- 211 TOPIC, THEME AND THE GERMAN INITIAL FIELD*
Wim Scherpenisse
This article deals with some constructions involving the
German Initial Field (henceforth IF), i.e. the position(s)
preceding the finite verb in main clauses. First, I will
demonstrate two different types of construction involving the
IF. Then I will present some possible structural analyses of
these constructions. In examining the constructions more closely,
I will try to decide which analysis has to be favored. Finally,
I will briefly deal with the distribution of the expletive
element es. The framework used will be the Government-Binding
theory throughout (cf. Chomsky (1981, 1982)).
1. The Initial
Field.
To begin with, consider the sentences in (1).
(1)a
Ein älterer Herr tritt an den Empfang.
(an elderly gentleman steps at the reception)
"An elderly gentleman steps up to the reception desk."
b An den Empfang tritt ein älterer Herr.
"Up to the reception desk steps an elderly gentleman."
In (1), the finite verb tritt
occupies second position. The first
position has been filled by a constituent from behind the finite
verb. In principle this may be any one constituent. This possibility of fronting any phrase might lead one to adopt the wellknown theme-rheme distinction for German main clauses. However,
it is not at all clear how to decide what is theme and what is
rheme. No single semantic or pragmatic feature suffices to
characterize a phrase as, say, the theme. Research surveys like
Lutz (1981) show that there is a whole scale of definitions of
the notion "theme", or rather opinions about this notion. For
this reason it seems to me that theories should not be based
on a semantic or contextual definition of "theme". The vagueness of the notion suggests that in this area there are ten-
- 212 dencies rather than rules. Instead, we should look for a structural definition or diagnosis.
In view of the fact that any constituent may be fronted to
the IF, we could regard the post-finite part of the sentence
as a predication of the phrase in the IF, simply identifying
the latter with the theme, so that sentences like (1) fit into
the schema (2).
(2) Theme - finite
verb
(V,.) - rest
of
the
sentence
(Rheme)
By defining Theme in this way, we have a purely structural
definition, which nevertheless strongly resembles one of the
more current traditional definitions: theme =
Satzgegenstand
(roughly, "that about which the sentence tells something").
In (1) and (2), we have one position in front of the finite
verb. This position is identified with the Theme. However, we
will have to refine our analysis because of the existence of a
sentence type with two IF positions. Consider (3).
(3)a
Den Kerl, den habe ich schon zu oft gesehen.
(that guy, that-one have I already too often seen)
"That guy I've seen too often now."
b Dieser Minister, warum ist der noch nicht da?
(this minister, why is that-one yet not there)
"This minister, why hasn't he shown up yet?"
In (3)a the finite verb habe is preceded by an NP and a
demonstrative pronoun, in (3)b the IF is filled by an NP and
a question word, warum. I defer an analysis of sentences like (3)
to a later section. Here it is sufficient to notice that some
German main clauses have structure (4).
(4)
X - I - V,. - rest
of
the
sentence
I will refer to X as the Topic of the sentence, for reasons
that will become clear later. Whatever analysis is proposed,
it will have to account for both (2) and (4).
2. Some
analyses.
Before we go on, let us consider some analyses that have been
proposed to account for structures like (2) and (4). To do that,
we will make two assumptions that are widely accepted.
The first assumption is that German and Dutch are SOV languages
- 213 underlyingly, i.e. that they are verb-final in their base
structure and in introduced dependent clauses (consider Koster
(1975) and Den Besten & Edmondson (1983)). The second assumption is that the finite verb in main clauses is in a position
identified as [+tense] (Den Besten (1983), Evers (1982)), schematically (5) .
(5)
s
+tense]
r«
We will take these assumptions as uncontroversial.
Having made these two assumptions, we can consider the
alternatives. With each alternative I will mention some authors.
This, however, does not mean that they have proposed the structure given in every detail. More specifically, I have taken
COMP to be a daughter of S and a sister of S in all structures;
moreover I simply skip the question of the position of INFL .
Each alternative will be accompanied by a Topic structure, analyzed in accordance with Chomsky (1977). The Topic is base
generated under S and a coreferent element (in German a so2
called d-pronoun ) is moved to COMP.
The first analysis is given in (6)1; cf. Haider (this volume)
and Scherpenisse (1984).
(6)1
This is in a way the simplest possible structure to deal with
(2) and (4). It is assumed in (6)1 that after placement of V f
in COMP a phrase from under S is Chomsky adjoined to the left of
COMP. In the Topic structure given on the right, the Topic is
- 214 base generated under S, and "something" is moved from S to COMP.
The dotted line is meant to indicate coreference, a continuous line
stands for movement.
Consider now the second analysis (cf. Cremers & Sassen
(1983), Platzack (1982)) :
(6)11
Here, the XP is taken not to adjoin to COMP, but to S. In the
Topic structure, however, the Topic is again supposed to be
under S, but the coreferent element is in a position under COMP
specified as d- in (6)11.
In the third alternative, the distinction between Theme and
Topic structures has been minimalized (cf. Koster (1978), Lenerz
(1982)):
S/E
(6)111
d
«T /0 i
The Theme in Theme sentences and the Topic in Topic sentences
are analyzed as one and the same structural position, called
Topic in (6)111. Again, a d-pronoun is fronted to a position
under COMP. If it remains there, we have a Topic sentence; if it
is deleted (by free deletion in COMP) we have a normal Theme
sentence.
To begin the comparison, let us look at XP movement, d-move3
ment and wh-movement in (6)1,11,111. In (6)1, these movements
cannot be distinguished a priori, because they are all movements
- 215 to COMP. In II and III, on the other hand, wh-movement and dmovement are different from XP movement: the first two are
movements to COMP, whereas XP movement is to an adjunction position of S (II) or nonexistent (III). Hence, only II and III
correctly predict that there will be no XP preposing in dependent
clauses but that wh- and d-movement is possible in these clauses.
