It has been suggested that constraining a natural language (NL) reduces the degree of ambiguity of requirement specifications written in that language. There is also a tendency to assume that an inescapable side effect of constraining a natural language is a subsequent reduction in its expressiveness.

The primary objective of this paper is to describe a technique that we have developed for empirically measuring the expressiveness of a Constrained Natural Language (CNL) when used to specify the requirements in a particular application domain.

Our simple yet practical and repeatable technique elucidates the individual contribution that each lexical entity of the CNL can make on the overall expressiveness of the CNL. This technique is particularly useful for designing new CNLs, as well as situations where tailoring or streamlining existing CNLs for particular application domains is needed.