References to developmental stages and continua seem to be part and parcel of investigations into the acquisition of new grammars. Nonetheless, there seems to be an equivocation in the literature about which is actually the most helpful way of explaining how learner grammars evolve through time. Some see development essentially as gradual growth and, in this perspective, talking about a sequence of discrete stages looks like an artificial way of organizing learner data imposed by researchers simply as a matter of convenience. Indeed, the evidence does lend support to the idea of a learner not only developing new solutions to how the L2 works very gradually, but also entertaining alternative solutions at the same time. At the same time, from a linguistic perspective, it often makes sense to see development as a stepwise movement from one rule system to another: the acquisition literature is consequently full of references to stages and evidence to back up this type of analysis. In this paper, we will attempt to resolve the apparent conflict by adopting a language processing perspective called MOGUL that allows for both stages and continua and we shall illustrate our proposals using empirical findings from the second language acquisition literature. © Oxford University Press 2005.