Function-based Translation Quality Assessment

Rudy Sofyan

, Bahagia Tarigan

Linguistics Department, Universitas Sumatera Utara, Jalan Abdul Hakim Kompleks USU, Medan, Indonesia

Keywords: Translation Quality Assessment, TQA Model, Translation Function.

Abstract: This article is aimed at (i) finding out the results of TQA using three TQA models, and (ii) developing a TQA

model. The research used a descriptive qualitative method using a process-oriented translation as the

approach. The data were the results of the assessment and the questionnaires from five raters. The raters were

translation experts and professional translators who were asked to assess the TT by using three different TQA

models. The first model (Model A) was proposed by Hurtado (1995), the second model (Model B) was

proposed by Waddington (2001), and the third model (Model C) was proposed by Nababan et al. (2012). The

TT assessed was entitled “Sejarah Awal Yellowstone National Park” translated by a professional translator.

This research found that (i) using the three different TQA models results in different level of quality of the

TT caused by two factors: the absence of text/sentence function as the quality aspect and the limited

description of quality level; (ii) the model developed in this article is called function-based translation quality

assessment using a holistic method of assessment in which the whole criteria of translation quality (accuracy,

finding equivalents, translation skill, text/sentence function and grammar and TT style) are assessed.

1 INTRODUCTION

Quality translation is an absolute condition to be

achieved by every translator because of which, as

Williams (2004) argues, translation quality has

become a central issue in a product-oriented

translation studies for such factors as aesthetic,

religious, political, pedagogical, administrative and

economic. Nevertheless, such factors generate

various questions concerning the criteria of quality

translation product. In addition, the definition of

translation quality as the translation producing a good

target text (TT), proposed by Schäffner (1997),

certainly brings about more various questions such as

“What is a good TT?”, “What are the characteristics

of a good TT?”, or “What are the elements of the

source text (ST) that need to be added, substituted or

omitted in the TT to produce a good TT”. Therefore,

it is necessary to design and develop a model that

hopefully would be able to distinguish different level

of translation quality.

To date, translation quality assessment (TQA)

continues to receive serious attention from the

researchers in translation studies. Studies on the

theory of translation assessment (Farahzad, 1992;

Hurtado, 1995; House, 1998; McAlester, 2000;

Bowker, 2001; Melis and Hurtado, 2001; Williams,

2001, 2009; Waddington, 2001; Khanmohammad

and Osanloo, 2009; Nababan, Nuraeni, and

Sumardino, 2012) produced several models of TQA.

In his research on the quality of translations into a

foreign language, McAlester (2000) states that the

methods to be used in TQA should be reliable, valid,

objective, and practical. In relation to McAlister’s

reliability, validity, objectivity and practicality in

assessing translation quality, Williams (2004)

proposes that TQA could be done by using either (1)

models with a quantitative dimension and (2) models

with non-quantitative dimension, text-logical models,

or both. The second type refers to House’s (1998)

models of TQA. In his further study, Williams (2009)

argues that TQA could be qualitative or quantitative

where the assessment could be based either on

mathematical or statistical measurements or on reader

response, interviews and questionnaires.

Although many experts in translation studies have

designed and developed various models of TQA, the

problem of TQA does not stop because certain TT

might have different quality when assessed by

different models of TQA. This is a serious problem

that needs serious attention and exploration to find

out the factors leading to such problem. This research

is aimed at (i) finding out the results of TQA using

three models of TQA, and (ii) developing a model of

TQA.

Sofyan, R. and Tarigan, B.

Function-based Translation Quality Assessment.

DOI: 10.5220/0010099317551764

In Proceedings of the International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches (ICOSTEERR 2018) - Research in Industry 4.0, pages

1755-1764

ISBN: 978-989-758-449-7

1755

Table 1: Hurtado’s error analysis in TQA.

