We have English and Japanese documents, which were formerly localized. Is it possible to use these legacy contents and reduce future localization cost?

Yes, it is possible. First, we will create a database (translation memory) from the formerly localized documents. Then compare (leverage) your new source with the database. Reusing the former contents, even if it is part of a sentence, you can reduce your localization cost. And you will be able to unite styles, tones referring to the former translation.