Content uploaded by Svetlana Mrs Gural
Author content
All content in this area was uploaded by Svetlana Mrs Gural on Sep 29, 2015
Content may be subject to copyright.
Available via license: CC BY-NC-ND 3.0
Content may be subject to copyright.
Procedia - Social and Behavioral Sciences 154 ( 2014 ) 360 – 363
Available online at www.sciencedirect.com
1877-0428 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of National Research Tomsk State University.
doi: 10.1016/j.sbspro.2014.10.163
ScienceDirect
THE XXV ANNUAL INTERNATIONAL ACADEMIC CONFERENCE, LANGUAGE AND
CULTURE, 20-22 October 2014
Analysis of Efficiency of Translation Quality Assurance Tools
Svetlana K. Gurala, Yan R. Chemezovb
*
abNational Research Tomsk State University, 36, Lenin Ave., Tomsk, 634050, Russia
Abstract
This study analyzes the efficiency of such computer programs and applications as translation quality assurance tools. These tools
may be used by a translator in order to improve the quality of translation. In order to elicit advantages and disadvantages of each
of them, several texts of technical and scientific subject matter were checked by means of these programs. In these texts spelling,
grammatical, terminological, and other mistakes were made deliberately. General results of the research are summarized.
© 2014 The Authors. Published by Elsevier Ltd.
Peer-review under responsibility of National Research Tomsk State University.
Keywords: Translation; equivalence; standards; quality assurance tools; applications; plug-ins; efficiency
1. Introduction
The problem of the translation quality assurance has been accompanying translation over the course of history.
But it was encouraged to be comprehended theoretically by means of the developing science of the theory of
translation in the middle of the 20th century. It should be emphasized that this problem attracts more research
attention due to the new developing movement in translation – machine translation, especially due to the
development of such technologies as machine quality assurance tools (QA Tools).
Translation is a science investigated by different approaches. When talking about “qualitative” translation,
researchers tend to “equivalent” translation making reference to the V.N. Komissarov’s theory of the five levels of
equivalence (Komissarov, 2004). An important contribution was made by Eugene Nida thanks to his theory of the
dynamic equivalence (Nida, 1964).
* Corresponding author. Tel.: +7-3822-529-695; fax: +7-3822-529-742.
E-mail address: yan_house@mail.ru
© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of National Research Tomsk State University.
361
Svetlana K. Gural and Yan R. Chemezov / Procedia - Social and Behavioral Sciences 154 ( 2014 ) 360 – 363
According to V. S. Vinogradov, equivalence or adequacy in translation should be understood as the most
complete and identical means to save genre diversity and a wide range of information of an original text in the
translation (Vinogradov, 2001). The quality of a translation depends on such factors as the translator’s skill,
peculiarities of comparable languages and cultures, the epoch of creation of an original text and a translation, a
method of translation, characteristics of translated texts etc.
Therefore, adequacy in translation has become “the first stage” of the problem of qualitative translation in the
modern science of translation. The next stage was a problem of formation of translation standards.
In 1998 a group of translators from America raised a question in Washington about the establishment of the
international translation standards aimed at the protection and enlightenment of users of translated materials. As an
example of a universal standard, there served a German standard DIN 2345, accepted in 1998.
Currently there exist such standards as a European standard 15038 – European Quality Standard BS EN
15038:2006 (prEN 15038, 2006), that has replaced the German standard DIN 2345, the Italian standard UNI 10574,
the Austrian standards Önorm D 1200 Standard and Önorm D 1201 Standard, the Danish standard Тaalmerk
Standard, the international standard ISO 12616 Standard. And what is more, many European companies went
through standardization procedures in accordance with ISO 9001. On the basis of the German standard DIN 2345
there has been created the American Standard ASTM F2575-06.
The modern global practice shows the necessity to find ultimate solutions to speed up working processes and a
control of the quality of tasks that are being performed on the competitive-oriented job market. Besides of machine
translation, there appear machine tools of translation quality assurance, which could be marked as the last stage in
the problem of developing of quality assurance tools (QAT). Translation Quality Tools (TQT), or Quality Assurance
Tools (QAT) have relatively young history beginning with a developing of such data bases Quality Memory Tools
(QMT, translation bases), that include already translated segments. The history of quality memory tools begins since
1985. The most popular tool in this area is SDL Trados. There are also known such programs as ABBYY Aligner,
Déjà vu, Omega T, SDLX, STAR Transit, Wordfast. New programs, plug-ins, and program functions that evaluated
the quality of translation began to emerge on their bases. This topic was touched upon and these programs were also
examined thoroughly by J. Makoushina (Makoushina, 2007).
