV, and they will therefore be
inflected as regular verbs (Booij, 1989).7
The reinterpretation of phrasal combinations as NV compounds took place on a much
larger scale in another Germanic language, Frisian where we find sentences like (Dyk, 1990:
30):
(40)
Hy noas-snutte wakker
He nose-blew
heavily
‘He blew his nose heavily’
Vikner (2005) proposed another explanation for the immobility of these NVs than that
proposed above. According to him, immobile verbs are simultaneously V’s and V*s, that is,
both words and small phrases. The only syntactic position in which these NVs can fulfill the
requirements of both structures is when N and V are adjacent. Hence, the verbal part cannot
be moved. There are two problems with this proposal, however. One is that assigning them Vstatus implies that the participial prefix ge- cannot occur in the middle, right before the
verbal stem. As we saw above, the prefix does occur in that position. Secondly, Vikner’s
analysis does not explain Ackema’s correct generalization that only verbs with argumental
nouns possess mobility. He has to stipulate for which NVs the double requirement holds. In
my analysis this difference is accounted for by providing two different structural
interpretations for combinations of verbs with argumental nouns only. A noun like zee in zee
zeilen is not a argument but an adjunct. Hence, it can only be licensed by the pseudoincorporation construction.
Immobility has been observed for two other classes of Dutch complex verbs (Koopman,
1995), and for similar German verbs (Vikner, 2005). The first class is the set of particle verbs
that are prefixed with the prefix her- ‘re-’, as in
(41)
her-in-dijk-en
her-in-voeren
‘to re-in-dike, to put within dykes again’
‘to re-in-lead, to reintroduce’
They do not exhibit the mobility of particle verbs, which is understandable since there is no
canonic position for a prefix-particle combination in root clauses:
(42)
(a) *Het parlement herinvoerde de doodstraf
‘The parliament reintroduced capital punishment’
(b) *Het parlement voerde de doodstraf herin
‘The parliament introduced capital punishment again’
(c) ... dat het parlement de doodstraf herinvoerde
‘... that the parliament capital punishment reintroduced’
(d) Het parlement wilde de doodstraf herinvoeren
‘The parliament wanted to reintroduce capital punishment’
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
18
The second class is exemplified by the verb voor-aan-melden ‘preregister’, a verb preceded
by two particles, voor and aan that both occur on their own with other verbs to form particle
verbs such as aan melden ‘to register’ and voor gaan ‘to take the lead’. Again, there is no
canonic position for a particle sequence. Thus, these verbs do not behave as regular particle
verbs and are immobile. Yet, they are phrasal in nature witness the fact that the prefix ge- of
the past participle appears in the middle, right before the verbal stem: her-in-gedijk-t, herin-gevoer-d, voor-aan-ge-mel-d. Their behaviour is predicted by giving them the same
structure in the lexicon as the cases of pseudo-incorporation discussed before, that is:
(43)
[her[in[voer]V]V’]V’
[voor[aan[meld]V]V’]V’
The prefix her- is a category-neutral prefix that preserves the syntactic category of its bases.
Hence, when it is prefixed to a V’, the resulting structure will also be dominated by a V’ node.
Structure (43) requires the preverbal elements to be structurally adjacent to the verb, as in
the case of NV sequences such as zee zeilen ‘to sail at sea’. And again, these words are phrasal
in nature given the form of their participle. This structure correctly predicts that these
predicates cannot appear in second position in root clauses, because that position does not
allow for V’, only for V.
4. More pseudo-incorporation
A particularly productive instantiation of (15) is found for the verb spelen ‘to play’. It often
combines with nouns that denote a musical instrument or a game:
(44)
(a) blokfluit / cello / gitaar spel-en
recorder / cello / guitar play-INF
‘recorder / cello / guitar playing’
(b) backgammon / playstation / rugby
spel-en
backgammon / playstation / rugby
play-INF
‘backgammon / playstation / rugby playing’
The NVs in (44) can be interpreted as regular VPs, or as cases of pseudo-incorporation, as
shown by the test with the aan het INF-construction:
(45)
Jan is gitaar aan het spel-en / aan
John is guitar at
the play-INF / at
‘John is playing the guitar’
het gitaar spel-en
the guitar play-INF
The most productive use of spelen in NV sequences is that with a diminutive noun that
denotes a particular role or institution:
(46)
dokter-tje
doctor-DIM
spelen
play
‘playing being a doctor’
kerkje
church-DIM
spelen
play
‘playing doing a church service’
moedertje
spelen
mother-DIM play
‘playing being mother’
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
Note that playing a role is by definition a conventional / institutionalized act. The use of bare
nouns as DPs in Dutch is quite regular for the expression of roles and capacities (de Swart et
al., 2007).
