Corpus

The SpeechReporting corpus

The SpeechReporting corpus contains corpora of traditional folk stories, annotated for a number of discourse phenomena using the ELAN-CorpA software and tools (Chanard 2015; Nikitina et al. 2019). It is updated regularly with newly available data, including data from new languages. All texts are transcribed, glossed, translated, and annotated.

The corpus was developed as part of the project “Discourse Reporting in African Storytelling”, funded by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 758232, PI Tatiana Nikitina). In addition to standard glosses and part of speech tags, it is annotated for instances of reported discourse; see our Annotation Guide.

Languages and their corpora

A narrative corpus of Bashkir

Language and storytelling tradition

Bashkir is a Turkic language spoken in Bashkortostan (Russia) by approximately 1,2 million people. Nowadays, the traditional way of storytelling no longer exists, and people do not gather to tell stories. However, there are people who still remember tales that were told to them by older people in their childhood. Moreover, a lot of tales are published and people might retell these written tales.

Corpus composition

The current version of the corpus contains about two hours of texts (10.500 words) and corresponding video files. 1 h 20 minutes were collected by Ekaterina Aplonova during fieldwork in 2019; the rest comes from the Spoken Corpus of Bashkir (without video). The texts were recorded in the following villages: Abzanovo, Aslayevo, Baimovo, Kyzgy, Mullakaevo, Rakhmetovo, Tuishevo, and Usakly.

Orthography

The texts are transcribed in the Latin script with some extensions, unlike standard Bashkir orthography, which is based on the Cyrillic script. The system employed here is mainly a transliteration of the standard Bashkir orthography, but it has some features of phonological transcription. Correspondences between the standard orthography and the transliteration system employed here are listed on the Spoken Corpus of Bashkir website.

List of parts of speech

List of glosses

Acknowledgements

I would like to thank my Bashkir language assistants who told me their stories and Fizaliya Makhanova for her help with transcription, as well as Lilya Buskunbaeva and Ramilya Karimova from Ufa Institute of Linguistics for their precious help during data collection and analysis.

Citing the sub-corpus:
Aplonova, Ekaterina. 2021. A narrative corpus of Bashkir. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Bandial (Jóola Eegimaa)

Language and storytelling tradition

Jóola Eegimaa is an Atlantic language spoken in Senegal by approximately 15.700 people. The practice of storytelling is largely extinct in Senegal; even in remote areas, the only people capable of telling stories are 'rememberers' of their grandparents' generation. Among the Jóola, many tales are likely borrowed from other linguistic-ethnic groups such as the Mandinka.

Corpus composition

The current version of the corpus contains about two hours of texts (about 12.000 words) and corresponding video files. The data was collected during a three month period in 2018-2019 by Abbie Hantgan-Sonko in the area known as The Kingdom in Casamance.

Orthography

The transcription is made according to the official Jóola (otherwise written as Jola or Diola) orthography which indicates long vowels by doubling the letter: aa ee ii oo uu , and advanced tongue root ([+ATR]) by an acute accent over the first vowel of a stem: á é í ó ú . Retracted or [-ATR] roots are unmarked.

List part of speech tags and glosses (adapted from Sagna 2008: 18)

Citing the Jóola Eegimaa corpus:
Hantgan-Sonko, Abbie. 2021. A narrative corpus of Jóola Eegimaa. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Chuvash

Language

Chuvash is a Turkic language spoken in Russia by approximately 1 million people

Corpus composition

The current version of the corpus contains examples of traditional storytelling recorded by A.K. Salmin in the 1980s (the original recordings are available in the archival collections of the Chuvash State Institute of Humanities) and folktales recorded by Elena Fedotova during her fieldwork in the summer of 2019. The total length of the currently available portion of the corpus is about 5 hours; the total number of words is about 25.000. The corpus is under construction, with new texts due to become available soon.

List of parts of speech

List of glosses

Citing the Chuvash corpus:
Nikitina, Tatiana. 2022. A narrative corpus of Chuvash. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Ginyanga

Language and storytelling tradition

Ginyanga is a North Guang language belonging to the Kwa language family spoken by a population of 16500 people in Togo and Ghana. Stories are still occasionally told during night gatherings, but they are less and less frequent. However, even young people generally know a couple of tales.

