2003 MannerAdverbs

Manner adverbs and information structure:
Evidence from the adverbial modification of verbs

of creation1
Regine Eckardt
Abstract
The paper investigates the pragmatic effects of word order variation in German,
specifically concerning the ordering of adverb and nominal arguments. The behav-
iour of verbs of creation in combination with adverbs and indefinite object NPs
proves that the allegedly neutral word order NP Adverb Verb is not really presuppo-
sition free. I claim that indefinite object NPs to the left of manner adverbs receive a
special kind of partitive reading which presupposes that the referent of NP is al-
ready contextually present before the reported event takes place. This will offer an
explanation for the fact that in a sentence like Bob hat ein Flugzeug stabil kon-
struiert (‘Bob has a plane solidly constructed’) the object NP cannot receive an
existential (indefinite) reading while the sentence Bob hat ein Flugzeug kunstvoll
bemalt (‘Bob has a plane artistically painted’) seems to be able to refer to a previ-
ously unmentioned plane.
1. Introduction
The empirical observation on which this paper is based concerns a peculiar

gap in the range of possible adverbial modifications in German. It is well-
known that manner adverbs in German can precede or follow the direct
object.2 Moreover, it apparently does not matter whether the object is a
definite noun phrase or an indefinite noun phrase. The examples in (1) and
(2) illustrate this fact: All sentences in (1) and (2) are equally well formed
and the respective (a) and (b) variants seem to convey the same truth condi-
tional contents.
(1) a. (daß) Hans vorsichtig eine Nuss öffnete

(that) Hans carefully a nut opened
b. (daß) Hans eine Nuss vorsichtig öffnete
(that) Hans a nut carefully opened
© Mouton de Gruyter 2003

E. Lang, C. Maienborn & C. Fabricius-Hansen (eds.): Modifying Adjuncts
(Interface Explorations 4), 261-305
262 Regine Eckardt
(2) a. (daß) Hans langsam die Tür öffnete

(that) Hans slowly the door opened
b. (daß) Hans die Tür langsam öffnete
(that) Hans the door slowly opened
Furthermore, there is a class of result-oriented adverbs (see note 2) that

specify the outcome of a certain event. These adverbs are restricted to posi-
tions after the direct object, and are acceptable with both indefinite and
definite object NPs. This is exemplified in (3) to (5).
(3) *(daß) Hans schwer einen/den Wagen belud

(that) Hans heavily a/the carriage loaded
(4) (daß) Hans den Wagen schwer belud

(that) Hans the carriage heavily loaded
(5) (daß) Hans einen Wagen schwer belud

(that) Hans a carriage heavily loaded
Adverbs of both type can also be applied to verbs of creation and coming
into existence. However, it turns out that if these verbs are combined with
an indefinite object (as in (6)), then manner adverbs are restricted to the
position preceding the object NP and, dramatically, result-oriented adverbs
become unacceptable altogether, as shown by (10), (11).
Manner adverbs
(6) (daß) Hans geschickt eine Flöte schnitzte
(that) Hans skilfully a flute carved
(7) (daß) Hans die Flöte geschickt schnitzte

(that) Hans the flute skilfully carved
(8) *(daß) Hans eine Flöte geschickt schnitzte

(that) Hans a flute skilfully carved
Result-oriented adverbs
(9) (daß) Hans den Pullover locker strickte
(that) Hans the pullover loosely knitted
(10) *(daß) Hans einen Pullover locker strickte

(that) Hans a pullover loosely knitted
Manner adverbs and information structure 263
(11) *(daß) Hans locker einen Pullover strickte

(that) Hans loosely a pullover knitted
The basic question then is, why are sentences like (8), (10) and (11) bad
while (7), (9) and (1b), (5) are acceptable?
I will start by reviewing some basic properties of manner and result-
oriented adverbs (Section 2) and verbs of creation (Section 3). In Section 4,
a wider and more systematic overview of the data will be given. In Sections
5 and 6, I will develop an account for the data in question that is based on
the semantic and pragmatic properties of the sentences in question. It will be
based on the assumption that object NPs that precede adverbs are topical in
the sense of Jäger (1996b). While being topical is a natural state for a defi-
nite NP, Jäger showed that indefinite NPs in topical positions can only re-
ceive one of several non-straightforward interpretations that differ from the
ordinary, presupposition-free purely existential interpretation of indefinites.
I will demonstrate that topical indefinites will lead to almost context-neutral
readings except with verbs of creation and coming into existence, where the
temporal properties of topics and created objects are in conflict.
A brief summary and discussion of open questions will conclude the
paper.
2. Adverbs
2.1. Manner adverbs
I will use the term “manner adverbs” to cover the class of adverbial modifi-
ers that specify the manner in which a certain event was performed. This
manner may come about due to a certain mood of the agent (wütend ‘an-
grily’, schüchtern ‘shyly’, fröhlich ‘gaily’), due to his/her intentions (vorsi-
chtig/aufmerksam ‘cautiously’/‘attentively’, schlampig ‘sloppily’, freiwillig
‘willingly’, ‘voluntarily’, absichtlich ‘intentionally’), be a mere matter of
speed (schnell ‘quickly’, langsam ‘slowly’), be a mixture of these (hastig
‘hastily’, ruhig ‘calmly’) or an indication of his/her dispositions and quali-
ties (fachmännisch ‘expertly’, intelligent ‘intelligently’). I will hence not
distinguish between “subject-oriented” adverbs that are related to a property
of an individual and purely event-related adverbs.3
Syntactically speaking, in German, these adverbs can occur in any posi-
tion after the subject NP and before the minimal verbal domain. The possi-
ble distributions are illustrated in (12) to (17).
264 Regine Eckardt
(12) (daß) ein Kind wütend/schnell/fachmännisch einen Stock absägte

(that) a child angrily/quickly/expertly a stick offsawed
(13) (daß) das Kind schüchtern dem Nikolaus eine Nuss überreichte
(that) the child shyly to-the St.Claus a nut handed-over
(14) (daß) das Kind dem Nikolaus schüchtern eine Nuss überreichte
(that) the child to-the St.Claus shyly a nut handed-over
(15) (daß) das Kind dem Nikolaus eine Nuss schüchtern überreichte
(that) the child to-the St.Claus a nut shyly handed-over
(16) (daß) ein Kind einen Stock schnell/wütend/vorsichtig/... absägte

(that) a child a stick quickly/angrily/carefully/ ... off-sawed
(17) *(daß) vorsichtig/wütend/... ein Kind einen Stock absägte

(that) carefully/angrily/ ... a child a stick off-sawed
Some of the listed adverbs can precede the subject, but then they receive a
different interpretation. Temporal adverbs in this position are usually inter-
preted not with respect to the speed or time of the event but rather they
measure the time that elapsed between some previous reference time point
and the beginning of the event:
(18) (daß) schnell ein Hase übers Feld hoppelte

(that) quickly a hare over-the field hopped
= the hare came soon after the reference time and may have hopped
slowly
(19) (daß) ein Hase schnell übers Feld hoppelte

(that) a hare quickly over-the field hopped
= it took the hare little time to cross the field
Some subject-oriented adverbs, typically those related to intentions, can

receive an interpretation where it is not the subject but some background
individual who shows the intention. An example is given in (20)/(21).
(20) (daß) ein Polizist absichtlich vor dem Eingang stand

(that) a policeman intentionally in-front-of the entrance stood
= it was the policeman’s intention to stand in front of the entrance
(21) (daß) absichtlich ein Polizist vor dem Eingang stand

(that) intentionally a policeman in-front-of the entrance stood
= it was someone else’s intention to place a policeman in front of the
entrance
Although it would be an attractive task to study the relation between pre-

subjectival and post-subjectival readings in these cases, I will for now sim-
ply ignore the pre-subjectival case.
Moreover, the statement about possible distributions of manner adverbs
is, for the time being, more an existential than a universal statement: For
each manner adverb, there are at least some verbs plus arguments that allow
the adverb between subject NP and object NP, and some verbs plus argu-
ments that allow the adverb between or after the object NPs. There may be
sentences where these possibilities are restricted; and there is a general,
well-known tendency for definite NPs to precede the adverb. In Sections 5
and 6, we will see that these effects are due to the different informational
status of the object NP before and after the adverb. This diagnosis will ex-
plain the varying behaviour of definites, among other things, because some
definites (the pope) are better out-of-the-blue than others (the man).
A notorious question in the literature on adverbs is the question of base
positions vs. derived positions. The present overview of the data (which is
very sparse in that respect, see the very detailed discussion in Frey and Pitt-
ner (1998), also Frey (this volume)) does not commit us to any answer with
respect to this question. I will take up the question briefly in Section 7 with-
out, however, attempting a final solution.
2.2. Result-oriented adverbs
Let me now turn to adverbs that specify the result of a certain event. Sen-
tence (22) illustrates the case.
(22) Miriam dressed elegantly.
Sentence (22) has a reading that states that the result, rather than the process
of dressing, is elegant – i.e. blouse and trousers match, are of a certain fash-
ion etc. Intuitively, this kind of state of affairs differs from the manner read-
ing of (22). In the manner reading, Miriam will have to make elegant moves
in dressing no matter how shabby the attire will look afterwards. I will now
recapitulate arguments why this intuitive difference must be mirrored by
two distinct semantic variants of adverbs like elegantly.
266 Regine Eckardt
Taking the Davidsonian approach to manner adverbial semantics (as I will

do), one might feel tempted to claim that these subtleties have no place in
logical semantics proper. We could assume that elegantly in (22) above is
simply a predicate of events λe.elegant(e) and it is an idiosyncratic case-to-
case question what aspect of an event should give rise to calling it elegant.
If you find this too coarse, remember that the same is common practice in
the semantic representation of adjectives like red where most semanticians
would also refrain from giving an exhaustive description as to when red N
means “the whole surface of N is red” (= red apples), “the entirity of N is
red” (= red cherries; note that white cherries are white only inside), “the
inside of N is red” (= red grapefruit) or “a specific part of N is red” (like in
the red tap is for hot water).
However, this strategy will fail for result-oriented adverbs. We find ad-
verbs that are ambiguous between a manner-oriented interpretation and a
result-oriented interpretation. Elegantly in English (and elegant in German
as well) is one of these. Sentence (22) can mean either that the resulting
attire is elegant or that Miriam’s movements and manner in dressing were
elegant (in spite of a shabby result). This by itself would still not show the
need for a finer semantic distinction of either reading: A red cherry can be
one which is red inside and outside – in contrast to white cherries that are
light red on the outside and whitish inside. Or it can be red on the outside no
matter what the inside looks like, for instance if contrasted with green cher-
ries that are still unripe. In both cases, we might claim that we master some
conceptual core notion (of redness, of elegance) and can leave it to the con-
text to decide which aspects or parts of an object or event should show the
respective property.
Yet this simple solution will predict that all readings are available if and
only if they get appropriate contextual support. Empirically, this is not the
case. Sentence (23) exhibits only the manner interpretation of elegantly, and
the German pair in (24a) and (24b) exhibits the same pattern.
(23) Miriam elegantly dressed. (= only manner reading)
(24) a. Sarah belud unsicher den Wagen mit Holz.

Sarah loaded uncertainly the cart with wood
= Sarah loaded the cart with wood in a manner that showed that
she was not certain whether or how this was to be done. (The load
itself might have been stable.)
b. Sarah belud den Wagen unsicher mit Holz.

Sarah loaded the cart uncertainly with wood
reading 1: as in (a)
reading 2: Sarah loaded the cart with wood in such a way that the
load was insecurely fixed on the cart afterwards. (Sarah herself
was quite confident about her action.)
We observe that adverbs of result modification in German are restricted to a

position directly preceding the verbal complex (in verb-final clauses) and
following the direct object NP. Adverbs that are ambiguous between man-
ner and result readings only show result modification in this position
whereas purely result-oriented adverbs are confined to this position in gen-
eral. This is exemplified in (24c), adapted from Geuder (2000):
(24) c. Sarah hat den Wagen schwer beladen.

