Papers › Sicilian Translator: A Recipe for Low-Resource NMT

Sicilian Translator: A Recipe for Low-Resource NMT

5 Oct 2021arXiv:2110.01938archive 2025-07-28

Eryk Wdowiak

With 17,000 pairs of Sicilian-English translated sentences, Arba Sicula developed the first neural machine translator for the Sicilian language. Using small subword vocabularies, we trained small Transformer models with high dropout parameters and achieved BLEU scores in the upper 20s. Then we supplemented our dataset with backtranslation and multilingual translation and pushed our scores into the mid 30s. We also attribute our success to incorporating theoretical information in our dataset. Prior to training, we biased the subword vocabulary towards the desinences one finds in a textbook. And we included textbook exercises in our dataset.

Paper PDF Code

Code

ewdowiak/Sicilian_Translator officialmxnetApache-2.0 report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Attribute Low Resource NMT Low-Resource Neural Machine Translation Machine Translation NMT Translation

Results from the paper archive 2025-07-28

Task	Dataset	Model	Metric	Value	Rank at snapshot	Leaderboard	Report
Machine Translation	Arba Sicula	Larger	BLEU (En-Scn)	35.0	#1 of 2	Archive leaderboard	report
Machine Translation	Arba Sicula	Larger	BLEU (Scn-En)	36.8	#1 of 2	Archive leaderboard	report
Machine Translation	Arba Sicula	Many-to-Many	BLEU (It-Scn)	36.5	#2 of 2	Archive leaderboard	report
Machine Translation	Arba Sicula	Many-to-Many	BLEU (Scn-It)	30.9	#2 of 2	Archive leaderboard	report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position Encodings Adam Attention BPE Dense Connections Dropout Label Smoothing Layer Normalization Linear Layer Multi-Head Attention Position-Wise Feed-Forward Layer Residual Connection Softmax Transformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections