Provided you save your localization module with utf8 encoding, just include use utf8; in that module. See perlunicode, and Tom Christiansen's most excellent Stackoverflow answer on unicode. Here's an example you can work from:

    use Encode;
    binmode STDOUT, ':encoding(UTF-8)';
    require 'russian.pl';

    my $en = 'Hello';
    printf "English: %s, Russian: %s\n",
        $en, MyPackage::ru::translate($en);
[download]

Output:

English: Hello, Russian: привет

If you run this exact code with russian.pl encoded in utf8, and do not see the expected Cyrillic script, check your terminal settings and/or try another. Also beware if you are piping the output through other programs that are not aware of the encoding.

russian.pl:

    package MyPackage::ru;

    use utf8; # Tells Perl to interpret this source file as utf8
    use Carp;

    # Very crude translation table
    my %trans = (
        # XXX - Note actual code contains the literal utf8 characters,
        #       not any kind of escapes. PerlMonks insists on converti
+ng
        #       them to HTML character entities.
        hello   => 'привет',
    );

    sub translate { $trans{ lc $_[0] } || croak 'No translation for '.
+$_[0] }
[download]

As a (sort-of) aside to your question, however, you might want to check out Locale::Maketext to save yourself the work of re-inventing a localization module.