EVALUATING THE QUALITY OF ENGLISH VIETNAMESE TRANSLATIONS CARRIED OUT BY GOOGLE TRANSLATE

This study aims to evaluate the quality of GT’s Vietnamese translations of English literary and technical texts to see which of these two genres Google Translate translates better.. The

QUY NHON UNIVERSITY

TRƯỜNG ĐẠI HỌC QUY NHƠN

TRẦN THỊ NGỌC GIÀU

ĐÁNH GIÁ CHẤT LƯỢNG CÁC BẢN DỊCH ANH-VIỆT

DO GOOGLE TRANSLATE THỰC HIỆN

Chuyên ngành: Ngôn Ngữ Anh

Mã số: 8.22.02.01

Người hướng dẫn: TS Lê Nhân Thành

I would like to thank all those who helped me complete this thesis First of all,

I would like to thank my supervisor, Dr Le Nhan Thanh, who has given guidance, patience, time, support, and help essential for the completion of this thesis Without his support, this thesis would never exist.

Besides, I would like to express my thankfulness to all the teachers of the Department of Foreign Languages, Quy Nhon University, for their useful lessons, guidance, experience, wholehearted instruction, enthusiasm, and sympathy to make

me a more knowledgeable person today.

This thesis would have never been fulfilled without the assistance and support from my classmates of the 19th MA Course on the English Language at Quy Nhon University.

Last but not least, I would like to send my warmest thanks to my family and friends, whose love and encouragement were a source of inspiration and motivation for me to complete this thesis successfully.

Google Translate may produce translations with different levels of quality depending on the kinds of text it translates This study aims to evaluate the quality of GT’s Vietnamese translations of English literary and technical texts to see which of these two genres Google Translate translates better The study tries to answer the research question (1) How does Google Translate translate literary texts and technical texts in terms of translation quality? and (2) Which of these two genres does Google Translate translate better? Based on Haijar Khanmohammd and Maryam Osanloo’s (2009) conceptual framework of five common types of translation errors including orthographic errors, morphological errors, lexical errors, semantic errors, and syntactic errors, the researcher analyzed Google Translate’s Vietnamese translations of three literary texts with 5,773 words and six technical texts with 4,884 words in total The findings indicate that GT is better with the technical texts than the literary texts It is understandable as literary texts are more closely related to culture than technical texts Although the study achieved its aim, there were some limitations For instance, there was no investigation on the influence of translation errors on TL readers, and the data for the study was not big enough for more reliable generalizations Some implications for teachers and students of translation, translators and other researchers are addressed

in the thesis.

CHAPTER 1 INTRODUCTION

1.1 Rationale

The increase in global business has provided a greater and more urgent need for quicktranslation An obvious tool which has been created to help meet this need is Google Translate,which is very well-known for free online machine translation Google Translate (GT) has become

an instrument in machine translation that has been claimed by its provider to be developing at agreat pace to achieve ever higher degrees of accuracy Because GT is freely available on theInternet and has its own app on computers, tablets, and smartphones, it is accessible anywhere theInternet and Google services are available, and it easily enables users to render a text in onelanguage into over 100 other languages with outcomes of varying quality and comprehensibility

Even though GT is a very popular tool used by millions of Internet users and it hasimproved a lot so far, it is clearly not perfect yet Therefore, we cannot rely on it blindly withoutknowing how accurate its translations are So this study is very necessary for users to know forwhich genre they can easily get good results and which genre they would get less accurateresults That would help users to decide how much trust they can put on GT when they need totranslate texts that belong to different genres such as tourism news, medical information, and theshort story

The first related study that has been carried out recently by Aiken and Balan (2011) is An analysis of Google Translate accuracy The aim of the research was to compare the quality of

GT’s translations and provide insights into the errors made by GT Aiken, Ghosh, Wee, and

