Past forms in >BIт>A etc. should be <plu> instead of <past>.

@varie, could you provide some information on each of these forms? They're coming up as common unanalysed forms.

• буолкалах
• буолааччы
• буоланнар
• буоллун
• буолуохтааҕын
• баргыа
• үөскүүр
• сыллардаахха
• суруллубут
• оҥоһуллубут

Most of them look like inflected forms of stems we have in the transducer, but they could also be derived forms. Either way, let's look at each one carefully and think about how to add to the transducer.

Sakha has "complex verb forms" with эбит.

Кэлбит эбиттэр "Они, оказывается, пришли"

Many other Turkic languages have this.

In Turkish, the etymologically identical and semantically similar (dub instead of evid) form is "gelmiş imişler"/"gelmişmişler":

"<gelmişmişler>"
	"gel" v iv past dub
		"i" cop aor p3 pl

In Kazakh, it's equivalent morphologically and semantically to "келген екен":

"<келген екен>"
	"кел" v iv past evid p3 pl

And Kyrgyz

"<келген экен>"
	"кел" v iv past evid p3 pl

There are a number of strategies for dealing with this sort of construction, and they were discussed to some extent on the apertium-turkic mailing list in 2014. A Turkic-wide solution should be found. In the meantime, I'm partial to doing it like Kazakh and Kyrgyz, and fixing everything all at once when a different strategy is decided on.

@varie, what are the negative forms of e.g., барыы, барааһын?

көр + ө +түс, суруйа + -а +оҕус; оҥор + -а(-ыы)+тарт; куттан + -ан+ыл, тыс + гын, малай + с гын, ходьой + х гын; дьүккүй + к гын

I also noticed that "көрсүөххэ" phonology doesn't work but "барыахха" works

Which of the following are passive forms? / What is the analysis of these forms, and the lemma?

• билиниллэр - билин ‘confess’ (?)
  • билиниллибит - (past)
  • билиниллэллэр - бил
• сиэниллэр - сиэ
  • past сиэниллибит, plural сиэниллибиттэр
• ыраастаныллар - ырастаа
• өйдөнүллэр - өйдөө
• өрдөнүллэр -
• ирдэниллэр - ирдээ
• хааччахтаныллар -хаччахтаа
• тарҕаныллыбыт - тарҕаа
• бааныллар - баай

We originally had just a <px3sp> in the lexicon, but it looks like there are distinct singular and plural forms, e.g.
барбытыгар/бар<v><iv><ger_past><px3sg><dat>
барбыттарыгар/бар<v><iv><ger_past><px3pl><dat>

However, some recent tests have forms like
барыахтарыгар/бар<v><iv><ger_fut><px3sp><dat>

Is барыаҕыгар a singular form to this plural form? Are there any possessives that don't have a distinction? Alternatively, are singular and plural readings ever ambiguous?

There are a number of what appear to be verb forms in {B}{I}т{I}н in the corpus. What do these mean / how are these used?

     58 ^буолбутун/*буолбутун$
     32 ^оҥорбутун/*оҥорбутун$
     21 ^кэлбитин/*кэлбитин$
     16 ^киирбитин/*киирбитин$
     15 ^тахсыбытын/*тахсыбытын$
     15  ^суруллубутунан/*суруллубутунан$
     15 ^барбытын/*барбытын$
     14 ^үөскээбитин/*үөскээбитин$
     14  ^сылдьыбытын/*сылдьыбытын$
     11  ^ылбытын/*ылбытын$
     11 ^хаалбытын/*хаалбытын$
     11 ^суруйбутун/*суруйбутун$
      9 ^суруйбутунан/*суруйбутунан$
      9 ^олорбутун/*олорбутун$
      9 ^биэрбитин/*биэрбитин$
      8 ^эстибитин/*эстибитин$
      8  ^элбээбитин/*элбээбитин$
      8 ^тутуллубутун/*тутуллубутун$
      8 ^төрөөбүтүн/*төрөөбүтүн$
      8 ^талбытынан/*талбытынан$

Note the ones with an extra {A}н.