Corpus composition

The data were collected in 2019-2023 by Katya Aplonova and Viktoria Pozdnyakova in Agbandi and Diguina in Blitta prefecture in Togo. The data were transcribed and glossed with a precious help of our language assistants: Yao Amédomé, Evégénou Tchala, Emmanuel Kontré and Atchou Edoh. The available part of the corpus corresponds to approximately 15 minutes.

List of parts of speech

List of glosses

A narrative corpus of Gizey

Language and storytelling tradition

Gizey is a Masa languoid (Chadic < Afro-Asiatic) spoken by approximately 19.000 people across Cameroon and Chad. Stories are still occasionally told during night gatherings. People generally know a couple of tales, which they modify freely to suit the storytelling context.

Corpus composition

The data were collected in 2019 by Guillaume Guitang and Dieudonné Soupoursou in Djougoumta and Maïda (Mayo-Danay, Cameroon). The data were partly transcribed by Victor Marheyna. In total, the corpus contains around 210 minutes. Only part of this corpus (around 30 minutes) is currently available.

List of parts of speech

List of glosses

Citing the Gizey corpus:
Guitang, Guillaume. 2022. A narrative corpus of Gizey. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Guro

Language

Guro is a South Mande language spoken in Ivory Coast by approximately 500.000 people. The practice of storytelling is endangered and undergoing some transformations. Even in remote areas it is largely abandoned as a family practice. Instead, in the later decades there emerged semi-professionalised groups performing traditional tales accompanied with instrumental music and singing. They are invited to weddings, funerals and similar occasions. In any kind of performance the main narrator is usually supported by a second narrator, who animates the storytelling by short interventions.

Corpus composition

The data was collected in 2019 by Olga Kuznetsova during two expeditions to the Zuenoula and Oume regions (Ivory Coast). Recordings from the Zuenoula region (over 10 hours) mostly represent group performances, and recordings from Oume region (about 2 hours) are performances in pairs (the main and a second narrators), sparsely accompanied by songs. Only parts of this collection are currently available; the corpus contains two tales from the Zuenoula region, around 45 minutes (around 6500 words) in total.

List of parts of speech

List of glosses

Citing the Guro corpus:
Kuznetsova, Olga. 2022. A narrative corpus of Guro. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Kafire

Language

Kafire is a Senufo language (Gur, Niger-Congo) spoken in Northern Côte d’Ivoire. Its speakers live in an area called ‘Kafigue’ which comprises the subprefectures of Sirasso, Nafoun and Kanoroba in the department of Korhogo. Kafire, like other Senufo languages, has noun classes / genders and it has a S(ubject)-Aux(iliary)-O(bject)-V(erb)-X (obliques, adverbials) word order. Verbs are marked for aspect (perfective vs imperfective) which is expressed through preverbal auxiliaries and on the verb as well.

Traditional narratives

In the community of Kafire speakers, i.e., the Kafibele, the narration activity is practiced by everyone (women, men, children…). It is learned during narration sessions which are held only at night (during the farm-keeping activity and around the fire in the village). Traditional narration requires at least two participants: a narrator (who narrates) and a responder (who assists him). But those roles are most of the time shifted during the session since people take turns narrating. Nowadays, the narration practice is endangered because people have changed their agricultural habits (they cultivate more products that do not require the farm-keeping activity) and children go to school (they do not gather with their elders to learn narration).

Corpus composition

This corpus is a portion of the data collected by Silué Songfolo Lacina from 2019 to 2021 in the community of Kafibele (Côte d’Ivoire, in the subprefectures of Sirasso, Nafoun and Kanoroba). The length of the total corpus is more than 10 hours. But the current annotated portion is 1h30 and it contains what the Kafibele narrate during a storytelling session, namely riddles, dilemma tales and tales. More annotated data will be uploaded progressively. In the ELAN files, morphemes are represented in two forms: the surface form (at tier mb) and the underlying form (at tier cf).

List of parts of speech

List of glosses

Citing the Kafire corpus:
Silué, Songfolo Lacina. 2022. A narrative corpus of Kafire. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Macedonian

Language

Macedonian is a South Slavic language spoken in North Macedonia, adjacent territories in Greece and Bulgaria, and diaspora communities, the largest of which are in Australia, the United States and Canada. The total number of speakers is about 2 million. Macedonian dialects are standardly divided into three major groups: Western, Northern, and Southeastern.

