15. Controlled Language

Controlled languages are subsets of natural languages restricted by explicit rules governing lexicon, grammar and style. They are essentially used to reduce ambiguity and complexity in written procedural or descriptive technical documents, and so can be viewed as extending good authoring practice. Their main declared purpose is to achieve cost savings through the simplification and standardization of document production, and through improved document usability. Human-machine interaction and the translation of documents, in particular by machines, are additional related aims. Controlled natural languages may be either empirical or logic-based. Many empirical controlled languages have been successfully developed in industry starting from existing in-house drafting practice and text corpora, the emphasis being placed on improving human readability. Logic-based controlled languages, by contrast, have been devised to form strictly unambiguous natural language subsets: these may have poor human readability, but unlike the less rigorously formalized empirical controlled languages, they can be used as processable machine input for knowledge representation. Language control raises some formal linguistic issues, particularly the question of whether sweeping restrictions on expression for the sake of simplicity and concision may unintentionally impede communication.