Using a Posner-type RT procedure 8 undergraduates responded "same" or "different" to visually presented letter pairs. Onset of the 2nd letter, relative to the 1st was delayed 0, 1125, 1500, or 2000 msec. Pair members were printed in either upper- or lowercase type, and selected from a population of either auditorily confusable or nonconfusable letters. Results show that RT was faster when S compared physical forms of letters (e.g., BB) than names of letters (e.g., Bb), but that this difference decreased with delay. Further, when names were compared, RT was longer for auditorily confusable names, and this effect increased with delay. These observations support a model in which the likelihood of auditory encoding increases over time, whereas maintenance of a visual code decreases. (35 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved)