1. sports
2. bodybuilding

Sandbjergseminar, 10.-11.01.2011
SLK's lingvistiske forskergruppe
Sætningsstruktur og negative polaritetsudtryk
Sten Vikner
Institut for Sprog, Litteratur og Kultur / Engelsk
Aarhus Universitet, DK-8000 Århus C
[email protected] - www.hum.au.dk/engelsk/engsv
Contents
1. Introduction, 1
1.1 X-bar structure, 1
1.2 Clause structure in tree analyses and in field analyses, 3
1.3 C-command, 5
2. Negative polarity items (NPIs), 6
2.1. Assertive vs. nonassertive vs. negative items, 6
2.2. Negated polarity items are not just idioms, 7
2.3. English and Danish NPIs, 8
2.4 Licensing of NPIs requires c-command, 10
3. Conclusion, 14
References, 14
Appendix. Extra examples of c-command, 15
1. Introduction
1.1 X-bar structure
In a generative analysis, syntactic constituents all have the same basic structure, namely the "X-bar
structure" shown in (1) (where the sequence of the head and the complement may vary):
(1)
specifier
MINIMAL PROJECTION
(HEAD)
(2)
MAXIMAL PROJECTION
(PHRASE)
XP
XP
X'
X°
= phrase
= X-bar
= head
X'
X°
complement
/ the maximal projection of X
/ the intermediate projection of X
/ the minimal projection of X (= e.g. a word or an even smaller unit)
Saying that XP and X' are projections of X expresses the idea that these constituents are built up
around X°, such that i.e. [PP across the hall] is built around [P° across].
A head is always the head of its own phrase (its own maximal projection), and all maximal
projections have a head (are endocentric). Inside a phrase, there is also room for two other phrases,
namely in the specifier position and in the complement position.
Vikner: Clause Structure & NPIs, 11.01.2011, p. 1 of 19
The position of the so-called specifier position is normally considered to be fixed, i.e. it is taken
to always be the left daughter of XP. The sequence of the head and the complement may on the other
hand vary, depending on the language.
X above may stand for one of the following categories:
(3)
lexical categories (word classes)
"functional" categories
N (noun)
V (verb)
P
(preposition)
Adj (adjective)
Adv (adverb)
C ("complementiser"
= subordinating conjunction)
I (inflection)
D (determiner)
etc.
Both heads and phrases (minimal and maximal projections) may move. Heads may only move
into other head positions, and phrases may only move into other phrase positions. X-bar constituents
(intermediate projections) may not move at all.
Both heads and phrases may be adjoined to other constituents. Heads may only adjoin to other
heads, and phrases may only adjoin to other phrases. X-bar constituents may not be adjoined at all.
Adjunction takes the following form, where the adjoined constituent, WP/W°, may be adjoined
either to the left, as shown, or to the right of the XP/X° that it modifies:
(4)
a.
XP
modifier
b.
XP
XP
modifier
XP
(ADJOINED
POSITION)
(ADJOINED
POSITION)
In a somewhat simplified generative analysis, the structure of a sentence (irrespective of whether it is
a main or an embedded clause) is as follows:
(5)
A clause is a CP,
the complement of its head (= C°) is an IP, and
the complement of the IP's head (= I°) is a VP.
For a sentence with no auxiliary verb and with a (mono-)transitive main verb the structure looks as
follows:
(6) Da.
CP
C'
XP
IP
C°
DP
I'
I°
VP
DP
V'
V°
DP
Vikner: Clause Structure & NPIs, 11.01.2011, p. 2 of 19
1.2 Clause structure in tree analyses and in field analyses
A fundamental difference between various approaches to clause structure is the one between
• tree analyses like the generative analysis that I advocate and
• field analyses like the sætningsskema analysis of Danish of Diderichsen (1946) and many
others or the topologische Modell analysis of German of Drach (1937) and many others.
The difference is to which extent the parts of the clause are seen as boxes inside other boxes or as
pearls after each other on a string. It is a question of extent, as it extent can neither be 0% 100%: Even
to Diderichsen (1946), not all constituents follow each other (e.g. the object is inside the content
field), and also in the generative analyses constituents may follow one another (otherwise trees would
only contain mothers and daughters and no sisters).
