Hi everone, To follow up on the segmnentation discussion: Here are some example of how segmentation could be marked up in an item. <trans-unit id="1"> <source xml:lang="en"><mrk mtype="seg">Sentence 1. </mrk><mrk mtype="seg">Sentence 2.</seg></source> </trans-unit> The mtype value 'seg' is not pre-defined (but it could be). A different mechanism would be to mark the break rather than the span: <source xml:lang="en">Sentence 1. <mrk type="seg/>Sentence 2.</source> Obviously in this case a special empty element would vbe better than <mrk>. There are some problems. Mostly mostly that the fact the trans-unit attributes apply to the whole content of <source> not parts of it. So for example a resname (possible unique ID) would apply to several segments. Keeping track of the segments within the sentence would be needed (so we would need additional attributes in <mrk>, or make a special element for the segmentation). That goes along with the match between the main <source> and the different <alt-trans> sources: how can we insure the sentence match? Since the data may come from different tools the segmentation may be different in the main <source> and inb the <source> of <alt-trans>. But at the same time, it's a nice way to have a 'paragraph-level' unit and be able to tell the translation tool how to break the content. Enforcing a specific segmentation would very useful, but it would have to be supported by the translation workbenches. cheers, -yves