Understanding Efficiency of Language Translation Software

This study analyzes the effectiveness of language translation software through survey data analysis. It discusses the background, research design, sampling methods, and outcomes.

Understanding efficiency of language translation software
(Survey on students)
FirstName Surname†
Department Name
Institution/University Name
City State Country
[e-mail]
FirstName Surname
Department Name
Institution/University Name
City State Country
[e-mail]
FirstName Surname
Department Name
Institution/University Name
City State Country
[e-mail]
ABSTRACT
The translation tools perform various translating activities with precise accuracy.
Besides, every tool is unable to complete the tasks correctly. The following study
demonstrates the effectiveness of the tools of language translating. Here, the data
analysis of various survey results is done. The survey was done on multiple
computer science students of computer science staying in the United Kingdom. The
report discusses the background situation. Then it demonstrates the research
design. Next, different sampling methods, data usages and experimental designs are
analyzed. At last the outcomes are evaluated in this report.
Keywords
Language translation software, software tools, survey
INTRODUCTION
In the present era of the Internet, the issue of translating from one
language to other is not a secure method. This word-for-word translating has never
been working effectively. This is because there have been different nuances in
language that are found to be getting lost during translation. These type of
inaccuracies has been notably changing the context of a specific message.
The ideal solution is that the translation tool can perform the tasks with a precise
amount of accuracy. These tools abound. However, not every tool has been
performing the task correctly. To go for an efficient translating tool for the website,
one can feature those helpful tools for translating the website content.
The following report highlights the efficiency of language translating
tools. For this, a survey is conducted on various computer science students at U.K.
Their age group lies within 25. At first, the background of the situation is analyzed.
Then the research design is studied. Then the sampling methods, experimental
designs and data usages are demonstrated. Lastly, the results are investigated in
this study.
DISCUSSION OF BACKGROUND LITERATURE
It is seen that the worldwide economy has expanded the potential
market ineffective way. This was also impossible ten years before. They have been
leveling the playground for big and small businesses [3]. Further, it has not been
devoid of some challenges. Here, one of the important one if the language barrier.
The accuracy of Google Translate has been comparable for human translators. This
is because of currently upgraded “Google Neural Machine Translation system”.
However, this has been found to be occasionally dropping words [1].
Further, it has mistranslated different non-standard sentences. Here the
object of the preposition has not been transparent. However, it has been not
entirely accurate overall. This has been appended to proper punctuation mark for
them. Besides, they have offered a guide of pronunciation as they need to know the
sounds of translated content [4].
As per the latest figures from the W3Techs, about 50% of the overall
content over the Internet has been written down in English. However, this has been
light years ahead of the residual top 5. It is 6.3% for Russian, 5.7% for German, 5.0%
for Japanese and 4.9% for Spanish [5]. However, this never indicates that almost
half of the Internet has been still inaccessible. This is unless one is fluent in
numerous languages. Here, various reasons have been there that one might require
to read the content in any other form of language. This also includes the
understanding of various local news stories for researching numerous places [6].
The Google translate has been using more than ninety supported languages,
versions of numerous browsers, 200 million daily users and operating systems.
Thus Google translate has remained as the undisputed ruler of translating [9].
The Microsoft Translator is regarded as a multilingual machine
translation service of a cloud. Microsoft provides this. This has been integrated into
various consumer, enterprise products and developer. This includes Bing [7]. It also
offers speech and text translation with the help of cloud services for the business.
Here, the service for text translation through Translator Text API has been ranging
from different free tier. This is done by supporting two million characters every
month [8].
RESEARCH DESIGN
The research is done through survey questionnaires. These questions
are asked to various students of age group 25. They are computer science learners
in the United Kingdom. Various steps are undertaken by research and development
team of various IT companies at U.K. who is in charge of the research. They have
undergone multiple stages to examine the survey data. As appropriately conducted,
the reporting processes and analysis has been delivering timely and accurate data.
This is about a huge population that has been unavailable otherwise. This is helpful
for the stakeholders of those IT companies to undertake decisions regarding large
practice or policy in agencies. The steps have included reviewing the plan of
analysis, checking and preparing data files, calculating the rates of responses,
measuring the summary statistics. Lastly, the results input into tables and charts as
shown below.
The students were already asked to test the latest products. Then the
survey questionnaire was sent to fetch what they have responded and where they
are to make improvements. The results are helpful to suggest changes and
recommend the products to others. The feedback received must be refining the
software tools.
UNDERSTANDING THE CONSIDERATION OF SAMPLING,
EXPERIMENTAL DESIGN AND DATA USAGE:
In this study, the stratified sampling is used. This includes the usage of
stratum and a subset of the targeted population. Here, the members have possessed
one and more common type of attribute. Here, the students mentioned above are
taken as a stratum. Furthermore, the sampling error has been lesser within
stratified sampling than random sampling [7].
THE EXPERIMENTAL DESIGN
The experiment is done by providing a questionnaire through the mail.
Rating on any specific scales is tested and tried a form of the question structure. It
is helpful as the researchers seek more open-ended questions than possible with
various multiple choice questions. This is complicated to analyze the reactions.
Moreover, one can assure that the scale has been permitting extreme views.
Different questions that are asked for opinions has been open-ended and the
subject is allowed for providing their reactions. The entrapment must be avoided
and appeared neutral as probable during the overall process. Here, the most critical
issue is that one needs to devise the numeric manner of assessing and statistically

WOODSTOCK’18, June, 2018, El Paso, Texas USA F. Surname et al.
evaluate the reactions. This has been leading to the biased view as any care is not
considered.
DATA USAGE
The survey research is helpful and the data can be used legitimately for
research that has clear-cut benefits. This is helpful to explore and describe various
constructs and variables of the interest. The data might have the potential for
various sources of errors. However, various strategies have been existing to
decrease the efficiency of error. Here, the researchers can determine the way and
how the analysis from the data is applicable in practice.
RESULTS
Data analysis for Question 1:
Responses
Participant 1 Yes
Participant 2 Yes
Participant 3 Yes
Participant 4 Yes
Participant 5 No
Participant 6 No
Participant 7 Yes
Participant 8 Yes
Participant 9 No
Participant 10 Yes
Number of "Yes" 7
Number of "No" 3
Number of "Yes" Number of "No"
Figure 1: “Analysis for Question 1”
(Source: Created by Author)
Data analysis for Question 2:
Responses
Participant 1 Yes
Participant 2 No
Participant 3 Yes
Participant 4 No
Participant 5 No
Participant 6 No
Participant 7 Yes
Participant 8 Yes
Participant 9 No
Participant 10 Yes
Number of "Yes" 5
Number of "No" 5
50%50%
Number of
"Yes"
Number of
"No"
Figure 2: “Analysis for Question 2”
(Source: Created by Author)
Data analysis for Question 3:
The responses shows that the language translators could be coped up
to reach larger audience. As the services and products are unable to meet the wider
audience, the translation in foreign nation is helpful to open up to markets that
have never existed before. Next, in the rising era of information technology, the
foreign firms taking the help of viable services of translation in having the
documents translated in the suitable language. Besides, the legal translations
helpful in pertaining the legal matters in foreign nations.
Data analysis for Question 4:

