Examining the acquisition of phonological word forms with computational experiments

Area of Science:

• Cognitive Science
• Computational Linguistics
• Psychology

Background:

• Theories suggest existing words aid new word learning.
• Dense word neighborhoods (many similar known words) may facilitate novel word acquisition.
• Sparse neighborhoods might pose learning challenges.

Purpose of the Study:

• To test the hypothesis that dense word neighborhoods enhance novel word learning.
• To investigate the role of lexical neighborhood density in word acquisition using computational models.
• To examine how processing resources affect this learning pattern.

Main Methods:

• Utilized a connectionist network model to simulate word learning.
• Manipulated the density of word neighborhoods (dense vs. sparse).
• Varied exposure conditions (all at once vs. gradual) and processing resource availability.

Main Results:

• Dense neighborhood words were learned better than sparse neighborhood words under standard conditions.
• This advantage persisted with gradual word exposure and varied processing resources.
• A sparse neighborhood advantage emerged only when dense neighborhoods were introduced later in training.

Conclusions:

• Computational models support the hypothesis that dense lexical neighborhoods facilitate word learning.
• The findings have implications for understanding language acquisition and processing disorders.
• Connectionist models offer valuable insights into psycholinguistic processes.