The quality of google translates vietnamese translations of the english personifications in la fontaines fables

This study is carried out i to identify and classify the personifications in La Fontaine's fables, ii to identify and classify the translation errors in GT's Vietnamese translations of t

THÁI THỊ THANH MINH

THE QUALITY OF GOOGLE TRANSLATE'S VIETNAMESE TRANSLATIONS OF THE ENGLISH PERSONIFICATIONS IN LA FONTAINE'S FABLES

Field: English Linguistics

Code: 8220201

Supervisor: Lê Nhân Thành, Ph.D

Binh Dinh, 2021

THÁI THỊ THANH MINH

CHẤT LƯỢNG DỊCH CỦA GOOGLE TRANSLATE KHI NHÂN HÓA TRONG CÁC TRUYỆN NGỤ NGÔN CỦA LA FONTAINE KHI DỊCH SANG TIẾNG VIỆT

Chuyên ngành: Ngôn ngữ Anh

Mã số: 8220201

Giáo viên hướng dẫn: TS Lê Nhân Thành

Bình Định, 2021

STATEMENT OF AUTHORSHIP

I, at this moment, declare this study is entirely my work carried out at Quy Nhơn University for the degree of Master of Arts in English Language and that it has not been submitted for a degree at any other Institution Where other sources of information are used, they are acknowledged

Binh Dinh, 2021

Thai Thi Thanh Minh

ACKNOWLEDGEMENTS

First and foremost, it is a pleasure to express my sincere gratitude to many people who made this thesis possible I wish to give my deepest gratitude and appreciation to my profound advisor, Ph.D Lê Nhân Thành, for his continuous support and encouragement Without his help, I would not complete this thesis Throughout my thesis-writing period, he encouraged me whenever I encountered problems with careful and inspiring advice, good teaching, and exciting ideas His extensive knowledge of Google Translate has given me valuable experience in conducting the study Especially in the data analysis, thanks to his helpful tutorials and recommendations, I had overcome the difficulties of the figure process and software In short, it is an honor for me to be instructed by Ph.D Lê Nhân Thành, who never accepted less than my best efforts

I would like to thank all professors at Quy Nhơn University, especially professors and staff in the Foreign Language Administration, who always do their best to make all thesis procedures as convenient to students as possible

Finally, my beloved family deserves a special mention for their unconditional love and support throughout the research study

To all of them, I dedicate this thesis

Bình Định, 2021

ABSTRACT

This thesis is conducted to investigate the quality of Google Translate's Vietnamese translations of the English personifications in La Fontaine's fables The data are the personifications in La Fontaine's fables and their Vietnamese equivalents in GT's translations The classification of personifications presented by Nguyễn Thế Truyền (2001) and the categorization of translation errors suggested by Farrús et al (2011) are used

as the conceptual frameworks for this study This study is carried out (i) to identify and classify the personifications in La Fontaine's fables, (ii) to identify and classify the translation errors in GT's Vietnamese translations of the English personifications, and (iii) to evaluate GT's translation quality in translating the English personifications into Vietnamese in terms of translation errors The results of the data analysis show that all three types of personifications are present in the fables The Activities and Characteristics personifications account for the most significant number The results also indicate that three out of five translation error types are found, and Semantic errors account for the highest rate of the errors committed by Google Translate Lexical errors rank second The third rank is Syntactic error Google Translate does not commit any Morphological and Orthographic errors The findings suggest that much care should be taken when using Google Translate to translate English personifications into Vietnamese Some suggestions and implications are proposed for EFL teachers and learners majoring in translation, human translators, and researchers interested in the related fields

Keywords: Google Translate, personifications, translation errors, Translation Quality Assessment, Vietnamese

