Cecilia Poletto Goethe Unversität Frankfurt Universitá degli studi di Padova The special status of negation Negation is the only operator which is ID: 830476
Download The PPT/PDF document "Negation in natural language" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Negation in natural language
Cecilia
Poletto
Goethe
Unversität
Frankfurt
Universitá degli studi di Padova
Slide2The special
status of negation
Negation
is
the
only
operator
which
is
universally
lexically
marked
and
cannot
be
substituted
by
a
syntactic
operation
(like
verb
movement
),
as
noticed
by
Moro (2015)
Negation
is
the
only
operator
that
triggers
an
agreement
procedure
like negative
concord
Slide3Negation running
wild
(1)
Ní
chuireann
sé
isteach
ar
phostanna
(
Irish
,
Mc
Closkey
2017:9)
Neg-fin
put.pres
he in on
jobs
He
does
not
apply
for
jobs
(2) Gianni
non
mangia
Gianni not
eats
Gianni
does
not
eat
(3) Alex hat das Buch
nicht
gelesen.
Alex
has
the
book
not
read
`Alex
didn’t
read
the
book
.’
(4) Alex hat
kein
Buch gelesen.
Alex
has
no
book
read
`Alex
didn’t
read
a
book
.’
Slide4˥ P?
Negation in
natural language
does
not
seem
to
reflect
the
expected
syntax
give
negation
in formal
logic
:
The
neg
-first
principle
(
Corblin
&
Tovena
2011)
is
systematically
violated
in
the
majority
of
the
languages
of
the
world
.
It
also
poses
a
problem
for an account
which
considers
the
semantic
operation
of
negating
a
clause
performed
in
natural
languages
as
the
same
of
formal
logic
.
Slide5Setting the problem...
The multiple positions of sentential negative markers constitutes a problem for the cartographic “one head=one feature” approach, where each projection has different semantic import.
Slide6Setting the problem…
The multiple forms and positions that negation can take in
natural
language
have
been
explained
(
cf
.
Zeijlstra
2007) by
assuming
that
at
least
some of
them
are
not
truly
negative and
that
the
real
negation
is
an
empty
operator.
I
will
propose
that
none of
them
is
negative, and
that
there
is
no
abstract
negative operator in the
clause
.
Slide7Negation is not a primitive
In
this talk I
will
try
to
pursue
an alternative account and
explore
the idea
that
the
operation
of
negating
a
clause
does
not
correspond
to the
one
of
formal
logic
,
but includes a series of semantic operations, some of which are also found in other configurations like Focus and quantification of the existential type.
Slide8A compositional
viewIf
negation
is
not a primitive but a
complex
set
of
operations
,
we
expect
languages
to
mark
one
or
the
other
operation
or
more
than
one
.
Each
of
these
operations
can
„stand
for
“
the
whole
cluster
and
thus
be
interpreted
as
negation
although
it
is
only
part
of
it.
Slide9Scattering
negation
This view
is
similar
to
the
one
proposed
by
Portner
and
Zanuttini
for
sentence
typing
:
the
phenomenon
is
compositional
and
involves
various
levels
of
the
sentence
, not
only
the
CP.
The internal
structure
of
„
NegP
“
corresponds
to
the
external
one
in
the
clause
.
Each
internal
projection
needs
to
be
checked
at
the
appropriate
level
in
the
clause
.
Slide10Negation in the
vP
I will provide
evidence
that
the
cluster
of
projections
of
the
„
big
NegP
“
starts
out in
the
vP
and
then
can
raise
to
the
positions
in
the
clause
that
correspond
to
the
semantic
operations
included
in
negation
.
It
has
been
proposed
(
see
Manzini
and
Savoia 2011, Bayer 2010)
that
negation
is
merged
in
the
direct
object
position
. I do not
have
any
specific
evidence
on
this
.
Slide11NegP does not exist
The set of semantic operations that are necessary to negate a clause are translated into syntax as a set of functional projections occurring at different levels of the syntactic tree.
E
ach of these operations is common to other operators.
NegP
does not exist as a single projection, this is only a cover term for a set of projections much like IP or CP.
Slide12A modular theory of complex operations
The different lexical realizations of negation in different languages are a function of the fact that only some of the internal projections of the complex "big NegP" are spelled out.
The different positions of the negative markers we observe are the same we find when the atomic parts compounding negation are used for other purposes.
Slide13Why is there so much variation?
