Can infants map meaning to newly segmented words? Statistical segmentation and word learning

Area of Science:

• Cognitive Psychology
• Developmental Psychology
• Linguistics

Background:

• Early word learning is a complex process involving both identifying word boundaries in continuous speech and associating those words with meanings.
• Statistical word segmentation, a key mechanism in early language acquisition, allows infants to parse fluent speech into discrete word units.

Purpose of the Study:

• To investigate the link between statistical word segmentation abilities and subsequent word learning in 17-month-old infants.
• To determine if prior exposure to word forms within a statistical segmentation task facilitates the learning of object labels.

Main Methods:

• Infants participated in a statistical word segmentation task using fluent speech.
• Immediately following segmentation, infants engaged in an object-label learning task.
• Labeling conditions varied: actual words from the segmentation task, novel nonwords, or familiar part-words.

Main Results:

• Infants successfully learned object labels when the labels were words previously identified in the segmentation task.
• Infants failed to learn object labels when presented with novel nonwords or familiar part-words with low internal probabilities.
• The findings indicate that prior segmentation opportunities, rather than mere frequency of exposure, facilitated word learning.

Conclusions:

• Exposure to word forms during statistical word segmentation significantly enhances subsequent word learning in infants.
• This study provides the first empirical evidence directly linking statistical word segmentation to facilitated word learning.
• The results highlight the importance of perceptual learning in early vocabulary acquisition.