Peak Scaling in Greek and the Role of Declination

The controversial role of declination in the scaling of accentual peaks was tested using Greek materials, consisting of sentences of increasing length (from two to

five accented words) in which successive accents were separated by either two or four unstressed syllables. It was hypothesized that if peak scaling is locally controlled (i.e. independent of declination), the peaks in the long sentences would not be scaled lower than peaks found in the same position in the short sentences. The results showed that peaks were scaled lower in the long sentences, but the effect was very small. On the other hand, multiple regression showed that the best predictor of peak scaling was the scaling of the preceding peak, while other factors played a very small role. These results largely agree with results from Spanish and English and support the view that declination does not play a major part in peak scaling.