ForFun 1.0: Prague Database of Forms and Functions – An Invaluable Resource for Linguistic Research

ForFun 1.0: Prague Database of Forms and Functions An Invaluable Resource for Linguistic Research Marie Mikulová, Eduard Bejček Charles University, Faculty of Mathematics and Physics, Institute of Formal and Applied Linguistics Malostranské náměstí 25, 118 00 Prague 1, Czech Republic {mikulova,bejcek}@ufal.mff.cuni.cz Abstract In this paper, we introduce the first version of ForFun, Prague Database of Forms and Functions, as an invaluable resource for profound linguistic research, particularly in describing syntactic functions and their formal realizations. ForFun is built with the use of already existing richly syntactically annotated corpora, collectively called Prague Dependency Treebanks. ForFun brings this complex annotation of Czech sentences closer to researchers. We demonstrate that ForFun 1.0 provides valuable and rich material allowing to elaborate various syntactic issues in depth. We believe that nowadays when corpus linguistics differs from traditional linguistics in its insistence on a systematic study of authentic examples of language in use, our database will contribute to the comprehensive syntactic description. Keywords: language resource, dependency syntax, semantic labeling, surface form, digital humanities 1. Motivation What is the difference between location expressions “walk in King Street”, “walk on King Street”, and “walk along King Street”? Should we use a different preposition when talking about destination rather than about direction and lo- cation? Or more precisely, what function does the preposi- tion “on” perform in contrast to the preposition “along” and which forms can express destination? Is the same form used in both spoken and written text? Is there any bias towards one form in translated text? For Czech, the answers can be found in a new database for inspecting thousands of real examples categorized by their form (e.g. by a prepositional case) as well as by their deep syntactic function. 2. Introduction In this paper, we present the first version of ForFun, Prague Database of Forms and Functions, as an invaluable resource for different linguistic issues, particularly for the descrip- tion of syntactic functions and their formal realizations. It takes advantage of several richly syntactically annotated corpora, collectively called Prague Dependency Treebanks (PDTs in the sequel) that have already been developed in Prague. Altogether, the treebanks contain around 180,000 sentences with their morphological, syntactic and semantic annotation. The ForFun database draws on the complex lin- guistic annotation of these corpora, arranges selected mor- phological and syntactical annotation into new shape, and offers a user-friendly access to a large resource of real ex- amples. 3. Related Work There is a wide range of corpora with rich linguistic anno- tation, e.g., Penn Treebank (Marcus et al., 1999), its suc- cessors PropBank (Kingsbury and Palmer, 2002) and Nom- Bank (Meyers et al., 2004); for German, there is Tiger (Brants et al., 2002) and Salsa (Burchardt et al., 2006), and many others. The ForFun database is unique in that it is compiled from four different treebanks of Czech, uniformly annotated using the same scenario, with data coming from text, speech and Internet sources. It offers a really large material with the deep syntactic manual annotation which is well and comprehensibly …

Content truncated to 500 words