Formalism Framework

Formalism Framework of Language

Formalism is a school of literary criticism and literary theory having mainly to do with structural purposes of a particular text. It is the study of a text without taking into account any outside influence. Formalism rejects (or sometimes simply "brackets," i.e., ignores for the purpose of analysis) notions of culture or societal influence, authorship, and content, and instead focuses on modes, genres, discourse, and forms.

In another definition, Formalism or formal linguistics is the study of the abstract forms of language and their internal relations. It fixes on the forms of languages as evidence of the universals without considering how these forms function in communication and the ways of social life in different communities.

In literary theory, formalism refers to critical approaches that analyze, interpret, or evaluate the inherent features of a text. These features include not only grammar and syntax but also literary devices such as meter and tropes.

The Formalist Approach

Formalism views the primary function of ordinary language as communicating a message by references to the world outside of language. The formalist approach studies the form of the work, as opposed to its content. This approach examines the formula or methodology in literature, and how it leads to deeper meaning with closer reading. New Critics believe that anything that is essential to interpreting the text must be found within the text itself (point of view, symbols, irony, language, etc.)—and not from what the reader brings to understanding the work through his or her own assumptions based on his or her own experiences and interpretive strategies.  They therefore particular attention to the literary devices used in the work and to the patterns these devices establish. As the narrative flows through its plot complications, it eventually reaches a climactic point, and all the details of the form fall into place in the dénouement.  These internal relationships gradually reveal a form—a principle by which all subordinate patterns can be seen. The formalist approach reduces the importance of a text’s historical, biographical, and cultural context.

Besides, Formalists favor an approach to the study of language which emphasizes abstract, quasi-mathematical theories of linguistic structure based primarily, but not always exclusively, on intuitions of grammaticality. These theories are usually, but not always, discrete: they do not employ statistical methods and avoid continuous structures. One strength of these theories, at least according to proponents, is that they take otherwise vague linguistic intuitions and make them precise and testable. However, there is no necessary divide between the two approaches. Functionalists can and sometimes do use formal techniques, and formalists can and sometimes do take communicative function into account. Often confused with generative linguistics, which is a subset of formal linguistics. Also often confused with Chomskyan linguistics, which is a subset of generative linguistics.

Formalism of Noam Chomsky

Chomsky has been described as the "father of modern linguistics" and a major figure of analytic philosophy. His work has influenced fields such as computer science, mathematics, and psychology. He is credited as the creator or co-creator of the Chomsky hierarchy, the universal grammar theory, and the Chomsky–Schützenberger theorem.

In the mid-1950s Noam Chomsky, developed the formalism of context-free grammars and also their classification as a special type of formal grammar (which he called phrase-structure grammars). What Chomsky called a phrase structure grammar is also known now as a constituency grammar, whereby constituency grammars stand in contrast to dependency grammars. In Chomsky's generative grammar framework, the syntax of natural language was described by context-free rules combined with transformation rules.

In formal language theory, a grammar (when the context isn't given, often called a formal grammar for clarity) is a set of formation rules for strings in a formal language. The rules describe how to form strings from the language's alphabet that are valid according to the language's syntax. A grammar does not describe the meaning of the strings or what can be done with them in whatever context—only their form.

Formal language theory is the discipline which studies formal grammars and languages, is a branch of applied mathematics. Its applications are found in theoretical computer science, theoretical linguistics, formal semantics, mathematical logic, and other areas.

A formal grammar is a set of rules for rewriting strings, along with a "start symbol" from which rewriting must start. Therefore, a grammar is usually thought of as a language generator. However, it can also sometimes be used as the basis for a "recognizer"—a function in computing that determines whether a given string belongs to the language or is grammatically incorrect. To describe such recognizers, formal language theory uses separate formalisms, known as automata theory. One of the interesting results of automata theory is that it is not possible to design a recognizer for certain formal languages.

