Wiktionary data extraction errors and warnings

Inflection check

List of different kinds of inflection tables. When wiktextract parses word heads and tables, it assigns the forms it encounters with tags that describe grammatical or contextual information. The tags and forms that are found in head sections and tables are kept separate from other head section and table tags, and later they are merged with other heads and tables into table types that all contain the same number of word forms with the same tags for those forms.

The information presented here is mostly for debugging, but it can also be used to find interesting word paradigms and to hunt down mistakes, typoes and badly formated Wiktionary entries. A table type that has only a few unique instances is quite likely to contain some kind of minor error in the original data.

Language ⏶ Table forms Errors (% affected words) Language Table forms ⏷ Errors (% affected words)
Aari 1 2 (100.00%) Nederlands 8137 190 (3.08%)
Abchazisch 1 2 (100.00%) Duits 37 20 (9.82%)
Achterhoeks 1 2 (100.00%) Middelnederlands 37 22 (70.11%)
Afrikaans 1 2 (100.00%) Fries 9 4 (59.09%)
Aino 1 2 (100.00%) Engels 4 4 (98.58%)
Albanees 1 4 (100.00%) Nedersaksisch 3 4 (99.81%)
Alemannisch 1 2 (100.00%) Veluws 3 2 (97.89%)
Angelsaksisch 1 4 (100.00%) Frans 3 2 (98.63%)
Anglo-Normandisch 1 2 (100.00%) Tsjechisch 3 6 (97.95%)
Atjehs 1 2 (100.00%) Universeel taalgebruik 3 8 (100.00%)
Beiers 1 2 (100.00%) Noors 2 6 (100.00%)
Bosnisch 1 2 (100.00%) Iers 2 2 (50.00%)
Bretons 1 2 (100.00%) Latijn 2 2 (78.26%)
Centraal-Nahuatl 1 2 (100.00%) Nynorsk 2 4 (100.00%)
Cherokee 1 2 (100.00%) Deens 2 4 (100.00%)
Chinees 1 8 (100.00%) Afrikaans 1 2 (100.00%)
Cimbrisch 1 4 (100.00%) Alemannisch 1 2 (100.00%)
Corsicaans 1 2 (100.00%) Beiers 1 2 (100.00%)
Deens 2 4 (100.00%) Bosnisch 1 2 (100.00%)
Drents 1 6 (100.00%) Cimbrisch 1 4 (100.00%)
Duits 37 20 (9.82%) Esperanto 1 2 (100.00%)
Egyptisch 1 2 (100.00%) Gotisch 1 2 (100.00%)
Engels 4 4 (98.58%) Hoogsilezisch 1 2 (100.00%)
Erzja 1 2 (100.00%) Lets 1 4 (100.00%)
Esperanto 1 2 (100.00%) Middelhoogduits 1 2 (100.00%)
Estisch 1 2 (100.00%) Middelnederduits 1 2 (100.00%)
Faeröers 1 2 (100.00%) Nedersorbisch 1 4 (100.00%)
Frans 3 2 (98.63%) Noord-Fries 1 2 (100.00%)
Fries 9 4 (59.09%) Oppersorbisch 1 2 (100.00%)
Gajo 1 2 (100.00%) Oudfrans 1 2 (100.00%)
Galicisch 1 2 (100.00%) Oudnederlands 1 2 (100.00%)
Gan 1 2 (100.00%) Oudsaksisch 1 2 (100.00%)
Gotisch 1 2 (100.00%) Pennsylvania-Duits 1 2 (100.00%)
Grieks 1 2 (100.00%) Pools 1 2 (100.00%)
Gronings 1 2 (100.00%) Saterfries 1 2 (100.00%)
Hebreeuws 1 2 (100.00%) Slowaaks 1 2 (100.00%)
Hindi 1 2 (100.00%) Westfaals 1 8 (100.00%)
Hongaars 1 2 (100.00%) West-Vlaams 1 2 (100.00%)
Hoogsaksisch 1 2 (100.00%) Zweeds 1 2 (100.00%)
Hoogsilezisch 1 2 (100.00%) Achterhoeks 1 2 (100.00%)
IJslands 1 2 (100.00%) Drents 1 6 (100.00%)
Iers 2 2 (50.00%) Luxemburgs 1 2 (100.00%)
Indonesisch 1 2 (100.00%) Oost-Fries 1 2 (100.00%)
Inuktitut 1 2 (100.00%) Riograndenser Hunsrückisch 1 4 (100.00%)
Inupiak 1 2 (100.00%) Sallands 1 2 (100.00%)
Italiaans 1 2 (100.00%) Schots 1 2 (100.00%)
Japans 1 4 (100.00%) Stellingwerfs 1 6 (100.00%)
Javaans 1 2 (100.00%) Twents 1 6 (100.00%)
Jiddisch 1 2 (100.00%) Urkers 1 2 (100.00%)
Kantonees 1 2 (100.00%) Gronings 1 2 (100.00%)
Kasjmiri 1 2 (100.00%) Limburgs 1 8 (100.00%)