Adverbs may also be pseudo-incorporated in verbal constructions, if the resulting phrase
denotes a conventional, institutionalized activity. Examples are listed in (47):
(47)
hard
fast
lop-en
walk-INF
‘running’
wild
wild
plass-en
urinate-INF
‘urinating in public’
Again, these word sequences can receive two structural interpretations. In the literal
interpretation, the adverb can be modified, and appear before the aan het + INFconstruction. If the word combination denotes a conventionalized activity, for instance a
particular sport, the adverb cannot be modified, and the word combination can appear with
the aan het + INF-construction:
(48)
(a) literal interpretation:
Ik loop hard
Ik loop erg hard
Ik ben hard aan het lopen
‘I am walking fast’
‘I am walking very fast
‘I am walking fast’
conventionalized activity interpretation
Ik loop hard
‘I jog’
*Ik loop erg hard
Ik ben aan het hard lop-en
‘I am jogging’
(b) literal interpretation
Jan plast wild
Jan plast erg wild
Jan is wild aan het plassen
‘John urinates wildly’
‘John urinates very wildly’
‘John is urinating wildly’
conventionalized activity interpretation
Jan plast wild
‘John urinates in public’
*Jan plast erg wild
Jan is aan het wild plass-en
‘John is urinating in public’
These facts follow if we assume a constructional schema for the (optional) incorporation of
Adv0 into a complex phrasal predicate that denotes an institutionalized or nameworthy
activity. That is, the occurrence of non-projecting adverbs is licensed by this constructional
schema for phrasal predicates. (Not all speakers accept the conventionalized activity
interpretation for Ik loop hard and Jan plast wild; such speakers will prefer the periphrastic
progressive construction when constructing root clauses.)
Adjectives that function as resultative predicates can also be pseudo-incorporated if the
activity is habitual or institutionalized. Examples are the predicates wit wassen ‘to white
wash, to launder money’ and schoon maken ‘to clean make, to clean’ that can be used in the
aan het INF-construction:
(49)
(a) De boef was (het geld aan het wit wass-en / het geld wit aan het wass-en)
The crook was the money at the white wash-INF/the money white at the wash-INF
‘The crook was laundering the money’
19
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
20
(b) De leerlingen zijn het lokaal (aan het schoon mak-en / schoon aan het mak-en)
The pupils are the classroom at the clean make-INF / clean at the make-INF
‘The pupils are cleaning the classroom’
(50)
(a) De boef waste
het geld
wit
The crook washed the money
white
‘The crook laundered the money’
(b) De leerlingen maakten het lokaal
The pupils
made
the classroom
‘The pupils cleaned the class room’
schoon
clean
These AV combinations can be treated on a par with NV combinations such as adem halen:
they can receive two structural interpretations, the regular one for resultative constructions
(for instance, a VP with the AP schoon, and one with pseudo-incorporation. This explains
why A and V can be separated in root clauses. In the aan het-INF-construction, A and V tend
not to be separated when the combination has an idiosyncratic meaning, as is the case for wit
wassen. In a Google search (Jan 22, 2008), I hardly found the word order wit aan het
wassen, and in the case of the semantically regular schoon maken the proportion of tokens
for the word order aan het schoon maken /schoon aan het maken is 26,200 / 1,190. Yet,
idiosyncracy does not block the separation of the words involved.
