Spelling performance on the web and in the lab.

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Cognitive Psychology

Background:

• Web dictionary usage provides extensive data on user-entered orthographic forms.
• Understanding spelling processes is crucial for cognitive and educational psychology.
• Large-scale data analysis offers new avenues for psychological research.

Purpose of the Study:

• To create a large-scale database of French spelling errors from web dictionary data.
• To assess the reliability of this big data methodology by comparing it with experimental data.
• To explore task-dependent factors influencing spelling accuracy.

Main Methods:

• Systematic recording of 200 million orthographic forms from a French web dictionary over nine years.
• Selection of 100 frequently misspelled words from the database.
• Conducting a spelling-to-dictation test with 100 French university students.

Main Results:

• A strong correlation (r = 0.82) was found between the web dictionary error data and student dictation test results for spelling frequencies.
• While error distributions were consistent, significant differences in correct response proportions were observed.
• Regression analyses identified task-dependent factors contributing to these differences.

Conclusions:

• The big data methodology using web dictionary entries is a reliable approach for studying spelling errors.
• Comparing big data findings with controlled experiments is essential for validating the methodology.
• This approach can significantly inform psychological theories of spelling processes.