ok, if i understood it correctly, its an easy differential code: each symbol stands for a number between 0 and 25. now, the first letter is encoded directly. Say if its a B, the symbol for 1 would be used (cause we satart with 0) if its a C the symbol for 2 is used, etc.
this only applys for the first letter. for the rest, the difference (position-wise in our alphabet) is encoded, so, if your last letter encoded was that B and you want to have a E next, you would encode 3, cause E is 3 letters away from B (B->C->D->E). thats basically it. the only thing needed now is a way to get a letter which is earlier in the alphabet, lets say ... C. for that purpose, our latin alphabet is looked at as a ring (mod 26) so, after Z it starts with A again, now, for the C after our encoded E: E-F-G-H-I-J-K-L-M-N-O-P-Q-R-S-T-U-V-W-X-Y-Z-A-B-C
so, its 24 letters to go. our encode of BEC would be 1 3 24. those numbers are now substituted for the alien letters according to the chart (or the subset of that, which is known by now) again to form the actual word.
Decoding is similar, first get the numbers for the word which you want to decode, lets say 1 3 24
now, add to each position its left neighbour, starting from the left:
1 4 28 (28 = 24 + 3 + 1) next step: for any number bigger than 25, substract 26 till its smaller than 26: 1 4 2, thats it, look up the letters in the alphabet at those positions (add one to each, cause we start with 0) and you are done: 1->B 4 ->E 2 ->C
if i understood it all wrong, well, go to hell then