Can BERT generate Quebec English? : exploring and improving the acceptability of regional variation in pretrained language models
Bibliographic record
Abstract
While pretrained language models have transformed natural language processing by generating contextually rich language representations, their performance can vary widely across different language varieties. Monolingual models, trained on large datasets in a single language, may struggle with language varieties that incorporate elements from multiple languages. This thesis examines the ability of pretrained language models, particularly BERT, to handle Quebec English, a regionally influenced variety of English shaped by contact with French. By comparing three different BERT models-one monolingual, one multilingual, and one fine-tuned on Quebec English-specific data-this study evaluates their effectiveness in generating Quebec English target words and English synonyms within a masked language modeling framework. Results suggest that fine-tuning improves performance for Quebec English-specific target words, outperforming the standard pretrained models. Additionally, findings indicate that tokenization, sentence context, and pretraining data substantially impact prediction accuracy, with all models struggling most with infrequent, region-specific expressions. This work contributes to the broader goal of developing natural language processing tools that inclusively represent diverse linguistic communities, underscoring the importance of fine-tuning in adapting language models to regional and minority language varieties.
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 | 0.007 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
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".