Automatic Focus Annotation: Bringing Formal Pragmatics Alive in Analyzing the Information Structure of Authentic Data

Proceedings of NAACL-HLT 2018, pages 117–128 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Automatic Focus Annotation: Bringing Formal Pragmatics Alive in Analyzing the Information Structure of Authentic Data Ramon Ziai Detmar Meurers Collaborative Research Center 833 University of T¨ubingen {rziai,dm}@sfs.uni-tuebingen.de Abstract Analyzing language in context, both from a theoretical and from a computational perspec- tive, is receiving increased interest. Com- plementing the research in linguistics on dis- course and information structure, in compu- tational linguistics identifying discourse con- cepts was also shown to improve the perfor- mance of certain applications, for example, Short Answer Assessment systems (Ziai and Meurers, 2014). Building on the research that established de- tailed annotation guidelines for manual anno- tation of information structural concepts for written (Dipper et al., 2007; Ziai and Meur- ers, 2014) and spoken language data (Calhoun et al., 2010), this paper presents the first ap- proach automating the analysis of focus in au- thentic written data. Our classification ap- proach combines a range of lexical, syntactic, and semantic features to achieve an accuracy of 78.1% for identifying focus. 1 Introduction The interpretation of language is well known to depend on context. Both in theoretical and com- putational linguistics, discourse and information structure of sentences are thus receiving increased interest: attention has shifted from the analysis of isolated sentences to the question how sentences are structured in discourse and how information is packaged in sentences analyzed in context. As a consequence, a rich landscape of ap- proaches to discourse and information struc- ture has been developed (Kruijff-Korbayov´a and Steedman, 2003). Among these perspectives, the Focus-Background dichotomy provides a particu- larly valuable structuring of the information in a sentence in relation to the discourse. (1) is an ex- ample question-answer pair from Krifka and Mu- san (2012, p. 4) where the focus in the answer is marked by brackets. (1) Q: What did John show Mary? A: John showed Mary [[the PICtures]]F . In the answer in (1), the NP the pictures is fo- cussed and hence indicates that there are alterna- tive things that John could show Mary. It is com- monly assumed that focus here typically indicates the presence of alternative denotations (denotation focus, Krifka and Musan 2012, p.8), making it a semantic notion. Depending on the language, different devices are used to mark focus, such as prosodic focus marking or different syntactic con- structions (e.g. clefts). In this paper, we adopt a notion of focus based on alternatives, as advanced by Rooth (1992) and more recently, Krifka and Musan (2012), who define focus as indicating “the presence of alternatives that are relevant for the in- terpretation of linguistic expressions” (Krifka and Musan, 2012, p. 7). Formal semantics has tied the notion of alternatives to an explicit relationship between questions and answers called Question- Answer Congruence (Stechow, 1991), where the idea is that an answer is congruent to a question if both evoke the same set of alternatives. Questions can thus be seen as a way …

Content truncated to 500 words