Bagou, O. , Gallot, S. & Frauenfelder, U.
Laboratoire de psycholinguistique expérimentale, FPSE,University of Geneva
Artificial word learning experiments generally have two phases: a learning and a testing phase. While the testing phase (classically forced-choice and/or lexical decision) is useful for measuring the result of learning, it provides no information about how learning actually takes place over time. This study develops a new method for studying the dynamics of lexical segmentation in novel word learning, using phoneme monitoring (PM) in which participants had to detect phoneme targets located either internally or finally in artificial words (AWs) that occurred frequently or rarely in an artificial language (AL). The first experiment compared the use of classical tasks with the PM during the testing phase and showed comparable effects of learning in both. We observed the expected interaction between phoneme position and word frequency, that is, item-final phonemes were detected faster in frequent than in rare AWs, while no effect of word frequency was found for internal phonemes. More interestingly, this result, taken to be a reflection of lexical activation, was also observed in the second experiment in which participants detected phonemes while listening to the AL. Assessing when and how this interaction emerges in the learning phase provides an excellent way of tracking word learning progress.