seetsnuett-VESCU LUUUELS, 2.2 VARIOUS GRAMMAR-BASED LANGUAGE MODELS Various computational grammars have been Proposed and studied, e, transformational grammar (Chomsky 1957), lexical functional gramm: (Kaplan and Bresnan 1982), government and binding (Chomsky 198! generalized phrase structure grammar (Gazdar et al, 1985), dependen grammar, paninian grammar, and tree-; adjoining grammar (Joshi 1985Language Modelling 23 This section focuses on lexical functional grammar (LFG), generalized phrase structure grammar (GPSG), government and binding (GB), and Paninian grammar (PG) and introduces various approaches to understand a language in a grammatical and rule-based format. It also introduces the dominant approaches to create statistical models of language and grammar. 2.2.1 Generative Grammars In 1957, in his book on Syntactic Structures, Noam Chomsky wrote that we can generate sentences in a language if we know a collection of words and rules in that language. Only those sentences that can be generated as per the rules are grammatical. This point of view has dominated computational linguistics and is appropriately termed generative grammar. ‘The same idea can be used to model a language. If we have a complete set of rules that can generate all possible sentences in a language, those rules provide a model of that language. Of course, we are talking only about the syntactical structure of language here. Language is a relation between the sound (or the written text) and its meaning. Thus, any model of a language should also deal with the meaning of its sentences. As seen earlier, we can have a perfectly grammatical but meaningless sentence. In this chapter, we will assume that grammars are a type of language models. 2.2.2 Hierarchical Grammar Chomsky (1956) described classes of grammars in’a hierarchical manner, where the top layer contained the grammars represented by its sub classes. Hence, Type 0 (or unrestricted) grammar contains Type 1 (or context- sensitive grammar), which in turn contains Type 2 (context-free grammar) and that again contains Type 3 grammar (regular grammar). Although this relationship has been given for classes of formal grammars, it can be extended to describe grammars at various levels, such as in a class-sub class (embedded) relationship. 2.2.3 Government and Binding (GB) As discussed in Chapter 1, a common viewpoint taken by linguists (not computational linguists, however) is that the structure of a language (or how well its sentences are formed) can be understood at the level of its meaning, particularly while resolving structural ambiguity. However, the sentences are given at the syntactical level and the transformation from meaning to syntax or vice versa is not well understood.Nat i ‘ural Language Processing and information Retrieval ‘Transformational grammars assume two levels OF at eh een One at the surface level and the other atthe GeeP TON ON! this oa TOL Be confused with the meaning level) ov a aan & (Cy theories have renamed them as sevel and °°" tified 4! a alled Phong’ More levels of representation (parallel to ¢2¢? ather) called pony. 2 be cone and logical form. Accor ding to GB theories, language can be Consiga,” for analysis at the levels shown in Figure 2+ d-structure s-structure Phonetic form Logical form : Figure 2.1 Different levels of representation in GB If we describe language as the representation of some ae ing ‘sound form, then according to Figure 9.1, these two. a is are the logical form (LF) and phonetic form (PF) respectively. The GB is concerned wig, LF, rather than PF. Chomsky was the first to put forward a GB theory (Peter Sells 1985). pore ‘Transformational grammars have hundreds of rewriting rules, which are generally language-specific and also construct-specific (say, different tules for assertive and interrogative sentences in English, or for active and passive voice sentences). Generation of a complete set of coherent rules may not be possible. The GB envisages that if we define rules for structural units at the deep level, it will be possible to generate any language with fewer rules. These deep-level structures are abstractions of noun-phras, verb-phrase, etc., and common to all languages. It is Possible to do if, as GB theory’ states, : oad learns its mother tongue because the human mind is ‘hard-wired’ with some universal structures. The existence of deep level, language indepen ‘al phonetic structures, and the expression of these in sutface level, mem lent, abstract structures with the help of simple rules i the main osva’, “nSUase-specific Let us take an example to explain d- and structures of GB theories. Example 2.1 5 Miikach wac Lillaet