Effects of vowel length and “right context” on the alignment of Dutch nuclear accents

We measured the alignment of f0 landmarks with segmental landmarks in nuclear “pointed hat” accents in controlled speech materials in Dutch. We varied the phonological length of the stressed vowel and the “right context” (syllable membership of following consonant, presence/absence of stress clash). The nuclear accented word was always followed by an unaccented content word. Based on previous work we expected that the alignment would be substantially affected by vowel length, stress clash and syllable membership, but the only important effect was that of vowel length. We believe this can be explained by the fact that most previous studies have dealt with prenuclear accents and/or with nuclear accents in utterance-final position, whereas we are dealing with nuclear accents that are not in utterance-final position. We also explored the effects of using different quantitative definitions of our dependent and independent variables, and of using Multiple Regression rather than ANOVA, and conclude that our findings are robust regardless of the variables or analysis technique used. An important methodological conclusion from our comparative analyses is that tonal alignment is best expressed relative to a nearby segmental landmark. Proportional measures may also be useful, but need further investigation.