Storytelling tradition

Traditional storytelling is no longer practiced in modern Macedonian society, but the stories are remembered by some people. Furthermore, many Macedonian folktales have been written down and published (most notably by Marko Cepenkov). Folktales have also been dramatized and televised and some are currently available on YouTube. Archived recordings and texts can also be found at the Marko Cepenkov Institute of Folklore and at the Macedonian Academy of Arts and Sciences in Skopje. Some of this archival material is digitalised and uploaded on YouTube by Dimitrije Buzarovski.

Corpus composition

The current version of the corpus contains 26 fairytales in either audio-only or audio and video format, recorded in 2016 and 2021 over a combined period of four weeks by Izabela Jordanoska. The total length comprises about 2 hours and it contains about 15.000 words.

The majority of data in this corpus is from the West Central dialect group and mostly recorded in the capital, Skopje, with speakers from Skopje itself and from: Veles, Sekirci (a village near Prilep) and Kičevo. Some recordings were also made in Prilep with local speakers. Only one fairytale is from a different dialectal group, the Northern group (recorded in Skopje with a speaker from Kumanovo), which contains major lexical, phonological and morphosyntactic differences from the Western Central dialects. Lexical differences are marked in the glosses. Recordings were also made in Štip, but are to date not annotated.

Most speakers were in their eighties at the time of recording.

Orthography

Standard Macedonian is codified in the Cyrillic script. The corpus contains both Cyrillic transcriptions and their corresponding romanization. For interlinearized words, morphemes and glosses, only Latin script is used.

List of parts of speech

List of glosses

Citing the Macedonian corpus:
Jordanoska, Izabela. 2023. A narrative corpus of Macedonian. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Mwan

Language

Mwan, ISO 639-3 {moa}, Glottocode mwan1250 (South Mande ‹ Mande ‹ Niger Congo) is a minor language of the republic of Côte d'Ivoire (West Africa). It is spoken in a small compact area in the central part of the country (Congasso and Counahiri sub-prefectures), and also in urban centers (mainly, Abidjan, Bouaflé).

The Mwan (also called Mona by the Jula) live in about 20 villages and number more than 20.000 people (my estimation based on the data published in Ethnologue-14 which estimated the number of Mwan speakers as 17.000 people in 1993). Most of the Mwan speak Jula, those living in the cities speak French. A small number of the Mwan are monolinguals.

Typologically, Mwan displays many properties characteristic for isolating languages: tones, tendency to single metric foot words, a fuzzy border between inflection and derivation, as well as between compound words and combinations of free words, and so on. There are three level tones in Mwan. A significant part of the inflectional morphology is tonal. Mwan grammar uses mostly analytical technics, however, significant elements of the agglutination are also present. There are instances of fusion.

Mwan has no official status. It is not used at school either as a subject or as a language of instruction. It is used partially in the church service, in communication with the family and neighbors. In other areas of life, French is used (school, administration) or Jula (trade, market). In the villages, Mwan is not endangered.

Corpus composition

The present corpus of interlinearized Mwan texts is based on a collection of texts which were gathered in the period 2004-2020 by Elena Perekhvalskaya. The texts in the corpus belong to the following genres: 1) oral transcribed texts (folk tales; tales of witchcraft; oral history; dialogues); 2) texts taken from published sources (folk tales; funny stories; proverbs; translations from French). Oral texts, especially dialogues, contain a large amount of incomplete sentences, hesitation pauses, discursive markers and loans from other languages, mainly, from French and Jula.

List of parts of speech

List of glosses

Citing the Mwan corpus:
Perekhvalskaya, Elena. 2023. A narrative corpus of Mwan. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Udihe

Language

Udihe is a Tungusic language spoken in the Russian Far East. The language is critically endangered; currently there are no more than 40 native speakers.

Corpus composition

The corpus contains written data that was collected in 1936 by E.N. Baskakova from the archives from the Peter the Great Museum of Anthropology and Ethnography (Kunstkamera, Russian Academy of Sciences). There is one text recorded in 2006 by the author of the corpus. The total number of tokens of the currently available corpus is about 9K words.

List of parts of speech

List of glosses

Citing the Udihe corpus:
Perekhvalskaya, Elena. 2021. A narrative corpus of Udihe. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Ut-ma'in

Language and its speakers

U̠t‑Ma'in is a Kainji (Benue-Congo, Niger-Congo) language spoken in northwestern Nigeria (Gerhardt 1989; Blench 2018). Most speakers live in Fakai Local Government Area near the border of Kebbi State and Niger State. The principal town of U̠t-Ma'in speakers is Mahuta, located on the road that runs from Dabai, near the town of Zuru, to the town of Koko on the main road to the Kebbi State capital Birnin Kebbi. The United Nations Office for the Coordination of Humanitarian Affairs (2017) present a projected 2016 population of Fakai Local Government Area as 161,365.

