Age Recommendation for Texts

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 1431–1439 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 1431 Age Recommendation for Texts Alexis Blandin1, Gw´enol´e Lecorv´e1, Delphine Battistelli2, Aline ´Etienne1,2 1Univ Rennes, CNRS, IRISA 2Universit´e Paris Nanterre, CNRS, MoDyCo Lannion, France Nanterre, France {alexis.blandin, gwenole.lecorve}@irisa.fr, {delphine.battistelli, aline.etienne}@parisnanterre.fr Abstract The understanding of a text by a reader or listener is conditioned by the adequacy of the text’s characteristics with the person’s capacities and knowledge. This adequacy is critical in the case of a child since her/his cognitive and linguistic skills are still under development. Hence, in this paper, we present and study an original natural language processing (NLP) task which consists in predicting the age from which a text can be understood by someone. To do so, this paper first exhibits features derived from the psycholinguistic domain, as well as some coming from related NLP tasks. Then, we propose a set of neural network models and compare them on a dataset of French texts dedicated to young or adult audiences. To circumvent the lack of data, we study the idea to predict ages at the sentence level. The experiments first show that the sentence-based age recommendations can be efficiently merged to predict text-based recommendations. Then, we also demonstrate that the age predictions returned by our best model are better than those provided by psycholinguists. Finally, the paper investigates the impact of the various features used in these results. Keywords: natural language processing, age recommendation, children, psycholinguistics, neural networks 1. Introduction The way in which an individual understands a text is com- plex. It depends both on the characteristics of the text and the abilities of the reader or listener. For instance, impor- tant abilities are the capacities to remember information, to position an event in a story line, to analyze the structure of a sentence, to understand a word or simply to read a text. For a given person, these various aspects are function of her/his degree of neuro-cognitive development, linguistic mastering (including the ability to read) and culture. Dur- ing childhood, all these aspects are highly and constantly evolving. Hence, when presenting a text to a child, it is crucial to assess if this text is adequate enough to be under- stood. This paper aims at automatically predicting age recommen- dations for texts with the objective to maximize its under- standing by a child. Potential applications are numerous. First, this may be helpful when one wants to provide texts to children, mainly in schools where teachers could be as- sisted when preparing their class, and in search engines where returned pages could be conditioned by the age of the user. Secondly, age recommendation could also be a precious tool for authors (be they professional or not) at present by analysing written drafts, or, in a further perspec- tive, by proposing reformulations. Finally, the automatic study of features could give useful feedback to psycholin- guists investigating children’s text comprehension abilities, e.g. …

Content truncated to 500 words