In (7), the two different analysis of Danish are shown, and in (8) the same for German:
(7)
a.
CP
Spec
C'
C°
IP
Spec
I'
I°
VP
AdvP
VP
Spec
V'
V°
VP
VP
Spec
AdvP
V'
V°
b. F
Nexus field
DP
Content field
F
v
n
a
V
N
A
Nu
har
Per
igen
poleret
bilen
med ståluld
Now has
Per
again
polished
car-the
with steel wool
c. Conj. f.
Nexus field
Content field
k
n
a
v
V
N
A
om
Per
igen
har
poleret
bilen
med ståluld
If
Per
again
has
polished
car-the
with steel wool
Vikner: Clause Structure & NPIs, 11.01.2011, p. 3 of 19
In (7), I have taken the structure in (6) and added the possibility that adverbials (etc.) may be adjoined
both on the left side and on the right side of a VP. The result is the generative tree structure in (7)a
which is compared to the simplified Diderichsen field model of constituent order in modern Danish,
Norwegian and Swedish, as illustrated in (7)b for main clauses and in (7)c for embedded clauses.
Collapsing the Diderichsen model for the main clause with the one for the embedded clause, as in
(7)b,c, was not done by Diderichsen himself but by Platzack (1985:71, fn 5) and Heltoft (1986:108).
For more details and more references, see Bjerre, Engels, Jørgensen & Vikner (2008).
In (8), there is a parallel illustration for German, first a generative tree structure in (8)a and then the
simplified field model / topological model analyses of German main and embedded clauses in (8)b,c.
For more details and more references, see Wöllstein-Leisten, Heilmann, Stepan & Vikner (1997).
(8)
a.
CP
Spec
C'
C°
IP
Spec
I'
I°
VP
AdvP
VP
Spec
V'
VP
Spec
V'
DP
Linke
Satzklammer
Vielleicht
hat
Perhaps
has
ob
if
Vorfeld
b.
c.
Mittelfeld
Peter
Peter
Peter
Peter
nie
never
nie
never
V°
ein Auto
a car
ein Auto
a car
V°
Rechte
Nachfeld
Satzklammer
poliert
polished
poliert hat
polished has
(9)
X
daughter of
mother of
Y
Z
DOMINANCE
Finally, in case you are not familiar with the mother, sister and daughter terminology:
sisters
PRECEDENCE
Vikner: Clause Structure & NPIs, 11.01.2011, p. 4 of 19
1.3 C-command
All constituents in the tree structure (except the very highest CP) are part of a larger constituent,
whereas the field models contain a number of fields which are not part of a larger field. This makes it
possible to define the following relation ("C-command") in the generative structure:
(10) C-command:
X c-commands Y if and only if
a. all constituents that contain X also contain Y,
b. neither X nor Y dominates the other.
In other words: if you can get from X to Y in the tree by taking one step upwards and then climb
downwards the rest of the way, then X c-commands Y.
C-command may be used to make the following apparently universal generalisation: A pronoun and a
DP may not be coreferential, if the pronoun c-commands the DP:
(11) En. a.
b.
NAME C-COMMANDS PRONOUN, COREFERENCE POSSIBLE
John thinks he is intelligent
He thinks John is intelligent
PRONOUN C-COMMANDS NAME, COREFERENCE IMPOSSIBLE
IP1
(12)
DP
John
I'
I°
VP
V'
V°
thinks
CP
C'
C°
that
IP2
DP
he
I'
I°
is
VP
V'
V°
AdjP
intelligent
Such generalisations would seem much more difficult to formulate within field model analyses.
I want to show, with illustrations from the area of negation and negative polarity items, that a purely
linear rule would not be able to make the right distinctions.
First, however, we need to examine negative polarity items in detail.
Note:
As for the actual position of negation, it will suffice for the purposes of this talk
to assume that it is left-adjoined to VP, just like igen in (7) and nie in (8).
owever, what I shall say is also comptatible with the assumption of a Negation
Phrase (NegP) along the lines sketched by Johanna in the following talk.
