Scripsit Helmut Wollmersdorfer:
A table is IMHO a little bit harder to read, if a term has more than
one translation.
Why? You can put the translations each on a line of its own inside a cell,
or you can have a row in the table for each pair of words so that for a word
with multiple translations, you use rowspan="..." in the first cell.
In practice, the desired visual appearance is important, since you cannot
turn a <dlinto tabular presentation or a table into non-tabular
presentation in CSS. (Tools might exist in principle but aren't supported by
the most common browser.)
<DLis the most appropriate. 'definition' is the usual term for the
right part of a dictionary or glossary. The left part is called
'subject', 'keyword', 'term'.
<dlis possible, since a translation can indeed be seen as a definition,
but it is debatable whether <tableis more logical.
There's a particular practical benefit in using a table. You (i.e., the
generic user) can copy the table into his favorite spreadsheet program
(i.e., MS Excel), which can interpret the table in a way that lets him
reorder it conveniently and e.g. produce a reverse glossary.
But look into TEI. AFAIR they have some XSLT for conversion of
TEI-dict.XML to HTML.
For a complicated dictionary, one might need to look at general structured
formats indeed, but for delivery in HTML format, much of the structure needs
to be lost anyway.
--
Jukka K. Korpela ("Yucca")
http://www.cs.tut.fi/~jkorpela/ 
