Discovering functional units in continuous speech.

Area of Science:

• Cognitive Science
• Auditory Perception
• Language Acquisition

Background:

• Language learning involves identifying functional units like words from continuous sound.
• Previous studies focused on isolated sounds, neglecting challenges of continuous input.
• Understanding category discovery in naturalistic auditory streams is crucial for language acquisition models.

Purpose of the Study:

• To investigate how listeners discover sound categories embedded in continuous, unfamiliar auditory streams.
• To explore the role of multisensory regularities and procedural learning in category acquisition.
• To determine if learning generalizes to novel contexts and isolated instances.

Main Methods:

• Adult participants engaged in a video game task with acoustically variable sound categories.
• Sounds were embedded in continuous, unfamiliar, and multisensory-rich streams.
• Learning was assessed after 100 minutes of gameplay through categorization and generalization tasks.

Main Results:

• Participants successfully learned to categorize target words embedded in familiar sound streams.
• Learned categories generalized to novel sound streams and isolated word instances.
• Discovery of predictive environmental regularities occurred without explicit attention or prior knowledge.

Conclusions:

• Listeners can discover functional sound categories within continuous streams, even with non-native speech and nonspeech signals.
• This learning process is robust and generalizes across different auditory contexts.
• Findings highlight the brain's capacity for implicit learning of auditory regularities crucial for language.