No. Evaluated Aspects

Minor

Errors (-1)

Serious Errors (-

1 Providing improper equivalents affecting the ST understanding

a. Contresens (-1) (-2)

. Faux sens (-1) (-2)

c. Nonsens

(

-1

)

(

-2

)

d. Addition

(

-1

)

(

-2

)

e. Omission

(

-1

)

(

-2

)

f. Unresolved extralinguistic references (-1) (-2)

g. Loss of meaning (-1) (-2)

h. Inappropriate linguistic variation (register, style, dialect, etc.). (-1) (-2)

2 Providin

er ex

ressions in the TT

a. S

ellin

(

-1

)

(

-2

)

. Gramma

(-1) (-2)

c. Lexical items (-1) (-2)

d. Text (-1) (-2)

e. St

(

-1

)

(

-2

)

3 Providing insufficient equivalents affecting the transfer of either the main

function or secondary functions of the ST.

(-1) (-2)

4 The plus point Good

Solutions

(+1)

Exceptionally

Good Solutions

(+2)

The first model of TQA is developed by Hurtado

(1995). The model is developed mainly based on error

analysis that classifies the error into either serious

error penalized with (-2) points or minor error

penalized with (-1) point. The errors are grouped into

three categories as shown in Table 1.

In addition to the three categories of errors, this

model also provides additional category containing

the plus points: good solutions to translation problems

are awarded +1 point and exceptionally good ones are

awarded +2 points.

Unlike the first model emphasizing on error

analysis, the second model developed by Waddington

(2001) adopts a holistic method of assessment.

Waddington (2001) designs a five-level scale of

assessment whose description of translation quality is

shown in Table 2.

Table 2: Waddington’s holistic model of TQA.

Level Quality of expression in TL

Degree of task

com

letion

Mark

Level

The ST content is completely

transferred. Only minor revision is

needed to reach professional standard.

Almost all the translation reads like a text

originally written in English. There may be

minor lexical, grammatical or spelling errors.

Successful 9, 10

Level

The ST content is almost completely

transferred. There may be one or two

insignificant inaccuracies. The TT

requires certain amount of revision to

reach professional standard.

Most of the translation reads like a text

originally written in English. There are a

number of lexical, grammatical or spelling

errors.

Almost

completely

successful

7, 8

Level

The transfer of the general idea(s)

contains a number inaccuracies. The

TT needs considerable revision to

reach professional standard.

Certain parts of the translation read like a text

originally written in English, but others read

like a translation. There are a considerable

number of lexical, grammatical or spelling

errors.

Adequate 5, 6

Level

The transfer of the general idea(s)

contains serious inaccuracies. The TT

needs a lot of revision to reach

rofessional standard.

Almost the entire text reads like a translation;

there are continual lexical, grammatical or

spelling errors.

Inadequate 3, 4

Level

The transfer of the ST content is totally

inadequate. The translation is not

worth revising.

The candidate reveals a total lack of ability

to express him/herself adequately in English.

Totally

inadequate

1, 2

ICOSTEERR 2018 - International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches

1756

The third method is developed by Nababan et al.

(2012). Like Waddington’s model, this is also a

holistic model which assesses the translation quality

through three instruments, i.e. accuracy, acceptability

and readability. Each of the instruments is divided

into three parts: translation categories, scores and

qualitative parameters. The score ranges from 1

(describing inaccuracy, unacceptability and low

readability level) to 3 (describing accuracy,

acceptability and high readability level). Even though

the model uses three different instruments, it is a

holistic assessment because the scores from each

instrument will be combined to produce the final

score which eventually describes the quality of the

assessed translation (see Table 3).

Table 3: Nababan’s holistic model in TQA.

No. Qualit

ects Evaluate

Mar

1 Accurac

n x 3

2 Acce

tabilit

n x 2

3 Readability n x 1

Letter ‘n’ in Table 3 represents the mark given for

the aspect of accuracy, acceptability and readability.

The highest score (x3) given to accuracy aspect

shows that Nababan’s model gives more emphasis on

accuracy in assessing the overall quality of the

translation. The final mark of the assessment is

obtained by dividing the sum of the quality aspects

evaluated with 6.

2 RESEARCH METHODOLOGY

This research used a descriptive qualitative method

using a process-oriented translation as the approach.