A final case to be mentioned are V +V combination such as roer bakken ‘to stir-fry’. This
combination behaves either as a real compound, or as a case of pseudo-incorporation,
witness its finite form roer-bak-t ‘stir-fry.3SG’ and the different participial forms:
(51)
ge-roer-bak-t, ge-roer-bakk-en, roer-ge-bakk-en
all found in the internet-corpus. The last form is the dominant one, but the variation shows
again that two structural interpretations are possible. Similar facts obtain for the VV
combination zweef vliegen ‘lit. glide-fly, glide’. Again, we find the finite verb form zweefvliegt
in root clauses, and a numner of different forms of the past participle. A Google search (25
January 2008) provided the following token numbers:
(52)
ge-zweef-vlieg-d
zweef-ge-vlog-en
zweef-ge-vlieg-d
ge-zweef-vlog-en
223
127
5
1
The uncertainty of Dutch speakers as to the proper structural interpretation of such word
combinations manifests itself also in the fact that we find quite a number of discussions on
internet about what the form of such participles should be.
In sum, pseudo-incorporation is a structure in which several types of words can be
incorporated: nouns (both argumental nouns and non-argumental ones), adjectives/adverbs,
and – though this is rare – verbs. In addition to scheme (15) we must therefore assume the
following cases of pseudo-incorporation:
(53)
[Adv V]V’ ‘nameworthy / habitual activity V in a manner indicated by Adv’
[A V]V’
‘nameworthy / habitual activity V resulting in the property A’
[V1 V2]V’ ‘nameworthy / habitual activity V2 by V1-ing’
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
21
5. Interaction with word formation
We have seen so far that lexical collocations such as the NV sequence adem halen can receive
two different structural interpretations:
• a VP with a bare noun that functions as a DP
• a V’ with the structure [N V]V’
A third structural interpretation seems to be excluded, that of [NV]V compound, because this
type of compound is unproductive in Germanic languages like Dutch. However, as I argued in
Booij (2002a; 2007), this pattern of compounding is productive when it is embedded in
complex words, compounds of the VN type and deverbal nouns. This is the phenomenon of
what I refer to as embedded productivity. Here are some examples from Booij (2002a):
(54)
[[[aardappel]N[schil]V]V[mesje]N]N ‘lit. potato peel knife, potato peeler’
[[[brand]N[blus]V]V[installatie]N]N ‘lit. fire extinguish installation, fire extinguisher’
‘lit. coffee make machine, coffee maker’
[[[koffie]N[zet]V]V[apparaat]N]N
In these nominal compounds the left constituent is a NV compound in which the N functions
as the Patient of the verb.
[NV]V compounding is not only boosted by VN compounding, but also by suffixation with
the deverbal suffixes -er, -ster, -ing and -erij:
(55)
(a) aandacht-trekk-er
brand-bluss-er
gif-meng-er
grappen-mak-er
‘attention drawer’
‘fire extinguisher’
‘poison mixer, poisoner’
‘lit. jokes maker, comedian’
(b) kinder-verzorg-ster
kranten-bezorg-ster
rokken-naai-ster
vee-hoed-ster
‘children’s care worker (fem.)’
‘newspaper deliverer (fem.)’
‘skirts sewer (fem.)’
‘cattle herd (fem.)’
(c) evangelie-verkondig-ing
hand-oplegg-ing
kinder-verzorg-ing
tempel-reinig-ing
‘gospel preaching’
‘hands imposition’
‘child care’
‘temple cleansing’
(d) bijen-houd-erij
bloem-kwek-erij
vlaggen-mak-erij
wijn-zuip-erij
‘bee keeping’
‘flower nursery’
‘flag makery’
‘excessive wine drinking’
Hence, the [NV]V structure is available for lexical combinations of a noun and a verb into a
compound when embedded in a complex word. For instance, a word like ademhaling can be
assigned the following structure:
(56)
[[[adem]N [hal]V]V ing]N
‘lit, breath taking, breathing’
It is quite obvious that the word ademhaling must be related semantically to the lexical
collocation adem halen. Does this mean that the phrasal structure of adem halen must be a
subpart of ademhaling? Not necessarily. The parts of an idiomatic expression need not be
adjacent in syntactic structure. What we should require is that the combinatory properties of
words are respected, but this may be done in more than one way.