Vikner: Clause Structure & NPIs, 11.01.2011, p. 5 of 19
2.
Negative polarity items (NPIs)
2.1 Assertive vs. nonassertive vs. negative items
Quirk et al. (1985:782) gives the following list of items that fall into one of three groups that
they call assertive, nonassertive or negative items.
(13)
Assertive
some
something
somebody
someone
somewhere
sometimes
already
still
too
to some extent
Nonassertive
any
anything
anybody
anyone
anywhere
ever
yet
any more
/any longer
either
at all
Negative
no / none
nothing
nobody
no one
nowhere
never
no more
/ no longer
neither
not at all
Assertive items do not change the polarity of a sentence:
(14)
En. a. Arnold just bought "Pride and Prejudice", didn't he?
b. *Arnold just bought "Pride and Prejudice", did he?
(15)
En. a. Arnold just bought some books, didn't he?
b. *Arnold just bought some books, did he?
Negative items turn a positive sentence into a negative one:
(16)
En. a. *Arnold didn't buy "Pride and Prejudice", didn't he?
b. Arnold didn't buy "Pride and Prejudice", did he?
(17)
En. a. *Arnold bought no books this week, didn't he?
b. Arnold bought no books this week, did he?
Nonassertive items do not change the polarity of a sentence either, but they require the sentence
to be negative (certain other contexts work as well: questions, conditional clauses,
comparatives, superlatives, ...):
(18)
En. a. *Arnold bought
any books this week, didn't he?
b. Arnold didn't buy any books this week, did he?
Quirk et al.'s (1985:782) nonassertive items in (13) are part of what is otherwise known as
"negative polarity items" (NPIs).
Vikner: Clause Structure & NPIs, 11.01.2011, p. 6 of 19
2.2 Negated polarity items are not just idioms
The adverbial expression En. at all / Da. overhovedet / Ge. überhaupt is a so-called "negative
polarity item" (NPI), as they would seem to need a negative element to be possible (or a
question/conditional/comparative ...). However, at first glance one might simply think this
dependence on negation comes from at all being part of an idiom that also includes not:
(19)
En. a. *Arnold
understands French at all
b. Arnold does not understand French at all
Da. c. *Arnold forstår
fransk overhovedet
d. Arnold forstår ikke fransk overhovedet
Ge. e. *Arnold versteht überhaupt
Französisch
f. Arnold versteht überhaupt nicht Französisch
But this would predict the following to be ungrammatical, as there is no not:
(20)
En. a. *Some students in this class understand German at all
b. No
students in this class understand German at all
Da. c. *Nogle studerende på det her hold forstår tysk overhovedet
d. Ingen studerende på det her hold forstår tysk overhovedet
Ge. e. *Einige Studenten in diesem Seminar verstehen überhaupt Deutsch
f. Keine Studenten in diesem Seminar verstehen überhaupt Deutsch
(21)
En. a. *Arnold understands one of the Scandinavian languages at all
b. Arnold understands none of the Scandinavian languages at all
Da. c. *Arnold forstår et
af de skandinaviske sprog overhovedet
d. Arnold forstår ingen af de skandinaviske sprog overhovedet
Ge. e. *Arnold versteht überhaupt eine von den skandinavischen Sprachen
f. Arnold versteht überhaupt keine von den skandinavischen Sprachen
(22)
En. a. *Arnold sometimes speaks German at all
b. Arnold never
speaks German at all
Da. c. *Arnold taler nogle gange tysk overhovedet
d. Arnold taler aldrig
tysk overhovedet
Ge. e. *Arnold spricht überhaupt ab und zu Deutsch
Deutsch
f. Arnold spricht überhaupt nie
The idiomatic expression hypothesis would run into even bigger problems with examples that
are not negative in any way at all, like yes/no-questions:
(23)
En. a.
b.
c.
Does
Arnold
understand French at all?
Does
Arnold not understand French at all?
Doesn't Arnold
understand French at all?
Da. d.
e.
fransk?
Forstår Arnold overhovedet
Forstår Arnold overhovedet ikke fransk?
Ge. f.
g.
Französisch?
