This works if the words to look for are UTF-8 (at least 4 chars long, as per specs), consisting of alphabetic characters of ISO-8859-15 (which is fine for Spanish, but also for English, German, French, etc.):
$n_words = preg_match_all('/([a-zA-Z]|xC3[x80-x96x98-xB6xB8-xBF]|xC5[x92x93xA0xA1xB8xBDxBE]){4,}/', $str, $match_arr);
$word_arr = $match_arr[0];
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…