Aspectuality Across Genre: A Distributional Semantics Approach

Abstract

The interpretation of the lexical aspect of verbs in English plays a crucial role for recognizing textual entailment and learning discourse-level inferences. We show that two elementary dimensions of aspectual class, states vs. events, and telic vs. atelic events, can be modelled effectively with distributional semantics. We find that a verb's local context is most indicative of its aspectual class, and demonstrate that closed class words tend to be stronger discriminating contexts than content words. Our approach outperforms previous work on three datasets. Lastly, we contribute a dataset of human-human conversations annotated with lexical aspect and present experiments that show the correlation of telicity with genre and discourse goals.

About Time: Do Transformers Learn Temporal Verbal Aspect?

“…Moving away from hard-coded annotations and lexical aspect, Peng (2018) uses two different compositional models to classify aspect, exploring the entire clause and not only the verb, with the use of distributional vectors and with-out annotated linguistic features, and highlights the importance of the verbal phrase and the verb's dependents in the interpretation of telicity. Kober et al (2020) propose modeling aspect of English verbs in context, with the use of compositional distributional models, and confirm that a verb's context and closed-class words of tense are strong features for aspect classification.…”

Proceedings of the 15th International Conference on Natural Language Generation

“…Pustejovsky et al (2010) described how aspect must be considered for event annotations and Baiamonte et al (2016) incorporated lexical aspect in the study of the rhetorical structure of text. Kober et al (2020) presented a supervised model for studying aspectuality in unimodal scenarios only in English. In this work however, we focus on image captions that enable us to better understand how humans describe images.…”

“…Our use of distributional semantic representations is furthermore motivated by the fact that they are readily available in numerous languages, and that they, contrary to manually constructed lexicons such as VerbNet (Schuler and Palmer, 2005) or LCS (Dorr and Olsen, 1997), scale well with growing amounts of data and across different languages. Furthermore, there is a growing body of evidence that models based on the distributional hypothesis capture some facets of aspect (Kober et al, 2020;Metheniti et al, 2022), despite the fact that aspect is represented in a very diverse manner across languages.…”

“…The aim of our first experiment is to establish that aspect can be classified for our set of languages with distributional representations in a supervised setting as has been shown on English data (Kober et al, 2020). Figure 3 shows the difference in Accuracy of our models in comparison to a majority class baseline.…”

A Kind Introduction to Lexical and Grammatical Aspect, with a Survey of Computational Approaches

“…While there has been a notable amount of work on modeling lexical and grammatical aspect in the computational linguistics community (Moens and Steedman, 1988;Siegel and McKeown, 2000b;Friedrich et al, 2016;Kober et al, 2020), this area is still a niche in natural language processing (NLP). In this paper, we survey the state of research in this area and argue that a good computational understanding of lexical and grammatical aspect is Figure 1: Aspect is like the camera lens of language, the device by which we focus on particular phases of a situation (Vendler, 1957;Smith, 1997).…”