Danielle Forget, Paul Hirschbuehler, France Martineau and
Maria-Luisa Rivero. (eds) Negation and Polarity. Amsterdam:
John Benjamin, 1997. Pp. 365, USD 83.00 / NLG 166.00.
ISBN 90 272 3660 7 (Eur.)/1-55619-871-X (US) (alk. paper)
Reviewed by Shravan Vasishth, The Ohio State University.
This book is a collection of seventeen papers (selected from a
total of twenty seven) presented at the conference, ``Negation:
Syntax and Semantics'', held May 11-13, 1995, at the University
of Ottawa, Canada. Two papers are in French, and the rest are
in English. The papers are arranged alphabetically by author, and
there are three very convenient indices towards the end of the
book, of authors, terms and concepts, and languages and language
The study of negation and polarity has gained importance in
linguistics over the last several decades, perhaps because
the issues raised by the syntactic, semantic, and pragmatic aspects
of negation and polarity directly affect linguistic theory as a
whole. The present collection reflects the diverse and
sophisticated approaches that have been brought to bear on this
subject and is thus a valuable contribution to the field. In this
review, for reasons of space, I will summarize and discuss articles
by topic, focusing on some issues that have implications for
linguistic theory in general.
A major theme in this collection is negative concord (NC).
The papers by Viviene Deprez, Liliane Haegeman, Daniel Valois,
Paul Rowlett, Joao Peres, and Jacob Hoeksema all touch on various
aspects of NC, and several take mutually conflicting positions on
the subject. The syntax papers on NC presuppose some familiarity
with GB syntax and in particular with Haegeman's Neg-criterion and
the related Wh-criterion (see, e.g., Haegeman 1995 for details).
Regarding the semantics of N-words (these are words such as
English `no-one' and its French equivalent `personne'), Joao Peres
(``Extending the notion of negative concord'') argues for a unified
view of NPIs and N-words as indefinites as argued for by Ladusaw
(1992). Jack Hoeksema (``Negation and negative concord in Middle
Dutch'') shows that evidence from Middle Dutch points to
N-words being ambiguous between a negative-existential and
existential reading. Both analyses converge towards Dowty's
(1994) treatment of N-words as being ambiguous between a negative
meaning and an existential interpretation.
Turning next to the several syntax papers on N-words, Viviene
Deprez (``A non-unified analysis of negative concord'') presents
data from French and Haitian Creole to show that Haegeman's
Neg-criterion is redundant in accounting for NC, and that,
following Ladusaw (1992), N-words are best treated as non-negative
indefinite NPs. Liliane Haegeman (``The syntax of N-words and the
Neg-criterion'') argues against Deprez's claim that N-words do not
have the NEG feature. Haegeman's refutation of Deprez's arguments,
although convincing, would have been even more persuasive if she
had discussed Haitian Creole and French (the languages Deprez bases
her arguments on) rather than limiting herself principally to West
Daniel Valois (``Neg-movement and Wh-movement'') explores another
aspect of the Neg-criterion; he shows that in spite of the
difference in behavior between N-words and Wh-traces, the
Neg-criterion can account for N-words if one follows Haegeman
(1995:234-269) in treating SpecNegP as both an A and A' position.
Paul Rowlett (``Jesperson, Negative Concord, and A'-binding'')
attempts to provide a purely syntactic account of NC and non-NC
languages by adopting and modifying Haegeman's and Progovac's
proposals. His main claim is that if one recasts the Spec-Head
agreement requirement as a weaker Spec-Head compatibility
requirement, NPI and negative quantifier licensing can be accounted
for in terms of A'-binding. This paper, although very insightful in
its treatment of data from European languages, faces the problem
that the analysis would be hard to motivate cross-linguistically.
For example, reliance on functional projections (FP) such as NegP
becomes difficult when one considers languages like Japanese,
Korean, and Hindi, where the presence of FPs in general is not
well-motivated (see. e.g., Kim and Sag 1995, Sells 1995,
Fukushima 1998). But even if one were to allow the functional
projection NegP in Hindi (contra Mahajan 1988), according to
Rowlett's analysis the Hindi negation marker would be predicted to
appear in the Spec-NegP position since Hindi is a non-NC language.
