Perfective and imperfective forms

In Dutch, diminutives are predictable and every noun has one, as long as semantics permit.

You can't compare it with paper dictionaries because they don't have categories, they only have soft redirects. Why would anyone want to look up only the imperfective verbs that are less common than the perfective equivalent? And if they do, would people not also want to look up the perfective verbs that are more common than the imperfective equivalent? If we're going to categorise, why categorise the "form" but not the "base"? If you want to be consistent, you should create a category for šmrkati too, then.

I still very much doubt the use of any of these categories, and I don't understand what use you see in them, despite all your attempts. You've yet to give a real use case that would lead to someone wanting to look up specifically all the imperfective verbs that are the less-common one of each pair of verbs. What's the use?

Not all languages are like Dutch. Usually there can be many ways to form diminutives.

One of the advantages of these categories is keeping the common base (part of speech) category less populated with contentless entries. Just like Dutch diminutives don't categorize into Category:Dutch nouns beside Category:Dutch diminutive nouns. Though Serbo-Croatian at this stage still lacks automatic categorization for perfectiveness in the headword templates. The other advantage is because it's interesting. I want to be able to look these alternative forms up. Fetch a list of them without having to parse the content of all verbs.

The reason why there is no category for the base lemma as well is because it's the unmarked form. Similarly, we don't have categories for "nouns from which diminutives are derived". Though I admit it's an intriguing idea. I do however want to see categories for verbs which are both perfective and imperfective, and for those lacking perfective/imperfective counterpart. These would be interesting because such behavior is unexpected. The default behavior would be each verb having a perfective/imperfective counterpart, with one of those forms being a base lemma, and that kind of assumed behavior needs no special marking through categorization. Everything that deviates from it does.

The use is the same as all the alternative forms categories - why not? We have many more useless categories which nobody every uses, like subdivision of nouns by gender, or by inflection such as Category:Spanish verb first-person forms.

Diminutives aren't necessarily "forms" though. Category:Dutch diminutive nouns also contains nouns that have no non-diminutive base form. The reason that it's done either-or is that most nouns have a diminutive, so if you categorised the diminutives in Category:Dutch nouns then that category would end up containing alternating pairs of nouns and their diminutive. That would make the category less useful overall. And there are in fact several ways to form diminutives in Dutch, but one of them is more or less universal.

And I agree that those categories are useless too. For Spanish even moreso because the "first person form" can be any of a number of forms, it's not just one form. I don't see that as a justification for more useless categories though, that's just a straw man argument.

I noticed that you mentioned this is also a derivational relationship. That changes things quite a bit. I have less of a problem with a category that lists imperfective verbs that were derived from their perfective counterpart. That's much more objective and less arbitrary, because it focuses on etymology and not on some random collection of requirements. Maybe Category:Serbo-Croatian secondary imperfective verbs is a better choice for a category?

But diminutives nouns are nouns, aren't they. They should be fetchable through a category that lists all nouns. Similar justification can be applied to the existence of alternative forms by perfectiveness - not to keep lesser entries co-existing with the base lemmas.

Yes it's a straw man. But so is your selective justification to delete these specific categories, while simultaneously an elaborate scheme of equally named categories persists with little or no use, unchallenged.

It's impossible to know which form is base and which derived. Distribution of attestation really proves nothing, and is merely an arbitrary indicator that can be utilized for practical purposes of deciding which form to lemmatize on. Perfectivization/imperfectivization are productive in both directions. In all Slavic languages there are several ways to do both. And they slightly change meanings as well, it's not merely a matter of completed vs. uncompleted action. And you can chain the process as well. SC alternative forms by perfectiveness are however only used for perfectivized/imperfectivized forms obtained by ablaut or infixes, and not by prefixes, because prefixation more obviously and more often introduces semantic changes than the former. The latter are normal derived terms with perfectiveness contrast to the baseword. I know it's a common practice to list them as perfective/imperfective counterparts in the headword line for some languages on Wiktionary, that practice is a simplification of how things are. It's the prefix that motivates and controls the meaning of derivation. The scope of the alternative forms category by perfectiveness is limited and well-defined, based on the nature of distribution of attested forms of pairs (or sometimes triplets) of perfective/imperfective verbal bases that are being inspected. Your suggested category name would indicate a subordinate nature of one specific form which is an unrecoverable relationship.

Ivan, you're probably too used to the way you structure Serbo-Croatian verbs and unwilling to change. If you take a look at Russian pairs делать/сделать, Polish robić/zrobić and Czech dělat/udělat (Belarusian and Ukrainian are modelled on Russian), Slovene seems to be done similarly. They use a different approach in displaying and categorisation. Isn't it better we unify approach for Slavic languages?

Prefixes such as u- and s- used in the verbs above can have a range of disparate meanings in their derivations such as "completing action inside something"; "moving an object away"; "separating objects"; "separating objects in a downward direction"; "finiteness"; "joining objects". The base and derived terms share the same underlying basic meaning, but they are slightly modified in derivations. For those ablaut/infix perfective/imperfective pairs this is usually not the case - they exists only in pairs (or sometimes triplets) always meaning the same thing but modified for perfectiveness of action, whereas prefixed forms can be many. сделать is really merely a derived form of делать, one of many possible.

