The relevance of the study of Latin & Greek (L&G) in secondary education is being questioned more and more. Proponents have claimed that it greatly improves linguistic abilities and even cognition beyond the linguistic domain, e.g. analytical thinking. Numerous American scholars have set up field tests, but they focused on the influence of Latin on English vocabulary acquisition by native speakers. Moreover, these studies all lack a theoretical framework for the supposed effects of L&G, so that the premises underlying hypotheses remain implicit. As a result, the debate on the educational value of L&G is still predominantly based on intuitive and ideologically-driven arguments rather than on empirical research. Therefore we propose a longitudinal study on the effects of L&G on Flemish pupils’ language competence (near transfer) and cognition (far transfer). Latin & Greek differ from the other languages standardly offered in Flemish secondary education because they are no longer spoken, but also because they are more complex. As such, the common didactic approach pays much more attention to grammatical parsing. We will investigate whether this intensive training in grammatical analysis results in cognitive transfer, which will be dissociated from a priori pupil characteristics. In addition to the main study, the features of L&G learning suspected to cause transfer will be simulated in a short-term artificial grammar learning experiment.