What do we do with what we learn? Statistical learning of orthographic regularities impacts written word processing

Area of Science:

• Cognitive Psychology
• Psycholinguistics
• Computational Linguistics

Background:

• Statistical learning is a fundamental cognitive process for detecting environmental patterns.
• The role of statistical learning in acquiring orthographic regularities for reading is debated and underexplored in current theories.
• Understanding which specific orthographic regularities readers learn is crucial for refining reading models.

Purpose of the Study:

• To investigate whether readers rapidly acquire sensitivity to orthographic regularities, specifically bigram frequency, in artificial scripts.
• To determine if this learned sensitivity influences subsequent reading-related tasks, such as wordlikeness judgments and letter detection.
• To explore the implications of these findings for models of visual word recognition and orthographic encoding.

Main Methods:

• Three experiments utilized artificial scripts with controlled bigram frequencies.
• Participants were exposed to artificial words for short durations (minutes) or longer periods (days) while learning word forms.
• Sensitivity to learned regularities was assessed using wordlikeness and letter detection tasks.

Main Results:

• Participants rapidly developed sensitivity to positional letter cluster frequency and bigram frequency after brief exposure.
• Learned orthographic knowledge significantly affected performance in the letter detection task, speeding up the detection of high-frequency letters.
• Sensitivity extended beyond single letter frequency, indicating learning of multi-letter patterns.

Conclusions:

• Readers can quickly learn and utilize orthographic regularities, including bigram frequencies, from environmental input.
• Statistical learning of orthographic patterns plays a more significant role in reading than previously assumed by some models.
• These findings necessitate revisions to current models of visual word recognition to incorporate rapid statistical learning mechanisms.