The data were the results of the assessment and the

questionnaires from five raters. The raters were

translation experts and professional translators who

were asked to assess the TT by using three different

models of TQA. The first model (Model A) was

proposed by Hurtado (1995), the second model

(Model B) was proposed by Waddington (2001), and

the third model (Model C) was proposed by Nababan

et al. (2012). The TT that was assessed was entitled

“Sejarah Awal Yellowstone National Park” which

was translated by a professional translator from

English whose title was “Early History of

Yellowstone National Park”. In the process of data

collection, before the raters assessed the TT, they

were given the explanation of how the three models

of TQA work. Having assessed the TT, they were

given an open questionnaire containing their

experience in applying those models. Then the results

of the assessment and questionnaires were analyzed

to find out whether their applications resulted in

similar level of the TT quality. Then, based on this

analysis, a new model of TQA was developed.

3 RESULTS AND DISCUSSION

The data analysis shows that assessing the translation

by using the three different models of TQA applied

by the five raters results in different level of quality

of the TT. Table 4 shows the results of assessment

using Model A.

Table 4: The results of TQA using model A.

No. Raters

Translation Qualit

Score Descri

tion

1 Rater A 82 Goo

2 Rater B 81 Goo

3 Rater C 81 Goo

4 Rater D 78 Fairly Goo

5 Rater E 83 Goo

The results of the assessment from the five raters

using Model a show the final scores of 82, 81, 81, 78

and 83, respectively. This means that four out of five

raters agree that the TT quality is good, and only

doe’s one rater state that the TT quality is fairly good.

Meanwhile, the results of the assessment from the

five raters using Model B (see Table 5) show the final

scores of 70, 61, 68, 60 and 61, respectively. This

indicates that four out of five raters agree that the TT

is almost completely successful.

Table 5: The results of TQA using model B.

No. Raters

Translation Qualit

Score Descri

tion

1 Rater A 70 Almost completely

successful

2 Rater B 61 Almost completely

successful

3 Rater C 68 Almost completely

successful

4 Rater D 60 Ade

uate

5 Rater E 61 Almost completely

successful

The results of TQA presented in Table 5 show that

none of the raters evaluates the TT as a successful

translation product, or a TT with a good quality. This

is reflected through the different score of the TT

displayed in Table 4 and Table 5.

Furthermore, by using the scoring system of

Nababan et al. (2012), four out of five raters agree that

Function-based Translation Quality Assessment

1757

the quality of the TT is average. Their scores are 2.50,

2.17, 2.33, 2.17, 2.33, respectively, as displayed in

Table 6.

Table 6: The results of TQA using model C.

No. Raters

Translation Quality

Score Description

1 Rater A 2.50 Good

2 Rater B 2.17 Average

3 Rater C 2.33 Average

4 Rater D 2.17 Average

5 Rater E 2.33 Average

These results lead to the finding that the TT has

different level of quality when assessed by using

different models of TQA. Although it is impossible to

get the exactly similar results of assessment by using

different models or methods of TQA, the range

should not be too wide because it will result in the

assessment uncertainty in terms of which translation

has good quality. Consequently, the finding needs to

be further explored since any models used in

assessing translation should arrive at the same or

nearly the same level of quality. The main cause of

such different assessment results is the absence of

evaluation on the text or clause function in the

parameters or instruments used in the three models.

Look at the translation in (1).

(1) ST : Native Americans have first claim on the

Yellowstone Plateau and lived in the area

in peaceful tranquility until the early

1800s--undisturbed by the presence of

white men.

TT : Penduduk asli Amerika pertama sekali

mendiami Yellowstone Plateau dan

tinggal di daerah itu dengan damai

sampai awal tahun 1800an--tanpa

diganggu oleh kehadiran orang kulit

putih.

Using Model A, Rater A found two serious errors

in the form of contresens (mistranslation) and gave -

4 points for the sentence in (1). The first contresens

was keeping ST phrase “Yellowstone Plateau” in the

TT. Keeping the ST “Yellowstone Plateau” in the TT

belongs to mistranslation because in bahasa Indonesia

has the equivalent of “Plateau”, i.e. “Dataran

Tinggi”. Although keeping the ST word(s) is

acceptable in translation, it is not proper to be used in

(1) because it might make the TT readers unable to

understand what the “Yellowstone Plateau” is exactly

(whether it is a highland, lowland, rocky area, etc.).

Meanwhile, the second contresens was writing

“pertama sekali mendiami” as the equivalent of the

ST “have first claim”. The mistranslation occurred

when the translator decided to omit or substitute the

meaning of “claim” which, according to the raters,

played a very important role in this sentence.

