Europarl3

A parallel corpus extracted from the European Parliament web site by Philipp Koehn (University of Edinburgh). Appr. 40 million words per language. The main intended use is to aid statistical machine translation research.

More information can be found at http://www.statmt.org/europarl/. The main difference in this release vs. the first release in 2002 and second release in 2003 is that it is larger and it comes with improved processing tools that allow the creation of parallel corpora between any two of the 11 languages. Some data is now tagged with the original language the text was spoken in.

11 languages, 55 bitexts
total number of files: 7,034
total number of tokens: 404.01M
total number of sentence fragments: 15.81M

Please cite the following article if you use any part of the corpus in your own work:
Jörg Tiedemann, 2009, News from OPUS - A Collection of Multilingual Parallel Corpora with Tools and Interfaces. In N. Nicolov and K. Bontcheva and G. Angelova and R. Mitkov (eds.) Recent Advances in Natural Language Processing (vol V), pages 237-248, John Benjamins, Amsterdam/Philadelphia

Download

Old version: Europarl2.

NEW: Dutch Europarl3 Treebank (parsed with Alpino).

Complete download (XML): Europarl3.tar.gz (3,5G )

Bottom-left triangle: download files

ces = sentence alignments in XCES format
leftmost column language IDs = corpus files in XML
TMX and plain text files (Moses): see "Statistics" below
lower row language IDs = parsed corpus files (if they exist)

Upper-right triangle: sample files

view = bilingual XML file samples
upper row language IDs = monolingual XML file samples
rightmost column language IDs = parsed XML file samples (if they exist)

	da	de	el	en	es	fi	fr	it	nl	pt	sv
da		view	view	view	view	view	view	view	view	view	view	da
de	ces		view	view	view	view	view	view	view	view	view	de
el	ces	ces		view	view	view	view	view	view	view	view	el
en	ces	ces	ces		view	view	view	view	view	view	view	en
es	ces	ces	ces	ces		view	view	view	view	view	view	es
fi	ces	ces	ces	ces	ces		view	view	view	view	view	fi
fr	ces	ces	ces	ces	ces	ces		view	view	view	view	fr
it	ces	ces	ces	ces	ces	ces	ces		view	view	view	it
nl	ces	ces	ces	ces	ces	ces	ces	ces		view	view	nl
pt	ces	ces	ces	ces	ces	ces	ces	ces	ces		view	pt
sv	ces	ces	ces	ces	ces	ces	ces	ces	ces	ces		sv
	da	de	el	en	es	fi	fr	it	nl	pt	sv

Statistics and TMX/Moses Downloads

Number of files, tokens, and sentences per language (including non-parallel ones if they exist)
Number of sentence alignment units per language pair

Upper-right triangle: download translation memory files (TMX)
Bottom-left triangle: download plain text files (MOSES/GIZA++)


language	files	tokens	sentences	da	de	el	en	es	fi	fr	it	nl	pt	sv
da	659	37.5M	1.6M		1.3M	0.7M	1.3M	1.3M	1.2M	1.3M	1.2M	1.3M	1.2M	1.1M
de	659	37.5M	1.5M	1.3M		0.6M	1.3M	1.3M	1.2M	1.3M	1.2M	1.3M	1.2M	1.1M
el	537	26.3M	1.0M	0.7M	0.7M		0.6M	0.7M	0.6M	0.7M	0.6M	0.7M	0.6M	0.6M
en	658	39.4M	1.5M	1.3M	1.3M	0.7M		1.3M	1.2M	1.3M	1.2M	1.3M	1.3M	1.1M
es	660	40.5M	1.5M	1.3M	1.3M	0.7M	1.3M		1.2M	1.3M	1.2M	1.3M	1.2M	1.1M
fi	611	26.4M	1.4M	1.2M	1.2M	0.6M	1.3M	1.2M		1.2M	1.2M	1.2M	1.2M	1.1M
fr	662	43.7M	1.5M	1.3M	1.3M	0.7M	1.3M	1.3M	1.3M		1.2M	1.3M	1.3M	1.1M
it	658	39.0M	1.4M	1.2M	1.2M	0.6M	1.2M	1.2M	1.2M	1.3M		1.2M	1.2M	1.1M
nl	659	39.5M	1.6M	1.3M	1.3M	0.7M	1.3M	1.3M	1.2M	1.3M	1.2M		1.3M	1.1M
pt	660	40.9M	1.4M	1.3M	1.3M	0.7M	1.3M	1.3M	1.2M	1.3M	1.2M	1.3M		1.1M
sv	611	33.4M	1.5M	1.2M	1.2M	0.7M	1.2M	1.1M	1.2M	1.2M	1.1M	1.2M	1.1M

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.