All negative markers
lexicali
ze features
which are internal to
a single complex “NegP”
and correspond to one of the semantic operations performed to obtain sentential negation.
Slide14This idea can potentially
explain:
Why in different languages the negative marker has a different etymology and a different position
Why in some languages you can find two, three or even four "negative markers" for one single semantic negation.
Slide15The Northern side of the puzzle
The Northern Italian dialects display four different types of negative markers located in four different positions and with different syntactic properties with respect to intervention in the verb movement path, clustering with clitics, negative concord, ordering with respect to adverbs.
Why is it so?
Slide16Clitic element
a. No
sai
(
Cencenighe
Agordino
(BL))
(
I) Not
know
b.
*
No
vienlo
?
(
Padua
)
Not
comes=he
c.
*
No
va
not
go+imperative
d.
No
'l è
lugà
nogugn
Not
he is come nobody
e
.
Nisun
no
vien
più
casa
mia
(
Venice)
Nobody
not comes more my home
f. U
min
sent
nent
/U
n li
sent
nent
(Cairo)
He
me.not
hears not/He not him hears not
Slide17Zanuttini’s findings
The preverbal clitic has the following properties
It interferes with V to C in interrogatives
It is not compatible with morphologically unambiguous imperative forms
It always requires negative concord with
postverbal
n-words (in some dialects even with preverbal ones)
It can reorder with subject and object clitics
Slide18Minimizer
a
. Al
sei
bic
(
Livigno
(SO
))
I=it
know not
b
.
Magnelo
mina
?
(
S. Anna (VE))
Eats=he
not
c
.
Movat
mia
!/
Mov
r
at
mia
!
(
S.Antonino
,/
Albinea
)
Move not
A
n
è
mina
rivà
nisun
(
Loreo
(RO))
It not is not come
nobody
e. A
l’a
pa
già
ciama
(Torino)
He
cl.has
not already called
f. At
crumpu
lu
opura
at
crumpi
millu
(
Borgomanero
)
You
buy.it
or
you
buy
not.it
Slide19Zanuttini’s findings
Postverbal
negative markers in this position are etymologically minimizers
A. They occurs before
AnteriorTP
signaled by “already”
B. They do not block V to C
C. They are generally compatible with true imperative forms (but see
Emilian
dialects for exceptions)
D. They do not require but can trigger negative concord
E. They only reorder with clitics in dialects with generalized
enclisis
Slide20“Nothing”
A
l’avia
già
nen
volu
‘
ntlura
He
it had already not wanted
then
A
l’ha
nen
dine
sempre
tut
He cl
has not
said.us
always
all
c.
Parla
nen
!
Talk
not
!
I
vni-ve
nen
?
You come-you not?
e
. A
parla
nen
cun
gnun
H
e
speaks not with
nobody
Slide21Zanuttinis’ findings
T
his type of negative marker generally etymologically derived from ‘nothing’ occurs after “already” but before “always”
A.
It never blocks V to
C in interrogatives
B. It is always compatible with true imperative verbs
C. It does not require negative concord, but in some cases it is possible
D. It never reorders with clitics
Slide22Pro sentence
a.
Su no
(
Milan)
(
I) know not
b.
L'è
rivà
nisun
It
is come
nobody
c.
Vusa
no
!
Shout+imp
not
d.
Te
la
cumpret
o
te
la
cumpret
no?
You it
buy.you
or you it
buy.you
not?
e
.
