By letters do they mean Latin/English alphabet, or hiragana/katakana characters?
Non-ASCII symbols are kind of non-canonical and don't parse well, even today. Japanese developers are deeply aware of that, so generated identifiers in Japanese systems are almost always either hyphenated numeric or occasionally ASCII alphanumeric. Only when said identifier are guaranteed to never need programmatic handling, or when the system can have complete control of input and output, non-ASCII characters start appearing.
