Wiktionary talk:Votes/pl-2020-04/Attestation of comparatives and superlatives

Rationale

There's almost nothing to see here. I feel the need to have this codified stems from the notion supported by some that inflected forms do not need to be individually attested; here we make it clear that comparatives and superlatives are the kind of thing that does need to be individually attested.

There are almost no votes running so this may be a good time for this kind of closing minor policy gaps.

--Dan Polansky (talk) 09:07, 10 April 2020 (UTC)

And how can you do as if this notion shall apply to all languages equally? I oppose this for Serbo-Croatian, and for German too it cannot really work. And so on and so on. Inflected forms do not need to be individually attested, and we do not need a written exception which is not common sense. Fay Freak (talk) 10:02, 10 April 2020 (UTC)

Please give specific examples of German words for which this does not work. The English Wiktionary has a long-term practice of not including comparatives unless they are attested, and that matches my common sense. --Dan Polansky (talk) 10:09, 10 April 2020 (UTC)

If it matches your common sense than it does not need a vote. The more text of rules there is the less people read it. In German one does not have this distinction of comparatives in -er and comparatives in more. Every adjective can have one in -er, and whether it is attested is irrelevant and a function of the frequency of a stepped-up sense. But you also voted for the inefficient work of attesting Romanizations of Sanskrit. Fay Freak (talk) 10:17, 10 April 2020 (UTC)

As for "Inflected forms do not need to be individually attested": that is controversial, not a written policy; I opposed this repeatedly and I saw some others oppose it, while I also saw multiple people support this. --Dan Polansky (talk) 10:12, 10 April 2020 (UTC)

I don’t trust what you see. I have not seen anyone opposing it, and for Latin everyone has supported it except some noobs who have not thought it through. Fay Freak (talk) 10:17, 10 April 2020 (UTC)

I honestly report to the reader that the thing is controversial. Let Fay Freak put it to vote and see how it fares. --Dan Polansky (talk) 10:33, 10 April 2020 (UTC)

And drop the crap, as per "I deny that there are people short of you who prefer no lang prefix. That’s just a reactionary reminiscence of the past when there was disorder of template names. Fay Freak (talk) 15:20, 10 November 2018 (UTC)" and the failed Wiktionary:Votes/2019-06/Language code into reference template names. Stop the nasty crap. --Dan Polansky (talk) 10:38, 10 April 2020 (UTC)

Plus you didn’t even say what requires attestation: The page creation or the link in inflection tables or headwords. If the later, it will break our template systems; if the former, automatical creation becomes difficult. Fay Freak (talk) 10:17, 10 April 2020 (UTC)

Please give me specific examples, words. Drop the personal, discuss the substance. Examples please. --Dan Polansky (talk) 10:33, 10 April 2020 (UTC)

You should drop the personal and discuss more substance, @Dan Polansky. You only eject value judgments without any rationale. Examples do not matter as need for rules derives from whole pictures. Fay Freak (talk) 12:55, 10 April 2020 (UTC)

We need examples. If we have no examples, we do not have the whole picture; we have a picture with zero pixels, or we have a white canvas with no paint on it. Let us continue the British empiricist tradition and let us abandon the continental idea that we learn about the world by looking into our own head. --Dan Polansky (talk) 09:00, 17 April 2020 (UTC)

The British have abolished themselves already soon after assuming this abandonment of caution. Prevention is better than acting after the damage is done. No experiments! Fay Freak (talk) 09:55, 17 April 2020 (UTC)

Let me address something not discussed above. One could argue that in order to determine which adjectives are comparable, one could look at the nature of the adjective and decide based on that, without looking at attestation; thus, when one would determine the adjective to be relational, one would conclude that the comparative and superlative are not there. Alas, that does not work since some relational adjectives do turn out to be comparable for a variety of idiosyncratic reasons; to wit, Czech papežský has the attested and frequently used comparative form papežštější. --Dan Polansky (talk) 06:02, 21 April 2020 (UTC)

Wording change proposals

If people have wording change proposals, please post them. Let us discuss possible problems and how they could be addressed by amending the wording. Thank you. --Dan Polansky (talk) 11:07, 10 April 2020 (UTC)

Which problem is addressed here?

Can you elaborate on which problem this addresses that is not already adequately covered by our current criteria? I think that if an editor added a comparative *usabler, or a superlative *magnifickest, it would soon end up on RfV and be shot down. We have a current RfV on the contested Latin comparative medior. Are there examples of a putative comparative or superlative form being contested, but an RfV being dismissed because "of course it exists even though it is not attested"? Also, how is this an issue specifically for these forms, and not for plurals, diminutives, and what not? Why single them out?  --Lambiam 12:38, 12 April 2020 (UTC)

I don't know it it meets CFI, but the comparative usabler certainly exists:

• "24 Mar 2014 - yes, the calibration routine of the cmps11 is very much usabler that the older one."
• "OK, both are practically same accurate and counts 50000/5000 but a 89 has lot more functions, is a usabler datalogger..." - https://www.vintage-radio.net/forum/showthread.php?t=147143

