The impact of using Machine Translation (MT)

on EFL students’ writing

Pushpendra
School of Information Technology and Engineering, Vellore Institute of Technology, Vellore, Tamil Nadu, India

Abstract
In the last modern years, we have seen an immeasurable increment in the field of technologies
especially in machine learning which includes machine translation also. This great invention
relaxes us in many ways like in the fields of health, education, research, etc. Now, coming to the
point, this article is all about the machine translation and its impressions on the writing of EFL
students. As we know that, these students are from non-English speaking country and learning it
as a foreign language which is somehow difficult, to solve this problem machine translation has
been introduced in their learning. In this article, there are some researches and valuable points
which proves how machine translation benefits these students in writing and learning their non-
official languages. It also shows that how effectively technology is participating in learning and
reducing the grammatical errors in the students’ writing, making their second language easier
and more attractive. Although, machine translation has founded as a useful aid, but there are also
some limitations and some negative impacts. Hence, teachers must aware of these things and
should always provide adequate guidance to the EFL students to achieve the maximum benefits
of machine translation in improving their learning and writing skills.

Introduction
In this rising era of globalization, interaction between the world has increased a lot which tends
to the growth of learning foreign languages for the communication, learning these languages
aren’t so easy so translation has been introduced in this. Translation is not new to human it is
known to be existed since the civilization of human, but now only the era of cross-language
communication in the form of translation came into existence. Earlier, machine translation was
not widespread context academically, this is mainly due to its reliability issues, there were lots of
doubts on its accuracy to translate the first language text in the required second language text.
However, there was a great development can be seen in past decades in the machine translation
which enhances its accuracy much better. Also, the grammatical errors and lexical accuracy is
improved a lot.
Machine translation, as we already know, is the process through which the source text is
translated into target text by the help of computer software. It is available for both devices PCs as
well as smartphones. It is well known for its convenience, efficiency, multilingualism,
immediacy, and also because of its free cost. MT is now being increasingly used by most of the
EFL students for various diverse educational motives in their day-to-day lives. These EFL
students also used machine translation for increasing vocabulary, reading comprehension and
writing assignments, thus MT has become a great supplementary tool for the learning. According
to literature, MT has benefitted learning by three perspectives: - linguistic, cognitive and
affective. The linguistic perspective promotes writing and reading comprehension and thereby
encouraging language learning. The cognitive perspective minimizes the intellectual burden as it
executes the elementary translations hence encouraging the self-directed learning. The affective
perspective let downs the language anxiety and rises the confidence and motivation thus building
a non-frightening learning environment.
The present-day situation can be summed up as follows: the need of MT in learning is rising but
due to its reliability it has not been fully firmed. Therefore, to produce positives outcomes of MT
in learning we need to regulate its use and investigate the facilitation of MT on basis of authentic
clues and concede whether it is a boon or bane in learning. Till now, some researches on these
issues have been done which says that MT as an unsatisfactory tool, having lexico- grammatical
errors at a large scale.
Nevertheless, this ground work will also consider MT as a perfect tool which can accommodate
EFL students a help to correct their mistakes of lexico-grammatical errors in their learning
language. This study avails of a peculiar function layout which helps the EFL students to
translate their source language to the target language on their own and then rectifying it using
machine translation (MT) for better comparison. The main objective of this research is to
scrutinize the usefulness of machine translation in improving EFL students’ writing.

Review of literature
Earlier studies prove that MT could be a productive learning tool while writing the target
language. There have been two studies by Pena & Garcia(2011) and Alireza and Ali(2014)
which declared that MT has benefitted learners by boosting up their writing speed with more
fluency and also having least errors. It helps in expressing themselves and interacting more in
their target language. As reported by the study of Godwin-Jones’s(2015), MT enhanced writing
of EFL students by giving a platform to their learning. Another study by Nino’s(2008)
demonstrated that MT has enhanced many of student’s skills like paraphrasing which is
necessary for writing. Similarly, another study of Meurers & Amaral(2011) declared that MT
can support while writing target language by providing linguistic modelling & awareness,
increasing lexical, semantic, syntactic & pragmatic knowledge.
Machine Translation can increase lexical fluency and studies have proved that they also
compared MT with other reference tools available and terminate to the point that machine
translation gives better results in respect of technical phrases and collocations as compared to
other traditional tools. Many studies also claimed that MT helps students in doing the revision of
their learning target language. Pena & Garcia(2011) professed that postediting of MT helps in
focusing and concentrating more on the processes of editing ang writing. A study of
Kliffer(2005) claimed that postediting of machine translation also helps students to put in their
prerequisite knowledge of learning about their target language and makes some improvement in
them. MT also works as an assistant to those students which struggles to obtain individualized
response regarding their writing by providing particularized feedback and also recognizing and
fixing errors. MT also focuses lexico-grammatical errors correction and also precise the EFL
students by forcing them to find errors in their writings, makes correct decisions, solve their
problems on their own, etc. Thus, they enhanced their individualistic and self-determined
learning skills related to their target language.
Studies also have showed some drawbacks of MT, they stated that whether it aids in growing
rhetorical intelligence and sensibility regarding the rules of grammar in background, but it fails
in offering grammatical errors solutions (Josefsson, 2011). Furthermore, machine leaning
supports language learning only at lexico-grammatical scale but not at discourse scale (Mundt &
Groves, 2015). A study led by Shei’s (2002a, 2002b) has raised MT lexico-semantic, idiomatic
and cultural limitations. Another statement given by him was that “In MT outputs the
grammatical errors are produced basically because of badly incorporated language input as well
as shortage of grammar systems in machine translation”. It was always controversial, but the
previous studies founded that MT has gained a lot in language learning.
Hence, this research explores MT considering a great tool in EFL students’ writing based on
various data origins. Whereas previous studies targeted mainly on MT postediting, this study
mostly queried EFL students to modify their target language writing with the help of machine
translation.