L’a
mangià
no
H
e
has eaten not
Slide23Pro sentence
This type of negative marker is found at the vP border after adverbs
It is compatible with true imperatives
It never requires or allows for negative concord
It never blocks V to C
It is generally found after the past participle and in some dialects even after the object
Slide24Summary
Clitic
Minimizer
Nothing
Pro-sentence
Position
preT
preAnteriorT
pregenericAsp
prevP
V
to C interference
+
-
-
-
Negative concord
+
+/-
-/(+)
-
Compatible with true imperatives
-
+/-
+
+
Reorders with clitics
+
-/(+)
-
-
Slide25Summary
Clitic
Minimizer
Nothing
Pro-sentence
Position
preT
preAnteriorT
pregenericAsp
prevP
V
to C interference
-
-
-
Negative concord
+
+/-
-/(+)
-
Compatible with true imperatives
-
+/-
+
+
Reorders with clitics
+
-/(+)
-
-
+
- - -
Slide26Summary
Clitic
Minimizer
Nothing
Pro-sentence
Position
preT
preAnteriorT
pregenericAsp
prevP
V
to C interference
+
-
-
-
Negative concord
+
+/-
-/(+)
-
Compatible with true imperatives
-
+/-
+
+
Reorders with clitics
-
+ +/-
- -
Slide27Summary
Clitic
Minimizer
Nothing
Pro-sentence
Position
preT
preAnteriorT
pregenericAsp
prevP
V
to C interference
+
-
-
-
Negative concord
+
+/-
-/(+)
-
Compatible with true imperatives
-
+/-
Reorders with clitics
+
-/(+)
-
+ +
- -
Slide28Summary
Clitic
Minimizer
Nothing
Pro-sentence
Position
preT
preAnteriorT
pregenericAsp
prevP
V
to C interference
+
-
-
-
Negative concord
+
+/-
-/(+)
Compatible with true imperatives
-
+/-
+
+
Reorders with clitics
+
-/(+)
-
-
-
+ +/- +/-
Slide29Zanuttini’s analysis
[
NegP1 non [
TP1
V+
Agr
[
NegP2
mica [
TP2
[
AdvP
already]
[
NegP3
niente
[
Asp
perf
.
V
past
part
[
Asp
gen/
progr
[
AdvP
always]
[
NegP4
NO [
VP
]]]]]]]]]
Each
negative
marker
has a different position in the clause where it is merged.
Slide30Etymological
types
Minimizer
(
pas
,
mica
,
brisa
,
bucca
,
filu
etc.)
Existential (
res
,
nen
,
nia
, non)
Focus (
no
,
neca
,
manco
)
Are
there
any
others
?
Slide31How many Negations?
The
situation
we
have
seen
in
Italian
varieties
is
similar
to
the
one
found
in
typological
work
:
Devos
et
alii
(2014): Bantu
(
mvûl
)
kà
-nák-ááŋ
pénd
(1.rain) neg1.1sc-rain-tam neg2
‘
it
does
not rain’
c. (mvûl) kà-nák-ááŋ-áp (kwénd)(1.rain) neg1.1sc-rain-tam-neg2 (neg3)‘it does not rain’
Slide32Three negative
markers
ka-zeby-áandi khúumbu ya ŋgúdy-áani kó
neg1.1sc-know-neg3 9.name 9.conn 1.mother-poss.1sg neg2
‘he does not know the name of my mother’
(Suundi H16b, Baka 1998:fieldnotes)
three
negative
markers
can
also
be
used
for
standard
negation
Slide33Negative strategies
in
Kanincinneg1+neg2
ki-... pend
neg1+neg2(+neg3)
ki-...-p(a) (kwend/kwaam)
p(a) is originally a minimizer
Pend is a compound for of p(a) plus end, a possessive pronoun used for emphasis
Kwend is a possessive pronouns typically used as a focus marker
Slide34Locatives
as
minimizers
Kanincin
dááŋ-âp
(
b)
dááŋ-ûkw
eat.imp-16loc
eat.imp-17loc
‘
eat a little’ ‘eat a bit (of it
)’
Kanyok
L32
dim
-ò-h
cultivate-IMP-16.LOC
“
Work on the land a little!”
(
Devos
& van
der
Auwera
(2013: 239))
Slide35Further etymologies
Austronesian
languages
: indefinite
articles
, partitives,
quantifiers
.
Arawan
,
Yuman
: modal
markers
of
Irrealis
Kru
languages
: „
leave
“ „
go
“ „lack“
Slide36Paamese:
quantification
Maile
vite
he+sau+tei
Mail
3sg.real.
say
3sg.dis.
sing
.part
“Mail said he would sing a bit.”
Ma+ani+tei
raise
1sg.imm.
eat
.part
rice
“
I would like to eat some rice.”
(
Crowley
(1982: 144))
Slide37Fulfulde: lack
(
a) o waas-ii
debbo
makko
.
He lose-TNS woman his
“He has lost his wife.”
(b)
ko
miin
waas-i
am-de.
FOC me NEG-TNS dance-INF
“It's me who did not dance.”
(Marchese 1986: 181)
Slide38The Hypothesis
Each etymological type of “negative markers” corresponds to a semantic operation necessary to achieve sentential negation.
The composition of all of them provides the semantics of negation.
Slide39The set of operations
Open a set of alternatives (Focus)
Identification (
Existential)
Exclusion
Negation would thus be a sort of anti-focus
Problem: what about minimizers? Do they constitute an alternative strategy implying
scalarity
?