2. Research design and methodology
2.1. Research Objectives
The purpose of this research is to compare quality assurance tools on the basis of eliciting their advantages and
disadvantages.
According to the purpose, the study has the following objectives:
‚ to analyze texts in programs and find out the amount of found, not found, and false mistakes;
‚ according to the results of the study, to count the level of mistakes;
‚ to consider different parameters of programs and elicit the most efficient quality assurance tools.
The methods of research used are contrastive-comparative, structural-sematic analysis of translations, as well as
quantitative-qualitative analysis of the mistakes and drawbacks found.
2.2. Method
QAT are divided into standalone applications, and QA-Plug-Ins and QA functions of TM (Translation Memory)
programs. The advantage of standalone applications is that they can work with different types of files, whereas plug-
ins are just limited with the format of the program.
Another criterion for classification can be an ability to enrich program bases in order to increase the quantity of
translation variants. Thus, there is an ability to enrich bases of Déjà vu, Trados QA Checker, while in programs like
XBench a user follows standard added-in by default into an application set of rules.
Applications can be classified on the basis of the set of rules by default.
362 Svetlana K. Gural and Yan R. Chemezov / Procedia - Social and Behavioral Sciences 154 ( 2014 ) 360 – 363
Checks performed by applications can be classified as well. Therefore, it can be elicited mistakes of what kind
these applications can find out and correct in translation. Programs check at the level of a segment (absent, missed,
incomplete, translated with mistakes, as well as segments that differ in the quantity of sentences in the source text
and the translation), compatibility checks (which elicit the same segments, that are translated differently, different
segments that are translated in the same way, as well as checks, that are performed at the level of words
compatibility), punctuation checks, numerals checks, terminology checks, etc.
Each of translation quality assurance tools performs a check according to the given parameters. Therefore, it is
possible to evaluate the effectiveness of programs in one or another type of check, taking into consideration such a
disadvantage of programs as finding “false” mistakes. By “false” mistakes we understand the ones which were
found by the program in those segments of the text where there actually were no mistakes.
In order to determine the effectiveness of a program, we can use a formula to determine the level of mistakes. It
looks in the following way: l = e * 100% / (d + n), where e = f + n. The level of mistakes – l, d – the number of
found mistakes, f – false positives, “false mistakes”, n – undetected mistakes, e – the sum of “false” mistakes
together with a number of undetected mistakes. The lower the level of mistakes, the higher the effectiveness of a QA
tool. Therefore, QA Distiller proved itself as the most reliable application. Moreover, during the research it became
clear that standalone applications proved themselves better than plug-ins of programs (SDL Trados Studio Migration
Guide, 2009).
In order to illustrate the abilities of programs as well as compare between them, 3 texts of a scientific and
technical direction, because of being interesting due to their difficulty. The difficulty of texts is the result of fixed
semasiological connections. A translator has to master special terminology. Technical texts are oriented at
professional representatives of definite spheres. Most often, these professions are clients of translation companies.
In texts, there was perversion of spelling, grammatical, lexical and other mistakes made deliberately in order to
make it clear whether these programs can find them. Another purpose was to check the reaction of programs to a
punctuation mismatch in a source text and in a translation, missed segments, repeated segments and another points
at issue, like eliciting “false mistakes”.
2.3. Results
Therefore, by checking three texts in five programs, we’ve got fifteen detailed analyses. General results of the
research are summarized in Tables 1 and 2.
Table 1. The results of working with programs (found and not found mistakes).
Programs
Mistakes (found)
Mistakes (not
found)
SDL Trados Studio 2009
28
11
QA Distiller
32
7
Déjà vu X2
18
23
Verifika
27
12
XBench
15
24
Table 2. The results of working with programs (“false” mistakes and level of mistakes).
