Module talk:la-verb

Future infinitive of deponents

Currently the future infinitive of deponent verbs ends in -um īrī. See for example hortor, vereor, loquor, mentior and orior. But according to Gildersleeve & Lodge 1895, §128, the future infinitive of deponent verbs should end in -ūrus esse. Allen and Greenough and Bennet confirm this. Please note that for non-deponents (e.g. amō), semi-deponents (e.g. fīdō) and others (e.g. ōdī) the future (active) infinitive already ends in -ūrus esse. Leen 94.212.253.98 13:09, 4 May 2015 (UTC)

@Kc kennylau This user is quite right. —JohnC5 19:04, 4 June 2015 (UTC)

@94.212.253.98, JohnC5 Thank you. Modified in accordance. --kc_kennylau (talk) 23:01, 4 June 2015 (UTC)

Some change to the apocope

I suggest the way of computing apocope be changed,

from

• apocope = mw.ustring.gsub(typeinfo.pres_stem,"[bcdfghjklmnopqrstvwxz]+$","")

to

• apocope = mw.ustring.gsub(typeinfo.pres_stem,"([bcdfghjklmnoprstvwxz]+|qu)$","")

since "qu" serves together as a single consonant.

I also wonder if verbs like scribo-scripsi should be in Category:Latin third conjugation verbs with perfect in -s- or -x-‎.

Huhu9001 (talk) 19:22, 8 June 2015 (UTC)

Maybe I should call for someone. [subst’d {{ping}} removed — I.S.M.E.T.A. 13:24, 10 June 2015 (UTC)]

Huhu9001 (talk) 09:41, 10 June 2015 (UTC)

@kc_kennylau Any thoughts? — I.S.M.E.T.A. 13:25, 10 June 2015 (UTC)

@I'm so meta even this acronym I'm not sure this is technically possible... --kc_kennylau (talk) 13:27, 10 June 2015 (UTC)

@kc_kennylau: Well, the suggested code means virtually nothing to me. Since q is always followed by u in Latin, does this even make a difference, in practice? — I.S.M.E.T.A. 13:30, 10 June 2015 (UTC)

@Huhu9001, I'm so meta even this acronym Modified accordingly. --kc_kennylau (talk) 13:42, 10 June 2015 (UTC)

To  I.S.: See coquo - coxi, and notice that u was not in "bcdfghjklmnopqrstvwxz".Huhu9001 (talk) 13:55, 10 June 2015 (UTC)

@kc_kennylau: Oh. Thanks. :-)  — I.S.M.E.T.A. 13:52, 10 June 2015 (UTC)

@I'm so meta even this acronym Concerning your question, concoquo was in the category of Category:Latin third conjugation verbs with irregular perfect because "concoqu" was viewed as the stem. My program deletes the consonants at the back, so for it to be categorized in Category:Latin third conjugation verbs with perfect in -s- or -x-, it has to be "concoqux-". Now, "qu" is viewed as one consonant, so it is "conco" + "x" -> "concox-" instead, which puts it in the right category. --kc_kennylau (talk) 13:56, 10 June 2015 (UTC)

@kc_kennylau: Ah, now I understand. Thanks for the explanation. — I.S.M.E.T.A. 13:59, 10 June 2015 (UTC)

+ fut. act. inf. of sum: fore

Could someone Lua-savvy please add fore as another future active infinitive form of sum (“I am”) to its conjugation, please? It was requested by Ēloquiō at Template talk:la-conj-irreg and L&S states “fut. inf. fŏre for futurum esse, very freq., and so always with partt.; cf. Madv. Gram. § 108”. — I.S.M.E.T.A. 02:13, 9 August 2015 (UTC)
@kc_kennylau? — I.S.M.E.T.A. 02:14, 9 August 2015 (UTC)

@I'm so meta even this acronym: Done. --kc_kennylau (talk) 02:28, 9 August 2015 (UTC)

@kc_kennylau: Thank you kindly. — I.S.M.E.T.A. 03:04, 9 August 2015 (UTC)

Specifying parameters

Why didn't this specify exstatūrus as the future active participle in that conjugation table? — I.S.M.E.T.A. 12:32, 29 October 2015 (UTC)

@I'm so meta even this acronym: Because it was not configured to have overriding. Feature added. --kc_kennylau (talk) 14:27, 30 September 2016 (UTC)