Slide40There is n
o “
NegP”
None of
this
elements
is
per se the
negation
in the
logical
sense
of ¬ P.
All
these
elements
have
different
features
and are
attracted
by
different
heads in the
clausal
domain,
therefore
they
occur
in
different
positions
although
they
have
been merged together.
Slide41Arguments for big NegP
I first present four arguments which show that
NegP doubling and DP doubling are the same phenomenon.
Then I show that the big
NegP
idea nicely complements
Zanuttini’s
theory in explaining some exceptions to her empirical generalizations
Slide421. A
rgument
Like DP doubling, NegP
doubling is clause bound:
Un
m’ha
detto
che
*(un)
viene
punto
(Florence)
not
to.me has said that not comes not
In order to license
punto
, the preverbal negative marker must be in the same clause
Slide432. Argument
DP doubling allows for
triplings. The same is true for negative markers,
you can have three negative
markers
No
la go
miga
magnada
NO
! (
V
enice
)
Not
it have not eaten not
‘
I did not eat it’
Slide443. Argument
If two negative markers are merged together, there must be cases where we see them together. This is the case of constituent negation, where there is no clausal structure to which the two formatives can independently move:
No
miga
tutti
(Padua)
not
not
all
Slide454. Argument
The advantage of this hypothesis is that it explains negative concord in a straightforward way. The reason why the two negative markers count as one is that they are merged together.
Again, the parallel with DP doubling where two elements share the same thematic role is striking.
Slide46Do we really need such a complex analysis?
What are the advantages if we assume this analysis with respect to simply assuming, as Zanuttini did, that there are four different
NegPs
?
First of all,
NegP
is at present the only exception to the view that sentence structure is a set of FPs all with different semantic features, at least phase internally.
Slide47Further
arguments: Relativized minimality
Why could it be potentially interesting to split negation into more elementary components?
X Z Y
* How do you wonder [ who behaved ___ ]
[+f] [+f] [+f]
*
Y is in a Minimal Configuration with X
iff
there is no Z such that
Z is of the same structural type as X, and
Z intervenes between X and Y.
Slide48Structural
types
(Rizzi 2004)
Argumental
: person, number, gender, case
Quantificational:
Wh
,
Neg
, measure, focus...
Modifier: evaluative, epistemic,
Neg
, frequentative,
celerative
, measure, manner,....
Topic
If negation occurs in two structural types, it means it has more than one feature.
Slide49How
is the
„
structural
type“
defined
?
On
the
basis
of
more
primitive
features
that
are
present
in all
the
members
of
a
single
class
.
For
instance
:
if
Focus,
wh-items
and
negation
belong to the same class, they must have the same primitive in their endowment. What we call „features“ might be basic semantic operations composing well known different types of phenomena (i.e. wh-items, focus and negation might have as common
Slide50It explains movement
It is well known that negative markers must be allowed to move to Focus in Italian varieties:
‘He did not come’
Non è
mica
venuto
Not is not come
Mica
é
venuto
Not is come
Slide51However, they can also be found lower than their assumed base position in
Zanuttini’s
schema, thus creating true exceptions to the hierarchy, unless we assume that the four positions have all been reached through movement from a lower one.
Slide52First exception
A NegP4
element
like
no
should
occur
on the right of ‘
always
’,
not
on the
left
.
However
,
this
is
not
always
true
:
i
an no
semper
durmi:d
they
have not always
slept
Hence, we have to assume that also the lowest negative markers can move higher
Slide53Second exception
Manzini
and Savoia
(2002; 2005)
notice that in some cases the minimizer type of negation is located lower than the adverb ‘already’:
jau
dorme
l
aun
bo
Müstertal
I sleep yet not
The negative marker
bo
, being a minimizer, should occur before ‘yet’ according to
Zanuttini’s
hierarchy.
Notice that this cannot be interpreted as constituent negation.
Slide54Third exception
There are cases in which two negative markers occur in front of ‘yet’, i.e. they cannot be constituent negation:
El
ciami
mia non
anmo
S. Angelo Lodigiano
him
I.call
not not yet
(Manzini& Savoia (2011):
27
If
we
assume
that
the
two
negative marker are
merged
as
a
unit
, the
problem
is
immediately
solved
.
