It has generally been accepted that L2 errors can give us information about the L2 acquisition process and its developmental stages. But dealing with interlanguage (IL) phenomena in terms of errors against target language norms can be regarded as problematic in several respects. Moreover, it does not seem adequate either to study rule‐governed IL utterances without explicitly formulating the rules in terms of the underlying psychological processes that characterize the learning. Without considering these processes, we are left merely interpreting our data, not being able to explain what causes them.In order to avoid these shortcomings, empirical research should be done into the psychological reality of the rules that underlie IL behavior. These rules should be studied as part of the mechanisms and strategies of the IL production process. The proper procedure should be to formulate hypotheses about underlying rules and, subsequently, to test the predictions that can be derived from these hypotheses.Hyltenstam's description (1977) of the developmental route of negative (NEG) placement in Swedish as an L2 will be used as an example of how empirical research on the acquisition of grammatical structures can be evaluated. Using Hyltenstam's data, I will propose an alternative that provides an adequate explanation of the route of acquisition of NEG placement in Swedish as a result of underlying principles which are located in the domain of a psychology of L2 learnin