Examples:
барыыһы
кэлииһи

1. According to Gavril Torotoev's table forms like "барбыт этим" are called Pluperfect (PLPF), such form is missing in the transducer.

2. According to Gavril Torotoev's table forms like "барбыппын" and "баран турабын" are called Resultative past tense Perfective 1st and 2nd respectively. Apertium-sah analyses form "барбыппын" as past and ger_past ⋅ px1sg ⋅ acc, "баран турабын" form is missing and those two words are analysed as separate tokens.

Conjugated NPs are not recognised by the transducer:
Иван is analysed,
but Ивана is no not.

бар + -ан+ис, ытаа + -ыы+сырыт, кэл + -э+тур, бар + ылаа, лыҥ + кынаа, ньаа + ҕынаа, баб + ыгыраа, бал + лыгыраа, хаб + дьыгыраа, кур + дьугунаа, кыч + чыгынаа, бычыгынаа + й

көрүө, кэлиэ, барыа (бар-, бараа-)

There are a lot of what appear to be {I}лл{I}б{I}т forms in the corpus:

    799  ^Ылыллыбыт/*Ылыллыбыт$
    158 ^оҥоһуллубут/*оҥоһуллубут$
    134 ^ыҥырыллыбыт/*ыҥырыллыбыт$
    133 ^суруллубут/*суруллубут$
    112 ^тэриллибит/*тэриллибит$
     91  ^ыытыллыбыт/*ыытыллыбыт$
     75  ^Туһаныллыбыт/*Туһаныллыбыт$
     75 ^иҥэриллибит/*иҥэриллибит$
     67 ^көмүллүбүт/*көмүллүбүт$
     48 ^туттарыллыбыт/*туттарыллыбыт$
     47 ^иитиллибит/*иитиллибит$
     44 ^тутуллубут/*тутуллубут$
     40 ^ылыллыбыт/*ылыллыбыт$
     38 ^бэриллибит/*бэриллибит$
     24 ^туруоруллубут/*туруоруллубут$
     24 ^бигэргэтиллибит/*бигэргэтиллибит$
     22 ^айыллыбыт/*айыллыбыт$
     20 ^ылыныллыбыт/*ылыныллыбыт$

There are also a bunch that have an extra {A} at the end:

    179  ^тэриллибитэ/*тэриллибитэ$
     73 ^иҥэриллибитэ/*иҥэриллибитэ$
     58  ^биллибитэ/*биллибитэ$
     55  ^ыытыллыбыта/*ыытыллыбыта$
     49 ^тутуллубута/*тутуллубута$
     37 ^оҥоһуллубута/*оҥоһуллубута$
     35  ^аһыллыбыта/*аһыллыбыта$
     29 ^арыллыбыта/*арыллыбыта$
     24 ^киллэриллибитэ/*киллэриллибитэ$
     21 ^уһуллубута/*уһуллубута$
     21 ^суруллубута/*суруллубута$
     21 ^бэриллибитэ/*бэриллибитэ$

What do these forms mean? They look almost like passive followed by the past {B}{I}т, but they seem to have an extra vowel. Is this all they are? And the ones with {A} are possessives of the past gerund of the passive?

This is a list of generator errors that @varie identified while working on the error analysis for the shared task. These forms were produced by the version of the generator from ~March 2021.

Incorrect inflections:

Other suggestions from @varie's consultants

автомобиль<n> -- suggested to change the lemma to аптамабыыл
гол<n> -- suggested to change the lemma to гуол
наследиенан:наследие<n><ins> -- suggested наследиянан

Which of these forms can be used in which ways? Complete the chart:

I'm finding a lot of forms with -дээх, mostly after №, like this:

27 №-дээх оскуолаҕа

What is this form?

