Self-learning is a difficult activity for those who prefer to study with friends. To make self-learning more accessible, Hadas mapped the behaviors observed in a Havruta – a traditional Jewish study approach where two students explore a given text together. One of the most basic acts observed in a Havruta is that of reading aloud. Hadas discovered that reading aloud supports active learning: while reading aloud, the student articulates, emphasizes and accentuates the words according to her understanding. She punctuates each sentence, processes and interprets the text, facilitating improved comprehension. In learning together, we feel comfortable reading aloud, because someone else is listening and responding to us. In a self-learning setting, who is listening to us, who is responding? In A Philosophy of Havruta, Holzer and Kent refer to text as if it were a Havruta member. It enables us to consider the self-learner as someone who is not alone: she is actually accompanied by the text. What if the text itself were to listen and react to the learner? What if the action of reading aloud could physically impact the written text? Trasmundi and Cowley, researchers in the field of environmental psychology, refer to the text as a written artifact, which we learn how to operate via social learning processes. How can we expand the properties of the written artifact and enable it to listen back?
The text persona in this project was characterized as tolerant and emphatic. The text tracks and follows the reader using a speech-to-text functionality. The text “agrees” to continue tracking, even when accent and articulation are inaccurate. The text “breathes” and gives the feeling it is “listening”. Volume and pitch detection impact, respectively, on the boldness and italicization of the font. This adapts incrementally, thanks to the use of variable fonts that permit control of their parameters on a continuous scale. A uni-width font was chosen to preserve readability. A "listen to the verbal pronunciation" feature was added, echoing the teacher-student Havruta mode.
Improved speech-to-text functionality that differentiates syllables and statistically detects words during their pronunciation will enable typographic changes within one word to reflect different pronunciation modes. It may also solve the synchronization problem in which we feel a delay in the text’s word-detection pace, compared to a human listener. In the future it will be possible to enrich the text’s response to additional reading-aloud qualities using an emotion detection mechanism that will rely not only on the content of the text and the vocabulary but also on vocal features.