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
22
A classic example of the situation in which selection requirements can be met in more
than one way is that of Dutch synthetic compounds discussed in Booij (1988; 2002a). For
instance, the deverbal noun verlen-er ‘provider’ is derived from the verb verleen ‘to provide’.
This is an obligatorily transitive verb that requires a Patient argument to be expressed. This
argument can be either expressed in the non-head position of a compound (that is, in
morphology), or in a complement PP (that is, in syntax). What is impossible, however, is that
the deverbal noun occurs on its own, without (morphological or syntactic) complement since
in that case its combinatorial properties are not respected.:
(57)
[[[hulp]N[verlen]V]V er]N / [verlener van hulp]DP / *verlener
help provider
provider of help
provider
As this example shows, the actual order of the elements involved is determined by the
structure in which they co-occur: N V in morphology, V … N in syntax.
On a less abstract level of collocation, consider an idiomatic phrase such as ten uitvoer
leggen ‘lit to the execution put, to execute’. In this idiom, the first part ten uitvoer contains
the archaic word ten, an inflected preposition with the feature [definite]. This lexical
collocation can also appear in a deverbal noun in -ing: ten-uitvoer-legg-ing ‘execution’.
Hence, the requirement that the verb leggen with this particular meaning ‘to execute’ must
occur with ten uitvoer can also be met within the domain of a complex word. We do not have
to assume that there is a VP ten uitvoer leggen that is incorporated in his complex deverbal
noun (see Den Dikken, 2001: 153 for a similar argumentation). Another example is the lexical
collocation het bevel voeren ‘the command have, to be in command’. The coresponding
complex deverbal agent noun is bevel-hebb-er ‘commander’. In the syntactic collocation the
present of the neiter definite singular determiner het is required, but a determiner is
excluded from being embedded in a complex word.
A parallel situation obtains for particle verbs. As we saw above, particles can occur either
before or after the verb, depending on the syntactic structure in which they occur. Hence, the
lexical representation of particle verbs has to specify that these are verbs that require the
presence of a particle, without specifying the exact position of the particle. If such particle
verbs feed word formation, the order of particle and verb is determined by the morphology.
For instance, the morphological structure that can accommodate the combination of a
particle verb and the deverbal suffix –er in aanvaller ‘attacker’ derived from the particle verb
aan val ‘to attack’ is one in which the verb follows the particle because compounding in
Dutch is right-headed:
(58)
[[[aan]Part[val]V]V]er]N
This structural analysis presupposes the insight that verbal compounding is productive when
embedded. Therefore, we do not have to represent aanval as a phrasal category here. What is
crucial is the presence of the particle aan since without the presence of that particle the verb
val ‘fall’ will not receive the meaning ‘to attack’.
The idea that the actual order of elements in a lexical collocation is determined by
independently given structures, morphological or syntactic, is confirmed by the behaviour of
particle verbs in Swedish and Norwegian. These languages are like English in that the particle
always follows the verb. Unlike English, these particles have to be strictly adjacent to the verb.
However, when these particle verbs feed word formation, the particle appears in front of the
verb. That is, the morphology dictates a different word order. The verb must be on the right
since word formation is right-headed in Germanic languages, and deverbal suffixes like –er
attach to a verb. Thus, we have the following facts for Norwegian (Hans Olav Enger, pers.
communication) and Swedish (Ackema and Neeleman, 2002: 103):
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
(59)
(a)
(b)
(60)
(a)
(b)
Norwegian
kjøre opp
oppkjøring
oppkjørsel
kaste ut
utkastelse
‘lit. to drive up; try to pass the car-driving exam’
‘car-driving-exam’
‘driveway’
‘to throw out’
‘fig. throwing out’
Swedish
stiga upp
trubba av
upp-stig-ning
av-trubb-ning
‘to rise’
‘to blunt’
‘ascent (of an aeroplane)’
‘blunting’
23
These data show that the actual order of elements in a lexical collocation depends on the type
of structure in which they are inserted. This argument is confirmed by an observation by
Peter Svenonius referred to in Zeller (2001, 288-289): in the past participle form of Swedish
particle verbs the particle precedes the verb when the participle has verbal status, but follows
the verb when the participle is used as an adjective, and hence is a case of word formation. In
the pair ned-hogne / hogne ned ‘chopped down’ the first order favours a stative
interpretation of the participle, whereas the second favours an eventive reading.
