Statistical learning by 8-month-old infants

Area of Science:

• Cognitive Science
• Developmental Psychology
• Linguistics

Background:

• Learners utilize both experience-independent and experience-dependent mechanisms for environmental information processing.
• Language acquisition theories predominantly highlight the role of innate, experience-independent mechanisms.
• The segmentation of words from continuous speech is a critical early language development task.

Purpose of the Study:

• To investigate if infants can segment words from fluent speech using solely statistical learning.
• To determine the minimal exposure required for infants to perform word segmentation via statistical learning.
• To assess the capability of infants' statistical learning mechanisms in processing linguistic input.

Main Methods:

• Exposing 8-month-old infants to 2 minutes of fluent speech.
• Analyzing infants' ability to distinguish word boundaries based on statistical regularities between speech sounds.
• Utilizing a statistical learning paradigm to evaluate auditory processing.

Main Results:

• Infants successfully segmented words from fluent speech.
• Segmentation was achieved based solely on statistical relationships between adjacent speech sounds.
• This word segmentation occurred after only 2 minutes of exposure to the speech input.

Conclusions:

• Infants possess a powerful, experience-independent mechanism for statistical learning in language acquisition.
• Statistical learning from auditory input is a fundamental component of early word segmentation.
• The findings challenge the overemphasis on innate knowledge, highlighting the power of statistical computation in language development.