Technology-enhanced environments need to be adaptive to learners’ needs and skills, even in real-time, in order to provide better support to learning. The prerequisite to adaptiveness, however, is the automated learner/learning assessment. To this end, a set of learners’ physiological data are captured, such as eye movements and EEG data, while the learners are interacting with a learning platform or with conventional learning material, e.g. a textbook. The analysis of these data conveys important information regarding the learner’s state, attention and cognitive load, thus providing the basis for assessing the learners’ level or for evaluating the learning platform.
Special emphasis is put in investigating the physiological and cognitive correlates of reading comprehension. By recording the reader’s reactions to a specific portion of text, the patterns correlated to the text difficulty are unveiled. Machine learning techniques are then employed to correlate these patterns to composite linguistic features of the respective text and, in effect, infer the parameters of the difficulty level of a text.
In parallel, the stream of learning data captured open up the way to learning analytics and can offer insight into how students learn, better understanding of the learning process, and the possibility to exploit these for improving the teaching practice.