However, existing research (e.g., Dwivedi 1991, Vasishth 1997, and
Bhandari 1998) has shown that if NegP were present in Hindi, the
negation marker would have to be in Neg_o position, not Spec-NegP.
Given the importance of cross-linguistic validity in GB syntax,
this makes Rowlett's analysis somewhat harder to justify.
If functional projections are indeed under-motivated in syntax,
the two papers summarized below are also open to the same
criticism. M. Teresa Espinal (``Non-negative negation and
Wh-exclamatives'') presents a syntactic analysis of the licensing
of non-negative or expletive negation in exclamative wh-sentences
(such as `How many people did you not deceive in your youth!'). In
order to account for various facts about exclamatives, Espinal
posits a functional projection, Int(ensifier)P, above CP, to which
the wh-element must raise at LF. A licensing condition, logical
absorption (Espinal 1992), accounts for negation being expletive in
exclamatives: briefly, an abstract intensifier operator which
lexically selects negation absorbs Neg in the configuration
[... Op [C [ ... Neg ...]]]
if Minimality is respected and no logical operator intervenes between
OP and Neg at LF.
Aafke Hulk and Ans van Kemenade (``Negation as a reflex of
clause structure'') look at negation in Old English (OE) and Old
French (OF) and conclude that negation occupies a fixed position
(Spec, NegP). They assume a clause structure as follows:
C AgrS Neg T (AgrO) V
Observing that a pronominal subject appears to the left of the
OE negative `na' while a DP subject appears to its right, they
propose a more articulated phrase structure where a functional
projection (FP) dominates NegP, Spec-FP providing a landing site
for the pronominal subject. Providing a separate position for
pronouns allows us to treat these not as clitics, as previous
analyses (Pintzuk 1993) have done, but as ``weak pronouns''
(Cardinaletti and Starke (1994)). This overcomes the problems with
Pintzuk's analysis, which presupposes that Spec-IP is a possible
topic position. A similar analysis is proposed for OF, based on
the distribution of `ne' and `pas' with respect to the finite verb
and the pronominal subject. As in OE, the verb moves to F of the
new functional projection FP, this movement later being lost in
The status of functional projections in syntax is
addressed rather decisively in two papers. Abeille and Godard
(``The syntax of French Negative Adverbs'') convincingly argue that
adverbs (including negative adverbs like `pas') in French do not
support verb movement and functional projections, contra Emonds
(1978) and Pollock (1989). They propose that adverbs are either
adjoined to VP or occur at the same level as a complement of the
VP. This view is formally treated within the Head-driven phrase
structure (HPSG) framework (Pollard and Sag 1994). Denis Bouchard
(``The syntax of sentential negation in French and English'') also
argues that functional projections are undermotivated; however, he
relies on his own version of Chomsky's Minimalist program (1995) to
account for the facts. Bouchard argues that the recent checking
based model (Chomsky 1995) has an element of redundancy since
functional properties appear both as heads in the syntactic
structure and as parts of a lexical item. He proposes an
alternative `minimal' account where functional projections are
unnecessary. Briefly, he assumes that (i) inflectional verbs are
composite lexical items, with no additional functional categories,
(ii) tense is in the highest projection of the sentence, (iii) the
sentential negation marker must scope over as much of the sentence
as possible, but not over tense. Bouchard's proposal is very
attractive, but it would have been helpful (at least to this
reviewer) if he had specified what a parameter is in the theory.
In the Minimalist Program, a parameter is apparently ``a choice of
STRONG/WEAK feature on a functional head'' (Fodor 1998:3). In the
above account, we have a parametric choice on the accessing of
composite heads: French allows the relevant parts of a composite V
to be accessed part by part, but English never allows part by part
access. This notion of parametrization may turn out to be
problematic in Bouchard's theory.
