User talk:GrounderUK

Re: reverted edit

Oh gosh, yes, it was absolutely by mistake. I don't know why it removed also your comment, I was only removing the repetition of the newsletter. I'll put your comment back immediately. I'm sorry for this mishap. Sannita (WMF) (talk) 20:28, 10 December 2023 (UTC)

Not a problem. Thank you for putting this right. GrounderUK (talk) 22:43, 10 December 2023 (UTC)

Comment on forms

Hi,

I don't want to flood Wikifunctions:Type proposals/Wikidata based types but not all lexemes have strictly one singular form (it could be 0 for plurale tantum like "clothes" and more than one for spelling variants like "organisation"/"organization", 12 being the highest number in English with d:L:L1014211 right now ; and I'm not even talking of languages where nouns don't follow the singular/plural paradigm). F1 is often the singular but not always (for instance, some people may have entered the plural first in F1 and the singular afterward in F2). Finally, sometimes people put several representations in the same form but that's a bad practice (because you can't add statements or link to a specific representation).

Cheers, VIGNERON (talk) 13:31, 11 July 2024 (UTC)

Thanks. Just to be clear, though, when I said “singular lemma” I was emphasising that “lemma” is singular, not that the form of a lexeme used as the lemma is a singular form. I’ve always thought that there is zero value in having a lemma when you have a unique identifier; now I think the value is probably negative! But the same applies to forms and representations: the plural of “colour” is “colours” and the plural of “color” is “colors”. It is not the case that the plural of color/colour is one of [“colors”, “colours”]. Or, to use your example, “organize” is perfectly good British English (and would be the lemma in an Oxford dictionary) and so is “organise”, but the present participle for “organize” (the form/representation) is always “organizing”. Or (harder case), the present participle of “focus” may be either “focusing” or “focussing” and the past tense/participle is “focused” or “focussed”, according to which variant of the present participle you prefer (and/or vice versa). GrounderUK (talk) 14:31, 11 July 2024 (UTC)

Thanks also.

Sadly, I'm not sure to exactly understand what you mean. What do you mean by « zero value in having a lemma when you have a unique identifier » and AFAIK, no one is saying « the plural of color/colour is one of [“colors”, “colours”] » (and certainly not d:L:L1347).

Cheers, VIGNERON (talk) 07:43, 12 July 2024 (UTC)

A lemma is a particular form of a lexeme chosen as its identifier. If you already have an identifier (and, I should have said, it is easy to access a lexeme by any of its forms) the lemma has no function. That is what I meant, but I’m happy to leave it there because we do, in fact, have “lemmas” and they will not, in practice, have the function of a lemma on Wikifunctions.

I don’t agree with you about L1347, however. It presents as colour/colour/color (“lemma”) with two forms. L1347-F2 is “the” plural form with literals “colours”, “colours” and “colors”. Of course, these have different language codes, so we can infer the correct plural if we happen to know the langage variant of the singular. And we can assume that a singular in a different or unknown variant will have the same plural as a known variant. But that is not what the lexeme actually says. All it says is that the plural is either “colours” or “colors”.

That’s too complicated. Sorry. “Colour” is a perfectly regular English noun and so is “color”. Each therefore has four forms. It is reasonable to consider them as a single lexeme, but then we need to recognise that the eight distinct forms divide into two distinct sets of four forms each. Language codes will not always be a convenient way to achieve this. See, for example, Dived vs. Dove: Usage Guide.

But, hey! That’s why we need Wikifunctions! 😎 GrounderUK (talk) 09:18, 12 July 2024 (UTC)

Ah yes, I think I see now, thanks. Lemma is maybe not the most useful but it has a few useful function, it's not just one form, it's the main form (the one we look up most of the time, it's how dictionaries works for centuries). Thus, it can make the search easier and faster (right now, there is ~1.4 million lemmas vs. 15 million forms ; and these numbers will get a lot bigger with time, especially the forms). It's not much but it's expected (just like labels for items).

The plural is not either “colours” or “colors”, it's “colours”@en-gb and “colors”@en-us ; languages code are not always enough but here they are. That said, as I said, I don't like this modelling grouping representations in one form, I prefer to split them like on d:L:L2127 (it's doesn't change much fundamentally, but it allows to put explicit statements in addition). Because in Breton, this is a case where languages code are not enough.

Anyway, I'm eager to see the lexemes working in Wikifunction and what we will be able to do (and it might impact on how we do store data in Lexemes but the model seems mostly sound to me).

Cheers, VIGNERON (talk) 18:51, 12 July 2024 (UTC)

Yes, the Breton example makes much more sense. Although I take your point about searching, I believe the more logical approach is to recognise that not all representations need to be searchable and then define which ones are. The type proposal does not suggest how we (or functions) can get from a string of some kind to a corresponding Wikidata lexeme, and that is an “interesting problem” for the future! GrounderUK (talk) 08:57, 13 July 2024 (UTC)