Negative Concord in Levantine Arabic

I show that negative concord occurs in the Levantine varieties of Arabic (Israel/Palestine, Jordan,
Lebanon, Syria), where negative concord is dened as the failure of an n-word (a word that can be used to
express sentential negation) to express negation in syntagm with another negative expression (a morpheme
that can be used to express sentential negation) (c.f. Giannikadou 2002, Watanabe 2004). I provide an
analysis of Levantine negative concord involving both syntactic and semantic elements. I treat negative
concord in Levantine Arabic as a purely semantic process, and therefore one involving interpretable features.
However, the distribution of negative concord congurations is constrained by syntactic and prosodic prop-
erties of the language. The analysis correctly predicts both the Tovena-Herburger generalization (Tovena
1996, Herburger 2001), as well as the fact that n-words express negation in fragment answers. The analysis
therefore supports approaches to negative concord that treat n-words as introducing negation, but improves
upon existing approaches by removing the need for additional formal devices such as quantier absorbtion
or factorization (e.g. Haegeman & Zanuttini 1996, de Swart & Sag 2002). Furthermore, a semantic analysis
has advantages over syntactic alternatives (e.g. Haegeman & Zanuttini 1996, Watanabe 2004, Zeijlstra 2004,
Hoyt 2006) because it captures without stipulation variation in the morpho-syntactic environments in which
negative concord can occur as well as subtleties involving scope interpretation.
Levantine Arabic has three categories of words that appear to be n-words: the still-words lissa- and
baQd- still, yet; the never-words Pabadan and bilmarra never, not at all; and wala-phrases, phrases
prexed by the particle wala not even, not a single. These dier from each other in several respects.
First, the still-words are ambiguous between negative (1) and positive interpretations (2) in both sentence
fragments and in full clauses, while the never-words and wala-phrases are never ambiguous. Never-words and
wala-phrase, in turn, dier in terms of the positions in full clauses in which they can express negation, and
in which they must be roofed (after Ladusaw 1992) by another negative expression. Never-words must be
roofed in all positions in both the pre- and post-verbal elds (3), while wala-phrases always express negation
in the pre-verbval eld (4), and in the post-verbal eld follow a generalization noted by (Tovena 1996, 230-
241) and (Herburger 2001, 301-05), according to which n-words must be roofed if they are interpreted in
thematically entailed positions, where thematically entailed means that the meaning of a predicate entails
the existence of an object satisfying the thematic role assigned to the position in question (5). Elsewhere,
wala-phrases express negation freely (6-8).
I show that wala-phrases presuppose alternatives ranging over cardinalities of noun phrase denotations,
negating the minimum cardinality (9), in contrast to noun phrases prexed with the NPI determiner Paiy
any, which introduce alternatives ranging over objects witnessing noun phrase denotations of varying car-
dinalities (10: c.f. Lahiri 1998). Likewise, wala-phrases answer constituent questions, sets of alteratives
ranging over argument positions. I then show that the Tovena-Herburger generalization follows directly
from a theory of logical form in which thematical roles are treated as modal operators modelling accessibility
relations between eventualities and their participants (c.f. Kruij 2001). The satisifaction conditions for
modal operators involve existential quantication over objects in a model, allowing wala-phrases to be inter-
preted as introducing variables rather than quantiers (c.f. Progovac 2000, Ladusaw 1992, Przepirkowski &
Kup 1999, Blaszczak 1998, a.o.), removing the need to stipulate scoping operations in the interpretation
of wala-phrases (c.f. Herburger 2001). The interpretation of a wala-phrase in a thematically-entailed posi-
tion negates the existence of an object satisfying the thematic role in question, creating a contadiction with
the entailments of the predicate. Negation of the predicate removes the contradiction, correctly predicting
that wala-phrases need be roofed only in thematically entailed positions, and that the positions in which
wala-phrases must be roofed and those in which in they need not be are in complementary distribution.
The dierence between wala-phrases and never-words in terms of whether they must be roofed in the
pre-verbal eld is treated as a dierence in the syntactic properties of the two kinds of expressions. Pre-verbal
wala-phrase, even those binding thematically-entailed positions, are interpreted as topics, with the remainder
of the clause predicated of the topic (c.f. Doron & Heycock 1999). The negation introduced by the wala-
phrase then scopes over what can be seen as a tri-partite structure, with the the common noun restriction
interpreted as the topic restriction, and the scope as the comment predicated of it. Pre-verbal never-words,
in contrast, are interpreted as focused phrases, and therefore interpreted in a reconstructed position, in which
they require roong. Contrasts between the Levantine dialects and and the Maghrebin dialects (Tunisia,
Algeria, Morocco) are argued likewise to follow from syntactic dierences: in the Maghrebin dialects, n-words
must be roofed in all positions, including in the pre-verbal eld. I argue that this follows from a syntactic
dierence: in Levantine, pre-verbal nominal expressions are interpreted as topics and therefore subject to
a well-known specicity condition, understood as discourse topichood, while in Maghrebi, pre-verbal noun
phrases are not subject to such a condition. Independent evidence for this conclusion is presented.
(1) Q: Pinta
you.ms
mitzawwa g?
married
A: ( la
no
) baQd-ni|lissa.
still-me|still
Are you married? (No) still [not], [not] yet.
(2) Q: Panta
you.ms
Qazab?
bachelor
A: ( Pa
yes
) baQd-ni|lissa.
still-me|still
Are you single? (Yes) I still [am].
(3) (Pabadan|bilmarra)
never in-the-once
*( ma-
not-
) baQrif
ind.1s.know
h

