Abstrakt
This paper addresses the automatic recog-nition of telicity, an aspectual notion. Atelicevent includes a natural endpoint (shewalked home), while anatelicevent doesnot (she walked around).
Recognizing thisdifference is a prerequisite for temporalnatural language understanding. In En-glish, this classification task is difficult, astelicity is a covert linguistic category.
Incontrast, in Slavic languages, aspect is partof a verb's meaning and even available inmachine-readable dictionaries. Our con-tributions are as follows.
We successfullyleverage additional silver standard train-ing data in the form of projected annota-tions from parallel English-Czech data aswell as context information, improving au-tomatic telicity classification for Englishsignificantly compared to previous work.We also create a new data set of English texts manually annotated with telicity.