Speech Segmentation and Cross-Situational Word Learning in Parallel

Area of Science:

• Cognitive Science
• Psycholinguistics
• Developmental Psychology

Background:

• Language learners utilize statistical information, specifically conditional probabilities, to identify words in continuous speech.
• Cross-situational word learning involves mapping words to objects across ambiguous contexts.
• Simultaneous application of speech segmentation and word mapping remains an underexplored area in language acquisition research.

Purpose of the Study:

• To investigate whether language learners can simultaneously perform speech segmentation and cross-situational word learning.
• To examine how conflicting statistical information in linguistic input affects these dual learning processes.

Main Methods:

• Adult participants (N=60 in Experiment 1, N=35 in Experiment 2) engaged in a combined task of segmenting continuous speech and mapping words to objects.
• Conflicting statistical properties were introduced into the speech stream.
• Speech segmentation was assessed using standard measures and a more sensitive measure in Experiment 2.

Main Results:

• Simultaneous performance of speech segmentation and word mapping led to better outcomes than individual task performance.
• When presented with conflicting statistics, participants successfully mapped words to objects but performed at chance level for speech segmentation.
• A more sensitive measure revealed participants could distinguish non-words but not between words and part-words, while mapping remained above chance.

Conclusions:

• Language learners can effectively track multiple statistical information sources for word discovery and object mapping in challenging, noisy environments.
• The findings highlight the brain's capacity for parallel statistical processing during language acquisition.
• Further research is needed to refine methods for measuring learning outcomes from complex statistical environments.