Wiktionary talk:Votes/pl-2009-03/Context labels in ELE

Too much information

An ad hoc label and the underscore can be explained at Template talk:context. I would want to see a really good use of the underscore before listing it there. Putting this information on ELE is controversial because, for instance, in this case it suggests that Northern,_,US is the correct way to regionalize the term. I submit it is not. DAVilla 19:23, 1 April 2009 (UTC)

I think it is the best way at the moment, since the most detailed relevant category is Category:US. But the example is at risk that Category:Northern American English were to be created. It does work for now, but I'll change it if a better one occurs (perhaps some regionalism so specific that it is unlikely to ever warrant a category).

This type of example helps stress the use of atomic labels in context's parameters: knowing which labels and categories exist, we can use {Northern|_|US} to insert cat:US, but I often see things like {Northern US} which leave entries uncategorized. (Creating a set of the basic regional categories and labels for anglophone countries is on my long list.)

Perhaps I'll add make sure the full treatment is in the template docs, then remove some cases from here. Might be a day or two before I have time. —Michael Z. 2009-04-01 19:36 z

Actually {{context|Northern US}} wouldn't be any worse because the template can always be created after the fact. It would be easier to create that than to sift through all the US entries. But come to think of it should be permissible to employ compass directions with regions, if that could be more correct. I just wish these were more settled, and don't really like advertising ad hoc solutions in the meantime. DAVilla 08:57, 2 April 2009 (UTC)

Compass directions worry me, because they are not regions, hence they do not categorize regionally.

For example, one template was being used to label linguistically and geographically unrelated Northern English and Northern Tatar. Carolina was thinking of using the same template for Northern Catalan, but I advised creating a new template. I separated the regions by creating {{Northern England}} and {{Northern Crimea}}. So now, Category:Northern England English and a hypothetical Category:Northern England Romany could both be placed in Category:UK (or Category:England or Category:Northern England).

Anyway, in the next week or two I will do a bit of research to define some subnational dialect categories for the British Isles and North America at least. I'll try to base it on real-world dictionary labels and dialectal atlases. I'd hope that would help moot the issue for a while. But maybe not, because even though we don't have the broader categories for English Midlands and West Country, we do have a few individual counties already in use.

I know, too much information. In summary, I agree with you that we should be permissive in defining dialect regions at a finer scale. —Michael Z. 2009-04-03 00:25 z

The problem I'm worried about is analogous to {context|Northern|_|UK} turning into (Northern British), especially since ight now there's talk about whether US is appropriate. Better to hash this out then try to generalize. I'm not knowledgeable enough to categorize the dialects, but it does need eyes like yours. DAVilla 21:48, 4 April 2009 (UTC)

I had a look over Template talk:context#Documentation. It mainly describes how to use {context|whatever}, rather than the standard practice of {whatever}. It would have to be rewritten, but is that really the right place for this? That template is just a technical vehicle, but we really need a general page about best practices and techniques for Wiktionary:Context labels.

Good idea, I'll go ahead and move some things around. It talks about {context} because it was initially set up that way only, not such that the tags could be used independently. DAVilla 21:48, 4 April 2009 (UTC)

But you are right, only the technical basics belong at ELE. I'll revise the proposal. —Michael Z. 2009-04-03 02:19 z

Categorization

This includes the admonition that context labels "must not be used merely for categorization". (Followed by a link which doesn't work.) But why not? I think that often context labels are an excellent way of doing exactly that. It's not so much an issue with English-language entries, where I agree context labels seem to indicate specialized jargon, but with foreign-language entries they are a very common and efficient way of showing different realms of usage. Consider a page like (deprecated template usage) manchon, where the context labels seem to me to be infinitely preferable to long-winded bracketed explanations, even though none of the senses is really jargon. Ƿidsiþ 07:13, 11 May 2009 (UTC)

These are called usage labels in lexicography, and they indicate restricted usage. Same in Wiktionary. See label, subject label, usage label.

In the example, these labels are not only being used to place the entry into categories, but also as a substitute for writing a clear definition. Both are wrong.

The example implies that the word manchon for muff is only used by clothing designers or something, or only when the topic of conversation is clothing, or only when wearing clothing, or something, and that it is only a coupler in the field of “technology”. This is confusing and misleading. I don't see how using these in a non-standard way in foreign-language headings can be preferable to writing a clear description including a phrase like “an article of clothing”. —Michael Z. 2009-05-11 13:53 z

Well, firstly, whether this is "non-standard" or not is up for discussion. Secondly, foreign-language entries are not supposed to have definitions. They are supposed to have translations. Ƿidsiþ 16:27, 11 May 2009 (UTC)

How is this not up for discussion? You are taking a context label and completely repurposing it. —Michael Z. 2009-05-12 19:22 z

And thirdly. The links you provide don't support what you're saying. The first asks, "in a bilingual dictionary, do the translation equivalents for each sene of the word need to be separately marked with corresponding labels?" Exactly what I am talking about. The third link notes that "some of these distinctions [between usage labels] do not have clear boundary lines....dictionaries differ widely in the scope and consistency of their labelling practices". Ƿidsiþ 16:34, 11 May 2009 (UTC)

You are quite misinterpreting the source. When it talks about a “bilingual dictionary,” it means the equivalent of our translations section in an English entry – but here we are not talking about whether we mark translations with context labels. We are talking about our foreign-language main entries: whether they are defined or translated, they belong to contexts, which we do mark using context labels. For example, magasiner#French is a Canadian French word meaning “to shop,” not a word used in France for “shopping in Canada.” tabarnak#French isn't an expletive name for Canada, it's a Canadian French expletive.