Omitting the meaning of “claim” shows the failure in

providing the right equivalent. Nevertheless, Rater A

also gave +2 points for good solutions in translating

2 ST phrases. The good solutions found in the TT

were writing “tinggal di daerah itu dengan damai” as

the equivalent of the ST phrase “lived in the area in

peaceful tranquility” and “tanpa diganggu oleh

kehadiran orang kulit putih’ as the equivalent of the

ST phrase “undisturbed by the presence of white

men”. Based on these minus and plus points, the

assessment for the sentence in (1) is -2 points.

Meanwhile, following Model B, Rater A found

almost complete meaning transfer due to several

insignificant inaccuracies requiring revisions to reach

professional standard. Such inaccuracies are due to

the use of literal translation technique, incorrect

equivalents and borrowing. Literally translating the

ST word “first” into the TT phrase “pertama sekali”

shows inaccuracy as the whole meaning of the

sentence in (1) has nothing to do with sequence of

actions. Similarly, the TT word “mendiami” is the

incorrect equivalent of the ST word “claim” because

the ST term “first claim” is commonly equivalent

with the TT term “memiliki hak”. Moreover,

borrowing the ST word “Plateau” is inaccurate

because it has its established equivalent in bahasa

Indonesia “Dataran Tinggi”. Due to such

inaccuracies, Rater A gave a score 61, and based on

the degree of task completion, it was viewed as

“almost completely successful” translation.

Furthermore, following Model C, Rater A found

meaning distortion that bothers the complete transfer

of meaning from the ST to the TT. Based on the

accuracy instrument, as there are only three levels of

accuracy (accurate, less accurate and inaccurate),

Rater A classified the sentence in (1) as less accurate

translation. Based on the data from the questionnaire,

the quality of sentence in (1) could have been more

than “less accurate” because it is between “accurate”

and “less accurate”. This is one of the weaknesses of

scoring system dividing quality only into three levels.

The results of assessment performed by Rater A

highlighted in the previous paragraphs represent the

results of assessments from the other four raters that

also show different level of quality of the TT when

assessed by using different models of TQA. Based on

the results of the questionnaire, Rater B reports that

the difference in grading the quality of the TT is

caused by the absence of detailed assessment criteria.

He says that Waddington’s holistic model is very

good because it assess the whole quality of the TT;

ICOSTEERR 2018 - International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches

1758

however, its description for each level of quality is

not as detailed as the one mentioned in Hurtado’s

error analysis model. For example, the quality

description “Almost all the translation reads like a

piece originally written in English” does not provide

a clear description of which quality that resembles or

does not resemble the ST. In addition, the absence of

assessing the text and sentence function is also the

problem that makes different models of TQA assess

the same TT in different quality.

The big differences found in the assessment

results can be resolved when the three models provide

one more quality aspect to be evaluated, i.e. text or

sentence function. By having this aspect, which

certainly has its own role in the scoring system, such

big differences can be minimized because the scoring

of the other quality aspects also needs to be adjusted.

In addition to the additional aspect of translation

quality (i.e. text/sentence function), the three models

also need more comprehensive scoring system

distinguishing the quality level within the same aspect

of quality. For example, when the highest score for a

certain quality aspect is 30, the description of how

such highest score is obtained should be described,

and the description should be further continued to the

other lower scores.

Based on these findings, this article proposes a

model of TQA called “Function-based translation

quality assessment”. This model is also a holistic

model since it evaluates or assesses the whole

translation quality criteria in the TT. Suggested by its

name, this model includes text or sentence function as

one of the criteria of TQA. This supports the idea that

the quality of the TT should not be assessed only

based on the elements constructing the text, but it

should also evaluates whether the TT has achieved

the function intended by the ST (Sofyan and Tarigan,

2018). In addition to scoring system and criteria of

quality proposed by Hurtado (1995), Waddington

(2001) and Nababan et al. (2012), this model provides

more detailed quality description, as suggested by

Farahzad (1992) and Khanmohammad and Osanloo

(2009), together with its more detailed scoring system

(Sofyan, 2016). In general, this model divides the

quality criteria into five broad categories: (i)

accuracy, (ii) finding equivalents, (iii) translation

skill, (iv) text/sentence function, and (v) grammar and

TT style.

