Crowdsourcing Piedmontese to Test LLMs on Non-Standard Orthography
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914332005105664 |
|---|---|
| author | Vico, Gianluca Libovický, Jindřich |
| author_facet | Vico, Gianluca Libovický, Jindřich |
| contents | We present a crowdsourced dataset for Piedmontese, an endangered Romance language of northwestern Italy. The dataset comprises 145 Italian-Piedmontese parallel sentences derived from Flores+, with translations produced by speakers writing in their natural orthographic style rather than adhering to standardized conventions, along with manual word alignment. We use this resource to benchmark several large language models on tokenization parity, topic classification, and machine translation. Our analysis reveals that Piedmontese incurs a tokenization penalty relative to higher-resource Romance languages, yet LLMs achieve classification performance approaching that of Italian, French, and English. Machine translation results are asymmetric: models translate adequately from Piedmontese into high-resource languages, but generation into Piedmontese remains challenging. The dataset and code are publicly released. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_14675 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Crowdsourcing Piedmontese to Test LLMs on Non-Standard Orthography Vico, Gianluca Libovický, Jindřich Computation and Language We present a crowdsourced dataset for Piedmontese, an endangered Romance language of northwestern Italy. The dataset comprises 145 Italian-Piedmontese parallel sentences derived from Flores+, with translations produced by speakers writing in their natural orthographic style rather than adhering to standardized conventions, along with manual word alignment. We use this resource to benchmark several large language models on tokenization parity, topic classification, and machine translation. Our analysis reveals that Piedmontese incurs a tokenization penalty relative to higher-resource Romance languages, yet LLMs achieve classification performance approaching that of Italian, French, and English. Machine translation results are asymmetric: models translate adequately from Piedmontese into high-resource languages, but generation into Piedmontese remains challenging. The dataset and code are publicly released. |
| title | Crowdsourcing Piedmontese to Test LLMs on Non-Standard Orthography |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2602.14675 |