This observation also has implications for the analysis of Dutch and German particle
verbs: we do not need the specific structure [Part V]V’- which is correct as one of the possible
structures of particle verbs - as the input form when particle verbs feed word formation. The
word order Particle – Verb follows from the requirements of the morphological system.
Hence, the role of Dutch and German particle verbs in word formation does not impede an
analysis of particle verbs as proposed in Müller (2002; 2006) in which they are represented
as verbs that require a particle to be present in their (syntactic or morphological)
environment.
6. Conclusions
In this article I have argued that Dutch has a construction of pseudo-incorporation in which a
single word of a lexcial category is combined with a verb to form a verbal predicate that has
phrasal status, but is different from a regular VP in that the incorporated word does not
project a phrase of its own. This construction does not have the unrestricted productivity that
is characteristic of regular syntax. In the case of incorporation of words of lexical categories,
the pseudo-incorporation structure invokes the interpretation of the activity as a habitual or
nameworthy one. Pseudo-incorporation is not restricted to nouns with a thematic role, but
extends to other classes of words such as non-argumental nouns, adverbs, and adjectives, and
occasionally even to verbs. This analysis makes correct predictions as to when the two words
involved can appear in non-adjacent positions, and as to the distribution of the negative
operators geen and niet, and accounts for the immobility of certain types of complex
predicates.
The analysis presented here is an illustration of what Dahl (2004) refers to as the growth
and maintenance of linguistic complexity”: “incorporating and quasi-incorporating patterns
arise by a gradual accretion of properties in subsets of members of syntactic constructions”
(Dahl 2004: 257). Linguistic systems become complicated because the different subsystems
tend to co-exist. For instance, pseudo-incorporation continues to exist and to function as a
productive mechanism beside compounding proper, and particle verb formation in Dutch is a
very productive system besides prefixation. Thus, the existence of pseudo-incorporation in
Dutch provides clear evidence in support of Dahl’s insight about linguistic complexity.
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
24
Notes
* I would like to thank Stefan Müller and Ariane van Santen for commenting on a previous draft of this
paper, and their willingness to share their insights with me.
1.
This is the type of incorporation referred to in Mithun (1984) as type I incorporation.
2. Cf. Dahl (2004, Chapter 10) for general discussion of pseudo-incorporation, which he refers to as
quasi-incorporation.
3. Similar combinations with bare nouns occur in Danish (Asudeh and Mikkelsen, 2000) and
Swedish (Dahl, 2004: 217). An example from Swedish is Vi har häst ‘lit. We have horse, we are horseowners’, in which sentence the bare noun häst ‘horse’ follows the verb, since Swedish is an SVO
language.
4. A similar threefold distinction between Ns that form part of lexical NV compounds, “sub-phrasal”
NV combinations, and regular phrasal combination of N V (where N forms a DP) is proposed for
Japanese in Iida and Sells (2007).
5. This type of pseudo-incorporation is also found in other Germanic dialects such as Lower-Saxon
and North-Frisian, see Booij (2004); Ebert (2000).
6. The non-neutrality of the plural marking in pseudo-incorporation has also been observed for
Hungarian (Farkas and de Swart, 2003) and Hindi (Dayal, 2007).
7. Similar variation is observed for German speakers by Vikner (2005).
References
ACKEMA, PETER. 1999. The non-uniform structure of Dutch N-V compounds. Yearbook of Morphology
1998, ed. by Geert Booij and Jaap van Marle, 127-58. Dordrecht: Kluwer.
ACKEMA, PETER and NEELEMAN, AD. 2002. Syntactic atomicity. The Journal of Comparative Germanic
Linguistics, 6.93-128.
ASUDEH, ASH and MIKKELSEN, LINE HOVE. 2000. Incorporation in Danish: Implications for interfaces.
Grammatical interfaces in HPSG, ed. by Ronnie Cann, Claire Grover and Philip
Miller, 1-15. Stanford: Stanford University.