The articles discussed next are somewhat specialized; they deal
with aspects of negation that presuppose some familiarity with the
One paper, on the semantics and pragmatics of `only' by
Laurence Horn (``Negative polarity and the Dynamics of Vertical
Inference'') is an earlier version of his arguments concerning
`only' in (Horn 1996). In the present paper, the principal
question addressed is: what are the relationships between a
sentence like (1)a and (1b),(1c)? (Note: the utterance (1a) has
nothing to do with the line appearing in the well-known poem by
Joyce Kilmer; the sentence here has its usual compositional
(1)a. Only God can make a tree.
b. God can make a tree.
c. No one distinct from God can make a tree.
While it is clear that (1a) entails (1c), the relationship
between (1a) and its ``prejacent'' (1b) is more controversial. Is
this relation one of entailment, semantic presupposition, pragmatic
presupposition, or possibly conversational implicature? He
concludes that (1b) is neither entailed, nor presupposed, but
merely the ``...existential import of the corresponding
universal...'' of (1a): ``...if no mortal can make a tree and the
set of tree-makers is non-null, then the truth of the prejacent
follows. No specific rule--entailment, presupposition, or
implicature--need be invoked to derive [(1b)] from [(1a)]'' (p. 168).
In addition, Horn also presents several arguments favoring the
view that `only' is downward monotone, which is consistent with its
NPI-licensing property. In this connection, and for the latest on
a long-standing dispute between Horn, Jay Atlas, and others, about
the (non-)downward monotone nature of `only' and its status as an
NPI licensor, also see Atlas (1997).
Eugene Rohrbaugh explores the relationship between focus and NPI
licensing (``The role of focus in the licensing and
interpretation of negative polarity items''). He provides
impressive arguments against the well-known treatment by Kadmon and
Landman (1993) of polarity sensitive (PS) and free-choice (FC)
`any'. Rohrbaugh claims that it is the focus or intonational
contour on `any' that is responsible for the widening effect
discussed by Kadmon and Landman. He claims that a crucial
difference between the two kinds of `any' (FC and PS) is that focus
plays a role in the licensing FC `any' but not PS `any'.
Interestingly, this analysis would predict correctly that in Hindi
NPI `any', `koi-(bhii)', the focus particle `bhii' is optional,
while in its homophonous PPI counterpart it is obligatory
(Bhatia 1995:27), and that minimizers in Hindi obligatorily require
focus particles in order to get the NPI interpretation (Vasishth
1998c). It would have been helpful, though, if Rohrbaugh had
spelled out his analysis in detail rather than simply stating that
focus plays a crucial role in NPI licensing.
Michael Isreal explores a central issue relating to NPI licensing
(``Scalar model of polarity sensitivity and aspectual
operators''): why do NPIs exist at all? He summarizes his theory of
polarity licensing as follows: ``Polarity in general is a matter of
scalar inferencing and polarity items are just scalar operators:
the proper expression of their lexical semantics depends on the
availability of a properly constructed scalar model'' (p. 217).
In this article, he assumes two scales, a q(uantificational) scale
and an i(nformational) scale. Propositions can be high or low on
the q- and i-scale. The q-scale value refers to the proposition's
position on the scale; a q-value of a proposition is high when the
asserted or text proposition (TP) is located higher on a
contextually determined scale compared to some alternative context
proposition (CP), and low if the TP is lower. The i-scale value
refers to its (relative) informativeness, where informativeness is
defined as follows: if the TP entails the CP, the i-value is high,
and low if the CP entails the TP.
An example of how this works is the minimizer `a wink' (minimizers
are expressions which, if they appear in a positive context, denote
a minimal quantity, and in negative contexts denote ``the absence of
a minimal quantity, and hence the presence of no quantity at all''
(2)a. Marianne didn't sleep a wink that night.
b.*Marianne slept a wink that night.
As an Emphatic NPI (see classification given below), this minimizer
has a low q-value and a high i-value when it is felicitous; a low
q-value because it indicates a minimal quantity, and a high i-value
because the TP `M didn't sleep the smallest amount' entails the CP
`M didn't sleep a normal amount.' (2b) is ungrammatical because the
CP, `M slept a normal amount', entails the TP, `M slept the
smallest amount', resulting in a low i-value.