Sarah has the cart heavily loaded
*Sarah hat schwer den Wagen beladen.
Sarah has heavily the cart loaded
Geuder (2000) investigates the necessary semantic analysis of result modifi-

cation in full detail. He argues that in order to maintain a coherent concep-
tual core of the respective adverbs (as well as their adjectival counterparts),
it is appropriate to assume that the relevant events (of loading, dressing,
decorating, etc. ) not only terminate in some unspecified “resultant state”
but also produce a more specific concrete “resultant object”. The resultant
state might also be subject to temporal modification (as in Sarah loaded the
cart with wood for 2 hours which implies that the wood remained on the
cart for 2 hours). The resultant object is denoted by nouns like the load, the
attire, or the decoration. If we assume that result-oriented adverbs predicate
– in one way or another – over these resultant objects, we can nicely capture
the fact that, unlike true manner modification, a sentence like (25) in its
result-oriented reading is not paraphrased by (25a) (which corresponds to
the true manner case) but by (25b).
(25) Olivia decorated the christmas tree elegantly.

a. The decorating of the christmas tree was conducted elegantly.
b. The decoration made the tree look elegant.
(26) Olivia decorated the tree for three hours.

a. The decoration remained on the tree for three hours.
b. *The decoration of the tree took three hours.
268 Regine Eckardt
(26b) shows that the nominal decoration does not denote an object that
would lend itself to temporal modification. Temporal modification of an
abstract resultant state is possible as in (26) under the paraphrase (a), which
offers further reason to distinguish between resultant state and resultant
object.4 For a more detailed discussion of result-oriented adverbs, I refer the
reader to Geuder (2000) and the sources quoted therein. In the remainder of
the paper, we will draw on the following aspects of result-oriented adverbs:
(i) Result-oriented adverbs require a semantic analysis distinct from

manner adverbs.
(ii) Result-oriented adverbs predicate over the resultant object of an event
(and are hence restricted to verbs that denote events with some such
object).
(iii) Result-oriented adverbs are restricted to positions to the right of the
direct object NP (“directly before the verb” or before the verbal com-
plex in terms of Frey and Pittner (1998)).
I will also use the terms resultative adverbs or result modifiers instead of
“result-oriented adverbs”.
3. Verbs of creation and coming into existence
So far, the result modifications that we have seen were predicated of results
as performed on some given object (the woman dressed, the tree decorated,
etc.). Verbs of creation and coming into existence – which I will abbreviate
as verbsCC – denote events which result in the creation of a new object
rather than in the modification of a given one. The (real) object is usually
denoted by the (grammatical) direct object in verbs of creation, and by the
subject in verbs of coming into existence. In (27) to (33), I list some exam-
ples where the NP denoting the created object is given in boldface.
(27) a. Eine Beule entstand.

A bump emerged
b. Ein Stau entstand.
A traffic jam built-up
c. Ein Ausschlag bildete sich.
A rash formed
(28) Die Verhandlungen führten zu einem Vertrag.

The negotiations led to a contract
(29) a. Andrea konstruierte eine Maschine.

Andrea constructed a machine
b. Andrea komponierte ein Lied.
Andrea composed a song
(30) a. Andrea hatte eine Idee.

Andrea had an idea
b. Dem Andrea kam eine Idee.
To-the Andrea came an idea
(31) Andrea schuf eine Statue.

Andrea created a statue
(32) Andrea setzte eine Unterschrift unter den Vertrag.

Andrea put a signature under the contract
(33) a. Andrea gab eine Antwort.

Andrea gave an answer
b. Andrea stellte eine Frage.
Andrea posed a question
It has been a matter of some debate what an appropriate semantic represen-

tation of verbsCC should look like. Following Dowty (1979), Parsons (1990)
and Zucchi (1999) pursued the idea that a verb like bauen ‘build’ should
relate the creation event to an agent and an effected object in the same way
as other transitive verbs like eat or love do. This strategy forces one to claim
that even the smallest initial parts of the created object (e.g. the first brick of
a future house) and the vague plan for such an object have to be in the ex-
tension of the respective nouns (e.g. house). In other words: Our ontology
would be crowded with virtual and partial houses, machines, songs, novels,
and so forth.
Von Stechow (2001) argues convincingly against such an approach. He
pursues the idea that a sentence like (34) states that the agent does some-
thing to cause the proposition in (35) to become true in the future. Sentence
(35) is not true while the event of creation is taking place.
(34) Sabrina baut eine Maschine.

Sabrina is-building a machine
(35) Es gibt eine Maschine, die von Sabrina gebaut wurde.

‘There is a machine that was built by Sabrina.’
270 Regine Eckardt
He points out that the relation between (34) and (35) is similar to the case of
intentionally opaque verbs like seek or owe. If the agent seeks a mushroom,
for example, she undertakes something to make the proposition “agent finds
mushroom” true. Examples (36)/(37) parallel (34)/(35).
(36) Sabrina sucht einen Pilz.

Sabrina is seeking a mushroom
(37) Es gibt einen Pilz, den Sabrina gefunden hat.

‘There is a mushroom that Sabrina has found.’
In order to stress this similarity, von Stechow (2001) classifies verbsCC as

“temporally opaque verbs”. He proposes to analyze a sentence like (38)
along the paraphrase given in (39).
(38) Andrea baute eine Maschine.

Andrea built a machine
(39) Andrea is the agent of a building event e such that during the time of
e, the proposition “there is a machine y at time t, and the occurrence
of e causes y to exist throughout time t” turns from false to true.
This analysis will imply the existence of a machine of the appropriate kind
after the event of creation has been completed, but without stating that any
kind of machine-entity is existent before or while the event takes place. In
particular, it is explicitly stated that no machine-created-in-e exists prior to
the event in question. While I have to refer the reader to the original paper
for the formal implementation of this idea, and the intricate arguments that
show why this semantic treatment is superior to one based on virtual ob-
jects, I will adopt this view in the subsequent discussion.
Of course, an analysis of verbsCC in terms of temporal opacity does not
exclude reference to plans in general. In (40a), the nouns one of the cakes
and the cakes will obviously refer to the plans (recipes) in the cookbook,
and a cake in (40b) can be understood in that way as well.
(40) a. After having consulted my cookbook for some time, I baked one of
the cakes.
b. After having consulted my cookbook, I decided to bake a cake
(rather than waffles or biscuits).
In the following, I will sometimes carefully distinguish between reference

to plans and reference to objects. While the option of reference to plans
may be available in some cases, it is not, and need not be, in general. We
find examples like (41) where any reference to previously existing intended
objects or plans seems extremely unlikely.
(41) a. Hans hatte eine Idee.

‘Hans had an idea.’
b. Andrea kritzelte einen Krakel aufs Papier.
‘Andrea scratched a scribble onto the paper.’
c. Beate improvisierte ein Lied.
‘Beate improvised a song.’
The NP eine Unterschrift (‘a signature’) in (42) also will refer to the created
object, and not to a plan, under the most common reading of the sentence.
(42) Cecile setzte eine Unterschrift unter den Vertrag.

Cecile put a signature under the contract
In this case, that is not so much due to the fact that there were no signature-
schemes: In fact there are. It is what is stored at your bank to make sure that
the checks signed by yourself can be distinguished from those faked by
others. Yet if we understood (42) such that the indefinite eine Unterschrift
referred to such plans, the sentence would suggest that Cecile could choose
amongst several such schemes when signing something. World knowledge
will tell us that this is not the case – there is only one signature plan that is
relevant for Cecile in all legal contexts.
4. Possible combinations
I will now review the result modification data for verbs of creation in detail.
Bearing in mind the survey of possible adverbial modifications in Section 2,
we will be better able to appreciate the crucial gap of impossible modifica-
tions.
We are concerned with (un)acceptable sentences that contain indefinite
NPs. It is well-known that indefinites have a wide range of possible inter-
pretations and that it is virtually always possible to understand any sentence
with some one of these possible readings. Usually, two to three readings for
indefinite NPs are distinguished in the literature. These are the following,
exemplified for the noun Hund ‘dog’:
272 Regine Eckardt
English German paraphrase reading

dogs, Hunde “every dog” generic
a dog ein Hund
(one dog) ein Hund “one of several known dogs” partitive
a dog ein Hund “some, possibly first, dog(s) out existential
dogs Hunde of the blue”
Generic and partitive readings are available in the same positions and con-
structions and, in the case that an indefinite does not allow all three readings
in some construction, we will always find either the existential reading, or
both the generic and the partitive reading. The distribution of readings as
well as possible driving forces for these interpretations have received wide
attention in the nineties, starting with the work of Kratzer (1989/1995) and
Diesing (1990, 1992), but also de Hoop (1992, 2000), Reinhart (2000), La-
dusaw (1994), Eckardt (1996) and Jäger (1996b). The present work will
mainly rely on the last source, offering further evidence in favour of the
more differentiated four-way distinction advocated by Jäger. We will come
back to this issue in Section 5. In the present section, we will still rely on
the simpler distinction between generic, partitive and existential interpreta-
tion of indefinite NPs.
Let us start with examples with unambiguous manner adverbs that mod-
ify a verb of creation with definite/indefinite object and that are positioned
to the right/left of the object NP. The full range of possible cases is given in
(43) to (46).
(43) Beate baute aufgeregt einen Drachen.

Beate built excitedly a kite
(44) Beate baute aufgeregt den Drachen.

Beate built excitedly the kite
(45) #Beate baute einen Drachen aufgeregt.

Beate built a kite excitedly
(46) Beate baute den Drachen aufgeregt.

Beate built the kite excitedly
Sentence (43) is most straightforward with the indefinite einen Drachen

understood existentially, and conveys the evident proposition that Beate,
excitedly, built a kite. Sentence (44) possibly requires a certain kind of con-
text that allows the definite den Drachen to occur post-adverbially, but the
effects are not very strong. The information conveyed is that Beate built the
– somehow known – kite, excitedly.5 Sentence (46) means the same as (44)
and they differ only in the kind of contexts in which one would rather use
one or the other. Sentence (45) forces the indefinite NP to receive either a
generic interpretation or a partitive interpretation.
(45’) generic: Usually when Beate built a kite, she did it excitedly.
(45’’) partitive: One of the kites she built, Beate built excitedly.
The existential interpretation of the indefinite is not accessible. In the re-

mainder of the paper, I will use ‘#’ to mark this constellation while un-
starred sentences with an indefinite mean “sentence acceptable with an exis-
tential interpretation of the indefinite”. The following sentences illustrate
the pattern further.
(47) a. Daphne entwarf wütend eine/ die Hose.

Daphne designed angrily a pair of/the trousers
b. Daphne entwarf die Hose wütend.
Daphne designed the trousers angrily
c. #Daphne entwarf eine Hose wütend.
Daphne designed a (pair of) trousers angrily
(48) a. Euridike konstruierte vorsichtig ein/das gleichschenklige(s)

Euridike constructed carefully an/ the isosceles
Dreieck.
triangle
b. Euridike konstruierte das gleichschenklige Dreieck vorsichtig.
Euridike constructed the isosceles triangle carefully
c. #Euridike konstruierte ein gleichschenkliges Dreieck vorsichtig.
Euridike constructed an isosceles triangle carefully
Is there anything unusual about this pattern? If all verbs behaved like verbs
of creation then the answer would clearly be “no”. In that case, adverbials
would simply be classed as indicator of the “VP boundary” in the sense of
Diesing (1992). The most famous such indicator is the particle ja doch. The
examples in (49) illustrate the general pattern: Indefinites to the right of ja
doch are preferably understood existentially, indefinites to the left are parti-
tive or generic.
(49) a. (daß) Daphne eine Hose ja doch flickte (gen/part, no exist.)

