Linguistics

German Texts

• The German National Corpus assembles various corpora from the 15th century to the present day.
• The Mannheim German Reference Corpus (Das Deutsche Referenzkorpus – DeReKo) contains corpora of contemporary written German at the IDS - Leibniz-Institut für Deutsche Sprache. 
• DDB (Deut­sche Dia­chro­ne Baum­bank) is a small (ca. 8000 to­kens) deep­ly syn­tac­ti­cal­ly an­no­tat­ed cor­pus con­sist­ing of three sub­-cor­pora (Old High Ger­man, Mid­dle High Ger­man, Ear­ly New High Ger­man).
• RIDG­ES proj­ect (Reg­is­ter in Di­a­chron­ic Ger­man Sci­ence) is an in­ves­ti­ga­tion in­to the de­vel­op­ment of the Ger­man sci­en­tif­ic lan­guage in the ear­ly mod­ern and mod­ern pe­ri­ods, rang­ing from the mid 15th to the 20th cen­tu­ry.
• German Political Speeches Corpus and Visualization consists of speeches by the German Presidents, Chancellors and a few ministers, all gathered from official sources. 

Spoken German

• Bavarian Archive for Speech Signals: Audio corpora of German speech, including read, spontaneous and telephone. Varieties include regional speech, adolescent speech and intoxicated speech.

German as a Foreign Language

• Falko: A freely available error-annotated learner corpus of German as a foreign language
• KanDeL (Kansas Developmental Learner corpus): A freely available longitudinal corpus of beginning to intermediate learners of German as a foreign language, 

Dutch Texts

• SUBTLEX-NL is a database of Dutch word frequencies based on 44 million words from film and television subtitles.

Spoken Dutch

• Spoken Dutch Corpus (CGN) contains 800 hours and nearly 9 million words of Dutch and Flemish speech.

Danish

• KorpusDK is a large Danish-language corpus containing 56 million words.
• Danish @ VISL Corpuseye is a research project based in the Department of Culture and Language, University of Southern Denmark.

Faroese

• FTS - Faroese text collection contains 1,142,289 tokens and was developed in cooperation with University of the Faroe Islands.
• The Faroese Danish Corpus Hamburg (FADAC Hamburg) is a corpus of spoken, informal Faroese and Faroese Danish. 440,000 words have been collected between 2005 and 2009 as part of the project Variation in Multilingualism on the Faroe Islands.

Icelandic

• The Icelandic Spoken Language Corpus (ISLC) contains four different sub-corpora: Spontaneous conversations, Group conversations, Parliamentary debates and Conversations of teenagers.
• The Icelandic Gigaword Corpus (IGC) consists of about 1,5 billion tagged running words of text from official documents, news media and various materials in the collection of the Árni Magnússon Institute for Icelandic Studies.
• The Icelandic Parsed Historical Corpus (IcePaHC) is a manually corrected treebank that contains around 1 million words from every century between the 12th and the 21st.

Norwegian

• NB Tale – Speech Database for Norwegian is a basic acoustic-phonetic speech database. It contains recordings of 380 speakers from 24 different dialect areas.
• Norwegian Newspaper Corpus contains texts from 1998 to 2019 with approximately 1,68 billion words for Norwegian Bokmål and about 68 million words for Norwegian Nynorsk.
• NoWaC (Norwegian Web as Corpus) is a large web-based corpus of Bokmål Norwegian currently containing about 700 million tokens.

• The Oslo Corpus of Tagged Norwegian Texts is based on works of fiction, factual prose and newpaper/magazines. Bokmål part contains some 18.5 million words, while the Nynorsk part contains about 3.8 million.

Swedish

• Corpora and resources from Stockholm University Department of Linguistics.
• Kubhist 2 is a diachronic collection of Swedish historical newspaper texts from the period of 1645–1926.

• Stockholm—Umeå Corpus (SUC) is a collection of various Swedish texts from the 1990s, totaling 1 million words.

Multilingual sources

• Corpus of American Nordic Speech is a speech corpus with speakers from USA and Canada speaking Norwegian and Swedish. There are 268 speakers from 63 places, with more than 774,000 tokens.
• The LIA Sápmi Corpus is a speech corpus with recordings from 1960 to 1990 of Sami dialects from the northern part of Norway, Finland and Sweden. The corpus contains about 190,000 tokens from 122 speakers in 19 locations.
• Nordic Dialect Corpus and Syntax Database is a corpus of Norwegian, Swedish, Danish, Faroese, Icelandic and Övdalian spoken languages. It consists of spontaneous speech data with 2.75 million words from a variety of sources recorded in 1998–2015.

