en | de | es | gl
|
Text resources
|
Publications
|
Team
|
Contact

About PaGeS


The German/Spanish parallel Corpus, PaGeS, is is part of an ongoing major project PaCorEs, Parallel Corpora Spanish, which aims to collect a series of bilingual parallel corpora with Spanish as the central language. So far, the project includes three other corpora at different stages of completion, all of them freely available online: Corpus PaEnS) Eglish < > Spanish, Corpus PaCheS Chinese < > Spanish and Corpus PaFreS French < > Spanish.

The core corpus is comprised of original texts in German and Spanish and their respective translations. It includes 229 works of fiction —novels and short stories, making up around 89% — as well as non-fiction — essays and popular science texts. The selected works are represented not by the full texts, but rather by samples, allowing for a better cross-section of the texts.

This part of PaGeS (s. below) contains nearly 42,000,000 tokens and 1.246.733 bisegments, i.e. pairs of aligned text chunks (sentences or subsentential units/segments).

To guarantee overall quality, the texts have been manually verified at different levels and the automatic alignment of the bisegments, performed by LF-Aligner, some of them by YouAlign or Gargantua has been manually reviewed. The German texts have been lemmatized and pos-tagged with Treetagger and the Spanish texts with Freeling. The tags of both have been subsequently mapped to the Universal POS tags, that mark the core part-of-speech categories.

For each occurrence, the original source is provided, which includes information on the author, title, year of the first publication, and — if applicable —the edition used and the part or chapter within the work to which the specific occurrence belongs. The complete bibliographic data of the works included in PaGeS can be found here.

The supplements contain a total of more than 80 million words. The supplements include so far:

  1. 18 works of fiction from German and Spanish texts translated from a third language. The alignment of these texts has been checked manually. The list of these works can be found here.
  2. Europarl v7, a corpus that collects the proceedings (Verbatim reports) of the European Parliament from 1996 to 2011. Segments with more than 350 characters (in Spanish and/or German) and those with a character imbalance exceeding a 2:1 ratio have been excluded. More details here.
  3. Ted-Talks, a corpus that collects the German and Spanish translations of the transcripts of 2859 TED-Talks from 2006 to 2020. More details here.
  4. OpenSubtitles v2018, a large collection of translated movie subtitles.

In the near future, new collections of bilingual texts of diverse origin are expected to be added.

Even though the initial impulse for the creation of PaGeS was to provide a broad source of data for contrastive linguistic research, the excellent reception by very diverse users has motivated us to strive for greater interoperability and standardization in order to make PaGeS a multifunctional resource able to meet the differentiated needs of our users. We aim at building a representative language resource for the language pair German / Spanish that can be exploited for multiple purposes such as general research in contrastive linguistics, linguistic typology, translation studies and bilingual lexicography, as well as the supply of training data to machine translation systems. PaGeS has also proven to be a very useful and widely used resource by translators and learners of German or Spanish as Foreign Languages at intermediate and advanced levels to obtain a multitude of translation suggestions made by humans and presented within examples of real language use.

For more detailed information about PaGeS, see the publications webpage and Doval, Irene / Sánchez Nieto, M.ª Teresa (en prensa): Parallel Corpora Spanish (PaCorES): A collection of multifunctional parallel corpora. In: RESLA. Revista Española de Lingüística Aplicada / Spanish Journal of Applied Linguistics.

Despite our best efforts, some mistakes have undoubtedly slipped through. If you come across any, please let us know by clicking here.

Notice:

If you use PaGeS in your work, please indicate it and let us know: corpuspages@usc.es. This way you contribute to the sustainability of the project.

Statistics PaGeS

Core Corpus

LANGUAGE CHARACTERS WORDS TOKENS MSTTRATIO* BISEGMENTS WORKS
German Original 48.949.765 9.419.402 11.049.990 0.577 702.731 114
Spanish Translation 46.634.694 10.083.202 11.581.354 0.542
Spanish Original 40.047.261 8.579.478 9.809.708 0,539 544.002 115
German Translation 44.210.720 8.494.546 9.924.406 0.572
Total 179.842.440 36.576.628 42.365.45 0.557 1.246.733 229

Supplements 1: Fiction translated from a third language

LANGUAGE CHARACTERS WORDS TOKENS MSTTRATIO* BISEGMENTS WORKS
German Translation 10.885.529 2.097.569 2.463.109 0.569 152.077 18
Spanish Translation 10.008.379 2.148.750 2.479.765 0.545
Total 20.893.908 4.246.319 4.942.874 0.557 152.077 18

Supplements 2: Europarl v7

LANGUAGE CHARACTERS WORDS TOKENS MSTTRATIO* BISEGMENTS
German 203.075.349 33.405.712 37.703.328 0.542 1.555.845
Spanish 190.849.843 37.694.928 41.518.229 0.481
Total 393.925.192 71.100.640 79.221.557 0.5115 1.555.845

Supplements 3: TED-Talks

LANGUAGE CHARACTERS WORDS TOKENS MSTTRATIO* BISEGMENTS
German 23.824.529 5.046.593 5.805.812 0.506 321.924
Spanish 25.344.941 4.754.108 5.599.587 0,543
Total 49.175.477 9.800.701 11.405.399 0.525 321.924

Supplements 4: OpenSubtitles v2018

LANGUAGE CHARACTERS WORDS TOKENS MSTTRATIO* BISEGMENTS
German 175.689.253 37.703.271 47.050.044 0,565 5.840.417
Spanish 159.707.973 36.664.180 46.709.094 0,575
Total 335.397.226 74.367.451 93.759.138 0.570 5.840.417

*MSTTR is the average TTR (Type/Token Ratio) for each non-overlapping segment of equal size (in this case 1000 tokens).


(Stand: 19/11/2024, Release 2.2)

                                                    
PaGeS Vers. 2.2
Last updated: 30.10.2024
ISLRN 300-741-224-666-2
ISSN 2605-5228 ©PaCorES
Creative Commons Licencia Creative Commons
University of Santiago de Compostela
This project is funded by the State Research Agency (AEI) of Spanish Ministry of Science, Innovation and University (PID2021-125313OB-I00).