The third link mentions the fact that the distinctions between contexts or language varieties do not have clear boundary lines (meaning that register, e.g. informal, may or may not overlap with correctness, e.g. non-standard), and laments that consequently dictionaries vary in the way they determine whether a usage label applies (e.g., by a lexicographer's opinion rather than objective statistics from the corpus, as detailed in Landau 2001, Dictionaries: The Art and Craft of Lexicography). But it is talking about problems with labelling of usage (under the glossary heading usage label), while you are interpreting it to be recommending that identical labels should be used for something else completely – as shortcuts for part of the definition. (And if your interpretation were correct, you'd have to note that the source is identifying a problem characteristic of dictionaries, so it's hardly justification for us to adopt a practice.)

What we do do in FL entries is disambiguate. We add notes identifying a sense, after a definition/translation. This must be done in either plain text or using {{sense}}, so as not to add restricted-context categories. For example, in водень#Ukrainian the notation “chemical element” identifies which sense of hydrogen is being referred to. It would be wrong to place the subject label {{chemistry}}, which would identify this as a technical term, whose usage is restricted to the context of chemistry, and place it into a category for such. Other examples: дума#Bulgarian, дума#Russian. —Michael Z. 2009-05-12 19:22 z

I've taken the liberty of editing manchon's definitions for clarity and correctness. I have no idea what “(technology) coupler” means, however, so I haven't touched that one.

I think that context labels should have a more distinct format to prevent this kind of confusion – something that can't easily be applied using easily-accessible wikitext code.

Also, we do suffer from differing “widely in the scope and consistency” of our labelling practices. We use context labels both to identify restricted usage of terms, a proper lexicographical function, and for thematic categorization, needlessly duplicating (in an inadequate way) Wikipedia's rich encyclopedic categorization of things and concepts. —Michael Z. 2009-05-12 19:41 z

When it talks about a “bilingual dictionary,” it means the equivalent of our translations section in an English entry. What makes you think so? All our foreign-language entries are equivalent to bilingual dictionaries, which translate from source language into English. That's why FL entries don't have translations sections themselves. We add notes identifying a sense, after a definition/translation. This must be done in either plain text... etc. Says who, where? I thought that's exactly what we were voting on. Look, my problem is not really with context labels per se, which I think basically work well. My problem is that your alternative in these situations looks unwieldy and amateurish to me. I can't imagine any print dictionary defining (deprecated template usage) manchon as ‘muff (an article of clothing)’, and for reasons that have little to do with lack of space. Furthermore it remains the case that many dictionaries do use context labels to indicate categories of usage, in exactly the way that you, apparently for rhetorical effect, pretend to have trouble understanding. Eg here, to take a relevant example. Ƿidsiþ 20:17, 12 May 2009 (UTC)

I don't think of FL entries as merely translations, because they can get quite full, and sometimes there is no translation; viz. горілка. But all right, even if we consider them to be translations, for Wiktionary the answer to the question “do they” is yes. We do use context labels in foreign-language entries, and {{context}} has support for a language parameter for this reason.

And I'm not trying to be difficult so please don't accuse me of it; I really haven't a bloody clue what “(technology) coupler” means (and my French is too poor to determine this from your link). I initially read it as a novice mistake in use of a context label leading to an ambiguous definition: does that mean technologists use the word, or it is restricted to technical contexts, or does it refer to a particular coupler technology? But if the template is applied the way you would have it, then we have more possibilities: we have two English definitions of coupler, one extremely broad, the other very specific, and I can imagine many more senses. Both senses relate to technology, and to specific technologies, at various levels, so “technology” doesn't refer to one of them (or, again, is it meant to?). So what is a manchon (“coupler”, technology)?

Does manchon just mean “coupler?” If so, then the misused context label removes all meaning, because it implies that there is more to it, but its meaning is completely vague.

Hold this vote pending CI's proposal in the BP?

Current BP discussion is about the validity of considering ELE immutable policy and includes a proposal by Conrad.Irwin to change ELE to a non-policy page and have a separate policy page with fewer clauses. Q.v. The result of that discussion may impact people's votes in this. Shall we put this vote on hold pending the resolution of the other issue?—msh210℠ 17:08, 13 May 2009 (UTC)

May as well continue with the schedule, and allow the result of this to follow any forthcoming changes in ELE. There seems to be support for the concept (but problems with the details), so let's carry on. I've finally tried to address that by making a much-abbreviated version 2 at Wiktionary:Votes/pl-2009-03/Context labels in ELE v2. —Michael Z. 2009-05-17 18:11 z