Wiktionary as CLDF
Content
- cldf1 and cldf2 contain cldf-conform data sets with a total of 2 377 756 entries about the vocabulary of all 1403 languages of the English Wiktionary.
- raw1 and raw2 together contain 1403 csv-files of 125MB size in total. File names are languages as they appear on the English Wiktionary. Each file consist of 4 columns: 'L2_orth' representing the orthographical form of the word, 'L2_ipa' its IPA-transcription, 'L2_gloss' its English explanation and 'L2_etym' its etymology iff it is borrowed from English
- lgs contains text-files with wordlists for every language that appears on the English Wiktionary. Files were created with WiktionaryParser.java.
- WiktionaryParser.java is a courtesy of Tomasz Jastrząb and was used to retrieve the wordlists found in the folder lgs
- lglist.txt is a complete list of languages that appear on the English Wiktioanry.
- lglist_full.txt is a copy of lglist.txt - since the latter serves as input for makedfs.py it can be modified according to one's needs without losing the full list.
- LICENSE: MIT
- makedfs.py - The parser with which the csv files where obtained. With a download speed of 144Mbps it needed 58 hours to parse all the languages from aari until zuni.
- makedfs.ipynb - Some notes, documenting the making-of of the parser
- parser.log - Documenting corrupted file names and handling of errors that occured while squeezing parsed data into data frames
- dfs is an empty folder into which the parser writes its results. Generated outputs were migrated to raw1 and raw2 due to Git's limitation of maximum 1000 files per directory
- changelog.txt - documenting manual deletion of false positive and insertion of false negative English loanwords
- cldf is an empty folder to which
dfs2cldf.pywrites it output. Generated output had to be migrated to folders cldf1 and cldf2 due to Github's limit of 1000 files per directory
remarks
- Sometimes the column "L2_etym" is not displayed by the csv-viewer in Github. This is likely the case whenever the first 100 lines of the column are empty. Clicking on "raw", the column can be seen again.
- The reason why columns have the "L2_" prefix is that this data was first used for baseline tests, where they served as pseudo-donor words (hence "L2" ~ second language ~ donor language), even though in the current setting they represent the recipient language (L1). The distinction L1-L2 is only internal.
Todo
- remove middle_english and old_english.csv, and generally anything with middle_ or old_ in it.
- add missing IPA transcriptions using epitran, copius_api, espeak-ng and potential other software
- Try to contribute those new IPA transcriptions to Wiktionary
570 Dec 31, 2022
29 Nov 26, 2022
37 Sep 5, 2022
1 Oct 10, 2022
2 Nov 5, 2021
1 Dec 13, 2021
1 Nov 11, 2021
0 Dec 22, 2021
42 Dec 27, 2022
11 Mar 28, 2022
2.6k Jan 08, 2023
2 Dec 27, 2022
24 Sep 11, 2022
1 Oct 14, 2021
1 Dec 22, 2021
141 Dec 30, 2022
2 Nov 13, 2021
26 May 27, 2022
342 Nov 21, 2022
2 Jun 24, 2022
63 Dec 14, 2022
456 Jan 06, 2023
2 May 17, 2022
20 Dec 13, 2022
91 Nov 29, 2022
32 Jul 29, 2022
110 Jan 07, 2023
481 Dec 30, 2022
1 Dec 10, 2021