thanks Bernd for adding the details to what I said. one comment on using eTeX (or friends) > What is needed to make this work is add the assignment of this new > e-TeX register when loading hyphenation patterns. In addition one has > to check the "correct" lccode settings for each language at that time. > > > But this will not solve the problem for standard TeX. For TeX the > current LaTeX approach is fine. that is correct and would resolve the problem that currently forces T1 as the only lc table in LaTeX (but it requires a different code base for language to deal with those patterns). but it wouldn't solve the problem of allowing \uppercase or \lowercase as those act on the input encoding which is not being managed by this eTeX mechanism and you can't chose one lc table during input encoding conversion and a different one during output encoding conversion as there may be no place to gain control again is the TeX primitive \lowercase is used directly in typesetting eg font encoding is T1 input encoding is UTF8 foo \lowercase{BAR} base \par frank