@kc_kennylau: Yay! Thank you. :-)  — I.S.M.E.T.A. 18:57, 2 October 2016 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ Hello again kc_kennylau. Do you know why the overriding feature hasn't worked in this case? — I.S.M.E.T.A. 11:31, 3 October 2016 (UTC)

@I'm so meta even this acronym: Because PEBCAK. --kc_kennylau (talk) 11:41, 3 October 2016 (UTC)

@kc_kennylau: :-D Thanks v. mo͝och. — I.S.M.E.T.A. 12:52, 3 October 2016 (UTC)

Fut. pas. inf. of deponents

@Kc kennylau Currently deponent verbs have a future passive infinitive, for example loquor has locūtum īrī, labor has lāpsum īrī, for has fātum īrī, sequor has secūtum īrī.

But according to Allen and Greenough, §190 c: The future infinitive is always in the active form: thus, sequor has secūtūrus (-a, -um) esse (not secūtum īrī) (emphasis added).

So: shouldn't the future passive infinitive of deponent verbs be empty? Leen 94.215.76.74 13:54, 30 September 2016 (UTC)

@JohnC5: What do you think about this? --kc_kennylau (talk) 13:59, 30 September 2016 (UTC)

@Kc kennylau: Makes sense to me. I think I've wondered this in the past, and I've certainly never seen a deponent verb with a future passive infinitive. —JohnC5 14:58, 30 September 2016 (UTC)

@JohnC5, Kc kennylau Yes, with the only caveat that ultum iri is a bit of a mediaeval set phrase, for example in Richerus (fideles et amicos hortans, ut hoc ultum iri accelerent) or especially regularly in William Tyre, (imperii fines violenter ingressus est, in res eorum violenter desaeviens, soceri necem, uxore stimulante, ultum iri desiderans, for example). I have never seen it used classically, however. Isomorphyc (talk) 16:29, 30 September 2016 (UTC)

• @JohnC5, Isomorphyc: Fixed. --kc_kennylau (talk) 01:17, 1 October 2016 (UTC)

@Kc kennylau: Thanks for looking into this -- Leen 94.215.76.74 17:50, 2 October 2016 (UTC)

Present passive infinitive

As mentioned in this discussion on Leasnam's talkpage, there are indeed obscure poetic present passive infinitives, such as laudārier, vidērier, and dīcier. I'm not sure what to do about these. In the past, we've discussed adding syncopated poetic forms (e.g. amāvistī > amāstī), but decided they appeared to randomly to be added generally. This seems like a similar situation. What should we do? —JohnC5 01:25, 3 October 2016 (UTC)

Pinging User:I'm so meta even this acronym and User:Isomorphyc. --kc_kennylau (talk) 10:34, 3 October 2016 (UTC)

@JohnC5, kc_kennylau, Isomorphyc: If they're unpredictable to add automatically, it seems to me that all we can do is add them manually, like this (once that works, of course). — I.S.M.E.T.A. 11:33, 3 October 2016 (UTC)

My thanks to kc_kennylau for making my solution work. — I.S.M.E.T.A. 12:53, 3 October 2016 (UTC)

Is there a way to add a note mentioning the presence of these forms? —JohnC5 15:07, 3 October 2016 (UTC)

@JohnC5, kc_kennylau, I'm so meta even this acronym: I don't know if this is helpful, but I was able to find 116 apparently attested forms and made a list here: User:Isomorphyc/Sandbox/Some_Attested_Latin_Poetic_Present_Passive_Infinitives. I think @Metaknowledge might be interested in this discussion too. Probably adding an argument such as p3inf=1 to the template might be a little bit cleaner than adding the forms to the templates as freeform text, though I have never really worked with the conjugation modules before. Edit: The usual caveats about problematic local corpora and stemming without parsing apply. Isomorphyc (talk) 16:31, 3 October 2016 (UTC)

@Isomorphyc: Is the variation always a change from -rī to -rier? — I.S.M.E.T.A. 16:53, 5 October 2016 (UTC)

@Isomorphyc: Also, what does “The usual caveats about problematic local corpora and stemming without parsing apply.” mean? I'm far from the most tech-literate around here… — I.S.M.E.T.A. 16:56, 5 October 2016 (UTC)

