All words are not created equal: expectations about word length guide infant statistical learning.

Area of Science:

• Cognitive Science
• Developmental Psychology
• Linguistics

Background:

• Infants are considered statistical learners, extracting linguistic structures from auditory input.
• Sequential statistical learning is crucial for language acquisition in early development.

Purpose of the Study:

• To investigate if prior knowledge influences infants' ability to track transitional probabilities in word segmentation.
• To determine if infants' sequential statistical learning is biased by previous experience with word length.

Main Methods:

• A laboratory simulation exposed 9- and 10-month-old infants to lists of either disyllabic or trisyllabic nonsense words.
• Following pre-exposure, infants heard a pause-free speech stream containing different sets of disyllabic or trisyllabic nonsense words.
• Infants' listening times were recorded to assess successful word segmentation from fluent speech.

Main Results:

• Successful word segmentation occurred only when word lengths (disyllabic or trisyllabic) were consistent throughout both experimental phases.
• Exposure to trisyllabic words during pre-exposure impaired segmentation of disyllabic words in the speech stream, and vice versa.
• This indicates a disruption in statistical learning when prior word length expectations conflict with subsequent input.

Conclusions:

• Prior knowledge of word length creates perceptual expectations in infants.
• These expectations facilitate more efficient processing and segmentation of subsequent language input.
• This finding highlights the role of experience in shaping early language learning mechanisms.