Help Hierarchial Cluster Analysis & Categorical data

Hi, I am trying to cluster my variables which are all the letters of the alphabet by the observations of many participants in a lipreading experiment. I used cosine method as it detects patterns, and I transformed the reponses from letters to numbers to input in SPSS (A=1, B=2 ect). My problem is the algorithem is clustering based on the change in pattern not on which categorie. Example if all the observations of the letter b are 5, and all the observations of the letter k are 6 the output will put the letter b and k in the same cluster. That is wrong! I don't know how to fix this I'm exhausted I tried all the different methods. Please help :shakehead

1. Should I use cosine method?
2. What is the best distance measure for my data? Furthest neighbour
3. Should I standardize and if so which choice? z score
4. Am I correct to cluster by variables my letters?

Thank you in advance