Back to list

Detail of contribution

Auteur: Elena GUERZONI

Co-Auteur(s): Yael SHARVIT, UCLA

Whether or not Anything but not Whether Anything or not

Abstract/Résumé: We propose that only NPIs that are overt in the scope of "not" are licensed in interrogatives. This is in line with Fauconnier (1975)’s and Lacdusaw (1997)’s theory of NPI-licensing, and with Guerzoni&Sharvit (2007)’s claim that no existing definition of questions that renders them downward-entailing is supported by native speakers’ intuitions. NEW DATA. We observe that an NPI is acceptable when it follows "whether or not" (as in "Mary wondered whether or not her students had ever read SS"), but not when it is “trapped” between "whether" and "or not" (as in "* Mary wondered whether her students had ever read SS or not"). This contrast correlates with Kayne’s observation that ellipsis of "any" is acceptable only when the overt "any" is in the negated clause (see Kayne 1994 n. 19, p. 146.) ANALYSIS. We propose a unified ellipsis-based approach to alternative questions and Yes/No questions (Cf. Larson'85 and Han&.Romero'04.) Specifically, we suggest that (A) a question like "whether or not John played chess or checkers" has the structure: [whether {John played chess} or not John played chess], where "{}" marks the elided material; (B) a question like "whether John played chess or not" has the structure [whether John played chess or not {John played chess}]; and (C) a question like "whether John played chess" has two possible analyses: (i) [John played chess {or not J. played chess}] and (ii) [{J. played chess} (or not) J. played chess]. Under this view "whether or not p"-questions and "whether p or not" -questions have the same underlying form but differ in the ellipsis site and "whether p"-questions have two ellipsis options, one like (Ci) above and the other like (Cii), where or not is ‘omitted’ for conversational reasons later in the derivation. PREDICTIONS. The new contrast we observed above is predicted because in "[...whether her students ever read SS or not {her students ever read SS}]" the overt NPI is not in the scope of negation but in "[...whether {her students ever read SS} or not her students ever read SS]" it is. Moreover, NPIs are predicted in "whether p"-questions because one of their possible analyses (i.e. [whether {her students ever read SS}(or not) her students ever read SS"]) adheres to the grammatical pattern. FURTHER IMPLICATIONS. Our analysis, paired with Guerzoni (2003)'s claim that Strongly Exhaustive (SE) WH-questions contain a "whether"- question, unlike Weakly Exhaustive (WE) ones, provides an explanation of why the former but not the latter admit NPIs. This is because in the SE question "[Which students]1 [[whether]2 [Q [t2 [t1 read SS (or not) t1 ever read SS ]]]]" the NPI is in the scope of negation but in the WE question "*[Which students]1 [Q [t1 ever read SS]" it is not.