Automatic prediction of aspectual class of verbs in context

Abstract

This paper describes a new approach to predicting the aspectual class of verbs in context, i.e., whether a verb is used in a stative or dynamic sense. We identify two challenging cases of this problem: when the verb is unseen in training data, and when the verb is ambiguous for aspectual class. A semi-supervised approach using linguistically-motivated features and a novel set of distributional features based on representative verb types allows us to predict classes accurately, even for unseen verbs. Many frequent verbs can be either stative or dynamic in different contexts, which has not been modeled by previous work; we use contextual features to resolve this ambiguity. In addition, we introduce two new datasets of clauses marked for aspectual class.

Automatic recognition of habituals: a three-way classification of clausal aspect

“…Table 2 shows the syntactic-semantic features, which we call CONTEXT-BASED as they are extracted from the context of each verb occurrence that we classify. This feature set comprises the features proposed by Mathew and Katz (2009) and the ones proposed by Friedrich and Palmer (2014). In addition, we use the features modal and negated.…”

“…This task predicts the aspectual class of a verb in context before any aspectual markers or transformations (such as use of the perfect or modals) have been applied. Siegel and McKeown (2000) propose the use of linguistic indicators (explained in Section 5.2); Friedrich and Palmer (2014) show the importance of using context-based features in addition. Zarcone and Lenci (2008) classify occurrences of 28 Italian verbs according to Vendlers classes state, process, accomplishment and achievement.…”

“…The distinction between dynamic and stative lexical aspect, as in examples (1a) and (1b) respectively, is fundamental (Vendler, 1957). Its automatic prediction has been addressed previously (Siegel and McKeown, 2000;Zarcone and Lenci, 2008;Friedrich and Palmer, 2014).…”

“…Linguistically, the determination of clausal aspect depends on the recognition of the verb's lexical aspectual class (stative or dynamic), and on the recognition of any aspectual markers or transformations, such as use of the perfect tense, negations or modals. Our work builds on results for the related subtasks (Mathew and Katz, 2009;Siegel and McKeown, 2000;Friedrich and Palmer, 2014), using both context-based and verbtype based information.…”

Classification of telicity using cross-linguistic annotation projection

“…Their classifier employs a similar but smaller feature set compared to ours. Other related work on predicting aspect include systems aiming at identifying lexical aspect (Siegel and McKeown, 2000;Friedrich and Palmer, 2014) or habituals (Mathew and Katz, 2009;Friedrich and Pinkal, 2015).…”