Module talk:ar-nominals

Much discussion is found elsewhere
See Module talk:ar-headword, Wiktionary talk:About Arabic. Benwing (talk) 18:59, 4 December 2014 (UTC)

Note, some of the discussion below was originally done in Module talk:ar-headword and moved here from there. Benwing (talk) 08:12, 30 December 2014 (UTC)

More declensions?
Is there a need for more declension paradigms? I don't see templates e.g. for diptotes, irregulars and some that were already discussed, including sound plurals (masculine and feminine). There are not so many but I don't know everybody's priorities. --Anatoli T. (обсудить/вклад) 23:37, 30 November 2014 (UTC)
 * There's now one template,, that could in theory handle all of them. It depends on whether they're added to the module. Benwing changed some things in the module now so I don't really know how it works anymore. —CodeCat 23:48, 30 November 2014 (UTC)


 * Everything you mentioned is already included in, if that's what you mean. --WikiTiki89 23:54, 30 November 2014 (UTC)


 * (before E/C) OK, I'll wait for you to sort this out. Please post here if I can help. Could you explain some changes? We need to add some common irregular nouns too, e.g.  (from  - the "five nouns" type).
 * (after E/C) If the template is ready to be used, then perhaps we need the documentations and examples. --Anatoli T. (обсудить/вклад) 23:58, 30 November 2014 (UTC)
 * The template was ready to use when I was finished with the triptote and sound plural inflections. All that remained after that was to just add more inflection types to it. I'd like to encourage you to add the template to as many entries as possible, and add to the rest. —CodeCat 00:02, 1 December 2014 (UTC)
 * I think we should have diptotes and irregulars there. Normally the construct form is not required but for nouns like it's needed, e.g. nominative construct is, genitive: , accusative . Yes, we should add more declensions but covering all exceptions first is important as the module/template and the documentation are incomplete.  --Anatoli T. (обсудить/вклад) 00:16, 1 December 2014 (UTC)
 * I updated some of the documentation at Template:ar-decl-noun/documentation. It seems that constructs of words like are not yet supported. I'm going to add support for it. --WikiTiki89 00:36, 1 December 2014 (UTC)
 * ✅. --WikiTiki89 00:45, 1 December 2014 (UTC)
 * Thanks! I guess, I'll have to do more Arabic now that most templates are done. --Anatoli T. (обсудить/вклад) 00:52, 1 December 2014 (UTC)
 * Can you create entries for and  and ? I can only guess at their meanings based on Hebrew: father-in-law, mouth, and I have no idea for the last one. Don't worry if the declension does not work yet for the latter two, I will fix it. --WikiTiki89 00:59, 1 December 2014 (UTC)
 * I'll make some when I have access to my books, so that I don't make mistakes and cover all the senses, etymology (roots) and use correct forms. Hans Wehr has all these words and they are in all grammar books. I've added to WT:RE:ar since I'm not confident without all the books I need but I will address it later, if nobody does. --Anatoli T. (обсудить/вклад) 01:08, 1 December 2014 (UTC)


 * I guess what's next is to support nouns with multiple plurals by allowing an indefinite amount of plurals to come after the singular in the template. --WikiTiki89 05:18, 1 December 2014 (UTC)
 * Indeed. Benwing (talk) 05:31, 1 December 2014 (UTC)
 * BTW, means something like "owner (of)" or "the one (with)", and always appears in the construct state. It is cognate with the Hebrew demonstrative zu. It is related to the Arabic demonstrative pronouns, e.g., which is composed of hā- "this" (cognate with Hebrew ha- "the") plus generic demonstrative ḏā (cognate with Hebrew zeh "this"). Benwing (talk) 05:42, 1 December 2014 (UTC)
 * Cool, so did I guess right on the other ones? Is simply the construct state of, or an alternative form? And if  is the construct state of , then how do you explain ? And what is the ʾiʿrāb ending of ? Also  mentions a "sixth" one  that I don't recognize either. As for the template, this means we need to incorporate irregular and missing states. --WikiTiki89 06:14, 1 December 2014 (UTC)
 * Indeed, you did guess right. I don't know what the deal with is. I've thought of this one before and tried to figure it out; my guess is it's a kind of pseudo-Classical Arabic that's influenced by the dialects, which have lost the forms like . Cf. Egyptian Arabic fumm "mouth", including in construct.  is a normal triptote outside of the construct state, so it takes tanwīn endings, triptote definite endings, etc.
 * As for, this is but I can't find any indication in the dictionary or grammar books that this noun has a lengthened construct state. A sixth noun that is irregular is , which has accusative  and genitive . Benwing (talk) 07:49, 1 December 2014 (UTC)


