Generative grammar
From Wikipedia, the free encyclopedia
In linguistics, generative grammar generally refers to a proof-theoretic framework for the study of syntax partially inspired by formal grammar theory and pioneered by Noam Chomsky. A generative grammar is a set of rules that recursively "specify" or "generate" the well-formed expressions of a natural language. This encompasses a large set of different approaches to grammar. The term generative grammar is also broadly used to refer to the school of linguistics where this type of formal grammar plays a major part, including:
- The Standard Theory (ST) (also widely known as Transformational grammar (TG))
- The Extended Standard Theory (EST) (also widely known as Transformational grammar (TG))
- Principles and Parameters Theory (P&P) which includes both Government and Binding Theory (GB) and the Minimalist Program (MP)
- Relational Grammar (RG)
- Lexical-functional Grammar (LFG)
- Generalized Phrase Structure Grammar (GPSG)
- Head-Driven Phrase Structure Grammar (HPSG)
Generative grammar should be distinguished from traditional grammar, which is often strongly prescriptive rather than purely descriptive, is not mathematically explicit, and has historically investigated a relatively narrow set of syntactic phenomena. In the "school of linguistics" sense it should be distinguished from other linguistically descriptive approaches to grammar, such as various functional theories.
The term generative grammar can also refer to a particular set of formal rules for a particular language; for example, one may speak of a generative grammar of English. A generative grammar in this sense is a formal device that can enumerate ("generate") all and only the grammatical sentences of a language. In an even narrower sense, a generative grammar is a formal device (or, equivalently, an algorithm) that can be used to decide whether any given sentence is grammatical or not.
In most cases, a generative grammar is capable of generating an infinite number of strings from a finite set of rules. These properties are desirable for a model of natural language, since human brains are of finite capacity, yet humans can generate and understand a very large number of distinct sentences. Some linguists go so far as to claim that the set of grammatical sentences of any natural language is indeed infinite.
Generative grammars can be described and compared with the aid of the Chomsky hierarchy proposed by Noam Chomsky in the 1950s. This sets out a series of types of formal grammars with increasing expressive power. Among the simplest types are the regular grammars (type 3); Chomsky claims that regular languages are not adequate as models for human language, because all human languages allow the embedding of strings within strings in a hierarchical way.
At a higher level of complexity are the context-free grammars (type 2). The derivation of a sentence by a context-free grammar can be depicted as a derivation tree. Linguists working in generative grammar often view such derivation trees as a primary object of study. According to this view, a sentence is not merely a string of words, but rather a tree with subordinate and superordinate branches connected at nodes.
Essentially, the tree model works something like this example, in which S is a sentence, D is a determiner, N a noun, V a verb, NP a noun phrase and VP a verb phrase:
The resulting sentence could be The dog ate the bone. Such a tree diagram is also called a phrase marker. They can be represented more conveniently in a text form, (though the result is less easy to read); in this format the above sentence would be rendered as:
[S [NP [D The ] [N dog ] ] [VP [V ate ] [NP [D the ] [N bone ] ] ] ]
However, Chomsky at some point argued that phrase structure grammars are also inadequate for describing natural languages. To address this, Chomsky formulated the more complex system of transformational grammar.
When generative grammar was first proposed, it was widely hailed as a way of formalizing the implicit set of rules a person "knows" when they know their native language and produce grammatical utterances in it. However Chomsky has repeatedly rejected that interpretation; according to him, the grammar of a language is a statement of what it is that a person has to know in order to recognise an utterance as grammatical, but not a hypothesis about the processes involved in either understanding or producing language. In any case the reality is that most native speakers would reject many sentences produced even by a phrase structure grammar. For example, although very deep embeddings are allowed by the grammar, sentences with deep embeddings are not accepted by listeners, and the limit of acceptability is an empirical matter that varies between individuals, not something that can be easily captured in a formal grammar. Consequently, the influence of generative grammar in empirical psycholinguistics has declined considerably.
Generative grammar has been used in music theory and analysis such as by Fred Lerdahl and in Schenkerian analysis. See: Chord progression#Rewrite rules.
| Automata theory: formal languages and formal grammars | |||
|---|---|---|---|
| Chomsky hierarchy |
Grammars | Languages | Minimal automaton |
| Type-0 | Unrestricted | Recursively enumerable | Turing machine |
| n/a | (no common name) | Recursive | Decider |
| Type-1 | Context-sensitive | Context-sensitive | Linear-bounded |
| n/a | Indexed | Indexed | Nested stack |
| Type-2 | Context-free | Context-free | Nondeterministic Pushdown |
| n/a | Deterministic Context-free | Deterministic Context-free | Deterministic Pushdown |
| Type-3 | Regular | Regular | Finite |
| Each category of languages or grammars is a proper subset of the category directly above it. | |||
