Production and perception of voicing contrast in assimilation contexts in Hungarian

This paper explores to what extent lexical factors, such as minimal pairhood and wordedness, affect the realisation of laryngeal features of the word-final fricatives /s/ and /z/ in Hungarian in potentially neutralising contexts, and whether the observed acoustic differences are perceptually salient enough to distinguish underlying voicing in non-minimal pairs and in minimal pairs in semantically ambiguous contexts. We show that in devoicing contexts the contrast between /s/ and /z/ in minimal pairs is more likely to be upheld than in non-minimal pairs in production, and this difference seems to map onto perceptual contrast, also that complete neutralisation can be prevented in devoicing contexts by durational cues. In the voicing environment, the acoustic difference between the fricatives is less likely to map onto a contrast in perception, indicating neutralisation. In the devoicing context, little voicing is enough to categorise the fricative as voiced: listeners compensate for phonological changes that correspond to existing rules in their language rather than for those that are only coarticulatory in nature.