BLOM, CORRIEN. 2005. Complex predicates in Dutch. Synchrony and diachrony. Utrecht: LOT.
BLOM, CORRIEN and BOOIJ, GEERT. 2003. The diachrony of complex predicates in Dutch: a case study
in grammaticalization. Acta Linguistica Hungarica, 50.61-91.
BOOIJ, GEERT. 1988. The relation between inheritance and argument structure: deverbal -er-nouns in
Dutch. Morphology and modularity. In honour of Henk Schultink, ed. by Martin Everaert, Arnold
Evers, Riny Huybregts and Mieke Trommelen, 57-74. Dordrecht: Foris Publications.
—. 1989. Complex verbs and the theory of level ordering. Yearbook of Morphology 1989, ed. by Geert
Booij and Jaap van Marle, 21-30. Dordrecht / Boston: Foris Publications.
—. 1990. The boundary between morphology and syntax: separable complex verbs in Dutch. Yearbook
of Morphology 1990, ed. by Geert Booij and Jaap van Marle, 45-63. Dordrecht: Foris.
—. 1993. Against split morphology. Yearbook of Morphology 1993, ed. by Geert Booij and Jaap van
Marle, 27-49. Dordrecht / Boston: Kluwer.
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
25
—. 1996. Inherent versus contextual inflection and the split morphology hypothesis. Yearbook of
Morphology 1995, ed. by Geert Booij and Jaap van Marle, 1-16. Dordrecht / Boston: Kluwer.
—. 2002a. The morphology of Dutch. Oxford: Oxford University Press.
—. 2002b. Separable complex verbs in Dutch: a case of periphrastic word formation. Verb-particle
explorations, ed. by Nicole Dehé, Ray Jackendoff, Andrew Macintyre and Silke Urban, 21-42. Berlin:
Mouton de Gruyter.
—. 2004. De aan het infinitief-constructie in het Nederlands. Taeldeman, man van de taal,
schatbewaarder van de taal, ed. by Georges de Schutter, Magda Devos and Jacques van Keymeulen,
97-106. Gent: Academia Press.
—. 2007. Construction morphology and the lexicon. Selected proceedings of the 5th Décembrettes:
Morphology in Toulouse, ed. by Fabio Montermini, Gilles Boyé and Nabil Hathout, 34-44. Somerville:
Cascadilla Press.
—. 2008. Constructional idioms as products of language change: the aan het + INFINITIVE
construction in Dutch. Construction grammar and language change, ed. by Alexander Bergs and
Gabriele Diewald. Berlin: Mouton de Gruyter.
CARLSON, GREG. 2006. The meaningful bounds of incorporation. Non-definiteness and plurality, ed. by
Svetlana Vogeleer and Liliane Tasmowski, 35-50. Amsterdam / Philadelphia: Benjamins.
DAHL, ÖSTEN. 2004. The growth and maintenance of linguistic complexity. Amsterdam / Philadelphia:
Benjamins.
DAYAL, VENEETA. 2007. Hindi Pseudo Incorporation. ms.
DE SWART, HENRIËTTE, WINTER, YOAD and
ZWARTS, JOOST. 2007. Bare nominals ansd reference to
capacities. Natural Language and Linguistic Theory, 25.195-222.
DEN DIKKEN, MARCEL. 2001. Review of Jochen Zeller (2001), Particle verbs and lexical domains. The
Journal of Comparative Germanic Linguistics, 4.145-69.
DYK, SIEBREN. 1990. Noun incorporation in Frisian. Leeuwarden: Fryske Akademy.
EBERT, KARIN H. 2000. Progressive markers in Germanic languages. Tense and aspect in the languages
of Europe, ed. by Östen Dahl, 605-53. Amsterdam / Philadelphia: Benjamins.
FARKAS, DONKA. 2006. The unmarked determiner. Non-definiteness and plurality, ed. by Svetlana
Vogeleer and Liliane Tasmowski, 81-105. Amsterdam / Philadelphia: Benjamins.
