Perception of voice-onset-time continua: a signal detection analysis

Area of Science:

• Phonetics and Speech Perception
• Psychoacoustics
• Auditory Neuroscience

Background:

• Understanding how humans perceive speech sounds is crucial for fields like linguistics and artificial intelligence.
• The perception of voice-onset-time (VOT) in stop consonants is a key area in speech research.
• Previous models have debated the relationship between speech sound identification and discrimination.

Purpose of the Study:

• To investigate the relationship between absolute identification and discrimination of synthetic syllables.
• To examine the role of voice-onset-time (VOT) in alveolar stop perception.
• To test predictions of different speech perception models, including dual-coding and continuous models.

Main Methods:

• Participants performed absolute identification and various discrimination tasks on synthetic syllables.
• Stimuli varied in voice-onset-time (VOT) and were initiated by alveolar stops.
• Signal detection theory (using d') was employed to analyze performance in both tasks.

Main Results:

• Discrimination performance (d') was significantly better than absolute identification performance.
• This difference was attributed to greater judgment instability in the identification task.
• A category boundary effect was observed, with peak performance near the voiced/voiceless boundary for alveolar stops.

Conclusions:

• The relationship between identification and discrimination is influenced by task constraints and stimulus properties.
• A category boundary effect is not solely dependent on equal performance in identification and discrimination tasks.
• A single-process continuous model may better explain the observed relationships than a dual-coding model.