The PNAV (Psycholinguistic, Neurocognition, Acquisitions and Variation) Team is interested in language functioning and language disorders in several speech and sign languages, in native and non-native speakers, by integrating behavioral, oculometric, and neurophysiological approaches. Beyond proposing empirical evidence on the mechanisms of language processing and production, we strive for the goal that our scientific results have a societal impact, both by developing assessment tools for clinicians, but also by assessing and supporting foreign language learning in the educational and social sector. For these diagnostic, assessment and learning support activities, the biomarkers tested in our EEG studies and validated more generally by the scientific community in different laboratories will be a valuable tool. Unlike behavioral markers, these biomarkers allow for a temporally fine-grained measure of ongoing linguistic processing in the brain. The Corpus-based Modeling and Linguistics team aims to develop and validate linguistic models in a bi-directional approach: on the one hand, by going from the facts attested in the corpora towards a model via the (semi-)automatic processing of generally annotated data; on the other hand, by focusing on the demonstrability of the hypotheses put forward and by testing the predictions resulting from the models. For this, three methods are used: (1) validation on corpus data: a qualitative or quantitative analysis of the data and in particular a fine-grained annotation of the corpus which allows both to test the models and to extract the relevant data for the elaboration of the models; (2) validation by simulation: the elaboration of a computer or mathematical model. This mode of validation is similar to the previous one when it comes to supervised learning on annotated data; (3) experimental validation, in collaboration with the PNAV team. This last one will be used in the AAC project.