Defining context-dependent syntax without using contexts

A method for defining context-dependent syntax is presented. The method, like many others, builds on a context-free grammar that defines a superset of a given language, mapping each string in the language to a derivation tree. The context-free language is subsequently restricted by imposing conformity criteria on derivation trees. These criteria are expressed with logic predicates, called context rules, which characterize the form of conformant derivation trees. A text belongs to the defined language if and only if its derivation tree is a model of all context rules. This approach avoids the use of explicit data structures for contextual information and tends to yield very concise formal definitions. It has been applied with good results in the formal definition of the programming language Oberon