Yes you should pursue this. It is a major issue and a current topic of conversation. Major users of major PLM software are standardising on English (no bad thing perhaps...) because whilst the software may well have localised langauge versions, the data is not localised. To store it separately brings lifecycle issues to the verification & validation of the data, as well as adding complexity to the database and to do it on demand not only would have the same issues as before, but as you have found, on-line translators are not reliable even for common language, much less engineering terminology and company standard abreviations (just for example). This was a hot topic amongst delegates at a European PLM summit conference in London earlier this year, sponsored by AMR research.