Readability of Twitter Tweets for Second Language Learners

Readability of Twitter Tweets for Second Language Learners Patrick JACOB and Alexandra L. UITDENBOGERD RMIT University - School of Science 124 La Trobe St Melbourne VIC 3000 patrick.jacob@rocketmail.com,sandra.uitdenbogerd@rmit.edu.au Abstract Optimal language acquisition via reading re- quires the learners to read slightly above their current language skill level. Identifying mate- rial at the right level is the essential role of au- tomatic readability measurement. Short mes- sage platforms such as Twitter offer the op- portunity for language practice while reading about current topics and engaging in conversa- tion in small doses, and can be filtered accord- ing to linguistic criteria to suit the learner. In this research, we explore how readable tweets are for English language learners and which factors contribute to their readability. With participants from six language groups, we col- lected 14,659 data points, each representing a tweet from a pool of 4100 tweets, and a judgement of perceived readability. Tradi- tional readability measures and features failed on the data-set, but demographic data showed that judgements were largely genuine and re- flected reported language skill, which is con- sistent with other recent studies. We report on the properties of the data set and implications for future research. 1 Credits We thank Klerke et al. (2016) for the Twitter cor- pus from their unpublished research. 2 Introduction Since the first half of the twentieth century re- searchers have analysed texts to determine their readability, that is, how easy the text is to read and comprehend, often expressed as levels of lin- guistic education, knowledge, age or experience. The findings around readability have been applied to education for selecting appropriate reading ma- terial for students, in communication with govern- mental bodies to reach a higher number of citizens (Temnikova et al., 2015) and in marketing/public relations of companies to increase the reach of their materials (Risius and Pape, 2016). While there have been many studies on read- ability of regular text for students and foreign lan- guage learners, the same is not true for the mi- croblog text genre. Twitter is a popular platform for reading current topics and engaging in so- cial interaction, providing a cross-cultural, cross- interest, and cross-language platform for read- ing social media posts of up to 280 characters in length, and a filtered feed has potential as a source of regular reading material for learners. While the Flesch readability of English language tweets has been analysed to discover demographic trends and to compare them to other modern text genres (Davenport and DeLine, 2014), and judgements of tweet clarity for emergency communication has been researched (Temnikova et al., 2015), to our knowledge tweets have not been studied in rela- tion to English as an Additional Language (EAL, a term that recognises that it may be a third lan- guage, for example). Our aim was to extend readability research to tweets for EAL learners. Tweets are different from ordinary text due to their short length, hashtags, mentions (user identifiers preceded by an @ sym- bol), links, and other …

Content truncated to 500 words