The role of associative learning has long been downplayed in research on language acquisition. However, the predictions of associative models are often misunderstood in the literature, and both the power and diversity of such models underestimated. For example, associative learning is often assumed to be inherently slow, or to necessarily predict that any two stimuli that occur together would be associated together. Recent years have seen a resurgence of associative accounts of many aspects of language acquisition, including phonetic categorization, phonotactics, word learning, and paradigmatic morphology. In this lecture, I will discuss the basic ideas of associative learning models, their predictions for language acquisition, and their limitations, both apparent and real. I would cover the differences between Hebbian, error-driven and reinforcement learning models, configural/analogical and elemental models, and the role of cue competition and selective attention in language acquisition. I’d draw on my own work, as well as work by the Tuebingen Quantitative Linguistics group (Harald Baayen, Michael Ramscar, Jessie Nixon, Fabian Tomaschek), and work by Linda Smith, John Kruschke, Chen Yu & Rob Goldstone at Indiana.