Following Nababan et al. (2012), accuracy is

given the highest percentage (30%) of the total score

(100) with the details described in Table 7.

Table 7: Description of accuracy.

Score

Range

Description

25-30 There are no identifiable problems of

comprehension to the TL readers. The original

message is completely transferred to the TT

without omissions or additions.

21-24 The problems of comprehension are only

related to the most highly specialized

vocabulary that do not affect the TL readers’

understanding. There are some partial

omissions and additions.

16-20 There is some difficulty experienced by the TL

readers in understanding the TT due to the

translator’s misunderstanding of some parts of

original message, apparent omissions and

additions.

11-15 The ideas are poorly expressed. The

translator’s serious problems in understanding

the ST disturb the transfer of the original

message. The TT is difficult to understand.

1-10 The translator’s serious problems in

understanding the ST terribly disturb the

transfer of the original message. The TL

readers are unable to understand what the

original writer intends to convey.

As accuracy determines 30 percent of the TT

quality, the highest score for accuracy is 30. Each

level of accuracy is given its own quality description

which clearly distinguishes itself from other higher or

lower level of accuracy. Besides, different quality

descriptions are given different score range. By using

such quality description, the quality criteria are more

comprehensible and the assessment on the accuracy

will be more representative.

Following Hurtado (1995), Waddington (2001)

and Khanmohammad and Osanloo (2009), finding

equivalent is considered as the second most important

criteria of assessing the quality of TT, and in this

model, it determines 25 percent of translation quality

whose detailed descriptions are presented in Table 8.

Table 8: Description of finding equivalents.

Score

Range

Description

20-25 All lexical and syntactic elements are

understood through precise vocabulary

usage. The words are chosen so skillfully that

the translation reads like a good publishable

text.

15-19 All lexical and syntactic elements are

generally understood through good usage of

a wide ran

e of vocabular

and structures.

Function-based Translation Quality Assessment

1759

Specialized vocabulary presents some

roblems with ina

riate e

uivalents.

10-14 Most of lexical and syntactic elements are

generally understood through a fair range of

vocabulary despite some observable gaps.

Some vocabulary is inappropriately used.

There is some evidence revealing translator’s

difficulties in finding equivalents,

perception, wordplay and other linguistic

features.

5-9 Comprehension of vocabulary and structures

shows quite observable gaps which blur

sense. There are problems in finding correct

vocabularies due to translator’s unability to

overcome s

ecialized vocabular

roblems.

1-4 The vocabulary is inapprpriately used.

Translator’s poor comprehension of the ST

seriously hinders the process of finding

equivalents. As a whole, translation makes

little sense.

The next quality is evaluated based on the

translation skill criteria described in Table 9.

Translation skill is evaluated based on the suggestions

from Hurtado (1995) who considers that good

translation skill should be given plus points to

complete his error analysis model of TQA. He divides

translation skill into two major skills: good solutions

and exceptionally good solutions to translation

problems. In addition, the description of translation

skill criteria presented in this model also adapts the

model proposed by Farahzad (1992),

Khanmohammad and Osanloo (2009) and Sofyan

(2016).

Table 9: Description of translation skill.

Score

Ran

Description

17-20 The translator demonstrates able and creative

solutions to translation problems and is skillful

in usin

resource materials.

13-16 The translator demonstrates consistent ability

in identifying and overcoming translation

problems. The TT contains only very few

minor errors. There are no obvious errors in

usin

resource materials.

9-12 The translator demonstrates a general ability

to identify and overcome translation problems.

There is a major translation error and/or an

accumulation of minor errors. The improper

use of reference materials is possibly reflected

in the TT.

5-8 The translator demonstrates some difficulty in

identifying and/or overcoming translation

problems. There are several major translation

errors and/or a large number of minor errors.

The improper use of reference materials is

reflected in the TT.

1-4 The TT reflects the translator’s inability to

identify and overcome common translation

problems. There are numerous major and

minor translation errors. The TT is seriously

flawed translation. The TT contains improper

use of materials and resources.

As described in Table 9, translation skill

determines 20 percent of the total score of the TT

quality; in other words, the highest score for

translation skill is 20. By using the score range

provided for each quality description, the TT

assessment in terms of translation skill is expected to

be more representative and reliable.

