Morse code learning for assistive technology access: engagement, learning time, and cue associations in a large...

To characterize Morse code learning speed and engagement in a public self-directed trainer, examining how cue settings affect alphabet recognition for assistive technology applications.

We analyzed 699,562 telemetry records from 34,641 anonymous users over five years. Learning time was estimated from 35,370 sessions; final accuracy for 9,476 engaged users was modeled using binomial regression.

While 51.6% of users showed no progress (casual traffic), a subset of users who progressed substantially reached full-alphabet recognition in a median of 55.6 minutes (0.47 characters/minute); this estimate is drawn only from sessions that showed substantial progress and does not represent the broader user population. Visual cueing significantly improved accuracy (90.3% vs 82.2%; Cohen's d = 0.79) and progress (51.6% vs 39.3%; p < 0.001). Regression showed visual cues were most effective initially (OR 1.29), but their advantage declined across the teaching sequence (OR 0.98 per step). Conversely, sound cueing became more beneficial as learning progressed (OR 1.03 per step), with a crossover at position 13.

Among users who progress far enough, the Morse alphabet can be learned in approximately one hour, though this reflects only the minority of sessions that accumulated substantial progress and should not be generalized to all new users. Findings suggest visual cues aid initial acquisition, while sound cues become more valuable later. This supports Morse code as a learnable access method and highlights the importance of adaptive, configurable cueing.