(that) Daphne a (pair of) trousers after all mended
b. (daß) Daphne ja doch eine Hose flickte (exist., gen/part possible)
(that) Daphne indeed a (pair of) trousers mended
274 Regine Eckardt
Adverbs, however, do not behave like ja doch. For instance, they do not
seem to mark a VP-boundary if they combine with verbs like flicken ‘mend’
or other verbs where an object is affected or destroyed. The following (b)
examples allow for an existential reading of the indefinite object although it
precedes the adverb.
(50) a. (daß) Daphne vorsichtig eine Hose flickte (exist. reading)

(that) Daphne carefully a pair of trousers mended
b. (daß) Daphne eine Hose vorsichtig flickte (exist. reading)
(that) Daphne a pair of trousers carefully mended
(51) a. (daß) Daphne wütend eine Hose zerriß (exist. reading)

(that) Daphne angrily a pair of trousers tore
b. (daß) Daphne eine Hose wütend zerriß (exist. reading)
(that) Daphne a pair of trousers angrily tore
If we derived the presence or absence of existential readings of indefinites

solely from their position inside or outside VP (Diesing’s mapping hypothe-
sis) then we would make the unorthodox prediction that a manner adverb
like carefully marks the VP boundary if combined with verbs of creation but
it does not if combined with any other kind of eventive verb. While it would
perhaps not be impossible to elaborate such a position, I will advocate a
different line of explanation in Section 6.
Let us now turn to result-oriented adverbs. From Section 2.2. we know
that they never occur to the left of an object NP. The picture for verbs of
creation is consistent insofar as we find that result-oriented adverbs cannot
modify a verb of creation plus an indefinite object NP. Sentences (52)–(53)
illustrate this gap.
(52) result: solide ‘solidly’

a. Beate baute den Drachen solide.
Beate built the kite solidly
b. #Beate baute einen Drachen solide.
Beate built a kite solidly
c. *Beate baute solide einen/den Drachen.
Beate built solidly a / the kite
(53) result: ausdrucksvoll ‘expressively’6

a. (daß) Nick den Hund ausdrucksvoll knetete
(that) Nick the dog expressively modelled
b. #(daß) Nick einen Hund ausdrucksvoll knetete
(that) Nick a dog expressively modelled
c. *(daß) Nick ausdrucksvoll einen/den Hund knetete

(that) Nick expressively a / the dog modelled
(manner reading, as far as one can model a figurine in an
expressive way)
Once again, the indefinite NP before the result modifying adverb must re-
ceive a partitive or generic interpretation (which yields rather odd asser-
tions). Some readers might find sentences like (52b) or (53b) acceptable in
an existential sense if ein Drachen or ein Hund refer to plans (e.g. in a do-it-
yourself book). Similarly, a discourse like (54) is acceptable if we under-
stand ein Rührkuchen to refer to the recipe in the book and not to the real
cake that resulted from Auguste’s baking (and we will see presently why the
cookbook with all its recipes is necessary for the sentence to work):
(54) Auguste bekam Besuch von Hans, der allergisch auf Milch war. Zu-
erst hat sie lange im Kochbuch geblättert.
‘Auguste expected the visit of Hans who was allergic to milk. First she read
the cookbook for a long time.’
Dann hat sie einen Rührkuchen milchfrei gebacken.
Then has she a sponge cake milkfree baked
Two things will turn out to be crucial for (54) to be acceptable: Firstly, the
plan for the sponge cake existed prior to the baking, and secondly, in the
given context the sponge cake plan is one in a set of already implicitly men-
tioned plans for cakes in the cookbook.
While reference to plans is easily available for things to be built, knitted,
kneaded, cooked or baked according to a given scheme, the respective NP
will almost certainly refer to the created object when we report the new
invention of the design or plan itself, like the design of a new costume or a
new car. Consequently, the following examples are clearly “#” according to
my intuition, and do not allow for a rescue in terms of reference to plans
like (54).
(55) a. (daß) Karl das Kostüm hauteng entwarf

(that) Karl the costume tightly designed
b. #(daß) Karl ein Kostüm hauteng entwarf
(that) Karl a costume tightly designed
c. *(daß) Karl hauteng ein/das Kostüm entwarf
(that) Karl tightly a / the costume designed
d. (daß) Karl ein hautenges Kostüm entwarf
(that) Karl a tight costume designed
276 Regine Eckardt
(56) a. (daß) Opel das Auto sparsam konstruierte

(that) Opel the car economically constructed
b. #(daß) Opel ein Auto sparsam konstruierte
(that) Opel a car economically constructed
c. *(daß) Opel sparsam ein/das Auto konstruierte
(that) Opel economically a / the car constructed
d. (daß) Opel ein sparsames Auto konstruierte
(that) Opel an economic car constructed
Similarly, (57) is resistant to a plan-denoting reading of eine Unterschrift.
(57) a. #Auguste setzte eine Unterschrift vorsichtig unter den Vertrag.

Auguste put a signature carefully under the contract
b. Auguste setzte die Unterschrift vorsichtig unter den Vertrag.
Auguste put the signature carefully under the contract
c. Auguste setzte vorsichtig die/eine Unterschrift unter den Vertrag.
Auguste put carefully the/a signature under the contract
Some verbs of creation do not find suitable result modifying adverbs, espe-
cially intransitive verbs. Yet they show very similar effects with suitable
temporal adverbs. Examples (58) to (60) show the behaviour of the verbCC
eine Idee haben ‘to have an idea’ under the adverb plötzlich ‘suddenly’. The
(b) variants with the verb kaufen ‘buy’, which is not a verbCC, are there to
show that the peculiar effects are once more not the general behaviour of the
adverb plötzlich but are restricted to the combination of plötzlich with
verbsCC.
(58) a. Sabine hat ganz plötzlich eine Idee gehabt.

Sabine has all suddenly an idea had
b. Sabine hat ganz plötzlich einen Hund gekauft.
Sabine has all suddenly a dog bought
(59) a. #Sabine hat eine Idee ganz plötzlich gehabt.

Sabine has an idea all suddenly had
b. Sabine hat einen Hund ganz plötzlich gekauft.
Sabine has a dog all suddenly bought
(60) a. Sabine hat die Idee ganz plötzlich gehabt.

Sabine has the idea all suddenly had
b. Sabine hat den Hund ganz plötzlich gekauft.
Sabine has the dog all suddenly bought
Sentence (59b) shows that the temporal adverbial plötzlich cannot be the
benchmark between existential and non-existential indefinites in general, as
the sentence can well report that Sabine bought a dog (suddenly) without
any other dogs being mentioned in the context (which would be the partitive
reading). The intransitive case is illustrated in (61).
(61) a. #(daß) ein Stau ganz plötzlich entstand

(that) a traffic-jam all suddenly built up
b. (daß) ganz plötzlich ein Stau entstand
(that) all suddenly a traffic-jam built up
c. (daß) der Stau ganz plötzlich entstand
(that) the traffic-jam all suddenly built up
Example (62) offers a reflexive case, this time with a suitable result-
oriented adverb. (We will ignore the marginal reading of (62c) “curiously, a
crystal built up”. This evaluative reading of “curiously” in English must be
translated to the German kurioserweise.)
(62) a. (daß) der Kristall sich kurios ausbildete

(that) the crystal itself curiously built up
b. #(daß) ein Kristall sich kurios ausbildete
(that) a crystal itself curiously built up
c. *(daß) sich kurios ein/der Kristall ausbildete
(that) itself curiously a / the crystal built up
Let me finally briefly summarize the range of data we have seen.
– If a transitive verbCC is combined with a manner adverb, the adverb has

to precede any existential indefinite object NP. Otherwise, the indefinite
will have to be understood partitively or generically.
– Hence, no transitive verbCC can be combined with a result-oriented ad-
verb and an indefinite object NP in existential reading.
– Some cases can be rescued if the indefinite NP is interpreted as referring
to a plan rather than an object. (The specific contexts that license these
readings will be reviewed presently.)
– If an intransitive verbCC is combined with an adverb, the adverb has to
precede any existential indefinite subject NP.
– Conjecture: if there were result-oriented adverbs for such verbs, they
could not be combined with an indefinite subject NP.
– None of the adverbs involved shows similar effects with event-denoting
verbs other than those of creation and coming into existence.
278 Regine Eckardt
Which leaves us with the puzzling question: Why does the combination of
adverbs and verbsCC differ from adverbial modification of other verbs?
5. Subtle topics
Jäger (1996b) proposes an extension of classical dynamic semantic frame-

works in order to capture various ways of “being topical” or “definite” in a
unified way. It would be beyond the limits and aims of the present paper to
recapitulate Jäger’s full formal system, but conveniently he proposed a ter-
minology inspired by Kamp’s Discourse Representation Theory. In this
section, I will recapitulate some of Jäger’s core ideas.7
Most importantly, Jäger distinguishes between discourse referents d1, d2,
... and “Understood Entities” pi, pj, ... (Jäger calls these “pegs” but Gro-
enendijk and Stokhof’s pegs have found quite different tasks to serve and I
hence propose this terminology to stress the originality of Jäger’s notion).
Every discourse referent automatically gives rise to an understood entity,
but we may have more understood entities than discourse referents. (Dis-
course referents also may “die” when becoming unsalient.) Discourse refer-
ents, as in Discourse Representation Theory, keep the record of individuals
that have been explicitly mentioned in a discourse. They, and only they, are
accessible antecedents for anaphoric (singular) pronouns.
Understood Entities, in contrast, are such individuals or objects that can
be inferred by the intelligent hearer without necessarily being mentioned.
The boldface definites in examples (63) and (64) pick up such Understood
Entities while the (unacceptable) pronouns show that they do not have the
status of discourse referents.
(63) Sue was stabbed. The criminal / *He had worn a black leather
jacket.
(64) Joe bought a new car. The horn / *It plays the beginning of
“Clementine”.
Yet we can use definites that do not even refer to such Understood Entities.
Notorious examples are the sun or the pope which can enter any discourse at
any point and refer to their referents much in the way of proper names.
Jäger calls these definites “referential” and claims that they, and only they
(a) can occur in thetic sentences

(b) must have a unique referent out there in the world.
Definites that refer to Understood Entities, on the other hand, have topical
status8 and need only be unique among the set of Understood Entities, not in
the world: Sentence (63), for instance, does not rely on the fact that there is
but one criminal in the world. The use of Understood Entities allows Jäger
to propose and model the following distinction:9
Kind of NP Status Reference

pronoun topical to discourse referent, need not
be unique
definite NP topical to Understood Entity, must be
(in categorical judgements) unique among the UEs
definite NP non-topical introduces new DR and UE,
(in thetic judgements) must be unique in the world
Interestingly, Jäger extends his approach to the discussion of topical and

non-topical indefinites, and he offers a truer picture of the full range of
readings of indefinite NPs in all kinds of constructions than can be found
elsewhere in the literature. He takes subjects of individual-level predicates
to be the test case of a topical NP10 and diagnoses the following range of
readings for German topical indefinite NPs in that position (following
common practice, I will use be blue eyed, be intelligent as prototype in-
stances of individual-level predicates). The words in boldface are meant to
bear main stress.
(65) generic reading:

a. Ein Schwede ist blauäugig.
A Swede is blue-eyed
b. Schweden sind blauäugig.
Swedes are blue-eyed
(66) partitive reading:

a. Ein Schwede ist blauäugig.
One Swede is blue-eyed
b. Zwei Schweden sind blauäugig.
Two Swedes are blue-eyed
c. Einige Schweden sind blauäugig.
Some Swedes are blue-eyed
d. Kein Schwede ist blauäugig.
No Swede is blue-eyed
280 Regine Eckardt
(67) in-group reading: Being one of a known set of people