The word also seems to exist as a QWERTY typo for 'usable'. --RichardW57 (talk) 15:12, 19 May 2020 (UTC)

Side note regarding English: I have, several times in the past, raised verb forms that do not exist (e.g. rare Shakespearean verbs that only occur in one or two forms) and the consensus has fairly clearly been that we want the whole verb even if not all forms are attestable. With plurals it hasn't been so clear (partly because countability can vary for nouns) and we also have the "no plural attested" {{en-noun|!}} template. I can't remember specifically challenging any compara/superlative but I have removed a few unlikely ones that seemed like wishful thinking on the creator's part. Equinox ◑ 12:06, 15 April 2020 (UTC)

The problem is that it is not expressly codified that comparatives and superlatives are only entered if, for well-documented languages, there is empirical evidence for their existence. For less well documented languages, CFI does not require empirical evidence for something to be attested. Some people may see the need for attestation as obvious, but others don't, and that makes codification worthwhile. --Dan Polansky (talk) 08:53, 17 April 2020 (UTC)

Let me give you an example. lumbální is a Czech non-comparable adjective. Should it carry lumbálnější as a comparative? No, it should not. But why should not? Because the empirical attestation evidence tells us so: the form lumbálnější is not attested. Let's take English lumbar: it is entered as "lumbar (not comparable)". How do we know that it is "not comparable"? We know that since *lumbarer and *more lumbar are not attested (later: of course, "one or more lumbar vertebrae" does not attest comparative "more lumbar"). The point is to codify the reasoning just shown. --Dan Polansky (talk) 08:57, 17 April 2020 (UTC)

'Attestations' from the web:

• "It is more lumbar than thoracic, one pair of full costal facets for articulation with the 12th rib, no transverse processes (only 3 tubercles." from https://www.studystack.com/flashcard-418794.
• "The 3 signs and 3 symptoms most frequently found included in a diagnostic scale were as follows: (symptoms) (1) axial or bilateral axial lumbar pain, (2) improvement with rest, and (3) absence of root pattern, which may have a pseudoradicular pattern; however, the pain is more lumbar than pain in the leg; and (signs) (1) Kemp sign, (2) pain induced in the joint or transverse process, and (3) facet stress sign or Acevedo sign." - from https://academic.oup.com/neurosurgery/article-abstract/66/Supplement_1/nyz310_704/5551845?redirectedFrom=fulltext

I can also imagine it turning up in discussion of evolution, as vertebrae change their role. Similarly, we have

• "This observation was in contrast with the findings of previous studies where breathing became more thoracic (14, 18)."

from behind a paywall to refute the claim that thoracic is 'not comparable'. --RichardW57 (talk) 15:00, 19 May 2020 (UTC)

As for plurals, diminutives, that could be for another vote; plurals should be uncontroversial and as for diminutives, I support attestation requirements but there was some opposition. I'd say, let us see how this fares with comparatives and superlatives, and move from there.

As for "Are there examples of a putative comparative or superlative form being contested, but an RfV being dismissed because "of course it exists even though it is not attested"?" I do not know of a specific RfV; this is a matter of principle. If someone creates Czech lumbálnější and I put this to RFV, what happens? Will I have to argue with continental editors that attestation is required? I don't know. I thought the case was clear that inflected forms could fail in RFV, but then editors started to claim that inflected forms do not need to be attested. This is an attempt to codify something at least for what someone called "segments". --Dan Polansky (talk) 09:14, 17 April 2020 (UTC)

Is it a “matter of principle” for specifically comparatives and superlatives, or equally for plurals, diminutives, genitives, past participles, optatives, imperatives, and Thrax knows what? If only the former, how are they so different from other inflected forms? If also the latter, I still don’t understand why compara-/superlatives are singled out in this proposal. Plurals can also be controversial; imaging someone adding a Latin plural viri to the entry virus. Many languages have singularia tantum.  --Lambiam 10:21, 21 April 2020 (UTC)

It seems to me the answers to your questions are already in my above answers. Let me quote one: "As for plurals, diminutives, that could be for another vote; ...", so yes, the problem exists for plurals as well, and plurals can be addressed in another vote. As for a possible vote to include only attested inflected forms, that seems too likely to fail or result in no consensus, and I am trying the practice art of the possible here. If you believe the current CFI already requires inflected forms to be attested, I would agree, but it does not do so explicitly and I have seen too many people claim inflected forms require no attestation. --Dan Polansky (talk) 17:01, 21 April 2020 (UTC)

If you believe there is no meaningful difference between individual inflected forms and blocks such as comparative, superlative, plural, etc., I see a pretty obvious difference: it is quite usual for the blocks not to exist at all, not because of poor attestation base (corpus, etc.), but because of deeper reasons; and thus, we have incomparable adjectives, nouns without plurals, etc. --Dan Polansky (talk) 17:10, 21 April 2020 (UTC)

