Wiktionary:Votes/pl-2018-12/Lemming principle into CFI

Lemming principle into CFIEdit

Voting on: Adding the following as the next-to-last paragraph of WT:CFI#Idiomaticity, and therefore, as the paragraph before the one starting with "In rare cases, a phrase that is arguably unidiomatic ...":

An attested term that appears to be a sum of its parts yet is included in at least two professionally published general monolingual dictionaries should be included. Such dictionaries include but are not restricted to Merriam-Webster, OED, AHD, Cambridge, Collins, Macmillan, Longman, German Duden and Spanish DRAE; dictionaries that do not count include WordNet.

Rationale: See vote talk page. The voters only vote on the proposed action, not on the rationale.


  • Vote starts: 00:00, 3 January 2019 (UTC)
  • Vote ends: 23:59, 3 March 2019 (UTC)
  • Vote created: Dan Polansky (talk) 08:14, 27 December 2018 (UTC)



  1.   Support. Let's face it, many of the dictionaries used as references are predecessors of Wiktionary in printed form. DonnanZ (talk) 11:38, 3 January 2019 (UTC)
  2.   Support It is a reasonable shortcut for including terms that are useful to readers, as other dictionaries seem more eager to include fixed phrases. Some SOP terms that can be kept via the Lemming principle are drinking water, local area network and white supremacist (though some could be kept by means of other rationales, others cannot). ←₰-→ Lingo Bingo Dingo (talk) 11:44, 3 January 2019 (UTC)
  3.   Support in general for attested terms. Acceptable dictionaries for specific languages to be agreed on separately. --Anatoli T. (обсудить/вклад) 13:33, 3 January 2019 (UTC)
  4.   Support, provided that some indicator of the SOP-ness of a term can be added, should a term be challenged in RFD. Andrew Sheedy (talk) 03:30, 4 January 2019 (UTC)
  5.   Support per rationale on the talk page. --Dan Polansky (talk) 07:21, 5 January 2019 (UTC)
    An example nomination: WT:RFDE#president-elect: I check president-elect at OneLook Dictionary Search, and off I go building the dictionary; case closed. I can also ponder independently on usefulness of "president-elect" and I see one, but I can be spared the trouble. --Dan Polansky (talk) 10:34, 5 January 2019 (UTC)
  6.   Support Mofvanes (talk) 18:51, 5 January 2019 (UTC)
  7.   Support We still require attestation, which is good and ensures we are documenting language as it is used. If independent groups of professional lexicographers think a term is worth explicitly defining, I'm fine with following that lead. The proposed rule allows us to rapidly close some RFD discussions, freeing up editors' time to invest in other things on the site. The application of both attestation and multiple dictionaries protects us from the risk of fictitious entry copyright traps. -Stelio (talk) 14:43, 15 January 2019 (UTC)
    Well put. --Dan Polansky (talk)
  8.   Support Matthias Buchmeier (talk) 07:29, 16 January 2019 (UTC)
  9.   SupportMnemosientje (t · c) 14:52, 17 January 2019 (UTC)
  10.   Support — justin(r)leung (t...) | c=› } 02:53, 18 January 2019 (UTC)
  11.   Support פֿינצטערניש (talk) 12:39, 20 January 2019 (UTC)
  12.   Support per Dan Polansky rationale. Richardb43 (talk) 01:12, 21 January 2019 (UTC) (See old persona at user:richardb)
  13.   Support. Generally I'd consider it not so much an argument in itself as an argument in favour of some other grounds for inclusion, one which is obviously recognised by professional lexicographers – but as a shortcut, I would support this. Ƿidsiþ 12:54, 2 February 2019 (UTC)
  14.   Support This is a good practical way to do things. John Cross (talk) 20:47, 6 February 2019 (UTC)


