RE: [Ltru] To invert or not to invert?

On 12/10/06, John Cowan <cowan at ccil.org> wrote:

Mark Davis scripsit:

> You may only be concerned with English,

Not what I said.  I am concerned with the LSR.

> I would want to find "Old Church Slavonic" listed under
>
> Old Church Slavonic
>  AND
> Church Slavonic, Old
>  AND
> Slavonic, Old Church

I think the third is very misleading, and would not wish to see it.
The language is the old variety of Church Slavonic, not the old-church
variety of Slavonic.  This is precisely why inversion cannot be done
mechanically.

> As Kent says, the inversion may be very different in other languages,
> so one MUST be able to extract the uninverted name mechanically.

I agree absolutely, which is why I want to see the inverted names used --
there is no method of inverting the uninverted form.

> 1. List both inverted and uninverted forms under the same tag; also
> ensure that the uninverted is first,

I continue to believe that order of field instances should never be
treated as significant.

And I continue to believe that they must be. When translating, we simply *can't* translate all of the lines, so I don't see any choice but to use the first one as the main description, and trust that that is a reasonable choice. You can make lives more difficult for users if they are in random order, that's all.

> 2. List the inverted forms under a separate field,

I could live with this.

> 3. Only list the (really "an") inverted form

This is what I favor.  I doubt that there are many languages, if any,
that could be inverted in more than one way with appropriate semantics.

> In all cases, we should verify that comma outside of parens *only*
> occurs in the case of inversion.

That has already been done by the Official Doug.

--
We call nothing profound                         cowan at ccil.org
that is not wittily expressed.                  John Cowan
        --Northrop Frye (improved)