liv4ever v1

This is the Livonian 4-lingual parallel corpus. Livonian is a Uralic / Finnic language with just about 20 fluent speakers and no native speakers (as of 2021). The texts and translations in this corpus were collected from all the digital text resources that could be found by the authors; scanned and printed materials are left for future work.

The corpus includes parallel data for Livonian-Latvian, Livonian-Estonian and Livonian-English; the data has been collected in 2021. After retrieval it was normalized in terms of different orthographies of Livonian and manually sentence-aligned where needed. It was collected from the following sources, with sentence counts per language pair:

• Dictionary - example sentences from the Livonian-Latvian-Estonian dictionary;
  liv-lv: 10'388, liv-et: 10'378
• Stalte - the alphabet book by Kōrli Stalte, translated into Estonian and Latvian;
  liv-lv: 842, liv-et: 685
• Poetry - the poetry collection book "Ma võtan su õnge, tursk / Ma akūb sīnda vizzõ, tūrska", with Estonian translations;
  liv-et: 770
• Vääri - the book by Eduard Vääri about Livonian language and culture;
  liv-et: 592
• Satversme - translations of the Latvian Constitution into Livonian, Estonian and English;
  liv-en: 380, liv-lv: 414, liv-et: 413
• Facebook - social media posts by the Livonian Institute and Livonian Days with original translations;
  liv-en: 123, liv-lv: 124, liv-et: 7
  
• JEFUL - article abstracts from the Journal of Estonian and Finno-Ugric Linguistics, special issues dedicated to Livonian studies, translated into Estonian and English;
  liv-en: 36, liv-et: 49
  
• Trilium - the book with a collection of Livonian poetry, foreword and afterword translated into Estonian and Latvian;
  liv-lv: 51, liv-et: 53
  
• Songs - material crawled off lyricstranslate.com;
  liv-en: 54, liv-lv: 54, liv-fr: 31
  

A small part of the corpus is held out in order to be used as a translation benchmark (dev and test set, see below for download). Its material was manually translated in order to have each sentence translated in all four languages (Livonian-English-Latvian-Estonian). The material sources and resulting sentence counts:

• Livones.net - web crawl of the Livonian culture, language and history web portal;
  163 sentences
  
• Stalte - 310 sentences
• Dictionary - 298 sentences
• Facebook - 189 sentences
• Satversme - 188 sentences
• Trilium - 172 sentences
• JEFUL - 120 sentences
  

5 languages, 4 bitexts
total number of files: 38
total number of tokens: 0.40M
total number of sentence fragments: 54.74k

Please cite the following article if you use any part of the corpus in your own work: Matīss Rikters, Marili Tomingas, Tuuli Tuisk, Valts Ernštreits, Mark Fishel: Machine Translation for Livonian: Catering to 20 Speakers. In Proceedings of ACL2022. and also acknowledge OPUS by citing J. Tiedemann, 2012, Parallel Data, Tools and Interfaces in OPUS. In Proceedings of the 8th International Conference on Language Resources and Evaluation (LREC 2012)

Download

Below you can download data files for all language pairs in different formats and with different kind of annotation (if available). You can click on the various links as explained below. In addition to the files shown on this webpage, OPUS also provides pre-compiled word alignments and phrase tables, bilingual dictionaries, frequency counts, and these files can be found through the resources search form on the top-level website of OPUS.

License: CC BY-SA

Other Download Packages

• Independent development set: liv4ever-dev.tgz
• Independent test set: liv4ever-test.tgz
• Original training set: liv4ever-train.tgz

Release history:

• liv4ever v1

Statistics and TMX/Moses Downloads

Upper-right triangle: download translation memory files (TMX)
Bottom-left triangle: download plain text files (MOSES/GIZA++)
Language ID's, first row: monolingual plain text files (tokenized)
Language ID's, first column: monolingual plain text files (untokenized)

Note that TMX files only contain unique translation units and, therefore, the number of aligned units is smaller than for the distributions in Moses and XML format. Moses downloads include all non-empty alignment units including duplicates. Token counts for each language also include duplicate sentences and documents.

Disclaimer

• We do not own any of the text from which the data has been extracted.
• We only offer files that we believe we are free to redistribute. If any doubt occurs about the legality of any of our file downloads we will take them off right away after contacting us.

Notice and take down policy

• Clearly identify yourself, with detailed contact data such as an address, telephone number or email address at which you can be contacted.
• Clearly identify the copyrighted work claimed to be infringed.
• Clearly identify the material that is claimed to be infringing and information reasonably sufficient to allow us to locate the material.
• And contact the OPUS project at the following email address: opus-project at helsinki.fi.