The dissertation has two main aims. 1. To examine nouns of foreign origin with sibilants at the end and the way they behave when Hungarian native speakers inflect them with the General Possession Marker (GPM) morpheme. Unlike in the case of other sibilant-final Hungarian nouns (identified by speakers as Hungarian nouns), some Hungarian native speakers (one of whose hobbies is to play video games in English) add a j-initial allomorph of the GPM morpheme to these nouns, which is a highly unexpected solution. 2. To examine nouns of foreign origin with strongly marked consonant clusters at the end and their behavior when Hungarian native speakers inflect them with the accusative ending. A significant part of Hungarian native speakers add the accusative ending to these nouns without a linking vowel (cf. native taps > tapsot [tɒpʃ] > [tɒpʃot] ‘applause, nom. > acc.’ vs. foreign Hanks > Hankst, nom. > acc.). My hypothesis is that Hungarian native speakers use an alienation method (resulting in an increase in transparency) when they produce these unexpected, marked forms. The analysis presented in the dissertation uses the framework of classical Optimality Theory. Within the OT framework, I demonstrated the process of this alienation method. With OT, it is simple to handle interpersonal variations. Following the demonstration within the classical OT framework, I repeat the demonstration using the Maxent Grammar Tool of the Maximum Entropy Model to determine the maxent value of the rival candidates, to better account for the persumably intrapersonal variations.