Versteht Arnold überhaupt
Versteht Arnold überhaupt nicht Französisch?
Vikner: Clause Structure & NPIs, 11.01.2011, p. 7 of 19
2.3 English and Danish NPIs
Here are a few English NPIs (in addition to the nonassertive items from (13) above):
(24)
all that + adj/adv
a red cent
stand (as verb)
bat an eyelid
bother to do something
budge an inch
do a thing about something
drink a drop
give a damn about something
hold a candle to someone
move a muscle
say a word about something
see a living soul
sleep a wink
And here are two lists of negative polarity items in Danish (the result of a quick check of the
electronic version of NuDansk Ordbog). In the first list the NPIs can be used both in negated
clauses and in questions:
(25)
a. Arguments
nogetsomhelst
skyggen af noget
en brik / en bønne / en disse / en dyt / en døjt / en flyvende fis / et hak / et klap / et
kuk / et kvidder / et kvæk / en lyd / et meter / et muk / en pind / et pip / en
pløk / en skid / en snus / spor / en stavelse / et suk
(money only) en hvid / en klink / en rød reje / en rød øre
b. Adjectives / Adverbs
nævneværdig(t)
synderlig(t)
c. Place Adverbials
nogen steder
nogetsteds
d. Time Adverbials
nogensinde
på noget tidspunkt
endnu
e. Measure adverbials
overhovedet
på nogen måde
i mindste måde
i det hele taget
så meget som ....
Vikner: Clause Structure & NPIs, 11.01.2011, p. 8 of 19
f. Verbs
behøve (with infinitive only)
fordrage
døje
orke
g. Full VPs
falde nogen ind at ...
finde hoved og hale på noget
fortænke nogen i noget
have en jordisk chance
have noget at skulle have sagt
kimse ad noget
lade noget sidde på sig
lade sig mærke med noget
se en hånd for sig
skulle nyde noget
tro sine egne øjne
vokse på træerne
være at foragte
være helt appelsinfri
være med nogens gode vilje
være nogens afdeling
være nogens kop te
være noget at rafle/tale om
være noget galt i at ...
være nogen ben i noget
være på talefod
være rigtig klog / være vel forvaret
være så dårligt/slemt
være til at blive klog på
være tilfældet at ...
In the second list of negative polarity items in Danish, the NPIs can be used in negated clauses
but not in questions:
(26)
a. Adjectives (inside arguments)
det mindste (maybe not so bad in yes/no-questions)
det fjerneste (maynot so bad in yes/no-questions)
b. Adverbials
ligefrem (not sooo bad in yes/no-questions)
heller
c. Full VPs
give fem flade øre for noget
give noget ved dørene
have en trevl på kroppen
have opfundet den dybe tallerken
have opfundet det varme vand
have opfundet krudtet
have tone i livet
Vikner: Clause Structure & NPIs, 11.01.2011, p. 9 of 19
krumme et hår på nogens hovede
kunne gøre en kat fortræd
kunne tage/snuppe/udstå noget
lægge fingrene imellem
lægge skjul på noget
lukke et øje
løfte en finger
sige noget to gange
stikke op for bollemælk
tro nogen over en dørtærskel
ville røre noget med en ildtang
være bleg for at ...
være lutter lagkage
være mange sure sild værd
være tabt bag af en vogn
være til at stå for
være ved sine fulde fem
2.4 Licensing of NPIs requires c-command
The negative polarity element must be c-commanded by the licenser, e.g. by the negative
element (e.g. Fromkin 2000: 223, 404, though see Hoeksema 2000 for problems with this
analysis).
The subject position (IP-spec) is not c-commanded by the negation, but the position of the
logical subject ("the associate") in there-constructions is. According to Vikner (1995:203-207),
this position is actually the object position, at least in Danish:
(27)
standing outside the door ...
En. a. *Because anybody wasn't
b. Because there
wasn't anybody standing outside the door ...
c. *Because there
was
anybody standing outside the door ...
Da. d. *Fordi nogensomhelst ikke stod
uden for døren ...
e. Fordi der
ikke stod nogensomhelst uden for døren ...
f. *Fordi der
stod nogensomhelst uden for døren ...
