Effects of consonant environment on vowel formant patterns

Area of Science:

• Speech perception research
• Acoustic phonetics
• Auditory processing

Background:

• Vowel identification is known to be influenced by spectral change patterns.
• Previous studies showed improved vowel category separability using formant pattern classifiers with multiple samples.
• Earlier research used a constant consonant-vowel-consonant (/hVd/) phonetic environment.

Purpose of the Study:

• To investigate if the relationship between vowel identity and spectral change patterns persists in varied consonant environments.
• To determine the impact of changing phonetic contexts on vowel perception.
• To assess the robustness of spectral change cues for vowel identification.

Main Methods:

• Recordings of eight vowels produced in isolation and CVC syllables by 12 speakers.
• CVC syllables included all combinations of seven initial and six final consonants.
• Formant frequencies (F1-F3) were measured every 5 ms using an interactive editing tool.

Main Results:

• Phonetic environment significantly affected formant patterns, especially for rounded vowels in alveolar contexts.
• Despite context-dependent shifts, spectral change information substantially improved vowel category separability.
• A pattern classifier using F0, duration, and two formant samples accounted for many listener behaviors.

Conclusions:

• Spectral change patterns remain crucial for vowel identification, even with varying consonant contexts.
• The study confirms the importance of dynamic spectral information over static formant patterns.
• Simple pattern classifiers incorporating spectral change effectively model aspects of human speech perception.