|Abstract:||Nine-month-olds are already sensitive to the distinction between licit and illicit forms drawn by the target adult phonotactics. Yet, they are still blind to alternations, as morphology still lags behind. How can they manage to acquire phonotactics from licit forms only? According to the error-driven phonotactic learning scheme, the model is trained on a stream of licit forms, starts from the most restrictive initial phonotactics, and slightly relaxes it whenever it fails on the current piece of data. Being trained on a stream of licit forms, it is easy to guarantee that the phonotactics learned by the model is consistent, namely that it correctly rules in every form which is indeed licit according to the target phonotactics. Yet, the learned phonotactics could be unrestrictive: it could incorrectly rule in also forms which are instead illicit according to the target phonotactics. In this paper, I tackle the problem of restrictiveness of error-driven learning within the framework of Optimality Theory (OT). Knowledge of phonotactics is modeled in OT through a ranking of markedness and faithfulness constraints. In the vast majority of cases, the relative ranking of the faithfulness constraints turns out to be irrelevant to describe a certain phonotactic patter (although it is crucial for the phonology). In this paper, I develop guarantees for restrictiveness when the relative ranking of the faithfulness constraints does not matter.