We were able to show that a cognitively plausible model based on the principle of "discriminative learning" can successfully distinguish between singular and plural forms. This model does not need to know any morphemes (e.g. endings or umlauts). Rather, it successfully learns to distinguish between singular and plural forms in German solely on the basis of sound combinations (biphones). The model was also able to reproduce the performance pattern of a patient with a speech disorder when distinguishing singular and plural forms.