(Die ganze Halle stand voller Menschen aller Nationen mit den
untypischsten Augenfarben.)
(‘The whole hall was filled with people of all nations with the most un-
typical eye colours.’)
a. Ein Schwede war schwarzäugig.
A Swede was black-eyed
b. Ein Schwarzafrikaner war blauäugig.
A black African was blue-eyed
...
The three readings differ characteristically in their intonation pattern. Sen-

tence (65) reports something about the typical Swede. The subject is left
unstressed. It is not entirely clear to what extent such sentences suggest that
one has been talking about Swedes all along, but there is general agreement
that the referent of indefinites in a generic interpretation can be “old” mate-
rial – if only by the fact that “the Swede” as a kind is obviously part of our
shared world knowledge.
If the noun is left unstressed but the determiner is stressed, as in (66), we
will understand the partitive (or “truly quantificational”) interpretation of
the NPs. The sentences in (66) presuppose a set of known Ns (Swedes in
our example), say something about some of these and strongly suggest that
there are still other Ns or at least that there might be.
We have already seen the last two readings in Section 4. The inter-
pretation of the indefinite in (67), however, has received little attention in
languages like German and English. Turkish marks this reading more point-
edly, and it has hence evoked more attention in the literature, e.g. in Enç
(1991). Moreover, it has recently been studied in language acquisition under
the label “source-set reading”, cf. Krämer (2000). At first glance, the read-
ing may seem like a purely existential reading. The sentence Ein Schwede
war schwarzäugig in the context of (67) does not presuppose the existence
of more Swedes, and neither does the sentence require a context where
Swede Discourse Referents or Understood Entities are available. Neverthe-
less, it needs a context in which a set of persons is explicitly or implicitly
known, and the Swede is one of these. In other words, the interpretations in
(67) are not available in a deserted discourse universe. I will call this inter-
pretation of ein N the in-group interpretation.11
The sentences in (67) bear a second stress on the adjective. This brings
them close to the hat contour pattern investigated by Büring, and in fact
Büring’s and Jäger’s data overlap without an explicit uniform treatment of
all cases being given in either work. In particular, note that (67) would be a
puzzling case for Büring (1996, 1997), who treats topic and focus semantics
as something in addition to an ordinary semantic interpretation of sentences:

Common semantic wisdom would have it that in a sentence like A Swede is
blue-eyed, the indefinite NP cannot receive a non-partitive and non-generic
interpretation. Such a reading is not available out-of-the-blue but rather
requires an appropriate context. In Büring’s (as well as Rooth’s, Krifka’s,
Jacobs’, von Stechow’s) sense, focussing and topic-marking are devices that
one can apply to a sentence, or one can leave the sentence as it is and inter-
pret it “ordinarily”. In the cases above (as well as the adverbial examples we
will presently come to), there is no “ordinary” interpretation of the sentence
prior to the indicated topic/non-topic distinction. In other words, a sentence
like Ein BÄCKer war unheimlich INTELLIGENT, when uttered after a TV
quiz show, can get an in-group interpretation if and only if the appropriate
topic structure is applied. The reading does not exist prior to topicalization,
unless we claimed that the sentence has readings that exist (namely one with
an existential interpretation of the subject indefinite) but that somehow
never surface in neutral contexts.
While Jäger thus offers a truer picture of the effects of topicality in his
examples, he has nothing to say with respect to the question of why and
how topical material associates (and even has to associate) with focus, and
how the overall structure relates to a topical question in the discourse at that
point. In the appendix, I will attempt to shed further light on the various
advantages and disadvantages of Jäger’s and Büring’s accounts and propose
a synthesis of their ideas. In the main body of the paper, I will restrict my
attention to the empirical side of topical indefinite NPs.
Summarizing, in German, we find the following interpretations of in-
definite NPs, depending on whether or not they are topical, and on the inter-
nal focal structure of the topical phrase. I use the noun Kind ‘child’ as an
example.
Indefinite + Intonation Status Interpretation

ein Kind non-topical (“rhe- existential: there is a new x
(accent depending on posi- matic”) that is CHILD,
tion in sentence) x is a new discourse referent,
x does not equal an old UE,
and it is not part of an old UE.
Ex.: Es stand ein Kind vor der Tür. (‘There was a child standing in front of the
door.’)
ein Kind topical generic: all children,
somewhat open status with
respect to UEs.
Ex.: Ein Kind braucht viel Liebe. (‘A child needs much love.’)
282 Regine Eckardt
ein Kind topical partitive: one of a set P of

CHILDren,
set P must be an old UE.
Ex.: Es spielten Kinder draußen. Ein Kind begann zu weinen. (‘There were
children playing outside. One of the children started crying.’)
ein Kind topical in-group: x is a CHILD,
x is new as a DR and UE,
x is part of an old UE.
The old UE was introduced as
a set of Q’s, CHILD ⊆ Q.
Ex.: Es saßen einige Patienten im Wartezimmer. Ein Kind hatte Schnupfen.
(‘There were some patients sitting in the waiting room. A child hat a cold.’)
*Es saßen einige Kinder im Wartezimmer. Ein Kind hatte Schnupfen.
(‘*There were some children sitting in the waiting room. A child had a cold.’)
Es saßen einige Kinder im Wartezimmer. Ein Kind hatte Schnupfen.
(‘There were some children sitting in the waiting room. One child had a cold.’)
Das Wartezimmer war noch völlig leer. *Ein Kind hatte Schnupfen / Ein Kind
kam herein.
(‘The waiting room was still empty. *A child had a cold. / A child entered the
room.’)
In the appendix, I will briefly (and without further justification) outline how
these readings can be derived in a compositional framework. For now, I will
take this refined notion of the nature of topical indefinites as a starting point
for another look at possible readings (and intonations) of indefinite NPs co-
occurring with manner and result adverbs.
6. Pre-adverbial indefinites – accommodating an environment
It is a well-known fact about German intonation that sentences with a pre-

adverbial object NP – in particular if it is an indefinite NP – have to bear
several accents in order to avoid any readings with narrow focus. I will use
(68) to demonstrate the case. (68a) bears a “neutral” stress pattern while
(68b, c, d) must be understood with narrow focus on Huhn, vorsichtig, and
vorsichtig gestreichelt, respectively. The reader is free to add more patterns
with one or two accents which will always result in narrow focus.
(68) Alicia hat ein Huhn vorsichtig gestreichelt.

Alicia has a chicken carefully stroked
a. Alicia hat ein Huhn vorsichtig gestreichelt.
b. Alicia hat ein Huhn vorsichtig gestreichelt.
c. Alicia hat ein Huhn vorsichtig gestreichelt.

d. Alicia hat ein Huhn vorsichtig gestreichelt.
This contrasts sharply with the inverse word order in (69) which shows the
neutral stress pattern of this type of sentence. Superficially speaking, one
will realize that for (69) one accent is sufficient while (68a) requires three.
(69) Alicia hat vorsichtig ein Huhn gestreichelt.

Alicia has carefully a chicken stroked
A very similar effect for English sentences with manner adverbs has been
described in Gussenhoven (1983) who provides an accent assignment mech-
anism for either case, yet without much discussion of any structural or
pragmatic reasons for this effect.
Jacobs (1992, 1993) describes accent-assignment in German as a mech-
anism that has to rely on a variety of conspiring factors rather than syntactic
structure alone – which many prominent prosodic theories take as their
starting point. In Jacobs’ terms, a constituent α can integrate into another
constituent β under certain circumstances. The constituents can be tied to-
gether by “integration”, in which case this “integrated” domain can receive
one single accent. Several factors support integration. For example, verbs
and objects can only integrate if the thematic relation of the object is appro-
priate (PATIENT-like rather than EXPERIENCER or STIMULUS). They can only
integrate if they are adjacent; and adverbs generally inhibit integration and
do not integrate themselves.
Sentences such as (68a) can be used as out-of-the-blue utterances and are
therefore commonly assumed to be neutral with respect to the preceding
context. I think that this assumption is wrong. Instead, I claim that object
noun phrases that stand to the left of manner and result adverbs are topical
in the sense of Jäger (1996b). Their status is driven not by the intonation as
such but by their syntactic position which will automatically impose certain
(yet unobtrusive) contextual restrictions on the sentences in question. How-
ever, the speaker still has the choice of marking part of the topical material
as new (in focus) by using one of several possible accent patterns. The ex-
amples in (70) show that we find exactly the three possible interpretations
of topical indefinite NPs that were listed in Section 5:
(70) a. generic: Alicia hat ein Huhn vorsichtig gestreichelt / vorsichtig

gestreichelt / vorsichtig gestreichelt.
b. partitive: Alicia hat ein Huhn vorsichtig gestreichelt.
c. in-group: Alicia hat ein Huhn vorsichtig gestreichelt.
284 Regine Eckardt
In sentence (70a), I list all possible accent patterns with unaccented object
NP. The object NP is always understood generically, i.e. referring to what
Alicia does to chickens in general. However, the overall focus structure of
(70a) still drives the generic statement in question (an effect that was de-
scribed in Rooth (1995)). The different variants in (70a) mean that “what
Alicia generally did to chickens was, to stroke them carefully” – “the way in
which Alicia generally strokes chickens was: carefully” and “what Alicia
generally did to chickens in a careful manner was: to stroke them”.12
In (70b) we find the partitive interpretation of the topical indefinite NP
that is triggered by the accent on the numeral. The sentence will require a
previously mentioned set of chickens in order to receive an interpretation.
Once more, we need a second – focus – accent later in the sentence. I give
the (most natural) case with focussed adverb. (70b) means: One of the
chickens that were stroked by Alicia was stroked carefully. (The readings
have already been described in Geilfuß (1992), Geilfuß-Wolfgang (1996)
and Eckardt (1999) but the eventual picture which I will propose will set
them in a different light.)
The sentence in (70c) has been held as the “neutral” topic-free variant of
sentence (68) so far. If I am right, however, and the indefinite ein Huhn is
generally topical, then ein Huhn should exhibit the in-group reading in (70c)
(= 68a). I will now argue for the following two points:
(a) The contextual restrictions that are set by the in-group reading of (70c)
are minimal. They require the hearer to accommodate an interest in the
question What did Alicia do in the situation that the speaker wants to
describe? This minimal interest will be supplied at least by politeness
and hence has never been diagnosed as a “presupposition” of the re-
spective sentences.
(b) The contextual restrictions still bear enough information to rule out
verbs of creation. Verbs of creation are hence the test case which shows
that the pattern is not really presupposition-free.
Addressing (a), let us have a closer look at the putative in-group reading of
(70c).
– The indefinite will introduce a new discourse referent x that has to be

part of an acquainted group P of individuals which have to be known not
as being a chicken but as being an animal.
– The construction interacts with the focus structure of the material that
follows. The non-topical background will tell us what Alicia did to the
animals, and the focus will tell us how Alicia acted on one particular
chicken among these animals.
The following discourse offers an example in which all these prerequisites

for (70c) are explicitly present.
(71) Alicia ging in den Stall und hat alle Tiere begrüßt.
‘Alicia went to the stable and greeted all animals.’
Sie hat ein Huhn vorsichtig gestreichelt, ( ... sie hat eine Kuh
She has a chicken carefully stroked ( ... she has a cow
zärtlich gestupst, und sie hat ein Pferd liebevoll gefüttert.)
tenderly pushed and she has a horse lovingly fed )
The focussed verbal complexes vorsichtig gestreichelt ‘carefully stroked’,

zärtlich gestupst ‘tenderly pushed’ and liebevoll gefüttert ‘lovingly fed’ are
all instances of the alternatives that were introduced by begrüßt ‘greeted’ in
the first sentence.
If no such explicit context is given, then the hearer will have to accom-
modate one, and at best a fairly minimal and plausible one. A minimal as-
sumption about the set of Ps is this: “The speaker wants to tell me about
Alicia in a certain situation. He mentions a chicken x which is part of some
known set P. Let me assume that the speaker wants to tell me how Alicia
acted on P, where P is the set of visible objects in her environment. It is
virtually always safe to assume that Alicia has such an environment.”
IGA: In-group accommodation:

Accommodation 1: P = {the visible things around Alicia}
Accommodation 2: Focus alternatives under debate
FA = {what did Alicia do to the things around her}
These two accommodations can also simply be described as “the speaker

wants to tell me what Alicia did to (some of) the things around her”. And it
is easy to see that this conversational minimum might be mistaken for the
“neutral context”.
Interestingly, subtle stylistic differences between sentences like (72) and
(73) lend support to my diagnosis. Intuitively speaking, sentence (72) starts
with Alicia “on a bare plane”, so to speak, and after it has been computed
we do not visualize more about the scene than Alicia in the emptiness, eat-
ing an apple. Sentence (73), in contrast, seems to evoke a scene where
Alicia is “somewhere” amidst things surrounding her: In front of a bowl of
fruit – or on a meadow below an apple tree – or simply at a table littered
with all kinds of junk.
286 Regine Eckardt
(72) Alicia hat vorsichtig einen Apfel gegessen.