@I'm so meta even this acronym: Please see Hale and Buck, 164.3. There are a variety of poetic forms in Latin in general, of course; I haven't got round to making a full catalogue, though I would like to. About the early/poetic present passive infinitive mentioned here in particular, I can't think of any others offhand. However, at the margins I tend to be unsystematic about guessing poetic (and in Greek, dialectical) forms from context when reading. I only spent a few minutes trying to answer the question asked here very specifically. Also, I realised the Perseus stemmer includes these forms, I imagine on an as-attested basis. Here is a list of residuals between that list and mine: User:Isomorphyc/Sandbox/Some_Attested_Latin_Poetic_Present_Passive_Infinitives_Residuals. Perseus has 20 forms which I was missing, and I have 18 forms which Perseus excludes.

I apologise for my convoluted sentence; it was not very well written. What I mean is that I keep a significant but not comprehensive Latin corpus around (similar in size to Perseus, but with various omissions and additions). I used a stemmer to find a list of forms which could match the form being discussed, but I did not do any parsing to ensure a valid syntactical tree exists in which the word has the form (present passive infinitive) implied by the stemmer. That's not likely to be a major issue with this list, because the form is quite distinctive; but it could be, especially if there are textual errors involving bad word boundaries, etc. I apologise if this is all a bit ad hoc. Isomorphyc (talk) 17:56, 5 October 2016 (UTC)

@Isomorphyc: Thank you for the explanations. Given that -ī → -ier is a regular, albeit rare, variation, I would say that justifies a |p3inf= or similar parameter. — I.S.M.E.T.A. 23:13, 9 October 2016 (UTC)

The most common examples of poetic forms of which I am aware are:

• The 2sg. synthetic passives (amāris > amāre, amābāris > amābāre, amāberis > amābere)
• The 3pl. perfect active (amāvērunt > amāvēre)
• Syncopated perfect stems (amāvistī > amāstī)
• And perfect passive participles (repositum > repostum since re.po.si.tum could never occur in dactylic hexameter).

—JohnC5 19:45, 5 October 2016 (UTC)

@JohnC5: Is there an attestation level above which one would want always to include a form, as I believe we already do for amāvēre, etc.? I believe these forms are much more common than the infinitives, though I have not counted them yet. If we include them on an as-attested basis, what is the best way to formulate the template arguments? Something like poetic_forms=2s_synth_pass+3p_perf_act+syncope_perf+perf_pass+inf_pass? It seems too much visually and too little grammatically. Isomorphyc (talk) 20:42, 5 October 2016 (UTC)

@Isomorphyc: Since we already have the 2s and 3p, I might ask for |sync_perf=1 and |inf_pass=1. —JohnC5 02:45, 6 October 2016 (UTC)

@JohnC5: So, the problem is, if we want to do this on an as-attested basis, I was able to find nine types of syncopated forms (-visse, -visti, -vistis, -visset, -vissent, -vissem, -vissemus, -visses, -vissetis) and Perseus adds three more (-vi, -vit, -vimus) in a few cases which are morphologically indistinguishable from the present tense forms. The list of (I think attested) syncopated forms extracted from the Perseus database is here: User:Isomorphyc/Sandbox/Some Latin Syncopated Forms. This is about 346 forms for 185 distinct verbs. Given that the average syncopated verb has only two forms, I would still suggest a lightweight delimited list, as:

• {{la-conj-1st|am|amāv|amāt|syncope=inf,1ss,2pi,2si,2ss}}
• {{la-conj-irreg|eo|ad|syncope=1si}}
• {{la-conj-1st|cēl|cēlāv|cēlāt|syncope=3ss}}
• {{la-conj-4th|aud|audīv|audīt|syncope=inf,1ps,1ss,2ps,2si,2ss,2si,3ps,3ss}}

With the syntax you propose, we could of course give all of the syncopated forms for every verb with any syncopated form. I don't think that's a terrible idea, but neither is simply including all the possible poetic forms whether attested or not. I'm not sure about the best way to approach this.

