Towards a multiplex network model of word associations and similarity in the human mind

Early language acquisition is a cognitive process mediating the learning of words according to heterogenous linguistic knowledge, e.g. semantics and phonology. Even if cognitive networks are insightful for understanding how conceptual associations influence word acquisition, current approaches do not account for the interplay between structure and word features, i.e. exploitable node metadata. In this work, we aim to investigate word learning by merging relational structure and word features, like frequency, length and polysemy, in a multilayer representation enhanced with vector spaces.