[YG Conlang Archives] > [katanda group] > messages [Date Index] [Thread Index] >


[Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [katanda] machine translation



"Garrett Jones" <alkaline@hidden.email> wrote:
> 
> So how much "noise" would show up in translations into katanda and
> back into the original language?
>

Obviously, this will depend on the particular item being translated.
Give me a specific example of a natlang subtlety that Katanda cannot
represent without periphrasis, and I'll tell you what I think the likely
outcome will be.

Here's an example off the top of my head:  Some natlangs have
evidentials that indicate the nature of the source of the information.
For example:

    It seems that John left.

where the word meaning 'seems' also indicates that the speaker saw John
leave with his own eyes.  The writer of the translation software
must decide whether to use:

    (a) Xedebe budinca Kajobyo = It seems that John left.
    (b) Pi vezdi mi budinca Kajobyo = I saw John leave.

This is a design decision.  If I were doing a first draft of the
software, I would probably choose (a) because of its simplicity;
otherwise, I'd choose (b).

BTW, I decided not to implement this particular subtlety in Katanda
because so few languages have an equivalent.

The important thing to keep in mind is that ALL of this kind of "noise"
will be the result of design decisions.  You can always do an accurate
translation if you really want to, although the result may be overly
verbose.


Regards,

Rick Morneau
http://www.srv.net/~ram
http://www.eskimo.com/~ram