9.7 The lang and letter-value Attributes

I'll mention a pair of attributes from number that I haven't yet discussed, but without concrete examples. These attributes have to do with the different way numbering is handled in different human languages. Speakers of English and other European languages are accustomed to numbering with so-called Arabic numerals, that is, with the ten digits 0-9. Some languages, such as Hebrew and Greek, use letters from their alphabets as numbers.

The lang attribute, like xml:lang, takes a language token as a value, such as en, fr, de, or es. This language token is supposed to signal to the XSLT processor what language is in use with regard to numbering. Another attribute, letter-value, takes the values alphabetic or traditional. These values are there to help distinguish between language-specific numbering systems that assign numerical values to alphabetical sequences or assign numerical values to letters in a traditional way. Unless you are familiar with a given language, such as Hebrew or Greek, some of these numbering schemes can be elusive.

The specification is somewhat loose in regard to this aspect of numbering, probably because the variations and ambiguities of numbering could make implementing number an overwhelming task. If you need to use traditional numbering, check the documentation provided with your XSLT processor of choice to find out how the processor handles these cases.