The Lexical Quality Hypothesis (Perfetti & Hart, 2002) suggests that the difficulties exhibited by poor readers cascade from deficient (impoverished, fuzzy) representations of phonological, semantic, and orthographic dimensions in lexical memory. If so, readers, even as adults, should vary in their ability to acquire new lexical representations. In our study, we examine the role of cross-modal (visual to phonological) associations in lexical learning. By pairing an artificial lexicon with novel objects, we aim to see whether learning implicit associations between new words and visual features of novel objects can be predicted by participants' performance in a number of visual and language-related assessments. We report intriguing preliminary results suggesting new relationships between recognition memory and ability for language learning and processing.