Automatic Metaphor Interpretation as a Paraphrasing Task

Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the ACL, pages 1029–1037, Los Angeles, California, June 2010. c⃝2010 Association for Computational Linguistics Automatic Metaphor Interpretation as a Paraphrasing Task Ekaterina Shutova Computer Laboratory University of Cambridge 15 JJ Thomson Avenue Cambridge CB3 0FD, UK Ekaterina.Shutova@cl.cam.ac.uk Abstract We present a novel approach to metaphor in- terpretation and a system that produces lit- eral paraphrases for metaphorical expressions. Such a representation is directly transferable to other applications that can benefit from a metaphor processing component. Our method is distinguished from the previous work in that it does not rely on any hand-crafted knowl- edge about metaphor, but in contrast employs automatically induced selectional preferences. Being the first of its kind, our system is capa- ble of paraphrasing metaphorical expressions with a high accuracy (0.81). 1 Introduction Metaphors arise when one concept is viewed in terms of the properties of the other. In other words it is based on similarity between the concepts. Sim- ilarity is a kind of association implying the presence of characteristics in common. Here are some exam- ples of metaphor. (1) News travels fast. (Lakoff and Johnson, 1980) (2) How can I kill a process? (Martin, 1988) (3) And then my heart with pleasure fills, And dances with the daffodils.1 In metaphorical expressions seemingly unrelated features of one concept are associated with another 1taken from the verse “I wandered lonely as a cloud” written by William Wordsworth in 1804. concept. In the example (2) the computational pro- cess is viewed as something alive and, therefore, its forced termination is associated with the act of killing. Metaphorical expressions represent a great vari- ety, ranging from conventional metaphors, which we reproduce and comprehend every day, e.g. those in (1) and (2), to poetic and largely novel ones, such as (3). The use of metaphor is ubiquitous in natural language text and it is a serious bottleneck in auto- matic text understanding. In order to estimate the frequency of the phenomenon, we conducted a cor- pus study on a subset of the British National Corpus (BNC) (Burnard, 2007) representing various genres. We manually annotated metaphorical expressions in this data and found that 241 out of 761 sentences contained a metaphor or (rarely) an idiom. Due to such a high frequency of their use, a system capa- ble of interpreting metaphorical expressions in unre- stricted text would become an invaluable component of any semantics-oriented NLP application. Automatic processing of metaphor can be clearly divided into two subtasks: metaphor recognition (distinguishing between literal and metaphorical language in text) and metaphor interpretation (iden- tifying the intended literal meaning of a metaphori- cal expression). Both of them have been repeatedly addressed in NLP. To date the most influential account of metaphor recognition has been that of Wilks (1978). Accord- ing to Wilks, metaphors represent a violation of se- lectional restrictions in a given context. Consider the following example. (4) My car drinks gasoline. (Wilks, 1978) 1029 The verb drink normally takes …

Content truncated to 500 words