Proceedings of the Third Workshop on Metaphor in NLP

Abstract

ii IntroductionCharacteristic to all areas of human activity (from poetic to ordinary to scientific) and, thus, to all types of discourse, metaphor becomes an important problem for natural language processing. Its ubiquity in language has been established in a number of corpus studies and the role it plays in human reasoning has been confirmed in psychological experiments. This makes metaphor an important research area for computational and cognitive linguistics, and its automatic identification and interpretation indispensable for any semantics-oriented NLP application. This year's workshop is the third Metaphor in NLP workshop, following the first workshop held at NAACL 2013 and the second workshop held at ACL 2014. In 2013, accepted papers dealt with metaphor annotation, features for metaphor identification, and with generalization of the techniques across languages. These themes were also represented in the 2014 workshop, along with interpretation, applications, and relationships with related phenomena. In 2015, prominent themes include creation and utilization of semantic resources for metaphor identification and interpretation; features for metaphor identification that capture properties of concepts such as concreteness, imageability, affect, and sensorial modalities; relationships between social dynamic and individual history and metaphor use; and metaphor generation. We received 13 submissions and accepted 10, based on detailed and careful reviews by members of the Program Committee.Creation and utilization of semantic resources to support metaphor identification is a recurrent theme in the 2015 workshop. An invited talk by Prof. Martha Palmer and Dr. Susan Brown about metaphor in VerbNet was followed by a number of contributions describing the creation of resources in support of metaphor identification and analysis. Li, Bai ,Yin, and Xu describe the construction of a resource where salient properties of concepts expressed by thousands of Chinese verbs and nouns are collected. Dodge, Hong, and Stickles describe MetaNet, a system combining a repository of metaphors and frames, and a metaphor detection component that utilizes the repository. Gordon, Jobbs, May, and Morbini describe an enhancement to their knowledge-based metaphor identification system that infers lexical axioms -rules which encode information about what words or phrases trigger particular source and target concepts.Gordon, Hobbs, May, Mohler, Morbini, Rink, Tomlinson, and Wertheim describe their ontology of commonly used source domains and release a corpus of manually validated annotations of linguistic metaphors about governance, economy, and gun control with source and target domains, as well as specific roles (slots) that support the interpretation of the metaphor. For example, according to the ontology, a metaphor drawing on the source domain of JOURNEY can be annotated with elements such as source, target, agent, goal, facilitator, barrier, change, and type of change (increase or decrease). The goal of the dataset is to support the an...

Comparing Distributional and Curated Approaches for Cross-lingual Frame Alignment

“…Despite progress on automatic recognition of metaphors (e.g. Veale 2016, Shutova et al 2015, Chakrabarty et al 2021, the kind of detail shown here is generally not retrievable computationally. Lane Lawley and Lenhart Schubert generate"Logical Story Representations via FrameNet + Semantic Parsing".…”