Also, I added counts for the syncopated and non-syncopated forms. It is unsurprising that the syncopated forms which are identical to present forms receive very high counts; but I am a little surprised that so many others without present tense twins overwhelm their base forms. It is possible I did something wrong, so I include this with these caveats. Isomorphyc (talk) 22:17, 7 October 2016 (UTC)

@Isomorphyc: Thanks for all this research. I'm fine with this option, though of course a footnote should also be added. —JohnC5 02:49, 8 October 2016 (UTC)

Hi @JohnC5, I'm so meta even this acronym, kc_kennylau, I have added a couple of parameters to this module per the foregoing discussion, and also at Talk:tiro about linking Google Books Search. Some issues remain, but this is the summary:

p3inf=1 : This parameter adds present passive infinitive forms in -ier. I have added it for the 123 unambiguous attestations I could find: User:OrphicBot/EditLogs/17November2016 - Poetic Present Passive Infinitive. I added a few more manually deciding amongst partly homonymous verbs by context: vador, misceo, comparo, and amicio.

sync_perf=poet : sync_perf=[y|n|yn] are preëxisting arguments applicable to fourth conjugation verbs' regular formation. Optional perfect syncopation also exists in eō and its prefixed variants. The orthogonal poetic forms apply to verbs which do not normally syncopate. I added this parameter to the 178 verbs attested irregularly to syncopate, mostly in poetry: User:OrphicBot/EditLogs/17November2016 - Latin Syncopated Forms. Reading through the examples, I came to agree with JohnC5's original idea that any verb which syncopates irregularly in any form should be marked as a syncopating verb, and all the syncopated forms should be given. Word families (puto, computo and disputo all syncopate) could inform a lower granularity still; but the forms are still so sparse this is more confusing than helpful.

search=Books : this feature does not work because I cannot get &lr=lang_la in the URL to restrict the universe of books to Latin; and the Google documentation does not imply Latin is supported. If anyone has made this work recently, I would appreciate any help.

search=site:(...) Instead of searching Google Books, searching with search=site:la.wikisource.org or search=site:thelatinlibrary.com are reasonable ways to cover much of the classical corpus. Unfortunately the links come out in five fascicles for verbs because Google restricts queries to 32 terms.

rudimentary footnote facility: For the two kinds of poetic forms, I added footnotes as suggested. I implemented a rudimentary footnote system modelled after Module:la-noun. The implementation is not currently wholly general.

minimal refactoring: I split the entry point into two in order to route the search queries separately (those functions are at the end of the file). Now the data structure is made in a separate function a couple of lines away. All of the other changes I made are in two if-statements at the end of the postprocess function.

For the time being, some usage examples are here: User:Isomorphyc/Sandbox3. I hope these changes are helpful. Please let me know what I can change to make anything more useful or conformant. Thanks! Isomorphyc (talk) 01:48, 18 November 2016 (UTC)

Edit: the two edit lists do not wholly reflect the current state of edits, but rather the original intended edits. The actual edits involved running and reverting one of the operations on the wrong list, some omissions and manual edits for ambiguities and unparsable wikitext, and a removal of sync_perf=poet from all prefixations of eo, and seven fourth conjugation verbs, since these are handled separately. I apologise for the messy process. Isomorphyc (talk) 01:57, 18 November 2016 (UTC)

@Isomorphyc: Thanks for instituting the |search= function. Two things:
1) Could you make |search=bgc and |search=b.g.c. aliases of |search=Books per {{b.g.c.}}, please?
2) Rather than producing a result like “Google Books forms of cohaerēscō : 1 - 2 - 3”, would it be possible to link each form individually from a visually identical conjugation table instead? I was thinking of something similar in appearance to {{la-decl-adj-table-m+f+n bgc}} / {{la-decl-1&2 bgc}}. As those three queries currently stand, they will produce inaccurate results (see Talk:auroleus#RFV discussion: October 2013–June 2014, specifically my post timestamped: 20:34, 13 November 2013); the queries need to be separated into one for each anisomorphic form. Also, re using the Boolean operator “OR”, note that “an OR unbookended by search terms in a search query causes that query to return fewer results than it otherwise would”, so you may wish to use | (pipes) instead (see Talk:auroleus#RFV discussion: October 2013–June 2014, specifically the thread involving Wikitiki89 and me, starting with my post timestamped: 20:34, 13 November 2013, and ending with Wikitiki89’s post timestamped: 17:40, 15 November 2013).
 — I.S.M.E.T.A. 18:16, 27 November 2016 (UTC)

Hi @I'm so meta even this acronym: Thanks for taking a detailed look at this. Also, apologies for the erroneous indentation in the previous note. To address each issue, the BGC aliases are easy to make. The table is a little bit more work, but I think it can also be done; I would assume the search-all function would be desired either as an option or a separate set of links. I believe that since I never use OR as an initial or terminal term, there is no need for the pipe syntax.

