Crossed random effect models: Brief considerations on an application in psycholinguistics