Exploring Lexical Patterns in Text: Lexical Cohesion Analysis with WordNet

Exploring lexical cohesion

On the basis of the annotated data, we have generated some statistics concerning the average chain lengths (in no. of sentences/words participating in a chain), according to register, of both all the chains and the dominant (i.e., the longest) chains and the distribution of types of lexical cohesion (repetition, synonymy, hyponymy, etc.) according to register.

As will be seen, the dominant chains in a text give a good indication of a text’s topic; also, the distribution of types of lexical cohesion turns out to be a possible measure for discriminating between registers.

Figure 5: Average length of dominant chains by register

1.