TABLE OF CONTENTS

STATEMENT OF AUTHORSHIP i

ACKNOWLEDGEMENTS ii

ABSTRACT iii

TABLE OF CONTENTS iv

ABBREVIATIONS viii

LIST OF TABLES ix

LIST OF FIGURES ix

CHAPTER 1 INTRODUCTION ix

1.1 Rationale 1

1.2 Aim and Objectives 2

1.2.1 Aim 2

1.2.2 Objectives 2

1.3 Research Questions 2

1.4 Scope of the study 2

1.5 Significance of the study 3

1.6 Structure of the thesis 3

CHAPTER 2 LITERATURE REVIEW 6

2.1 Theoretical background 6

2.1.1 Personifications 6

2.1.2 Translation theory 10

2.1.3 Machine Translation 21

2.2 Previous studies 26

2.3 Jean de la Fontaine's life and works 29

CHAPTER 3 RESEARCH METHODOLOGY 36

3.1 Research Method 36

3.2 Data collection 37

3.3 Data analysis 38

CHAPTER 4 FINDINGS AND DISCUSSIONS 43

4.1 Types of personifications 43

4.1.1 Verbal personifications 43

4.1.2 Naming personifications 44

4.1.3 Activities and Characteristics personifications 44

4.2 Translation errors made by Google Translate 45

4.2.1 Morphological errors 46

4.2.2 Orthographic errors 46

4.2.3 Syntactic errors 47

4.2.4 Lexical errors 49

4.2.5 Semantic errors 54

4.2.6 No errors 57

CHAPTER 5 CONCLUSIONS, LIMITATIONS, AND IMPLICATIONS 62

5.1 Summary of the findings 62

5.2 Limitations of the study 63

5.3 Implications 64

5.3.1 Implications for EFL teachers and students of translation 64

5.3.2 Implications for translators 64

5.3.3 Implications for future researchers 65

REFERENCES 66 APPENDIX

LEX Lexical translation errors

MOR Morphological translation errors

SEM Semantic translation errors

SMT Statistical machine translation

SYN Syntactic translation errors

TQA Translation Quality Assessment

TT Target Text

VB Verbal personifications

LIST OF TABLES

Table 3.1 Sample data preparation 38

Table 3.2 Sample of translation errors analysis 39

Table 3.3 Analyzing and comparison types of errors 40

Table 4.1 The identified personifications 43

Table 4.2 Types of errors 46

Table 4.3 The number of the Syntactic errors 47

Table 4.4 The number of the Lexical errors 49

Table 4.5 The number of the Semantic errors 55

Table 4.6 The number of occurrences of no translation errors 57

Table 4.7 The statistics of translation errors committed by GT 59

LIST OF FIGURES

Figure 2.1 The dynamics of translation 11

Figure 2.2 The Translation process 12

Figure 2.3 Translation approach 14

Figure 2.4 Farrús et al.'s 2011 classification translation errors 21

Figure 2.5 The Vauquois Triangle 24

Figure 2.6 Timeline of Machine Translation Evolution 24

Figure 2.7 Languages supported by Goole Translate 26

Figure 4.1 Occurrences of translation errors according to personification types 60

of Pinocchio‖ published in 1883, is a children's novel by Italian, ―The brave lead soldier‖ a fairy tale by Danish author Hans Christian Andersen…Aesop‘s fables are published in Greece, and Jean de La Fontaine's fables are written in France, published in 1668 With the written form flexibly, plentifully, and creatively, the author brings purely, innocently, warmly, and an imaginary world for children and highly educational lessons

In Vietnam, the name La Fontaine hasn‘t been a stranger In the early years of the twentieth century, his name and works are familiar with famous translators like Nguyễn Văn Vĩnh, Huỳnh Lý, Nguyễn Đình, Đỗ Khắc Siêm,

Hà Khắc Nguyện and the others Fables express one's feelings such as happy, sad, angry, or disappointed, which are assigned to inanimated objects

to show the implicit feelings of the author, and use many figurative and hidden meanings These reasons cause problems in translating fables

Translating the personifications in fables by the human translators is a challenge Translating the personifications by GT's translation is more challenging It is just because GT must concentrate on seeking equivalence

meaning Therefore, the researcher has examined GT's in its Vietnamese translations of English personifications into Vietnamese

1.2 Aim and Objectives

1.2.1 Aim

This study aims to investigate GT's quality when translating the English

personifications in La Fontaine's fables into Vietnamese

1.2.2 Objectives

To achieve this aim, the researcher will try to:

- to identify and classify the personifications in La Fontaine's fables,

- to identify and classify the errors in GT's Vietnamese translations of the English personifications, and

- to evaluate GT's translation quality in translating the English personifications into Vietnamese in terms of translation errors

1.3 Research Questions

This study tries to answer the following questions to fulfill the aim and objectives mentioned above:

1 What are the English personifications in La Fontaine's fables?

2 What translation errors does GT make when it translates the English personifications into Vietnamese?

3 How well does GT translate the English personifications into Vietnamese in terms of translation errors?

1.4 Scope of the study

This study investigates the quality of GT's Vietnamese translations of

the English personifications in La Fontaine's fables 41 La Fontaine's fables are translated into Vietnamese by different famous translators Of all the fables collected by La Fontaine, the researcher chose 41 fables, with the 174 personifications identified using the theoretical framework of Đinh Trọng

Lạc (2001) and Nguyễn Thế Truyền (2001) For this reason, the English sentences which contain personifications are translated into Vietnamese by

GT Still, only those errors related to GT‘s Vietnamese translations of the English personifications are examined following the classification of

translation errors suggested by Farrús et al (2011)

1.5 Significance of the study

The study has theoretical and practical significance

Regarding the theoretical aspect, this research helps teachers and learners deeply understand the stylistic devices in teaching and learning literature Furthermore, this assists students in developing their knowledge of translation For translators, knowledge about personifications support them to produce better accurate translations For the researcher, this aids in identifying different kinds of personifications used in La Fontaine‘s fables It also informs GT‘s quality in translating English personifications into Vietnamese

by identifying the translation errors

Regarding the practical aspect, this thesis raises the awareness of people who use GT They should be careful of the errors that GT usually commits

