Previous research suggests that short-term memory (STM) processes are separable into at least two buffers: a lexical-semantic and a phonological buffer. While there are multiple tasks used to measure phonological STM, only one task is commonly employed to test semantic STM, the category probe task. The current study used a novel paradigm, the concurrent probe paradigm (Shivde & Anderson, 2011), to measure semantic and phonological maintenance in aphasia patients. In Experiment 1, which evaluated semantic maintenance, we replicated the findings of Shivde and Anderson (2011) with older adults and revealed dissociations in patient performance depending on the type of STM deficit. The concurrent probe paradigm provided converging evidence with the category probe task in measuring semantic STM deficits. In Experiments 2 and 3, we applied the task to phonological maintenance. We replicated the findings of Shivde and Anderson (2011) with older adults, but for patients the results were less clear.