XLF translation misses bullet points when re-importing into Rise - why?

We are exporting Rise courses as XLF, converting to Word (using a Mac), translating the text and converting back to XLF again. 

In the word file, the text loses all formatting and appears as a single block of text.  If we then add returns in the translated text side of the table, such as putting a bullet list on separate lines within the same cell of the table, this converts back into the XLF with no problem.  However, once the XLF is re-uploaded into Rise, the formatting disappears again and becomes a solid block of text which we have to reformat manually.

This in itself is irritating enough as it is time-consuming to reformat. 

The bigger problem is that even though all text is present in the XLF, it does not all get reimported - it misses the last sentence each time.  If there are 4 bullets in the list, only 3 will be reimported.  If there are 3 bullets plus a 4th line, the 4th line will be missing.  Even if no bullet symbol is used, just returns at the end of the line, the same thing is true.  We have to cross check against the word table to ensure everything has been imported properly.

If we take out all the returns in the word table, it becomes much harder to reformat the lists properly in the re-imported Rise module (especially when the translation is in Chinese/Korean alphabets), so is no better in terms of time-saving.

Had anyone experienced a similar problem, or can anyone suggest a way of avoiding this?  

Thanks in advance