Neutralisation: phonetic or phonological? An acoustic and perceptual study of vowel harmony in two dialects of Hungarian

The aim of this project is twofold. First, to gather reliable articulatory, acoustic and perceptual data from Standard Hungarian and Slovakian Hungarian. Second, I plan to address two main theoretical questions. Does the lexicon contain only abstract phon ological representations, only detailed phonetic representations, or both? What is the nature of the mapping between phonological and phonetic representations: one-to-one, one-to-many, or many-to-many? More specifically, to what extent does the abstract p honological representation of a word determine its phonetic realisation, and vice versa? The empirical area I plan to investigate is anti-harmonic stems in Hungarian: stems containing the vowels i, i: or E, e: are divided into two classes: one class takes suffixes with front vowels, the other class takes suffixes with back vowels, with seemingly no phonetic differences between the two classes of stems. Benus et al. (2004); Benus & Gafos (2007) claimed that the vowels in stems triggering back harmony are p ronounced with a more back articulation than the vowels in the stems that trigger front harmony. However, their experimental setup raises doubts as to the validity of their results. First, they only measured 4 speakers of Hungarian, second, the only signi ficant differences between the two groups of stems showed when the stems were recorded in suffixed forms, where the phonetic difference could be caused by co-articulation. I plan to replicate and extend these experiments to determine whether the claims ho ld. Furthermore, since the two dialects show a phonetically different set of transparent vowels, the aim is to investigate whether these result in differences in phonological behaviour.