Abstract A study was conducted in which aphasic patients, age-matched normals, and normal young adults performed five types of matching judgments for object pictures. These required matching for physical identity, basic object identity, and membership in the same superordinate category. Spoken name-to-picture matching was tested for the last two conditions. An analogous set of conditions was presented for letters. Latency patterns across the conditions showed general slowing for the aphasic patients, but with a differential decrement in the conditions that involved auditory (spoken name) input for the matching task. Results showed that variations in semantic judgment capability among the aphasics did not predict the patients' object naming ability.