Transcription and Diffusion. The "True" Picture of ei- Diphthongization via Diffusion against the Background of Transcriber Effects

In the IPA-transcripts of the Goeman-Taeldeman-Van Reenen (GTR) database we study the nucleus traject(s) in the dialect vowel spaces of one of the Dutch diphthongs. This diphthong developed from the highest front vowel /i/, Westgermanic * long i & merged with the product of Westgermanic *ai in the Standard language. In certain dialects, /i/ remained, while in others /i/ has been diphthongized. Although the leaders of the GTR-project aimed at comparable & consistent IPA transcriptions, the IPA standards in Belgium & the Netherlands are slightly different in three respects: (1) transcripts in Belgium are broader, while in the Netherlands they show more phonetic details; (2) /a/ is the lowest front vowel in the Flemish tradition, while it is /ae/ in the SIL-standard used by Van Reenen in the Netherlands; (3) generally, /e/, /o/ & /o/ may be used for higher realizations in the Flemish transcripts than in those of the Netherlands, where /e/ & /(epsilon)/ are at equidistant points within the /i/-/a/ continuum. As an outcome of normal human perception, the height of the diphthong's nucleus is perceived as either higher, or lower than measured; according to kind of diphthong respectively either centralizing, or upgliding (Van Heuven, van Bezooijen & Edelman 2005). Although in both countries all transcribers had interiorized the same abstract extreme anchorpoints /i/, /a/ or /ae/, /(script a)/ & /u/, the fact that they come from different regions of the language area, being exposed to different vowel systems & to specific regional vowel realizations, will have an impact on their transcription decisions. In order to calibrate their transcriptions of the diphthong nucleus we used mean normalized F1 & F2 measures for the corresponding monophthongs from their region of provenance, because we assume them to make up their mental image of the vowel space. These measures (Adank 2003, & Verhoeven & Van Bael 2002), normalized for gender differences, proved to retain the regional differences. The Neary normalized F1/F2 values are used to study the diphthong nucleus' trajects & the geographical traject of diphthongization in different distributional environments (before e.g. labial fricative). Finally, we inspect anew the circle diffusion model proposed by Goeman, Van Reenen & Wattel (1993) because we have now more data, specifically the data from Belgium that were not yet available at that time. Van Reenen taught the SILIPA standard to the Dutch transcribers, Taeldeman the 'B