Theoretical accounts of pure word meaning deafness are rare; accounts of its rehabilitation are virtually non-existent. We contrast the effects of two therapies in a patient with pure word meaning deafness. One therapy required only implicit auditory access from the patient (silent reading comprehension exercises). The second required explicit auditory access (auditory comprehension exercises), and thus appeared to be more suited to the exact locus of the patient's impairment. Improvement was observed after both types of therapy. However, improvement on implicit access therapy was influenced by the use of a compensatory strategy developed by the patient. In contrast, improvement on explicit access therapy was more durable, and appeared to be due to a direct effect on the audition–semantics link, rather than to compensation. We conclude that pure word meaning deafness is amenable to treatment, and that cognitive models can be useful in designing such therapy studies.