Vanjani (2009) have studied An evaluation of the accuracy of online translation systems in

communications of the IIMA This study gave an estimate of how good a potential translationmight be achieved by GT The result of this study showed that translations between Europeanlanguages were usually good, while those involving Asian languages were often relatively poor.Further, the vast majority of language combinations probably provided sufficient accuracy forreading comprehension in college Several studies on the use of GT in translation have focused

on computational analysis For example, Carl (2019) investigated Machine translation errors and the translation process: a study across different languages in The Journal of Specialised

Translation The findings clearly showed that expressions with MT accuracy issues into onelanguage (English-to-Spanish) were likely to be difficult to translate also into another languagefor humans and for computers; whereas this did not hold for MT fluency errors Also, Ghasemi &

Hashemian (2016) conducted a study investigating A comparative study of Google Translate translations: An error analysis of English-to-Persian and Persian-to-English translations Its

main findings were of no significant differences between the qualities of GT from English toPersian and Persian to English Moreover, there have been a lot of master theses carried out based

in the light of machine translation Most of them have focused on the comparison of translationsdone by machines or translations done by machines and those done by humans in differentlanguages There has not been any study of the accuracy that GT could achieve in translatingliterary and technical texts This study hopes to fill in this research gap

1.2 Aim and Objectives of the Study

1.2.1 Aim

This study aimed to investigate the quality of Google Translate’s translations of literarytexts and technical texts from English into Vietnamese

1.2.2 Objectives

To fulfil the above aim, the researcher:

- Evaluated the quality of Google Translate’s Vietnamese translations of English literary texts and technical texts

- Compared the quality of the translations of literary texts and that of technical texts in order to see which of these two genres Google Translate translates better

1.3 Research Questions

In order to achieve the above aim and objectives, the researcher collected data and analyzed

it in order to answer the following questions:

1. How does Google Translate translate literary texts and technical texts in terms of translation quality?

2. Which of these two genres does Google Translate translate better?

1.4 Scope of the Study

This study examined the quality of the translations carried out by Google Translate in twogenres, including literary texts and technical texts Other genres were not examined in this study.Only the quality of English – Vietnamese translations of texts of these two genres would beinvestigated For each genre, the author selected 3 texts of about 4000 words in total and thesetexts were translated by GT

1.5 Significance of the Study

The study has both practical and theoretical significance For theoretical aspects, theresearch offers insights into the quality of GT’s translations of literary texts and technical texts.Moreover, this thesis also provides knowledge about errors made by GT in its translations ofliterary texts and technical texts

For practical aspects, the study helps students, teachers, and translators get acomprehensive overview of errors committed by GT and the occurrence of such errors in literarytexts and technical texts translated by GT In addition, after analyzing errors committed by GT inthe literary texts and technical texts, the study gives a conclusion about which genre is translatedbetter by GT In fact, this is a useful note for the students, teachers, and translators when using

GT in their work

1.6 Organization of the Thesis

The thesis was organized into five chapters Chapter 1 (Introduction) addressed the

rationale, aim and objectives, research questions, the significance of the study, the scope of the

study, and the organization of the study Chapter 2 (Literature review) presented an overview of

the translation theory, translation methods, translation shift, translation quality assessment,translation equivalence, errors in translation, machine translation, Google translation, and genres

Chapter 3 (Methods) addressed the research methods and the description of the procedures for data collection and data analysis Chapter 4 (Findings and discussions) reported and discussed the results of data analysis Chapter 5 (Conclusion) summarized the major findings and offered

implications for practice and further studies

CHAPTER 2 THEORETICAL BACKGROUND

In this chapter, the theoretical background for the present study was presented Inparticular, the author focused on the definition and classification of translation and identificationand translation errors of GT

2.1 Translation Theory

In this section, some main contents including definition, methods, shifts, equivalence,errors of translation were described The section was also inclusive of translation qualityassessment

2.1.1 Definition of translation

Translation is defined in many ways and may be understood differently Translation is one

of the diverse means of communication It sets up an association between at least two languagesand their culture Through translation, the characteristic elements from one language aretransferred into another Translation focuses on the translator’s role, from taking a source text andturning it into one in another language, and it also concentrates on the specific product created bythe translator

