Bahlmann et al. (2006) reported an experiment on event-related brain potentials of sequences of syllables obeying two rules, one defined by AnBn and the other by (AB)n, where the As and Bs are different classes of syllables. They interpreted their findings on the assumption that AnBn are parsed according a center-embedded phrase-structure grammar. In fact, such sequences are much more likely to be parsed in terms of the repetition of element types, without reference to phrase structure. This raises a general issue about attempting to study syntactic processing independently of semantics.