Template talk:sw-conj/sandbox

We're getting started

@Metaknowledge: This is a start, and it is clearly not done yet. Could you give comments? —JohnC5 20:59, 26 March 2017 (UTC)

@JohnC5: Thank you, I'm excited to see the templates so close to deployment! -Leta is irregular with respect to the imp. sg.; it seems it's correct by mistake (if it were a regular verb, it would simply be the stem). Also, the imperative of monosyllabics keeps the initial ku-. Why are the tables of such different lengths, and is there are a way to prevent the ugliness of words having their last letter or two cut off and shoved to the next line? I was unclear in my writeup (I'll fix this), but for some of the forms with multiple options, you only chose the less standard one where both should be given (footnotes 1 and 2 in the TAM markers section). The negative subjunctive uses the positive concords, rather than the negative ones. I think that's all of the issues, though! —Μετάknowledge^{discuss/deeds} 02:53, 27 March 2017 (UTC)

@Metaknowledge: Ok, I'll fix these ones, and tell me when the write up is fixed. As to table sizes, I'll try and figure that out. The text-wrapping problem I'm still working on. I may need some help with that one. —JohnC5 03:08, 27 March 2017 (UTC)

@JohnC5: I clarified those points already; your confusion is helping it become a better writeup. (And please feel free to edit, reörganise, or beautify it if you have any ideas.) —Μετάknowledge^{discuss/deeds} 03:11, 27 March 2017 (UTC)

@Metaknowledge: Also, based on the other negatives in -si-, I'd expect "if not", -sipo-, to take positive concords. Is this not the case? —JohnC5 03:18, 27 March 2017 (UTC)

@JohnC5: Crap, how did I miss that? You're right. Before we deploy, I'll show these tables to one of the native or near-native speakers I know so they can make sure nothing else is amiss. —Μετάknowledge^{discuss/deeds} 03:33, 27 March 2017 (UTC)

@Metaknowledge: And what about the general relative negative? Should this take positive concords as well? —JohnC5 03:45, 27 March 2017 (UTC)

@JohnC5: Yes, it should. Oh, and I forgot to add that "if not" needs -ku- before the stem for monosyllabics. —Μετάknowledge^{discuss/deeds} 03:48, 27 March 2017 (UTC)

@Metaknowledge: Fixed. I'll work on the other optional stuff now. —JohnC5 03:55, 27 March 2017 (UTC)

@Metaknowledge: Ok, I think I've gotten everything. Tell me if anything is missing. Also, does hamu- undergo the same shortening that mu- → m- does? —JohnC5 04:15, 27 March 2017 (UTC)

@JohnC5: Yes, but people don't usually reflect that in writing, in my experience, and there are some phonological rules conditioning it that I need to add... I think it's actually probably better to leave that out for both mu- and hamu-. Anyway, everything looks good except for the known problems with text-wrapping and autocollapsing. I think I'd prefer if the conjugation type weren't given in the header. Also, how will it work for -enda with displaying both a normal and a monosyllabic conjugation table? —Μετάknowledge^{discuss/deeds} 06:00, 27 March 2017 (UTC)

@Metaknowledge: Oh, I missed the fact that -enda should be both. How would you like to handle it?

• I can code the exception manually with a note.
• I can remove the monosyllabic exceptions -enda and -isha then give you a |mono=1 param.
• I can leave the monosyllabic exceptions -enda and -isha then give you a |no_mono=1 param.

Also, are there any categories you'd like this to generate? I think having some categories for the monosyllabic and Arabic (and irregular?) verbs would be nice (I like categories). I don't know how to name the categories, though. —JohnC5 06:24, 27 March 2017 (UTC)

Any of those work! The second option is the most intuitive to me, but you can choose whatever you like. Categories are fine, though I feel a bit uncomfortable due to the fact that I'm in the territory of using my personal terminology, since the grammarians haven't really agreed on any. Having a category for monosyllabic verbs is obviously odd when two of them aren't, strictly speaking, monosyllabic. Similarly, having a category for irregular verbs is odd when they may be regular for most speakers. —Μετάknowledge^{discuss/deeds} 06:48, 27 March 2017 (UTC)

Could I entreat you to come up with names for the categories? I'm going to sleep now. Also, where are we going to list derivative verbs? I'd completely forgotten about them. —JohnC5 06:53, 27 March 2017 (UTC)

I can't think of good names... maybe CAT:Swahili verbs in the monosyllabic conjugation? (And one for Arabic, but not one for normal.) As for derivative verbs, I've got some ideas — I'll start a discussion elsewhere on that. —Μετάknowledge^{discuss/deeds} 06:55, 27 March 2017 (UTC)

@Metaknowledge: So are we hot to trot? —JohnC5 02:58, 28 March 2017 (UTC)