From the perspective of Carford (1965), translation means "the replacement of textualmaterial in one language (SL) by equivalent textual material in another language" (p 20) Itmeans that translation is the replacement of the word and grammar structures in the ST with theequivalent words and grammar structures in the TL According to Nida and Taber (1982, p 12)translation includes reproducing in the receptor language the closest natural equivalent of thesource-language message, firstly in aspects of meaning, and secondly in aspects of style Fromthe explanation above, translation can be simply defined as transferring the message from the SL

into the TL in terms of meaning and style Hatim and Mason (1990, p 3), on the other hand,focus more on the communication purpose of translation saying that “translation is acommunicative process which takes place within a social context.” Newmark (1988) simplydefines translation as “rendering the meaning of a text into another language in the way that theauthor intended the text” (p 5).

These definitions, in spite of their slight differences in wording, share one thing incommon: Translation means to find equivalents in the target language’s lexis, grammaticalstructures, cultural contexts, and communication situations

Communicative translation

Figure 1 List of translation methods

(Newmark, 1985, p 45)Following are the main features of these translation methods

- Word-for-word translation: This method is often demonstrated as an interlinear

translation In this method, the SL word order is preserved and the words are translated singly by

their most common meanings, out of context Cultural words are translated literally The main use

of this method is both to understand the mechanics of the source language and construe a difficulttext as a pre-translation process (Newmark, 1985, p 45)

- Literal Translation: In this method, the SL grammatical constructions are converted to

their nearest TL equivalents However, the lexical words are again translated singly, out ofcontext This method helps indicate the problems to be solved in the translation process(Newmark, 1985, p 45)

- Faithful translation: Following this method, the translator tries to reproduce the precise

contextual meaning of the original text within the constraints of the TL grammatical structures.S/he transfers cultural words and preserves the degree of grammatical and lexical “abnormality”(i.e., deviation from SL norms) in the translation S/he tries to be completely faithful to the SLwriter’s intention and the text-realization (Newmark, 1985, p 45)

- Semantic translation: This translation method differs from faithful translation in that the

translator has to take more account of the aesthetic values of the SL text (Newmark, 1985, p 46)

- Adaptation: Newmark (1988) said that this is the “freest” form of translation This

method is mainly used to translate plays and poems In this method, the SL culture is converted tothe TL culture, and the text is rewritten (Newmark, 1985, p 46)

- Free translation: Following this method, the translator tries to produce the TL text

ignoring the style, the form, and even the content of the original S/he paraphrases the originaltext using the TL Therefore, the TL text is usually longer than the original, and the TL text iscalled an “intralingual translation” (Newmark, 1985, p 46)

- Idiomatic translation: In this method, the translator reproduces the message of the

original text, but s/he tends to distort the aspects of meaning owing to his or her preference ofcolloquialisms and idioms which do not exist in the original text (Newmark, 1985, p 46)

- Communicative translation: Following this translation method, the translator attempts to

reproduce the exact contextual meaning of the original text in a way that both content andlanguage are readily acceptable and comprehensible to a TL readership (Newmark, 1985, p 46)

The eight translation methods mentioned above provide translators and researchers intranslation with a comprehensive background related to translation methods However, in thisresearch, the author investigated translation errors committed by GT Thus, the identification ofthe translation errors committed by GT in the literary texts and technical texts was based on aframework of translation errors rather than on a framework of translation methods

2.1.3 Translation shifts

According to Catford (1965, p 73), a shift in translation means a departure from formalcorrespondence in the process of going from a source language to a target language In addition, ashift is said to occur if, in a given target text, a translation equivalent other than the formalcorrespondent occurs for a specific source text element According to Bell (1993), to shift fromone language to another is, by definition, to alter the forms From the previous explanations abouttranslation shift, it can be concluded that translation shift is used to describe changes occurs when

a source text is translated into a target language

Catford (1965) states that there are two major types of shift: level shifts and categoryshifts

