Listeners quickly adapt to novel accents. There are three main hypotheses for how they do so. Some suggest that listeners expand their phonetic categories, allowing more variability in how a sound is pronounced. Others argue that listeners shift their categories instead, only accepting deviations consistent with the accent. A third hypothesis is that listeners shift and expand their categories. Most work has supported the category expansion hypotheses, with the key exception of Maye et al. (2008) who argued for a shifting strategy. Here, we apply the ideal adaptor model from Kleinschmidt & Jaeger (2015) to reexamine what conclusions can be drawn from their data. We compare adaptation models in which categories are shifted, expanded, or both shifted and expanded. We show that models involving expansion can explain the data as well as, if not better than, the shift model, in contrast to what has been previously concluded from these data.