Portuguese Text 

• O Corpus Brasileiro: 1 billion words of contemporary Brazilian Portuguese in a variety of genres and registers.
• O Corpus do Português: Consists of the Historical Corpus that contains more than 45 million words in almost 57,000 Portuguese texts from the 1300s to the 1900s; the Web Corpus with about 1 billion words of Portuguese taken from 1 million web pages from four Portuguese-speaking countries (Brazil, Portugal, Angola and Mozambique); and the NOW (News on the Web) Corpus that has about 1.1 billion words from web-based newspapers and magazines in four Portuguese-speaking countries from 2012-2019.

Spoken Portuguese

• Arquivo Dialetal do CLUP (Dialectal Archive of the Center of Linguistics of the University of Porto): A database of recordings of European Portuguese collected during the last two decades, spanning both Mainland Portugal and islands. 

Bi-directional

• COMPARA (Portuguese-English Parallel Corpus): A bi-directional parallel corpus based on an open-ended collection of Portuguese-English and English-Portuguese source-texts and translations. 

Italian Texts

• CORIS (CORpus di Italiano Scritto): A reference corpus of written Italian, CORIS contains 150 million words.
• DiaCORIS (Diachronic Corpus of Written Italian): Project that aims at the construction of a diachronic corpus comprising written Italian texts produced between 1861 and 1945, extending the structure and possibilities of CORIS.
• BoLC (Bononia Legal Corpus): A multilingual comparable legal corpus developed at the University of Bologna.
• PAISA (Piattaforma per l’Apprendimento dell’Italiano Su corpora Annotati): A large corpus of authentic contemporary Italian texts from the web.

Spoken Italian

• BADIP (Banca dati dell'italiano parlato): Database of spoken Italian. Contains an online edition of the 500,000 word LIP-Corpus. The edition is being enriched with POS-tags and lemmata, more data are being added continuously. Deactivated as of 2019.
• CLIPS (Corpora e Lessici dell'Italiano Parlato e Scritto) is a corpus of spoken and written Italian. It contains about 100 hours of speech, equally represented by female and male voices. A section of the corpus is transcribed orthographically and a smaller section has been phonetically labeled.

Sicilian

• Text Archive of the Ancient Sicilian: A corpus of 736 medieval Sicilian texts from 13th to 16th centuries

• The Polish National Corpus: A reference corpus of Polish language containing over 1.5 billion words. The corpus is searchable by means of advanced tools that analyze Polish inflection and Polish sentence structure.

Czech Texts

• The Czech National Corpus (Český národní korpus): Large electronic corpus of written and spoken Czech with more than 4 billion tokens.

• Czech Academic Corpus (Český akademický korpus): A morphologically and syntactically annotated corpus of the Czech language consisting of approximately 650,000 words in continuous texts.

Spoken Czech

• Prague DaTabase of Spoken Czech (PDTSC): Authentic spoken Czech from the Prague area and its surroundings with over 7,000 minutes of spontaneous speech recorded during several related projects.

Slovak

• Slovak National Corpus: An electronic database of contemporary Slovak language texts (1955-present), covering a broad range of language styles, genres, areas, regions etc.

• Corpus of Slovene Language (FIDA) contains just over 100 million words of Slovene texts found in the press, the Internet and speech transcripts from the late 1990s.
• Slovenian National Corpus (Gigafida) aka Korpus pisne standardne slovenščine is an upgrade of the older FIDA corpus that currently contains over 1 billion words.
• Slovene web corpus (slWaC) is a web corpus collected from the .si top-level domain. The current version of the corpus (v2.0) contains 1.2 billion tokens and is annotated with the lemma and the morphosyntax layer.

Bosnian

• Bosnian web corpus (bsWaC) is a web corpus collected from the .ba top-level domain. The 1.0 version of the corpus contains 429 million tokens and is annotated with the lemma, morphosyntax and dependency syntax layers.
• The Oslo Corpus of Bosnian Texts compiled at the University of Oslo consists of approximately 1.5 million words from various texts, including fiction, legal texts, newspapers and journals, mostly published in the 1990s.

Croatian

• Croatian National Corpus (Hrvatski nacionalni korpus, HNK): A corpus of modern Croatian with over 2.5 million words.

• Croatian Learner Corpus (CroLTeC) contains texts collected from learners of Croatian as a second and foreign language with over 1 million tokens.

• Croatian web corpus (hrWaC) is a web corpus collected from the .hr top-level domain. The current version of the corpus (v2.0) contains 1.9 billion tokens and is annotated with the lemma, morphosyntax and dependency syntax layers.

Serbian

• The Corpus of Serbian Language (CSL) was compiled from a sample of 11 million words and spans the Serbian language from the 12th century to the contemporary period.