Thus, the correct predictions can only be made without further
stipulation when wh- and d-movement on the one hand and XP
movement on the other are distinguished structurally. So this
first comparison already favors II and III over I, like the
more detailed analysis of Topic structures given in the next
section.
3. Topic
structures
and deletion
in COMP.
Consider the following sentences:
(7)
Den Kerl, den habe ich schon zu oft gesehen,
(that guy, that-one have I already too often seen)
"That guy I've seen too often now."
(8)
Dieser Minister, warum ist der noch nicht da?
(this minister, why is that-one yet not there)
"This minister, why hasn't he shown up yet?"
(9)a Diesen Jungen, niemand darf den beleidigen,
(this boy, no-one may that-one insult)
"This boy, no-one may insult him."
b *Diesen Jungen, niemand den darf beleidigen,
(this boy, no-one that-one may insult)
(10)a ?Diese Frau,
(this woman,
"This woman,
b *Diese Frau,
(this woman,
c ?Diese Frau,
(this woman,
dieser Mann schlägt die zusammen,
this man beats that-one together)
this man is beating her up."
dieser Mann die schlägt zusammen,
this man that-one beats together)
dieser Mann, der schlägt die zusammen.
this man, that-one (masc.) beats that-one (fem.)
together)
The type of topic structure that occurs most frequently is given
in (7,8). In this type the second IF position is filled by a whphrase, or by a d-pronoun coreferent with the topic. When such a
d-pronoun is fronted, it is impossible to have more than one XP
- 216 ((9)b, (10)b). If, however, the d-pronoun remains in place, two
XPs are acceptable ((9)a, (10)a), even if the second one is accompanied by its own d-pronoun ((10)c). These facts can be interpreted as follows:
First, consider the configuration in which the d-pronoun has
been moved to COMP, in which case we may take it to act as a
predication operator. When an XP is moved up to a position adjoined to S, this adjunction creates a second operator. If we
assume that there can only be one operator per sentence, the
ungrammaticality of structures like (11) is immediately accounted for.
(11)
VfAe]L-[e]>
On the other hand, when the d-pronoun remains in situ, it
does not function as an operator. Therefore another XP may be
fronted without violating a principle of grammar: there is
only one movement to an adjunction position. (Remember that
we consider Topics base generated under S.) Sentences with dpronouns in situ are always somewhat marked. This probably
follows from the fact that d-pronouns are operators in the unmarked case and therefore tend to be locally bound.
Only analyses (6)11 and III are consistent with the restriction that the second IF position is filled by a wh- or d-element
in the unmarked case, whereas I needs stipulation. So, II and
III are favored over I also by the analysis given. I would now
like to present some evidence that II, in its turn, has to be
favored over III. The evidence concerns deletion in COMP.
In German and Dutch the COMP position must be lexicalized
in finite sentences. I will not go into the question why this
- 217 is so but simply state the fact. This means that we can only
investigate deletion in COMP when COMP contains more than one
element. This is probably the case in some subordinate structures with wh- and d-elements. It turns out, now, that wh- and
d-elements which bind an empty category in S may never be
deleted, cf. (12,13):
(12)a
Ich weiß, [ rnMP wer daß] [e] gekommen ist. (D-structure)
(I know,
who that
come has)
"I know who came .ii
b Ich weiß, [ rnMP wer] [e] gekommen ist.
c *Ich weiß, l-p^ü daß] [e] gekommen ist.
-COMP
(13)a Der Mann, [ .
dem daß] ich [e] mein Geld geliehen habe,
(D-str.)
(the man,
who that I
my money lent have)
b Der Mann, [roMP dem] ich [e] mein Geld geliehen habe,
c *Der Mann, [
daß] ich [e] mein Geld geliehen habe.
COMP
Deleting daß from the D-structures (12)a and (13)a yields a grammatical output, but deleting wer from (12)a or dem from (13)a
leads to an ungrammatical sentence with an unbound EC (the ccases). So, summarizing we may say that in tensed clauses only
base generated COMP items may be deleted, provided this does
not yield an empty COMP. But if we accept this, then we must
also conclude that d-pronouns may NOT be deleted freely as in
alternative (6)111: this would leave an EC to be related to a
position outside S. Clearly, this analysis of deletion in COMP
favors II over III.
As I said above, if a d-pronoun does not move, it does not
act a predication operator and no local binding is required.
Also there is no EC. In that case, the coreference relation
between the d-pronoun and the Topic can cross several clause
boundaries, witness (14):
(14)
Diesen Mann, es ist wohl klar, warum ich den nicht mag.
(this man, it is PRT clear, why I that-one not like)
"It should be clear why I don't like that man."
The grammaticality of (14) confirms what has been said so far.
4
- 218 -
4. Theme
structures.
I now turn to sentences involving one IF position, the theme
structures. I will formulate some restrictions on the fronting
of an XP to the S adjunction position.
The first restriction seems to be that only one non-verbal
phrase may be fronted. See (15-17).
(15) *Umziehen sie können nun endlich,
(move they can now finally)
"Now, finally they can move to a new house."
(16) *Der Mann aus der Hand schlug seiner Frau die Teigrolle,
(the man out-of the hand beat his wife the rollpin)
"The man beat the rollpin out of his wife's hands."
(17) ^Gestern mit dem Ball hat er mir am Strand ins Gesicht geworfen,
(yesterday with the ball has he me at-the beach into-the face
thrown)
"Yesterday on the beach, he threw the ball in my face."
