On 10/3/2011 11:16 PM, Daniel Bowman wrote:
> I have decided that Angosey is not complicated enough after all, and I'm
> going to throw evidentiality into it also.  I'll post the results once I get
> a system worked out.
>
> While thinking about implementing this, though, I started wondering.  How do
> natlangs and conlangs with evidentiality incorporate it in questions and
> subjunctive-type sentences?  For example, how would obligatory evidentiality
> handle the following:
>
> Do you want to go get something to eat?

What I think Tirelat speakers might say in this case is something like

Rĕdaladanezaz my każavit?
r-dala-da-ne-za-z my każa-vit
2s.SG-go.get-DES-EXP.NP-QU-IPF ACC.SG eat-thing

with the tense/evidential suffix -ne (nonpast, direct personal 
experience). In Tirelat though, the tense/evidential suffix is not 
obligatory when a mood suffix like the desiderative -da is used, so 
you'd more likely hear

Rĕdaladazaz my każavit?

If you're asking about someone else (do they want...?), you might use 
nonpast opinion -łi.

I guess the key to questions (at least in a language that works like 
Tirelat) is to think about what evidential the listener would be 
expected to use in an answer. So in a question, all the evidentials are 
defined relative to the listener, not the speaker.