Translation shifts

Level shifts Category shifts

Structure shifts Class shifts Unit shifts Intrasystem shifts

Figure 2 Types of translation shifts

+ Level shift: Catford (1965, p 73) speaks of a level shift when a source text item has a

textual equivalent on a different linguistic level Level shifts, however, can only occur betweenthe levels of grammar and lexis This restriction, according to Catford’s understanding of

translation equivalence from his structuralism point of view, is not based on a sameness ofmeaning, for meaning is defined as “the total network of relations entered into by any linguisticform” (Catford, 1965, p 35 ) and consequently cannot be the same across languages

The second categorization is category shifts Category shifts mean change from theformal correspondence in translation, which refers to unbounded and rank-bounded translation(Catford, 1965, p 80 ) Catford classifies category shifts into four subtypes:

+ Structure shift: A structure is defined as the patterned way in which a unit is made up

of lower-rank units A structured shift thus occurs when the target structure contains differentclasses of elements or when it contains the same classes of elements but arranges themdifferently According to Catford (1965, p 77), structure shifts are the most frequent amongcategory shifts As an example, Catford presents the translation of an English clause consisting of

the elements of subject, predicate, and complement into a Gaelic clause that is composed of the elements of the predicate, subject, complement, and adjunct.

+ Class shift: We define a class as that grouping of members of a given unit which is

defined by operation in the structure of the unit next above Class shift, then, occurs when a TLitem is a member of a different class from that of the original item (e.g., a noun in the ST istranslated into the TT using a verb)

+ Unit shift: By unit-shift, we mean changes of rank – that is, departures from formal

correspondence in which a unit at one rank in the SL is translated into a unit at a different rank inthe TL

+ Intra-system shift: A system consists of a closed number of elements among which a

choice can be made In fact, the terms available in each system in one language can showfundamental differences from the terms of the same system in another language Intra-systemshifts refer to those changes that occur internally within a system They are regarded only on theassumption that formal correspondence between the two languages should possess approximatesystems All languages have their own systems of number, articles, etc Intra-system shifts occurwhen a term is singular in the source text but its textual equivalent is plural in the TL, or viceversa (e.g., a change in number even though the languages have the same number system)

2.1.4 Translation equivalence

The concept of equivalence can be said to hold a central position in translation studies, andmany different theories of the concept of equivalence have been elaborated within this field inthe past fifty years

Nida (1964) argues that there are two different types of equivalence, namely formal equivalence, which is referred to as formal correspondence and dynamic equivalence According

to Nida and Taber (1982), formal correspondence “focuses attention on the message itself, in bothform and content”, while dynamic equivalence is based upon “the principle of equivalent effect”(p 159).

Formal correspondence occurs when a TL item represents the closest equivalent of an SL

word or phrase Nida and Taber (1982) make it clear that there are not always formal equivalentsbetween language pairs They, therefore, suggest that these formal equivalents should be usedwherever possible if the translation aims at achieving formal rather than dynamic equivalence

Dynamic equivalence is defined as a translation principle according to which a translator seeks to

translate the meaning of the original in such a way that the TL wording will trigger the sameimpact on the TL audience as the original wording does upon the ST audience They argue that

“frequently, the form of the original text is changed; but as long as the change follows the rules ofback transformation in the source language, of contextual consistency in the transfer, and oftransformation in the receptor language, the message is preserved and the translation is faithful”(Nida & Taber, 1982, p 200)

Unlike, Nida and Taber (1982), Koller (1979) presents five types of equivalence asfollows

1. Denotative equivalence: This orients towards the extra-linguistic content transmitted

by a text

2. Connotative equivalence: This respect indicates that individual expressions in the

textual context have not only a denotative meaning but also additional values which mean various

or synonymous ways of expressions

3. Text-normative equivalence: This has to do with text-type specific features or text and

language norms for given text types To put it another way, the SL and TL words are used in thesame or similar context in their respective language

4. Pragmatic equivalence: This means translating the text for a particular readership (i.e.

the receiver to whom the translation is directed and to whom the translation is tuned in order toachieve a given effect)

5. Formal equivalence: This aims to produce an analogy of form in the translation by

exploiting the formal possibilities of the TL or even by creating new forms if necessary