But there are also sentences in which the IF seems to be doubly
filled:
(18)
Mit dem Ball ins Gesicht hat er mir geworfen,
(with the ball into-the face has he me thrown)
"Throw the ball in my face he did."
(19)
Gestern am Strand hat er mit ein Geheimnis gesagt. (18,19
from Haider (1982)
(yesterday at-the beach has he me a secret said)
"Yesterday on the beach he told me a secret."
(20)
Nach Dänemark zum Essen sollte man nicht fahren. (Joseph
Bayer, p.c.)
(to Denmark to ea-t should one not drive)
"You shouldn't come to Denmark to eat."
I will come to these below.
Second, a non-finite verb may be fronted with or without
objects; a single modal may never be fronted:
(21)
Backen hätte ich den Kuchen schon gerne wollen,
(bake had I the cake PRT PRT want)
_
"I would have liked to bake the cake."
(22)
Den Kuchen gebacken hätte ich gerne,
(the cake baked had I PRT)
"Bake the cake, I would have liked to."
(23) *Wollen würde ich den Kuchen gerne backen,
(want should I the cake PRT bake)
- 219 Another NP may never be fronted together with the verb:
(24) *Mein Vater backen möchte gerne einen Kuchen,
(my father bake would-like PRT a cake)
"My father would like to bake a cake."
We would like to explain these fronting (im)possibilities in
a non-arbitrary way.
The first restriction that comes to mind vis-a-vis sentences
like (15-17, 24) is that the string to be fronted must be continuous. In fact this restriction is straightforward: a discontinuous string would bind more than one trace, which is a
violation of the Bijection Principle (Koopman & Sportiche (1982)).
But clearly, this restriction does not suffice. The following
sentences have continuous strings fronted, but are nevertheless
ungrammatical:
(25) *Dem Kunden einen Ladenhüter hat er verkauft,
(to-the customer a shopkeeper has he sold)
"He sold the customer a shopkeeper."
(26) *Johann gelacht hatte. (Den Besten, p.c.)
(Johann laughed had)
"Johann had laughed."
So there has to be a second restriction. The most common formulation of this restriction is that the phrase fronted must be
a constituent, i.e. it must resort under one node. It need not
be a maximal projection, cf. (18-21, 27).
(27) [N Milch] ist [
keine [N e]] mehr da.
( milk
is
none
more there)
"There isn't any milk left."
(18-20) involve PPs and/or adverb combinations, (21) a single
V and (27) a noun that is fronted whereas its determiner, keine3
remains in place.
Turning now to cases like (18-20), I suppose there must be
some kind of restructuring here, such that the phrases end up
under one single ADV node or perhaps an ADV projection. In the
case of (19) this may look like (28), for example:
(28)
ADV'
gestern
am Strand
- 220 -
I should add that it is not at all clear to me presently if
there is something like an "ADV projection". Perhaps it is simply
a property of adverbs that they may cluster together to form new
ADV constituents, cf. the well-known English examples (29, 30)
from Jackendoff (Craig Thiersch, p.c.):
(29) Out of the building into the street ran Mary.
(30) Down the aisle with Marlene Dietrich walked John.
In analyses of the German examples restruturing cannot be avoided,
because the ADVs and PPs may also be fronted by themselves, cf.
(31).
(31)
Zum Essen sollte man nicht nach Dänemark fahren,
(to eat should one not to Denmark drive)
same meaning as (20)
I would like to conclude this section with some remarks on V1
sentences, i.e. sentences in which no XP is fronted so that V_
is in first surface position. We may take COMP to bear the
feature [+wh] in such sentences. [+wh] is an operator defining
scope, which means that fronting of an XP would create a structure
with two operators; this is ungrammatical, as I argued in section
3. In independent questions the finite verb lexicalizes COMP, in
dependent questions one finds a special complementizer, ob (- English ifa
whether).
Some languages also lexicalize [+wh] in independent questions, cf. the French formula est-ce
que.
5. Expletive
es.
I will now try to account for the distribution of the expletive
element es in German, which corresponds to both English it and
there.
I will begin by listing four different types of es; the
reasons for thus dividing the different types of es will become
clear below.
Type
type
type
type
a1: weather-es, see example (32);
a1: the impersonal construction, ex. (33);
b1: the impersonal passive construction, ex. (34), and
b2: the existential construction, ex. (35).
- 221 (32)
Es regnet,
(it rains)
(33)
Es überläuft mich kalt,
(it wailks-over me cold)
"I shiver."
(34)
Es darf gelacht werden,
(it may laughed be)
"Laughing is allowed."
(35)
Es ritten drei Reiter zum Tor hinaus,
(it rode three horsemen to-the gate out)
"Three horsemen rode out of the gate."
I start with types a1/2. In these two constructions we have
a lexical item that does not assign a 0-role to its subject as
an inherent property - in other words, it is an idiosyncracy of
those elements that they do not assign a subject 9-role, but
there is no structural reason why there could not be a subject.
Therefore the subject position is filled by a non-referential
pronoun, es, with which the finite verb agrees.
For the b-cases the story is different. Let us first look at
b1. Impersonal passives are passives of intransitive verbs, that
is to say verbs that normally assign one 0-role, namely to their
subjects. However, as usual passivization absorbs the subject
©-role. That is, the verb ends up assigning no 6-role at all.
The subject position is empty and there is no other NP that
could be fronted to it as in normal passives of transitive verbs.
So we see that there are clear structural reasons why the subject
position must be empty. But this leads to the question: where
does the es in (34) come from?
I will assume that in sentences like (34) there is an existential
operator, call it E, which binds the empty subject. I take E to
be in a clitic or adjunction position, adjoined to S. Now look at
structure (36)a:
(36)a
S
/
COMP
\
V.
