Wikidata:Property proposal/Alternative form

alternative form

Originally proposed at Wikidata:Property proposal/Lexemes

   Done: alternative form (P8530) (Talk and documentation)

Description	Link to spelling variant between two or more forms
Data type	Form
Domain	Form
Example 1	højtaler/højttaler (L229388)-F1 → L229388-F2 as per [1]
Example 2	L229388-F2 → L229388-F1
Example 3	hilsen (L307241)-F3 → L307241-F5
Example 4	L307241-F5 → L307241-F3

Begrundelse

In Danish (Q9035), certain lexemes may have multiple different spellings with the same grammatical features. For instance, at least in Den Danske Ordbog (Q1186741) and Retskrivningsordbogen (Q3398246), the Danish word for loudspeaker may be spelt with one or two t's: højtaler/højttaler. In Retskrivningsordbogen (Q3398246), the alternative forms may be indicated with a parenthesis, e.g., hils(e)ner for hilsener or hilsner. It is not possible to use the "spelling variant" feature in the Wikidata lexeme datastructure because the language code is the same (here "da"). Thus what I have done is to put up multiple forms for the alternative spellings. The spelling variant could, I suppose, be identified by looking at forms with the same grammatical features. A more direct way would be to explicitly state the link with a property. Instead of "alternaltive form" it could also be called "spelling variant" (note there are a rejected Wikidata:Property_proposal/spelling variant proposal for another issue). Finn Årup Nielsen (fnielsen) (talk) 15:10, 31 July 2020 (UTC)

Discussion

• Yes, I've found this a bit annoying in the past. English has a lot of words with alternate noun pluralizations, where I've just added both as forms. Is a property necessary for linking them though, or do the common grammatical features imply directly they are alternatives? ArthurPSmith (talk) 17:08, 31 July 2020 (UTC)

  Common grammatical features should imply that they are alternatives, as far as I can tell. However, a SPARQL query that query on same grammatical feature to identify alternative form(s) will need to be somewhat complicated, — also as far as I can tell. For instance, the query below is an inadequate attempt

  SELECT ?form (COUNT(?grammaticalFeature) AS ?grammaticalFeaturesCount) {
    ?lexeme ontolex:lexicalForm wd:L307241-F5, ?form .
    wd:L307241-F5 wikibase:grammaticalFeature ?grammaticalFeature .
    ?form wikibase:grammaticalFeature ?grammaticalFeature .
  }
  GROUP BY ?form
  HAVING (?grammaticalFeaturesCount = 2)
  

  Try it!

  — Finn Årup Nielsen (fnielsen) (talk) 13:35, 1 August 2020 (UTC)

•  Support This is a good idea for alternative spellings that are guaranteed to have identical grammatical properties. However, I think it would be confusing to speak of forms of forms. Perhaps we should call them written variants, reminiscent to how CJKV characters have national written variants?  – Minh Nguyễn ^💬 18:02, 3 August 2020 (UTC)
•  Comment In Wikidata talk:Lexicographical data#Abbreviations, I also raised the point that we may need to qualify this statement with determination method (P459), because abbreviations can vary not only by dialect but also by context.  – Minh Nguyễn ^💬 18:02, 3 August 2020 (UTC)
•  Support I don't really see how could make statements about forms when they are all listed as the same Form (note the "F"). --- Jura 21:11, 10 August 2020 (UTC)

@Fnielsen, ArthurPSmith, Mxn, Jura1:  Done alternative form (P8530) Pamputt (talk) 15:47, 16 August 2020 (UTC)

• I'm aware this sail has already shipped, but I would like to point out than an alternative way to specifying these alternative forms, would be to use a "private" language code, such as "da-x-Q59342809" (using for instance alternative form (Q59342809), but other qualifications are of course possible). This has the advantage that then you can simply list the alternative spelling as a spelling variant in the same form, and also at the lexeme entry. For the sake of the demonstration, I have done this at the header of højtaler/højttaler (L229388). AGutman-WMF (talk) 09:22, 22 June 2022 (UTC)