 * I have added and . As Benwing said,  is the construct form of .  and  need inflection tables.  is currently in Category:form_of/unknown. The full paradigm for  is on "A Student Grammar of Modern Standard Arabic", by Eckerhard Schults, p.134. --Anatoli T. (обсудить/вклад) 11:55, 1 December 2014 (UTC)

Proposed modifications to Template:ar-decl-noun
I'm proposing to change the way we indicate singular and plural declensions in Template:ar-decl-noun so that instead of e.g. we write. The reason for this is that in the large majority of cases, the declension type is automatically inferable (inferrable?) from the ending of the noun, so that in practice we could just write, or with all the plurals given (note, the arguments appear reversed due to R2L issues); the use of a prefix such as tri: will be fairly rare. The alternative with specifying the declension type as a separate argument would require writing something like, which is easy to mess up, especially with the L2R/R2L issues. This would also make it possible to specify something like with the declension type alone given, and the form inferred from the singular. The potential ambiguity between declension type and form is easily resolved since the former is always Latin and the latter always Arabic.

Also, we will eventually have to support explicitly specifying a manual transliteration to allow for the declension of nouns like, and I'd rather not stick this as yet another argument and require that people write in the common case where neither a declension type nor manual translit needs to be given. (Using colons, the manual translit could be specified e.g. as with declension types inferred, or  with explicit declension types, or  with the plural form and translit inferred.) Benwing (talk) 05:31, 1 December 2014 (UTC)
 * I support this 100%, except that I think it makes more logical sense if the declension type is given after the word, like, although that might get ugly when coupled with transliterations. --WikiTiki89 05:39, 1 December 2014 (UTC)
 * What can be done for transliterations is something like . --WikiTiki89 05:42, 1 December 2014 (UTC)
 * I would use just tr for the singular, and pltr, pl2tr, ... for the plurals. (Or pltr, pltr2, ...?) I still like better the idea of incorporating the translit directly into the argument; if you really think having the declension type after is good, either we can specify manual translit by forcing it to be the third item (two colons after the form if necessary), or use a different separator, like /. E.g. in full form,  abbreviated. Benwing (talk) 07:56, 1 December 2014 (UTC)
 * Maybe even . I guess we'll have to vote on which one is best. --WikiTiki89 08:09, 1 December 2014 (UTC)
 * I'm not so happy with introducing custom syntax for templates this way. It make things harder to learn and remember. We should re-use the existing template parsing if possible. —CodeCat 15:38, 2 December 2014 (UTC)
 * Actually this makes things infinitely more convenient, which trumps the "re-use existing X" philosophy. --WikiTiki89 16:34, 2 December 2014 (UTC)


