From each text type a passage of approximately 400 words was taken one from story, user guide and magazine. According to his model, texts can either have expressive function, informative function, or vocative function. To observe whether text types have any influence on accuracy of MT system or not, different genres have been selected based on Peter Newmark’s (1988, p.15) model of text type. In order to propose a framework for analysis and classification of these errors that was the main goal of the present study, three passages were translated by two different MT systems and a translator. There are two types of evaluating MT systems, Human Evaluation and Automatic Evaluation, but due to the complexity of natural language, researcher decided to evaluate it manually. And finally identifying the cause of errors in MT output is not straightforward, either. One error can lead to another, sometimes. Second, defining the boundaries of errors in MT output is often difficult. First, a text has several different translations, all of which are. Flanagan (1994) points out that there some reasons for complexity of evaluating MT systems. It aims to discover criteria for improving the performance and quality of MT systems. The purpose of the present study is to propose a framework for error analysis and classification of rule-based machine translation outputs. But the limitation is that it cannot handle semantic and contextual problems. This system can translate all kinds of sentences. It uses the output of the lexical analyzer and the parser to make Bengali sentence. The generator generates a Bangla sentence, which is equivalent to the given input English sentence. The parser parses the input sentence and identifies the types of it and finds tense, phrase, clauses, etc. The lexical analyzer uses the English to Bangla dictionary and word morphology for finding lexical information. The lexical analyzer reads the English sentence, tokenizes into words, and stores information into a stack. The system performs translation procedure in three steps. The system incorporates itself with the dictionary and gives the corresponding Bengali meaning. Here, we have been developing a top-down parsing program. The implemented translation system determines the relationship among different forms of English and Bengali sentences and makes appropriate correspondence between English and Bengali grammar.

In this translator, the source language is English, and the target language is Bangla. This paper deals with the design and development of an expert sentence translation system.

English to bengali letter translation