Analyzing Reading Patterns with Webcams: An Eye-Tracking Study of the Albanian Language

Eye-tracking technologies have proven to be an important tool for understanding human behaviour. Unfortunately, high quality eye-tracking devices are not always affordable, especially in the case of studying low-resource languages. In this study we aim to explore the usage of low-cost opportunities, such as the usage of the WebGazer JavaScript library, utilizing normal webcams, to study an under resourced language such as Albanian. We present a study of the reading patterns in Albanian, with a primary aim to enlarge available datasets in our language and evaluate the accuracy and trustworthiness of the WebGazer library in capturing reading patterns. Our results demonstrate that the usage of webcams is an effective solution in the case where high-cost devices are not available. Various limitations in the webcam technology indicate the necessity to make further improvements in webcam eye-tracking models.