 * Can we have automatic triptote-diptote detection based on ʾiʿrāb (ḍammatān vs ḍamma)? --WikiTiki89 14:04, 2 December 2014 (UTC)
 * And for sound plurals, so we can do this: --WikiTiki89 14:10, 2 December 2014 (UTC)
 * Yes, all of this is already implemented, with my recent changes. I follow your suggestion of writing the :tri or whatever directly after the Arabic form, which is allowed to have ʾiʿrāb or not. There's a complex auto-detection mechanism that detects diptotes based on ʾiʿrāb if it's present (ḍammatān vs ḍamma), and otherwise according to the known diptote patterns. An example is, which has the following declension declaration: . The first :lc annotation is necessary because this is an irregular "long-construct" noun; the pl. is correctly detected as triptotic. Another example is , where the pl. is of a known diptotic pattern and is hence declined correctly as a diptote, whereas the singular is not, and is declined as a triptote. You could also add the ʾiʿrāb explicitly, which overrides any pattern-based auto-detection. Benwing (talk) 10:49, 3 December 2014 (UTC)
 * I also have now implemented multiple plurals, explicitly specified (potentially multiple) duals, and various irregular nouns (,, , . I've mostly implemented manual translit using a slash to separate Arabic from translit in the form like Wikitiki89 proposed, e.g. , but it needs some work before it works properly. Benwing (talk) 12:17, 3 December 2014 (UTC)
 * I don't like the way you handled irregular nouns. The details of irregular nouns should be specified outside of the module. I was thinking something like this: and, and for single-state nouns, something like this:  and . That would work for everything except for , but that could be fixed with manual overrides for each individual form (which would only rarely be used). Also, I don't like the way multiple plurals are displayed: I think each one should have its own sub-table. In case I sound overly critical, I want to thank you very much for doing all this great work! --WikiTiki89 14:01, 3 December 2014 (UTC)
 * Hmmm. There's some justification to your state= because we will have to deal one way or another with definite-only nouns. The other changes seem more trouble than it's worth given the small number of nouns involved and the need to handle additional irregularities (e.g. has two construct-state possibilities). As for the way multiple plurals are displayed, you're welcome to change the code to make it work this way -- I don't think it will be very hard. (I am running out of time to work on this farther.) Benwing (talk) 18:56, 4 December 2014 (UTC)
 * But is it really only a small number, or are more exceptions going to keep cropping up? --WikiTiki89 01:12, 5 December 2014 (UTC)
 * I think this is an exhaustive list of all irregular nouns; this is all the grammar books mention. Benwing (talk) 22:47, 5 December 2014 (UTC)


 * Since it seems we have completely moved to ʾiʿrāb in the headers and in declension tables, perhaps we could consider displaying the pausal forms as well? Users may want to see "ḡurfa" and "ḡuraf" in or it's the transliteration they are more comfortable with? Or ʾiʿrāb could be reserved for declensions only but pausal forms used in the headword? What do you think (regardless of whether greying out ʾiʿrāb is implemented)? --Anatoli T. (обсудить/вклад) 21:57, 4 December 2014 (UTC)
 * I think the whole purpose of graying out the ʾiʿrāb is to display the non-pausal and pausal forms at the same time. --WikiTiki89 01:12, 5 December 2014 (UTC)
 * I think the headword should have pausal forms. I'm ok with having pausal forms in the headword and ʾiʿrāb in the declension tables. The declension tables only apply to very formal Arabic in any case; normal spoken Arabic doesn't bother. Benwing (talk) 22:47, 5 December 2014 (UTC)
 * Yes, please reconsider. ʾiʿrāb will definitely cause confusion for users who are accustomed to Hans Wehr and pausal forms. My preference is to have pausal forms in the tables. Greying out, if doable, could also be added to headwords, declension tables, not sure about Module:ar-translit for any other case with automatic transliteration. --Anatoli T. (обсудить/вклад) 00:36, 10 December 2014 (UTC)
 * "in the tables" you mean in the headwords? Benwing (talk) 10:50, 10 December 2014 (UTC)
 * No, in declension tables. It won't make sense showing both forms in the headwords with ʾiʿrāb and without it. One or the other should be chosen. --Anatoli T. (обсудить/вклад) 12:42, 10 December 2014 (UTC)
 * OK, I'm confused. What exactly are you proposing? I strongly believe that headwords should have pausal forms (except for nouns ending in -in or -an); this is consistent with other dictionaries (including the exception for -in/-an). Do you agree? Presumably you there should be extra entries as well in the declension tables showing pausal forms? How should these entries appear? Keep in mind that the pausal form mostly eliminates case distinctions but not entirely, so if we want to display all pausal forms we might have to double the number of entries. I suggest instead that we have a "dictionary form" row, which shows the same form as appears in the headwords (based on the nominative, with ʾiʿrāb omitted except for -in/-an). Benwing (talk) 22:14, 10 December 2014 (UTC)
 * Sorry for not getting back earlier. So, the simplest way is just use pausal form (without ʾiʿrāb) in the headwords? OK, I'm fine with that. --Anatoli T. (обсудить/вклад) 23:31, 10 December 2014 (UTC)
 * BTW Wikitiki, if you look at and  you'll see they're no longer defined using special "irregular noun" declensions, but use the regular mechanism, now that 'state=' is supported. Irregular noun declensions dhu and dhat are no more. Benwing (talk) 09:49, 12 December 2014 (UTC)