Alicia has carefully an apple eaten
(73) Alicia hat einen Apfel vorsichtig gegessen.

Alicia has an apple carefully eaten
Although I am convinced that these effects exist, one would clearly not
want to build an entire semantic analysis of (72) and (73) on them. On one
hand, one might object that even the contents of sentences like (72) are usu-
ally not understood to take place in the empty space, and once we start to
imagine a suitable environment for (72) it will fit (73) as well and vice
versa. On the other hand, even those who share my judgement about the
intuitively felt differences might conjecture that these differences are not
due to the formal interpretation of prosodic patterns but rather are a side
effect of a style of narration (say, one that is rhythmic, slow, contemplative,
allows for accents and breaks, and hence signals the speaker’s intention to
colourfully describe a scene), while the unpretentious (72) is used to convey
pure information rather than to narrate.
Matters change once we use sentences with verbs of creation. Let us turn
to example (74) with the – corresponding – accents, as in (74a). If our story
so far is right then (74a) should evoke the accommodations in (75). And
(76) elucidates why these accommodations will at once contradict the con-
tent conveyed by (74).
(74) Alicia hat eine Taschenlampe vorsichtig konstruiert.

Alicia has a torch carefully constructed
a. Alicia hat eine Taschenlampe vorsichtig konstruiert.
(75) In-group reading of eine Taschenlampe with standard accommo-
dation of set P: There was a torch among the visible things surround-
ing Alicia.
Focus: What did she do to that torch? She constructed it carefully.
(76) The torch created by Alicia only came into existence after the event
of creating it. “Creating something” is not an act that you can perform
on the visible things around you.
Hence, verbs of creation offer a case where the required accommodations

fail to make sense and yield no sensible out-of-the-blue reading of the re-
spective sentences.13
We will now turn to those variants of adverbial modification of verbsCC
that do make sense and will discuss why. It will turn out that IGA is so
prominent an accommodation in out-of-the-blue contexts that one will
rather class a sentence as ungrammatical than turn from this very loose con-
textual restriction to a stronger one. The discourse in (77) offers a suitable
context for the last sentence which is, hence, acceptable.
(77) Es standen überall selbstgemachte Gefäße herum.

‘There was a lot of home-made pottery standing around.’
Alicia hatte eine Schüssel sorgfältig getöpfert, aber alles andere
Alicia had a bowl carefully modelled but everything else
sah ziemlich schief aus.
looked fairly awkward
The presupposed set P is the set of home-made pots, and the salient alterna-
tives are all well-formed answers to Which pot did Alicia make how? The
sentence is straightforward then, but the context is evidently more than zero.
Another possibility to turn IGA into a presupposition that coheres with
the content of the sentence is to interpret the NP as a plan-referring one.
Consider the discourse in (78).
(78) Robert schmökerte lange in der Handarbeitszeitschrift herum. Zuerst

fand er die Vorschläge alle ziemlich dröge, aber dann hatte er die
geniale Idee:
‘Robert read in the handcraft journal for a good while. At first, he found all
the suggestions fairly dull, but then he had an ingenious idea:’
Er hat einen Pullover mit einer superdicken Häkelnadel
He has a pullover with an extremely thick crocheting needle
gehäkelt
crocheted
(und dann sah das Modell völlig hip aus).
(‘and the result looked really hip’)
As häkeln ‘to crochet’ is a verb of creation and the PP mit einer superdicken
Häkelnadel ‘with an extremely thick crocheting needle’ cannot be a noun
modifier (*Pullover mit einer superdicken Häkelnadel) we would expect the
sentence to be bad. However, here the NP ein Pullover can refer to the gar-
ment proposed in the journal. The instruction for the pullover clearly exists
prior to the event of crocheting, and the set P of instructions for garments of
which the pullover is an element is clearly an understood object in the given
context. And finally, due to the fact that such journals usually offer instruc-
tions for knitting, crocheting and perhaps sewing, we even find reasonable
alternative ways of construction that can serve as focus alternatives. This
matches with the fact that the sentence in (78) allows smoothly for a focus
on the whole verbal complex mit einer superdicken Häkelnadel häkeln. The
288 Regine Eckardt
sentence in (77) contradicts this focus structure, and there seems to be no

reasonable way to think of alternative ways of creating pottery. This effect
is even more dramatically visible for rather neutral verbs of creation like
bauen ‘build’, entwerfen ‘design’, konstruieren ‘construct’.
All the effects listed so far can hence be explained if we assume that the
object NP to the left of an adverb (of manner, of result) in German is topi-
cal, and verbs of creation and verbs of affective actions differ only insofar
as one and the same information structure (topical indefinite object in an in-
group reading) can be met by an extremely low-effort accommodation in
one case but needs contexts that look very restricted in the other case.
Definite object NPs in topical position are unproblematic. If they are
completely unaccented, then they can refer to a previously mentioned entity.
We can leave it to the logical and linguistic competence of the speaker to
ensure that the respective discourse referent or understood object was intro-
duced in a coherent manner.
(79) Paulina mußte in der Schule ein Schiffchen und eine Taschenlampe
konstruieren.
‘Paulina had to construct a ship and a torch at school.’
Sie hat die Taschenlampe sparsam konstruiert,
she has the torch economically constructed
und kriegte auch eine gute Note dafür.
and got also a good grade for it
We can also get an in-group reading if the definite NP is not anaphoric but
licensed as an understood object as in examples (63)/(64) in the previous
section.
(80) Paulina mußte ein Schiffchen bauen.

‘Paulina had to build a ship.’
Sie hat die Steuerung sehr ökonomisch konstruiert
she has the steering very economically constructed
(und bekam dann einen Preis dafür).
(‘and got a prize for it’)
The explanation sketched here can also help us to spot other cases where
topical indefinite NPs may be problematic. If we separate subjects of verbs
like entstehen from the verb by adverbial modification, they become topical
and the same effects arise as in (74).
(81) #(daß) ein Stau rasch entstand

(that) a traffic-jam quickly built up
presupposed: The jam is one of a set of known objects on the scene.
focussing: What did happen to this traffic-jam?
Assertion: It built up.
Contradiction: Existing objects cannot build up.
Once more, the indefinite in non-topical position is as acceptable as a defi-

nite in topical position:
(82) (daß) rasch ein Stau entstand

(that) quickly a traffic-jam built up
(daß) der Stau rasch entstand
(that) the traffic-jam quickly built up
Interestingly, we find parallel behaviour for verbs that were already equated
with verbs of creation by von Stechow (2001). Opaque verbs like suchen
‘seek’ and schulden ‘owe’ can refer to virtual objects which need not exist
anywhere near the events or states reported in the sentence. Adverbs once
more require the NP to their left to be topical and lead to IGA to allow for
an in-group reading of indefinites in out-of-the-blue contexts. Sentence (83)
can only be understood to refer to a given mushroom that is already present
somewhere on the scene. It implies the existence of a mushroom, while
sentence (84) does not.
(83) Hans hat einen Pilz fieberhaft gesucht.

Hans has a mushroom feverishly sought
presupposed: There is a mushroom somewhere at the scene of the
event.
focussing: What did Hans do to that mushroom?
Assertion: He was seeking it.
Consequence: There exists a mushroom that Hans seeks.
(84) Hans hat fieberhaft einen Pilz gesucht.

Hans has feverishly a mushroom sought
no mushrooms are implied to exist
Finally, note that in-group readings do not always literally require the
physical presence of the group at stake. Verbs like loben ‘praise’ allow for
objects that are remote or even dead. The wider restriction here seems to be
that they must exist independently of the event reported.
290 Regine Eckardt
(85) Juliane hat Goethe / einen Dichter wortreich gelobt.

Juliane has Goethe / a poet eloquently praised
Neither Goethe nor the referent of einen Dichter have to be part of a group
of physically present persons. In the case of verbs of saying and thinking, it
seems to suffice that these persons existed prior to the event (here: of prais-
ing). Nevertheless, this does not mean that no restrictions hold: Sentence
(86) is again bad because the poet heralded (‘prophesied’) cannot be imag-
ined out-of-the-blue as one of a number of somehow given artists. (87)
shows the admissible variant.
(86) *Marcel Reich-Ranicki hat ein neues Genie wortreich prophezeit.

Marcel Reich-Ranicki has a new genius eloquently heralded
(87) Marcel Reich-Ranicki hat wortreich ein neues Genie prophezeit.

Marcel Reich-Ranicki has eloquently a new genius heralded
Let me finally summarize the core findings of this section.
– Object NPs in the position to the left of a manner or result adverb are
topical in German. Subject NPs of verbs of coming into existence in the
position to the left of manner (and temporal) adverbs are topical in Ger-
man.
– Matching the results of Jäger (1996a, b), we found that topical indefinite
NPs can be interpreted generically, partitively or in the in-group reading.
– All three readings associate with focus later in the sentence.
– The in-group reading with verbs of affective action requires contexts
which are very often tantamount to saying “we assume a common inter-
est in what the subject did to some thing in his/her environment”. These
sentences have previously been mistaken as out-of-the-blue sentences
without pragmatic side effects.
– The in-group reading with temporally or modally opaque verbs does not
allow for such contexts. The sentences are acceptable in restricted con-
texts but have never been mistaken to be out-of-the-blue sentences.
These are also the main results of the present paper. In the final section, I
will relate these findings to other positions in the literature.
7. On word order
There is an ongoing debate in the investigation of the grammar of adverbs