The subject c-commands the object, but not vice versa:
(28)
En. a. No one understood anything
b. *Anyone understood nothing
Da. c. Ingen
forstod nogetsomhelst
d. *Nogensomhelst forstod ingenting
Vikner: Clause Structure & NPIs, 11.01.2011, p. 10 of 19
The negative polarity element can also be licensed by a negative element in a different clause,
provided there is c-command:
(29)
En. a. *Because he
thought that I would ever understand it
that I would ever understand it
b. Because he didn't think
c. *Because he ever
thought that I would not understand it
Da. d. *Fordi han
troede at jeg nogensinde ville forstå det
troede at jeg nogensinde ville forstå det
e. Fordi han ikke
f. *Fordi han nogensinde troede at jeg ikke
ville forstå det
Ge. g. *Weil er
geglaubt hat, dass ich es jemals verstehen würde
h. Weil er nicht geglaubt hat, dass ich es jemals verstehen würde
i. *Weil er jemals geglaubt hat, dass ich es nicht verstehen würde
A topicalised object is not c-commanded by the negation (i.e. the situation that counts is the one
after movement of the object from the object position to the left edge of the clause):
(30)
En. a.
Arnold did not understand a single word
b. *A single word Arnold did not understand
Da. c. Arnold forstod
ikke et klap
d. *Et klap forstod Arnold ikke
Ge. e.
Arnold
Arnold
f. *Einen
A
g.
hat
has
keinen Ton verstanden
no
tone understood
Ton hat Arnold nicht
tone has Arnold not
Keinen Ton hat Arnold
No
tone has Arnold
verstanden
understood
verstanden
understood
In the following, I shall compare the c-command condition on NPI-licensing with an alternative
condition that requires precedence instead. In the first set of examples the NPI ever is both ccommanded and preceded by the licenser:
(31)
a. Da.
b. En.
c. Ge.
Selvom vi altid forberedte os, bestod ingen nogensinde eksamen
Though we always did our homework, nobody ever passed the exam
Obwohl wir uns immer vorbereitet haben, hat niemand die Prüfung
jemals bestanden.
If the NPI is preceded but not c-commanded by the licenser, it is not possible:
(32)
a. Da. *Selvom vi aldrig forberedte os, bestod alle *nogensinde eksamen
b. En. *Though we never did our homework, everybody *ever passed the exam
c. Ge. *Obwohl wir uns nie vorbereitet haben, haben alle die Prüfung
jemals bestanden.
The opposite case (c-command without precedence) is not straightforwardly possible, at least
not in the Germanic and the Romance languages.
Vikner: Clause Structure & NPIs, 11.01.2011, p. 11 of 19
In the following set of examples the NPI is both c-commanded and preceded by the licenser,
although the licenser is in a different clause:
(33)
a. Da.
b. En.
c. Ge.
Hvis du aldrig forbereder dig, kan du ikke forvente
[at du nogensinde vil bestå eksamen]
If you never do your homework, you cannot expect
[that you will ever pass the exam]
Wenn du dich nie vorbereitest, kannst du nicht erwarten,
[dass du die Prüfung jemals bestehen wirst]
If the embedded clause in the above examples is topicalised, then the NPI is neither ccommanded nor preceded by the licenser (cf. also the examples in (30)):
(34)
a. Da. *[At du nogensinde vil bestå eksamen], kan du ikke forvente, hvis
du aldrig forbereder dig
b. En. *[That you will ever pass the exam], you cannot expect, if you
never do your homework
c. Ge. *[Dass du die Prüfung jemals bestehen wirst], kannst du nicht
erwarten, wenn du dich nie vorbereitest
Consider a sligthly different set of examples where the NPI a word is both c-commanded and
preceded by the licenser:
(35)
a. Da.
b. En.
c. Ge.
