I was able to test on 5.8.7 (an old freebsd box), as well as 5.10.1 (linux) and 5.12.3 (macosx), and all three behaved the same -- including a problem in the output from Test::More::is() (I don't know why unicode characters get converted to "?").

Here's my "util.pm":

use strict;
package util;

sub beautify {
    my ($in) = @_;
    my $tmp;
    foreach (split(/\s/o, lc($in))){
        $tmp .= ucfirst($_);
        $tmp .= ' ';
    }
    $tmp =~ s/\s$//;
    return($tmp);
}
1;
[download]

Here's my test code:

use utf8;
use util;
use Test::More tests => 4;

is( &util::beautify( "àisTheWord"), "Àistheword", "àisTheWord - specia
+l character changes case." );
is( &util::beautify( "ùisTheWord"), "Ùistheword", "ùisTheWord - specia
+l character changes case." );
is( &util::beautify( "üisTheWord"), "Üistheword", "üisTheWord - specia
+l character changes case." );
is( &util::beautify( "ÿisTheWord"), "Ÿistheword", "ÿisTheWord - specia
+l character changes case." );

print util::beautify( "ÿisTheWord") . "\n";
[download]

And here's the output I got on all three versions:

1..4
ok 1 - ?isTheWord - special character changes case.
ok 2 - ?isTheWord - special character changes case.
ok 3 - ?isTheWord - special character changes case.
ok 4 - ?isTheWord - special character changes case.
Ÿistheword
[download]

(I put in that last print statement just to check that it would correctly show what was expected on a utf8-capable terminal. I also left out the "u with circumflex because for some reason I couldn't post its upper-case form correctly -- very strange.)

UPDATE: I should add that the command line I used was "perl -C31 test-util.t" (and it seems "-CS" would do the same thing.) Anyway, the tests do pass for me, despite the test message getting munged. So what are you doing that's different?

SECOND UPDATE: I added use Test::More::UTF8 as per mje's suggestion below, and that fixed the output messages -- thanks++!!