Three experiments tested predictions of a neural network model of phonological short-term memory that assumes separate representations for order and item information, order being coded via a context-timing signal [Burgess, N., & Hitch, G. J. (1999). Memory for serial order: A network model of the phonological loop and its timing. "Psychological Review, 106", 551-581005D]. Predictions were generated for long-term sequence learning and tested using the Hebb Effect, the improvement in immediate serial recall when a list is repeated. Results confirmed predictions that the Hebb Effect would be (1) insensitive to phonemic similarity and articulatory suppression, variables that impair immediate recall without affecting the context-timing signal and (2) reduced if the context-timing signal is altered by varying the temporal grouping pattern of the repeated list. Results highlighted an interesting shortcoming of the model in that participants were able to learn more than one sequence simultaneously. However, this problem was addressed by extending the model to include multiple context representations and a sequence-recognition process.