Tidligere forsøgte ingen at forstå et ord i timerne
Earlier nobody tried to understand a word during the classes
Früher hat niemand versucht im Unterricht ein Wort zu verstehen
In the following examples, the NPI is preceded but not c-commanded by the licenser, because
the negation is inside the topic:
(36)
a. Da. *[For ikke ret lang tid siden] forsøgte jeg at forstå et ord i
timerne
b. En. *[Not long ago] I tried to understand a word during the classes
c. Ge. *Vor nicht langer Zeit] habe ich versucht im Unterricht ein Wort
zu verstehen
This contrasts with the following examples, where the NPI is preceded and c-commanded by
the licenser, because here the entire topic counts as negative:
(37)
a. Da.
b. En.
c. Ge.
[På intet tidspunkt] forsøgte jeg at forstå et ord i timerne
[At no point in time] did I try to understand a word during the
classes
[Zu keiner Zeit] habe ich versucht im Unterricht ein Wort zu
verstehen
Vikner: Clause Structure & NPIs, 11.01.2011, p. 12 of 19
Finally, I want to look at a type of NPI where c-command might seem not to be required, NPIverbs. The NPI-verbs are auxiliary need in English (with an infinitive without to) and
behøve/brauchen in Danish and German (when they embed an infinitive, the behøve/ brauchen
that take a DP-object are not NPI-verbs, see (39)). In none of (38a,c,e) are the NPI-verbs ccommanded by the negation. But still, it cannot be the case that these verbs do not need to be
licensed by e.g. a negation, as shown by the ungrammaticality of (38b,d,f). Instead, I think that
what is necessary is that the NPI-verb was c-commanded by the licenser before movement:
(38)
En. a. Maybe Arnold need not read these books
read these books
b. *Maybe Arnold need
Da. c. Måske behøver Arnold ikke læse de her bøger
læse de her bøger
d. *Måske behøver Arnold
Maybe needs
Arnold (not) read these books
Ge. e. Vielleicht braucht Arnold diese Bücher nicht zu lesen
zu lesen
f. *Vielleicht braucht Arnold diese Bücher
Maybe
needs
Arnold these books (not) to read
(39)
Da. a.
b.
Måske behøver Arnold ikke de her bøger
Måske behøver Arnold
de her bøger
Maybe needs
Arnold (not) these books
Ge. c.
d.
Vielleicht braucht Arnold diese Bücher nicht
Vielleicht braucht Arnold diese Bücher
Maybe
needs
Arnold these books (not)
Constituent negation inside the object does not c-command the verb itself which is why the
NPI-verbs stand/fordrage/abkönnen are only possible with sentential negation, (40), not with
constituent negation, (41):
(40)
En. a. Arnold can't stand vodka
b. *Arnold can
stand vodka
Da. c. Arnold kan ikke fordrage vodka
d. *Arnold kan
fordrage vodka
Ge. e. Arnold kann Wodka nicht ab
f. *Arnold kann Wodka
ab
(41)
En. a. *Arnold can
stand [not vodka but tequila]
b. Arnold drinks
[not vodka but tequila]
Da. c. *Arnold kan
fordrage [ikke vodka men tequila]
d. Arnold drikker
[ikke vodka men tequila]
[nicht Wodka sondern Tequila] ab
Ge. e. *Arnold kann
f. Arnold trinkt [nicht Wodka sondern Tequila]
Also constituent negation inside an adverbial does not c-command the object:
(42)
En. a. *Arnold understood a single word
b. Arnold understood something
[not long ago]
[not long ago]
Da. a. *Arnold forstod et kvæk [for ikke ret lang tid siden]
b. Arnold forstod noget
[for ikke ret lang tid siden]
Vikner: Clause Structure & NPIs, 11.01.2011, p. 13 of 19
3. Conclusion
The objective here was mainly to illustrate a particular difference (c-command) between two kinds of
approaches to clause structure, namely tree analyses like the generative analysis that I advocate and
field analyses like the sætningsskema analysis of Danish of Diderichsen (1946) and many others or
the topologische Modell analysis of German of Drach (1937) and many others.
Having said this, I hope that the first part of my talk also illustrated that these approaches have a
number of properties in common. I continue to believe that syntacticians would be well advised to
look further than the surface of the different formal and functional approaches. Despite the
occasionally polemic tone, the various approaches actually have much in common, which also means
that they may learn from each other's insights.
