Diphones-fr: a French database of diphone positional frequency.

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Speech Science

Background:

• Understanding sequential probabilities in spoken language is crucial for speech perception and production.
• Existing resources for French diphone frequencies, particularly positional ones, are limited.
• Speech segmentation relies on processing phonetic sequences.

Purpose of the Study:

• To describe a novel database of diphone positional frequencies in French.
• To provide comprehensive frequency data for French diphones in various positions (word-initial, -internal, -final).
• To offer new indicators for psycholinguistic research on French speech.

Main Methods:

• Extraction of words from a large subtitle corpus (50 million words) from French movies and TV series.
• Calculation of diphone frequencies based on their position within words (initial, internal, final).
• Analysis of intra- and inter-syllable, as well as inter-word diphone frequencies.

Main Results:

• A comprehensive database of diphone positional frequencies in French has been compiled.
• Frequencies are provided for word-initial, word-internal, and word-final diphones.
• Intra-syllable, inter-syllable, and inter-word diphone frequencies are also available.

Conclusions:

• The created database is a unique tool for psycholinguists studying French.
• The new indicators can facilitate research on speech segmentation and sequential processing.
• This resource addresses a gap in available tools for the analysis of French spoken language.