U̠t-Ma’in speakers comprise a minority group among minority language groups in the area. The U̠t-Ma’in language area is bordered by Hausa ([hau], Chadic, Afroasiatic) spoken to the north and west, by C’Lela [dri] and Gwamhi-Wuri-Mba [bga] spoken to the northeast and east, by U̠t-Hun [uth] to the south, and U̠s-Saare [uss] to the southeast. With the exception of Hausa, all other neighboring languages belong to the Northwest Kainji Cluster. Most native speakers of all Northwest Kainji languages are bilingual in Hausa; Hausa is used between speakers who do not share a common first language.

A more detailed description of the language can be found here.

Language name

U̱t‑Maꞌin is a recently applied cover term for seven mutually‑intelligible varieties previously known in the literature by their individual varietal endonyms Kag-Fer-Jiir-Koor-Ror-Us-Zuksun (e.g., Ror for U̱t‑MaꞌRor), by the label Kag Cluster (Gerhardt 1989: 362-363), and by exonyms Puku-Geeri-Keri-Wipsi from the neighboring C’Lela [dri] language, as well as Fakkanci, Gelanci, etc., in the regional language Hausa [hau]. U̠t-Ma'in roughly translates as ‘our (incl.) language’.

Corpus composition

Compiled and annotated by Rebecca Paterson. Texts contributed by Ibrahim Tume Ushe, Mama Iliya, and Ibrahim Yohanna. Free translations completed in consultation with Sunday John.

Select U̠t‑Ma'in texts annotated for speech reports are taken from approximately seven and a half hours of recorded and translated data; the larger corpus contains texts in various genre including folk narratives, personal narratives, pear story retellings, and conversational data. These texts were collected during twelve months of fieldwork conducted in 2005–2007, 2013 and 2017. The annotated texts come from the Ror and Jiir varieties.

List of parts of speech

List of glosses

Citing the Ut‑Ma'in corpus:
Paterson, Rebecca, comp. 2022. A collection of U̱t-Maꞌin narrative texts annotated for reported speech. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

A narrative corpus of Wan

Language

Wan is a South Mande language spoken in central Côte d'Ivoire. It is traditionally divided into two dialects: Mian and Ken. Other languages spoken in the same area with which Wan is in contact are Mwan, Guro, Baule, and Dyula. Wan is not endangered and has a thriving storytelling tradition.

Corpus composition

The corpus contains traditional narratives from a variety of sources: audiorecordings made by Tatiana Nikitina in the village of Kounahiri and in Abidjan in the period of 2001-2016; videorecordings made by Alexandre Djamala in a number of different villages in 2019; as well as transcriptions from Philip L. Ravenhill’s archive, of recordings made in several different villages in 1973-1974 (Philip L. Ravenhill papers, National Anthropological Archives, Smithsonian Instituion).

List of parts of speech

List of glosses

Citing the Wan corpus:
Nikitina, Tatiana. 2023. A narrative corpus of Wan. In Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

Citing the corpus and tools

Corpus:

Nikitina, Tatiana, Ekaterina Aplonova, Izabela Jordanoska, Ekaterina Biteeva, Abbie Hantgan-Sonko, Guillaume Guitang, Olga Kuznetsova, Rebecca Paterson, Elena Perekhvalskaya & Lacina Silué (eds.) 2022. The SpeechReporting Corpus: Discourse Reporting in Storytelling. CNRS-LLACAN & LACITO, http://discoursereporting.huma-num.fr/index.html

Template:

Nikitina, Tatiana, Hantgan-Sonko Abbie & Chanard Christian. 2019. Reported speech annotation template for ELAN (The SpeechReporting Corpus). Villejuif-Paris: LLACAN.

Elan-CorpA:

Chanard, Christian. 2015. ELAN-CorpA: Lexicon-aided annotation in ELAN. In Amina Mettouchi, Martine Vanhove & Dominique Caubet (eds.), Corpus-based Studies of Lesser-described Languages: The CorpAfroAs corpus of spoken AfroAsiatic languages (Studies in Corpus Linguistics 68), 311–332. Amsterdam: John Benjamins Publishing
Company. https://doi.org/10.1075/scl.68.10cha.
https://benjamins.com/catalog/scl.68.10cha (2 October, 2020).