TL;DR
This paper investigates the timing of the "low plateau" (L2) in Dutch falling-rising pitch contours, a key element of intonation. Through two production experiments, the researchers found that L2's timing is remarkably consistent with the end of the utterance or the end of the final rise, rather than being anchored to the last stressed syllable or the last syllable itself. This challenges previous assumptions and has implications for understanding how speakers produce and listeners perceive intonation.
What Problem Does This Paper Address?
Intonation, the melody of speech, plays a crucial role in conveying meaning and structure. Within intonation, specific turning points in pitch are often thought to be "anchored" to particular points in the speech signal, like stressed syllables. The falling-rising pitch contour, common in Dutch questions and statements, features a falling movement followed by a rise. The timing of the second low point (L2) in this contour has been a subject of debate. Some studies suggested it aligns with the last stressed syllable, while others found it more variable. This paper aims to clarify the precise anchor point for L2 in Dutch falling-rising contours, specifically investigating whether it's tied to stress patterns, syllable boundaries, or the utterance's end. They also explore whether the level of stress (primary vs. secondary) influences this alignment.
Methods
The researchers conducted two production experiments with native Dutch speakers.
- Experiment 1: Speakers produced sentences with falling-rising contours where the distance between the last stressed syllable and the end of the utterance was systematically varied. This was achieved by using target words with different post-nuclear stress patterns (s, sw, sww). The goal was to see if L2's position shifted with these variations.
- Experiment 2: This experiment focused on the role of stress level. Speakers produced sentences with minimal pairs of words that differed in whether primary or secondary stress fell on a particular syllable (e.g., "Hogelonen" vs. "hoge lonen"). This allowed them to test if L2 alignment differed when primary stress was in a different position relative to secondary stress.
Acoustic analysis involved precise measurement of pitch (F0) turning points (L2 and H2) and syllable boundaries using specialized software (Praat) and statistical analysis.
Datasets
The paper does not introduce new datasets. Instead, it uses controlled experimental stimuli created by the researchers for their production experiments. These stimuli consist of carrier sentences and target words designed to manipulate specific prosodic features.
Key Insights and Ideas
- Utterance End as the Primary Anchor: The most striking finding is that L2's timing is remarkably stable relative to the end of the utterance and the end of the final rise. This suggests a global, rather than local, anchoring mechanism for this intonational feature.
- Stress is Not the Key Anchor: Contrary to some previous assumptions, the last stressed syllable (whether primary or secondary) does not consistently dictate the timing of L2. The researchers found that varying the distance to the last stress did not affect L2's alignment with the utterance end.
- Stress Level Matters for Duration, Not Timing: While the level of stress (primary vs. secondary) did influence syllable duration (primary stressed syllables were longer), it did not affect the timing of L2. This suggests that the acoustic realization of stress differs from its role in anchoring intonational events.
- Challenging Previous Findings: The results appear to contradict some earlier studies that suggested a stronger link between L2 and stressed syllables. The authors propose that differences in experimental design, particularly the length of the post-nuclear stretch, might explain these discrepancies.
Key Results
- Experiment 1: Mean distances from L2 to the end of the utterance or the end of the final rise (H2) varied by a maximum of only 4 ms across different stress patterns (s, sw, sww). In contrast, distances to the last stressed syllable or the last syllable varied significantly.
- Experiment 2: The timing of L2 was not determined by the position of the primary stress in the tested stress patterns. Mean distances from L2 to the vowel onset of primary stress differed by as much as 252 ms between the two conditions ('swsw vs. sw'sw).
- Syllable Duration: Primary stressed syllables were found to be longer than secondary stressed syllables in the post-nuclear position.
Limitations and Future Work
- Experimental Control: While efforts were made to control for factors like rhythmical patterns, word boundaries were not always in the same position, which could be a minor limitation.
- Naturalness: The data comes from read speech, which may not perfectly reflect spontaneous speech.
- Future Directions: The authors suggest that further research could explore how these findings apply to other languages and intonational contours. They also highlight the need to understand the perceptual implications of these stable alignment patterns for listeners. The distinction between primary and secondary stress in terms of duration, but not timing, also opens avenues for further investigation into the acoustic correlates of stress.
Summary updated