consider using <mod> instead of <qst>

[jonorthwash]

<qst> is not a part of speech tag used in Apertium, and I see no reason to categorise -mI/MA differently from -ǵo/Goy/ku/etc.

[jonorthwash]

@ftyers says that the first argument isn't valid: the tagset page is "is completely out of date and unreliable", and <qst> is widely used.

He also points out that the distributions might be different. E.g., -mI and ya in Turkish have [slightly] different distributions:

• gitmiş miydin
• gitmiştin ya

I don't think the latter is a good argument for classifying them as different parts of speech. Perhaps it justifies different subcategory tags, though?

In any case, we need to discuss the distribution of different modal particles and the tags we use for them. In Kazakh we have these, for example:

%+ау%<mod_emo%>:%-ау # ;
%+ай%<mod_emo%>:%-ай # ;
%+ғой%<mod_ass%>:% %{G%}ой # ;
%+гөр%<mod_ass%>:% гөр # ;
%+ма%<qst%>:%>% %{M%}%{A%} # ;
ма:%~ма QST ; ! "" Dir/RL
ма:ма QST ; ! "" Dir/LR
ма:ба QST ; ! "" Dir/LR
ма:па QST ; ! "" Dir/LR
ма:ме QST ; ! "" Dir/LR
ма:бе QST ; ! "" Dir/LR
ма:пе QST ; ! "" Dir/LR
ма% не:%~ма% не QST ; ! "" Dir/RL
ма% не:ма% не QST ; ! "" Dir/LR
ма% не:ме% не QST ; ! "" Dir/LR
ше:ше QST ; ! ""
ғой:%~ғой MOD-ASS ; ! "" Dir/RL
гөр:гөр MOD-ASS ; ! ""
шығар:шығар MOD ;

In Kyrgyz, we have

%+беле%<qst%>:% беле # ;
%+бекен%<qst%>:% бекен # ;
%+чы%<qst%>:%>ч%{I%} # ;
%+го%<mod_ass%>:% го # ;
%+да%<mod_tru%>:% да # ;
%+дыр%<mod_ind%>:%>%{D%}%{I%}р # ;
%+имиш%<mod_dub%>:% имиш # ;
бы%<qst%>:%~бы # ; ! Dir/RL 
бы%<qst%>:бы # ; ! Dir/LR 
бекен%<qst%>:%~бекен # ; ! Dir/RL
бекен%<qst%>:бекен # ; ! Dir/LR
го%<mod_ass%>:%~го # ; ! Dir/RL
го%<mod_ass%>:го # ; ! Dir/LR
чы%<qst%>:чы # ; 
сыяктуу:сыяктуу POST-DECL ;

And окшо as just a verb.

[jonorthwash]

I think this should probably apply for <emph> too. See https://github.com/apertium/apertium-chv/issues/17 .