We're slowly changing that usage, but that will not change the one-to-one correspondence to the Latin transliteration (actually not one-to-one to Latin *letters*, which may be what you meant). The question at that point is which script one chooses to consider the cipher and which the (or a) "true" script.I think such a collection of symbols A becomes a cipher for a true script B when it replicates the usage of symbols in B, irregularities and all. In the Pigpen cipher, there is a symbol for C and one for T and one for H, and C+H and T+H are slapped together *exactly* as they are in Latin to spell English words. In a substitution cipher, the word "chime" is spelled with those five letters, c-h-i-m-e; they just look different.
I was surprised and a bit dismayed to see Klingon dragged into this thread. Klingon was rejected not because it's a cipher, but because it's used only for decoration, not for communication.
~mark