E
I
I
darf
S
/ \
/\
S
NP
e
VP
gelacht
werden
- 222 This structure, where V f has been fronted to COMP, is grammatical. But if we try to front the EC in subject position to
the IF, an ungrammatical structure ensues:
(36)b
NP
e
darf
e
gelacht werden
This is so because there is no binder for the upper EC. Hence
this EC is illicit, hence it must be lexicalized. In this structure, too, the lexicalization is es. Alternatively, we may consider es a scope marker for the operator E, generated in place.
This becomes more plausible when we take into account that es
never shows up in post-COMP position in sentences like (34),
cf. (37).
(37) *Darf es gelacht werden?
(may it laughed be)
Deletion of es from (37) yields the correct question form of (34)
The base generation hypothesis is still more plausible with
type b2. In sentences like (35), es may be co'indexed with the
subject or, alternatively, be a scope marker for the unrealized
operator E. Thus it would be comparable to French ne which acts
as a scope marker for negations (cf. Kayne (1983)).
Sidestepping to Dutch for a moment, we find support for the
analysis given above. In Dutch, the expletive element in the
a-type sentences is het,
a pronoun historically related to the
German es, whereas it is er in the b-type. Er, which has no
expletive German counterpart, also shows up in post-COMP position, see (38) :
en P
o 3
3* 0)
fl> H
P *<
13 co
CD p3
en
Pw 0H l
CO
fD
^^
TO
03
—1
vo
00
£*
P
CO
—' P
Hl
0
l-i
CO
e
0
3*
en P
H
CO
3*
0
c
H
OJ
3
OJ
fD
P
0)
3
P
P"
Hl
0
P
fD
co X
p- r t
CO P
• CU
•O
0
^<
(0
fD
OJ
CO
fD
3
rt
fD
3
O
fD
CO
«*•
O
Cu
CO
13
rt
3*
Pu
rt
0)
3
fD1
P
H
rt
0
o
fD
3*
0
H
2
0)
P1
CO
0
fD
3
rt
p0
3
131
P
P
Cu
co
<
DJ
.—. r t
0
13
*
O
Prt
tr
P
<
(ü
O
0
3
CO
o
P0
c
co
M
><
O
3
prt
rt
fD
DJ
P
3
3*
Cu
rt
rt
tr
fD
•
•• ^
VD
H
5*3 13 H l 3
Cu O P- CU
CO i Q
CO fD i Q
0
fD p- DJ fD
0
P
rt
H
13
p- PCu1 ^ - v 0
r t 13
—i
P
3
O
vo
co1 CO
OJ
co O P
P0
*. Cu p- r t
•—•
3
IQ
p3
P
3" 0
p- (D Cu r t
3
3
X
P" P" *—
r t CO *<
P- fD O •
p3
3
CO
•—
.
Cu
CO
co
c;
s'
fD
Cu
3
Cu1
P
*:
CO
pco
>
cr
fD 3
Cu
3
Cu1
P
Hl
0
0
3
D.
•<
P3
CO
PCO
O
P
fD
P3
£
s:
tr
rt
3'
Cu
rt
TO
Prt
\->
CU •
er
3
P3
—
. » 3 fD
Hl
"*
CU
CO
O
Cu 0
3 13
fD
cr
fD
Cu
OJ
I_I.
0
p3
fD
OJ
rt
o
I-!
Cu
rt
0
P.
DJ
fD
Hl
P3
P3
i£|
fD
W
p-
o
fD
rt
3*
TO
Cu
hi
—>.
DJ
<J> CO
rjo
J^ CU
•— 3
.
<
Cu
3
Cd
Cu
P
rt
ßo
fD
:
3
ö
rt
*<
CO
0
fD
X
rt
fD
3
DJ
fD
DJ
Cu
3
cr 03
Cu
3
fD
X
iQ
13
P"
P- fD
3
rt
prt <
3" fD
fD *
H
Cu
3
O
0
3
<
P3
O
fD
DJ
rt
tr
Cu
rt
rt
3"
Pco
P
3
r t Cu1
3" P
fD »<
co
CO pO co
1
O 0
Cu1 I-h
P
M a
fD c
OJ
rt
s:
Cu
O
DJ
p1
CO
p3
OJ
i_i-
fD
O
rt
cr
fD
s
0
(D
prt
13
O
0
CO
3
rt
l
rt O
3" O
fD
*d
P
P- o
uQ p"
t r prt rt
prt O
P
Cu 'O
O 0
X
co
• prt
H P0
3* 3
Cu ^
§
Hl
•
el-
*-* Cu
fD
P
I
n
3
rt
tr
fD
P
(D
=
>—
^
<-
s
p-
o
3"
3"
Cu
CO
rt
tr
fD
CO
Cu
hi
f1
fD
3
fD
i-J
N
O
0
3
3
(ü
O
rt
co
O
rt
3"
fD
0
H.
*<
ê
H3
3*
fD
13
H.
