diphone_probability

Scripts for calculating and analyzing diphone probability data, based on the framework of Warner et al. (2005). Used in research on the role of phonotactic probability and duration in speech perception.

Overview

Diphone probability reflects how likely a particular two-phone sequence is in a given language. These scripts compute and apply diphone probability values to experimental speech perception data, supporting analysis of how listeners use statistical regularities during spoken word recognition.

Scripts

calculateProbability.py

Computes diphone (bigram) probabilities from a phonological corpus.

Usage:

python calculateProbability.py

calculateRespProbability.py

Computes response-level probability metrics for experimental stimuli.

Usage:

python calculateRespProbability.py

add_probability_to_data.py

Merges computed probability values into experimental datasets for statistical analysis.

Usage:

python add_probability_to_data.py

Related Publications

• Park, S. & Warner, N. (2023). "The role of probability and duration in perception of speech sounds." Speech Communication 152, 102950.
• Park, S., Hoffmann, M., Shin, P.Z. & Warner, N.L. (2018). "The role of segment probability in perception of speech sounds." JASA 143(3_Supplement), 1920.
• Warner, N.L. et al. (2005). Original diphone probability framework.

Requirements

• Python 3.6+
• NumPy, pandas

Author

Seongjin Park — seongjinpark.com

License

MIT