Indexed language
From Wikipedia, the free encyclopedia
Indexed languages are a class of formal languages discovered by Alfred Aho;[1] they are described by indexed grammars and can be recognized by nested stack automata.[2]
Indexed languages are a proper subset of context-sensitive languages.[1] They qualify as an abstract family of languages (furthermore a full AFL) and hence satisfy many closure properties. However, they are not closed under intersection or complement.[1]
The class of indexed languages has practical importance in natural language processing as a computationally affordable[citation needed] generalization of context-free languages, since indexed grammars can describe many of the nonlocal constraints occurring in natural languages.
Gerald Gazdar (1988)[3] and Vijayashanker (1987)[4] introduced a mildly context-sensitive language class now known as linear indexed grammars (LIG).[5] Linear indexed grammars have additional restrictions relative to indexed grammars. LIGs are weakly equivalent (generate the same language class) as tree adjoining grammars.[6]
Examples
Properties
Hopcroft and Ullman tend to consider indexed languages as a "natural" class, since they are generated by several formalisms, such as:[9]
- Aho's indexed grammars[1]
- Aho's one-way nested stack automata[10]
- Fischer's macro grammars[11]
- Greibach's automata with stacks of stacks[12]
- Maibaum's algebraic characterization[13]
Hayashi[14] generalized the pumping lemma to indexed grammars. Conversely, Gilman[7] gives a "shrinking lemma" for indexed languages.