A Powerful Grammar Formalism

The grammar formalism is closely related to other formalisms currently in use in computational linguistics. These formalisms are known as `unification-based', `constraint-based', `information-based' and `feature-logic based'. Members of this class are for example Definite Clause Grammars, PATR II, Functional Unification Grammar and formalisms underlying linguistic theories such as Generalized Phrase Structure Grammar, Lexical Functional Grammar, Unification Categorical Grammar, Categorical Unification Grammar and Head-driven Phrase Structure Grammar.

They also show how the nice properties of logic programming languages carry over to a whole range of such constraint-based formalisms, by abstracting away from the actual constraint language that is used. I define such a constraint-based formalism in which the underlying constraint language, consists of path equations. The most important characteristics of the formalism are:

• The formalism consists of definite clauses, as in Prolog; instead of first-order terms the data structures of the formalism are feature structures.
• The formalism does not assume that concatenation is the sole string-combining operation (in contrast to FUG, DCG, PATR II, LFG, GPSG and UCG).
• The formalism is defined in an abstract framework, which facilitates the extendability of the techniques I develop in later chapters, to formalisms based on other (more powerful) constraint-languages.

Each of these points will now be clarified in turn.

Firstly, the principal `data-structures' of the formalism are feature structures, rather than first-order terms such as in Prolog. The motivation is that such feature structures are closer to the objects usually manipulated by linguists. Furthermore, in writing grammars the use of first-order terms becomes rather tiresome because it is necessary to keep track of the number of arguments functors take, and the position of sub-terms in such terms. Using path equations to define feature structures achieves some sort of data abstraction. As an example consider a program which manipulates terms such as the following

and suppose furthermore we want to refer to a specific part Cat of such a term T. In Prolog we are then forced to mention all intermediate functors, and for each functor we need to mention all its arguments (possibly using the `anonymous' variable `_'):

On the other hand, using path equations, it is possible to refer to such an embedded term by its path, in this case the value of Cat is obtained by the equation Cat T syn head cat.

This said, it should be stressed though that the difference between the two approaches is not very decisive. In fact first-order terms may be used in an implementation of such graph-based formalisms, and data-abstraction can also be achieved by other means such as syntactic macro's, or by auxiliary predicates.

From a linguistic point of view, the second characteristic is the most salient one. The formalism to be proposed, does not enforce that concatenation is the sole operation to combine strings. This choice can be motivated by:

• Increased symmetry of parsing and generation
• Increased expressive power
• Other applications

Dropping the concatenative base can be motivated from the desire to use grammars in a reversible way. From a reversible viewpoint, it is attractive to view a grammar simply as a definition of the relation between strings and logical forms. To give a different status to the phonology attribute, seems to destroy the inherent symmetry somewhat. Thus, the formalism does not prescribe how the value of the phonology attribute is to be composed, just as it does not prescribe how the value of the semantics attribute is composed.

If we do not incorporate a concatenate base, then we allow for investigation of other types of string combinations in natural language grammars. Several researchers, have noted that analyses of a whole range of linguistic phenomena (most notably those involving discontinuous constituents) may be simplified by assuming other types of string operations. If no assumptions about the construction of phonological representations, or semantic representations, are defined, then the parsing and generation problem of the formalism is generally not decidable. An important theme of this thesis is, to investigate parsing and generation procedures which can be applied usefully, for linguistically motivated grammars.

Another reason for developing a formalism which is not based on concatenation is the observation that other (non-linguistic) problems can be encoded in a unification-grammar as well, if we are not forced to manipulate strings. Furthermore, the formalism is also used to define meta-interpreters in -- this usage of the formalism also entails that no assumptions about string construction are built-in.

The third characteristic states that the resulting formalism is a member of a class of constraint-based formalisms. Therefore, results that hold for this class carry over to the present formalism. In the other direction, it is easy to see how the current formalism can be extended to allow for other, perhaps more complex constraints. This is very useful as in the last few years a whole family of different constraints has been proposed that do extend formalisms such as PATR II. In a somewhat idealized view, parsing and generation algorithms defined for a member of the class of constraint-based formalisms, can be used for other members of this class, provided the appropriate constraint-solving techniques are available for the constraints incorporated in these other formalisms.