I have a few other questions. I was not able to use the Google Books syntax to limit the corpus to Latin. If the syntax I am using does not work, I do think the only simple option to restrict to Latin works would be to write ("est" AND "sum") AND (x OR y OR z) or similar (assuming the parenthesis syntax works as expected). Are there other ways to do this?

My view is that a longer term project could be to OCR archive.org to extract the Latin ourselves, perhaps putting it in Wikisource. To do this accurately of course is significantly more work than we are doing now.

I will probably make these changes after I have finished working on another set of modules I am editing; but I sholud be able to do these not immediately, but reasonably soon. Isomorphyc (talk) 18:44, 27 November 2016 (UTC)

@Isomorphyc: Yes, feel free to add a search-all function; I just feel I should warn you that I felt my work was wasted when I added such a function to {{la-decl-adj-table-m+f+n bgc}} / {{la-decl-1&2 bgc}}. (Consider this number-of-hits discrepancy: (google books:"sum" [5,290K] + google books:"es" [6,730K] + google books:"est" [4,130K] + google books:"sumus" [453K] + google books:"estis" [379K] + google books:"sunt" [980K]) = 17,962K ≫ google books:"sum" OR "es" OR "est" OR "sumus" OR "estis" OR "sunt" [7,330K].)
Re “us[ing] the Google Books syntax to limit the corpus to Latin”, it might be worth asking Google to institute this feature. Adding other terms to a search query will inevitably cause the omission of valid texts (AFAICT, all of a query's conjunctive terms must occur on a single page for that text to be included in a query's search results); I would hope that those other terms would be omitted by default. Latin often omits sum and its forms by ellipsis, so forms of that verb may not be the best choice for the purpose you intend; I would imagine that common prepositions like ad, in, dē, and ex would turn up more often than forms of sum. I have a copy Mark A.E. Williams' Essential Latin Vocabulary: The 1,425 most common words Occurring in the actual writings of over 200 Latin authors that could inform our speculations with some statistics; if you're interested, I could dig it out.
Re text-searching Archive.org's Latin corpus, they do their own OCRing (though, I grant you, it is not what I would call "accurate"). See these 66,809 results for starters. I don't know why we'd bother to OCR those texts ourselves and then add them to Wikisource, when Wikisource (like all the Wikimedia projects) has built-in problems with reliability and verifiability.
Thank you for all your hard work. — I.S.M.E.T.A. 01:52, 2 December 2016 (UTC)

@I'm so meta even this acronym: This is very interesting that all of the search terms must be on the same page; you are right this makes Google Books much more difficult to use. For OCR: I don't think Archive.org's Latin OCR is configured in all details for Latin. Normally the quality of OCR depends a lot on frequency lists and conditional frequency lists for n-grams. I think if we make these for Latin, with some post-processing of the results, it should be possible to get closer to modern-language quality OCR results than the results one normally sees in online book repositories. It's a significant task, but not perhaps so large as it sounds; the frequency files are normally only a few megabytes, though they can be iteratively refined. Isomorphyc (talk) 05:28, 2 December 2016 (UTC)

@Isomorphyc: I think we might be writing at cross-purposes. When I wrote that “Wikisource (like all the Wikimedia projects) has built-in problems with reliability and verifiability”, I was referring to the way in which nothing is uneditable in them. I – and I don't think I'm alone in this – will never truly trust something written in one of these projects unless it's backed up by something reliable and very difficult to fabricate; that is why I don't think an archive of OCR'd texts with their sources obscured will do much good for this project's credibility. Now, if we could do what some of Wikisource does, and have “scanned edition[s] used to establish…text[s]” (see, for example, their copy of Keats’ “Ode on a Grecian Urn”), then I think that would be a great idea. — I.S.M.E.T.A. 16:50, 6 December 2016 (UTC)

