Accents régionaux en français : perception, analyse et modélisation à partir de grands corpus
Bibliographic record
Abstract
Large oral corpuses including regional accents of French become today available: their data offer a good base to begin the study of accents. The tools of automatic treatment of the word allow to treat quantities of data more important than the samples that the experts linguists, phoneticians or dialectologues can examine. The French language is spoken in numerous countries worldwide. Our study concerns French of continental Europe, so excluding territories as Quebec, French-speaking Africa or still French overseas departments. We shall study regional accents of France, Belgium and Swiss French. What are the geographical limits inside which it is possible to assert that the speakers have the same accent? The answer to this question is not evident. We adopted the following terminology, adapted to our data: we shall speak about accent when we shall make reference to a precise localization such as a city or a given region; we shall use the term variety to indicate a vaster group. Although numerous studies describe the peculiarities of the accents of French, there are fewer works describing the variation of the language in general, and even less from the point of view of the automatic treatment. Numerous questions remain opened. How many accents can a listener native of French identify? What performances could an automatic system reach for an identical task? Can the indications described in the linguistic literature as characteristics of certain accents be measured in a automatic way? Are they relevant to differentiate varieties of French? Shall we discover the other measurable indications on our corpuses? These indications can be put in connection with the perception? During our thesis, we approached the study of regional varieties of French from the point of view of the human perception as well as of that of the automatic treatment of the word. Traditionally, count of studies in linguistics focus on the study of a precise accent. The automatic treatment of the word allows to envisage the joint study of several varieties of French: we wanted to exploit this possibility. We can so examine what differs from a variety in the other one, what is not possible when a single variety is described. We are lucky to have at our disposal a successful system of automatic alignment of the word. This tool, which allows to segment the sound flow following a phonemic transcription, can show itself precious for the study of the variation. The automatic treatment allows us to consider several styles of word and numerous speakers on quantities of important data with regard to those who were able to be used in linguistic studies led manually. We automatically extracted characteristics of the signal by various methods; we tried to validate our results on two corpuses with accents. The parameters which we held allowed to classify automatically the speakers of our two corpuses.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".