The next quality criteria developed in this model

are related to text/sentence function, i.e. to assess

whether the TT has met the text or sentence function

intended in the ST (see Table 10). This is the main

difference of this model from other previous models

of TQA. The inclusion of sentence function is to show

that this model, in addition to its nature as a holistic

model, also assesses the quality at the sentence level

(the function of the sentences that construct the TT).

At the text level, the assessment is done to evaluate

whether the TT is corresponding to the ST function

(informative, expressive or operative) (Reiss,

1977;1989). At the sentence level, the evaluation is

whether the sentence is used to inform, retell, report,

describe, persuade or discuss.

Table 10: Description of text/sentence function.

Score

Range

Description

13-15 Meeting the text/sentence function of the ST;

creative inventions and skillful solutions to

achieve the function of the ST; corresponding

to the text function based on the TL

erspective

10-12 Almost meeting the text/sentence function of

the ST; some inventions to achieve the

function of the ST; corresponding to the text

function based on the TL perspective

7-9 Inconsistency in meeting the text/sentence

function of the ST; awkward structure in

achieving the ST function; not fully

corresponding to the text function based on the

TL perspective

4-6 Less attention to the text/sentence function of

the ST; not corresponding to the text function

ased on the TL

ers

ective

1-3 In contrast to the text/sentence function of the

ST; not corresponding to the text function

ased on the TL

ers

ective

ICOSTEERR 2018 - International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches

1760

As described in Table 10, text/sentence function

determines 15 percent of the total score of the TT

quality; in other words, the highest score for

text/sentence function is 15. This means that if the TT

should reach the highest level of quality, it must meet

the text function of the ST corresponding to the TL

perspective of text function.

The last quality criteria described in this model are

related to the use of good grammar and TT style.

Following Waddington’s (2001) suggestions

considering language (grammar and style) error as a

minor error, the highest score for good language

performance in translation is 10, meaning that

grammar and style determine only 10 percent of the

total score. The detailed description of each level of

quality in grammar and style is presented in Table 11.

Table 11: Description of grammar and TT style.

Score

Ran

Description

9-10 The TT needs no improvement in terms of its

grammatical and stylistic points though one or

two natural failings are possibly observable.

The TT shows native-like fluency in gramma

7-8 The TT needs a little improvement in terms of

its grammatical and stylistic points. The TT

maintains advanced proficiency in grammar;

There are some grammatical problems, but

the

do not affect the messa

5-6 The TT tends to have awkward grammatical

usage due to literal rendering of the ST

message, but it does not significantly hinders

the message transfer. There are some attempts

to reflect stylistic features of the TL. Some

grammatical problems are apparent and have

ative effects on messa

e transfer.

3-4 The TT is awkward due to nonsensical

grammatical usages. The TT sounds unnatural.

There is a little attempt to reflect stylistic

features of the TL. There is evidence of clear

difficulties in following the TL style.

Grammatical review of some areas is

obviously required.

1-2 Little sense of the TL style. Knowledge of TL

grammar is inadequate. The use of TL

grammar is inadequate. Serious grammatical

roblems trul

affects the messa

e transfer.

As shown in Table 11, the highest score in

grammar and TT style (10) will be obtained if the

sentences in the TT are grammatically correct and

corresponds to the TT style. The other lower levels of

quality have limitations and are graded based on the

number of limitations contained in the TT. The

complete model of TQA developed in this article,

function-based translation quality assessment, can be

seen in the appendix.

4 CONCLUSIONS AND

SUGGESTIONS

The findings elaborated in this article show that using

different models of TQA in assessing the TT results

in different level of quality. The TT has a good quality

when assessed by Model A and Model C, but it has

lower level of quality when assessed by using Model

B. These findings need to be further explored to find

the factors leading to such differences as using any

model of TQA in assessing the TT should have

arrived at the same or nearly the same level of

translation quality. The results of analysis found that

the absence of the quality aspect in the form of

text/sentence function and the limited description of

quality level are the main causes leading to such

different assessment results. These findings,

therefore, provide a chance to develop a model that

accommodates the absence of the two main factors.