Baker (1992) approaches the concept of equivalence differently by discussing the notion

of non-equivalence at word level and above word level, grammatical equivalence, textualequivalence, and pragmatic equivalence

Non-equivalence at word level means that the target language has no direct equivalent for

a word which occurs in the source text Common problems of non-equivalence then involve suchcases as culture-specific concepts, when the SL concept is not lexicalized in the target language,the SL word is semantically complex, the SL and TL make different distinctions in meaning, the

TL lacks a superordinate, the TL lacks a specific term (hyponym), differences in physical orinterpersonal perspective, differences in expressive meaning, difference in form, differences infrequency and purpose of using specific forms, and the use of loan words in the source text

Non-equivalence above word level is closely related to the differences in the collocation

patterning of the SL and TL These differences create potential pitfalls and can pose variousproblems in translation

Grammatical equivalence is concerned with the differences in the grammatical structures

of the SL and TL These differences often result in some change in the information content of themessage of a translation This change may take the form of adding to the target text someinformation which is not expressed in the source text This can happen when the TL has agrammatical category which does not exist in the SL Likewise, the change in the informationcontent of the message may be in the form of omitting information specified in the source text Ifthe TL lacks a grammatical category which exists in the SL, the information expressed by thatcategory may have to be ignored

Textual equivalence is achieved through the realization of cohesion or cohesive devices

such as reference, substitution, ellipsis, and conjunction and lexical cohesion from the source textinto the target text

Pragmatic equivalence is realized by means of studying and translating coherence and

implication from the SL to the TL

2.1.5 Translation errors

In this study, the researcher examined the quality of GT’s translations in two genres (i.e.,literary texts and technical text) Therefore, it was necessary to look at the errors made by GTwhen the selected texts of these two genres were translated The following sections discuss issuesrelated to translation errors

2.1.5.1 Definition of translation errors

Mossop (1989) argues that “a given rendering will be deemed to be non-translation if itfails to conform to the concept of translation predominant in the target culture” (p 55) Heidentifies translation errors in terms of cultural norms and formal equivalence It means thattranslation errors defined by Mossop (1989) includes the achievement of formal equivalence but

excludes other critical factors such as smoothness, readability, and consistency in TL texts.Besides, formal equivalence, as defined by Nida and Taber (1982), is achieved by translatingliterally and preserving rhythm, special stylistic forms, expressions in syntax and lexis,metaphors, wordplay, and so on; therefore, formal equivalence is mainly used in translatingpoems and songs A broader notion of errors in translation is proposed by Pym (1992) Thisscholar supposes that translation errors may be attributed to lack of comprehension; misuse oftime; inappropriateness to readership, language, pragmatics, culture; over-translation; undertranslation; and discursive or semantic inadequacy Compared to the definition by Mossop(1989), Pym (1992) suggests more kinds of translation errors However, these kinds of errors arenot systematically classified.

Ten years later, Aveling (2002) presents a more comprehensive and systematicclassification of translation errors According to Aveling (2002), translation errors occur when thetranslator fails to gain equivalence, adequacy, and accuracy This definition is morecomprehensive as it stresses that equivalence covers many different types Besides, it is moresystematic because Aveling (2002) emphasizes that translation errors can be divided into “dumbmistakes” and “deliberate mistakes” The former is due to the translator’s lack of competence,and the latter occurs when the translator poses a purpose to recreate the text

2.1.5.2 Classification of translation errors

Not only the definition but also the classification of translation errors has attracted amyriad of efforts from scholars and researchers Nevertheless, due to the complexity of thispractice, it remains intricate to establish a single comprehensive list of all the translation errorshaving been observed

House (1977) suggests that researchers should prepare separate profiles for the ST and the

AT. House also argues that when the source text's and the translation text's profiles do not match, there is an error House describes two types of errors:

A covert error is an error which results from a mismatch of one situational dimension with

a similar one in the TT, and a covert error is a mismatch of the denotative meaning of the ST andthe TT element or a breach of the target language system

