In this paper

an enhanced backing-off language model is proposed

based on traditional backing-off language model combining Turing-discounting technique.The enhanced backing-off language model given above presents the algorithms in computing the probabilities of bigrams and unigrams and solving OOV（out-of-vocabulary） problem differed from related approaches.Experimental results show that our approach outperforms the traditional one by dealing with the problem of computing the probability of bigrams

unigrams and OOVs effectively.