All syntacticians, regardless of theoretical persuasion, are ultimately interested in explaining
language data. Given the complex subject matter of the discipline, we need all the help we can get,
and therefore none of us can afford to ignore the results reached within ‘the opposite camp’.
It should be emphasised that this does not mean that linguists should forget all the differences
between the two approaches, but merely that they should not forget that in spite of such differences,
there are areas where the two approaches can learn from each other and build on each others' insights.
At the end of the day, linguists from the two approaches will still set out in different directions
when it comes searching for an explanation, and this is as it should be, given that "the growth of
knowledge depends entirely upon disagreement" (Popper 1994:x).
This quote is further explained in Popper (1994:93-94): "Since the method of science is that of
critical discussion, it is of great importance that the theories discussed should be tenaciously defended.
For only in this way can we learn their real power. And only if criticism meets resistance can we learn
the full force of a critical argument."
References
Bjerre, Tavs, Eva Engels, Henrik Jørgensen, & Sten Vikner:2008, "Points of convergence between
functional and formal approaches to syntactic analysis" in Working Papers in Scandinavian
Syntax 82, 131-166, <www.hum.au.dk/engelsk/engsv/papers/bjer08a.pdf>
Diderichsen, Paul: 1946, Elementær Dansk Grammatik, Gyldendal, Copenhagen. 3rd edition 1962,
Reprinted 1984.
Drach, Erich: 1937, Grundgedanken der deutschen Satzlehre, Diesterweg, Frankfurt am Main.
Reprinted 1963, Darmstadt: Wissenschaftliche Buchgesellschaft.
Fromkin, Victoria (ed.): 2000, Linguistics, An Introduction to Linguistic Theory, Blackwell, Oxford.
Heltoft, Lars: 1986, "Topologi og syntaks. En revision af Paul Diderichsens sætningsskema" in Lars
Heltoft & John E. Andersen (eds.), Sætningsskemaet og dets stilling - 50 år efter, Nydanske
Studier og Almen Kommunikationsteori, NyS 16/17, 105-130.
Hoeksema, Jack: 2000, "Negative Polarity Items, Triggering, Scope and C-Command", in Laurence
Horn & Yasuhiko Kato (eds.), Negation and Polarity - Syntactic and Semantic Perspectives,
Oxford University Press, Oxford, pp 115-145.
Platzack, Christer: 1985, "A Survey of Generative Analyses of the Verb Second Phenomenon in
Germanic" in Nordic Journal of Linguistics 8.1, 49-73.
Popper, Karl: 1994, The Myth of the Framework, Routledge, London.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, & Jan Svartvik: 1985, A Comprehensive
Grammar of the English Language, Longman, London.
Vikner, Sten: 1995, Verb Movement and Expletive Subjects in the Germanic Languages, Oxford
University Press, New York.
Wöllstein-Leisten, Angelika, Axel Heilmann, Peter Stepan, & Sten Vikner: 1997, Deutsche
Satzstruktur, Stauffenburg, Tübingen.
Vikner: Clause Structure & NPIs, 11.01.2011, p. 14 of 19
Appendix. Extra examples of c-command
In this appendix, you will find more examples of c-command, this time with trees, drawn from the
area of coreference in Danish.
As in (11) above, the baisc rule is (still) that a pronoun and a DP may not be coreferential, if the
pronoun c-commands the DP:
In this way, the difference between the various examples in this section may be accounted for,
whereas a purely linear rule would not seem to be able to make the right distinctions.
(43) Da.
CP
C'
PP
C°
var3
P'
P°
Uden
IP
DP
Anne
og Bo1
CP
C'
C°
at
IP
DP
de1
I'
I°
Spec
t1
VP
V'
I'
I°
t3
VP
DP
t1
V'
V°
t3
VP
V'
DP
t1 V°
blevet
DP
t1
VP
V'
V°
DP
V°
DP
t1
fotograferet
vidste det
Without that they knew it had Anne and Bo been photographed
In (43) at least one constituent contains de 'they' without also containing Anne og Bo, 'Anne and Bo',
e.g. the embedded clause that de is the subject of.