Responses
Participant 1 Very likely
Participant 2 Likely
Participant 3 Not sure
Participant 4 Very likely
Participant 5 Very unlikely
Participant 6 Likely
Participant 7 Unlikely
Participant 8 Likely
Participant 9 Very likely
Participant 10 Likely
Number of "Very Likely" responses: 3
Number of "Likely" responses: 4
Number of "Not sure" responses: 1
Number of " unlikely" responses: 1
Number of "very unlikely" responses: 1
0
1.5
3
4.5
Figure 3: “Analysis for Question 4”
(Source: Created by Author)
Data analysis for Question 5:
Responses
Participant 1 often
Participant 2 sometimes
Participant 3 Very often
Participant 4 often
Participant 5 never
Participant 6 often
Participant 7 Very often
Participant 8 Very often
Participant 9 often
Participant 10 often
Number of "Very often" responses: 3
Number of "often" responses: 5
Number of " sometimes" responses: 1
Number of "never" responses: 1
0
2
4
6
Figure 4: “Analysis for Question 5”
(Source: Created by Author)
Data analysis for Question 6:
The various reactions proves that accuracy is many times not offered
by the tools in consistent way. The translations are done on word to word basis.
This is not without any comprehension of the data that is needed to be corrected in
manual way. Besides, formal and systematic rules are not followed by machine
translations. Hence, it can never concentrate on the context. Further, it is unable to
solve ambiguity and never uses the metal outlook or experience that is done by
human brain.
Data analysis for Question 7:
Responses
Participant 1 No
Participant 2 Yes

WOODSTOCK’18, June, 2018, El Paso, Texas USA F. Surname et al.
Participant 3 Yes
Participant 4 Yes
Participant 5 No
Participant 6 Yes
Participant 7 Yes
Participant 8 Yes
Participant 9 No
Participant 10 Yes
Number of "Yes" 8
Number of "No" 2
Number of "Yes" Number of "No"
0
5
10
Figure 5: “Analysis for Question 7”
(Source: Created by Author)
Data analysis for Question 8:
Responses
Participant 1 50+
Participant 2 31-35
Participant 3 26-30
Participant 4 50+
Participant 5 15-20
Participant 6 31-35
Participant 7 21-25
Participant 8 50+
Participant 9 21-25
Participant 10 50+
Number of "15-20" responses: 1
Number of "21-25" responses: 2
Number of "26-30" responses: 1
Number of "31-35" responses: 2
Number of "50+" responses: 4
0
2
4
Figure 6: “Analysis for Question 8”
(Source: Created by Author)
Data analysis for Question 9:
Responses
Participant 1 Very easy to use
Participant 2 easy to use
Participant 3 Very easy to use
Participant 4 easy to use
Participant 5 easy to use
Participant 6 Okay
Participant 7 easy to use
Participant 8 Very easy to use