  1.   Oppose: Other general dictionaries are bad. The less one has to use them the better. Fay Freak (talk) 01:13, 3 January 2019 (UTC)
    Re: "Other general dictionaries are bad": Untrue generalization, with zero substantiation. A substantiation would consist e.g. in giving example entries that are bad, in other dictionaries, and these would be bad sum of parts entries since attestation is not given up by the proposal. --Dan Polansky (talk) 07:21, 5 January 2019 (UTC)
  2.   Oppose: Most of the English entries I've created are supported by two or more lemmings, but I still want people to feel free to RFD them if they have good reason to think we shouldn't have them. Similarly, I don't want to be shut down with a peremptory Keep per WT:LEMMING when I nominate entries for deletion. In other words, let reasoned discussions keep happening. By the way, more doesn't necessarily equal better. Per utramque cavernam 01:21, 3 January 2019 (UTC)
    Let me add that I concur with -sche. Per utramque cavernam 22:52, 3 January 2019 (UTC)
    "More doesn't necessarily equal better". I have added a reference to the more the merrier. DonnanZ (talk) 12:53, 3 January 2019 (UTC)
  3.   Oppose Different dictionaries can have different "words" in them. That's a good thing. We should aim to describe language and not other dictionaries. I particularly object to sentiments like "for the sake of competition, we should include everything multiple professional dictionaries include". Finally I think that the list of acceptable sources is slightly nonsensical and I imagine that it will lead to many disagreements in the future about "acceptable" dictionaries. DTLHS (talk) 16:22, 3 January 2019 (UTC)
    1) The aim indeed is to describe language, not dictionaries, and to indicate that an attested term, using attestation methods of en wikt, means such and such; that is not a description of other dictionaries. 2) The "for the sake of competition" rationale is not part of the main rationale presented on the vote page. It cannot possibly be that a proposal is opposed only because of opposition to one of multiple independent rationales supporters have: a theorem is proven when one correct proof is found regardless of how many incorrect proofs were submitted. 3) The list of sources is there to give an idea of a broader term "professionally published", and that it is so is reinforced by "include but are not restricted to" phrasing; I don't see how it is "nonsensical", and disagreements about the scope of "professionally published" can be there either way, and the examples merely reduce the possible disagreement. --Dan Polansky (talk) 07:21, 5 January 2019 (UTC)
  4.   Oppose Equinox 20:03, 3 January 2019 (UTC)
  5.   Oppose. I think lemmings can guide us in our determinations of whether or not something is NISOP or otherwise inclusion-worthy, but I don't want a rule that indiscriminately includes anything other [pairs of] ["acceptable"] dictionaries have. - -sche (discuss) 22:10, 3 January 2019 (UTC)
  6.   Oppose. This rule could allow a lot of entries that really don't need to be included. In my opinion, an entry "needs" to be included if someone might come to Wiktionary looking for its definition, and although some of the entries the lemming principle allows will meet this requirement, all of them definitely won't. We shouldn't have to include a term that no one will ever look up just because another dictionary does, and this still applies even if several other dictionaries decide to unnecessarily include that same term. Implementing this policy would make editors of professional dictionaries indirectly responsible for judgments about what belongs in Wiktionary instead of allowing Wiktionary editors to directly make those judgments. Making a policy like this feels dangerous to me because it takes some power of judgment away from us as editors. It might make RFDs quicker, but quicker doesn't necessarily mean better if it robs us of a piece of our independence. —Globins (yo) 08:29, 5 January 2019 (UTC)
    I find it likely that professionally published dictionaries include additional terms because people keep on looking them up; Wiktionary editors keep on proposing for deletion useful entries that people keep on looking up, from my experience. The notion that lemming-included entries will include those that "no one will ever look up" seems rather unlikely, and in any case, unsubstantiated: not a single example has been provided. And even if the lemming criterion would lead to inclusion of some terms that are not looked up, that is compensated by additional inclusion of terms that people do keep on looking up, and would not find in Wiktionary otherwise. Put differently, adding meat and protein is good even if it means adding a little amount of fat and carbs as well (the analogy is not so good since we do need fat and carbs, as well as dietary fiber). --Dan Polansky (talk) 08:56, 5 January 2019 (UTC)
    I agree that it's better to have unnecessary entries than to not have necessary entries, but my point is that we shouldn't trust other dictionaries to make decisions for us about which entries are the necessary ones; we should be the ones making those decisions independent of what other dictionaries are doing. —Globins (yo) 09:13, 5 January 2019 (UTC)
    That's a matter of cost benefit analysis. This proposal is a partial outsourcing proposal (attestation is not given up), and also a proposal to include more, never to include less because of lemmings. It is not an utter and complete delegation of inclusion criteria to other dictionaries. An alternative is to keep on developing all criteria in house. As a wiki lexicographer, I am doing my best to stubbornly try to think clearly and look at and search for evidence, but I do not have the time to devote to figuring out criteria and their repercussions, and I would much prefer to have more of my energy spared for expanding the dictionary rather than having to discuss and show evidence in RFD when useful entries are being threatened with deletion, sometime by students who have not seen a paying customer in their lives, and have no sense of economy of effort and delivery of results. --Dan Polansky (talk) 09:33, 5 January 2019 (UTC)