Dear Fran,
After (almost) completing the Uzbek tagging, I started to work on two other Turkic languages: Tatar and Yakut. A rather strange error message, however, appears:
www-data@unesco1:/tmp$ apertium -d /usr/local/share/apertium sah-tagger sah_ar1600y_b2_0.h2~ >sah_ar1600y_b2_0.i2~
Error: Invalid dictionary (hint: the left side of an entry is empty)
I firstly suspected a bug in the lexicon file, but in both lexicons at the same time? Any suggestions?
Have a healthy New Year,
Vlado B, 10:30

from https://www.jehsmith.com/1/2021/03/verbal-mood-in-sakha-a-comprehensive-analysis.html

"Affirmative mood"

Foorms like барыыһы+copula endings, 3 negative forms: барымыыһы+copula endings, барыа суох+copula endings, and барыа+~endings суох.

"Debitive mood"

• Aр>DAAx (should be realised at moment of speaking; must be realised after moment of speech)
• IAх>TAAх (must be completed in the future)
• Aр>DAAx этэ / IAх>TAAх этэ (should have been completed in the past but was not)

"Possible mood"

• ((I)м)>AAйA / ((I)м)>AAрAй
• should have similar tags to Kyrgyz бараармын/барбаасмын

суруй + -ан+кэбис, бар + -ан+хаал; сит + эн ыл; бырах + ан биэр

Opening an issue per @ftyers' request.

I see a lot of syncretic forms in noun paradigms:

тириэрдиилэринээҕэр:тириэрдии<n><pl><px3pl><comp>
тириэрдиилэринээҕэр:тириэрдии<n><px3pl><comp>
тириэрдиилэринээҕэр:тириэрдии<n><pl><px3sg><comp>

Are <px3sg> and <px3pl> always expressed with the same morpheme? If yes, why are they annotated separately and not with <px3sp> like in Tuvan?

• ыарый/ыалдьар

• оҕус/охсор

• араҕыс/арахсар

• сайын/сайдар

• умун/умнар

• үтүгүн/үтүктэр

• эбилин/эбиллэр

• бэрилин/бэриллэр

• иҥэрилин/иҥэриллэр

• иһилин/иһиллэр

• талылын/талыллар

• туһанылын/туһаныллар

• төнүн/төннөр

• кытын/кыттар

• орт/ордор

• тарт/тардар

• тириэрт/тириэрдэр

• сыһын/сыстар

What do the <pii> forms correspond to? The generator gives me

барабыт:бар<v><iv><pii><p1><pl>
бараҕыт:бар<v><iv><pii><p2><pl>
бараллара:бар<v><iv><pii><p3><pl>
барарым:бар<v><iv><pii><p1><sg>
барарыҥ:бар<v><iv><pii><p2><sg>
барара:бар<v><iv><pii><p3><sg>

Most forms match the Ubriatova book paradigm for imperfect past (прошедшее незаконченное время, "unfinished past", pp. 228, 315):

But the generator has барабыт:бар<v><iv><pii><p1><pl> instead of барарбыт and бараҕыт:бар<v><iv><pii><p2><pl> instead of бараргыт. Not sure if this is an error or it is actually a different inflection.

Beginning of imaginary phonology was implemented in e3abeab. Now we need some real morphotactics for numbers, including the % symbol.

Example: сууйун

Example:
хампахтаа

хаамп + -махтаа/-мэхтээ/-мохтоо/-мөхтөө
туур + -бахтаа/-бэхтээ/-бохтоо/-бөхтөө
-пахтаа/-пэхтээ/-похтоо/-пөхтөө

I'm trying to understand the range of things -TAх does. As we discussed on IRC, the information on pp. 236 to 237 of Убрятова suggests there are two main "moods" that use this suffix: второе условное and предположительное. There are also verbal adjective and verbal noun uses.

So my questions are:

1. Are второе условное and предположительное moods both finite?
2. Are there some good examples of each?
3. Could we see a full paradigm of each?

What's буолуон mean? Is this a form of буол? If so, how is the suffix {I}{A}н used?

modes.xml includes some modes with install="yes", but the required
files aren't installed.

Some generic suggestions:

• -lexc and -twol modes probably aren't useful to users

• -spell modes should depend on --enable-ospell

• .deps files are never installed, so any modes using them shouldn't be
  installed.

