MegaIntensionality

Much recent work in the formal semantics literature has attempted to relate lexically triggered inferences to predicates’ intensional properties. In particular, the properties of a predicate related to belief or desire of one or both participants give rise to doxastic and bouletic inferences, respectively. For example, (1a) gives rise to the inference (2a), while (1b) gives rise to the inference (2b).

1. a. Jo knew that Bo left.
   b. Jo liked that Bo left.

2. a. Jo believed that Bo left.
   b. Jo wanted Bo to have left.

These inferences are of particular interest due to the patterns that emerge in how the inferences are affected by contexts such as negation. Doxastic and bouletic inferences are said to be “foregrounded” if they are at-issue and targeted by negation, and “backgrounded” if they are insensitive to negation. For example, while (3a) gives rise to the inference (4a), (3b) gives rise to (2a), suggesting that the doxastic inference for like is backgrounded. In contrast, the bouletic inference in this case is foregrounded, as evidenced by the fact that (3b) gives rise to (4b). The question of which correlations between these properties (and other lexical properties) are possible has received considerable attention in recent work.

3. a. Jo didn’t know that Bo left.
   b. Jo didn’t like that Bo left.

4. a. Jo didn’t believe that Bo left.
   b. Jo didn’t want Bo to have left.

The MegaIntensionality dataset consists of slider-based judgments of doxastic and bouletic inferences for 725 finite 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.

Data

References

Researchers

Aaron Steven White

Benjamin Kane

William Gantt