OPUS Neapolitan Translations: Nearly 1 Million Italian-English-Neapolitan Sentence Pairs

OPUS Neapolitan Translations provides nearly 1 million parallel translation examples across Italian, English, and Neapolitan. The dataset was created by author Gdacciaro, starting from an OPUS English-Italian parallel corpus and generating Neapolitan translations using a translation model. It was last updated on June 14, 2026.

Use Cases

• Train machine translation models based on the parallel Italian-English-Neapolitan sentence pairs.
• Benchmark translation quality for the Neapolitan language based on the generated translations.
• Study linguistic patterns and code-switching based on the trilingual parallel corpus.
• Create language learning resources for Neapolitan based on the aligned translations.

Strengths

• Contains nearly 1 million parallel translation examples, providing substantial data for model training.
• Offers trilingual alignment (Italian, English, Neapolitan), which is less common than bilingual pairs.
• Based on the established OPUS corpus, suggesting a foundation of curated translation data.

Limitations

• The Neapolitan sentences were machine-generated, which may introduce translation errors or artifacts.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific scaling needs.

Provenance

Source

OPUS English-Italian parallel corpus.

Collection Method

Original Italian-English pairs used as base; Neapolitan sentences generated using a Neapolitan translation model.

Freshness

Last updated 2026-06-14 20:28:50; freshness should be verified.