@JohnC5: Pretty much! Text-wrapping remains an issue; at the very least, stuff on the side like "positive" is of constant length, so that should be possible to have on one line. As for the rollout, I'll have to do some of it by hand, but I guess the rest could be botted. Also, I want to get someone to look it over, hopefully in the next week. Did my ping at Module talk:sw-derivations not go through, or are you putting that one off for now? —Μετάknowledge^{discuss/deeds} 03:14, 28 March 2017 (UTC)

@Metaknowledge: The text stuff I'm not super worried about at the moment since it can be fixed later (if not in this table format, then by just substituting in a different table format whenever we want). In terms of rollout, except for irregular cases, it should just be slapping this baby down on all the verb lemma pages. I'm sure some bot could be conscripted, if not @CodeCat's. I did receive your ping for derivations, but I'm allowing myself a little Swahili coding break. Also, I implemented the |mono= override parameter, which should make verbs that aren't obviously monosyllabic behave like monosyllabic verbs. Also, in case you missed it, the irregular kw- verbs isha, enda, and anza are directly coded in, so they should behave correctly. I have done nothing with the copula at the moment. If you want I could code all the weird forms in for that one, I guess, but not now. —JohnC5 04:36, 28 March 2017 (UTC)

@JohnC5: You certainly deserve a Swahili coding break after all this! I was just worried that my formatting had nullified the ping. There's a mistake in the general relative of the monosyllabic verbs: the way it should be is that the positive form lacks a -ku- and the negative form has it. (This is the sort of thing that makes me feel that I've got to have someone else take a look before rollout.) And the copula is really a mess... I think it's too complicated to be supported by this table, unfortunately, but I'll work on figuring out exactly what needs to be done with it. —Μετάknowledge^{discuss/deeds} 04:42, 28 March 2017 (UTC)

@Metaknowledge: I fixed the general relatives for monosyllabics, maybe? —JohnC5 04:45, 28 March 2017 (UTC)

@JohnC5: Yep, you did it right. The negative present and past conditionals need -ku- as well... I realised I've been getting that wrong. I'll make a table to explain this more clearly in the appendix. —Μετάknowledge^{discuss/deeds} 05:06, 28 March 2017 (UTC)

@JohnC5: Also, I tried it out in the mainspace at -weza, and the category didn't behave right. —Μετάknowledge^{discuss/deeds} 06:13, 28 March 2017 (UTC)

@Metaknowledge: That's my bad. Fixed. —JohnC5 06:25, 28 March 2017 (UTC)

@JohnC5: Thanks. I also noticed that because the imperative is usually the same as the stem, it gets bolded... that's not a behaviour we want, right? —Μετάknowledge^{discuss/deeds} 06:50, 28 March 2017 (UTC)

@Metaknowledge: I did it intentionally so that there is no self-linking, but we could change it if you want. I think in every other inflection template I've worked on we've opted to prevent self-linking. —JohnC5 06:54, 28 March 2017 (UTC)

@JohnC5: Could just not have a link, though, right? But it doesn't matter, so whatever fits the standard around here, I guess. —Μετάknowledge^{discuss/deeds} 07:11, 28 March 2017 (UTC)

• @JohnC5: The negative present and past conditionals for the monosyllabics never got fixed! They need -ku- right before the stem. —Μετάknowledge^{discuss/deeds} 16:34, 28 March 2017 (UTC)
  • @Metaknowledge: ...Yes they did. Where are you seeing this? And to be clear you mean the negative present and the positive or negative past conditional? —JohnC5 17:26, 28 March 2017 (UTC)
  • @Metaknowledge: Oh, do you mean the -ngali- form of the negative past conditional needs a -ku-? —JohnC5 17:42, 28 March 2017 (UTC)

    @JohnC5: Both -ngali- and -nge- forms, in the negative only, need -ku-. —Μετάknowledge^{discuss/deeds} 17:46, 28 March 2017 (UTC)

    @Metaknowledge: Ok, to be clear, I have just added the -ku- to the present negative, and the non-si- forms of the present and past negative conditionals. —JohnC5 17:50, 28 March 2017 (UTC)

    @JohnC5: Sorry, this is on me; I haven't been very good about communicating this. The -si- forms need it too. —Μετάknowledge^{discuss/deeds} 17:53, 28 March 2017 (UTC)

    @Metaknowledge: Oh really? Ok, easy enough. The current list of -ku- forms are:
    past (p, r), pres (p, r, n,), fut (p, r, n), cond_pres (p, n x 2), cond_past (p, n x 2), cond_cond (n), perf (p), alr (p), ifn (n), rel (n).
    —JohnC5 18:01, 28 March 2017 (UTC)