(D
CO
fD
3
rt
P
3
P
pj
c •<
CO
3
fD
rt
O
3*
pco
OJ
TO
PCO
rt
H
P-
hS
0
rt
0
s:
fD
CO
0
Hl
TO
03
p3
rt
t-i
P
3
co
prt
p-
<
fD
CO
fD
3
rt
fD
3
O
fD
CO
O
0
3
CO
fD
X
PCO
rt
fD
3
rt
PP
P1
O
0
3
CO
rt
H
C
O
rt
p0
3
P3
O
fD
H
3
P
3
r t **•
prt
(D
CO
C
r t 13
r t (D fD
0
CO O
P1
tr
P
P
(D
P"
3 13
P1
fD H. '<
h
0
N cr1 r t
P
3*
—. fD (D
3
el- c
cr
0
o
r t s'
fü 3"
p0
3
P
CO
O
fD
3
Cu
3
TO
hs Cu
DJ
cr <fD
e
r t H ..^ 3
vo
pco
3
0
rt
tr
TO
UI
CTi
U)
Hl
Hl
•
e
co
fD
DJ
cr
p-
P
M
Rj
ö
^^
II
rjo
to
•»-'
^
fD
CO
13
fD
O
PP
\->
<<
Hl
0
H.
el-
s'
P-
CO
0
O
O
C
i-i
H.
fD
3
O
fD
3
13
Hco
PCO
H3
P3
co
C
et
3"
fD
cr
i_i.
fD
O
rt
O
0
i-i
H.
fD
O
etHi
0
H
w
3
<
13 13 ^
0
H
•
H3
3*
Pco
O
p
3
3
0
rt
cr
fD
3
P
p3
rt
P
P3
fD
CO
Prt
P0
3
P3
Pcn
3*
•
50
fD
C1
P
P
fD s
DJ
X
Ps~*
co
rt _J.
fD VO
00
3
00
rt
,
P- —'
P
P 1 3*
o
co P 1
(ü OJ
3
CO
rt
DJ
(D r t
!-h 3
3"
0
O fD
M fD
CO 13
fD ^
0
CO
<
fD Z Pl-i s- r t
^< p- PH 0
r t fD 3
•<:
'O
fD
0
Hl
rt
3"
fD
CO
C
cr
LJ.
fD
O
rt
rt
tr
P
rt
TO
03
O
(ü
cr _—k
n
rt
<D
00
3
P
fD
0
P
p3
M
Hl
*»
*
P
3
P
3
p
P-
••—
rt
tr
fD
^
OJ
0
fD
W
•^ 50 •<:
13 fD co
fD C pM co
3
rt-4 P
0
S
3 13
r t TO a H
hS
fD
O TO ^—s co
—1
P
fD
H. Cr* ^D 3
TO
00
I-!
rt
K:
CO
fD
^ *-* O0
tJ .
V
<
(D *—^ s: fD
H C p- H
H- p j fD
rt
rt p j
0
t r p- P0
P
CO
O
d
3 OJ
fD r t co
H.
fD
3 < rt <
P
13 P
P3 ^
X
P
fD
3
<•
CO etTO
P
\->
rt
tr
0
d
vQ
3*
Het
3
P
^<
cr
fD
PCO
P
3
>
C
X
'*-'
• ^
t,
V
PO
tr
(D
H.
Hi
Hl
pj
rt
OJ
n
i-3
tr
fD
«-
Ä*
TO
hs
TO
1
O
0
3
CO
rt
H
C
O
r+
p-
o
3
CO
rt
0
Hi
H.
0
3
cr
0
rt
3*
s:
ppj
P
1
pP
3
co
rt
tr
fD
H
^
pCO
c
3
Hl
pP 11
P
fD
DJ
•
tr r t
fD V
r t fD
p- 55
r t fü 0
fD 13
H 3
fD
fD
P
X
P
np- tr r t
0
fD 0
P
P
pJ
rt
P- €
W
N 0
fD
PCO P* r t
P
CO
OJ
3
fD1
P- iD
P
P. C H l
fD P
•
O ÜQ
r t1 fD cn
P
CO 0
P
CO
•<
P
CO
Hl
P3
O
0
O
(D
P
*»•
rt
tr
fD
rt
tr
P
rt
3
fD
P
3
co
rt
V
fD
3
P
3
fD
0
13
fD
P
?r
fD
CO
P
P
rt
0
P
C
CO
(D
^
3
0
Hl
P
CO
O
0
'O
fD
M
cr
e
rt
S3
3*
fD
P
fD
1
3
P-
M
OJ
K:
p
3
p3
p-
3
P1
P
co
fD
^
P
3
O
fD
P.
c
rt
O
3"
c
p
3
*
O
fD
P
TO
co
&
I_I.
fD
0
rt
13
0
co
prt
PO
3
3
e
co
rt
cr
fD
(D
3
hS 13
rt
P ^<
P"
fD
X
P-
Hl
0
P
O
P
rt
tr
fD
(D
N
P
fD
3
O
P
co
P
OJ
P-
Hl
Hl
fD
oo
00
••
K
r t1 fD
to ^
0
fD 3
rt P
P
t r (D ^<
t r p- o
P
co 0
< 0 3co
P3
,_,
O
H
P-
rt
P0
3
0
Hl
3
fD
p
(ü
p
co
0
3
co
>—•
P
tr
= ^^
3
co p
H
S
p
• < l£l
M
P r t fD
C tr p
kQ fD
t r P >jq
P- (D fD
3
M
iQ H P
P O
P c tr
H i n fD
M tr s
0 fD
s: O J S
fD
0
DJ cr P
••0 fD DJ
= —' fD
3
•o
s
F
P
C
iQ
3"
p.