Passive impersonal forms

@CodeCat, JohnC5 Can you correct "pass-3only" and "pass-impers" to show only 3rd person forms in perfect/pluperfect/future perfect passive indicative and perfect/pluperfect passive subjunctive? -GuitarDudeness (talk) 11:41, 13 September 2017 (UTC)

@GuitarDudeness: the perfect passive infinitive still has to be fixed: "ventus esse" > "ventum esse". --Barytonesis (talk) 13:30, 7 December 2017 (UTC)

@Barytonesis: right you say. I wanted to confirm that. And there is no "ventus", right? But only the neuter form "ventum"... So the ventus page for "having come" should reflect this...or be removed as it has not only this signification... --GuitarDudeness (talk) 16:57, 7 December 2017 (UTC)

@GuitarDudeness: No, I don't think "ventus" exists, and even if it does, it doesn't belong in the conjugation table. Maybe we could find attestations in Medieval Latin/New Latin texts, but I'm not interested in doing that; and the noun ventus will make it harder... --Barytonesis (talk) 12:12, 8 December 2017 (UTC)

conest

@Benwing2 I have just created conest, a tricky little impersonal verb. Would it be possible to make it show only the 3s forms? Currently the 3p forms are also visible. Even better, the verb is only attested in perfect and future tenses, so there's more that could be hidden. But at least can we make it impersonal? This, that and the other (talk) 10:37, 1 January 2022 (UTC)

@This, that and the other In order to make those tenses disappear I probably have to hack the module a bit to make it recognize this verb specially. This is possible but if it's attested only in the perfect and future, what should the lemma be? conest wouldn't be attested. Benwing2 (talk) 01:15, 2 January 2022 (UTC)

Also, per Lewis and Short it looks like only cōnfuerit, cōnfutūrum and cōnfore exist, i.e. even cōnfuit doesn't exist. Is that right? Definitely the entry (under whatever lemma it ends up at) should quote all the extant examples. Benwing2 (talk) 01:19, 2 January 2022 (UTC)

@Benwing2 thanks for this; that subtype wasn't mentioned at Template:la-conj/documentation, so I added it there.

I made the lemma conest because it was all I could make the template display. I don't really have a strong view on what the lemma should be, although Lewis and Short's choice of confore doesn't make a lot of sense for me. Also it has to be noted that L&S are not covering medieval and later Latin, which may (for all I know) use this verb in other tenses. Hence, in cases like this I am a little skeptical of attempts to artificially limit the range of tenses displayed. This, that and the other (talk) 01:32, 2 January 2022 (UTC)

Also while I've got you here, I added two irregular nouns: canephoros and chrysophrys which don't fit any existing paradigm. Does {{la-ndecl}} allow you to manually specify forms using parameters? The documentation doesn't appear to say anything about this. This, that and the other (talk) 01:37, 2 January 2022 (UTC)

@This, that and the other It looks like it does allow you to override individual forms using parameters like |nom_pl= or |abl_sg= for nouns, and |gen_pl_f= etc. for adjectives. Separate multiple alternatives with a slash. See User:Benwing2/test-la-ndecl for some examples; this is an old test page of mine. From looking at the code it appears that you can explicitly specify that a given form doesn't exist by setting its value to - or —. Apologies for the sorry state of the documentation. Benwing2 (talk) 01:56, 2 January 2022 (UTC)

@Benwing2 when you have a moment is there any possibility you could flesh out Template:la-conj/documentation (and the noun one too) with a complete list of available subtypes? I just happened upon another undocumented subtype (or way of combining subtypes?) pass-3only which is needed for dētergeō. I'm doing a lot of work at the moment on adding missing verbs and I'd find it immensely useful to know what options exist in the module. This, that and the other (talk) This, that and the other (talk) 13:48, 5 January 2022 (UTC)

@This, that and the other If we use an unattested form as the lemma, L&S’s choice of “consum” seems better to me than “conest”. The prefix “con-“ usually drops the nasal before a vowel, so it’s questionable whether “conest” would even be the right hypothetical form.-Urszag (talk) 02:11, 2 January 2022 (UTC)

Doubly irregular

It seems that compounds of irregular verbs are being marked as "irregular conjugation, irregular": see abeō. This, that and the other (talk) 05:47, 22 January 2022 (UTC)

And even sum has this double label. Is it on purpose? It just feels weird. This, that and the other (talk) 05:47, 22 January 2022 (UTC)