Xposition: An Online Multilingual Database of Adpositional Semantics

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1824–1830 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1824 Xposition: An Online Multilingual Database of Adpositional Semantics Luke Gessler Austin Blodgett Joseph Ledford† Nathan Schneider Georgetown University U.S. Army Research Lab Georgetown University lg876@georgetown.edu austin.j.blodgett@gmail.com {jcl240, nathan.schneider}@georgetown.edu Abstract We present Xposition, an online platform for documenting adpositional semantics across languages in terms of supersenses (Schneider et al., 2018). More than just a lexical database, Xposition houses annotation guidelines, structured lexicographic documentation, and annotated corpora. Guidelines and documentation are stored as wiki pages for ease of editing, and described elements (supersenses, adpositions, etc.) are hyperlinked for ease of browsing. We describe how the platform structures information; its current contents across several languages; and aspects of the design of the web application that supports it, with special attention to how it supports datasets and standards that evolve over time. Keywords: adpositions, prepositions, case, lexical semantics, supersenses, interfaces, meaning representations 1. Introduction Linguistic annotation is a challenging process that re- quires the coordination of annotation guidelines, pri- mary datasets, and derived datasets such as lexical inven- tories. Annotation guidelines virtually always require iterative revision as the process of annotation reveals flaws and edge-cases. Datasets must subsequently be brought in line with guidelines as they change, and an- alytical, derived datasets like lexical inventories must also be kept synchronized. Here we describe Xposition,1 a linguistic database and website that has helped the SNACS project (Schnei- der et al., 2018) address these issues in the process of annotating several multilingual corpora for adpositional semantics. Prepositions, postpositions, and case mark- ers across the world’s languages serve to mediate gram- matical relationships while often signaling the seman- tics of the relation. Across languages, such elements convey broadly the same kinds of meanings (related to space, time, causality, etc.), though the conceptual boundaries of particular adposition lexemes are highly variable across languages (e.g., Bowerman and Choi, 2001). SNACS provides a unified metalanguage for characterizing the major classes of meanings typically expressed with adpositions, allowing for multilingual study of their polysemy at the lexical level as well as disambiguation in corpora. We show how Xposition has addressed our need for an accessible presentation of SNACS corpora and guidelines for SNACS annotators and researchers across languages, supported by auxiliary documentation and lexical inventories. We focus particularly on our strategy of keeping all major parts of a multilingual annotation project—corpora, guidelines, and derived resources— together in a single web application and a single in- tegrated database. This strategy yields synergistic ef- †Work performed while at Georgetown University. 1http://xposition.org/ Figure 1: Xposition’s page for the DIRECTION supersense, one of 50 supersenses used by SNACS corpora to describe the meaning of an adposition. The page describes the meaning of the label, gives common examples which are hyperlinked to their occurrences in SNACS corpora (see Figure 2), and con- trasts it with similar supersenses. Like all pages in Xposition, the page is version-controlled and has …

Content truncated to 500 words