Lang*Reg: A multi-lingual corpus of intra-speaker variation across situations

The Lang*Reg corpus records intra-speaker variation across languages and different situational-functional contexts, presumed to result in different registers. It has been prepared in the SFB1412 Register with data collections taking place in 2021-2022 for the following languages included in this version: German, Persian, Kurdish. The data sets for each language comprise the speech of the same language users in a variety of spoken conversations and one written interaction. A minimum of 12 participants per language traversed a course of 6 situations in which they were asked to produce language in three types of activities: telling a story to a friend, talking freely with various interlocutors (friend, stranger, taxi driver) and engaging in an interview with a (university) professor. Moreover, our design included the storytelling in two modes, which allows for the comparison between spoken and written modes of the same language user. 

Lang*Reg has a basic syntactic segmentation (one matrix clause and all its dependent clauses per segment). v0.1.0 includes the data sets with transcriptions, normalizations and tokens for each language as well as additional language-specific annotations. We prepared each data set also for use with the browser-based search and visualization architecture ANNIS. For further language-specific morpho-syntactic and sociolinguistic annotations, refer to the respective data set description. For an overview of all data set characteristics, please see the corpus documentation in each data set.

Citation:

Please cite the whole corpus as follows:

Adli, Aria, Verhoeven, Elisabeth, Lehmann, Nico, Mortezapour, Vahid & Vander Klok, Jozina (eds). 2023. Lang*Reg: A multi-lingual corpus of intra-speaker variation across situations (v1.0.0). [Data set]. Berlin, Köln: Humboldt-Universität zu Berlin, Universität zu Köln. DOI: 10.5281/zenodo.7646320

For individual data sets, please cite as indicated in the description of each data set.

German:

The German Lang*Reg data set includes ELAN files with transcriptions, normalizations and tokens as well as CONLLU files with tokens, POS tags and dependency annotations. An ANNIS version with all annotations is also included.

Cite as: Lehmann, Nico, Verhoeven, Elisabeth, Müller, David, Andreas, Sina, Anlauf, Luka, Deichsler, Florian, Renard, Victor & van Stek, Zacharias. 2023. German Lang*Reg data set. In Adli, Aria, Verhoeven, Elisabeth, Lehmann, Nico , Mortezapour, Vahid & Vander Klok, Jozina (eds). Lang*Reg: A multi-lingual corpus of intra-speaker variation across situations (v0.1.0). Berlin, Köln: Humboldt-Universität zu Berlin, Universität zu Köln.

Persian:

The Persian Lang*Reg data set includes ELAN files with transcriptions, normalizations and tokens as well as annotations for clause type, constituent type, subject type and subject characteristics. An ANNIS version with all annotations is included.

Cite as: Mortezapour, Vahid, Adli, Aria, Sameri, Motahareh, Ilghami, Negar & Taherkhani Mehrnoosh. 2023. Persian Lang*Reg data set. In Adli, Aria, Verhoeven, Elisabeth, Lehmann, Nico , Mortezapour, Vahid & Vander Klok, Jozina (eds). Lang*Reg: A multi-lingual corpus of intra-speaker variation across situations (v0.1.0). Berlin, Köln: Humboldt-Universität zu Berlin, Universität zu Köln.

Kurdish:

The Kurdish Lang*Reg data set includes ELAN files with transcriptions, normalizations and tokens as well as annotations for code switching. An ANNIS version with all annotations is included.

Cite as: Farokhnejad, Zahra. 2023. Kurdish Lang*Reg data set. In Adli, Aria, Verhoeven, Elisabeth, Lehmann, Nico, Mortezapour, Vahid & Vander Klok, Jozina (eds). Lang*Reg: A multi-lingual corpus of intra-speaker variation across situations (v0.1.0). Berlin, Köln: Humboldt-Universität zu Berlin, Universität zu Köln.