Closed LinguList closed 4 years ago
Not all are real problems, but to narrow this down, we need to start by:
marking all language glosses by "gloss" in one of the main fields, e.g., "other", then I can exclude them from the text (right now, I have no apparent way of telling from the column description in norare, what "english" refers to, but we need to know that it is a gloss for a given language here.
marking ID columns (nel_id, stedt_id) as "identifier", so we can exlude them as well.
resolving the other parts (lgc_sd, for example, is ambiguous with some other aspect there, etc.).
You have to git-pull the most recent version, @AnnikaTjuka, from pynorare, to be able to use the norare check
command. This is a first step towards having more testing.
Not sure if I understand the check correctly. Does it use the datasets as a basis or norare.tsv? For example, in norare.tsv we do not include "SD" columns.
It uses norare.tsv.
Wait, this may be different for concepticon concept lists. I'll check.
My error, I did check for all columns, will try and correct this now.
Perfect, now it is fixed:
$ norare check
INFO norare at /home/mattis/data/datasets/cldf/concepticon
INFO checking Bond-2013-OMW
INFO checking Alonso-2015-AoA
INFO checking Brysbaert-2009-Frequency
INFO checking Brysbaert-2011-Frequency
INFO checking Brysbaert-2014-Concreteness
INFO checking Brysbaert-2019-Prevalence
INFO checking Cai-2010-Frequency
INFO checking Cuetos-2011-Frequency
INFO checking Desrochers-2009-SubjFrequency
INFO checking Engelthaler-2018-Humor
INFO checking Juhasz-2013-SER
INFO checking Keuleers-2010-Frequency
INFO checking Kuperman-2012-AoA
INFO checking Riegel-2015-AffectiveRatings
INFO checking Scott-2019-Ratings
INFO checking StadthagenGonzalez-2017-ValenceArousal
INFO checking Starostin-2000-Sense
INFO checking Warriner-2013-AffectiveRatings
INFO checking Cortese-2008-AoA
INFO checking Keuleers-2012-LexicalDecision
INFO checking Ferrand-2010-LexicalDecision
INFO checking GonzalezNosti-2014-LexicalDecision
INFO checking Tsang-2018-LexicalDecision
INFO checking Keuleers-2015-Prevalence
INFO checking StadthagenGonzalez-2018-DiscreteEmotions
INFO checking Alonso-2016-AoA
INFO checking Imbir-2016-Ratings
INFO checking Ferre-2017-DiscreteEmotions
INFO checking Wierzba-2015-DiscreteEmotions
INFO checking Alonso-2011-OralFrequency
INFO checking Lynott-2019-Sensorimotor
INFO checking Kapucu-2018-EmotionRatings
INFO checking Briesemeister-2011-DiscreteEmotions
INFO checking Mandera-2015-Frequency
INFO checking Moors-2013-Ratings
INFO checking Wu-2020-CoreVocabulary
INFO checking Mohammad-2018-AffectiveRatings
INFO checking Mohammad-2018-EmotionIntensity
INFO checking Clark-2004-ImageryFamiliarity
INFO checking Abdaoui-2017-EmoLex
INFO checking Matisoff-2015-STEDT
INFO checking Kiss-1973-EAT
INFO checking Wikidata
INFO checking Webster
INFO checking OmegaWiki
INFO checking LEGO
INFO checking Babelnet
INFO checking Numerals
INFO checking Pagel-2007-200
INFO checking Winter-2016-300
INFO checking DiezAlamo-2018-750
INFO checking Hill-2015-999
INFO checking Izura-2005-499
INFO checking Yao-2017-1100
INFO checking Rzymski-2020-1624
INFO checking Verheyen-2019-1000
INFO checking Maciejewski-2016-100
INFO checking Baroni-2011-200
INFO checking Dellert-2018-1016
INFO checking Luniewska-2019-299
WARNING non-unique value en-ratings-AoA-mean- in Luniewska-2019-299 / english_aoa
WARNING non-unique value en-ratings-AoA-mean- in Luniewska-2019-299 / southafricanenglish_aoa
INFO checking Lewis-2016-499
INFO checking Schroeder-2012-824
INFO checking Lynott-2009-423
INFO checking Pagel-2018-200
INFO checking Desrochers-2010-330
INFO checking Monnier-2014-1031
INFO checking Gampe-2017-48
INFO checking Lynott-2013-400
INFO checking Xiao-2012-213
INFO checking Luniewska-2016-299
WARNING non-unique value en-ratings-AoA-mean- in Luniewska-2016-299 / southafricanenglish_aoa
Yes, this looks much better :)
Concerning the Warnings in the Luniewska data set. I couldn't find an ISO gloss for southafricanenglish. Should I just shorten it to "seen" or "ae"?
No, I suggest you add "South-African English" to "other". Since I use the two-letter-codes to create the flags in the app...
Ok!
I just added a check, it spits out the following problems: