Here is another approach, operate the second TMemo in mock virtual mode. No scroll bars, just, apparently, a window into the first TMemo.
The second one, refreshed as the first changes, shows only as much translation as can fit, starting with the first visible content in from the first. I am assuming here that end user only edits first Memo.
Not sure if TMemo tells us what the visible content is so, might need to use a more advanced memo such as KMemo.
(cannot try it, not typing well after some very minor hand surgery)
Davo