The learning theory described here addresses the learning of lexical entries for certain predicational terms. Through characterizing input to the language learner in terms of conceptual structure-syntactic structure pairs, and evidence in terms of the technically defined construct detectable lexical error, the process outlined by the theory functions through error detection based on positive-only evidence. Learning is achieved when the learner has no occasion to reject newly formed lexical entries; this is the situation when lexical entries for all designated morphology (affixes and prepositions) are correctly acquired. The functioning of the theory is exemplified through a discussion of the learning schema, with particular attention to varying lexicalization patterns.