I'm afraid this rings slightly hollow for me.

We can likewise claim about C that "0 in C is typed in theory and sort of untyped in practice" because constant expressions of integer type that evaluate to 0 can serve as null pointers of any type:

   char *p = 0;        // no cast required
   if (p == 0) ...     // no cast required
   time_t t = time(0); // fine
   char *q = 1;        // error; requires (char *) 1.

We can't think of this 0 as an object; it's just a source code token that provides a null pointer of the right type, which is statically obvious from the use.

The only time it's a problem is in unsafe interfaces:

   int old_style_string_fun();
   int variadic_fun(char *first, ...);

   old_style_string_fun(0);
   variadic_fun("first", 0);

Here, the wrong type is inferred; the 0 just behaves like the ordinary constant of type int. A null pointer to char was intended, which requires (char *) 0 must be used.

None of this challenges the fact that we can have a variable of type int, whose value is 0.