What about semelfactive verbs - стукнуть, кинуть, сдохнуть, they are not formed using prefixes? There are too many prefixes and verbs differ in perfective/imperfective without adding anything - спрашивать/спросить, взвешивать, взвесить, they may not have a pair or have multiple opposite equivalents. Imperfective may be more common than perfective and vice versa, making them absolutely equal, so that one can't be really explained as "a form" of the other. They exist in pairs.

доде́лать, переде́лать, вы́делать, отде́лать, заде́лать - all add additional senses to делать, not just making them perfective forms but separate verbs, even if these in turn can have additional imperfective forms, e.g. доде́лывать.

Those existing in pairs should be treated like Serbo-Croatian treats them - make the most common one a lemma, and redirect the other one using the templates which I try to keep alive. Treat other variant and secondary stems as alternative forms of one of the infinitives. You notice how usually such pairs don't differ in meaning other than perfectiveness part? :)

Prefixed forms are truly derived words, and have a separate treatment in Serbo-Croatian as well. They add and significantly (almost always unpredictably) modify the original meaning. Each of those prefixes can introduce half a dozen semantic modifications which are listed in grammar books.

There are of course always exceptions and special cases (usually one of the forms shares all of the meanings of another, but has some special ones on its own, which I usually handle by listing them below the redirection template), but the base argument seems to be solid and applicable to the majority of cases.

I cannot agree with you about lemma and derivations. They are not forms, they are separate words. Nouns, adjectives, adverbs are also formed using suffixes and prefixes. This info useful in etymologies, e.g. сделать = с + делать but it doesn't need to be classified as a perfective form of делать, they form a pair. In many cases, the perfective verb is much more common and would be defined as lemma with your criteria (чокнуться) or may not have an imperfective form at all (кануть) or they use suppletive forms - сесть/садиться.

You can't exactly disqualify my argument based on my unwillingness or lack of initiative to clean up all similar cases. That's a fallacy in itself.

The distribution isn't really a very objective criterium either. It can even be dialect-specific for all we know. How do we know whether someone who comes along sometime in the future will not "grasp" your reasoning and decide to apply different criteria, whichever they think makes sense? If the requirements for an entry to appear in the category are so arcane that others can't recover it, then the category is just not well-defined enough to exist. That's why I proposed an etymological basis, that's more objective. And it's no less "subordinate" than your own proposal is, or at the very least no less asymmetrical.

But here again it seems like you're accidentally shooting your own argument in the foot. If the perfective/imperfective distinction is not just binary, but more detailed and subtle, like you say, then that's a very strong argument against using {{imperfective form of}}, because that would be just as much a simplification of reality. Such subtle details would need a proper definition to describe them, not just a form-of template. So really, here's what I see:

• If the distinction is binary, and the derivational relationship is either secondary or unrecoverable, then your two templates would be ok. But it would be arbitrary which verb it's placed on, and any category it adds would likewise be arbitrary and only useful for statistical purposes or as a curiosity (your own, it seems).
• If the distinction is not binary, but has various shades of meaning that might not be clear from just the "imperfective" or "perfective" label, then your templates would not be an accurate representation of reality and should be replaced with proper definitions which can show all the nuances.

Either way, the categories don't seem to have much use beyond your own personal interest.

I haven't made up this criterion. It was adopted from paper dictionaries that use it, and that have been using it for a very long time. It is not at all a matter of debate. Requirements are not arcane - they're obvious. Speakers instinctively perceive forms such as šmrknuti, jeknuti and muknuti as secondary because they are very rarely used as such. Corpora search is an exact and reliable criterion to establish precedence.

As I've explained, etymological basis is arbitrary. šmrknuti can just as easily derive from šmrkati, and vice versa. Perfectivization and imperfectivization can go both directions. Yes my way is asymmetrical but that's how things are. Languages don't fit into some perfect mathematical formula. They are full of inconsistencies. Some verbal pairs of the exact same shape will have a form of one perfectiveness as a base lemma, and other verbs will have another. For a user looking up such verbs, in both cases the most common form will be shown with definition lines.

The distinction is orthogonal to meanings. There are many ways to perfectivize or imperfectivize a verbal stem. That's why I'm arguing that the way Russian and others are handling perfective/imperfective variants in the headword line with prefixed forms is too simplistic. However, for SC, the only verbs who are redirected are the ones where there is no semantic discrepancy between the pairs. That is your first bullet point (except for the last sentence).

You're making it seem like it's my own personal interest. It's interesting how when you're left without arguments you tend to personalize arguments by interlocutors. The bottom line is:

• These are indeed alternative forms in the all of the other dictionaries. When you look up šmrknuti you're redirected to šmrkati [1]. The only modification in meaning is whether the action has been completed or not.
• We categorize all other alternative forms that are lemmas, in their own special categories.