I am hoping to understand your problem better. You are transcribing Ugaritic texts (using EpiDoc? or tei_all?) in oXygen, yes? And you would like to indicate a Latin tansliteration for each … word? character? sentence? If that is the case, a new attribute may not be necessary (although might still be something you want to do just for ease of encoding), as that sort of 1-to-1 correspondence is exactly what @corresp
is intended for.
E.g. (I know nothing about Ugaritic, so the following example has no meaning):
<w type="adj" ana="silly" correspUnic="">blort</w>
If that is the case, then a standard TEI encoding might be
<w type="adge" ana="silly" xml:lang="ugr-Ugar">
<c corresp="#ctl_b"></c>
<c corresp="#ctl_l"></c>
<supplied xml:lang="ugr-Latn">o</supplied>
<c corresp="#ctl_r"></c>
<c corresp="#ctl_t"></c>
</w>
Where the pointer "#ctl_t" point to a <char>
element in the teiHeader/encodingDesc/charDecl. The (minor) disadvantage here is that you do have to set up a list of <char>
elements to define characters that (for the most part) everyone already knows and are kinda obvious.
Worth noting (with no insult intended to Stackoverflow, which is wonderful), the normal place to post such a question is the TEI mailing list, where hundreds of TEI users, many of whom know much more about epigraphy than I, are likely to see it.
If kind of approach, or some other standard TEI mechanism will not do, or if you really want a @van:latin
attribute just to make encoding faster and easier to proofread, it is not hard to do in ODD, and many folks on TEI-L would not mind stepping you through that.