The function-based translation quality assessment

uses a holistic method of assessment in which the

whole criteria of translation quality are assessed. This

model assesses the quality of translation under five

aspects with different weight depending on how big

they contribute to a good quality of translation. The

five aspects are: accuracy (30%), finding equivalents

(25%), translation skill (20%), text/sentence function

(15%) and grammar and TT style (10%). Each of the

aspects is provided with detailed descriptions to make

the results of assessment more representative.

This model of TQA has not yet been tried out;

therefore, it is suggested for other researchers to

conduct a study on TQA by applying this model. The

results of the studies applying this model are expected

to give contributions in the improvement of this

model.

ACKNOWLEDGEMENTS

The authors’ gratitude is addressed to the Research

Institute of the University of Sumatera Utara for

funding this research under the TALENTA 2018

Research Grant.

Function-based Translation Quality Assessment

1761

REFERENCES

Bowker, L. 2001. Towards a methodology for a corpus-

based. Approach to translation evaluation. Meta, 46(2),

345–64.

Farahzad, F. 1992. Testing achievement in translation

classes. In Dollerup, C. & Loddergard, A. (Eds.),

Teaching translation and interpreting. Benjamins,

Amsterdam/ Philadelphia. pp. 271-278.

House, J. 1998. Quality of translation, In Baker, M. (Ed.),

Routledge Encyclopedia of Translation Studies.

Routledge, London.

Hurtado, A. A. 1995. La didáctica de la traducción.

Evolución y estado actual. In Fernández, P. (Ed.) X

Perspectivas de la Traducción. Universidad de

Valladolid, Valladolid. pp.49-74.

Khanmohammad, H. & Osanloo, M. 2009. Moving toward

objective scoring: A rubric for translation assessment.

JELS, 1(1), 131-53.

McAlester, G. 2000. The evaluation of translation into a

foreign language. In Schaeffner, C. & Adab, B. (Eds.),

Developing translation competence. Benjamins

Translation Library, 38. Benjamins, Amsterdam. pp.

229-42.

Melis, N. M. & Hurtado, A. A. 2001. Assessment in

translation studies: Research needs. Meta, 46(2), 272–

87.

Nababan, M., Nuraeni, A. & Sumardiono. 2012.

Pengembangan model penilaian kualitas terjemahan.

Kajian Linguistik dan Sastra, 24(1), 39-57.

Reiss, K. 1977/1989. Text-types, translation types and

translation assessment (Translated by A. Chesterman).

In Chesterman, A. (Ed.), Readings in translation

theory. Oy Finn Lectura Ab, Helsinki. pp. 105-15.

Schäffner, C. 1997. From ‘good’ to ‘functionally

appropriate’: Assessing translation quality. Current

Issues in Language and Society, 4(1), 1-5.

Sofyan, R. 2016. Translation process and translation

quality: A study of Indonesian student translators.

Doctoral Dissertation. Universitas Sumatera Utara,

Medan.

Sofyan, R. & Tarigan, B. 2018. Theme markedness in the

translation of student translators. Indonesian Journal of

Applied Linguistics, 8(1), 235-43.

Waddington, C. 2001. Different methods of evaluating

student translations: The question of validity. Meta,

46(2), 311–25.

Williams, M. 2001. The application of argumentation

theory to translation quality assessment. Meta, 46(2),

326-44.

Williams, M. 2004. Translation quality assessment: An

argumentation-centered approach

. University of

Ottawa Press, Ottawa.

Williams, M. 2009. Translation quality assessment. Mutatis

Mutandis. 2(1). 3-23.

APPENDIX

Table A1: The function-based translation quality

assessment.

Score

Ran

Description Rating

1 Accurac

(

30%

)

25-30

There are no identifiable

problems of comprehension to

the TL readers. The original

message is completely

transferred to the TT without

omissions or additions.

Score:

Note:

21-24

The problems of

comprehension are only

related to the most highly

specialized vocabulary that do

not affect the TL readers’

understanding. There are some

partial omissions and

additions.

16-20

There is some difficulty

experienced by the TL readers

in understanding the TT due to

the translator’s

misunderstanding of some

parts of original message,

apparent omissions and

additions.