Participant 9 Very hard to use
Participant 10 hard to use
Number of "Very
easy to use"
responses: 3
Number of "easy to
use" responses: 4
Number of "okay"
responses: 1
Number of "hard to
use" responses: 1
Number of "very
hard to use"
responses:
1
0
1.5
3
4.5
Figure 7: “Analysis for Question 9”
(Source: Created by Author)
Data analysis for Question 10:
The reactions highlights the fact that accuracy of translation is relying
on the translated languages. For instance, converting from French to English is
simple. This is because the two languages has been sharing much. However,
languages where a greater jump is needed, the accuracy cannot be maintained
always. For examples, translation from Chinese to English can be considered here.
DISCUSSION
The above results proved that the overall amount of the created
content and sustained around the enterprises for various initiatives is vast. It has
been continuing to be growing every day. Since translation done by human beings
has been costly, most of the contents have been remaining in one language. The
software translation tools can customize different specific business goals and
domain-specific subjects. They are helpful for the enterprise and provide contents
for users across the world in over sixty languages. All the companies, irrespective
of the size can monetize and leverage the present information and gain an extra
return on investment. This can be done through multilingual content from a single
source document with the solutions of software translation tools.
CONCLUSION
The researchers have needed to understand the cost-effective solution
from the responses of the students. The study is helpful to understand how to
deliver efficient translation quality along with saving time. This is commonly
applied to the environment of the enterprise. This indicates that the translation
should be reaching the quality threshold that is defined by the business. It is
achievable in various ways. This includes customizing the software to impose
consistency of language and controlling the terminology. Here, the initial cost has
been to reach the quality threshold for each domain if the target. This has been
leveraging every linguistic resource available like present translated texts,
translation memories and glossaries. However, the organizations have also required
to consider various related ongoing expenses for improving the quality of
translation of the current domains. This might also include the quality of the
currently targeted domains. Thus it can be said that the investments can be cost-
effective for the IT companies. In this way, the software tools can be easy to
maintain, quick over the present hardware, accurate and provide savings on the
current costs of translations.
REFERENCES
[1] D, Chen. A Linguistic Evaluation of the Output Quality of'Google
Translate'and'Bing Translator'in Chinese-English Translation (Master's thesis,
NTNU), 2017.
[2] I. Ramati and A. Pinchevski. Uniform multilingualism: A media genealogy of
Google Translate. New Media & Society, 2017, p.1461444817726951.
[3] A. Hosseinzadeh Vahid, Arora, Q. Liu and G.J, Jones. A comparative study of
online translation services for cross language information retrieval. In Proceedings
of the 24th International Conference on World Wide Web, 2015 (pp. 859-864). ACM.
[4] R. Roller, M. Kittner, D. Weissenborn, and U. Leser, U. Cross-lingual Candidate
Search for Biomedical Concept Normalization. arXiv preprint arXiv:1805.01646,
2018.
[5] J. Halpern. Very Large-Scale Lexical Resources to Enhance Chinese and
Japanese Machine Translation. In Proceedings of the Eleventh International
Conference on Language Resources and Evaluation (LREC-2018), 2018.
[6] B. Reyes Ayala, R. Knudson, J. Chen, J., G. Cao and X. Wang. Metadata records
machine translation combining multi‐engine outputs with limited parallel data.
Journal of the Association for Information Science and Technology, 69(1), 2018, pp.47-
59.
[7] D. Katan. Translation at the cross-roads: Time for the transcreational turn?.
Perspectives, 24(3), 2016, pp.365-381.
[8] G. Medvedev. Google translate in teaching English. Journal of Teaching English
for Specific and Academic Purposes, 4(1), 2016, pp.181-193.
[9] F.A. Thorne-Ortiz and N.V. Gilman. Librarians’ Everyday Use of Google Tasks,
Voice, Hangouts and Chat, Translate, and Drive. The Complete Guide to Using
Google in Libraries: Instruction, Administration, and Staff Productivity, 1, 2015, p.235.
[10] A. Freitas, S. Barzegar, J.E. Sales, S. Handschuh and B. Davis. Semantic
relatedness for all (languages): A comparative analysis of multilingual semantic
relatedness using machine translation. In European Knowledge Acquisition
Workshop, 2016, (pp. 212-222). Springer, Cham.
APPENDIX
The survey questions are listed hereafter.
1. Have you ever used Google translate?
YES [ ] NO [ ]
2. Have you ever used a translation app when in a foreign country?
YES [ ] NO [ ]
3. If you don’t have access to a Language translator how will you cope in
a foreign country?

WOODSTOCK’18, June, 2018, El Paso, Texas USA F. Surname et al.
[ Free Text ]
4. How likely are you to use a language translation software/app or
service in the next twelve months?
Very likely [ ] likely [ ] not sure [ ] very unlikely [ ] unlikely [ ]
5. How often do you use a particular Language Translation Software or
service?
[ Free Text ]
6. What difficulties do you face whilst using a Language Translation
Software or service?
[ Free Text ]
7. Would it be helpful if Language translation done today is fluent and
real-time?
YES [ ] NO [ ]
8. What age groups are more likely to face problems regarding
understanding a particular language?
15-20 [ ] 21-25 [ ] 26-30[ ] 31-35[ ] 50+ [ ]
9. Considering the last software/app/service you used to translate-how
would you rate its user friendliness?
Very easy to use [ ] easy to use [ ] okay [ ] hard to use [ ] very hard
to use [ ]
10. How accurate is todays Language Translation Software or service?
[Free Text ]