Input and Age-Dependent Variation in Second Language Learning: A Connectionist Account.

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Second Language Acquisition

Background:

• Previous studies suggested limited improvement in second language (L2) rule knowledge despite increased language exposure.
• Variations in rule testing across studies may have obscured the relationship between input and L2 knowledge.

Purpose of the Study:

• To re-examine the impact of linguistic input on L2 rule acquisition, accounting for rule-specific variations.
• To develop and test a computational model explaining the interaction of input frequency and sensitive periods in L2 learning.

Main Methods:

• Reanalyzed grammaticality judgment scores from a previous L2 learner study using rule-specific predictors.
• Developed a connectionist model simulating L1 Korean and L2 English learning with age-dependent learning rates.
• Modeled differential learning rates for syntax and lexical components to capture age-related differences.

Main Results:

• L2 knowledge increased with years of input, even with a sensitive period effect.
• Input frequency of specific grammar rules negatively correlated with knowledge of those rules.
• The connectionist model successfully replicated observed L2 learning patterns, including input sensitivity differences.

Conclusions:

• Linguistic input significantly enhances L2 rule knowledge, modulated by a sensitive period.
• A connectionist approach with L1 transfer and age-dependent learning rates can model L2 acquisition dynamics.
• This framework provides an explicit account of how input and sensitive periods interact in second language learning.