Indonesian Journal of Electrical Engineering and Computer Science Vol. 1, No. 1, January 2016, pp. 182 ~ 190 DOI: 10.11591/telkomnika.v1i1.pp182-190  182 Received July 18, 2015; Revised November 4, 2015; Accepted December 1, 2015 A Review on Machine Translation Approaches Benson Kituku* 1 , Lawrence Muchemi 2 , Wanjiku Nganga 3 1 Department of Computer science, Dedan Kimathi University of Technology, Kenya 2,3 School of computing and informatics, University of Nairobi, Kenya *Corresponding author, e-mail: nebsonkituku@gmail.com 1 , lmuchemi@uonbi.ac.ke 2 , wanjiku.nganga@uonbi.ac.ke 3 Abstract The frequent domestic and international exchanges have created an opportunity for machine translation to flourish since human translation cannot cater for the translation demand. As a result, Machine translation has made tremendous stride since inception in 1940 with emergence of many architectures and approaches. This review present overview of the start of art of machine translation approaches, architectures and taxonomy of machine translation based on the background theory of each approach Keywords: Corpus, Hybrid, Multilingual, statistical, Machine translation, Rule-based Copyright © 2016 Institute of Advanced Engineering and Science. All rights reserved. 1. Introduction: Machine translation (MT henceforth) is a branch of computational linguistics which is defined as an automatic process by a computerized system that convert a piece of text (written or spoken) from one natural language referred to as a source language (SL) to another natural language called the target language (TL) with human intervention or not, and with the objective of restoring the meaning of the original text in the translated text [1, 2, 3, 4]. The issues of machine translation has been in existence since 1940[2] and over the time a lot of improvement has been witnessed in the approaches and architectures used to build the systems. However, despite the effort, the translation performance in terms of fluency, fidelity, post edit and precision is quite low compared with that of human translation though quite encouraging for computerized systems. Today machine translation has diversified from just text based to speech based translation. Translation whether machine or human, comes with a cost which can be divided into three segments [5]. Firstly, the linguistics knowledge of particular languages involved. Secondly, theoretical frameworks for the system to be constructed and finally, the programming skills. Note, the actual cost of each segment depend on the methodology used to implement the translation. 1.1. Motivation The need for machine translation for the over 7000 world living languages [6] cannot be under estimated for example: need for software localization [5], dissemination and assimilation of data and information over the internet, marketing etc. Therefore, each languages has got the best approach for translation based on language resources available and linguistic endowment. The motivation of the paper was to summaries all available approaches, their requirements and classify them. This would enable researchers pick the appropriate translation paradigm for a specific language weighting on the analysis of a language resources, linguistic richness of the language versus the paradigm requirement. 1.2 Methodology The methodology involved documents reviews mainly journals and conference papers and books on Machine translation plus examination of the various tools or prototypes which has been built using the approaches, Triangulation procedure was carried to ensure reliability and viability, establishing categories patterns, features and themes that are outstanding and then Pattern matches them was done at reviews stage making use of Qualitative research [7]. The