Keep it Surprisingly Simple: A Simple First Order Graph Based Parsing Model for Joint Morphosyntactic Parsing in Sanskrit

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 4791–4797, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 4791 Keep It Surprisingly Simple: A Simple First Order Graph Based Parsing Model for Joint Morphosyntactic Parsing in Sanskrit Amrith Krishna1, Ashim Gupta2, Deepak Garasangi3∗, Pavankumar Satuluri4 and Pawan Goyal5 1Department of Computer Science and Technology, University of Cambridge 2School of Computing, University of Utah 3Independent Researcher 4School of Linguistics & Literary Studies, Chinmaya Vishwavidyapeeth 5Department of Computer Science and Engineering, IIT Kharagpur ak2329@cam.ac.uk, ashim@cs.utah.edu, pawang@cse.iitkgp.ac.in Abstract Morphologically rich languages seem to ben- efit from joint processing of morphology and syntax, as compared to pipeline architectures. We propose a graph-based model for joint morphological parsing and dependency pars- ing in Sanskrit. Here, we extend the En- ergy based model framework (Krishna et al., 2020), proposed for several structured predic- tion tasks in Sanskrit, in 2 simple yet signifi- cant ways. First, the framework’s default input graph generation method is modified to gener- ate a multigraph, which enables the use of an exact search inference. Second, we prune the input search space using a linguistically moti- vated approach, rooted in the traditional gram- matical analysis of Sanskrit. Our experiments show that the morphological parsing from our joint model outperforms standalone morpho- logical parsers. We report state of the art re- sults in morphological parsing, and in depen- dency parsing, both in standalone (with gold morphological tags) and joint morphosyntactic parsing setting. 1 Introduction Morphology and syntax are often inextricably inter- twined for morphologically rich languages (MRLs). For such languages, it might be unrealistic to design dependency parsers that expect correct morpholog- ical tags to be provided as input (More et al., 2019; Bohnet et al., 2013). Jointly modelling morpholog- ical parsing (MP) with dependency parsing (DP) has shown to be effective for several MRLs (More et al., 2019). In this work, we present multigraph- EBM (MG-EBM), a joint model for morphosyntac- tic parsing, i.e. joint MP and DP, in Sanskrit. Morphosyntactic parsing has been successfully applied to several MRLs. Bohnet et al. (2013) pro- posed a transition based joint parser, extending the ∗Work done while at IIT Kharagpur joint POS tagger and dependency parser of Bohnet and Nivre (2012). Similarly Seeker and C¸ etino˘glu (2015) proposed a joint graph based parser for Turk- ish. Here, two different models, one predicting a morphological path and the other a dependency tree, are made to reach an agreement using dual decomposition. More et al. (2019) proposed a tran- sition based joint parser for Hebrew, where it aims to maximise a global score over both morphologi- cal and dependency transitions. Sanskrit is an MRL which shows high degree of syncretism and homonymy in its morphological paradigm. About 90.96 % of the tokens in a dataset of 115,000 Sanskrit sentences (Krishna et al., 2017) show syncretism with an average of 3.62 morpho- logical tags per token. Morphological features, especially case, are indicative of the syntactic roles that a word (nominal) can assume in a …

Content truncated to 500 words