TL;DR
This paper investigates the production and perception of Vietnamese short vowels, particularly their distinction from classical vowels when followed by a semivowel. The study reveals that while vowel duration alone isn't always sufficient for differentiation, dynamic acoustic features like formant transition slopes play a crucial role in distinguishing these vowels, both in production and perception.
What Problem Does This Paper Address?
Vietnamese has a set of "special" short vowels that cannot be produced in isolation and always require a following consonant or semivowel. This obligate final part raises questions about its role in distinguishing these short vowels from their longer, classical counterparts. Previous research has hinted at dynamic acoustic differences, but a comprehensive analysis of how the final semivowel influences the production and perception of these vowel pairs (e.g., /a/ vs. /ă/) in a vowel-semivowel (VS) context was lacking. This paper aims to fill that gap by examining vowel duration, formant transition duration, and formant transition slopes in VS and CVC contexts, and then validating these findings with perception experiments.
Methods
The researchers collected a corpus of Vietnamese vowel-semivowel (VS) and consonant-vowel-semivowel (CVS) syllables produced by four native male speakers. They analyzed acoustic features such as:
- Vowel duration (Vdur): The length of the vowel sound.
- Formant transition duration (VSdur): The time it takes for the formants to transition from the vowel to the semivowel.
- Final semivowel duration (Sdur): The length of the semivowel.
- Formant transition slopes: The rate of change of the first three formants (F1, F2, F3) during the vowel-semivowel transition.
Statistical analysis, including One-way ANOVA, was used to compare the acoustic features between classical and special vowel pairs. For perception experiments, synthesized V1V2 (vowel-vowel) syllables were created by manipulating vowel durations and then presented to listeners to assess their ability to distinguish between classical and special vowels.
Datasets
The study utilized a newly recorded corpus of Vietnamese VS/CVS syllables from four native speakers. This corpus included both actual lexical items and non-lexical items, totaling 440 tokens. For perception experiments, synthesized V1V2 syllables were generated based on the acoustic analysis.
Key Insights and Ideas
- Dynamic Features Matter: The study strongly suggests that dynamic acoustic features, specifically formant transition slopes, are critical for distinguishing Vietnamese short vowels from their classical counterparts, even when static features like target formants are similar.
- Final Semivowel's Role: The final semivowel in a VS context is not merely a passive follower; its duration and the way it interacts with the preceding vowel (via formant transitions) actively contribute to the perception of vowel identity.
- Duration Ratios as Indicators: A clear relationship exists between vowel duration and final semivowel duration. When the vowel is short (especially special vowels), the semivowel duration tends to be longer, and vice-versa, suggesting a compensatory mechanism.
- Perception Mirrors Production: Perception experiments confirm that listeners rely on similar acoustic cues (like duration ratios and dynamic transitions) to distinguish between classical and special vowels as observed in production.
Key Results
- In VS/CVS contexts, formant transition duration (VSdur) remained relatively stable, while vowel duration (Vdur) and final semivowel duration (Sdur) varied significantly.
- A duration ratio was observed: Vdur = 1.6*Sdur for classical vowels, and Sdur = 1.6*Vdur for special vowels.
- Classical-special vowel pairs (/a/-/ă/ and /४/-/४/) were consistently discriminated by at least one formant transition slope (F1, F2, or F3) in the VS context.
- Perception tests showed that listeners could distinguish between classical and special vowels based on duration ratios in V1V2 contexts. For example, the classical vowel /a/ was perceived when V1dur was at least twice as long as V2dur, while the special vowel /ă/ was recognized when V1dur was short and V2dur was longer than V1dur.
Limitations and Future Work
- Limited Speaker Pool: The production corpus was recorded by only four male native speakers from the Hanoi dialect. This limits the generalizability of the findings to other dialects or speaker groups.
- Focus on Specific Vowel Pairs: While the study focused on key classical-special vowel pairs, a more extensive analysis across all twelve Vietnamese vowels in various contexts could provide a more complete picture.
- Future Work: The authors suggest that further research could explore the role of the final consonant in distinguishing special vowels and investigate the influence of tone on these distinctions. The study also highlights the importance of dynamic acoustic characteristics in speech production and perception, suggesting this as a fruitful area for future investigation.
Summary updated