The compiler is designed to allow only 8-bit characters to be used in the PROGRAM COLLATING SEQUENCE IS phrase.
It is meaningless to define a collating sequence using an ALPHABET clause in which the literals contain double-byte characters. This is because double-byte appears as two separate 8-bit characters. If you attempt to use double-byte in this way, double-byte characters are collated as two single bytes – any double-byte meaning is ignored.
You can, however, use non-ASCII characters in the ALPHABET clause literals when you define the program collating sequence. For example, in a Japanese environment single-byte Katakana can be used in a Shift-JIS environment as they are stored and displayed as one byte. If you are using EUC you cannot use single-byte characters in the ALPHABET IS phrase, because such characters are hybrid two-byte characters.