    I had some trouble with this (compounded by the fact that many Swahili speakers have trouble with it too), but the rule comes down to the fact that only the morphemes ku, ki, ka, si, hu, a, ja, and subject/object concords can be in the penultimate syllable for monosyllabic verbs. —Μετάknowledge^{discuss/deeds} 18:05, 28 March 2017 (UTC)

@Metaknowledge, CodeCat, I just remembered: Should we add example forms in the toggle bars and if so, which forms? —JohnC5 19:12, 28 March 2017 (UTC)

I suggested putting the bare stem in there, remember? —CodeCat 19:13, 28 March 2017 (UTC)

Rather than the macrostems, which is what I assume you mean, it might be better simply to have the morpheme given, so that it shows Past -li- or something like that. —Μετάknowledge^{discuss/deeds} 19:26, 28 March 2017 (UTC)

But they can differ by polarity. What then? —JohnC5 19:36, 28 March 2017 (UTC)

Then they open up the toggle bars and see the forms! (Or maybe I misunderstood you.) —Μετάknowledge^{discuss/deeds} 19:39, 28 March 2017 (UTC)

I mean that -li- is only the morpheme for the past positive forms and not the negative forms. —JohnC5 20:09, 28 March 2017 (UTC)

Yeah, grammars that use morphemes instead of names will still tend to call that "the negative of the LI tense" rather than give it a separate name, but I see your point. I don't know what's best. —Μετάknowledge^{discuss/deeds} 02:44, 29 March 2017 (UTC)

Bot run

What exactly needs to be done for the bot run? I haven't really followed the discussions much, so can you give a concise list of changes that should be made? —CodeCat 13:41, 28 March 2017 (UTC)

@Metaknowledge will have to correct me if I'm wrong, but I believe the only thing that will be necessary is adding or replacing the inflection section with just {{sw-conj}}. The only things that need to happen before that are to convert the handful of old transclusions of {{sw-conj}} to the new format and perhaps to generate a list of irregular verbs to be added manually. Both of these things I'd ask Meta to do. —JohnC5 14:08, 28 March 2017 (UTC)

I'll putter through those, and then it'll be a simple find-and-replace. I have some extra work coming up in meatspace, but hopefully I can get this done soon, and I'll tell you once I have. —Μετάknowledge^{discuss/deeds} 15:42, 28 March 2017 (UTC)

Looking at it, there's few enough transclusions that this could be done more easily by hand. —CodeCat 16:44, 28 March 2017 (UTC)

I'm doing it by hand. The point is that the template name will have to be switched over once the old one is orphaned. —Μετάknowledge^{discuss/deeds} 16:52, 28 March 2017 (UTC)

Ok, renaming a template is easy, I have a pre-made script for that already. I just input the old and the new name. —CodeCat 16:53, 28 March 2017 (UTC)

Wait a second. Don't change anything over. I'm gonna make it so that you can use {{sw-conj}} as the new template but {{sw-conj-aff}} and {{sw-conj-neg}} will be the old template. That way, you just have to change things to {{sw-conj}}. The bot will be used to just mass add {{sw-conj}} to entries. —JohnC5 17:12, 28 March 2017 (UTC)

Ok @Metaknowledge, please start using {{sw-conj}} in the mainspace and convert over all instances of {{sw-conj/sandbox/template}} to that. All the remaining ones to be converted are at {{sw-conj/old}}. —JohnC5 17:19, 28 March 2017 (UTC)

@Metaknowledge: Thanks for moving everything over. Coul I ask you to write some documentation for {{sw-conj}} (I'm lazy)? The only parameters are |mono=, |1= (the lemma if different than pagename), and all the form codes. Do we want @CodeCat to use a bot to add this template to all the remaining Swahili verbs or hold off? —JohnC5 18:27, 28 March 2017 (UTC)

Sure, I'll write it. And as for bot-adding the template, that should only be done for entries that are not in CAT:Tbot entries (Swahili). —Μετάknowledge^{discuss/deeds} 18:30, 28 March 2017 (UTC)

Relative?

This is purely for my own interest, but what is the row called "relative"? —CodeCat 14:24, 28 March 2017 (UTC)

@CodeCat: So, based on Appendix:Swahili verbs#Relative and Meta's explanation, verbs in relative clauses are marked either in three tenses (positive past relative, positive present relative, positive future relative), or tenselessly (positive and negative). This tenseless relative marking Meta's calling the general relative. —JohnC5 15:29, 28 March 2017 (UTC)

If I compare this to Zulu, in that language you create relative clauses by prefixing the augment (essentially a copy of the vowel in the subject prefix) and then a relativiser particle a-. Thus, mina ngiyabona (“I see”) > mina engiyabona (“I who see”). I'm not sure how it works for other tenses, or for the negative, but I guess the idea is similar. —CodeCat 15:39, 28 March 2017 (UTC)