In (44), on the other hand, no constituents contain de, 'they', without also containing Anne og Bo,
'Anne and Bo'.
Thus the pronoun c-commands the name in (44) but not in (43).
Vikner: Clause Structure & NPIs, 11.01.2011, p. 15 of 19
(44) Da. *Jeg tror ikke ...
CP
C'
C°
at
IP
DP
de1
I'
I°
VP
Spec
t1
V'
V°
vidste
C°
at
CP
C'
IP
DP
Anne
og Bo1
I'
I°
VP
DP
t1
V'
V°
var
VP
V'
DP
t1 V°
blevet
DP
t1
VP
V'
V°
DP
t1
fotograferet
I think not that they knew that Anne and Bo had been photographed
Similarly, in (45) at least one constituent contains deres, 'their', without also containing Anne og Bo,
'Anne and Bo', e.g. the DP of which deres is the determiner
In (46) on the other hand, no constituents contain deres, 'their', without also containing Anne og Bo,
'Anne and Bo'.
Thus the pronoun c-commands the name in (46), but not in (45).
Vikner: Clause Structure & NPIs, 11.01.2011, p. 16 of 19
(45) Da.
CP
C'
PP
C°
var3
P'
P°
Uden
DP
IP
DP
Anne
og Bo1
D'
D°
deres1
I'
I°
t3
NP
VP
DP
t1
N'
V'
V°
t3
N°
tilladelse
VP
V'
DP
t1 V°
blevet
DP
t1
VP
V'
V°
fotograferet
DP
t1
Without their permission had Anne and Bo been photographed
(46) Da. *Jeg har ikke hørt om...
DP
D'
D°
deres1
NP
N'
PP
N°
tilladelse
P'
P°
til
C°
at
CP
C'
IP
DP
Anne
og Bo1
I'
VP
I°
DP
t1
V'
VP
V°
kunne DP
V'
t1
V°
DP
t1
fotograferes
I have not heard about their permission to that Anne and Bo could be-photographed
Vikner: Clause Structure & NPIs, 11.01.2011, p. 17 of 19
Another generalisation involving c-command is that a DP may only be coreferential with a pronoun
that it c-commands if the pronoun is inside a different clause.
In (47), Anne og Bo, 'Anne and Bo', and dem, 'them', are in the same clause and Anne og Bo
c-commands dem. The example is therefore correctly expected to be ungrammatical.
(47) Da. *Jeg tror ikke...
CP
C'
C°
at
IP
DP
Anne
og Bo1
I'
VP
I°
DP
t1
V'
V°
har
VP
DP
t1
V'
V°
DP
fotograferet dem1
I think not that Anne and Bo have photographed them
In (48) Anne og Bo, 'Anne and Bo', and dem, 'them', are also in the same clause, but here Anne og Bo
does not c-command dem. The example is therefore correctly not expected to be ungrammatical.
(48) Da. Jeg tror ikke...
CP
C'
C°
at
IP
DP2
I'
D'
D°
det
I°
NP
N'
N°
billede
PP
P'
P°
af
VP
DP
t2 V°
vil
V'
VP
DP
t2
V'
V°
imponere
DP
dem1
DP
Anne
og Bo1
I think not that that picture of Anne and Bo will impress them
Vikner: Clause Structure & NPIs, 11.01.2011, p. 18 of 19
In (49) Anne og Bo, 'Anne and Bo', does c-command dem, 'them', but as the two expressions are not
inside the same clause, this example is therefore correctly not expected to be ungrammatical either.
(49) Da. Jeg tror ikke ...
CP
C'
C°
at
IP
DP
Anne
og Bo1
I'
I°
VP
Spec
t1
V'
V°
vidste
C°
at
CP
C'
IP
DP
de1
I'
I°
VP
DP
t1
V'
V°
var
VP
V'
DP
t1 V°
blevet
DP
t1
VP
V'
V°
DP
t1
fotograferet
I think not that Anne and Bo knew that they had been photographed
Such generalisations, which are not only valid for Danish, would seem much more difficult to
formulate within field model analyses.
Vikner: Clause Structure & NPIs, 11.01.2011, p. 19 of 19