Rename Module:ar-nouns to Module:ar-decl?
Should we rename this module? The use of "nouns" plural doesn't jive with Module:ar-verb or Module:ar-headword (both singular), but an even bigger issue is that much of the same code will be reused for adjective declension, so it probably makes the most sense to have both types of decl in the same module, with two entry points show_noun and show_adj. Benwing (talk) 12:17, 3 December 2014 (UTC)
 * Prefer Module:ar-nominals in that case. See Module:fi-nominal. —CodeCat 22:38, 4 December 2014 (UTC)
 * Support Module:ar-nominals. --WikiTiki89 01:13, 5 December 2014 (UTC)
 * Should it be singular (Module:ar-nominal)? Benwing (talk) 22:47, 5 December 2014 (UTC)
 * I think plural sounds better, because it's a module that deals with nominals. Compare Module:links and Module:translations. —CodeCat 23:42, 5 December 2014 (UTC)
 * Maybe, but it's in conflict with the naming scheme for Arabic and other languages. We already have Module:ar-verb, Module:ar-prep and Module:ar-headword, all of which are singular. Take a look at Category:Inflection modules and you'll see a predominance of singular terms -- the only plural terms are Module:ar-nouns and Module:se-nouns/Module:se-verbs/Module:se-adjectives, all of which you created in the last month. In comparison, I count 12 modules named *-verb, 5 modules named *-noun (one of which is Module:el-decl-noun, the only module with the word "decl" in its name), and 6 modules named *-adjective or *-adj (or *-adj-irreg) -- all singular. Unless you're prepared to rename all the dozens of modules out there, I think we should stick with the singular. Module:ar-nominal is also consistent with Module:fi-nominal and Module:fiu-fin-nominal. Also, So I definitely think we should go with the singular, by weight of convention. Benwing (talk) 07:06, 6 December 2014 (UTC)
 * But if not for convention, what do you prefer? —CodeCat 14:23, 6 December 2014 (UTC)
 * I'd prefer Module:ar-nominal or Module:ar-decl. Perhaps the latter because it corresponds to the template Template:ar-decl-noun and because of the many existing modules named Module:foo-conj, but I could go either way. Benwing (talk) 12:36, 7 December 2014 (UTC)
 * I've started to add adjective support; could you rename the module to Module:ar-nominal? Benwing (talk) 22:25, 10 December 2014 (UTC)
 * I'd prefer the plural Module:ar-nominals. --WikiTiki89 22:53, 10 December 2014 (UTC)
 * Why? See my comments above -- all the existing modules are named with the singular. Benwing (talk) 05:34, 11 December 2014 (UTC)
 * I have already seen the comments above. The singular-named modules are modeled directly after templates. For template names the singular makes sense because of how templates are used. Modules are not visibly invoked the way templates are and therefore if a module does not directly correspond to a template, then it is not bound by the naming conventions of the template. In this case, the module handles nominals, so it makes sense to call it Module:ar-nominals. I guess I would support Module:ar-nominal if "nominal" were actually the name of a part of speech, but nouns and adjectives are differentiated despite having identical declension logic. --WikiTiki89 19:48, 11 December 2014 (UTC)
 * We're not going to agree. How about I let you win this argument if you agree that headwords not contain ʾiʿrab -un/-u endings? Benwing (talk) 05:14, 12 December 2014 (UTC)
 * That sounds like a deal (as long as the endings are not removed before a declension table is added, which I'm sure you'd agree with). One minor detail though is how to handle taa marbuta. I would have liked -a(t), but that looks kind of bad when the transliteration itself is in parentheses (-a(t)). I also completely disagree with the h in our transliteration of ـَاة. It would be better to use -ā(t), or whatever we may or may not decide to use instead of (t). As for the module name, even you are willing to concede, I would still like to know why you disagree with me. --WikiTiki89 06:43, 12 December 2014 (UTC)
 * I'd prefer to let Benwing have his way, as the current primary editor for Arabic. These discussions over minor things often drag on and are somewhat useless. No offence, Tiki, Benwing will do the job. --Anatoli T. (обсудить/вклад) 05:22, 12 December 2014 (UTC)
 * You can't make decisions based on who is currently editing more. I also don't understand your consonant desire to suppress discussion. --WikiTiki89 06:43, 12 December 2014 (UTC)
 * I really don't much like the use of -a(t) although I see why you might like it. We should normally cite words in their "dictionary form", which should be a single form, commonly one of the forms used when the word stands alone. -a(t) violates this by trying to combine two different forms (indef/def and construct). I like -a because it's pretty much unambiguous, whereas the potential -ā for ـَاة is totally ambiguous and IMO the worst of all possibilities. I'm not sure why you oppose -āh, which follows what Hans Wehr does (both in -a and -āh). The actual informal pronunciation of such words is -āt, and that would be another possibility, although still not totally unambiguous esp. since it looks like a sound feminine plural. But why do you object to -āh? Would you object equally if we wrote -a as -ah? I'm ok with that, although I prefer just -a. (So in summary I prefer -āh then -āt then -ā(t) then -ā, and -a then -ah then -a(t).) Anatoli, what do you think?