• The Corpus of Modern Serbian Language (SrpKor2013) contains 122 million words collected from literary and administrative texts, newspapers, magazines and the web.

• Serbian web corpus (srWaC) is a web corpus collected from the .rs top-level domain. The 1.0 version of the corpus contains 894 million tokens and is annotated with the lemma, morphosyntax and dependency syntax layers.

• The Macedonian Spoken Corpus was developed at the University of Zurich and comprises transcriptions of audio files collected mostly during a series of field research trips to the Prespa, Bitola and Debar regions between 2012 and 2019.

Hungarian Texts

• The Hungarian National Corpus (HNC) currently contains 187.6 million words of Hungarian language spoken in Hungary, Slovakia, Subcarpathia, Transylvania and Vojvodina. It is divided into sub-corpora based on regional variants and text genres.
• The Linguistics Research Center (A Nyelvtudományi Kutatóközpont) databases include Old Hungarian Corpus, Hungarian Historical Corpus and many more.

Spoken Hungarian

• Beszélt nyelvi Adatbázis (BEA) or Spoken Language Database is a phonetically grounded multifunctional database of spontaneous Hungarian speech. It contains data from nearly 500 speakers, representative in age, sex, educational background and dialect in the central region of Hungary.

• Budapesti Szociolingvisztikai Interjú (BUSZI) or Budapest Sociolinguistic Network is a large-scale survey that provides reliable data and analyses of the varieties of the Hungarian language spoken in Budapest.

• Mari Language is a digital collection of resources in and about Mari language varieties developed by researchers at the University of Vienna in collaboration with indigenous scholars and native speakers.

• Meadow Mari Corpora contain the corpus of contemporary written literary texts in Meadow Mari variety, as well as the corpus of social media materials.

Erzya

• Erzya Corpora contain the corpus of contemporary written literary Erzya and the corpus of social media and forums in Erzya language.

Moksha

• Moksha Corpora contain the corpus of contemporary written literary Moksha and the corpus of social media and forums in Moksha language.

• Ob-Ugric Languages (OUL) is a collection of online descriptive resources, including text corpora, of two related and endangered Ob-Ugric languages: Khanty (Ostyak) - Kazym and Surgut dialects - and Mansi (Vogul) - Northern and Eastern dialects.
• Ob-Ugric Database (OUB) is a project related to OUL with analysed text corpora and dictionaries for less described Ob-Ugric dialects: Yugan Khanty and Western Mansi.

Komi

• Corpus of the Komi Language (Коми кыв корпус / Korpus komi jazyka) is an electronic resource developed by the Interregional Laboratory of Information Support for the Functioning of Finno-Ugric Languages (FU-Lab).

• The Komi media collection is an electronic reference information system based on an annotated recordings of Komi dialects in video and audio formats with translations into Russian and English

• The Komi-Zyrian Corpora contain the corpus of contemporary written literary texts in Komi-Zyrian variety, as well as the corpus of social media materials.

Udmurt

• Udmurt Corpora contain the corpus of contemporary written literary Udmurt, the corpus of Udmurt-language social media and the sound-aligned corpus of Udmurt dialects.

Kamas

• The INEL Kamas corpus has been created within the long-term INEL project. It consists of two parts: folklore texts collected by Kai Donner in 1912–1914, and transcribed audio recordings of the now-extinct Kamas language's last speakers made between 1964 and 1970.

Nganasan

• The Nganasan Spoken Language Corpus (NSLC) includes 177 communications, 136 of which contain an aligned audio recording, with glossed and annotated transcripts from 57 speakers. All texts have been translated into Russian and English, some also into German.

Selkup

• The INEL Selkup Corpus has been created within the long-term INEL project. It is composed of archival material from between 1962 and 1977.
• The Selkup Language Corpus (SLC) contains 144 communications with glossed and annotated transcripts from 53 speakers. All texts have been translated into English and German, most of them are also available in Russian.

• Corpora of Uralic Volga-Kama Languages contains texts published on the internet. There are two corpora for each language: the Main corpus and the Social Media corpus.
• Endangered Languages and Cultures of Siberia is a multimedia collection of linguistic and cultural information about endangered languages and cultures of Siberia.
• INEL - Grammars, corpora and language technology for Indigenous Northern Eurasian Languages, 2016–2033 is a long-term project funded jointly by the federal and state governments and coordinated by the Union of German Academies of Sciences.
• Uralonet - Uralic Etymological Database is a platform for studying protolanguage connections, reconstructed protoforms and many other aspects of 25 Uralic languages.
• Uralic Languages Under the Influence Database (UraLUID) is a linguistically annotated database of written and spoken sources in Udmurt, Tundra Nenets, Synja and Surgut Khanty languages designed for the study of Uralic–Russian language contacts.

