Not really, because a file might not even contain any header information that indicates its encoding. There are tools out there that will "guess" at what the encoding is which might help, eg

[oracle@db192 ~]$ file -i emp.csv
emp.csv: text/plain; charset=us-ascii

You could take advantage of this by using *another* external table to query this result and then make a decision on whether you need to modify the true external table, eg

  create table file_check
   (    ftype varchar2(64)
   )  
   organization external
    ( type oracle_loader
      default directory temp
      access parameters
      ( records delimited by newline
        preprocessor  bin:'run_file_command.sh'
   )
      location
       ( temp:'dummy.txt'
       )
    )
   reject limit unlimited

where "run_file_command.sh" is something like:

file -i emp.csv | sed 's/.*charset=//g'