Building Analyses from Syntactic Inference in Local Languages: An HPSG Grammar Inference System

Building Analyses from Syntactic Inference in Local Languages: An HPSG Grammar Inference System Kristen Howell, University of Washington and LivePerson Inc., USA kphowell@uw.edu Emily M. Bender, University of Washington, USA ebender@uw.edu Abstract We present a grammar inference system that leverages linguistic knowledge recorded in the form of annotations in interlinear glossed text (IGT) and in a meta-grammar engineering system (the LinGO Grammar Matrix customization system) to automatically produce machine-readable HPSG grammars. Building on prior work to handle the inference of lexical classes, stems, afixes and position classes, and preliminary work on inferring case systems and word order, we introduce an integrated grammar inference system called basil that covers a wide range of fundamental linguistic phenomena. System development was guided by 27 genealogically and geographically diverse languages, and we test the system’s cross-linguistic generalizability on an additional 5 held-out languages, using datasets provided by field linguists. Our system out-performs three baseline systems in increasing coverage while limiting ambiguity and producing richer semantic representations, while also producing richer representations than previous work in grammar inference. 1 Introduction Machine-readable grammars for human languages that are grounded in theoretical syntactic formalisms can be useful tools in the context of endangered language doc- umentation and revitalization. First, they support tree- banking (Oepen et al., 2002), which in turn supports data exploration (Letcher and Baldwin, 2013; Bouma et al., 2015); and second, they facilitate the development of tools such as grammar checkers (da Costa et al., 2016) and automated tutors (Hellan et al., 2013). In spite of these advantages, the use of such grammars is hindered by the time-consuming process of developing them to- gether with the need of a specific skillset required for grammar engineering, which is distinct from the skills involved in documentation itself. We are therefore mo- tivated to investigate whether we can create machine- readable grammars automatically.1 Endangered lan- guages represent scenarios where the type of resources required for typical natural language processing tech- niques are scarce to non-existent. Furthermore, the out- put we are targeting goes well beyond simple labels or even structured representations, but rather must be a coherent and well-formed formal object — a grammar. Fortunately, we have two rich sources of linguistic 1This is similar in spirit to the work of Sarveswaran et al. (2019) who present an efort to create FSMs to provide computational bene- fits in the context of morphological analysis without requiring addi- tional technical skillsets. knowledge from which to work: The first is corpora of interlinear glossed text (IGT), annotated by field lin- guists during the process of documentation and anal- ysis. Due to the eforts of field linguists and archivists, a number of archives (many of which we list in Ap- pendix A) make IGT data publicly available. An ex- ample from Chintang [ISO 639-3: ctn] is shown in (1). Such annotations are linguistically rich, showing what grammatical information is marked morphologically and providing further information implicitly via a trans- lation into a language of broader communication (in all examples we work with, …

Content truncated to 500 words