Is this efficient?

When compared with what?

Also, how can I do [ucfirst(lc($word))] for swedish words?

It should work as-is for Swedish words.

use open ':std', ':locale';
use charnames ':full';
my $word =
    "\N{LATIN CAPITAL LETTER A WITH RING ABOVE}" .
    "\N{LATIN CAPITAL LETTER A WITH DIAERESIS}" .
    "\N{LATIN CAPITAL LETTER O WITH DIAERESIS}" .
    "\N{LATIN CAPITAL LETTER E WITH ACUTE}";
print($word, "\n");
print(ucfirst(lc($word)), "\n");
[download]

ÅÄÖÉ
Åäöé

Of course, if the words are coming to you encoded (i.e. from a file handle), you need to decode them first.

You probably won't run into this problem, but if the characters in the range U+0080..U+00FF are left unchanged, precede the expression with

utf8::upgrade( $word );
[download]

That bug will be fixed in 5.12 (although it might require use 5.012;).