The aim of this experiment was to use behavioral and electrophysiological methods to compare university students with dyslexia and matched skilled readers in a novel word learning experiment that included phonological categorization tasks, a word learning phase and a test phase with matching and semantic tasks. Specifically, we aimed at disentangling two hypotheses. If phonological processing drives novel word learning and if phonological processing is impaired in students with dyslexia, they should perform lower than skilled readers not only in the phonological categorization tasks but also in the matching and semantic tasks. By contrast, if students with dyslexia use semantic knowledge to compensate for their phonological deficits, should be able to reach the same level of performance and show similar enhancements of the N200 and N400 components than skilled readers in the matching and semantic tasks. Results at both behavioral and electrophysiological levels showed that the phonological deficits evidenced in the phonological tasks did not impede students with dyslexia to learn the meaning of novel words, possibly because they mobilized more frontal resources than skilled readers. These results are discussed within a general framework of semantic compensation in adults with dyslexia.