• Farsdat (Farsi Speech Database) comprises recordings of 300 Iranian speakers representing ten different dialects. 6,000 utterances were segmented and labelled phonetically and phonemically.

• Uppsala Persian Corpus (UPC) is a large, tagged and freely available Persian corpus. It is a modified version of the Bijankhan corpus and contains almost 3 million tokens.

• TS Corpus Project is a free and independent project that aims to build Turkish corpora, develop natural language processing tools and compile linguistic datasets. The 14 currently published corpora together contain over 1.3 billion tokens derived from various sources: online newspapers, forums, social media, academic papers etc.
• Turkish National Corpus (TNC): With a size of 50 million words, TNC is a balanced and a representative corpus of contemporary Turkish. It consists of samples of textual data across a wide variety of genres covering a period of 24 years (1990-2013). It was developed by researchers at Mersin and Çukurova University.
• METU Spoken Turkish Corpus is a selection of 18 Turkish audio recordings and archives from Radyo ODTÜ hosted by the Department of Foreign Language Education at Middle East Technical University (METU).

Text / Written Corpora

• Balanced Corpus of Contemporary Written Japanese (BCCWJ): BCCWJ is a corpus that attempts to grasp the breadth of contemporary written Japanese. It contains extensive samples of modern Japanese texts in order to create as uniquely balanced a corpus as possible. The data is comprised of 104.3 million words, covering genres such as general books and magazines, newspapers, business reports, blogs, internet forums, textbooks, and legal documents among others. Random samples of each genre were taken.

Spoken Corpora

• Corpus of Spontaneous Japanese (CSJ): The CSJ corpus contains approximately 650 hours of spontaneous contemporary Japanese speech recorded between 1999 and 2003. Two kinds of speech, academic presentation speeches (APS) and simulated public speaking (SPS), were the main sources, but some material was also taken from interviews with the subjects about their APS or SPS, and from recordings of the subjects reading short passages aloud.
• Japanese Speech Corpora of Major City Dialects consists of a series of recordings made in the 1990s.
   

• National Institute of the Korean Language Corpus: Dataset with information on the frequency of use of modern Korean language.

• Open Korean Corpora: A living document for Korean NLP dataset curation.

• Mongolian National Corpus contains 966 texts with 1,155,583 words.

• Classical Tibetan Corpus consists of a small number of Classical Tibetan texts that were linguistically analyzed and annotated.
• Tibetan Corpus contains audio and video recordings of Tibetan native speakers.

• Hanoi Dialect Corpus contains recordings made during the study of glottalization in Vietnamese tones.
• SEAlang Library Vietnamese Text Corpus consists of Vietnamese newspaper, literary, and Wikipedia texts published on the Internet.

• Cardamon Khmer Corpus contains recordings of the Cardamon dialect made by ethnologist Marie Alexandrine Martin in 1960s-70s.
• SEAlang Library Khmer Text Corpus consists of Khmer texts published on the Internet.

• SEAlang Library Burmese Text Corpus consists of Burmese texts published on the Internet.

• Indonesian Conversational Speech Corpus consists of 4.54 hours of transcribed Indonesian conversational speech.
• SEAlang Library Indonesian Text Corpus contains Indonesian texts retrieved from a variety of Internet sources.

• Malay-Dataset contains various linguistic datasets collected from the internet.
• Malay Language Word Corpus for Primary School (MyBAca) contains common words that occur in the Malay language textbooks used in Malaysian schools and the related linguistic information.
• SEAlang Library Malay Text Corpus consists of Malay texts retrieved from a variety of Internet sources.

• SEAlang Library Tagalog Text Corpus contains more than 2 million words taken from various internet sources, as well as Ramos Tagalog-English Dictionary and Tagalog Literary Text collection.
• Tagalog Corpus contains a number of one-hour recordings of interactions between different groups of Tagalog-speaking adults (Adult language corpus), and children and their guardians (Child language corpus).
• UP Filipino Language Corpus (UP-FLC) is a work in progress to create a corpus of standardized written and spoken texts.

• Corpus for Afroasiatic Languages (CorpAfroAs) is the first fully-annotated and searchable corpus of lesser-described Afroasiatic languages.

• Legacy Documentation of Soo and Nyangi consists of digitized audio recordings of Soo and Nyangi produced in northeastern Uganda between 1964 and 1972.
• Nuer Collection presents data recorded between 2015-2019 with speakers of Lou Nuer around the Akobo town in South Sudan.
• Shilluk & Dinka Datasets contain a number of recordings from native speakers of Shilluk and Dinka languages.