An overt error is an error which results from a non-dimensional mismatch Over errors can be divided into seven categories as follows

Translation Errors

Covert Errors: result

from a mismatch of one

situational dimention with

a similar one in TT

Overt Errors: result from

a non-dimensional mismatch, and can bedivided into:

1. Not Translated

2. Slight Change in Meaning

3. Significant Change in Meaning

If a translation error is defined as a failure to carry out the instructions implied in thetranslation brief and as inadequate solutions to a translation problem, then translation errors can

be classified into four categories (Nord, 1997, p 75)

Pragmatic translation errors are caused by inadequate solutions to pragmatic translation

problems such as a lack of receiver orientations

Cultural translation errors are made by an inadequate decision with regard to the

reproduction or adaptation of culture-specific conventions

Linguistic translation errors are caused by an inadequate translation when the focus is on

language structures

Text-specific translation errors, which are related to a text-specific translation problem

like the corresponding translation problems, can usually be evaluated from a functional orpragmatic point of view

As linguistic-based typologies tend to offer more information about the types of errorsfound, the classification scheme employed in this study derived from the aforementioned errortypology by Cabeceran et al (2010) It has a hierarchical structure as shown in Figure 3 At the

first level, errors are split into five major categories: orthographic, morphological, lexical, semantic, and syntactic errors.

17

orthographic errors

morphological errors

Category errors

Figure 4 Hierarchical structure of the error typology

This figure presents the hierarchical structure of the error typology The definition of eacherror type is provided in the following paragraphs

Orthographic errors can be identified in three subcategories: punctuation, capitalization,

and spelling Punctuation errors are inappropriate usages of punctuation marks in the TL text.Capitalization errors are incorrect upper or lower cases Spelling errors occur when words aremisspelt This main category excludes incorrectly inflected words, which are categorized as

morphological errors The errors mentioned above induce only minor disturbances of readers’understanding of the TL text.

The next category is morphological errors, which concern the form of TL words.

Morphological errors are divided into three subcategories: incorrect forms of verbs, nouns, orwords of any other part-of-speech (POS) The first two subcategories do not seriously affect thereader’s understanding of the TL text as a whole, for English users can easily identify these errorsand replace them with the right forms It is worth noting, though, that the impact of an errorcategory differs across target texts

Some linguistic errors could have more influence than others on the overall quality of thetext outputs (Cabeceran.F et al., 2010) The last four categories are more important than the firsttwo in terms of their impact on readers’ understanding of the TT The first category includeslexical errors One broad definition of such errors is inappropriate lexical choices For theclassification scheme employed in this study, the definition is narrowed down to erroneousoccurrences and non-occurrences of target words, i.e., extra words and missing words Extraword errors include any word that should be omitted and cannot be substituted by another word

of a more appropriate meaning; if the erroneous word can be replaced by another word, it should

be counted as a semantic error Missing words include only missing content words, for they carryessential meanings for the text to be understood Missing prepositions, conjunctions, and articlesare classified as syntactic errors so that more information about errors in sentence formation can

be provided

Semantic errors are found when the system chooses the wrong word to render a sourceword The meaning of this TL word can be irrelevant as well as related to that of itscorresponding SL word For instance, it can be a hypernym or hyponym of the source word

Five subcategories are distinguished in syntactic errors, which concern the structuring ofsentence elements Conjunction, preposition, and article errors are all errors of the respectivePOS Category errors occur when words are rendered in the wrong POS In the English language,

it is not easy for a computer program to identify the POS or class of a word, as an English wordcan sometimes be either a verb or a noun, depending on its context The last subcategory is acomplicated notion Syntactic element reordering errors occur when a sentence needs to bereordered or restructured To avoid confusion in the number of errors of this category, this errortype is only counted once per sentence

