(Pre-)Annotation of Topic-Focus Articulation in Prague Czech-English Dependency Treebank

International Joint Conference on Natural Language Processing, pages 55–63, Nagoya, Japan, 14-18 October 2013. (Pre-)Annotation of Topic-Focus Articulation in Prague Czech-English Dependency Treebank Jiří Mírovský, Kateřina Rysová, Magdaléna Rysová, Eva Hajičová Charles University in Prague Faculty of Mathematics and Physics Institute of Formal and Applied Linguistics Czech Republic {mirovsky|rysova|magdalena.rysova|hajicova}@ufal.mff.cuni.cz Abstract The objective of the present contribution is to give a survey of the annotation of information structure in the Czech part of the Prague Czech-English Dependency Treebank. We re- port on this first step in the process of building a parallel annotation of information structure in this corpus, and elaborate on the automatic pre-annotation procedure for the Czech part. The results of the pre-annotation are evalu- ated, based on the comparison of the automatic and manual annotation. 1 Introduction In the past three or four decades, topic-focus ar- ticulation (known also as sentence information structure) is a language phenomenon that has at- tracted an enormous interest in linguistics and has become a “hot” topic of linguistic studies. No wonder then, that these days several linguis- tic teams (e.g. at the University of Potsdam, Uni- versity of Berlin, University of Stuttgart, Charles University in Prague) have attempted to include the annotation of information structure in the an- notating schemes they propose. Among corpora that contain also annotation of information struc- ture or such type of annotation is planned in them there are e.g. ANNIS database (Annotation of Information Structure, see Dipper et al., 2004), The English Switchboard Corpus (see Calhoun et al., 2005), the corpus DannPASS (Danish Phonetically Annotated Spontaneous Speech, see Paggio, 2006) and the Prague De- pendency Treebank (for the information on PDT, see Hajič et al., 2006). There are also several types of annotation guidelines and schemes for the different corpora, based on various linguistic theories dealing with information structure (e.g. Hajičová et al., 2000; Nissim et al., 2004; Dipper et al., 2007; Don- hauser, 2007; Cook and Bildhauer, 2011). In our paper, we present the annotation of topic-focus articulation in the Czech part of the Prague Czech-English Dependency Treebank, based on the theory of topic-focus articulation as developed withing the Praguian Functional Gen- erative Description. It is the first step in the process of building a parallel Czech-English cor- pus annotated with this type of linguistic infor- mation.1 1.1 Topic-Focus Articulation in Prague Treebanks The first complex and consistent theoreti- cally-based annotation of topic-focus articulation was already fully applied in the first Czech cor- pus from the Prague corpora family, the Prague Dependency Treebank (PDT; Hajič et al., 2006, updated in Bejček et al., 2012), and is available for the linguistic community. PDT is a large col- lection of Czech journalistic texts, (basically) manually annotated on several layers of language description (more than 3 thousand documents consisting of almost 50 thousand sentences are annotated on all the levels). Detailed annotation guidelines that constitute the basis of the handling with the language mate- rial were developed (Mikulová et al., 2005) based on the theoretical assumptions of the Func- …

Content truncated to 500 words