Apparently identical verbs can be represented differently: comparing L1–L2 inflection with contingency-based measure ΔP

We propose a method that models unidirectional, contingency-based association scale ΔP in order to analyse the different degrees of morpheme productivity in apparently identical L1–L2 inflected pairs. The method has the potential to uncover differences in how in L1–L2 inflected items are represented by L2 learners and native speakers. Such differences are at risk of remaining invisible if one considers only frequency, distribution and rank of predicates.