2.1.6 Translation quality assessment (TQA)

The quality of a translation is a serious concern for Translation Quality Assessment(TQA) approaches The main issue is how to measure and express this quality There have beenmany attempts to find ways to tackle these issues and evaluate the quality of a TL text but notmany models have been introduced to assess translation quality assessment We can mentionWaddington’s model of translation quality assessment in 2011 This model seems less subjectiveand translation was evaluated according to some pre-set criteria After analyzing twentyparagraphs selected randomly and compared and contrasted precisely with the paralleltranslations, Waddington’s model turned to be incomplete, relating to translation shifts andadditions However, the model does not sound promising One of the promising approaches wasthe model provided by House (1996)

House's (1996) assessment model is based on Halliday’s Systemic-Functional Theory(SFT), but it also draws eclectically on Prague School’s ideas, speech act theory, pragmatics,discourse analysis and corpus-based distinctions between the spoken and written language It

provides the means for the analysis and comparison of an original text and its translation on threedifferent levels: Language/Text, Register (Field, Mode, and Tenor) and Genre.

through the text

Mode

How the text iscommunicatedhow its parts fit

Figure 5 Three different levels of an original text and its translation

According to House (1977), the equivalent sought should be an equivalent of function;that is, both SL and TL texts must present the same function and the text's function can only bemade explicit through a detailed analysis of the text itself

This is the basis for the model, and what makes it different from other criteria forestablishing equivalence is the fact that those criteria rely either on the writer's intention (an itemthat is not open to the empirical investigation) or on the reader's responses, which presentproblems to be measured The function of a text would then be "the application or use of what thetext has in the particular context of a situation" (House, 1997, p 37).

Thus, each text is an individual text embedded in a unique situation, and in order tocharacterize the text's function, it is necessary to refer a text to a particular situation Toaccomplish this, the notion of the situation has to be broken down into the following specificsituational dimensions (register) analysis: (House, 1997, p 45)

a Field refers to the subject matter and social action and covers the specificity of lexical

items

Subject Matter: It can be a Novel, Poem, Play, etc

Social Action: It can be Specific, General, Popular, etc

b Tenor refers to who is taking part, to the nature of the participants, and to the addresser

and the addressee and the relationship between them This dimension includes the addresser’stemporal, geographical, social provenance as well as his intellectual and emotional stance, i.e.,his personal viewpoint vis-a-vis the content he is portraying and the communicative task he isengaged in The ‘social’ role relationship may be either symmetrical (marked by the existence ofsolidarity or equality) or asymmetrical (marked by the presence of some kind of authority)

c Mode refers to both the channel – spoken or written (which can be “simple”, e g.,

“written to be read” or “complex”, e g., “written to be spoken as if not written”), and the degree

to which potential or real participation is allowed for between the interlocutors (House, 1997, p.109)

Participation can also be either simple or complex An example of simple participationwould be a monologue with no addressee-participation while complex participation involvesvarious addressee involving mechanisms characterizing the text, e.g., “characteristic use ofpronouns, switches between declarative, imperative and interrogative sentence patterns or thepresence of contact parentheses, and exclamations” (House, 1997, p 40).

Medium is Simple if it is written to be read and Complex if it is written to be heard Participation is Simple if it is in a monologue and Complex if it addresses a large

community

House (1997) defines genre as “a socially established category characterized in terms ofoccurrence of use, source and a communicative purpose or any combination of these” (p 107).When discussing the function of the text, House uses the notions “ideational function” (i.e., usinglanguage to describe things in the external world and to present and evaluate arguments andexplanations) and “interpersonal function” (i.e., using language as an expression of a speaker’sattitudes and his influence on the attitudes and behavior of the hearer)

In this part, the author has addressed the definition of translation, methods of translation,translation shifts, translation equivalence, translation errors, and translation quality assessment

2.2 Machine Translation

In this part, the author presents the history of machine translation and Google Translate.This is the important knowledge for the author to get a comprehensive viewpoint about GoogleTranslate and the establishment and development of machine translation