1.6 Structure of the thesis

The thesis is designed with five main parts placed in five chapters:

Chapter I, Introduction, the study's rationale, aim and objectives, the

research questions, scope, significance, and the thesis structure

Chapter II, Literature Review, addresses the definitions and

characteristics of personifications It also provides a review of crucial concepts of translation, Machine Translation, Google Translate, La Fontaine‘s life and works, and the previous studies related to the subject of the thesis

Chapter III, Methodology Research, is concerned with a detailed

description of the procedures for data collection and data analysis in the present study.

Chapter IV, Findings and Discussion, presents the findings and

discussion of the study The first section presents the findings about personifications types The second offers the errors made by GT The degree

of translation accuracy will be analyzed in the final part based on the occurrences of the translation errors

Chapter V, Conclusions, Limitations, and Implications, summarizes the

study's main findings and addresses the implications for different stakeholders Simultaneously, this chapter also puts forwards the limitations

of the study

CHAPTER 2 LITERATURE REVIEW

This chapter addresses the definitions and characteristics of personifications It also provides a review of crucial concepts of translation, Machine Translation, Google Translate, La Fontaine‘s life and works, and the previous studies related to the subject of the thesis

2.1 Theoretical background

This part addresses theoretical background for the study, focusing on personifications and classifications, translation theory, translation errors, La Fontaine‘s Fables, and previous studies that help justify the research‘s aim and objectives

2.1.1 Personifications

In this part, the author demonstrates the theories and concepts relevant

to stylistic devices, the personifications and classifications, the comparison of personifications and metaphors, and the denotation of fables

2.1.1.1 Rhetorical measures/ stylistic devices

Stylistic devices (or figures of speech) play a significant role in analyzing literary text The term "figure of speech" is frequently used for stylistic devices that use a figurative meaning of the language elements and create a rich image It connects two things and sees the world in new ways In

"Stylistics," Galperin (1977, p 30) defined that "stylists device is a conscious

and intentional intensification of some typical structural and Semantic property of a language unit (neutral or expressive) promoted to a generalized status and thus becoming a generative model."

According to Lehtsalu et al (1973), personifications belong to the stylistic devices based on circumlocution Stylistic devices make speeches and essays more interesting, lively and they keep the reader's attention

2.1.1.2 Definition of personifications

Personifications are one of the fundamental stylistic devices in any language Animals, inanimate objects, or abstractions are represented as having human characteristics such as behavior, feelings, and character

Lakoff & Mark (1980), as cited Dorst (2011, p.114), has stated that personifications belong to metaphorical rhetoric, cross-mapped to an inanimate object or entity to indicate a person

In Viet Nam, Đinh Trọng Lạc (1994, p.63) in "99 phương tiện và biện

pháp tu từ tiếng Việt" (99 means and the Vietnamese language's Rhetorical

measures); Cù Đình Tú (1994, pp 289-293) in "Phong cách học và đặc điểm

tu từ Tiếng Việt" (Stylistics in studying and Rhetorical Features Vietnamese

language) have defined personifications as rhetorical devices which are used

to assign human beings' characteristics, features, qualities, and actions to animals and objects

In "Polish and personifications in classical Athenian Art,"

Mylonopoulos (2013, p.13) has pointed out that personifications refer to something as a human being while it is not A similar viewpoint with

Mylonopoulos, in the book "The poetic of personifications," Paxson (2009,

p.1) has stated that personifications are the visible path through which a human characteristic is assigned to as a non-human being Especially, Lakoff

& Johnson (2008, p.33) has considered his thoughts about personifications:

"non-human is human."

Personifications are employed widely in La Fontaine's fables, so it is essential to understand personifications For example, when we say, "The Fox invited neighbour Stork to dinner," we are giving the Fox the ability "to invite

- to dinner," which is a human activity Thus, we can say that "invited - to dinner" has been personified in the given sentence

In conclusion, there are many definitions of personifications Different

writers have a similar viewpoint about personifications

2.1.1.3 Personification classification

In Vietnam, Nguyễn Thế Truyền (2013, p.153) in "Phong cách học

Tiếng Việt hiện đại" (Stylistics in studying Modern Vietnamese Language)

the classification and functions of personifications have been discussed The

researcher has classified personifications into three kinds:

a Verbal personifications

In this first type, things and animals talk and speak while

communicating with each other as if they were human beings For example:

In English: "The Oak said one day to a river Reed "

(The Oak and the Reed)

In Vietnamese: " Một ngày nọ, Sồi nói với một cây sậy trên sông,"

(Cây Sồi và cây Sậy)

b Naming personifications

In this second type, animals and things are given names similar to those

given to people For example:

