The grammar of inclusion. The evaluation of soft skills for European Mobility.

The growing emphasis on lifelong learning and competence-based education has highlighted a fundamental limitation of current assessment practices: while European frameworks such as LifeComp, DigComp, EntreComp and GreenComp provide valuable conceptual models, they do not offer sufficiently operational definitions to support objective, transparent and comparable assessment of transversal competences across different linguistic and cultural contexts. This dissertation addresses this gap by proposing a multilingual semantic framework for the operationalization and assessment of transversal competences, with particular attention to inclusion. The study combines educational theory, comparative linguistics and Natural Language Processing (NLP) to reconstruct competence definitions starting from their semantic components rather than from traditional taxonomies alone. The proposed methodology integrates lexical analysis, multilingual semantic embeddings, clustering algorithms and Bloom's revised taxonomy to identify the cognitive, affective, behavioural and metacognitive dimensions underlying competence definitions. Official European documents constitute the primary corpus, while semantic relationships are analysed across several European languages through vector representations and similarity measures. The resulting semantic structures are subsequently translated into measurable assessment descriptors and analytic rubrics capable of supporting educational evaluation. Empathy is presented as a case study demonstrating the entire methodological pipeline, from multilingual semantic reconstruction to the development of an assessment rubric based on observable indicators and progressive levels of mastery. This process illustrates how linguistic variability can be transformed into a pedagogically meaningful evaluation model without reducing competence to simplistic behavioural checklists. Beyond the specific case of empathy, the research proposes a generalisable methodological framework applicable to a wide range of transversal competences. By integrating computational linguistics with pedagogical assessment, the dissertation contributes to overcoming the ambiguity that often characterises competence-based education and provides a replicable model for constructing transparent, measurable and culturally aware assessment instruments. The proposed framework aims to support educational institutions, researchers and policy makers in developing more reliable competence assessment practices, fostering greater consistency between European competence frameworks, educational design and evaluation processes.