Volume 4, Issue 2, 2022, Pages 22 - 42

Author(s) : Samin Salajegheh* ¹

Abstract :
Abstract Online machine translators provide users with translator models based on their previously incorporated corpus. Different works have been done in this field yet machines have some deficiencies needing to improve. The goal of this paper was to evaluate and compare two free translation machines which are Google (https://translate.google.com/) and Targoman (https://targoman.ir/) in terms of translating English adverbs into Persian and suggest the more qualified machine. It was done according to Doug Arnold’s quality assessment method which considers elements of intelligibility, accuracy, error analysis and test suit. The study contained a test suite of 64 English sentences which were categorized based on the types of adverbs they included. In some cases, Targoman was more similar to the acceptable or reference translation but in total, Google was more accurate in specifying the adverbs and could present more pleasing equivalences. The ultimate correctness percentage of the translated adverbs of Google were 93.75% and this percentage for Targoman was 79.68%. In some cases, the translated sentence was natural and acceptable in the target language (Intelligible as Arnold says) but having a partial view to the translation of adverb, it was not properly translated (the element of accuracy) or even vice versa. Meanwhile, both machines were helpful in most of the cases alongside their problems.

Keywords :
Adverb, Google translate, Machine, Reference, Targoman translate