Jaime,
The problem is that you are finding both sets of the English text, so that when you apply formatting to what you find, you apply that to everything. You could be using a positive look behind command to find only the Spanish text, but then it won’t have the English text to replace it with.
Without seeing your document, it looks like this may require a multi-step process to isolate the Spanish text, because if there is only one return between each set of paragraphs, and they just alternate, Find/Change may not be able to tell which ones to trap and which to ignore. Is there anything unique about the Spanish? Sounds like it has unique formatting – if you can search for that and apply a unique paragraph style to it, that would be a good start.
Then, you can find all of the Spanish and add something unique before it, such as ???, then you could do the (.+)\r???(.+) and replace with $1\r???$1 (replace what would be $2 with $1. Then you could reapply the Translation paragraph style to all paragraphs that begin with ???, and then remove all of the ???.
Again, without seeing your document, this is an approximation, but the idea is that you need to do something to make the Spanish text unique so that you can replace it.