diff --git a/NLP/Scores.hs b/NLP/Scores.hs
--- a/NLP/Scores.hs
+++ b/NLP/Scores.hs
@@ -27,6 +27,8 @@
     , ari
     , mi
     , vi
+    -- * Strength of association
+    , logLikelihoodRatio
     -- * Comparing probability distributions
     , kullbackLeibler
     , jensenShannon
@@ -43,7 +45,8 @@
     , countJoint
     , countFst
     , countSnd
-    -- * Extracting lists of values from 'Counts'
+    , countTotal
+      -- * Extracting lists of values from 'Counts'
     , fstElems
     , sndElems
     )
@@ -115,6 +118,25 @@
 vi cs@(Counts _ cx cy) = entropy (elems cx) + entropy (elems cy) - 2 * mi cs
   where elems = Map.elems
 
+
+-- | Log-likelihood ratio for two binomial distributions.
+-- H_0: P(x|y) = p = P(x|~y)
+-- H_1: P(x|y) = p1 =/= p2 = P(x|~y)
+logLikelihoodRatio :: (Ord a, Ord b) => Counts a b -> a -> b -> Double
+logLikelihoodRatio cs x y =
+  let p   = nx / n                     -- relative count of x
+      p1  = nxy / ny                   -- relative count of xy among _y
+      p2  = (nx - nxy) / (n - ny)      -- relative count of xnoty among noty
+      n   = countTotal cs
+      nx  = countFst x cs
+      ny  = countSnd y cs
+      nxy = countJoint x y cs
+      b k n p = p**k * (1-p)**(n-k)
+      {-# INLINE b #-}
+  in   log (b nxy nx p)  + log (b (nx - nxy) (n - ny) p)
+     - log (b nxy nx p1) - log (b (nx - nxy) (n - ny) p2)
+
+
 -- | Kullback-Leibler divergence: KL(X,Y) = SUM_i P(X=i) log_2(P(X=i)\/P(Y=i)). 
 -- The distributions can be unnormalized.
         
@@ -130,8 +152,10 @@
 -- | Jensen-Shannon divergence: JS(X,Y) = 1\/2 KL(X,(X+Y)\/2) + 1\/2 KL(Y,(X+Y)\/2).
 -- The distributions can be unnormalized.
 jensenShannon :: (Eq a, Floating a, T.Traversable t, T.Traversable u) => t a -> u a -> a
-jensenShannon xs ys = 0.5 * kullbackLeibler xs zs + 0.5 * kullbackLeibler ys zs
-  where zs = zipWithTF (+) xs ys
+jensenShannon xs ys = 0.5 * kullbackLeibler xs' zs + 0.5 * kullbackLeibler ys' zs
+  where zs = zipWithTF (+) xs' ys' 
+        xs' = normalize xs
+        ys' = normalize ys
           
 -- | Adjusted Rand Index: <http://en.wikipedia.org/wiki/Rand_index>
 ari :: (Ord a, Ord b) => Counts a b -> Double
@@ -204,6 +228,9 @@
 -- | Count of second element
 countSnd :: Ord k => k -> Counts a k -> Count
 countSnd y = Map.findWithDefault 0 y . marginalSnd
+-- | Total element count
+countTotal :: Counts a k -> Count
+countTotal = F.sum . joint
 
 -- | List of values of first element
 fstElems :: Counts k b -> [k]
@@ -219,3 +246,6 @@
 zipWithTF h t f = snd . T.mapAccumL map_one (F.toList f) $ t
   where map_one (x:xs) y = (xs, h y x)
         
+-- | @normalize xs@ divides each element of xs by the sum of xs.
+normalize :: (Fractional b, Functor f, F.Foldable f) => f b -> f b        
+normalize xs = let s = sum xs in fmap (/s) xs
diff --git a/nlp-scores.cabal b/nlp-scores.cabal
--- a/nlp-scores.cabal
+++ b/nlp-scores.cabal
@@ -7,7 +7,7 @@
 -- The package version. See the Haskell package versioning policy
 -- (http://www.haskell.org/haskellwiki/Package_versioning_policy) for
 -- standards guiding when and how versions should be incremented.
-Version:             0.6.2
+Version:             0.7.0
 
 -- A short (one-line) description of the package.
 Synopsis:            Scoring functions commonly used for evaluation in NLP and IR