3
l£3
, .
rt
3*
fD
p
fD
W
P
3
P
iQ
3
p lQ
•<: fD
PM
01 P P )
P O
P c tr
M i O fD
H 3* 3
0 fD
S OJ «
(D
0
DJ cr P
• fD OJ
= —' fD
3
•
13
P
fD
S3
tr
p-
0
tr
DJ
P-
O
rt
rt
tr
0
co
fD
OJ
PHl
Hl
(ü
P
fD
3
O
(D
CO
p-
co
Hl
P
0
3
rt
fD
OJ
rt
0
O
O
3
13
^
p
3
pj
OJ
*-
P-
fD
rt
tr
fD
p
fD
cr
•<
3
pco
co
p
^Q
fD
3
fD
P
P1
P
PN
CO
p-
$3
fD
fD
3
co
e
0
tr
(D
P
3
DJ
0
rt
tr
fD
P
DJ
•*
OJ
0
3
PP"
rt
0
P
P
S3
tr
cr
H
o
fD
P
fD
M
rt
^<
rt
tr
p-
3
P
fD
s
tr
i
fD
0
fD
P"
Hl
rt
tr
fD
3
fD
3
rt
CO
DJ
p-
p-
3
Hl
Hl
fD
P
fD
3
O
fD
CO
O
fD
P
fD
p-
3
uQ
cr
P
co
fD
PCO
rt
P
P-
cr
0
rt
P-
0
3
fD
3
(D
P
P
rt
fD
DJ
^I
0
3
M
rt
P
W
fD
co
P
<
fD
P
•<!
«
rt
P
3*
PCO
^
H
CO
P-
O
rt
fD
OJ
co
fD
rt
Hl
fD
fD
p-
co
p
3
0
rt
tr
fD
P
0
T3
fD
P
P
rt
0
P
P
3
P
3
rt
tr
fD
P
fD
•<
P
fD
i-3
POJ
P
3
P
co
fD
s
0
iQ
tr
^
cr P
fD 3
o OJ
P
c OJ
3
<
p-
p-
cr
oQ
«
3
P
LQ
c
^""N
K
0
c
pj
CO
P-
(D
3
3
0
3
CO
fD
3
rt
co
P
O.
fD
H
co 13
tr
p
CTt p
co
*H—'
fD
co
Hl
•
fD
rt
p3 13
P
p
fD
3 13
P1 0
P
co
*J fD
P
rt
P-
cr P
0
Hl
H
P
fD
P
DJ
H
fD
PCO
fD
P"
3
fD
3
rt
0
13
0
fD
p
P
P
cni r t
0
sC P
fD co
CO
p-
rt
0
3
>
0
P
rt
3
rt
P
3
0
13
fD
P
P
rt
0
P
DJ
OJ.
0
tl
O
rt
p0
3
13
0
CO
p-
rt
p0
3
co
•
03
o '<
i-3
13
O
P
CO
P
«
S
13
Hl
O
P
fD
P
rt
fD
0
P
3
O
O
M
p-
tr
3*
fD
>
fD
fD
DJ
n>
3
rt
CO
P
3
<T3 S3
P
tr
fD pCO
- • — *
O
P
3
<< ^
p-
^
en
0
P
c
co
fD
P
O
0
3
Hi
H
P-
P
H
S3
0
0
M
OJ
P1
P-
X
fD
rt
0
co
rt
P
fD
co
co
0
rt
tr
fD
3
P-
3
tQ
P
13
tr
O
O
3
3
fD
O
rt
P
p-
0
3
p.
<
h3
<*
fD
«
0
CO
rt
fD
P
tr
fD
rt
S3
fD
(D
3
rt
S3
0
P
(D
H
P
rt
p-
0
3
co
P1
p-
P
rt
rt
fD
P
0
3
CO
^
S3
fD
CO
30
0
M
OJ
OJ
co
0
fD
00
hJ
co
\—i 0
p
<£> r t
CO
oo
rt
0
p- 13
3
tr
rt 0
H
0
^
—* 3
KD
O
•—
Hl
p
3
P
cr
rt
3"
PCO
DJ
P-
3
0
rt
P-
0
3
<D
00
* — •
*^
p"
>—'
0
3
0
rt
,—N
.
O
p1
P
p-
rt
^<
O
0
3
co
rt
P
P
p3
rt
•
H
rt
P-
CO
3
K:
o
P1
P
p-
3
rt
tr
p
rt
p3
p-
co
rt
co
rt
P
P
O
rt
—J>
p
co
fD
rt
O
P
p-
co
P-
3
131
OJ
OJ
OJ.
0
rt
Hl
13
0
fD
X
P
rt
tr
fD
P
3
O.
rt
P
•<
rt
O
0
P
OJ
fD
P
rt
0
3
0
3
1
3
0
<
fD
3
(D
3
rt
P
fD1
P
P
rt
P0
3
CO
^
rt
tr
fD
Hl
0
p
3
fD
P
cr
fD
p3
<-Q
3
3
** 0
P
fD
3
fD
P
3
P-
3
uq
Hi
C
H
P
fD
P
(ü
CO
rt
P
p-
o
rt
(D
OJ
C
3
P-
13
P
fD
Hl
OJ
>< P0
1
rt
tr
P
3
rt
tr
fD
co
3*
0
S3
co
rt
3"
P
rt
rt
tr
fD
P
fD
p
p
(D
P-
3
13
0
P
rt
P
3
rt
DJ
i-3
tr
fD
P
3
P
M
•<
CO
p-
co
0
rt
tr
fD
<
fD
P3
3
p-
3
DJ
rt
tr
fD
3
P
3
H
*J
•*
fD
CO
'0
fD
O
P-
P
PCO
rt
P-
3
O
3
131
P
p-
O
prt
p-
3
cr
fD
rt
S3
fD
fD
3
fö
3
rt
P
3
OJ
CO
rt
P
C
O
rt
0
P
fD
CO
^
S3
tr
P
rt
DJ
H1
P
fD
r t i-3
S3 0
fD 13
fD PO
3
OJ
p-
fD
3
>< 0
<
0
CD
cr H l 3
P
3
0
O
3
P-
fD
fD
3
rt
CO
13
P
fD
O
(D
rt
CO
3
0
<
fD
3
>
pJ
OJ
O
fD
P
P-
O
tr
P
Hl
Hl
Hl
(D
P
(D
3
H
O
0
P
fD
hh
fD
P
fD
3
0
fD
•
CO
PCO
P
3
0
rt
tr
fD
P
CO
z
rt
P
C
O
rt
0
P
fD
co
rt
P
0
o
rt
0
P
fD
rt
tr
P
rt
O
P
3
CT
fD
p-
co
•<
P-
P-
3
3
<
M
P-
P
3
P
P
P
3
rt
0
3
CO
tr
0
co
0
3
O
rt
p0
3
H
H
3
0
rt
^<
<
P
p
rt
pO
•
Hl
CO
3
3
P
P
rt
O
fD
P
uQ
p-
13
P
p-
0
3
P-
13
P
P
i-3
pelp-
en
«
p-
Cfl
en
~
co
p
3
13
0
Prt
.