FARKAS, DONKA F. and DE SWART, HENRIËTTE. 2003. The semantics of incorporation. From argument
structure to discourse transparency. Stanford Calif.: CSLI Publications.
HAESERYN, WALTER, ROMYN, KIRSTEN, GEERTS, GUIDO, DE ROOIJ, JAAP and VAN DEN TOORN, MAARTEN.
1997. Algemene Nederlandse spraakkunst. Groningen / Deurne: Martinus Nijhoff / Wolters Plantyn.
IIDA, MASAYO and SELLS, PETER. 2007. Mismatches between morphology and syntax in Japanese
complex predicates. Lingua.
KIEFER, FERENC. 1992. Compounding in Hungarian. Rivista di Linguistica, 4.61-78.
KLAMER, MARIAN. 2001. Phrasal emotion predicates in three languages of Eastern Indonesia. Yearbook
of Morphology 2000, ed. by Geert Booij and Jaap van Marle, 97-122. Dordrecht: Kluwer.
KOOIJ, JAN and MOUS, MAARTEN. 2002. Incorporation: a comparison between Iraqw and Dutch.
Linguistics, 40.629-45.
GAGL 46 (2008)
Booij, Pseudo-incorporation in Dutch
26
KOOPMAN, HILDA. 1995. On verbs that fail to undergo Verb-Second. Linguistic Inquiry, 26.137-63.
MASSAM, DIANE. 2001. Pseudo noun incorporation in Niuean. Natural Language and Linguistic Theory,
19.153-97.
MINER, K. 1986. Noun stripping and loose incorporation in Zuni. International Journal of American
Linguistics, 52.242-54.
MITHUN, MARIANNE. 1984. The evolution of noun incorporation. Language, 60.847-94.
—. 2000. Incorporation. Morphologie / Morphology. Ein internationales Handbuch zur Flexion und
Wortbildung / An international handbook on inflection and word formation (in collaboration with
Wolfgang Kesselheim and Stavros Skopeteas) ed. by Geert Booij, Joachim Mugdan and Christian
Lehmann, 916-28. Berlin: De Gruyter.
MITHUN, MARIANNE and CORBETT, GREVILLE G. 1999. The effect of noun incorporation on argument
structure. Boundaries of morphology and syntax, ed. by Lunella Mereu, 49-71. Amsterdam /
Philadelphia: Benjamins.
MÜLLER, STEFAN. 2002. Complex predicates: Verbal complexes, resultative constructions and particle
verbs in German: Studies in constraint-based Lexicalism. Stanford: CSLI.
—. 2003. The morphology of German particle verbs: solving the bracketing paradox. Journal of
Linguistics, 39.275-325.
—. 2006. Phrasal or lexical constructions? Language, 82.850-83.
RADFORD, ANDREW. 2004. Minimalist syntax. Exploring the structure of English. Cambridge:
Cambridge University Press.
RIEHL, CLAUDIA MARIA and KILIAN-HATZ, CHRISTA. 2005. Structure and function of incorporation
processes in compounding. Studies in African linguistic typology, ed. by F. K. Erhard Voeltz, 361-76.
Amsterdam / Philadelphia: Benjamins.
SPENCER, ANDREW. 2005. Word-formation and syntax. Handbook of word-formation, ed. by Pavol
Stekauer and Rochelle Lieber, 73-97. Dordrecht: Kluwer.
VAN MARLE, JAAP. 2002. Dutch separable compound verbs: words rather than phrases? Particle verb
explorations, ed. by Nicole Dehé, Ray Jackendoff, Andrew MacIntyre and Silke Urban, 211-32. Berlin:
Mouton de Gruyter.
VIKNER, STEN. 2005. Immobile complex verbs in Germanic. Journal of Comparative Germanic
Linguistics, 8.83-115.
WEGGELAAR, C. 1986. Noun incorporation in Dutch. International Journal of American Linguistics,
52.301-05.
ZELLER, JOCHEN. 2001. Particle verbs and local domains. Amsterdam / Philadelphia: Benjamins.
author's address
University of Leiden, Faculty of Humanities
P.O. Box 9515
2300 RA Leiden
[email protected]
tel +31 71 5272509
fax +31 71 5272205