User talk:Mathglot
Latest comment: 6 months ago by Mathglot in topic Removals of French Terms
- If I left you a message: please answer on your talk page, linking to me (
{{Ping|Mathglot}}
), so I will be notified. - If you leave me a message: I will answer on my talk page, linking to your username, so you will be notified.
- Please click here to leave me a new message.
Hi Mathglot. I recommend that you move that discussion to the WT:Tea room; not many people pay attention to the talk pages of individual entries. — Ungoliant (falai) 13:44, 2 May 2017 (UTC)
Removals of French Terms
[edit]Please read WT:CFI. Wiktionary is a descriptive dictionary based on usage, not on authoritative sources. We have a process for challenging whether words should be included, but a quick Google Books search on "asinin" (along with "les" to filter out non-French texts) seems to show enough French usage to pass. Chuck Entz (talk) 19:39, 3 May 2024 (UTC)
- @Chuck Entz:, I take your point, and I can see that some of these might pass the threshold for appearance here. The statement under CFI shortcutted as WT:ATTEST seems resaonable, and if followed would probably eliminate a lot of problems. But in the case of words identified as foreign, shouldn't evaluations of ATTEST depend on people who actually either understand the language, or are willing to go through the sources? Just finding hits on Google seems a very weak threshold, as who knows what kind of content you might be hitting.
- Maybe a more complex query would do better than the one you proposed at weeding out a lot of the chaff, but let's just go with that one for now. Searching in Books (not web, because of self-published sources and all sorts of garbage) gives this search: asinin les, and what I see there is books in French talking about Algonquian language in most cases. It appears that asinin is the French phonetic spelling of some word in Algonquian. See for example, Mémoires Et Comptes Rendus de la Société Royale Du Canada:, page 117 near the bottom, Chapter 19, Syntax, #346. There seems to be an overrepresentation of Algonquian in a lot of the books listed under that query.
- I'm not saying this holds for every case, but it seems pretty cavalier to include a word based on this kind of evidence, and it gives a word a kind of stamp of approval of appearing in Wiktionary. I'm not that active here, but if this appeared in Wikipedia, I'd send it to AFD straightaway. It does make me wonder how many words here are cases like this one. It just seems like it waters down the whole feeling of reliability of Wiktionary if bots can run around adding words like this found in books identified (correctly, in this case) as "French" by Google, but where the passage from which the word is extracted may have been Algonquian (or any of a few hundred other language-learning books for native-French language learners. If the bot were clever enough to extract a meaning from the French text in such away as to satisfy ATTEST, then I'd be okay with it; but currently, it seems like the dumbest kind of pattern matching, and it's just as likely to match asinin from a French book "Learn Albanian" where asinin means "doorknob", or from a French book "Learn Basque" where asinin means hubcap. (Okay, I just made those up; but you get the point.) I'm very dubious about the current operation of those bots currently. Perhaps this is one way where the current blossoming of LLMs will help, but the current versions do not seem adequate to the task. If it were under my control, I'd turn them off. Mathglot (talk) 07:02, 6 May 2024 (UTC)
- You can't write your own rules. Feel free to discuss this at the Beer parlour, but everyone else (including native speakers) follows the rules and we can't make an exception for you. Chuck Entz (talk) 12:18, 6 May 2024 (UTC)
- Chuck Entz, I don’t wish to write my own rules, nor do I desire any exceptions to be made in this case. What I am doing is pointing out the consequences of some bots not following your own rules as described at WT:ATTEST, and that consequence is that you are taking glosses like asinin which are not attested with meaning in French, and adding them to Wiktionary in contravention to your own rules, which I have no interest in changing. In a word: your bots are running wild. I am glad to hear that everyone else follows the rules, and maybe that applies to your human editors, but from what I have seen, it does not seem to apply to your bots. Not sure how it works here at Wiktionary, but users that repeatedly flout the rules are blocked at Wikipedia, whether they are human or not. Mathglot (talk) 22:25, 6 May 2024 (UTC)
- You can't write your own rules. Feel free to discuss this at the Beer parlour, but everyone else (including native speakers) follows the rules and we can't make an exception for you. Chuck Entz (talk) 12:18, 6 May 2024 (UTC)