Huge can of worms

What are we going to do, stick a whole bunch of asterisks in front of millions of inflections that don't appear in google books? Is it a question of semantics and grammar? If something technically could be an inflected form but would never actually be used, these cases can already be RFVed so what is this accomplishing? Is it saying they shouldn't appear in inflection tables at all (which would make providing tables for many entries impractical). DTLHS (talk) 12:25, 15 April 2020 (UTC)

To put it another way, there's a big difference between saying something "requires attestation" (big change with unknown implications, see above), vs saying something can be challenged through RFV (reinforces existing status quo). DTLHS (talk) 16:45, 15 April 2020 (UTC)

"Requires attestation" are "can be challenged through RFV" is the same thing, by my lights. And this is not about inflected forms in general (which I expect to be very controversial), only about comparatives and superlatives. --Dan Polansky (talk) 08:39, 17 April 2020 (UTC)

Anyway, to address the above, I changed the wording to "Adjectival and adverbial comparatives and superlatives can only be included in the mainspace if they are attested in the sense of CFI; it does not suffice that the base form is not enough." This is a phrasing approximately parallel to CFI's "This in turn leads to the somewhat more formal guideline of including a term if it is attested and, when that is met, if it is a single word or it is idiomatic", italics mine.

The point is to subject comparatives and superlatives to the same attestation standard as lemmas. Their treatment should be the same as for lemmas: lemmas can be entered without attesting quotations and can be challenged in RFV, and when it happens, a mere derivation plausibility of a lemma is no defense in RFV. If you think the above change in phrasing does not make this clear enough, please propose a phrasing that you are happy with. --Dan Polansky (talk) 09:08, 17 April 2020 (UTC)

All languages vs. some languages

An editor objected that applying this to all languages indiscriminately would lead to problems. The editor has given not a single example causing a problem, and I know of no such example. Still, in general, there is some plausibility to this objection, so I added an option to apply this to only some languages, starting with English in the list. --Dan Polansky (talk) 08:49, 17 April 2020 (UTC)

Czech is the problem too, including the abovementioned lumbální. Inflection tables include these kinds comparatives irrespectively of whether they exist, so it is at least in Serbo-Croatian adjective entries automatically. Theoretical but morphologically correct comparatives for these languages are useful; imagine one is an author and now one wants to use such a comparative for the first time; and still there is no reason why the comparatives and superlatives are singled out here. There are a lot of forms in many languages, for example in Bantu languages like Swahili which is in the list of so-called “well-documented languages” and it would be tedious work with negligible gain to check all attestations. It ends up to be a form of racism towards these African languages as content in them is less accessible, so that coverage is more hard in them, and towards the European languages as most of them are included in WT:WDL. In Czech Dan Polansky omits the inflection tables so they do not include comparatives or anything, this is telling about him not understanding the problem of “sticking a whole bunch of asterisks in front of millions of inflections” and his vote breaking our template use. Work for others but not for him, the classical leftist parasitic stance. Which is the same as that of those armchair-empiricist liberals which abolished England. Fay Freak (talk) 10:01, 17 April 2020 (UTC)

I now added an infection table to Czech lumbální; the inflection table does not contain comparatives and superlatives, and it should not. As before, let us drop the personal and talk the substance. Again, we need examples of specific words that would be negatively impacted. We do not have a single example of a template that would be broken: we need an example to have a look what is at stake. Let us recall that languages that are not WT:WDL have weaker attestation criteria per WT:CFI, and for them, "attested" means something slightly different. --Dan Polansky (talk) 10:18, 17 April 2020 (UTC)

Furthermore, the above post does not even relate to the heading since this section reports introduction of option 2 to address the concerns that some languages could be negatively impacted. --Dan Polansky (talk) 10:21, 17 April 2020 (UTC)

Request for clarification of wording

In the proposal, I don't understand what "included in the mainspace" means. As opposed to included somewhere else / somehow else? Perhaps this could be clarified. Mihia (talk) 01:53, 19 April 2020 (UTC)

Would changing "included in the mainspace" to "included" make the vote better for you? Are you considering to support at least one of the two options? --Dan Polansky (talk) 05:47, 21 April 2020 (UTC)

I went ahead and dropped "in the mainspace": CFI also says "included" without being explicit about the namespace. --Dan Polansky (talk) 05:54, 21 April 2020 (UTC)

@Dan Polansky: Sorry, it may be a bit late for this now, but I wondered whether "in the mainspace" had something to do with the word appearing in inflection tables/templates versus having its own individual entry. Dropping "in the mainspace" removes the question of what "in the mainspace" means, but I am still not clear whether "included" means that a word has to have its own separate page. Sorry for not getting back sooner about this. Mihia (talk) 23:11, 3 May 2020 (UTC)

Starting

@Dan Polansky Does this vote still need changes, or is it ready to start? It's going to be two days past the current starting date soon. — surjection ⟨??⟩ 20:28, 20 April 2020 (UTC)

Surjection: Thank you; I postponed the vote to be able to address Mihia's input above. --Dan Polansky (talk) 05:48, 21 April 2020 (UTC)