In English: " But Mr Rat runs up; a mesh or two "

(The Weasle in the Granary)

In Vietnamese: " Nhưng anh Chuột chạy lên, một hoặc hai lưới"

(Chồn lột vào Kho)

c Activities and Characteristics personifications

In this third type, the author referred to an inanimate object or animal

as an action or state usually connected with human activity and

characteristics For example,

In English: "The Fox trotted up, very servile and sly "

In Vietnamese: " Con Cáo chạy lon ton, rất ngoan cố và ranh mãnh;"

(Triều đình Vua Sử tử) The above classifications help the researcher identify and categorize the personifications for assessing the GT's translation quality

2.1.1.4 The personifications and metaphors

In "The powers of personifications: Rhetorical purpose in the Book of

wisdom and the letter to the Romans," Dodson (2008, p 34) has declared that

a metaphor is beautiful trope that compares two different things in a figurative sense The same viewpoint with Dodson, Newmark (1998, p.104) has said that metaphors transfer the abstraction of personifications, applying a word or

a collocation to what it does not denote

Personifications can be regarded as a kind of metaphor It is an artistic device conveying figurative meaning to symbolize the image of reality If personifications are associated with moral tendencies, it becomes allegory That is why Dodson (2008) has advised learners to concentrate on the subject

of the sentence related to an object, concepts, or impersonal beings and a verb-link to human action to avoid arguable personifications Personifications are used a lot in La Fontaine‘s fables So, personifications in this study work independently of metaphors in this thesis

2.1.1.5 Fables (allegory)

Jon Whitman, cited in J.R Dodson (2008, p 34), has said that allegory

is to say one thing, but it means another

Samuel Taylor Coleridge, cited in J R Dodson et al (2008, p.35), provides a more detailed explanation of allegory:

“Allegorical writing as the employment of one set of agents and images with actions and accompaniments correspondent, so as to convey, while in disguise, either moral qualities or conceptions of the mind that are not in themselves objects of the senses, or other images, agents, actions, fortunes,

and circumstances, so that the difference is everywhere presented to the eye

or imagination while the likeness is suggested to the mind; and this connectedly so that the parts combine to form a consistent whole.”

2.1.2 Translation theory

This section addresses the theory about translation consisting of the definition of translation, translation process, types of translation, machine translation, Google Translate, translation errors, and translation quality assessment

2.1.2.1 Definition of translation

From the perspective of Reiß in Lauscher (2000, p 151), translating has created a TT while constantly referring back to the ST They mean finding equivalents for ST items in the target language and individual text unit levels

A translation is thought good if it achieves the finest equivalence, or

"considering the linguistic and situational context, the linguistic and stylistic

level, and the author's intention, ST and TT units have the same 'value' as the text unit in the SL." Lauscher (2000)

According to Dalvai (2012, pp 378-379), translation creates a new text with equivalent meaning to its ST

With the same viewpoint, in "Seleskovitch and Lederer's Interpretive

Theory of Translation," Nida and Taber (1969), cited in Dalvai (2012, p

379), has defined that the translation is a process of finding the closest natural equivalence of the source language, in terms of both meaning and style A good translator must comprehend the importance of a text, formulating the resources in another language but satisfying the recipient's needs and the translation purposes (Hurtado Albir, 2011)

Larson (1997) has emphasized that translation includes studying the

vocabulary, grammatical structures, the communicative contexts, and the ST's cultural context to discover the meaning; re-expressing the purpose in the

receptor language reproduces the equivalent meaning

Similarly, in "The Handbook of Translation Studies," Dalvai (2012) has

highlighted that translation is creating a text that is equivalent to its ST

In summary, translation means converting the source text (ST) into

another language with the criteria of producing a similar response To have

the best translation product, Cape (1968, p.54) in Newmark (1981, p.38) has

stressed that that product should have eight criteria: "give the words of the

original, give the ideas of the original, read as a unique work, reflect the style

of the original, possess the style of the translation, read as the contemporary

of the translation, add (or never add) from the original A translation of verse

should be in prose or inverse"

Larson (1997, pp 519-527) has also stressed that translation is a

complicated process To obtain qualified translation products, a translator has

understood how to analyze the knowledge of ST and TT, the translation

methods, and handle texts, sentences, and other units, to discuss the relationship

between meaning, language, culture (or the socio-cultural factors)

9 The truth (the fact of the matter)

TEXT

4 SL setting 8 SL setting and

and tradition tradition

10 Translator

Figure 2.1 The dynamics of translation (Newmark, 1998, p.4)

2.1.2.2 Translation Process

Newmark (1981, p.144) has classified three basic translation processes: analyzing the ST, corresponding Syntactic structures, reformulating the author's thoughts, the readers' expectations, and the appropriate norm of the

