ELTE Folk Song Corpus is a database developed by the Department of Digital Humanities at Eötvös Loránd University. Currently, the corpus contains 2390 Hungarian folk songs, the sound devices of the texts and the grammatical features of words in XML format (in TEI and non-TEI XML format).
- number of folk songs: 2390
- number of words: 113 565
- number of tokens: 148 524
The source of the corpus was the digital version of the book Magyar népdalok (Hungarian Folk Songs) edited and selected by Gyula Ortutay and Imre Katona, published in 1976. The HTML files of the book were downloaded from the Hungarian Electronic Library.
- The HTML files from the Hungarian Electronic Library were converted into TEI XML format based on the Text Encoding Initiative.
- 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.
- 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).
- 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).
<head>: title<lg>: stanza<l>: line<p>: subtitle, epigraph, separator, editorial note
<w>: word<pc>: punctuation mark@lemma: lemma@pos: part of speech@msd: morphosyntactic features (Universal Dependencies)
@rhyme: rhyme pattern@real: rhythm (0: short syllable; 1: long syllable)<spanGrp type="phonStructures">: standoff annotation of the phonological features of words<span>: standoff annotation of the phonological features of a word content of<span>: phonological representation of the wordc: consonantb: short back vowelB: long back vowelf: short front vowelF: long front vowel
@subtype: syllable number@type: type of vowelslow: only back vowels in the wordhigh: only front vowels in the wordmixed: front and back vowels in the word
@target: thexml:idof the annotated word<linkGrp type="rhymePairs">: standoff annotation of rhyme pairs<link>: standoff annotation of a rhyme pair@target: xml:id of the two words in a rhyme pair<spanGrp type="alliterations">: standoff annotation of alliterations<span>: standoff annotation of an alliteration@target:xml:idof the words in the alliteration@type: structure of the alliterationa: alliterating wordn: non-alliterating word (only one non-alliterating word can be between two alliterating words)
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.
@div_numStanza: number of stanzas in the poem@div_numLine: number of lines in the poem@div_numWord: number of words in the poem@div_numSyll: number of syllables in the poem@div_numShortSyll: number of short syllables in the poem@div_numLongSyll: number of long syllables in the poem@div_rhyme: the rhyme pattern of the poem@div_syllPattern: syllable numbers of lines in the poem@lg_numLine: number of lines in the stanza@lg_numWord: number of words in the stanza@lg_numSyll: number of syllables in the stanza@lg_numShortSyll: number of short syllables in the stanza@lg_numLongSyll: number of long syllables in the stanza@lg_syllPattern: syllable numbers of lines in the stanza@l_numWord: number of words in the line@l_numSyll: number of syllables in the line@l_numShortSyll: number of short syllables in the line@l_numLongSyll: number of long syllables in the line@w_numSyll: syllable number of word (conversion of level3's@subtypein<span>)@phonType: type of vowels in the word (conversion of level3's@typein<span>)@phonStruct: phonological representation of the word (conversion of level3's<span>content)<rhymePairs>: standoff annotation of rhyme pairs (conversion of level3's<linkGrp type="rhymePairs">)<rhymePair>: standoff annotation of a rhyme pair (conversion of level3's<link>)<firstRhyme>,<secondRhyme>: standoff annotation of the first and second word of a rhyme pair- content of
<firstRhyme>and<secondRhyme>: the rhyming word form
- content of
@rhyme_lemma: the lemma of the rhyming word@rhyme_pos: the part of speech of the rhyming word@rhyme_msd: the morphosyntactic features of the rhyming word (Universal Dependencies)@rhyme_numSyll: the syllable number of the rhyming word@rhyme_phonType: the type of vowels in the rhyming word@rhyme_phonStruct: the phonological representation of the rhyming word<alliterations>: standoff annotation of alliterations (conversion of level3's<spanGrp type="alliterations">)<alliteration>: standoff annotation of an alliteration (conversion of level3's<span>)- content of
<alliteration>: the alliterating word forms
- content of
@allStruct: structure of the alliteration (conversion of level3's@typein<span>)@posTags: the parts of speech of the alliterating words@msdTags: the morphosyntactic features of the alliterating words@lemmas: the lemmas of the alliterating words
- Péter Horváth (design, annotation scripts)
- Gábor Palkó (design, data model)
If you use ELTE Folksong Corpus, please cite the following article:
Horváth Péter – Kundráth Péter – Palkó Gábor 2022. ELTE Népdalkorpusz – magyar népdalok gépileg annotált adatbázisa. In: Tick József – Kokas Károly – Holl András (szerk.): Valós térben – Az online térért: Networkshop 31: országos konferencia. Budapest: HUNGARNET Egyesület. 276-283.
The content of the repository is licensed under the CC BY-NC-ND license.
All texts of the corpus are in the public domain.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent