OK, so I wrote sentence segmentator on top of po4a. It works this way: 1. po4a-gettextize creates the initial .po with paragraph-level segments. 2. The attached po4a-segment.pl -S(plit) splits the paragraphs (currently by dot+space, but abbrevs and rules should be added). Each sentence belongs to a block, so that it can be merged back to paragraphs at the end. 3. Tranlators work with this sentence-level file. My wife is a translator and don't tell me sentence is a poor context - you guys were just lazy to code the sentence splitter :) 4. Once the translators are done, the po4a-segment.pl -M(erge) is run to revert to original po4a paragraph segments. Quick tested with a libreoffice file (attached 1.odf with content.xml inside) and latex. Seems to work for the attached file - I am able to split, translate, merge back to paragraphs and at the end succesfully po4a-translate! :) The code is well documented and I hope to maintain the code if you merge it with po4a package. Of course there are features missing, like more rules for sentence splitting, picking proper latex (and other formats) environments which can be translated, encodings, etc. But basic work is done and if it's accepted I can work more on that. -- Karol Kreński