TL However, these processes are too general, hard to apply

Source Language Receptor Language

Text to be translated Translation

Meaning

Figure 2.2 The Translation process (Larson, 1997, p.4)

Larson (1997, pp 519-527) has given eights detailed translation process steps, and these are appropriate in this study

1 Preparation: Before beginning the assignment, the translator should have received linguistics and translation principles training Then, preparation activities such as reading the whole book, reviewing background material, and studying linguistic concerns are followed to become familiar with the text

2 Analysis: the translator writes out the keywords to find a good equivalence in the receptor language The translator should pay close attention to reading the text's start and end

3 Transfer: It is a method of assessing the semantic structure of a translation's first draft In this step, the translator tries to figure out the receptor language's lexical equivalence and culture

4 Initial Draft: The translator had set aside the manuscript for some time to gather more reading material to find the best equivalent and then returned to revise it

5 Reworking the initial draft: With an emphasis on the meaning of language, the translator examines for correctness and naturalness After the translators receive a second draft, they may repeatedly work on it until they complete a final draft

6 Testing the translation: the translator aims to test the accuracy and natural translation product The translator has responsibility for himself Besides, the group of consultants, testers, and reviewers needed to assess the outcome

7 Polishing the translation: rechecking and polishing the translation to ensure the TT's naturalness and acceptability

8 Preparing the manuscript for the publisher: The translation process comes to a close at this stage The book is ready to be prepared for the publisher after the final draft is done

2.1.2.3 Translation methods

To gain acceptable products, the translators master translation theory Hartono (2015, p 734) has defined that the translation method affects the outcome of text translation This part introduces several translation methods

In "On Linguistic Aspects of Translation" by Roman Jakobson, Cruz (2013, p.2) has divided three models of translation: (1) Intralingual

translation (or rewording) is an interpretation of Verbal signs employing

other signs in the same language; (2) interlingual translation (or translation

proper) is an interpretation of Verbal signs employing some other language;

(3) intersemiotic translation (or transmutation) is an interpretation of Verbal

signs employing signs of nonVerbal sign systems that usually occur when the

speaker and the hearer do not know the same language The (2) interlingual translation (or translation proper) model describes the transfer from ST to TT

Newmark (1998, pp 45-47) has suggested two approaches to translation: "Semantic" and "communicative" approaches Here is Newmark's

V diagram:

SL emphasis TL emphasis

Word for word translation Adaption Translation

Literal Translation Free Translation

Faithful Translation Idiomatic Translation

Semantic Translation Communicative Translation

Figure 2.3 Translation approach (Newmark, 1998, p.4)

He has stated that the translation method relates to the text used for

sentences and smaller language units The translation methods that emphasize

the ST include Word-for-Word translation, Literal translation, Faithful translation, and Semantic translation The other method emphasizes the TT, including Adaptation, Free, Idiomatic, and Communicative

Word-for-word translation: The words are translated one-by-one

single word, using their most common meanings, out of the text

Literal translation: The translator converts the SL grammatical

constructions to their nearest TL equivalents, but the Lexical words are still translated singly, out of context

In Faithful translation: cultural words are transferred, but the degree

of grammatical and Lexical "abnormality" in the translation can still be seen The translator is faithful to the writer's intentions and the text's realization

Semantic translation: is more flexible, admitted 100% faithful, and

allows the translator's understanding of the original

Adaptation: is the "freest" form of translation It is used chiefly for

poetry, dramas The SL culture is converted to the TL culture, and the text is rewritten

Free translation: reproduces the content rather than the form of the ST

A free translation is often longer than the original

Idiomatic translation: reproduces the original's meaning but tends to

use popular expressions and idioms which do not exist in the original

Communicative translation: renders the exact contextual meaning of

the ST to fit the readership of the TT

The communicative approach creates in its the same effect that the original readers may feel On the other hand, Semantic translation observes both the Semantic and Syntactic constructions of the second language Newmark (1988) has concluded that the communicative approach is more economical and effective than the Semantic one because it may create in its readers the same effect that the original readers may feel

However, in "Meaning-based Translation," Larson (1997, p.17) has

disagreed with Newmark's viewpoint She divided translation into two models: form-based translation attempts to follow the form of the SL (or Literal translation), and meaning-based translation makes every effort to communicate the meaning of the ST in the natural conditions of the receptor language She has declared that idiomatic translation is the best here

Translation classification breaks down objective criteria for evaluating

a translation product A translation model has to fit the type of document being translated If it is a scientific text with an information function, the translator should choose the Literal translation If literature text with a linguistic function, the author should choose the typical approach of Larson

In conclusion, Larson (1984), with two translation models, has declared

that idiomatic is better than literal At the same time, Newmark (1988) has suggested eight types of translation methods

Regarding function-based equivalence, Nida (1964) has distinguished

two kinds of equality Formal equivalence enables target readers to slip into the inside of source readers Dynamic equivalence adapts ST according to the

target culture

On the contrary, Kade (1968) has offered four equivalence types Total: two terms in different languages coincide entirely Facultative: one term corresponds to many in another language Approximative: one term only finds partial correspondence in a term belonging to another language Null or nil

equivalence: culture-specific terms that have no correspondence in another

language

According to Watts et al (2001), Koller (1979) has suggested five

equivalence kinds using meaning-based equivalence Denotative equivalence refers to the extralinguistic reality Connotative equivalence is regarded as a type of Verbalization Text-normative equivalence is concerned with the textual and language norms in question Pragmatic equivalence refers to the target receiver Finally, formal equivalence regards good words and figures

of speech Non-equivalence is a translation error or means that the target

language has no direct equivalent for a term in the ST

Different authors classify equivalents into specific kinds based on different approaches For example, Nida (1964) has used function-based

equivalence with two types Kade (1968) has classified four types of equal, and Koller (1979) has illustrated meaning-based equivalence and introduced five equivalence kinds Koller has mentioned Non-equivalence

2.1.2.5 Translation Quality Assessment (TQA)

TQA has been used popularly in translation evaluation As a result, several efforts have been made to discover proper ways to evaluate translation

products

In "A translation robot for each translator," Joke (2016, pp 20-22) has shared two methods: the ranking method means presenting reviewers,

consultants, all translation products of an ST and ranked according to quality

The scoring method has been used for the quality assessment of MT

output It consists of quality scales for fluency (target language) and adequacy (the ST's meaning) (Fiederer & Sharon, 2009) The most common scale is a five-point scale that has been used by the Advanced Research Project Agency (White et al 1994) The researcher has used the scoring method for analyzing the test (Garcia, 2010) "Accuracy," "quality of language," and "application of good practices" are the most crucial criteria;

However, these two methods have their limitations Most scoring and ranking methods just have been conducted at the sentence level The two methods can not understand a whole text Additionally, they cannot distinguish how different translation products are Suppose the testers have given this translated product three marks and another one receives four marks

In that case, we cannot determine exactly which product's quality is better than the rest Finally, human annotators can use score scales in different ways (Wisniewski et al., 2013) If a sentence contains one serious error, it does not show a "horrible translation." This surveyor's prejudice is the main criticism

of TQA approaches (Lauscher, 2000b) In summary, ranking or scoring

methods do not tell whether the first translation's quality is worse than that of the second

In another research, Elvis et al (2019, p 21) has used TQA to identify translation problems and evaluate MT systems' output and translators' work

J & Scobell (2012, pp 382-536) argued that the critical component in assessing the quality of a translation product is the theory of translation Different assessments make different thoughts of translation quality and of ensuring quality in translation products He has suggested the following three

criteria for TQA Accuracy: a client is confident that the translation is accurate Suitability is manifested in using appropriate cultural concepts,

meeting the client's requirements, and following the social norms concerning

texts Usability: the translation is clear, straightforward, and is easy for

(2011) as the framework of this study because of its specificity and

particularity

a Definition of translation errors:

In the Handbook of Translation Studies of Dalvai (2013), an error

means something has gone wrong, misunderstandings and involvements

happen during the ST's transferring of meaning and form process to the TT

Neubert and Shreve (1995), cited in Phạm Thị Kim Cúc (2017), have described non-equivalence as the standpoint for translation errors that can be

measured as the non-adequacy of the TT

According to House (1997), a translation error occurred if there was a mismatch between a TT and the ST

b Classification of translation errors:

The classification of translation errors is essential in error analysis House

(1997) has divided translation errors into two kinds of errors A covert error is a mismatch such as a genre, field, tenor, model An overt error is a grammatical

error, wrong word errors, and inappropriateness errors This model has just been used for the aim of description and explanation However, these models are so complicated to apply to identifying translation errors

Vilar et al (2006, pp 697-699) have confirmed five types of errors

that fit MT Missing words: Some words in the generated sentence are missing Word Order: A correct sentence can be generated by moving individual words Incorrect words interrupt the sentence's meaning Another kind of error occurs when an Unknow Word is used Furthermore,

Punctuation Errors represent only minor disturbances for languages without

fixed punctuation

Weiss and Ahrenberg (2012) have started six errors are missing words,

word order, incorrect words, and unknown words;

The other types include Not translated error: expression from the ST

appears in the translation instead of a proper target language counterpart, and

Incorrect form: a word of the translation has an upright stem but wrong

inflection

Based on the functional and purposeful viewpoint, Nord (2017) classifies

translation errors into four categories Pragmatical errors occur when there is a lack of attention to readers of the TT Second, cultural errors occur when cultural conventions are reconstructed Third, language errors appear when

much emphasis is put on linguistic structures rather than communicative or

functional appropriateness Finally, Text-specific translation errors are related to

the TT in that they are unsuitable for TT readers

Popescu (2013, p 244) has divided translation errors into three

categories First, language errors include grammatical and syntactic errors

Second, reading errors appear because of misinterpretation of the syntax

Finally, translation errors include misrepresenting the meaning or content of

the original text, errors in incorrect expression of the intention of words

According to Farrús et al (2011), there are five types of errors First, Orthographic errors consist of three sub-categories: punctuation errors, capitalization errors, and spelling errors Second, Morphological errors are related to features of verbs, nouns, others Third, Lexical errors are the consequences of unnecessary adding and missing words Fourth, Semantic errors happen when selecting the wrong words Finally, Syntactic errors are divided into five sub-categories: conjunction, preposition, article, Syntactic element reordering, and category errors Farrús et al.'s (2011) model was chosen as a framework for this study because this translation error classification has been explained in detail and is easy to use for data analysis

Figure 2.4 presents the classification of translation errors The definition of each error type is provided in the following sections

Figure 4 Translation errors 1

Figure 2.4 Farrús et al.'s 2011 classification translation errors

In conclusion, the author has addressed translation, translation methods, translation equivalence, translation quality assessment, and translation errors

in this chapter This knowledge is essential for understanding and analyzing the data in this present study

2.1.3 Machine Translation

In this part, the researcher presents the history of machine translation and Google Translate This discussion is essential to get a comprehensive understanding of the establishment and development of MT, MT's software, and its approach, especially of GT

2.1.3.1 Definition and MT software's history

Gustiar and Basari (2014, p.3) have defined MT as directly translating a

text from SL to the TL with no or minimum help from human users

Hutchins (2001), Afshin and Alaeddini (2016, pp 41-42), and Koehn

et al (2020, pp 11-35) have defined that MT has been describing from the 1940s to the present day However, researchers have focused on developing high-quality translation systems requiring human assistance and strategies for translating electronic mails, Web pages, and other Internet documentations

Chéragui in Hutchins (2001) has summarised the history of MT's

software as follows:

First Period (from 1948 to 1960)

In 1949, Warren Weaver offered his first opinion on the application of

published the first journal article, namely "Mechanical translation devoted to

translating languages by the aid of machines

Second Period (from 1960 to 1966)

In the early 1960s, numerous grammar and dependency grammar were

submitted for researchers to develop MT

In 1961, David G Hays in Los Angeles invented Computational grammar

In 1964, the first ALPAC (Automatic Language Processing Advisory Committee) was established In 1966, ALPAC declared no advantage in using

MT systems Funding should go into basic linguistic research and the development of methods to improve the human translation

Third Period (from 1966 to 1980)

In 1970, Russian researchers commenced the project REVERSO, and Peter Toma improved SYSTRAN1 (Russian-English), which was developed into GT later

From 1976 to 1978, FUJITSU company developed ATLAS2 for translating Korean into Japanese and vice versa

Fourth Period (from 1980 to 1990)

In 1982, SHARP company vended its translator DUET (English -

Japanese);

In 1983, the NEC company advanced the PIVOT translation system

In 1986, PENSEE by OKI3 (Japanese-English translator) was promoted And Hitachi improved his translation system based on an approach obtained by transfer (Japanese-English)

Fifth Period (since 1990)

During the 1990s, the widespread desktop processor systems led to computer-aided translation systems for human computer-aided translation translators by companies such as Trados

In 1993, C-STAR (Consortium for Speech Translation Advanced Research) was developed

In 1998, REVERSO was developed by Softissimo In 2000, ALPH (Japanese-English and Chinese-English) was advanced

Besides, Offline et al (2020, p.23) has shown the Vauquois Triangle in MT's operation The linguistic vision analyzes the meaning of a source sentence into a language-independent sense and then generates the target sentence—the illustration of Figure 2.5

Interlingual

Analysis Semantic Transfer Generation

Syntactic Transfer Lexical Transfer

Figure 2.5 The Vauquois Triangle (Offline et al.,2020, p.23)

M S Maučec and G Donaj (2013, p.4) have divided the development

of MT's approaches into four significant for MT's operation

Figure 6: Timeline of MT Evolution 1

Figure 2.6 Timeline of Machine Translation Evolution

Rule-based This approach uses grammar and language rules such as

rules for Syntactic analysis, Lexical laws, regulations for linguistic transfer, rules for Syntactic generation, rules for Morphology, etc

In example-based MT, a sentence similar to the input sentence is found and makes suitable changes to its stored translation It belongs to the

corpus-based approach because examples are taken out from extensive

collections of bilingual corpora

Statistical machine translation (SMT) does not depend on linguistic

rules and words; it learns how to translate by analyzing many existing human translations All major translation service providers have used SMT IBM and Microsoft develop commercial SMT systems

Hybrid machine translation, guided by rule-based MT, uses statistical

MT to recognize the set of suitable translation applicants Hybrid systems, determined by statistical MT, use directions at pre-/post-processing steps

