Evaluating Hierarchies of Verb Argument Structure with Hierarchical Clustering

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 986–991 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Evaluating Hierarchies of Verb Argument Structure with Hierarchical Clustering Jesse Mu and Joshua K. Hartshorne Department of Psychology Boston College Chestnut Hill, MA {muj,hartshoj}@bc.edu Timothy J. O’Donnell Department of Linguistics McGill University Montreal, Canada timothy.odonnell@mcgill.ca Abstract Verbs can only be used with a few spe- cific arrangements of their arguments (syn- tactic frames). Most theorists note that verbs can be organized into a hierarchy of verb classes based on the frames they admit. Here we show that such a hierar- chy is objectively well-supported by the patterns of verbs and frames in English, since a systematic hierarchical clustering algorithm converges on the same structure as the handcrafted taxonomy of VerbNet, a broad-coverage verb lexicon. We also show that the hierarchies capture meaning- ful psychological dimensions of general- ization by predicting novel verb coercions by human participants. We discuss limi- tations of a simple hierarchical represen- tation and suggest similar approaches for identifying the representations underpin- ning verb argument structure. 1 Introduction Why can Sally like to read but not *appreciate to read? Key to the grammar of sentences are verbs and the arguments with which they appear. How children learn the constraints that govern the ways verbs and arguments combine is a central question in language acquisition. Theorists have long noted that verbs can be or- ganized into classes based on their syntactic con- structions and the events they express (see Levin and Rappaport Hovav, 2005 for review). Verb classes are included in most theories of argument structure acquisition, whether as first class objects (Perfors et al., 2010) or mere epiphenomena of other claims about the structure of form-meaning mappings (Pinker, 1989; Goldberg, 1995). Most theories also propose further structure be- tween classes. One common assumption is that verb argument structure can be at least partially described by a hierarchy: Each verb belongs to a class, which itself may belong to a number of broader superclasses. While many theories predict more complex structure (e.g. cross-cutting categories; Levin and Rappaport Hovav, 2005), providing (psy- cho)linguistic evidence for a simple hierarchy of verbs is an important starting point for investi- gating more complex theories. VerbNet (Kipper et al., 2008), the largest English verb argument structure resource,1 organizes verbs and classes into a shallow hierarchy, but its structure has been handcrafted incrementally over time (starting with seminal work by Levin, 1993). On the other hand, recently-developed, state-of-the-art machine learning methods offer a unique alternative ap- proach to constructing such a hierarchy. In this paper, we first conduct a broad-coverage analysis of how verbs might be hierarchically ar- ranged by comparing VerbNet’s handcrafted hi- erarchy to structure systematically inferred by a Bayesian hierarchical clustering algorithm. We find that the two arrive at similar structure, thus substantiating both methods (i.e. intuition vs. clus- tering) and the common hierarchy they find. Second, we investigate the psychological valid- ity of this representation: if …

Content truncated to 500 words