Developing a Spell and Grammar Checker for Icelandic using an Error Corpus

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 4644–4653 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 4644 Developing a Spell and Grammar Checker for Icelandic Using an Error Corpus Hulda Óladóttir1, Þórunn Arnardóttir2, Anton K. Ingason2, Vilhjálmur Þorsteinsson1 1Miðeind ehf., Fiskislóð 31 B/303, 101 Reykjavík, Iceland 2University of Iceland, Sæmundargata 2, 102 Reykjavík, Iceland 1{hulda, vt}@mideind.is, 2{thar, anton}@hi.is Abstract A lack of datasets for spelling and grammatical error correction in Icelandic, along with language-specific issues, has caused a dearth of spell and grammar checking systems for the language. We present the first open-source spell and grammar checking tool for Icelandic, using an error corpus at all stages. This error corpus was in part created to aid in the development of the tool. The system is built with a rule-based tool stack comprising a tokenizer, a morphological tagger, and a parser. For token-level error annotation, tokenization rules, word lists, and a trigram model are used in error detection and correction. For sentence-level error annotation, we use specific error grammar rules in the parser as well as regex-like patterns to search syntax trees. The error corpus gives valuable insight into the errors typically made when Icelandic text is written, and guided each development phase in a test-driven manner. We assess the system’s performance with both automatic and human evaluation, using the test set in the error corpus as a reference in the automatic evaluation. The data in the error corpus development set proved useful in various ways for error detection and correction. Keywords: spell checker, grammar checker, spelling error correction, grammatical error correction, SEC, GEC, error corpus, Icelandic, morphologically rich languages, low-resource languages, medium-resource languages 1. Introduction Spell and grammar checking of text is a well- established task within language technology, and checking tools are readily available for major lan- guages. However, this task has to date not been suf- ficiently addressed with open and accessible solutions for Icelandic, a low-resource language (Rögnvaldsson, 2022). Spell and grammar checking was therefore in- cluded as one of the key components of the Icelandic government’s strategic 5-year Language Technology Programme for Icelandic (Nikulásdóttir et al., 2020), where all deliverables are open-source and made avail- able in the Icelandic CLARIN repository. We describe our approach to the task, which is repro- ducible for other low- or medium-resource languages and especially well-suited to morphologically rich lan- guages. The method employs different language re- sources, including the newly-created Icelandic Error Corpus (Arnardóttir et al., 2021), to guide the develop- ment of GreynirCorrect, a pre-existing Icelandic spell and grammar checker. The checker is available in the Icelandic CLARIN repository and on GitHub under the MIT license (Þorsteinsson et al., 2020). The paper is structured as follows. Section 2 discusses methods for developing and evaluating spell and gram- mar checkers, the state of the art for Icelandic, and language-specific issues. Section 3 focuses on lan- guage resources used for the development of the Ice- landic spell and grammar checker. …

Content truncated to 500 words