Syntax and Semantics in a Treebank for Esperanto

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 5120–5127 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 5120 Syntax and Semantics in a Treebank for Esperanto Eckhard Bick University of Southern Denmark Odense, Denmark eckhard.bick@mail.dk Abstract In this paper we describe and evaluate syntactic and semantic aspects of Arbobanko, a treebank for the artificial language Esperanto, as well as tools and methods used in the production of the treebank. In addition to classical morphosyntax and dependency structure, the treebank was enriched with a lexical-semantic layer covering named entities, a semantic type ontology for nouns and adjectives and a framenet-inspired semantic classification of verbs. For an under-resourced language, the quality of automatic syntactic and semantic pre-annotation is of obvious importance, and by evaluating the underlying parser and the coverage of its semantic ontologies, we try to answer the question whether the language's extremely regular morphology and transparent semantic affixes translate into a more regular syntax and higher parsing accuracy. On the linguistic side, the treebank allows us to address and quantify typological issues such as the question of word order, auxiliary constructions, lexical transparency and semantic type ambiguity in Esperanto. Keywords: Treebanks, Esperanto, Dependency Grammar, Constraint Grammar, Syntactic parsing, Free word order languages 1. Introduction Treebanks satisfy important needs in both language technology and descriptive linguistics, allowing the latter to identify and quantify linguistic patterns, and the former to train and evaluate machine-learned parsers. With a general change of focus towards language technology, dependency treebanks have become more prevalent at the expense of constituent treebanks, driven by methodological considerations such as implementability in mathematical models (graphs). Historically, dependency syntax has roots in the linguistic tradition of Slavic languages, one of its strengths being the handling of free word order and discontinuities, while constituent grammar departed from English with its fixed word order and reliable subject-predicate pairs. Thus, the first and largest dependency treebank was published for Czech (Böhmová et al. 2003, Bejček et al. 2013), while major English treebanks like the Penn Treebank were originally annotated with phrase structure and converted to the dependency format only later (Johansson & Nugues 2007), by the machine-learning (ML) community. A third approach was used for the Danish Arboretum treebank (Bick 2003), where a rule-based Constraint Grammar (CG) parser was used to create shallow dependency trees that were then converted to constituent trees, using manual revision at both stages. As an artificial language with a non-negligible living speaker community and several generations of native speakers, Esperanto is a linguistically interesting language, albeit under-resourced in terms of both development/research funding and existing NLP resources. Our treebank project intends to address this issue at both the linguistic and NLP levels. We decided on a dependency format not only because of the current focus of the research community, but also because of the purported free word order-characteristics of the language. In addition, the only available parser was a CG dependency parser, and we needed to minimize (unfunded) human …

Content truncated to 500 words