ada
one
| *( ma-
not-
) baQrif
ind.1s.know
h

ada
one
(Pabadan|bilmarra)
never in-the-once
I never know anyone, I dont know anyone at all.
(4) wala
not.even
yo:m
day
Qa gab-ni
pleased-me
l-Ekil
the-food
| wala
not.even
yo:m
day
ma-Qa gab-ni
not-pleased-me
l-Ekil.
the-food
Not one day did the food please me, Not one day did the food please me.
(5) *( ma-
not
) h

ake:t
spoke.1s
wala
not.even
maQ
with
h

ada
one
minhum.
from-them
I didnt speak with even one of them.
(6) Pili:sa
Elisa
maGru:re
conceited.fs
Qala
upon
wala
not.even
Pisi.
thing
Elisa is conceited for nothing at all.
(7) Q: mi:n
who
bedd-ak
want.2ms
tazawwi g?
2.marry
A: Pamma
either
h

ani:n
Hanin
Paw
or
wala
not.even
wah

di.
one.fs
Who do you want to marry? Either Hanin, or nobody at all.
(8) Pinta
you.ms
wala
not.even
Pisi!
thing
You are nothing at all!
(9) fiss
exist.neg
wala
not.even
( sabb
boy
| *saba:b
boys
| *sabbe:n
boy.dual
).
There isnt a single boy.
(10) fiss
exist.neg
Paiy
any
( sabb
boy
|
|
saba:b
boys
|
|
sabbe:n
boy.dual
).
There isnt any boy, There arent any boys, There arent any two boys.
(11) Q: X:lid,
Khalid
gel-l-ak
said-to-you
isi?
thing
A: la,
no
la,
no
wala
not.even
isi.
thing
Khalid, did he say anything to you? No, no, not a thing.
References
Blaszczak, J. (1998), Towards a binding analysis of negative polarity items, in Linguistics in Potsdam 4, Institute for Linguistics,
University of Potsdam.
de Swart, H. & Sag, I. A. (2002), Negation and negative concord in romance, L&P 25, 373417.
Doron, E. & Heycock, C. (1999), Filling and licensing multiple speciers, in D. Adger, S. Pintzuk, B. Plunkett & G. Tsoulas,
eds, Speciers: Minimalist Approaches, Oxford University Press (Oxford), pp. 6989.
Giannikadou, A. (2002), N-words and negative concord, in H. van Riemsdijk & R. Goedemans, eds, Linguistics Companion,
Blackwell.
Haegeman, L. & Zanuttini, R. (1996), Negative concord in west emish, in A. Belleti & L. Rizzi, eds, Parameters and Functional
Heads, Oxford, pp. 117179.
Herburger, E. (2001), Negative concord revisited, Natural Language Semantics 9, 289333.
Hoyt, F. M. (2006), Long-distance negative concord and restructuring in palestinian arabic, in Proceedings of the Workshop
on Concord and the Syntax-Semantics Interface.
Kruij, G. M. (2001), A Categorial-Modal Logical Architectyure of Informativity: Dependency Grammar Logic and Information
Structure, PhD thesis, Charles University, Prague.
Ladusaw, W. (1992), Expressing negation, in C. Barker & D. Dowty, eds, Proceedings of SALT II, number 40, pp. 237259.
Lahiri, U. (1998), Focus and negative polarity in hindi, Natural Language Semantics 6, 57123.
Progovac, L. (2000), Coordination, c-command and logophoric n-words, in L. Horn & Y. Kato, eds, Negation and Polarity:
Syntactic and Semantic Perspectives, Oxford University Press, pp. 88114.
Przepirkowski, A. & Kup, A. (1999), Eventuality negation and negative concord in Polish and Italian, in R. D. Borsley &
A. Przepirkowski, eds, Slavic in HPSG, CSLI Publications, Stanford, pp. 211246.
Tovena, L. M. (1996), Studies on Polarity Sensitivity, PhD thesis, University of Edinburgh.
Watanabe, A. (2004), The genesis of negative concord: Syntax and morphology of negative doubling, LI 4, 559612.
Zeijlstra, H. (2004), Sentential Negation and Negative Concord, PhD thesis, Amsterdam University.