- Word features Words feature describe different aspects of the words present in the buffer or in the stack. They are composed of an address function identifying a particular word and an attribute function selecting a specific attribute of the word, such as its form, lemma, part of speech ... Such features are represented as strings that are the concatenation of the address function and the attribute function. Address function can be one of the following:
- b0 the current word in the buffer
- b1 the word following b0
- b2 the word following b1
- b3 the word following b2
- bm1 the word preceding b0
- bm2 the word preceding b1
- bm3 the word preceding b2
- s0 the word on the top of the stack
- s1 the word below s0
- s2 the word below s1
- s3 the word below s2