••
—J.
00
Prt
P-
co
P-
Ol
3
rt
*-" 3p j
. *i
•
0
*<
CO
et
P
0
3
'<
o
O
rt
p-
0
3
rt
tr
P
rt
3
P
3
fD
co
S3
0
0
pj
0.
O
rt
co
0
0
P u3
P 1 kQ
P
fD
CO
rt
fD
H
3
co
^
rt
^
rt
tr
(D
co
fD
z
Hl
P-
fD1
13
P
P
P1
OJ
P
P1
en
fD
P
•*
=
H3
0
13
P-
O
M
tc
0
3
iQ
P
3
P
P
OJ
P-
*]
P
3
O
O
0
CO
^
S3
tr
p-
rt
S3
0
co
(ü
fD
P
Hl
P-
5*4
p-
P
3
3
fD1
co
co
P
5T
p3
iQ
CO
fD
0
Hl
CO
fD
3
P
3
rt
PO
^ - N
P
OJ
CO
0
tr
1
iQ
0
P
vQ
fD
^~^
VD
00
cn
rt
—i
H
3
vO
0
•*1
co
tr
fD
DJ
13
0
p
fD
H
•<
tr
fD
OJ
fD
en
0
P
P-
fD
OJ
cr
p3
0
rt
0
3
pj
p-
5V
fD
3
13
0
P
rt
P
3
rt
O
O
cr
0
<+
p0
3
•
^<
O
P
3
P
O
O
p-
0
3
0
Hl
fD
X
131
P
fD
rt
p-
<
O
t-3
rt
0
P
P
0
13
0
3
0
p
rt
0
rt
tr
fD
H
P1
3
fD
P
3
PO
CO
^< r t
• P
0
H
O
3
rt
0
P
fD
cn
p-
cn
fD
P
3
<
p-
co
O
P
3
cr
fD
3
p-
rt
tr
fD
0
P
P
3
O-
P
3
P
P-
en
rt
P
0
rt
Hl
r t 'X3
3" P
fD fD
P1
tr
fD
3
cr
DJ
P
fö
en
»13
en fD
tr o
CD r t
en ^
cr
i-3
>
rt
tr
fD
CO
p-
cn
rt
P
0
co
fD
P
rt
tr
P
rt
P-
fD
rt
rt
S3
0
PCO
OJ
O
0
3
rt
rt
tr
PCO
rt
tr
fD
p-
O.
*-*
p
— ppj
. tQ3
en
0
0
co P-
OJ
P
P-
—•»
O
P
3
P
3
\0
DJ
P-
rt
0
S3
prt
tr
0
0
rt
0
= ••Q
fD
P- ^~. ^
O
0
3
3
'a 13 P <
P0
•O 3OJ
P
rt
P
3
rt
H
OJ
ÜJ.
0
P
13
0
P
fD
O
H
0
cr
fD
co
0
OJ
cr
P
SJ
S3
3
Hl
P
0
3
rt
tr
P-
co
fD
P
P
3
P
rt
'<
OJ
M
s3
p-
rt
tr
pj
rt
M
fD
OJ
TO
03
P-
rt
co
ü
0
rt
O
tr
fD
^
0
P-
p-
Hl
Hl
<
fD
P
fD
3
P
O
fD
p1
fD
3
rt
CO
*w
CO
0
P
3
Hl
DJ
Hl
cn
rt
tr
fD
p-
0
0
3
0
P
DJ
co
p-
Hl
P
3
OJ
rt
tr
P
rt
H
O
p
P
3
pj
pj
DJ
fD
OJ
H
CTl
3
*
rt
tr
p.
co
s
H
H
H
H3
•
tr
(D
Cd
P
P
rt
3
p-
fD
0
fD
DJ
P
3
fD
OJ
H
i-3
3P
CD
CO
p-
pj
CO
rt
tr
P
rt
^
s3
fD
tr
P
<
fD
vO
P-
<
fD
3
P
3
P
O
O
0
0
3
rt
0
Hl
rt
tr
fD
OJ
PCO
rt
P
pl
0
13
PO
<
fD
CO
cn
rt
P
0
O
rt
0
P
fD
cn
Hl
P
<
0
p
co
p
3
P1
P
•<
co
tr
0
3
rt
tr
P
rt
P
3
P
3
P
pj
•<
co
p-
co
0
Hl
p-
co
^O l
• — •
H
H
0
<
fD
P
^-^
S3
tr
p
rt
H
tr
P
<
fD
Co
3
3
ft
hä
«i
*
- 225 FOOTNOTES.
* I would like to thank Hans den Besten, Eric Reuland, and Sjaak
de Mey, as well as the participants of the Sixth Groningen
Grammar Talks for discussions about earlier versions of this
paper.
1
INFL is considered to be sentence-final by some authors, e.g.
Reuland (1983), and to be identical to COMP by others, e.g.
Cremers & Sassen (1983), Haider (this volume).
