NyLLex: A Novel Resource of Swedish Words Annotated with Reading Proficiency Level

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1326–1331 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1326 NyLLex: A Novel Resource of Swedish Words Annotated with Reading Proficiency Level Daniel Holmer, Evelina Rennes Department of Computer and Information Science Link¨oping University, Link¨oping, Sweden {daniel.holmer, evelina.rennes}@liu.se Abstract What makes a text easy to read or not, depends on a variety of factors. One of the most prominent is, however, if the text contains easy, and avoids difficult, words. Deciding if a word is easy or difficult is not a trivial task, since it depends on characteristics of the word in itself as well as the reader, but it can be facilitated by the help of a corpus annotated with word frequencies and reading proficiency levels. In this paper, we present NyLLex, a novel lexical resource derived from books published by Sweden’s largest publisher for easy language texts. NyLLex consists of 6,668 entries, with frequency counts distributed over six reading proficiency levels. We show that NyLLex, with its novel source material aimed at individuals of different reading proficiency levels, can serve as a complement to already existing resources for Swedish. Keywords: lexicon, easy language, reading proficiency, text complexity 1. Introduction Being able to decode and comprehend text is an impor- tant skill for participating in the modern society. How- ever, there are many groups of people that, for various reasons, struggle with reading, and could be helped by more accessible texts. In order to help writers create more accessible texts, easy language guidelines have been developed. For Swedish, such guidelines include the easy-to-read guidelines proposed by Swedish Agency for Acces- sible Media (MTM, 2021), whereas commonly re- ferred international initiatives include the IFLA guide- lines (Misako Nomura and Tronbacke, 2010), the In- clusion Europe guidelines (Inclusion Europe, 2020), and the Plain Language guidelines (PLAIN, 2011). Some of these initiatives are more focused on specific target audiences, and some adopt a broader approach, aiming to create more accessible texts for all. Although the various sets of guidelines differ, depend- ing on, for instance, intended target audience or type of text, one advise that is common for all easy lan- guage writing is to use simple words. This is, how- ever, expressed in slightly different ways. For example, the MTM guidelines advise to use ”simple and prefer- ably short words. If a more difficult word is needed, it should be explained in the text where it occurs”. The IFLA guidelines, on the other hand, say that one should ”avoid difficult words but use language that is adult and dignified. Unusual words should be ex- plained through context clues”. Thus, it is clear that easy language texts should use sim- ple words, and avoid difficult words. But how can we know, then, if a word is simple or not? A common way of determining this is to look at the relative fre- quency of a word in a corpus. This gives an indica- tion …

Content truncated to 500 words