Interactions of segmental and prosodic parameters

The topic of this talk is the mutual dependence of segmental properties and phonetic context. A context-embedded speech sound has its own segmental properties and is at the same time a constituting element of global prosody. Both in production and perception, contextual parameters affect individual segments and vice versa. Relevant signal parameters are temporal organization, fundamental frequency, intensity and spectral composition. Some examples will be given demonstrating the relevance of these factors to perceived segment identity as well as prominence. The main focus of my presentation will be on the effect of fundamental frequency contour on perceived segment duration in Norwegian. During the last fifty years or so, a number of studies on this issue have been performed for several languages. In spite of these efforts, there are some remaining questions yet. It has been speculated that effects might be language-dependent. Due to the tonal system of Norwegian, effects may be different from those found for other Germanic languages like English or German. One research question was if there are any indications that the Norwegian word accents 1 and 2 could impose certain constraints on the interplay of fundamental frequency and perceived vowel duration. So far, results from acoustic measurements suggest that this is not the case. Most investigations of the effect of f0 on perceived duration use real monosyllables or monosyllabic logatomes. The experimental design of the study presented here involved identification of a Norwegian disyllabic word with either a short or long accented vowel (/haker/; ‘[he] hoes’ vs. /haːker/; ‘chins’). In addition to manipulation of vowel duration, f0 contours in both the first and the second syllable were manipulated. Word identification rather than vowel-duration estimation was chosen as a more natural task. Findings will be presented and discussed.