 * And BTW I'm ok with not removing endings before a declension table is added.


 * As for the module name, it just looks wrong to me given the way the others are all singular. And I don't really get your distinction about "nominal" not being a part of speech -- it's arguably still a part of speech, just a less-specific one, similarly to how "stative verb" is a more specific part of speech than "verb" but still a part of speech. It's well known in Chinese, for example, that adjectives are really a subclass of verbs, so the whole concept of a single cross-linguistic level of speech parts isn't really defensible. Benwing (talk) 09:08, 12 December 2014 (UTC)


 * I also understand Anatoli's frustration, and I do think I deserve some extra say since I've been doing the majority of the work (although not the ability to simply dictate policy to others). Benwing (talk) 09:45, 12 December 2014 (UTC)


 * After thinking about it, I see there are two ways of looking at it: Either the module should be named after its output (in which case the singular makes more sense) or after the content it handles (in which case the plural makes more sense). Either way is equally valid, so if you feel strongly enough about it then go with the singular. One thing in programming that I have always been undecided on is whether to name an array using a plural or singular. The plural looks better when handling the array as a whole, but the singular looks better when accessing individual elements. As for اة, there is no actual [h] sound. We should either use the pausal form -ā or the spoken form -āt. If we use (t), it is not in order to combine the non-construct and construct forms, but just as a way of representing the taa marbuta itself. --WikiTiki89 21:02, 12 December 2014 (UTC)
 * There is an [h] sound; it's the traditional pausal pronunciation. Why do you think tāʾ marbūṭa is in the form of a letter h with the two dots of tāʾ on top? It's because [h] is the traditional pausal pronunciation. That's why you frequently see feminine -a represented as -ah as well. IMO the h is a perfectly fine representation of tāʾ marbūṭa. Both -ā and -āt are ambiguous with other declensions, that's why Hans Wehr uses -āh. Benwing (talk) 08:45, 13 December 2014 (UTC)

