SYN version 10

Name SYN version 10
Position Number of tokens 5 887 514 992
Number of tokens without punctuation 4 881 700 519
Number of word forms 11 033 112
Number of lemmas 7 432 413
Structures Number of documents 131 383
Number of texts 18 158 945
Number of sentences 374 658 105
Other information Referential YES
Representative NO (predominantly journalism)
Publication year 2022

Every SYN corpus contains all the synchronic written corpora of the SYN series published up until the time of the given version's publication. The corpus SYN version 10 therefore contains the corpora SYN2000, SYN2005, SYN2006PUB, SYN2009PUB, SYN2010,SYN2013PUB, SYN2015 and SYN2020; additionally, it contains a journalistic component predominantly from the years 2010–2019 (already included into SYN version 4, SYN version 5, SYN version 6, SYN version 7 and SYN version 8) and as yet unpublished journalistic texts from 2020 in yearly volume exceeding 150 mil. words (however, only after the publication of the whole corpus we found out that part of the journalistic data from 2020 was inadvertently incorrectly lemmatized, for more details see below).

The SYN corpus is not representative; the dominant component is journalism, which is the result of the predominance of journalistic corpora SYN2006PUB, SYN2009PUB, SYN2013PUB and the journalistic component from the years 2010–2020.

The SYN version 10 corpus is referential, and will remain accessible to users even after newer versions have been published. It is however necessary to keep in mind that the linguistic information will become outdated, as a natural result of the referential nature of the corpus. Individual versions of the corpus SYN will continue to be published regularly every year with the addition of current journalistic data, and every new addition will be given the attribute value <doc syn> equal to the version of the SYN corpus in which the given text first appeared; for example a subcorpus corresponding to the above mentioned (as yet unpublished) journalistic component can be created from the SYN version 10 with the help of the condition syn="v10". However, it was during the preparation of journalistic data from 2020 (syn="v10") that an error ocurred: forms with variant sublemmas (eg. the form voknem) were assigned a lemma identical to the sublemma (vokno) instead of a special lemma (okno). This error affected about 1.8% of tokens in this part of the corpus. If your work with the SYN version 10 corpus depends on accurate statistics, we recommend you to exclude the subcorpus matching the condition syn="v10" from the search.

Composition of the corpus SYN version 10

Referential written language corpora (synchronic and general) ordered by date of creation
corpus size (words) lemmatization morphological tags publication year corpus description
SYN2020 100M 2020 representative corpus, mainly texts from the years 2015–2019
SYN2015 100M 2015 representative corpus, mainly texts from the years 2010–2014
SYN2013PUB 935M 2013 corpus of journalistic texts from the years 2005–2009
SYN2010 100M 2010 representative corpus, mainly texts from the years 2005–2009
SYN2009PUB 700M 2010 corpus of journalistic texts from the years 1995–2007
SYN2006PUB 300M 2006 corpus of journalistic texts from the years 1989–2004
SYN2005 100M 2005 representative corpus, mainly texts from the years 2000–2004
SYN2000 100M 2000 representative corpus, mainly texts from the years 1990–1999

Journalism in SYN version 10

The composition of the journalistic part of the corpus SYN version 10 covers the production of most of the national daily newspapers (Mladá fronta DNES, Lidové noviny, Právo, Hospodářské noviny, Blesk, Sport), regional daily newspapers (chiefly Deníky Bohemia and Moravia published by Vltava Labe Media) and non-specialized magazines (Reflex, Respekt, Týden) from 1998–2020; the total number of journalistic titles is almost 200. The following graphs show the composition of the SYN corpus based on the main text types over the years and offer a closer look at the composition of the journalistic section.

Composition of the corpus SYN version 10
Composition of the journalistic part of the corpus SYN version 10

Structure and annotation of SYN version 10

Generally speaking, structure and annotation of SYN version 10 are based on that of the SYN2020 corpus. In particular, hierarchy of structural tags for SYN version 10 has been taken over from SYN2020, as well as the lemmatization and morphological tagging. SYN version 10 shares all these aspects with its predecessor, SYN version 9.

This correspondence of structure and annotation between SYN version 10 and SYN2020 only has the following exceptions:

How to cite SYN version 10

Křen, M. – Cvrček, V. – Hnátková, M. – Jelínek, T. – Kocek, J. – Kováříková, D. – Křivan, J. – Milička, J. – Petkevič, V. – Procházka, P. – Skoumalová, H. – Šindlerová, J. – Škrabal, M.: Corpus SYN, version 10 from 22. 2. 2022. Ústav Českého národního korpusu FF UK, Praha 2022. Available online: https://www.korpus.cz.

Hnátková, M. – Křen, M. – Procházka, P. – Skoumalová, H. (2014): The SYN-series corpora of written Czech. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14), 160–164. Reykjavík: ELRA. ISBN 978-2-9517408-8-4.

Jelínek, T. – Křivan, J. – Petkevič, V. – Skoumalová, H. – Šindlerová, J. (2021): SYN2020: A new corpus of Czech with an innovated annotation. In: K. Ekštein – F. Pártl – M. Konopík (eds.), Text, Speech, and Dialogue. TSD 2021. Lecture Notes in Computer Science, vol. 12848. Cham: Springer, 48–59.

Křivan, J. – Šindlerová, J. (2022): Změny v morfologické anotaci korpusů řady SYN: nové možnosti zkoumání české gramatiky a lexikonu. Slovo a slovesnost, 83, 2/2022, 122–145.

Michal Křen

Related links