Determining vocalic qualities from linguistic atlas data

Abstract

In the past twenty years, the acoustic analysis of vowel qualities has become a standard approach in dialect research in general, and in dialect geography in particular. Usually, normalized vowel formant values are mapped and statistically analysed. While little has been done so far to describe the acoustic characteristics of non-standard varieties of Hungarian, there are more than one million data instances from computerised dialect atlases, mainly The Atlas of Hungarian Dialects (AHD), where articulatory details are reflected by the narrow phonetic transcription. Vocalic qualities are denoted in AHD with basic symbols and diacritics as needed, according to the standard Hungarian transcription system. The present study attempts to define, in a quantitative framework, the typical qualities of some vowels, based on the subjectively transcribed atlas data, and to assess the spatial evolution of these vocalic qualities. To this end, estimated F1 an F2 values are attributed to a number of combinations of vocalic symbols and diacritics. At each location of AHD, typical F1 and F2 values are calculated in function of the frequency of different vowel variants, as transcribed in the atlas, for the three most open vowels /a?/, /?/ and /?/. These typical F1 and F2 values are mapped so as to show the spatial differences in the phonetic quality of each vowel under investigation. For a better understanding of spatial patterns in vowel articulation, a large-scale dialect research focusing on acoustic measurements of vowels would be indispensable.