Israel posits four types of polarity items, Emph(atic) and
Understat(ed) NPIs and PPIs (an example of the first is given above):
Emph NPIs Understat NPIs Emph PPIs Understat PPIs
q-scale low high high low
i-scale high low high low
This analysis is applied to apparently non-quantificational
aspectual operators, such as `yet', `already', `still', and
`anymore', which are also treated as scalar operators.
For a more fully worked out version of Israel's research, the reader
should consult (Israel 1996). Although Israel's treatment is
extremely insightful, its cross-linguistic validity would
be enhanced further once it is extended to account for NPIs in
languages like Hindi and Japanese, where `focus particles' crucially
affect the behavior of NPIs (see Vasishth 1998a and 1998b).
Elizabeth Pearce (``Negation and indefinites in Maori'') addresses
the problem of NPI licensing in Maori, a VSO language. She follows
Progovac's (1994) binding approach to account for the salient
facts. The reader may find it useful to consult Horn and Lee's
(1995) critique of Progovac's theory in this regard.
Jacques Moeschler (``La negation comme expression procedurale'')
addresses a central problem relating to metalinguistic negation
(MN): is MN part of a pragmatic account that recognizes
truth-conditionality (Horn 1989) or is truth-conditionality to be
excluded (Ducrot 1972)? Moeschler develops Carston's (1996)
analysis of MN, which is itself based on Relevance Theory (Sperber
and Wilson 1986). He proposes a contextually driven approach
to MN: negation provides instructions for building a context
necessary for interpreting utterances, and external and
metalinguistic negation give rise to different contexts. This is
implemented as an algorithmic procedure that allows us to give a
uniform account of the various kinds of MN, as presented below.
1. If there is a proposition P' such that (not(P), P') corresponds
to the formal structure of the phrase,
1.1 If (P --> not(P')), then conclude not(P) --> P';
1.1.1 If (P' --> not(P)) then you have ``negation abaissante''
1.1.2 If (P' --> P) then you have ``negation majorante''
1.2 If (E(P) --> not(E(P'))), conclude (not(E(P)) --> E(P')),
1.3 If (not(P) --> Q) (P --> Q) and (P' --> not(Q)), conclude
(not(P) --> not(Q)) (presuppositional negation)
else go to 2.
2. Find an available proposition Q:
2.1 If Q is available in the `cotext':
2.1.1 If (Q --> P) and not(P), conclude (P and not(P)) and
nullify not(P) (concessive negation, polemic)
2.1.2 If (Q --> P) and not(P) then conclude not(Q) and
nullify Q (refutative negation, polemic)
2.2 If Q is available in the `co-text':
2.2.1 From (P --> Q) and not(P), conclude not(Q) (inferential
An example of how this works is the case of the `more-than' reading
of negation (his term is ``negation majorante''). Consider the
following instance of ``negation majorante'', which is a kind
of MN (in this connection, see also Horn 1989:204):
Max is not tall, he is gigantic.
By (1), we have P=Max is tall and P'=Max is gigantic. By (1.1),
since (P --> not(P')), by invited inference, we can conclude
(Not(P) --> P'). By (1.1.1), since P' --> P, this is ``negation
majorante''. Horn points out (personal communication) that the
Relevance-theoretic position that Moeschler follows Carston (1996)
in supporting, is in fact quite close to Horn's. Carston's
forthcoming paper (1999) also apparently makes this point.
To conclude, we have not been able to look closely at each article
in this collection (indeed, I have not discussed some articles
here), but, clearly, many of these articles address important
issues of general linguistic interest, and several shed new light
on various aspects of negation and polarity. This volume should thus
prove to be an important source of reference for linguists.
Note: The references given below are also available in BibTeX
format from the following ftp site:
Atlas, Jay David, 1996. `Only' Noun Phrases, Pseudo-Negative
Generalized Quantifiers, Negative Polarity Items, and
Monotonicity. Journal of Semantics, Vol. 13, pp. 265--329.
Bhandari, Rita, 1998. On the role of tense for negative polarity
item licensing, Paper presented at the 1998 Annual Meeting
of the Linguistic Society of America, New York City.
Bhatia, Tej K., 1995. Negation in South Asian Languages.
Indian Institute of Language Studies, Patiala, India.