• Messages for package app-dicts/apertium-sah-9999:

• Failed to find '/usr/share/apertium/apertium-sah/.deps/sah.twol.hfst' in install image.

• QA: missing files required for mode sah-twol.

• Failed to find '/usr/share/apertium/apertium-sah/.deps/sah.LR.lexc.hfst' in install image.

• QA: missing files required for mode sah-lexc.

There are verb forms in -{A}{A}т that are identified by Johanson (2021) as a type of converb. Examples: биэрээт, түһээт, ылаат.

(An English gloss of these forms appears to be "as soon as X Y-ed" or "upon having Y-ed"? In Kyrgyz these semantics are expressed with a VN+possessive ending followed by a postposition, e.g. берээрим менен - "as soon as I gave". So no help to the following questions.)

I guess we need a verbal adverb label (gna_???) for these. Aside from the question of what the label should be, I'm wondering if these might behave as infinitives too (that is, occur with auxiliaries, prc_???). @varie?

бараар, сылдьаар, диэр

Examples:
бараайа/-ээйэ/-өөйө/-оойо
бараайык/-ээйик/-өөйүк/-оойук

күл + эн бар; ытаа + бытынан бар; оҥор + он киирэн бар; абар + ан тур; улаат + ан кэл; ытаа + ан эр

Currently forms like барбыт are analysed as <ger_past><nom>, but it's not clear that they can be used as bare verbal nouns. Forms like барбыта (<ger_past><px3sp><nom>) are fine.

Meaning, distribution

The form "илик" only occurs in verb forms and adjusts the meaning to something like "not yet":

Оҥоро иликпин "Я ещё не сделал"
Оҥоро иликкин "Ты ещё не сделал"
Оҥоро иликпит "Мы ещё не сделали"

Aside from its use in finite verb complexes, it can also be attributive (verbal adjective):

бара илик киһи "A person who hasn't gone yet"

Similar things in other languages

Kyrgyz is the only language where we've implemented something similar:

"<бара элекмин>"
	"бар" v iv notyet p1 sg

"<бара элек>"
	"бар" v iv gpr_notyet
"<киши>"
	"киши" n nom

There's also a gerund reading, in constructions like:

"<бара электе>"
	"бар" v iv ger_notyet loc

and

"<бара элегимде>"
	"бар" v iv ger_notyet px1sg loc

The form бара appears in the transducer as a noun; investigate where this came from and see if it needs to be removed or left.

I'm using .deps/sah.RL.hfst to generate inflected verbs. It generates several <neg><imp><p2><pl> forms that differ only by one vowel. Is this behavior intended?

булкуспааҥ:булкус<v><iv><neg><imp><p2><pl>
булкуспаеҥ:булкус<v><iv><neg><imp><p2><pl>
булкуспаоҥ:булкус<v><iv><neg><imp><p2><pl>
булкуспаөҥ:булкус<v><iv><neg><imp><p2><pl>
булкуспаэҥ:булкус<v><iv><neg><imp><p2><pl>
булкуспаяҥ:булкус<v><iv><neg><imp><p2><pl>

There is a request to treat MWEs and when analysed give an information that it is an MWE along with the each word's analysis.

барыы - ?
барааһын - бараамааһын?

бар + ыс + ан = барсан
суруй + ус + ан = суруйсан
<v><recp><converb_past>

The testing section of the README file needs a good example sentence; ideally one that has some ambiguity that we can resolve, even if fairly simple (like p3.sg/pl). (It doesn't have to be currently resolved.)

Examples:
барыталаа
арартаа

• Various issues in aorist.yaml
  • Especially таҕыс:тах%{y%}с%{ы%} V-IV ; ! "выходить"
• Development process using GitHub issues
• evidential and related stuff

• <ger1> (-IAн) should be <ger_cond> or similar
  • көрүөн, кэлиэн
  • compare to көрүө (<ger_pot>) and көрүөх (<ger_fut>)
• Figure out <ger2> (#14) - maybe <ger>? ("act of ...")
  • барааһын, көрөөһүн, кэлээһин