Authors:Alexandra Markó, Tekla Gráczi and Judit Bóna
Hungarian represents a particularly fruitful ground for exploring voicing assimilation. Although this topic has been extensively analysed, a contradiction can be observed between most phonological descriptions and acoustic-phonetics-based studies of voicing assimilation. Theoretical works suggest that this process in Hungarian speech is a purely regressive, obligatory and categorical phenomenon, but in practice divergent realisations can be observed. In the present paper three case studies of voicing assimilation are performed. CCC clusters, CC clusters interrupted by pause and partially voiced realisations were analysed. The results showed that in the first two cases the speech planning process and the degree of self-monitoring were the most influential factors, while the various concomitances of voicing and devoicing arising due to aerodynamic and articulatory reasons resulted in partially voiced realisations. The variability of the data confirms the hypothesis that Hungarian voicing assimilation is a gradient and sometimes only partly regressive process. Even if it operates mainly obligatorily, several factors can override it.