Traditional approaches to aphasia therapy in the last century have varied depending upon the theoretical assumptions underlying the therapy. These approaches include: reactivation of weakened representations, reorganization of function, reteaching of lost information, and functional compensation. In more recent years, there has been a move toward framing issues of rehabilitation within models of cognitive neuropsychology (Riddoch & Humphreys, 1994). This approach has proved quite useful in helping to focus remediation attempts at a more precise functional localization of the deficit and has clearly made a significant contribution to aphasia therapy. What we need to do in the next century is to go one step further, in focusing not on what needs to be fixed, but rather on what is the best procedure for fixing it. That is, we need to examine the role of the specific learning paradigm that is employed in aphasia therapy, independent of the nature of the deficit or the theoretical goal of the treatment. It may be useful to look to the current literature on learning and memory for help with this endeavor. The past two decades of this century have seen an explosion in the attention paid to different kinds of memory and learning. A new dichotomy of memory has become popular, based largely on research with amnesic patients. These patients perform poorly on memory tasks involving explicit (or, declarative) recognition or recall, i.e., the conscious recall or recognition of facts and events, including knowledge of explicit rules (‘‘i before e, except after c.’’) or procedures. However, it has been noted that new learning can be demonstrated in these amnesic patients on tasks of an implicit (or, nondeclarative) nature. In these tasks, the learning of new information is demonstrated by changes in performance. Such changes can take the form of in