Exploiting Frame-Semantics and Frame-Semantic Parsing for Automatic Extraction of Typological Information from Descriptive Grammars of Natural Languages

Proceedings of Recent Advances in Natural Language Processing, pages 1247–1256, Varna, Bulgaria, Sep 2–4, 2019. https://doi.org/10.26615/978-954-452-056-4_143 1247 Exploiting Frame Semantics and Frame-Semantic Parsing for Automatic Extraction of Typological Information from Descriptive Grammars of Natural Languages Shafqat Mumtaz Virk1, Azam Sheikh Muhammad2, Lars Borin3, Muhammad Irfan Aslam4, Saania Iqbal5, and Nazia Khurram6 1,3Språkbanken, University of Gothenburg, Sweden {shafqat.virk, lars.borin}@svenska.gu.se 2Department of CS & Eng., Chalmers University of Technology, Sweden, azams@chalmers.se 4University of Skövde, Sweden, irfan.aslam@hotmail.com 5,6GIFT University, Pakistan, sania__iqbal@hotmail.com, nazia.yousaf@gmail.com Abstract We describe a novel system for automatic extraction of typological linguistic infor- mation from descriptive grammars of nat- ural languages, applying the theory of frame semantics in the form of frame- semantic parsing. The current proof-of- concept system covers a few selected lin- guistic features, but the methodology is general and can be extended not only to other typological features but also to de- scriptive grammars written in languages other than English. Such a system is ex- pected to be a useful assistance for au- tomatic curation of typological databases which otherwise are built manually, a very labor and time consuming as well as cog- nitively taxing enterprise. 1 Introduction There are more than 7,000 living languages in the world and grammatical descriptions1 are available for some 4,000 of these (Seifart et al., 2017). A central concern of the academic discipline of lin- guistics is to classify languages along different dimensions. The subbranch of linguistics which deals with classification and comparison of lan- guages based on their structural and functional characteristics is known as linguistic typology (or typological linguistics. In addition to comparing the world’s languages, practitioners of linguistic typology aim to explore the distribution of vari- ous structural and functional patterns among lan- guages and to explain them in historical and/or universal terms. (Song, 2010) 1Grammatical descriptions are plain text descriptions of various phonological, morphological, and syntactic charac- teristics of languages. To achieve its goals, typological linguistics has relied largely on manual reading of available de- scriptive material about languages for the extrac- tion of pertinent feature values for comparison of these across languages. For example, in their sen- tence structure, different languages favor different word orders (e.g. subject-object-verb, verb-object- subject, or object-verb-subject, etc). If word order is to be used as one structural feature for language comparison, the word order of all the languages to be compared has to be found out manually by reading available material on those languages. This is doable, if the scope of comparison is to be limited to a few features spanning across a few lan- guages. If one aims to extend the scope from a few to a few hundred features spanning across thou- sands of languages, the manual reading and com- parison strategy seems simply unfeasible. With the availability of large amounts of digital data, and the recent advances in natural language process- ing, automatic extraction of typological features from grammatical descriptions seems a plausible task. There have already been a few attempts to au- tomatically extract typological and other linguis- …

Content truncated to 500 words