as to whether adverbs have syntactic base positions. I will not be able to
cover all relevant opinions in this section but want to relate my findings to
two positions that are potentially affected.
There is a substantial body of literature that in synopsis suggests a strong
correlation between “moved NPs” and “topical NPs” in German. The dis-
cussion that I will refer to was originated by the work of Kratzer
(1989/1995) and specifically Diesing (1990, 1992) who demonstrated for
German that indefinite NPs outside VP must be interpreted generically or
partitively14 while those that can receive an existential reading were diag-
nosed to be situated in their base position (inside VP). Ladusaw (1994) was
the first to propose that the true effect of these movements (in German) was
to turn non-topical material into topical material. As the original function of
indefinite NPs is to introduce new discourse referents into the discourse
universe, Ladusaw argues, they have to undergo specific reinterpretations if
they are topical. By this explanation, the distinction between existential and
generic indefinites (and related issues like the well-known individual-
level/stage-level distinction) is linked to deeper semantic-pragmatic mecha-
nisms. This is to be preferred to the somewhat ad-hoc representational se-
mantic interpretation mechanism proposed by Diesing.
Jäger (1996a, b) elaborated Ladusaw’s proposal and formulated an ex-
plicit synthesis between Diesing’s and Ladusaw’s findings, equating NPs
that are scrambled out of VP, NPs that are topical, and NPs that receive
generic, partitive or (Jäger 1996a, b) in-group readings.
Eckardt (1996, 1998) demonstrates that there is a coherent correlation
between indefinite NPs that are covered by the focus domain created by
“neutral” sentence accent and indefinite NPs that are interpreted existen-
tially. Viewed from this angle, it turns out that the movements (out of VP)
as diagnosed by Diesing can equally well be seen as movements out of the
focus domain of the “neutral” sentence accent. These observations finally fit
in coherently with the classical findings of Höhle (1982) that, in German,
scrambling of NPs disturbs focus projection (≈ accents can only indicate
narrow focus domains).
Against this background, the data discussed in the present paper lend
considerable support to the hypothesis that both manner adverbs and result-
oriented adverbs have their base position between the subject NP and object
NPs for transitive verbs; left of the subject NP for intransitive verbs of com-
ing into existence (erscheinen ‘appear’) and right of the subject NP for in-
transitive verbs otherwise (which were not tackled in this paper). (88) shows
292 Regine Eckardt
the transitive case. This hypothetical base order will be called AOV (adverb
object verb) order in the following.
(88) NPsubj Adv NPobj (NPiobj) Verb
Subsequently, NPobj can be moved to the left of Adv (if Adv is a manner
adverb) or even has to be moved there (if Adv is a result-oriented adverb).
If NP is moved, it becomes topical. Luckily though, even for indefinite NPs,
there are ways to deal with their topicality that allow for a sentence meaning
that will suit in virtually any context – i.e. no strongly visible pragmatic
restrictions will arise. Verbs of creation and coming into existence do not
support this unobtrusive kind of topicality and consequently offer a good
test case to demonstrate that NPobj to the left of Adv are indeed topical.
This position has, however, been seriously challenged in recent work of
Frey and Pittner (Frey and Pittner 1998; Frey 2000, this volume). They
claim that the German base position for manner adverbs is immediately to
the left of the verbal complex and after all nominal arguments. I will abbre-
viate it as OAV (object adverb verb) order:
(89) Frey and Pittner: NPsubj NPobj (NPiobj) Adv Verb
I want to discuss this assumption in some more detail, mainly addressing the
article by Frey (this volume). Three tests for base positions are proposed by
Frey (2000):
(I) Existential wh-phrases in German can only appear in base position.

(II) Wide focus projection indicates that all material is in base position.
(III) Scope reversal arises only under reconstruction (i.e. if a scope bearing
element appears in a scrambled position and can return to its base po-
sition before interpretation).
Test (II) here will have to be taken with a grain of salt. Frey and Pittner later
allow for another reason for wide focus projection, namely so-called inte-
grated constructions (Jacobs 1992, 1993). Let me first review the data that
are offered in order to lend support to OAV in the light of tests (I) to (III).
Frey (this volume) presents (90) as evidence in favour of OAV, accord-
ing to test (I).
(90) Peter will jetzt was konzentriert lesen.

Peter wants now s.thing concentratedly read
If wh-phrases have to occur in base position, he argues, then the base posi-
tion of the object NP (was) is to the left of the adverb (konzentriert). How-
ever, consider German examples like (91) which are as well formed as (90)
and would support AOV, according to test (I).
(91) a. Alicia hat dann gierig was gegessen.

Alicia has then greedily s.thing eaten
b. Peter hat dann vorsichtig wen gefragt.
Peter has then carefully s.one asked
c. Claudia hat demonstrativ was gelesen.
Claudia has ostentatiously s.thing read
d. Eberhard zog schüchtern was aus.
Eberhard took shyly s.thing off
...
The result of these findings is (somewhat disappointingly) that either test (I)
cannot be trusted or that there are in fact two base positions for manner
adverbs.
I will not make use of test (III) because manner adverbs do not exhibit
clear scope effects. Frey proposes to use in jeder Weise ‘in every way’ but I
hold it to be unclear what “ways” or “manners” are in play here; after all,
the question in welcher Weise can be answered with manner adverbials,
subordinated sentences and with instrumental PPs or mental-attitude ad-
juncts (which both occupy a syntactic class different to manner adverbs,
according to Frey (this volume)).
(92) In welcher Weise hat er die Frau umworben?

In which way has he the woman courted?
a. Er hat sie umworben, indem er ihr Blumen auf den Balkon warf.
He has her courted by he her flowers onto the balcony threw
b. Er hat die Frau schmierig umworben. (manner)
He has the woman slimily courted
c. Er hat die Frau mit Liedern und Geschenken umworben.
He has the woman with songs and gifts courted (theme?)
d. Er hat die Frau mit einer Balalaika umworben. (instrument)
He has the woman with a balalaika courted
e. Er hat die Frau bereitwillig umworben. (mental attitude)
He has the woman eagerly courted
Hence we remain with test (II). It has already been mentioned several times
in this paper that the word order in (88) allows for wide focus projection,
294 Regine Eckardt
and Frey and Pittner do not deny this. This is once more exemplified in
(93).
(93) Alicia hat vorsichtig ein Huhn gestreichelt.

Alicia has carefully a chicken stroked
Frey and Pittner offer two counter-arguments. Firstly, they claim that sen-
tences like (93) exhibit “integration” of object NP and verb in the sense of
Jacobs (1993). If integration occurs, then a single focus accent is enough to
cover the whole integrated domain; and – as Frey and Pittner (1998) pro-
pose – the integrated material forms a verbal complex. A test for this is
whether or not the verb alone can be moved to sentence-initial position.
Indeed, sentences like those in (94) pass this second test for integration as
well as the focussing test.15
(94) a. ???Gestreichelt hat Alicia vorsichtig ein Huhn.

stroked has Alicia carefully a chicken
b. ???Gelesen hat Cecile aufmerksam ein Buch.
read has Cecile carefully a book
Yet both focus projection and (possibly) the verb’s resistance to front with-
out its object NP could also be explained by the AOV base position. Hence,
we will need evidence in favour of undisturbed focus projection with the
OAV word order. Frey offers (95) as a second argument in favour of OAV
base position:
(95) Peter hat den Artikel sorgfältig gelesen.

Peter has the article carefully read
This is, however, a problematic example. The pragmatic behaviour of defi-

nites was discussed in Section 5 and we saw that the definite NP den Artikel
in this position has to be topical, i.e. refer to a known object in the common
discourse universe. But then imagine the natural flow of a conversation
about Peter: If we already know that there is Peter and that there is this arti-
cle, then the questions What happened then? (a common test for wide focus)
and What did Peter do to the article? (a common test for narrow focus) are
interchangeable. In other words: the question What happened then? is not a
very reliable test for wide focus if we are dealing with known agents and
objects. Tellingly, these matters change once we use indefinite NPs:
(96) Was geschah dann?

‘What happened then?’
a. *Peter hat ein Buch sorgfältig gelesen.
Peter has a book carefully read
(narrow focus, or the described generic interpretation of indefinite)
b. Peter hat ein Buch sorgfältig gelesen.
(ok, in-group reading cf. Section 6)
c. Peter hat sorgfältig ein Buch gelesen.
Peter has carefully a book read
(ok, due to base position and/or integration)
The data in (96) seriously challenge the OAV word order proposed in (89).
Let me finally elaborate Frey’s integration hypothesis for (89) and (96c).
It emerges that it will face problems once we extend it to resultative adverbs
and verbs of creation. It was shown in Section 2 that resultative adverbs
have to stay close to the verb:
(97) a. *Berenike hat üppig den Baum dekoriert.

Berenike has lavishly the tree decorated
b. Berenike hat den Baum üppig dekoriert.
Berenike has the tree lavishly decorated
As integration is viewed as a relation between verb and argument NPs,

(97b), in which the resultative adverb must intervene, should indicate that it
is forbidden to integrate den Baum and dekoriert. However, sentences like
(98) falsify this expectation:
(98) Berenike hat vorsichtig den Baum dekoriert.

Berenike has carefully the tree decorated
If the possibility of integration were driven by the semantic nature of verb,

object, and the way they relate then we should expect that den Baum deko-
rieren can form an integrated unit no matter what kind of adverb comes
next.
In response to this objection, Frey might claim that the relation between
verb and resultative adverb differs from that between verb and manner ad-
verb. The resultative adverb, in some sense, is still oriented towards the
referent of the object NP and, in this sense, it might change the thematic
structure of the verb.16 Common responsibility for the overall thematic
structure of the predicate is – following Frey and Jacobs – another licenser
296 Regine Eckardt
for integration. Perhaps, NP + verb cannot integrate because resultative

adverb and verb are inclined to integrate to an even higher degree?
But again, this cannot be the case because resultative adverb and verb do
not form a common focus domain.
(99) a. Berenike hat den Baum üppig dekoriert. (narrow focus)

b. Berenike hat den Baum üppig dekoriert. (narrow focus)
c. Berenike hat den Baum üppig dekoriert. (narrow focus)
d. Berenike hat einen Baum üppig dekoriert. (NP topical,
outside focus)
e. Berenike hat einen Baum üppig dekoriert. (still no straight
wide focus, as proved in Section 6)
Finally, the behaviour of verbs of creation and resultative adverbs would

certainly appear odd under Frey’s proposal. Verbs of creation and their ob-
jects apparently can “integrate”:
(100) Desdemona hat konzentriert ein Kleid entworfen. (wide focus)

Desdemona has concentratedly a dress designed
However, they resist integration once a resultative adverb is in play, al-

though this would be the one and only chance for the indefinite NP to enter
the sentence at all (in an existential reading as in (100))!
(101) *Desdemona hat hauteng ein Kleid entworfen.

Desdemona has tightly a dress designed
Let me summarize the pros and cons for the two word order hypotheses in
(88) (AOV: universally Adv – Obj – Verb) and (89) (OAV: Obj – Adv –
Verb). It emerges that neither one is cost-free: Where Frey has to diagnose
that integration is blocked, the AOV hypothesis forces us to claim that
resultatives force NPs to move. But while OAV will predict that NPs can
have base positions so unusual that existential indefinites cannot survive
there, AOV keeps the picture that indefinites are fine in all base positions
but sometimes may not stay there – below resultatives, in subject position of
individual-level predicates, in object position of attitude verbs like love or
hate. Moreover, if they are moved, they uniformly become topical.
I will leave it to the readers to decide whether topical base positions or
obligatory movement would make the better choice. I hope to have contrib-
uted to the debate by drawing attention to the syntactic behaviour of resulta-
tive adverbs, and to the topical nature of pre-adverbial NPs. The paper also
offers a moral to the investigating researcher in pragmatics: Not all sen-
tences that can answer the question What happened? necessarily are out-of-
the-blue. There are subtle topics.
Appendix: Referential and denotational topics
The range of topical NPs that played a role in the paper is treated by two authors,
Büring (1996) and Jäger (1996a, b) which I will take as the referential poles in the
discussion, thereby ignoring a wealth of other literature for the sake of a brief expo-
sition, appropriate for an Appendix.
Sentences (102) to (105) set the range of data to be treated, accents are given.
The / symbol marks rise accents, \ indicates a fall.
(102) Ein Schwede hat Schweiß\füße.

