The Semantics of Gradable Modal Expressions

Annotation of Modal Meaning

Paul Portner, Aynat Rubinstein, and Graham Katz organized a workshop on the annotation of modal expressions at 10th International Conference on Computational Semantics at the University of Potsdam in Potsdam, Germany.

Representing the semantics of modal verbs (must), adjectives (likely) and nouns (chance), verbs of propositional attitude (believe, want), hedging adverbs (maybe, probably) and other expressions whose meaning is related to non-actual possibilities has been an important domain of research for philosophers of language, logicians, natural language semanticists, and computational linguists for decades. In recent years, researchers in computational semantics have started to focus their attention on the interpretation of these expressions, as it has become clearer that identifying modal meaning has important consequences for such practical tasks as the computation of textual entailments, factuality status and sentiment analysis.

The Workshop on the Annotation of Modal Meaning (WAMM) brought together researchers to discuss recent progress in developing techniques for the annotation of modal meaning in texts. The program included eight talks by speakers from Europe and the United States, including two groups from Georgetown Linguistics, one focusing on English and the other on Mandarin Chinese.

This workshop is part of an NSF-funded project on the semantics and use of gradable modal expressions (NSF BCS-1053038) led by Paul Portner, Elena Herburger and Graham Katz. Modals expressions are those which allow us to talk about possibilities that are not necessarily real – as in You can stop smoking, which is true regardless of whether or not you do stop smoking. Modal properties, like being something you should do, are not absolute; they can come in degrees, like I am barely able to go an hour without smoking. The most well-studied gradable modal language concerns probability: phrases like more probable/likely, and 80% probability, but the phenomenon is pervasive. The goal of this research is to better understand the meaning and function of modal language by focusing on its use in gradable expressions.