Kasjoebisch 1 2 (100.00%) Roemeens 1 2 (100.00%)
Koerdisch 1 2 (100.00%) Spaans 1 2 (100.00%)
Koreaans 1 2 (100.00%) Vietnamees 1 2 (100.00%)
Ladino 1 2 (100.00%) Xhosa 1 2 (100.00%)
Latijn 2 2 (78.26%) Middelengels 1 14 (100.00%)
Lets 1 4 (100.00%) Oudfries 1 2 (100.00%)
Limburgs 1 8 (100.00%) Zeeuws 1 2 (100.00%)
Lombardisch 1 2 (100.00%) Papiaments 1 2 (100.00%)
Luxemburgs 1 2 (100.00%) Surinaams 1 2 (100.00%)
Maleis 1 2 (100.00%) Javaans 1 2 (100.00%)
Mandarijn 1 2 (100.00%) Servisch 1 2 (100.00%)
Middelengels 1 14 (100.00%) Servo-Kroatisch 1 2 (100.00%)
Middelhoogduits 1 2 (100.00%) Faeröers 1 2 (100.00%)
Middelnederduits 1 2 (100.00%) Estisch 1 2 (100.00%)
Middelnederlands 37 22 (70.11%) IJslands 1 2 (100.00%)
Mongools 1 2 (100.00%) Hongaars 1 2 (100.00%)
Montenegrijns 1 2 (100.00%) Ripuarisch 1 2 (100.00%)
Myceens 1 2 (100.00%) Indonesisch 1 2 (100.00%)
Nederlands 8137 190 (3.08%) Galicisch 1 2 (100.00%)
Nedersaksisch 3 4 (99.81%) Portugees 1 2 (100.00%)
Nedersorbisch 1 4 (100.00%) Yola 1 2 (100.00%)
Negerhollands 1 2 (100.00%) Italiaans 1 2 (100.00%)
Noord-Fries 1 2 (100.00%) Bretons 1 2 (100.00%)
Noors 2 6 (100.00%) Welsh 1 2 (100.00%)
Nyanja 1 2 (100.00%) Reto-Romaans 1 2 (100.00%)
Nynorsk 2 4 (100.00%) Centraal-Nahuatl 1 2 (100.00%)
Oost-Fries 1 2 (100.00%) Corsicaans 1 2 (100.00%)
Oppersorbisch 1 2 (100.00%) klassiek Nahuatl 1 2 (100.00%)
Oudfrans 1 2 (100.00%) Siciliaans 1 2 (100.00%)
Oudfries 1 2 (100.00%) Albanees 1 4 (100.00%)
Oudgrieks 1 4 (100.00%) Maleis 1 2 (100.00%)
Oudhoogduits 1 2 (100.00%) Swahili 1 2 (100.00%)
Because subject class concord and object class concord uses the same form (for example "c5"), we don't have yet a way to distinguish them. The meaning is human-parsable because we can look at the alignment of the headers (horizontal=subject concord, vertical=object concord), but the parser does not (yet) have this kind of memory or spatial awareness. /////// Sep 9th 2022: I've recently implemented a way to transform row or column tags into other tags based on language, but it doesn't solve an underlying issue with Swahili tables that I'd missed or forgotten about: The horizontal column headers for class and person don't get inherited through the whole template because it is chopped up into several subtables that don't directly inherit from above. It's an issue with 'bleed' again. /////// Later (comment in Dec 2022): forgot to put this here, but Swahili tables have now gotten their own big systems that make them work, including a "save to register" feature that can keep column headers in memory to be used in a later subtable within a template.
Oudnederlands 1 2 (100.00%) Anglo-Normandisch 1 2 (100.00%)
Oudnoords 1 2 (100.00%) Atjehs 1 2 (100.00%)
Oudsaksisch 1 2 (100.00%) Oudhoogduits 1 2 (100.00%)
Oudtsjechisch 1 2 (100.00%) Angelsaksisch 1 4 (100.00%)
Paltsisch 1 4 (100.00%) Koerdisch 1 2 (100.00%)
Papiaments 1 2 (100.00%) Ladino 1 2 (100.00%)
Pennsylvania-Duits 1 2 (100.00%) Inupiak 1 2 (100.00%)
Perzisch 1 2 (100.00%) Cherokee 1 2 (100.00%)
Pools 1 2 (100.00%) Perzisch 1 2 (100.00%)
Portugees 1 2 (100.00%) Chinees 1 8 (100.00%)
Reto-Romaans 1 2 (100.00%) Kantonees 1 2 (100.00%)
Riograndenser Hunsrückisch 1 4 (100.00%) Mandarijn 1 2 (100.00%)
Ripuarisch 1 2 (100.00%) Hebreeuws 1 2 (100.00%)
Roemeens 1 2 (100.00%) Russisch 1 2 (100.00%)
Russisch 1 2 (100.00%) Thai 1 2 (100.00%)
Sallands 1 2 (100.00%) Hindi 1 2 (100.00%)