2
Here I only consider Topic structures with d-pronouns, and not
the so-called hanging topic structure exemplified by (i) (cf.
Van Haaften et al. (1983)):
(i) Die man, ik ken hem niet.
(Dutch)
(that man, I know him not)
"That man, I don't know him."
3
In German and Dutch there is a special class of demonstrative
elements that have almost exactly the same behavior as whelements; these I will refer to as d-elements throughout.
Therefore,what is typical of wh-movement also goes for dmovement in German and Dutch.
The class of d-elements contains d-pronouns (exemplified
in exx. (7-10, 14)) and probably also some full NPs that can
refer to a Topic, as in (i) :
(i) Johann, diesen Halunken kann ich nicht ertragen.
(Johann, this rascal can I not bear)
"I can't stand that rascal Johann." (Hubert Haider p.c.)
4
The acceptability of (9)a, (10)a,c also favors II over III,
because such structures are not derivable in III, where there
is no XP preposing.
(20-22) especially occur in southern varieties of German.
c
In northern German, (i) is possible with contrastive stress on
gewollt:
(i)
Gewollt hätte ich den Kuchen backen,
(wanted had I the cake bake)
"What I would have liked was to bake the cake."
7
Clements et al. (1983) have shown that string-vacuous rule application should not be excluded from grammar, so this cannot
be used to rule out (26). This is desirable because of the
grammaticality of sentences like (i):
(i) [g er^ [g [C0Mp kommt.J [g e.. e ± ] ] ]
(he comes)
- 226 -
REFERENCES.
Abraham, W. (ed.) (1983). On the formal
syntax
of the
Westgermania.
Amsterdam.
Bart, P. van & R. Kager (1984). "Er is hier - een alternatief
voor een diagnose." In: De Haan et al. (eds.).
Besten, H. den (1983). "On the interaction of root transformations
and lexical deletive rules." In: Abraham (ed.).
Besten, H. den & J.A. Edmondson (1983). "The verbal complex in
Continental West Germanic." In: Abraham (ed.).
Chomsky, N. (1977). "On WH-movement." In: P. Culicover, T. Wasow
& A. Akmajian (eds.): Formal Syntax.
New York.
Chomsky, N. (1981). Lectures
on Government
and Binding.
Dordrecht.
Chomsky, N. (1982). Some concepts
and consequences
of the
Theory
of Government
and Binding.
Linguistic Inquiry Monograph Six.
Clements, G.N., J. McGloskey, J. Maling & A. Zaenen (1983).
"String-vacuous rule application." Linguistic
Inquiry
14,1:
1-17.
Cremers, C. & A. Sassen (1983). "On V1, GB and INFL." In: H.
Bennis & W.U.S. van Lessen Kloeke (eds.): Linguistics
in the
Netherlands
1983. Dordrecht.
Drewes, H.J., E. Jacobs, F. van de Maagdenberg, J. Veld & M. de
Wolff (1984). "Crossing R-graphs." In: De Haan et al. (eds.).
Evers, A. (1982). "Twee functionele principes voor de regel 'verschuif het werkwoord'." Glot 1: 11-30.
Haaften, T. van, R. Smits & J. Vat (1983). "Left dislocation,
connectedness, and reconstruction." In: Abraham (ed.).
Haan, G.J. de, M. Trommelen & W. Zonneveld (eds.) (1984). Van
periferie
naar kern.
Dordrecht.
Haider, H. (1982). "Dependenzen und Konfigurationen. Zur deutschen
V-Projektion." GAGL 21: i-60.
Haider, H. (this volume). "Topic, focus and verb-second in German."
Kayne, R. (1983). "Chains, categories external to S, and French
complex inversion." Natural
Language and Linguistic
Theory 1:
107-139.
Kiss, K. (1981). "Structural relations in Hungarian, a "free"
word order language." Linguistic
Inquiry
12,2: 185-213.
Koopman, H. & D. Sportiche (1982). "Variables and the Bijection
Principle." The Linguistic
Review 2: 139-160.
Koster, J. (1975). "Dutch as an SOV lanauage." Linguistic
Analysis
1,2: 111-136.
Koster, J. (1978). Locality
principles
in syntax.
Dordrecht.
Koster, J. (1982/1983). "Enthalten syntaktische Repräsentationen
Variablen?" Teil 1, Linguistische
Berichte
80 (1982): 70-100,
Teil 2: Linguistische
Berichte
83 (1983): 36-60.
- 227 Koster, J. (1984). "Die konfigurationelle Matrix." In: H.
Czepluch & H. Janßen (eds.): Syntaktische
Struktur
und
Kasusrelation.
Tübingen.
Lenerz, J. (1982). "Syntaktischer Wandel und Grammatiktheorie.
Eine Untersuchung anhand von Beispielen aus der Sprachgeschichte des Deutschen." Unp. Habilitationsschrift, Univ.
Münster.
Lutz, L. (1981). Zum Thema "Thema". Hamburger Buchagentur.
Platzack, C. (1982). "Germanic word order and the COMP/INFL
parameter." Working Papers in Scandinavian
Syntax 2.
Reuland, E. (1983). "On the subject of nonargument subjects."
In: Abraham (ed.).
Scherpenisse, W. (1984). "Die Satzstruktur des Deutschen und des
Niederländischen im Rahmen der GB-Theorie. Eine Reaktion auf
Marga Reis' Doppelkopfkritik." In: W. Abraham (ed.): Erklärende Syntax
im Deutschen.
Tübingen.
Williams, E. (1984). "There-insertion." Linguistic
Inquiry
15,1:
131-153.
Wim Scherpenisse
Germanistisch Instituut RUG
Grote Rozenstraat 15
9712 TG Groningen
The Netherlands