Degree estimates as a measure of inference calculation

Scalar inference (SI), e.g., utterances containing the quantifier some being enriched to mean but not all, is a central topic in semantics-pragmatics. Of recent interest experimental literature phenomenon of scalar diversity: that different lexical scales exhibit variation how likely they are lead SI. However, studies diversity have almost exclusively relied on particular task: task. In this paper, we argue task suffers from number shortcomings: namely, it biases by providing participants with stronger alternative and obscures pragmatic inferences other than Instead offer as an degree estimate investigate terms. We use reassess previous task-based findings two manipulations (discourse context only) affect likelihood calculation. Our results show tasks produce differ each subtle important ways.