Article

Polarity

Anastasia Giannakidou

in Linguistics

ISBN: 9780199772810
Published online October 2011 | | DOI: http://dx.doi.org/10.1093/obo/9780199772810-0035
Polarity

More Like This

Show all results sharing these subjects:

  • Linguistics
  • Anthropological Linguistics
  • Language Families
  • Psycholinguistics
  • Sociolinguistics

GO

Preview

Polarity phenomena in natural language are pervasive and diverse. Polarity items (PIs) are expressions of various syntactic categories, such as nominals, adverbials, verbs, particles, and idioms, with limited distribution: that is, they do not occur in a positive sentence in the simple past. The English word “any,” superlatives like “the faintest sound,” and minimizers such as “lift a finger” are among the first studied PIs. Because of their apparent sensitivity to the presence of negation, they were labeled negative PIs (NPIs). English NPIs are also contrasted with positive polarity items (PPIs) in the earlier literature, words such as “some” and “already,” which are claimed to avoid the negative context. Negative concord words (Overview Papers on N-words) in various languages are treated as NPIs, since these too need negation for well formedness. Typically, “any” and more idiomatic NPIs are viewed as scalar items, rhetorical devices for the manipulation or strengthening of discourse information. PPIs, on the other hand, are argued to be nonscalar. Other NPIs are narrow-scope, nonspecific indefinites that cannot be linked to discourse referents in the way “regular” indefinites are. Such NPIs cross-linguistically do not exhibit the rhetorical effects (emphasis, focus) observed with “any” and minimizers and are typically nonemphatic. The subjunctive mood is often treated as a temporal NPI of this kind. Free choice items (FCIs) are sensitive to the modal, generic, or quantificational properties of the context and seem to disfavor negation. “Any” has both NPI and FCI usages, but it is fairly common in languages to tease the two paradigms apart morphologically. For example, FCIs are usually wh-based (see, e.g., English “whoever”) and contain special morphology (unlike “any”). NPIs and FCIs are sensitive to the logical properties of the sentences in which they occur and are subject to licensing. Licensing typically postulates that the PI be in the syntactic or semantic scope of the licensor—and this captures the narrow scope property of most PI classes. What logical property unifies licensors as a natural class has been a matter of intense debate, with proposals including negation, Licensing and Downward Entailment, and Negative Polarity Items and Nonveridicality. Empirically, nonveridicality captures best the wide range of polarity environments and offers a flexible framework to capture the variation across NPI and FCI classes in a number of languages, predicting possible (but not identical) distributions in negative, downward entailing, and nondownward entailing nonveridical contexts, such as modal contexts, questions, generic sentences, and disjunctions. William Ladusaw views NPI licensing as a global constraint on grammatical representations, but nowadays the limited distribution and the ensuing need for licensing are explained by appealing to the lexical semantic or pragmatic properties of individual PI classes (e.g., scalarity, referential deficiency, or a free choice component), which result in most cases in nonfully identical yet predictable distributions. These major considerations and a solid cross-linguistic perspective will guide the presentation of the material in this article.

Article.  14009 words. 

Subjects: Linguistics ; Anthropological Linguistics ; Language Families ; Psycholinguistics ; Sociolinguistics

Full text: subscription required

How to subscribeRecommend to my Librarian

Buy this work at Oxford University Press »