Talking about motion events in L2 is done in different ways by different speakers on different occasions. This is due to multiple factors, typological, psycholinguistic and sociolinguistic, which interact and play a role in L2 acquisition and use. These factors can sometimes lead the same types of L2 speakers to produce very different outputs and sometimes very different L2 speakers to produce the same or similar outputs. In order to capture this diversity of outcomes the CASP (Complex Adaptive System Principles) for Bilingualism model was proposed and we illustrate how this model helps us formulate predictions about motion event verbalizations, set up experiments and account for results in a holistic manner, taking into consideration the relevant multiple factors. Furthermore, a lot of effort in the field has gone into contrasting monolingual and bilingual populations while more knowledge is needed about how different bilingual populations compare. These different bilinguals, including L2 learners, need to be tested under different conditions in which they use their languages (e.g., when only one or both is actively used with vs. without the possibility to code-switch) in order to understand the variability of L2 verbal behaviours and the underlying factors at play under different circumstances of acquisition and use. This perspective paper provides both theoretical and empirical indications how this can be done, with the key message that future research into L2 acquisition (and bilingualism in general) must be based on a multi-factor approach.