Home

Awesome

ELTE Poetry Corpus

ELTE Poetry Corpus is a continuously expanding database developed by the Department of Digital Humanities at Eötvös Loránd University. Currently, the corpus contains the complete poems of 52 Hungarian canonical poets, the sound devices of the poems and the grammatical features of words in XML format (in TEI and non-TEI XML format).

Size

For more information of the size of subcorpora and the authors' year of birth and death, see the files subcorpus_sizes.tsv and poets_birth_and_death.tsv

TEI Levels

The source of the corpus was the collection of the Hungarian Electronic Library, which contains numerous poetic oeuvres in digitized form.

  1. The texts from the Hungarian Electronic Library were converted into TEI XML format based on the Text Encoding Initiative.
  2. The automatically converted poems containing the annotations of structural units were checked manually (level1).
  3. Then, we tokenized the poems and annotated the grammatical features of words by using e-magyar, an NLP tool chain for Hungarian texts. The level2 folder contains the TEI XML files in which the morphosyntactic features (values of the msd attributes) are annotated in the format of universal dependencies, while the level2_emMorph folder contains the same files in which the morphosyntactic features are annotated in its own, emMorph format of e-magyar.
  4. After the grammatical annotation, we also annotated the rhyme patterns, the rhyme pairs, the rhythm of lines, the alliterations and the phonological features of words (level3).
  5. Finally, we added further annotations of poetic features to the corpus and changed the name and the position of some elements and attributes, using a non-TEI XML format defined for the project (level4).

poem_texts

The poem_texts folder contains the poems in TXT format, without the XML annotations. This version of the corpus was generated from the level1 files. The TXT files contain the editorial notes related to date and place, which are in <p> elements in the TEI versions.

Elements and attributes

Level1 -- annotation of structural units

Level2 -- tokenization and annotation of grammatical features of words

Level3 -- annotation of sound devices

Level4 -- conversion of the TEI format into non-TEI format

By changing the name and the position of certain elements and attributes in level3 and by adding further annotations to the corpus, it is easier to process but cannot be expressed in valid TEI XML format.

Contributors

Citing and License

If you use ELTE Poetry Corpus, please cite one of the following articles:

Horváth Péter – Kundráth Péter – Indig Balázs – Fellegi Zsófia – Szlávich Eszter – Bajzát Tímea Borbála – Sárközi-Lindner Zsófia – Vida Bence – Karabulut Aslihan – Timári Mária – Palkó Gábor 2022. ELTE Verskorpusz – a magyar kanonikus költészet gépileg annotált adatbázisa. In: Berend Gábor – Gosztolya Gábor – Vincze Veronika (szerk.): XVIII. Magyar Számítógépes Nyelvészeti Konferencia. Szeged: Szegedi Tudományegyetem TTIK, Informatikai Intézet. 375–388.

Horváth, Péter – Kundráth, Péter – Indig, Balázs – Fellegi, Zsófia – Szlávich, Eszter – Bajzát, Tímea Borbála – Sárközi-Lindner, Zsófia – Vida, Bence – Karabulut, Aslihan – Timári, Mária – Palkó, Gábor 2022. ELTE Poetry Corpus: A Machine Annotated Database of Canonical Hungarian Poetry. In: Calzolari, Nicoletta – Béchet, Frédéric – Blache, Philippe – Choukri, Khalid – Cieri, Christopher – Declerck, Thierry – Goggi, Sara – Isahara, Hitoshi – Maegaard, Bente – Mariani, Joseph – Mazo, Hélène – Odijk, Jan – Piperidis, Stelios (eds.): Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022). Paris: European Language Resources Association (ELRA). 3471–3478.

The content of the repository is licensed under the CC BY-NC-ND license.

All texts of the corpus are in the public domain.