Template ar-decl-adj
Template now works. It's designed to take parameters similar to the headword template. Generally if you just copy the template and change the name, you should get the correct declension (provided in most cases that you specify the masculine plural; otherwise you'll get the sound masculine plural in -ūn, which often isn't correct). Benwing (talk) 00:09, 18 December 2014 (UTC)
 * Do you have working examples for mod and idafa therein? The documentation of Template:ar-decl-adj claims the parameters to be present and working like in the noun declension template, so not giving separate examples (perhaps because they are rare in the actual language …), but the things which worked with ar-decl-noun did not work with ar-decl-adj at, so you see I just used the noun template as a workaround. It is that idafa at the adjective template actually recognizes certain values as valid while giving errors for others (made up ones etc.) but when it is valid the modifier is just omitted. (In this case, I tried to use the whole table of plus  as a modifier.) Fay Freak (talk) 20:04, 26 June 2021 (UTC)

Lots of new features
,, : I've added a whole bunch of features to the declension tables, and created declension tables for , , and. Among these features is something Wikitiki has wanted for awhile: You can override arbitrary forms, and as a result I've eliminated all the special-cased irregular nouns. I've also added an additional "case" called "informal", used in speech, and made the lemma display in the head of the declension table, and added the param  to suppress the otherwise automatic inclusion of the definite article in the definite state (e.g., , and  are all considered definite-only but either don't have an article or have it already embedded). You can already control which states and numbers display, and this should be enough for proper nouns, definite-only nouns, etc. I'm about to vanish from Wiktionary for awhile (RL issues), so if you have any comments, make them now. Benwing (talk) 08:27, 30 December 2014 (UTC)
 * I like the new format of nouns and adjectives, well done! It's good that you have added the "Informal" row. --Anatoli T. (обсудить/вклад) 22:08, 30 December 2014 (UTC)

المملكة العربية السعودية
When using in المملكة العربية السعودية the parameter |mod2head2= is not properly handled. Please see the entry. فين أخاي (talk) 20:09, 14 September 2020 (UTC)
 * Looking into this. It's been ages (relatively) since I wrote this module so I have to re-familiarize myself with it. Benwing2 (talk) 01:09, 15 September 2020 (UTC)

خطء, دفء and the like.
Hi, there is a spelling error in the declension of these two nouns (and the like). The hamza should be written on the in the accusative indefinite form. — فين أخاي ( تكلم معاي · ما ساهمت ) 18:09, 19 December 2020 (UTC)
 * Are you sure about this? I looked into all this stuff in detail before implementing the module, consulting four books (John Mace "Arabic Verbs and Essential Grammar" https://www.amazon.com/Arabic-Verbs-Essential-Grammar-John/dp/0844226858, Haywood and Nahmad "A New Arabic Grammar of the Written Language" https://www.amazon.com/New-Arabic-Grammar-Written-Language/dp/085331585X, Raymond Scheindlin "201 Arabic Verbs" https://www.amazon.com/201-Arabic-Verbs/dp/0812005473, and Kristen Brustad "Al Kitaab fii Ta'allum al-'Arabiyya" https://www.amazon.com/Al-Kitaab-fii-Taallum-al-Arabiyya-Beginning/dp/1589017366). I wrote up the results here: . Per these rules, the hamza should be on the line. It's possible these rules represent an older norm that's no longer strictly followed, but in that case we should at least list the normative forms as alternatives. Benwing2 (talk) 21:23, 19 December 2020 (UTC)
 * I agree with Benwing2. As far as I know, tāʾ marbūṭa and hamza don't get an alif in the accusative indefinite and ـً is written over them. I couldn't find an additional specific reference, though. --Anatoli T. (обсудить/вклад) 23:02, 20 December 2020 (UTC)
 * Actually, what about the case with ? Isn't it governed by the same rule or I am missing something? Please ignore my previous post, I am not sure any more, until we find a definite answer. --Anatoli T. (обсудить/вклад) 00:40, 21 December 2020 (UTC)