(103) Ein/ Schwede hatte Schweiß\füße.
(104) Ein blonder/ Schwede hatte Schweiß\füße.
(105) Ein Schwede/ hatte Schweiß\füße.
‘A (blonde) Swede had (has) sweaty feet’
The accents on Schweissfüsse turn out to be focus accents upon brief reflection, and
the subject NPs are topical in all cases (here both Jäger and Büring agree).
Büring’s account treats the rise accent / as topical accent and offers a semantic
interpretation [[ . ]]T that resembles focus semantics but moreover links the content
of the sentence (and its alternatives) to a salient topical question. Importantly, the
alternatives that play a role in [[ α ]]T are alternative meanings of the same logical
type as the denotation of α (this is as in focus semantics). I will call this “denota-
tional topic”. This will lead us to consider alternative predicates in (104), alterna-
tive quantifiers in (105) and alternative determiners in (103). These last alternatives
are appropriate only in limited cases and certainly not here, and hence Büring’s
account is tailored for (104) and (105) while missing (many) cases like (103).
Jäger assumes that the referential argument of a topical NP has to be a known
object p in the discourse context. The / accent within topic marks focussed material
within topic, i.e. information that is new about the known object p. The Swede p
must hence be known (as part of a set of known individuals) in all cases (102)–
(105), but is known as a Swede in (103), known as a Swede but newly asserted to be
blond in (104), and known as a person but newly asserted to be a Swede in (105). I
will call this notion of topic “referential topic”. – In contrast to Jäger, I will claim
that the accent / in (103) does not signal new information but simply serves the job
of signalling that the noun Schwede contributes old material (in the spirit of
Schwarzschild (1999)). Otherwise we would predict that the narrow focus-in-topic
in (104) implies that the cardinality information ein ‘one’ is known information –
which is not correct.17
Not much has been said with respect to (102) so far, and I will follow common
wisdom here in assuming that completely unfocussed indefinite NPs are completely
298 Regine Eckardt
topical, hence generic, and that generic sentences can generally associate with focus
(as Rooth (1995), among others, proposes).
I think that the distinction between denotational topics and referential topics lies
at the heart of a proper understanding of the full range of (103) to (105). Büring’s
account allows remarkably strong predictions about the kinds of contexts that sup-
port certain sentences (including the famous Scope Inversion Conspiracy (Büring
1997) but fails in (103) and possibly even (105).18 Jäger, on the other hand, does not
have much to say about the focus following the topic. While remarking that “topics
can associate with focus” in a vague way, he does not explain why they have to
associate with focus and neither how the overall construction relates to a discourse
topic (topical question).
A full account of referential topicality will hence have to exhibit at least the
following features (remember that Understood Entities are objects that are accessi-
ble to all anaphoric constructions apart from pronouns; see Section 5):
[[ α ]]refT = the referential argument p of [[ α ]] has to be an old Understood

Entity.
Parts of old Understood Entities are also old Understood Entities.
(Expressions that do not introduce a referential argument cannot be referential top-

ics, only denotational topics.)
[[ [ β γ ]F δ ]]refT = “focussed material within refT contributes new infor-

mation about old p.”
Referential topics associate with focus to create the “referential topical question”.
The referential topical question basically equals the non-focal non-topical material
in the sentence where we lambda-abstract over the focal argument and lambda-
abstract over discourse referents over the referential topical argument. Let me give
an impressionistic example:
(106) [ Eine Frau/]refT hat [eine Lampe\]F gekauft.

a woman has a lamp bought
Topical question: “Who amongst the set of known discourse entities P of
which p = a woman is part – who of these bought what?”
λpλx. (p<P & ∃e( buy (p,x,e)))
The case contrasts with Büring’s denotational topical questions like the one in ex-
ample (107) below. Let me stress the fact that the pure intonation pattern will not
tell anything about the kind of topicality that is in play. It has to be derived from
other factors (like the availability of a referential argument and speaker’s interests).
(107) [Eine Frau/]denT hat [eine Lampe\]F gekauft.

Topical question: “Who bought what?”
λyλx. (∃e( buy (y,x,e)))
The formula in (106) will evidently only be defined in a framework that allows for
lambda-abstraction over discourse referents; Muskens (1996) offers this possibility.
Intuitively, the topical question in (106) is concerned with individuals that we al-
ready know in context, while (107) asks for customers in general without referring
to any given group of customers. Note that (107) cannot be restricted to equal (106)
even if we allow for further silent predicates that restrict the variable y: Without
reference to discourse referents, we will not be able to ensure that y is instantiated
with those individuals we had been talking about already.
I will not offer a full elaboration of [[ . ]]refT here; this task would require a
paper in its own right. Let me finally give another example where the difference
between referential topicality and denotational referentiality is highlighted, and an
explanation of why created objects are bad referential topics if the focus concerns
the respective act of creation.
(108) [Drei/ Aufgaben]denT hat er gelöst.