Saterfries 1 2 (100.00%) Mongools 1 2 (100.00%)
Schots 1 2 (100.00%) Koreaans 1 2 (100.00%)
Schots-Gaelisch 1 2 (100.00%) Grieks 1 2 (100.00%)
Servisch 1 2 (100.00%) Jiddisch 1 2 (100.00%)
Servo-Kroatisch 1 2 (100.00%) Japans 1 4 (100.00%)
Siciliaans 1 2 (100.00%) Lombardisch 1 2 (100.00%)
Silezisch Duits 1 2 (100.00%) Abchazisch 1 2 (100.00%)
Slowaaks 1 2 (100.00%) Erzja 1 2 (100.00%)
Spaans 1 2 (100.00%) Kasjoebisch 1 2 (100.00%)
Stellingwerfs 1 6 (100.00%) Paltsisch 1 4 (100.00%)
Surinaams 1 2 (100.00%) Nyanja 1 2 (100.00%)
Swahili 1 2 (100.00%)
Because subject class concord and object class concord uses the same form (for example "c5"), we don't have yet a way to distinguish them. The meaning is human-parsable because we can look at the alignment of the headers (horizontal=subject concord, vertical=object concord), but the parser does not (yet) have this kind of memory or spatial awareness. /////// Sep 9th 2022: I've recently implemented a way to transform row or column tags into other tags based on language, but it doesn't solve an underlying issue with Swahili tables that I'd missed or forgotten about: The horizontal column headers for class and person don't get inherited through the whole template because it is chopped up into several subtables that don't directly inherit from above. It's an issue with 'bleed' again. /////// Later (comment in Dec 2022): forgot to put this here, but Swahili tables have now gotten their own big systems that make them work, including a "save to register" feature that can keep column headers in memory to be used in a later subtable within a template.
Inuktitut 1 2 (100.00%)
Tamazight 1 2 (100.00%) Wymysoojs 1 2 (100.00%)
Thai 1 2 (100.00%) Oudgrieks 1 4 (100.00%)
Tsjechisch 3 6 (97.95%) Yucateeks 1 2 (100.00%)
Twents 1 6 (100.00%) Schots-Gaelisch 1 2 (100.00%)
Universeel taalgebruik 3 8 (100.00%) Gan 1 2 (100.00%)
Urkers 1 2 (100.00%) Oudnoords 1 2 (100.00%)
Veluws 3 2 (97.89%) Aino 1 2 (100.00%)
Vietnamees 1 2 (100.00%) Oudtsjechisch 1 2 (100.00%)
Walserduits 1 2 (100.00%) Egyptisch 1 2 (100.00%)
Welsh 1 2 (100.00%) Myceens 1 2 (100.00%)
West-Vlaams 1 2 (100.00%) Gajo 1 2 (100.00%)
Westfaals 1 8 (100.00%) Montenegrijns 1 2 (100.00%)
Wymysoojs 1 2 (100.00%) Kasjmiri 1 2 (100.00%)
Xhosa 1 2 (100.00%) Aari 1 2 (100.00%)
Yola 1 2 (100.00%) Tamazight 1 2 (100.00%)
Yucateeks 1 2 (100.00%) Hoogsaksisch 1 2 (100.00%)
Zeeuws 1 2 (100.00%) Negerhollands 1 2 (100.00%)
Zweeds 1 2 (100.00%) Walserduits 1 2 (100.00%)
klassiek Nahuatl 1 2 (100.00%) Silezisch Duits 1 2 (100.00%)

This page is a part of the kaikki.org machine-readable dictionary. This dictionary is based on structured data extracted on 2025-03-06 from the nlwiktionary dump dated 2025-03-02 using wiktextract (b81b832 and 633533e). The data shown on this site has been post-processed and various details (e.g., extra categories) removed, some information disambiguated, and additional data merged from other sources. See the raw data download page for the unprocessed wiktextract data.

If you use this data in academic research, please cite Tatu Ylonen: Wiktextract: Wiktionary as Machine-Readable Structured Data, Proceedings of the 13th Conference on Language Resources and Evaluation (LREC), pp. 1317-1325, Marseille, 20-25 June 2022. Linking to the relevant page(s) under https://kaikki.org would also be greatly appreciated.