[00:49:48] Singular they for a specific person who hasn't specified "they" also leaves half the English native population screaming (re @Nikki: and the same goes for pronouns, in english the obvious choice is singular they, in german you run away screaming) [00:50:33] which is silly of them because everyone uses it that way, whether they realise or not [00:50:49] It aims at generating a best guess of what a human writer would do, not the most politically correct handling of ambiguity (re @moebeus: I'm a little surprised Google doesn't offer both actually, seems like a cheap and easy solution) [00:51:20] Only for hypothetical people (re @Nikki: which is silly of them because everyone uses it that way, whether they realise or not) [00:51:51] so what do you do for non-hypothetical people whose gender you don't know? [00:52:12] And older native speakers default to "he" for hypothetical singular human [00:52:31] "I'm meeting up with a friend later" "what's his slash her name?" [00:53:22] Find out their preferred pronoun (or someone else's words referring to them) before trying to speak about them (Note: I'm aware I've used "they" here because I'm meta-analysing the situation) (re @Nikki: so what do you do for non-hypothetical people whose gender you don't know?) [00:53:43] "and then maybe you want to invite him slash her over for lunch next week, would be nice to meet him slash her" [00:53:54] Or avoid using pronouns (always use the noun form you know) [00:54:09] It's rude to speak out of your ignorance! [00:54:17] haha, like people do that [00:54:22] they just use "they" and then claim they don't [00:55:34] For speakers without a "specific singular they" internal grammar, the response question is "Which friend?" (re @Nikki: "I'm meeting up with a friend later" "what's his slash her name?") [00:55:49] Or "What's your friend's name?" [00:56:15] I use "they" all the time and I've never *ever* been asked what someone's pronouns are [00:56:21] It's easy to avoid pronouns when you're referring to a specific person. [00:56:44] I think my point still stands, it really isn't political correctness that wants me to get the pronouns right when I'm writing or reading something. It's not misunderstanding/not looking foolish when conversing. (re @conrick: It aims at generating a best guess of what a human writer would do, not the most politically correct handling of ambiguity) [00:57:34] If Google offers a 49/50 chance of getting it right, that leaves quite a bit of room for improvement [00:57:51] If Google offers a 49/51 chance of getting it right, that leaves quite a bit of room for improvement [00:57:57] There's no "not looking foolish" here (I'm in the "this problem is intractable" camp of the debate). You're missing a key piece of information, without which it's impossible for you to sound like a native speaker in the situation. (re @moebeus: I think my point still stands, it really isn't political correctness that wants me to get the pronouns right when I'm writing or reading something. It's not misunderstanding/not lookin [00:59:47] This is about languages being different from one another and their syntaxes forcing writers to specify certain unemphasized details, in this case gender of the human being discussed. [01:00:38] When translating from e.g. Cantonese to German you run in to not just gender, but also time. [01:01:28] anyway you can write a different function that doesn't use singular they if you really want, it doesn't change my point that the functions will have to make *some* decision about what to do when gender is unknown [01:01:36] And even pragmatics [01:04:02] I think we need to broaden the discussion because I don't actually think unknown gender while translating from genderless languages to gendered languages is going to be the biggest problem with pronouns we need to worry about [01:05:04] There are languages that *drop* pronouns when the subject is clear from context, e.g. Greek [01:05:59] And then there are languages for which the use of pronouns for a non-hypothetical person is outright rude e.g. Japanese [01:11:31] We're going to have a bigger problem deciding whether to use the name, or the pronoun, or to drop. This will cause us more awkwardness than the few cases where we need to generate a sentence with a pronoun but the gender of the subject is unknown. [01:16:34] (The Japanese anti-pronoun logic is something like this: [01:16:34] 0. Sentences are not required to contain a subject. [01:16:36] 1. If the subject is obvious from context, omit it altogether. [01:16:37] 2. If the subject is unclear from context, say it out loud. [01:16:39] 3. Pronouns are therefore used for a human only when you want to talk about a person but avoid saying the name out loud, i.e. trying to keep emotional distance [01:16:50] (The Japanese anti-pronoun logic is something like this: [01:16:51] 0. Sentences are not required to contain a subject. [01:16:52] 1. If the subject is obvious from context, omit it altogether. [01:16:54] 2. If the subject is unclear from context, say it out loud. [01:16:55] 3. Pronouns are therefore used for a human only when you want to talk about a person but avoid saying the name out loud, i.e. trying to keep emotional distance) [01:24:48] Abstract Wikipedia is not going to translate from any natural language to any other, whether gendered or not (re @conrick: I think we need to broaden the discussion because I don't actually think unknown gender while translating from genderless languages to gendered languages is going to be the biggest problem with pronouns we need to worry about) [01:24:54] (as far as I understand) [01:25:04] so I'm not sure why this discussion is happening tbh [01:28:24] Agreed [02:22:43] What Nikki and Lucas said. The abstract representation will be able to store the information, including a preferred pronoun, no matter whether specific languages then have just a single genderless pronoun or drop the pronoun altogether. [02:26:40] If a language needs the pronoun but doesn't have it, that generation would be blocked and maybe a fallback is used (eg the name) [07:57:27] Can you please define "frame"? (re @markus_goellnitz: For AW the frames won't be generated from free text. So, I guess, the data which entity should be used with which lexeme and which gender will be hopefully more predictable.) [11:52:49] Agree! (re @vrandecic: If a language needs the pronoun but doesn't have it, that generation would be blocked and maybe a fallback is used (eg the name)) [12:14:24] good idea, agreed. (re @vrandecic: If a language needs the pronoun but doesn't have it, that generation would be blocked and maybe a fallback is used (eg the name)) [14:16:12] My understanding in this context of the term was: the natural language text rendered/generated from the abstract content (re @amire80: Can you please define "frame"?) [22:07:05] Hello, am a software developer, and semi active Wikimedia editor (User:Shushugah) [22:11:02] hello! [22:39:48] Welcome!