my (off the top of my head) comment is:
don't use regex...convert each character into ASCI or your "true" alphabetically mapped sequence number (eg. aA->1 etc.) and loop through the string, keeping the highest number seen in a variable. As soon as you hit a lower value, exit the loop and print "not alphabetical". This should beat regexp for longer/complex strings, and you can make your own map of exactly what "alphabetical" is (in Chinese if you like). The solution ends up more portable, flexible (not tied to regex syntax), and runs faster. Sorry, but am about to disembark and go home, so i'll post some code later.