Neural machine translation is a new translation method where a

computer uses deep learning to form an artificial neural network to demonstrate how to translate between languages It uses these neural networks to translate entire sentences without breaking them down into smaller parts

In conclusion, this section discusses MT history at diverse stages, different MT approaches, and MT operations developed over time Currently, the most promising approach is a neural machine translation Although effective, it also faces some issues, such as a more extensive vocabulary Researchers have been continuing to solve these problems and make MT a better service accessible to everyone

2.1.3.2 Definition and GT software's history

GT is a free online application offered by Google Inc It allows users to translate a section of a text or a web page into another language

Zafitri and Harida (2017, p 82) have said that Google Translate was first developed in 2007 using a system called SYSTRAN From February

2009 to August 2009, Turkish, Thai, Hungarian, Estonian, Albanian, Maltese, Galician, Africa, Belarusian, Icelandic, Irish, Macedonian, Malay, Swahili, Welsh, and Yiddish were added From 2008 until 4.2021, GT has translated over 109 languages worldwide

Figure 7 GT upports languages 1

Figure 2.7 Languages supported by Goole Translate

Figure 9 GT graphical user interface 1

GT has lots of practical functions With the hearing and talking function, users can use GT for learning the pronunciation of a word Additionally, GT is developed as an online language translator Moreover, GT can also be helpful as a lexicon or reference to the meaning of a word In conclusion, with artificial intelligence assistance, and NMT technology, GT is gradually becoming more comprehensive than ever

GT is growing to be a complex MT There are many new GT features However, translation results from GT need to be further studied, especially in the linguistic field

2.2 Previous studies

There have been some studies carried out employing different theories for assessing the quality of GT

Huỳnh Hà Mi (2020) carried out a study entitled with the title: "The

quality of GT when translating English metaphors into Vietnamese." Based on

Lakoff and Johnson's (1980) and Nord's (1997) theory, the result shows GT committed pragmatic errors at the highest rate, followed by linguistic and text-specific errors

Trương Tiểu Mi (2020) examined the translation of English relative clauses in literary and technical texts into Vietnamese by Google Translate Murray et al (1995) framework was used to examine the errors in GT's translations of the English relative clauses in these texts The data included English relative clauses collected from "Rip Van Winkle" and "Medical Astrology." The findings show that GT's translations of the relative clauses in the literature are not as good as those in the technical text This study raises awareness of using GT to translate texts of different kinds GT users need to know what type of texts are translated well by GT There are some limitations

to this thesis Only two genres (literature and medical texts) were used in this study Therefore, it does not reflect the entire quality of GT's translation

Another study was carried out by Võ Mỹ Thư (2019), who used Nord's

theory (1997) in her study entitled "The quality of Google Translate's

Vietnamese translations of English idioms with words denoting time." The

researcher collects 128 English sentences containing idioms with words denoting time from "Oxford Dictionary of Idioms" (Siefring, 2004) and four online dictionaries The findings point out that GT cannot accurately translate pure and semi idioms with symbolic meanings GT gains better quality in translating literal idioms GT's highest rate of errors is founding in translating English pure idioms into Vietnamese translations (making up 75%) With the limitation of the data, the errors may not be extensive enough to provide the most accurate and comprehensive judgment of the GT's quality in translating

English-Vietnamese idioms

Another study conducted by Trần Thị Ngọc Giàu (2019) evaluated the quality of English-Vietnamese translations Google Translate This theory was based on the theory of Cabeceran et al (2010) The result shows that GT's translations of the technical texts are better than those of the literary texts Although GT is not an excellent machine translator, GT's translation quality is quite suitable for those who do not speak a specific language to understand the meaning of a website The researcher also indicated two limitations Firstly, the author did not investigate the influence of translation errors on TL readers Secondly, the data is not big enough to offer a detailed overview of the quality of the English-Vietnamese translations carried out by GT

Halimah and Suryakancana's (2018) study is entitled "Error analysis In

English-Indonesian Machine Translate." The goals of this study are (i) to

categorize English-Indonesian MT errors into three types: Semantic errors, syntax errors, and Morphological errors, and (ii) to describe the dominant kind of translation errors produced by GT The result indicates that MT still has many shortcomings and does not produce accurate translations

Another study administered by Jimmy (2015) evaluated "the

Translation Quality of English-Indonesian by using GT." This research aims

to describe the translation techniques and the quality assessment that covers correctness, satisfactoriness, and readability of the sentences of scientific articles by GT The analysis shows that GT could not determine suitable techniques to produce a quality translation in translating sentences found in scientific papers The translation by GT is highly less correct, less acceptable, and less readable in the target language

A thesis on the translation of stylistic devices performed by Đào Thị

Xuân Kiều (2018) investigated "Aesop's Fables by Laura Gibbs and their