Inferences that are triggered (at least in part) by particular lexical items provide a rich test bed for distinguishing the relative semantic contribution of lexical items and functional structure. One class of such inferences that has garnered extended attention is neg(ation)-raising, wherein negation on a predicate can be interpreted as though in that predicate’s subordinate clause. For example, a neg-raising inference is triggered by (1) while one is not triggered by (2).

  1. Jo doesn’t think that Bo left. ⇝⇝ Jo thinks that Bo didn’t leave.

  2. Jo doesn’t know that Bo left. ⇝/⇝ Jo knows that Bo didn’t leave.

The MegaNegRaising dataset (MegaNeRd) consists of slider-based veridicality judgments aimed at capturing patterns like that in (1) and (2) as well as ordinal acceptability judgments for 925 clause-embedding verbs of English with a variety of subordinate clause structures, matrix tenses, and matrix subjects. For a detailed description of the dataset, the item construction and collection methods, and discussion of how to use a dataset on this scale to address questions in linguistic theory, please see the references below.


Sentences Predicates Frames Download Citation
7936 925 24 v1 (zip) An & White 2020


White, Aaron Steven. accepted with revisions. On believing and hoping whether. Semantics & Pragmatics. [pdf, code]
An, Hannah Youngeun, and Aaron Steven White. 2020. The Lexical and Grammatical Sources of Neg-Raising Inferences. In Proceedings of the Society for Computation in Linguistics 3: 220–233. [pdf, doi]


Aaron Steven White bio photo
Aaron Steven White
Hannah Youngeun An bio photo
Hannah Youngeun An