Programs
Mistakes (False)
Level of mistakes
SDL Trados Studio 2009
7
48,72%
QA Distiller
2
23,08%
Déjà vu X2
0
56,1%
Verifika
4
41,03%
XBench
7
79,49%
363
Svetlana K. Gural and Yan R. Chemezov / Procedia - Social and Behavioral Sciences 154 ( 2014 ) 360 – 363
So far as some of the details of the research are concerned, it can be pointed out that Trados Studio 2009 and
Verifika find two of two spelling mistakes, the rest of programs (QA Distiller, XBench и Deja Vu) don’t. When
correcting translation bases and programs’ dictionaries taking into consideration terminological mistakes in six
segments, only Trados Studio 2009 and QA Distiler find them. Only Verifika can find non-translated segments
correctly. Different variants of a translation of the same word in a text (2 examples) can recognize only QA Distiller,
Déjà vu, XBench, and partly Verifika.
Analyzing the results of a comparison and collation analysis of translated texts, checked with Trados Studio
2009, QA Distiller, Déjà vu, Verifika, Xbench, the following can be said:
‚ totally, programs detected 28, 32, 18, 27, 15 mistakes correspondingly; 11, 7, 23, 12, 24 mistakes
correspondingly were not detected; 7, 2, 0, 4, 7 “false” mistakes were detected correspondingly;
‚ the level of mistakes was calculated for each program and equals to 48,72%, 23,08%, 56,1%, 41,03%, 79,49%
correspondingly;
‚ on the assumption that the most effective program is the one that has the lower level of mistakes, the most
effective translation quality assurance tool is QA Distiller with a level of mistakes of 23,08%. Such parameters as
price, convenience, and program interface were also taken into account. Though these parameters were evaluated
subjectively by the author of this paper, it can be noticed that QA Distiller has a good index in the price/quality
ratio. It should be pointed out that the most successful interface has the program Verifika, the most complex –
Trados Studio 2009. Generally, all of these applications are convenient to work in a continuous matter.
3. Conclusion
Programs that have been examined promote an optimization of an analysis of translatable texts. Where a person
can make a mistake, a program precisely finds the wrong segment if it was configured thoroughly. Programs can
optimize the work of a translator, but the translator has to make efforts to prepare a program in advance. The more
translations one has in one’s translation base, the faster and the more qualitative a process of translation can be. The
same stands for terminological bases and dictionaries. The more terms and words you’ve entered in advance, the
better is the result of one’s following translations. The results of the study showed advantages and disadvantages of
programs in each type of check.
References
Barkhudarov, L. S. (2004). Language and translation. Higher School Publishing House.(rus)
GOST 7.36–2006. (2006). System of standards on information, librarianship and publishing. Unpublished translation. General requirements and
rules for typescript. Standard-inform.
Knyazheva, E. A. (2010). The Quality of Translation assessment: in Theory and Practice. Vestnik of VSU. Series: Linguistics and cross-cultural
communication, 2, 190–195.
Komissarov, V. N. (2004). Modern Translation Science. Moscow: ETS Publishing House and Poluglossum.
Kovalchuk, E. A. (2010). Translation quality control: a search for effective methods, standards and parameters. In: Bulletin #11-2 (2) (pp.81–85).
Komsomolsk-on-Amur: Komsomolsk-on-Amur State Technical University.
Lipatova, V. V. (2010). On the importance of national and international translation quality standards. Moscow: People’s Friendship University
of Russia publishing house.
Minchenkov, A. G. (2008). Criteria for Translation Quality Assessment and Types of Translator’s Errors. In: Bulletin of the Saint-Petersburg
University № 2, P. II (pp. 166–175). Saint-Petersburg: Saint-Petersburg State University Publishing House.
Minyar-Beloruchev, R. K. (1980). General Theory of Translation and Oral Translation. Moscow: Voennizdat.
Nida, E. A (1964). Toward a Science of Translating. Leiden: E.J. Brill.
Makoushina, J. (2007). Translation Quality Assurance Tools: current state and future approaches. Tomsk: Palex Languages and Software.
prEN 15038. (2006). Translation services – Service requirements. CEN.
SDL Trados Studio Migration Guide (2009). SDL pl.
Vinogradov, V.S. (2001). An introduction to translation studies (general and lexical aspects). Institute of General Secondary Education
Publishing House. (rus).