Methodology
The research was done on basis of a survey which was held at a university in Korea for about
one and half months. There was total 34 students out of which 13 are males and 21 are females,
also English was majored in all. Their first language was Korean and target language was
English. The students were given an assignment of writing about the language used by the young
ones of present generation who were besmeared in these mentioned steps:
I. Watching a TED video of 15 minutes in a topic.
II. Writing some paragraphs about any topic in their first language.
III. Translating their own writing into English by barring the service of MT.
IV. Translating their own writing again into English by machine translation.
V. Modifying their own English translation by checking with the writing translation done
using MT.
Throughout step III, the EFL students were granted permission for using the traditional assets
like dictionary and books whereas step V allowed them to use MT. At conclusion, it was
observed that all of them used Google Translate except four of them who preferred another MT
tool named Papago which was developed in Korea as they were familiar to it. After finishing
every work, the EFL students were individually interviewed through writing conferences and
they also come up with conclusion in the form of reflection papers at end.
This study consists of a mixture of methods which includes an analysis of EFL students’ writing
in a quantitative way, also the submitted reflection papers of students and their given interviews
are analyzed in a qualitative way. We can see the summarized collection of data and analysis in
the below chart:
The writings of EFL students was also inspected by the analysis of texts in multiple processes. In
first process source text was examined for checking overall quality of their writing seeking more
attention to the content and their meaning as these were probably same in both students’ writing
and MT translated texts. Throughout the next steps, both the texts of students’ writing and MT
translation are checked with errors, the results are calculated and classified into two: - lexical
errors and grammatical errors. While past researches used only analysis based on errors to assess
the precision of machine translation (Balan & Aiken, 2011 and Mundt & Groves, 2015), this
modern abstraction of research is exclusive of all as it uses the analysis of mistakes and errors to
count the percentage of improvement in writing of student’s target language. The past studies
evaluated only grammatical errors whereas this study evaluated the errors according to their
levels (use of symbols, words, phrases, clauses, sentences, etc.) and types (lexical or
grammatical).
The final step was rectifying the differences of MT texts and recognizing each difference
according to its level and purpose. Two of reputed university professors have marked students’
writing outcomes and also identified their mistakes. For the translation without MT, the
reliability was found to be 91% and by using MT it was 93%. At end, the reflection papers and
interviews data were also collected and analysis was done based on their performances. The 10
minutes interview covered all the pros and cons of MT in writing student’s target language,
approaches done for correcting error while revising, intention which are underlying the changes.
This has also helped the students to clarify their cognition about machine translation and their
purpose in making the changes in their writings according to MT texts, which couldn’t have been
determined.

Result Analysis
The result outcomes of this study were glorifying and acceptable. Most of the students had
produced decent and adequate writing quality in their first language, also their initial scores of
target language varies between 2-6 having ‘3.76’ as an average on a 6-pointer scale(4 students
scored 2 marks ; 10 students scored 3 marks ; 14 students scored 4 marks ; 2 students scored 5
marks ; and 4 students scored 6 marks). While comparing with initial scores, the average of their
final one which uses MT has increased significantly by 0.83, it was calculated as ‘4.59’.
The mean of total errors(grammatical = 15.67 ; lexical = 5.97) in former was ‘21.94’ while the
later one has only a mean of ‘13.64’ which is far below, as grammatical error falls to 9.82 and
lexical to 3.82, this was only due to comparison and modification done by students with their
own writing and that of using MT. The mean of total changes made by EFL students in final MT
text was found to be 16(lexical level mean = 8.57 ; phrasal level mean = 3.55 ; clause/sentence
level mean = 3.91 ; symbol/paragraph level mean = 0 ‘no changes’). It was also observed that
during revision the EFL students found difficulties in adopting machine translation in their
learning.