Feedforward, -backward, and neutral transparency measures for British English

Area of Science:

• Psycholinguistics
• Cognitive Psychology
• Computational Linguistics

Background:

• Orthographic transparency metrics are crucial for understanding reading processes in deep languages like English.
• Existing metrics often focus narrowly on feedforward and feedback directions, potentially oversimplifying reading influences.

Purpose of the Study:

• To analyze five orthographic transparency metrics for Southern British English, including novel non-feedforward/feedback metrics.
• To investigate the impact of corpus size and origin on these metrics.
• To explain feedback effects observed in children's reading accuracy.

Main Methods:

• Calculation and analysis of five orthographic transparency metrics for Southern British English.
• Examination of metric structures across various corpus sizes and origins.
• Comparison of adult and child corpus profiles.

Main Results:

• Demonstrated complex interrelationships among the five transparency metrics.
• Identified metrics that are neither feedforward nor feedback, offering new insights.
• Found that large corpus sizes do not significantly enhance metric value compared to smaller samples.
• Observed highly similar metric profiles between adult and child corpuses.

Conclusions:

• Orthographic transparency in English is more complex than previously modeled by feedforward/feedback metrics alone.
• The findings provide a basis for understanding feedback effects in children's reading.
• Metric reliability is not substantially improved by using very large corpora; smaller, representative samples are effective.