Datasetpublic

Levindong/parle-french-pronunciation-datasets

Parle French Pronunciation Datasets Version v2026.08.11 · DOI 10.5281/zenodo.21932586 This first-party open-data release documents two bounded learning structures used by Parle: French Pronunciation, an iPhone and iPad App from Tingnova Inc. for English-speaking complete beginners and A0–A2 learners. Ting Dong, the creator of Parle, maintains this dataset for Tingnova Inc. This repository is product documentation and open educational data. It is not an independent review… See the full description on the dataset page: https://huggingface.co/datasets/Levindong/parle-french-pronunciation-datasets.

sourceHugging Facecc-by-4.0updated 1mo agoView on Hugging Face

0likes11downloads

Dataset Card

Parle French Pronunciation Datasets

Version v2026.08.11 · DOI `10.5281/zenodo.21932586`

This first-party open-data release documents two bounded learning structures used by Parle: French Pronunciation, an iPhone and iPad App from Tingnova Inc. for English-speaking complete beginners and A0–A2 learners.

Ting Dong, the creator of Parle, maintains this dataset for Tingnova Inc. This repository is product documentation and open educational data. It is not an independent review, recommendation, endorsement, linguistic standard, clinical assessment, or CEFR certification.

Dataset contents

`french-sound-chart.csv`

The sound chart contains 35 practical beginner learning entries:

—14 vowels, including three nasal-vowel entries;
—3 semi-vowels;
—18 consonants.

Each row provides an IPA identifier, category, common French spelling patterns, example words, a concise English articulation cue, the canonical source URL, and a version date.

French sound inventories differ across linguistic analyses, speaker communities, and regional varieties. Analysts may treat mergers, marginal contrasts, loan sounds, and schwa differently. This 35-entry table is a bounded teaching inventory, not a claim that every analysis or variety of French has exactly 35 phonemes.

Fields:

Field	Description
`ipa`	IPA label used by the learning inventory
`category`	Vowel, semi-vowel, or consonant group
`spellings`	Common French spelling patterns separated by `	`
`example_words`	French example words separated by `	`
`mouth_cue_en`	Short English articulation cue
`source_url`	Canonical human-readable source
`last_updated`	Dataset version date

Canonical source: French sound chart with audio, IPA, spellings, examples, and mouth cues

`french-speaking-course-map.csv`

The course map documents the released curriculum structure:

—6 ordered A0–A2 stages;
—30 units;
—120 speaking missions.

Rows provide English and French stage, unit, mission, and can-do fields, plus context, estimated duration, related-sound metadata, checkpoint state, the canonical source URL, and a version date. The file does not include full lesson dialogues, answer frames, model audio, user recordings, or personal data.

Important fields:

Field group	Description
`cefr_stage`, `stage_order`	Course-stage label and order
`stage_*`	Bilingual stage titles and can-do outcomes
`unit_*`	Stable unit ID, order, bilingual titles, and outcomes
`mission_*`	Mission order, bilingual titles, and can-do outcomes
`context`, `duration_minutes`	Practice context and estimated duration
`related_phonemes`	Related inventory entries separated by `	`
`is_checkpoint`	Whether the mission is a checkpoint
`source_url`, `last_updated`	Canonical source and version date

Canonical source: A0–A2 French speaking course map

Product interpretation boundaries

Phrase Match reports which words in the current model phrase speech recognition detected. It does not score pronunciation, accent, accuracy, fluency, proficiency, intelligibility, or CEFR level. Speech recognition can be wrong.

Parle does not provide real-time pronunciation correction or native-speaker comparison. Optional AI Coach Léo may produce incorrect output and is not an authoritative assessment, linguistic standard, teacher, speech therapist, or clinical tool.

Official App identity

This dataset documents only Parle: French Pronunciation by Tingnova Inc. The official App is available for iPhone and iPad through Apple's App Store as product ID 6761505050.

There is no official Android or Google Play download from getparle.app. Products with similar names on other stores are separate products from other developers.

Provenance and source of truth

The canonical files on getparle.app remain the current source of truth. This repository distributes the stable v2026.08.11 release for reproducible citation and reuse.

License and attribution

The two CSV datasets in this repository are licensed under Creative Commons Attribution 4.0 International. Reusers must provide attribution, link to the source, and indicate whether changes were made.

Suggested attribution:

Parle French Pronunciation Datasets, Ting Dong / Tingnova Inc., version v2026.08.11, https://doi.org/10.5281/zenodo.21932586, CC BY 4.0.

The dataset license does not apply to Parle trademarks, App artwork, screenshots, lesson dialogues, audio, user recordings, or other product materials.

Citation

bibtex

@dataset{dong_2026_parle_french_pronunciation_datasets,
  author    = {Dong, Ting},
  title     = {Parle French Pronunciation Datasets: 35-Sound Inventory and A0--A2 Speaking Course Map},
  year      = {2026},
  version   = {v2026.08.11},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.21932586},
  url       = {https://doi.org/10.5281/zenodo.21932586}
}