How to create a customized translation table when using the command translate-region?

Question

When using the command translate-region, the command asks to use which table. There are already some predefined tables (by pressing TAB) such as

cp51932-decode cp51932-encode eucjp-ms-decode eucjp-ms-encode japanese-ucs-cp932-to-jis-map japanese-ucs-glibc-to-jis-map japanese-ucs-jis-to-cp932-map japanese-ucs-jis-to-glibc-map jisx0213-to-unicode ucs-normalize-hfs-nfd-table ucs-normalize-nfd-table ucs-normalize-nfkd-table unicode-to-jisx0213

What if someone likes to create its own table?

For example, assume that one would like to create a conversion table like below:

1 --> 2
a --> x
n --> m
8 --> 0
σ --> s

What I have tried: based on one of the above tables, I defined a table as bellow, but I get the error apply: Wrong type argument: integerp, x

(let ((map
       '(
     (1 . 2)
     (a . x)
     (n . m)
     (8 . 0)
     (σ . s)
     )))
        (define-translation-table 'my-conversion-decode map)
  (mapc #'(lambda (x)
        (let ((tmp (car x)))
          (setcar x (cdr x)) (setcdr x tmp)))
    map)
  (define-translation-table 'my-conversion-encode map))

Answer 1

The map should be a char-to-char alist. In Emacs Lisp, a character is a number, the character x can be written as ?x or its Unicode code point 120. If you just write x, it's a symbol.

The following works from here:

(let ((map '((?1 . ?2)
             (?a . ?x)
             (?n . ?m)
             (?8 . ?0)
             (?σ . ?s))))
  (define-translation-table 'my-conversion-decode map))