Slide55Fourth exception
Florentine
punto is a minimizer and should be located in NegP2, i.e. higher than adverbs like yet/already:
Un ha
ancora
dormito
punto
(Florence)
Not has yet slept not
Here
punto
occurs lower than
ancora
and again
this cannot be treated as constituent
negation.
Slide56Summing up
Northern Italian has four different types of negations which have different properties, but they are not merged in the position where we actually see them in the majority of the dialects, but lower.
If we assume that they are all merged as
a unit in the vP,
we can account for these exceptions to Zanuttini’s generalizations.
Slide57The Southern side of the puzzle
In the Southern
I
talian dialects no standard
postverbal
negative marker has been
developed.
Why is this so?
Slide58The Southern side of the puzzle
Southern Italian dialects have developed new preverbal negative markers without undergoing any doubling stage, i.e. violating
the Jespersen
cycle:
Neg
V
Neg
V
neg
V
Neg
Slide59Rionero
mankə
In the dialect of
Rionero
in Vulture the usual form
non
has been replaced by ‘
mankə
’, related to the verb ‘lack’ and the adverb ‘less’:
Vivə
spessə
se
mankə
vu
caré
malatə
drink
often if not want.2sg fall.inf
ill
Slide60mankə versus
non
Mankə
behaves like
non
in
Occurring after the subject and before object clitics:
Mankə
tə
rə
dikə
pecché
mankə
rə
saccə
.
not
you=it=tell.1sg because not it=know.1sg
b) Being incompatible with true imperative forms
Mankə
u
piglià
!
not it=take.inf
Slide61mankə versus
non
c) Requiring negative concord with
postverbal
n-words
Mankə
je
venutə
nesciunə
.
not
is come nobody
d) Not being the pro-sentence negation
Hai
vistə
a
Pierə
?
No / *
Mankə
.
have.2sg
seen to P. no
Slide62Focus negation
Why has there been a substitution without changing the position of the negative marker in the clause?
The negative marker is the spell out of another portion of the internal structure of the big NegP, namely the one related to Focus, similar to sentence final/initial NO.
So, it is the internal position in the NegP that has changed.
Slide63The solution to the puzzle
We conclude that the Jespersen cycle is not the only way in which a new negative marker can be created.
New negative markers can be the spell out of a lower or higher position inside the “big NegP” and do not necessarily require doubling.
This explains both the Jespersen cycle and its exceptions.
Slide64Concluding remarks
The big
NegP hypothesis explains
Why there is negative concord with two negative markers
Why
Neg
-doubling is strikingly similar to DP doubling
Why there are exceptions to
Zanuttini’s
schema in the position of the negative markers
Why some dialects recreate the negative marker without an intermediate doubling stage and without changing its position.
Slide65Thank
you for your attention
Slide66Future developments
If
negative doubling
is
an
effect
of
big
NegP,
what
is
negative
concord
?
i.e.
can
negative
concord
also
be
treated
as
deriving
from
a
single
constituent
which
is
split
?
Slide67It might be the case that at least in some languages also negative concord can be treated as the result of the splitting of a unity containing the negative marker and the n-word.
Gianollo
(2015): in
the
Romance
languages
the
change
from
double
negation
to
negative
concord
is
accompanied
by
a
change
in
the
form
of
the
n-
words
.
Negative
concord
might
be
a
necessary
(
though
not sufficient) condition to have multiple negative markers.
Slide68What about negative concord?
If we assume that negative concord is an instance of negative doubling in the sense that the negative markers start out as a unit, we might want to apply this also to negative concord with n-words
Is there any evidence that this might be the case?
Slide69Chiomonte, Provençal dialect in Piedmont
Postverbal n-words:
Lù’ei pâ arrivá gî
It has not arrived anybody
La m’a pâ vî gî
It me has not seen anybody
Preverbal n-words
Pagî o minjá la soupo
Not anybody has eaten the soup
Pagî ou’m capî
Not anybody me understands
Slide70From non strict to strict negative concord
A research on Zemignanese, which has an intermediate system where strict negative concord is not always obligatory has shown that strict NC is sensitive to the following factors
A) type of n-word
B) type of thematic role
C) presence of a subjunctive
Slide71Strict NC with preverbal n-adverbs
Slide72Strict NC with prevergal object niente
Strict NC with preverbal subject niente
Slide74Strict NC with preverbal nessuno subject
Slide75Strict NC with preverbal nessuno object
Slide76This multifactorial situation about strict NC points towards a complex analysis of the phenomenon that is not simply to be captured in terms of agreement.