  7.   Oppose --{{victar|talk}} 06:47, 8 January 2019 (UTC)
  8.   Oppose*i̯óh₁n̥C[5] 23:41, 14 January 2019 (UTC)
  9.   OpposeJulia 18:17, 24 January 2019 (UTC)
  10.   Oppose with a heavy heart. This does not accomplish much because of the "multiple respectable" dictionaries restriction. As unintended consequences often flow from the application of legalistic argument to novel wording, The small benefit does not seem to me worth the risk. DCDuring (talk) 12:36, 1 February 2019 (UTC)
    @DCDuring: Could you clarify? The wording used is "at least two professionally published", not "multiple respectable". The wording is a result of Wiktionary talk:Votes/pl-2018-12/Lemming principle into CFI#How to gauge dictionary quality?. Without such a requirement or similar, amateur monolingual dictionaries would count as well. And what is the "risk" you are talking about? What makes you think "at least two professionally published" significantly reduces the benefit? --Dan Polansky (talk) 12:06, 2 February 2019 (UTC)
    I don't. I just see it as having insufficient benefit. It will not help in adding terms from the worlds of work and crafts, where we are notably deficient, because the glossaries that contain such terms are not "respectable" and too few in number in the numerous special areas involved. I view any additional wording added to our rules, coupled eith a decided tendency toward legalistic argumentation, as providing too much opportunity for unintended consequences. DCDuring (talk) 12:33, 2 February 2019 (UTC)
    @DCDuring The proposal says "professionally published", not "respectable". Do you have a particular glossary in mind that is not "professionally published" and yet worthwhile? --Dan Polansky (talk) 12:37, 2 February 2019 (UTC)
    In your Beer parlour proposal years ago, you wrote: "In the interest of speeding up RfDs for English terms and senses [...]". Does the proposal fail to achieve the speeding up that you intended? Also, you wrote: "If it can't be enacted into something virtually automatic, it wouldn't do the job I was looking to have it do." If a proposal is automatic, it has to contain something that prevents inclusion of unreliable amateur works, doesn't it? --Dan Polansky (talk) 12:41, 2 February 2019 (UTC)
    Absolutely. So, on mature reflection, having seen how marginal most new English entries are, and recognizing that our ability to have this principle operate automatically at the definition level will not be very good, and for the reasons stated above, I oppose this. DCDuring (talk) 13:48, 2 February 2019 (UTC)
    The rate of speeding up RFD does not seem to depend on new entries but rather on those that end up in WT:RFDE. And in WT:RFDE, my experience is that there is plenty of use of WT:LEMMING. --Dan Polansky (talk) 14:05, 2 February 2019 (UTC)
  11.   Oppose—The lemming test is a useful diagnostic, but it shouldn't give an automatic pass. If several dictionaries list a multi-word phrase as an entry, Wiktionarians should be able to articulate why and use that argument to keep the phrase, rather than simply pointing to those dictionaries and calling it sufficient. —Mahāgaja · talk 10:45, 2 February 2019 (UTC)
  12.   Oppose – We don’t know which criteria the editors of some professionally published dictionaries use to include some (in my eyes occasionally obviously Ni) SoP entries; we should not be bound by them and be free to apply our own judgement to the question of idiomaticity.  --Lambiam 23:46, 2 February 2019 (UTC)
  13.   Oppose: Inclusion in other dictionaries may be taken into account, but ultimately we should make our own decisions about what to include in Wiktionary. Mihia (talk) 18:35, 8 February 2019 (UTC)
  14.   OpposeUngoliant (falai) 11:22, 14 February 2019 (UTC)
  15.   Oppose per Mahāgaja. — Droigheann (talk) 19:18, 17 February 2019 (UTC)
  16.   Oppose I'd like a way for us to include SOP terms, but I wouldn't want them being first-class citizens and I loathe rulings that reduce our autonomy. Crom daba (talk) 23:12, 3 March 2019 (UTC)


  1.   Abstain good arguments on both sides, can't decide. – Jberkel 12:42, 2 February 2019 (UTC)


If the vote fails, what will happen of Wiktionary:Idioms that survived RFD § Lemming test? Per utramque cavernam 21:37, 10 January 2019 (UTC)

I don't think this should change the results of past RFDs. If the word passed, then I don't see a reason for it to be deleted, especially considering that the lemming principle has never been an official policy. —Globins (yo) 02:03, 11 January 2019 (UTC)
@Globins: I'm talking specifically about the "Lemming test" section on the page I linked to, not so much about the entries that have been kept thereby. Should it be kept there, or should it be removed entirely? Per utramque cavernam 13:36, 11 January 2019 (UTC)
It can stay until there's a discussion (and consensus) that's about deleting it. Most of the other tests on that page aren't backed by CFI either; the point of the page is "to rationalize how and why some terms are idiomatic when others are not". Several of the people opposing making this a rigid policy explicitly think it's fine as an advisory guideline, and the page does say "tests can be used as guides during RFD, but they are not hard/fast rules". - -sche (discuss) 19:02, 11 January 2019 (UTC)
Understood. Per utramque cavernam 22:40, 11 January 2019 (UTC)