Building on the ideas of NetWallah and choroba:

c:\@Work\Perl\monks>perl -wMstrict -MData::Dump -le
"my @dict = ('???', qw(a abbey zymophore zymophoric));
 ;;
 my %encode = map { $dict[$_] => $_ } 0 .. $#dict;
 dd \%encode;
 ;;
 chomp(my $sentence = <STDIN>);
 print qq{sentence: '$sentence'};
 my @words = split ' ', lc $sentence;
 dd \@words;
 ;;
 my @encoding = map { $encode{ exists $encode{$_} ? $_ : '???' } } @wo
+rds;
 dd \@encoding;
 ;;
 my @decoding = @dict[ @encoding ];
 dd \@decoding;
"
{ "???" => 0, "a" => 1, "abbey" => 2, "zymophore" => 3, "zymophoric" =
+> 4 }

a crumbling zymophoric abbey
sentence: 'a crumbling zymophoric abbey'
["a", "crumbling", "zymophoric", "abbey"]

[1, 0, 4, 2]

["a", "???", "zymophoric", "abbey"]
[download]