Here try this:
import nltk
pattern = [(r'(March)$','MAR')]
tagger = nltk.RegexpTagger(pattern)
print tagger.tag(nltk.word_tokenize('He was born in March 1991'))
You have to tokenize the words.
This is the output I get:
[('He', None), ('was', None), ('born', None), ('in', None), ('March', 'MAR'), ('1991', None)]