(109) [Drei/ Aufgaben]refT hat er gelöst.
Three problems has he solved
The first sentence will refer to the topical question For which quantifier q is it true
that he solved q problems? It might arise in a context where candidates have to
solve a certain minimum of problems in order to pass the exam but where all prob-
lems count as equal.
The second interpretation refers to the problems the candidate actually did
solve. The second, but not the first, interpretation allows us to pick up the solved
questions with a pronoun: ... and they were fairly tricky, too. The second, but not
the first, interpretation determines the meaning of die anderen ‘the others’ as in ...
die anderen hat er nicht einmal probiert ‘but the other ones, he did not even look
at’.
The first interpretation, but not the second one, allows the speaker to continue:
... aber vier/ Aufgaben hat er nicht gelöst, also kriegt er kein “sehr gut” ‘but he
didn’t solve four problems, hence he will not get an A’. In interpretation one, three
solved and four unsolved problems will not imply that there were seven problems
altogether. In interpretation two, the two sentences together will necessitate that the
candidate had seven problems to solve.
What will happen in a sentence like (110) (= sentence (79)) where referential
topic meets a focussed verb of creation?
(110) Alicia hat [eine Taschenlampe/]refT [ökonomisch konstruiert]F
refT requires that the referential argument p of “a torch” is part of a known group
individual P available in the discourse context. The referential topical question is
What happened to elements of P? In a normal narrative, the reference time R is
located at a point where P is a known object at the non-modal level of the narrative.
Hence P exists at the beginning of R. The verb of creation ‘construct economi-
cally’, however, asserts that p is constructed at reference time R (within or around
300 Regine Eckardt
depending on aspect, see Klein (1994) and Hinrichs (1985). Adopting the semantics
of verbs of creation in von Stechow (2001) will imply that p only exists after the
event time, i.e. not at the beginning of reference time R. This contradicts the com-
mon background information above, that P as a whole (and hence in all its parts)
did exist at the beginning of R. Which shows the hearer that she has accommodated
incoherent background information and hence failed to spell out the conversational
maxim of quality: “Assume that the speaker intends to tell the truth.”
Notes
1. I would have preferred “How to create things in a certain way” for a title, but
this one covers about any key noun that will be used in the article. The paper
was written as part of SFB 471 Wandel und Variation im Lexikon, and al-
though it does not specifically deal with variation, the contents were shaped by
the lively and inspiring environment at Konstanz. Specifically, I want to thank
Willi Geuder, Renate Musan and Arnim von Stechow for useful discussions.
2. I will use “manner adverb” and “result-oriented adverb” in this introduction
without offering a definition, relying on the reader’s intuitive understanding of
the terms. Section 2 will be devoted to clarifying this terminology. I will
mostly use German verb-final sentences in order to avoid further complica-
tions that arise by deriving V2.
3. Originally, Jackendoff (1975) used the term “subject-oriented adverbs” to refer
to those adverbs that always predicate over the grammatical subject of the sen-
tence (no matter what thematic role it maintains) in contrast to those adverbs
that invariably orient towards the agent of an event and cannot refer to any
other participant. Frey (this volume) discusses the case under the label “men-
tal-attitude adverb”. I will follow the terminology of Geuder (2000) at this
point.
4. This argument is not discussed in Geuder (2000).
5. Tastes of German linguists differ as to how bad they find definite NPs that are
“unduly far to the right”. Some scholars accept definites in virtually any posi-
tion while others tend to star sentences quickly. Ultimately, these tendencies
correlate with the ease with which less-than-out-of-the-blue contexts are ac-
cepted as “normal” by speakers. In the present section, I will follow a liberal
strategy to accept definites while the next section will review Jäger’s more re-
fined picture as to which NPs presuppose what kind of discourse universe.
6. Note that expressively can sometimes also be a manner adverb, e.g. in dancing
expressively. Eventually, the range of possible readings is determined by the
combination of verb + adverb and only few adverbs are “manner” or “result”
per se.
7. Let me warn the reader against taking the following boxes as Jäger’s full for-
mal framework which is actually more in the spirit of Groenendijk and Stok-
hof’s Dynamic Intensional Logic than Kamp and Reyle’s Discourse Represen-
tation Theory. For details, the original thesis (Jäger 1996b) should be con-
sulted.
8. Jäger’s work is concerned with the theme/rheme distinction, also sometimes
discussed as topic/focus, or as the distinction between thetic and categoric sen-
tences. This distinction is not immediately the same as the notion of focus that
was discussed in the literature on association with focus by Rooth, Krifka,
Büring, Jacobs etc. It is easy to see this by noting, for example, that even defi-
nite NPs that refer to Understood Entities (and should hence be “topical”) can
still be focussed in the latter sense, as in:
Joe bought a car which is entirely red. Even [the steering wheel]focus is red.
9. Let me make it clear that I do not intend to defend this classification as the
ultima ratio in the interpretation of noun phrases. Still, Jäger’s treatment – in
spite of possible shortcomings and simplifications – is one of the most elabo-
rate and differentiated accounts available.
10. Cf. Ladusaw (1994), Jäger (1992), Eckardt (1996).
11. Jäger calls both readings in (66) and (67) partitive readings. This would, how-
ever, be misleading for the purposes of the paper. Let me stress that specificity
is orthogonal to the issues that I discuss here. In particular, an indefinite can
have an in-group reading although the speaker is not acquainted with the refer-
ent, and an indefinite can receive a specific interpretation (“a certain”) without
necessarily being one of a known group.
12. In order to be complete with respect to the readings of at least the core exam-
ples in this paper, I have to add the following readings for (70a): The accented
material can be understood as a contrastive focus. If so, then the respective
sentence will contrast with a similar sentence that differs form (70a) in the fo-
cussed material. Importantly, the contrasted sentence will get the same reading
as the original sentence had, and the original sentence will have received its
reading according to one of the possibilities that are discussed in the set in
(70). Similar possibilities exist for (70b) and (70c).
13. This is not to say that contradictory sentences are always ungrammatical –
which they certainly are not. Yet the hearer allows for accommodation only in
order to rescue what would otherwise be an incoherent discourse, and s/he
does so because she assumes that the speaker will have something rational to
say. If the accommodation leads to contradictions with the literal content of the
sentence then it must be a wrong accommodation: It was made on the assump-
tion of a rational speaker and the result of the process contradicts this very as-
sumption.
14. ... or in-group. Diesing herself did not pay attention to in-group readings, but
this gap has been closed by Jäger (1996a, b) since.
15. The sentences are borderline cases between “need a very odd context” and “are
bad indeed”. I will for the sake of the argument behave as if they were really
bad.
16. This observation offers an attractive basis to integrate result adverbs in the
general programme of Ernst (this volume) according to which all adverbs find
their syntactic positions in dependence on their semantic properties – specifi-
302 Regine Eckardt
cally, the NPs they predicate over. If the result-oriented adverb predicates over
the referent of the object NP, then it will have to have scope lower than this
NP, hence to its right (in German, and, surprisingly, in English as well ...). I
will not be able to elaborate these ideas here.
17. As promised, I am very brief here. The reader might want to check this predic-
tion by going through the original proposal by Jäger and testing more cases
with higher cardinals, like five blonde/ Swedes had stinky feet. – Even if we
know that there are Swedes around, this sentence does not require contexts
where the hearer is certain that the number of Swedes exceeds five.
18. I think that (105) should be continued ... a Dane had a blister on his toe, and
another Swede had a corn on the foot. Yet, Büring would assume that (105)
exhausts the range of relevant true propositions that we can make with A
Swede had ... . Once more, the problem is that he cannot account for the fact
that the same NP ein Schwede can very well refer to different Swedes.
References
Büring, Daniel
1994 Topic. In Focus and Natural Language Processing, Vol. II, Peter
Bosch and Rob van der Sandt (eds.) Heidelberg: Working Papers of the
Institute of Logic and Language. Reprinted in Focus, Peter Bosch and
Rob van der Sandt (eds.), 142–165. Cambridge: Cambridge University
Press.
1996 The 56th Street Bridge Accent. London: Routledge.
1997 The great scope inversion conspiracy. Linguistics and Philosophy 20:
175–194.
Chomsky, Noam, and Morris Halle
1968 The Sound Pattern of English. New York: Harper & Row.
Cinque, Giuliemo
1993 A null theory of phrase and compound stress. Linguistic Inquiry 24:
239–297.
Diesing, Molly
1990 The syntactic roots of semantic partition. Ph.D. dissertation, University
of Massachusetts, Amherst.
1992 Indefinites. Cambridge, Massachusetts: MIT Press.
Dowty, David
1979 Word Meaning and Montague Grammar. Dordrecht: Reidel.
Eckardt, Regine
1996 Intonation and Predication. (Arbeitspapier 77 des Sonderforschungs-
bereichs 340.) University of Tübingen and Stuttgart.
1998 Events, Adverbs, and Other Things. Tübingen: Niemeyer Verlag.
1999 Focus with nominal quantifiers. In Focus, Peter Bosch and Rob van der
Sandt (eds.), 166–186. Cambridge: Cambridge University Press.
Enç, Mürvet
1991 The semantics of specificity. Linguistic Inquiry 22(1): 1–25.
Ernst, Thomas
2003 Semantic features and the distribution of adverbs. In this volume.
Fintel, Kai von
1994 Restrictions on quantifier domains. Ph.D. dissertation, MIT.
Frey, Werner
2000 Syntactic requirements on adjuncts. In Approaching the Grammar of
Adjuncts, Ewald Lang, Cathrine Fabricius-Hansen, and Claudia Maien-
born (eds.), 107–134. (ZAS Papers in Linguistics 17.) Berlin: Zentrum
für Allgemeine Sprachwissenschaft.
2003 Syntactic conditions on adjunct classes. In this volume.
Frey, Werner, and Karin Pittner
1998 Zur Positionierung der Adverbiale im deutschen Mittelfeld. Linguisti-
sche Berichte 176: 489–534.
Geilfuß, Jochen
1992 Nominal quantifiers in association with focus. ConSole I Proceedings,
138–147. Amsterdam: SoLe Publications.
Geilfuß-Wolfgang, Jochen
1996 Über gewisse Fälle von Assoziation mit Fokus. Tübingen: Niemeyer.
Geuder, Wilhelm
2000 Oriented adverbs. Issues in the lexical semantics of event adverbs.
Ph.D. dissertation, University of Tübingen.
Groenendijk, Jeroen, and Martin Stokhof
1991 Dynamic predicate logic. Linguistics and Philosophy 14(1): 39–100.
Groenendijk, Jeroen, Martin Stokhof, and Frank Veltman
1994 Update Semantics for Modal Predicate Logic. Technical Report, ILLC,
University of Amsterdam.
Gussenhoven, Carlos
1983 Focus, mode, and the nucleus. Journal of Linguistics 19: 377–417.
Hinrichs, Erhard
1985 A compositional semantics for aktionsarten and NP reference in Eng-
lish. Ph.D. dissertation, Ohio State University.
Höhle, Tilman N.
1982 Explikationen für ‘normale Betonung’ und ‘normale Wortstellung’. In
Satzglieder im Deutschen, W. Abraham (ed.), 75–153. Tübingen: Nie-
meyer.
Hoop, Helen de
1992 Case configuration and noun phrase interpretation. Ph.D. dissertation,
University of Groningen.
2000 Optional scrambling and interpretation. In Interface Strategies, H. Ben-
nis, M. Everaert, and E. Reuland (eds.), 153–168. KNAW, Amsterdam.
304 Regine Eckardt
Jackendoff, Ray
1975 Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT
Press.
Jacobs, Joachim
1983 Fokus und Skalen. Zur Syntax und Semantik der Gradpartikeln im
Deutschen. Tübingen: Niemeyer Verlag.
1992 Integration. (Arbeitsbericht Nr. 13, SFB 282 “Theorie des Lexikons”.)
Universities of Düsseldorf and Wuppertal.
1993 Integration. In Wortstellung und Informationsstruktur, Marga Reis
(ed.), 63–117. Tübingen: Niemeyer Verlag.
Jäger, Gerhard
1992 Diskursverknüpfung und der Stadien-/Individuenkonstrast. MA thesis,
University of Leipzig.
1996a The stage/individual level contrast revisited. Proceedings of WCCFL
15: 225–240. CSLI Stanford.
1996b Topics in dynamic semantics. Ph.D. dissertation, Humboldt University
Berlin. In print: University of München, CIS-Bericht 96–92.
Klein, Wolfgang
1994 Time in Language. London: Routledge.
Krämer, Irene
2000 Interpreting indefinites. An experimental study of children’s language
comprehension. Ph.D. dissertation, Utrecht/Nijmegen.
Kratzer, Angelika
1989/1995 Stage-Level and Individual-Level Predicates. In Genericity in Natu-
ral Language, M. Krifka (ed.) (Proceedings of the 1988 Tübingen Con-
ference, SNS-Bericht Nr. 42.) Tübingen University. Published in The
Generic Book, Greg Carlson and F. Jeff Pelletier (eds.), 125–176. Chi-
cago: Chicago University Press.
Krifka, Manfred
1991 A compositional semantics for multiple focus constructions. In Pro-
ceedings of the 1991 SALT Conference, 127–158. Dept. of Modern
Languages and Literatures, Cornell University.
Kuroda, Sige-Yuki
1972 The categorical and the thetic judgement. Foundations of Language 9:
153–185.
Ladusaw, William
1994 Thetic and categorical, stage and individual, weak and strong. Proceed-
ings of the 1994 SALT IV Conference, 220–229. Dept. of Modern Lan-
guages and Literatures, Cornell University.
Muskens, Reinhard
1996 Combining Montague semantics and discourse representation. Linguis-
tics and Philosophy 19: 143–186.
Parsons, Terence
1990 Events in the Semantics of English. Cambridge, MA: MIT Press.
Reinhart, Tanya
1981 Pragmatics and linguistics. An analysis of sentence topics. Philoso-
phica 27: 53–94.
2000 Strategies of anaphora resolution. In Interface Strategies, Hans Bennis,
M. Everaert, and E. Reuland (eds.), 295–324. North Holland, Amster-
dam.
Rooth, Mats Edward
1985 Association with focus. Ph.D. dissertation, University of Massachusetts
at Amherst.
1992 A theory of focus interpretation. Natural Language Semantics 1: 75–
116.
1995 Indefinites, adverbs of quantification, and focus semantics. In The Ge-
neric Book, Greg Carlson and Francis Jeffrey Pelletier (eds.), 265–299.
Chicago: Chicago University Press.
Sasse, Hans-Jürgen
1987 The thetic/categoric distinction revisited. Linguistics 25: 511–580.
Schwarzschild, Roger
1999 GIVENness, AvoidF and other constraints on the placement of accent.
Natural language semantics 7: 141–178.
Stechow, Arnim von
2001 Temporally opaque arguments. In Semantic Interfaces: Studies Offered
to Andrea Bonomi on the Occasion of his Sixtieth Birthday, Carlo
Cechetto, Gennaro Chierchia, and Maria Teresa Guasti (eds.), 278–319.
Stanford: CSLI Publications.
Zimmermann, Ede Thomas
1993 On the proper treatment of opacity in certain verbs. Natural Language
Semantics 1: 149–179.
Zucchi, Alessandro
1999 Incomplete events, intensionality and imperfectivity. Natural Language
Semantics 7: 179–215.

2003 MannerAdverbs

Uploaded by

Copyright:

Available Formats

2003 MannerAdverbs

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2003 MannerAdverbs

Uploaded by

Copyright:

Available Formats

Manner adverbs and information structure:

Evidence from the adverbial modification of verbs

The empirical observation on which this paper is based concerns a peculiar

(1) a. (daß) Hans vorsichtig eine Nuss öffnete

© Mouton de Gruyter 2003

(2) a. (daß) Hans langsam die Tür öffnete

Furthermore, there is a class of result-oriented adverbs (see note 2) that

(3) *(daß) Hans schwer einen/den Wagen belud

(4) (daß) Hans den Wagen schwer belud

(5) (daß) Hans einen Wagen schwer belud

(7) (daß) Hans die Flöte geschickt schnitzte

(8) *(daß) Hans eine Flöte geschickt schnitzte

(10) *(daß) Hans einen Pullover locker strickte

(11) *(daß) Hans locker einen Pullover strickte

2.1. Manner adverbs

(12) (daß) ein Kind wütend/schnell/fachmännisch einen Stock absägte

(16) (daß) ein Kind einen Stock schnell/wütend/vorsichtig/... absägte

(17) *(daß) vorsichtig/wütend/... ein Kind einen Stock absägte

(18) (daß) schnell ein Hase übers Feld hoppelte

(19) (daß) ein Hase schnell übers Feld hoppelte

Some subject-oriented adverbs, typically those related to intentions, can

(20) (daß) ein Polizist absichtlich vor dem Eingang stand

(21) (daß) absichtlich ein Polizist vor dem Eingang stand

Although it would be an attractive task to study the relation between pre-

2.2. Result-oriented adverbs

(22) Miriam dressed elegantly.

Taking the Davidsonian approach to manner adverbial semantics (as I will

(23) Miriam elegantly dressed. (= only manner reading)

(24) a. Sarah belud unsicher den Wagen mit Holz.

b. Sarah belud den Wagen unsicher mit Holz.

We observe that adverbs of result modification in German are restricted to a

(24) c. Sarah hat den Wagen schwer beladen.

Geuder (2000) investigates the necessary semantic analysis of result modifi-

(25) Olivia decorated the christmas tree elegantly.

(26) Olivia decorated the tree for three hours.

(i) Result-oriented adverbs require a semantic analysis distinct from

3. Verbs of creation and coming into existence

(27) a. Eine Beule entstand.

(28) Die Verhandlungen führten zu einem Vertrag.

(29) a. Andrea konstruierte eine Maschine.

(30) a. Andrea hatte eine Idee.

(31) Andrea schuf eine Statue.

(32) Andrea setzte eine Unterschrift unter den Vertrag.

(33) a. Andrea gab eine Antwort.

It has been a matter of some debate what an appropriate semantic represen-

(34) Sabrina baut eine Maschine.

(35) Es gibt eine Maschine, die von Sabrina gebaut wurde.

(36) Sabrina sucht einen Pilz.

(37) Es gibt einen Pilz, den Sabrina gefunden hat.

In order to stress this similarity, von Stechow (2001) classifies verbsCC as

(38) Andrea baute eine Maschine.

In the following, I will sometimes carefully distinguish between reference

(41) a. Hans hatte eine Idee.

(42) Cecile setzte eine Unterschrift unter den Vertrag.

English German paraphrase reading

(43) Beate baute aufgeregt einen Drachen.

(44) Beate baute aufgeregt den Drachen.

(45) #Beate baute einen Drachen aufgeregt.

(46) Beate baute den Drachen aufgeregt.