while ( $line =~ s/([[:alpha:]][[:alpha:]\']{0,44})[_\-,\.\"\'\)\?!:;\/&]{0,5}([ \t\n\r]|$)/ / ) { update_word($self,$1, $encoded, '', '[_\-,\.\"\'\)\?!:;\/ &\t\n\r]') if (length $1 >= 3);