Carston, Robyn, 1996. Metalinguistic Negation and Echoic Use.
Journal of Pragmatics, Vol. 25, pp. 309--330.
Carston, Robyn, (to appear). Negation, `Presupposition', and
Metarepresentation: A Response to Noel Burton-Roberts.
Journal of Linguistics, Vol. 35, pp. ??--??
Chomsky, Noam, 1995. The Minimalist Program, MIT Press,
Dowty, David, 1994. The role of Negative Polarity and Concord
Marking in Natural Language Reasoning. MS, Ohio State
Ducrot, Oswald, 1972. Dire et ne pas dire. Hermann, Paris.
Dwivedi, Veena. 1991. Negation as a functional projection in Hindi.
WECOL 4 Proceedings, edited by K. Hunt, T. Perry, and V.
Samiian, pp. 88--100.
Emonds, Joseph E., 1978. The Verbal Complex V'-V in French.
Linguistic Inquiry. Vol. 9, pp. 151--175.
Fodor, Janet Jean, 1998. What is a Parameter? Annual Meeting of the
LSA, New York City.
Fukushima, Kazuhiko, 1998. Compositional, Inherent and Frozen
Negation: Lexicalism versus Functional Categories.
The Proceedings of the Salford Conference on Negation.
Haegeman, Liliane, 1995. The Syntax of Negation. CUP, Cambridge, UK.
Horn, Laurence R., 1989. A Natural History of Negation. UCP, Chicago.
Horn, Laurence R. and Young-Suk Lee, 1995. Progovac on Polarity.
Journal of Linguistics, Vol. 31, pp. 401--424.
Horn, Laurence R., 1996. Exclusive Company: Only and the Dynamics
of Vertical Inference. Journal of Semantics, Vol. 13, pp.
Israel, Michael, 1996. Polarity Sensitivity as Lexical Semantics,
Linguistics and Philosophy, Vol. 19, pp. 619--666.
Kadmon, Nirit and Fred Landman, 1993. Any. Linguistics and
Philosophy, Vol. 16. pp. 353--422.
Kim, Jongbok and Ivan A. Sag, 1995. English and French Negation: A
Lexicalist Perspective, MS, Stanford, CA.
Ladusaw, William A., 1992. Expressing Negation. SALT II Proceedings,
edited by Chris Barker and David Dowty. Ohio State University
Working Papers in Linguistics No. 40.
Mahajan, Anoop Kumar. 1988. Word Order and Negation in Hindi.
MS, MIT, Cambridge, MA.
Pollard, Carl and Ivan A. Sag, 1994. Head-driven Phrase Structure
Grammar, UCP, Chicago.
Pollock, Jean-Yves, 1989. Verb Movement, Universal Grammar, and the
Structure of IP. Linguistic Inquiry, Vol. 20, pp. 365--424.
Progovac, Ljiljana, 1994. Negative and Positive Polarity, CUP,
Sells, Peter. 1995. Korean and Japanese Morphology from a Lexical
Perspective. Linguistic Inquiry, Vol. 26, pp. 277--325.
Vasishth, Shravan, 1997. The NEG-Criterion and Negative Polarity
Licensing in Hindi and English. Osaka University Journal of
Language and Culture, Vol 6, pp. 159--176.
Available by ftp:
Vasishth, Shravan, 1998a. Boolean properties of focus particles and
NPIs in Japanese. Paper presented at the Annual Meeting of the
Linguistic Society of America, New York City.
Available by ftp:
Vasishth, Shravan, 1998b. Monotonicity constraints on negative
polarity in Hindi. OSU Working Papers in Linguistics, Vol. 51,
Ohio State University, edited by Mary Bradshaw, Dave Odden,
and Derek Wyckoff, pp. 147-166.
Available by ftp:
Vasishth, Shravan, 1998c. Focus Particles and Negative Polarity in
Hindi, The Proceedings of the Salford Conference on Negation.
Short biography of the reviewer:
Shravan Vasishth a (2nd year) graduate student in the Linguistics
department of the Ohio State University. His research interests
include the formal syntax, semantics, and pragmatics of word order,
which subsumes issues relating to negation and polarity.