Two experiments examine the effect of regular feedback on the appearance of the word-length and the irrelevant sound effects in immediate memory. A reliable effect of learning was observed but both effects persisted across multiple learning trials, contrary to suggestions that they could be diminished either by feedback-inspired strategy change or by dual-task learning. Learning improved performance most noticeably at the end of the serial position curve but there was no sign that the irrelevant speech eliminated the word-length effect at any stage of learning, consistent with Tremblay et al. (2000) but inconsistent with Neath et al. (1998). Implications for immediate memory effects, and for models of intra-individual variation in immediate memory, are considered.