Do we agree that {ro da zo'u da jo'u da mintu da}?
In the abstract, the letter "o" exists only once, just like there is only one number "1".
When speaking abstractly about the letters in a word, there would then however be two of the same "thing", and that would mean there are only distinct four referents.
Talking about instances of the abstract letter is the only easy way out of this, but then there still remains the slight problem that both of those instances are identical to one another. They differ in nothing, except - you might argue - in the property {lo ka du ma kau}.
However, does this solve the problem? The word "brodo" is spelled correctly no matter which of those two "o" letters you place first, and I think that's because there aren't two distinct "o" letters, since there is nothing really to tell them apart.
If you say that {me'o .obu} has two referents, which you must if you claim {ci moi lo mu lerfu}, then you claim that the _expression_ "o" exists more than once. This is fine by me, but it requires a special domain to work, one in which there are only instances of letters (even though each instance can itself be a kind). I would probably feel more at ease (for a general solution) if it involved {mupli}, but then the aforementioned problem still remains: The two instances differ in nothing, and either of them can be the third letter in "brodo". And these so-called instances of the letters are still very abstract, since words aren't always written down; they can simply be in our minds.
This intricacy is what kept me from proposing a solution similar to yours yesterday (I had thought about the _expression_ {by jo'u ry jo'u re boi .obu jo'u dy} and found it very strange (with or without {me'o})!).