11-15

The ideas are poorly

expressed. The translator’s

serious problems in

understanding the ST disturb

the transfer of the original

message. The TT is difficult to

understan

1-10

The translator’s serious

problems in understanding the

ST terribly disturb the transfer

of the original message. The

TL readers are unable to

understand what the original

writer intends to conve

2 Findin

uivalent

(

25%

)

20-25

All lexical and syntactic

elements are understood

through precise vocabulary

usage. The words are chosen

so skillfully that the translation

reads like a good publishable

text.

Score:

Note:

15-19

All lexical and syntactic

elements are generally

understood through good

usage of a wide range of

vocabulary and structures.

Specialized vocabulary

presents some problems with

ina

riate e

uivalents.

ICOSTEERR 2018 - International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches

1762

10-14

Most of lexical and syntactic

elements are generally

understood through a fair

range of vocabulary despite

some observable gaps. Some

vocabulary is inappropriately

used. There is some evidence

revealing translator’s

difficulties in finding

equivalents, perception,

wordplay and other linguistic

features.

5-9

Comprehension of vocabulary

and structures shows quite

observable gaps which blur

sense. There are problems in

finding correct vocabularies

due to translator’s unability to

overcome specialized

vocabulary

roblems.

1-4

The vocabulary is

inapprpriately used.

Translator’s poor

comprehension of the ST

seriously hinders the process

of finding equivalents. As a

whole, translation makes little

sense.

3 Translation Skill (20%)

17-20

The translator demonstrates

able and creative solutions to

translation problems and is

skillful in using resource

materials.

Score:

Note:

13-16

The translator demonstrates

consistent ability in identifying

and overcoming translation

problems. The TT contains

only very few minor errors.

There are no obvious errors in

usin

resource materials.

9-12

The translator demonstrates a

general ability to identify and

overcome translation

problems. There is a major

translation error and/or an

accumulation of minor errors.

The improper use of reference

materials is possibly reflected

in the TT.

5-8

The translator demonstrates

some difficulty in identifying

and/or overcoming translation

problems. There are several

major translation errors and/or

a large number of minor errors.

The improper use of reference

materials is reflected in the TT.

1-4

The TT reflects the translator’s

inabilit

to identif

and

overcome common translation

problems. There are numerous

major and minor translation

errors. The TT is seriously

flawed translation. The TT

contains improper use of

materials and resources.

4 Text/Sentence Function (15%)

13-15

Meeting the text/sentence

function of the ST; creative

inventions and skillful

solutions to achieve the

function of the ST;

corresponding to the text

function based on the TL

erspective

Score:

Note:

10-12

Almost meeting the

text/sentence function of the

ST; some inventions to

achieve the function of the ST;

corresponding to the text

function based on the TL

erspective

7-9

Inconsistency in meeting the

text/sentence function of the

ST; awkward structure in

achieving the ST function; not

fully corresponding to the text

function based on the TL

ers

ective

4-6

Less attention to the

text/sentence function of the

ST; not corresponding to the

text function based on the TL

ers

ective

1-3

In contrast to the text/sentence

function of the ST; not

corresponding to the text

function based on the TL

erspective

5 Grammar and TT style (10%)

9-10

The TT needs no improvement

in terms of its grammatical and

stylistic points though one or

two natural failings are

possibly observable. The TT

shows native-like fluency in

ramma

Score:

Note:

7-8

The TT needs a little

improvement in terms of its

grammatical and stylistic

points. The TT maintains

advanced proficiency in

grammar; There are some

grammatical problems, but

they do not affect the message.

5-6

The TT tends to have awkward

grammatical usage due to

literal rendering of the ST

message,

ut it does not

Function-based Translation Quality Assessment

1763

significantly hinders the

message transfer. There are

some attempts to reflect

stylistic features of the TL.

Some grammatical problems

are apparent and have negative

effects on message transfer.

3-4

The TT is awkward due to

nonsensical grammatical

usages. The TT sounds

unnatural. There is a little

attempt to reflect stylistic

features of the TL. There is

evidence of clear difficulties in

following the TL style.

Grammatical review of some

areas is obviousl

uired.

1-2

Little sense of the TL style.

Knowledge of TL grammar is

inadequate. The use of TL

grammar is inadequate.

Serious grammatical problems

truly affects the message

transfer.

ICOSTEERR 2018 - International Conference of Science, Technology, Engineering, Environmental and Ramiﬁcation Researches

1764