When does subtitle viewing become reading? Identifying the onset of reading in dynamic multimodal environments.

Most eye-movement research on reading begins once participants are already engaged with text. In multimodal environments such as subtitled video, however, reading is optional: viewers must shift attention from the image to text, and entering the subtitle area does not necessarily mean that reading has begun. This talk presents an eye-tracking study of 95 L1 and L2 English viewers examining the transition from subtitle onset to progressive reading. We distinguish attentional orienting, indexed by time to first fixation, from a data-informed estimate of reading onset that accounts for corrective eye movements following subtitle entry. The findings show that some subtitles are never entered, that entry precedes estimated reading onset, and that early adjustment patterns vary systematically with reader and subtitle characteristics. I use these findings to raise a broader methodological and theoretical question: should reading onset be treated as a discrete event, or as a gradual transition into a latent reading state? The talk concludes by considering what this might mean for understanding when exposure to text in dynamic digital environments becomes meaningful reading experience.