bigrams.py

#!/usr/bin/env python                                                                                                                                
import json
import urllib

def estimated_count_for(search_term):
  url = 'http://ajax.googleapis.com/ajax/services/search/web?v=1.0&%s' % urllib.urlencode({'q': search_term})
  results = json.loads(urllib.urlopen(url).read())
  try:
      return results['responseData']['cursor']['estimatedResultCount']
  except KeyError:
      return 0

tweet =  "Is it just me or has the Haskell just recently figured out that statistical conjugacy can be interpreted algebraically"
tokens = tweet.split(' ')
for i in range(len(tokens) - 1):
    bigram = " ".join(tokens[i:i+2])
    print "%s\t%s" % (bigram, estimated_count_for(bigram))

run.sh

$ bigrams.py | sort -t'   ' -k2 -n result.tsv  | head -n5

Haskell just  1070000
statistical conjugacy	1420000
interpreted algebraically	1430000
the Haskell	1820000
conjugacy can	1970000