Web Readability and Computer-Assisted Language Learning

Proceedings of the 2006 Australasian Language Technology Workshop (ALTW2006), pages 99–106. Web Readability and Computer-Assisted Language Learning Alexandra L. Uitdenbogerd School of Computer Science and Information Technology RMIT University Melbourne, Australia Abstract Proficiency in a second language is of vi- tal importance for many people. Today’s access to corpora of text, including the Web, allows new techniques for improv- ing language skill. Our project’s aim is the development of techniques for presenting the user with suitable web text, to allow optimal language acquisition via reading. Some text found on the Web may be of a suitable level of difficulty but appropriate techniques need to be devised for locating it, as well as methods for rapid retrieval. Our experiments described here compare the range of difficulty of text found on the Web to that found in traditional hard-copy texts for English as a Second Language (ESL) learners, using standard readability measures. The results show that the ESL text readability range fall within the range for Web text. This suggests that an on-line text retrieval engine based on readability can be of use to language learners. How- ever, web pages pose their own difficulty, since those with scores representing high readability are often of limited use. There- fore readability measurement techniques need to be modified for the Web domain. 1 Introduction In an increasingly connected world, the need and desire for understanding other languages has also increased. Rote-learning and grammatical ap- proaches have been shown to be less effective than communicative methods for developing skills in using language (Higgins, 1983; Howatt, 1984; Kellerman, 1981), therefore students who need to be able to read in the language can benefit greatly from extensively reading material at their level of skill (Bell, 2001). This reading material comes from a variety of sources: language learning text- books, reading books with a specific level of vo- cabulary and grammar, native language texts, and on-line text. There is considerable past work on measuring the readability of text, however, most of it was originally intended for grading of reading material for English-speaking school children. The bulk of readability formulae determined from these stud- ies incorporate two main criteria for readability: grammatical difficulty — usually estimated by sentence length, and vocabulary difficulty, which is measured in a variety of ways (Klare, 1974). Publishers later decided to use the readability mea- sures as a guideline for the writing of texts, with mixed success. However, new reading texts cater- ing for foreign language learners of various lan- guages are still being published. Most of these use specific vocabulary sizes as the main crite- rion for reading level. Others are based on spe- cific language skills, such as the standard devel- oped by the European community known as the “Common European Framework of Reference for Languages” (COE, 2003). The goal of our research is to build an applica- tion that allows the user to improve their language skills through accessing appropriate reading mate- rial from the Web. This may incorporate person- alised retrieval …

Content truncated to 500 words