This work is framed into the Statistical Machine Translation field, more specifically into the language modeling challenge. In this area, have classically predominated the n-gram approach, but, in the latest years, ...