2.2.1 The history of machine translation

According to Farah (2017), the term “machine translation” (MT) refers to computerizedsystems responsible for the production of translations with or without human assistance It

excludes computer-based translation tools which support translators by providing access to onlinedictionaries, remote terminology databanks, transmission and reception of texts, etc It was firstdeveloped in the 1950s as a computer system that performed automatic translation In thebeginning, the system worked when the whole text in a source language (SL) was translated into

a target language (TL) as a single task without human intervention The source text outputproduced by machine translation is known as “raw output” as it provides a quick translation ofthe original (Farah, 2017) These “raw outputs” usually offer informative translation It means theoutput produced by machine translation only provides a surface translation of the target textwithout human involvement, as it is a statistical machine translation It is deemed to translatewhat is input into the system Machine translation is one of the oldest applications and has beenused in computer science Nonetheless, due to the development of global needs for transferringknowledge and information, it has been used in language and linguistic fields as well.Additionally, the main objective of machine translation during the early stages was to replacehuman translators as it was expected to do the translation work (Farah, 2017) However, theunsatisfactory output produced by the system and problems that could not be solved due tolexical ambiguities produced by machine translation made the enthusiasm among experts decline.Many companies that developed machine translation at an early age started to admit that theirsystems were not able to produce perfect translations Therefore, due to the failed system andunsatisfactory output produced, it has led to the development of computer-assisted translation(CAT) (Farah, 2017) Subsequently, computer-assisted translation enables human intervention inthe machinery system, thus, helping translators to work quickly It helps assist humantranslation’s work and gives a human translator an extended control over the process

2.2.2 Google Translate

Machine translation is a sub-field of natural language processing and the area ofinformation technology In general, it is based on computer technology that uses software to

translate one natural language into another Google Translate is an automatic machine-translation

service provided by Google Inc It translates one written source language into another directly orwith English as a medium It is a free translation service that currently provides instanttranslations between 103 different languages In addition, GT can translate words, sentences andweb pages between any combinations of the supported languages GT has been created with theexpectation to make useful information universally accessible, regardless of the language inwhich it has been written (Google, 2019) When GT generates a translation, it searches forpatterns from hundreds of millions of documents to help make a decision on the best availabletranslation By identifying patterns in documents that have already been translated by humantranslators, GT can make quick decisions as to what a suitable translation could be This

procedure of seeking patterns in a large number of texts is called Statistical Machine Translation

(SMT) and the quality of a translation is likely to vary across languages (Google, 2019.) Thefollowing figure presents all 103 languages currently supported by GT

Figure 6 Languages supported by GT (Google 2019)

As seen in Figure 1, the variety of the languages supported by GT is rather extensive.However, Google is trying to make them function better Google has the intention of supportingother languages as well, as soon as the translation quality is good enough Currently, there arevarious online MT systems (such as Microsoft Translator, Google Cloud Translation API, BingTranslator, Amazon Translate, IBM Watson Language Translator and etc) but all of them arecommercial applications Therefore, they are not able to compete with GT, which is free andwhich supports many languages

Translations produced by GT can be improved by selecting the wanted alternative fromthe given alternative translations For example, when the translator encounters a translation thatdoes not seem good enough, s/he can simply click the phrase in question and choose a betteroption By clicking the option, GT will learn from the translator’s feedback and continue toimprove over time In addition, the translator has the option of using Google Translator Toolkit to

upload translation memories online When the translator logs in to Google, the personallyuploaded data will be taken into consideration while translating documents The next figuredisplays Google’s free online MT system interface in its present form.

Figure 7 GT graphical user interface

Google’s Graphical User Interface, as shown in Figure two, has been designed to looksimple, but it actually has surprisingly many features regardless of the plain design The ST boxhas been placed on the left, and the TT box on the right Any text can be just copied and pastedinto the box The SL and the TL can be selected, but in case the user is uncertain of the SL, GT isable to automatically detect it The translation direction can be easily reversed by clicking on thereverse button A link of a website can also be pasted to the box, which will lead the user to theposted site, but with the desired TL instead Thus, the design of the webpage remains untouched,but the language of the text changes Translations can be rated by the user according to three

different categories: helpful, not helpful or offensive In addition, the word is highlighted in bothtexts when the mouse cursor is moved onto a specific word This makes it easier for the humantranslator or the user to spot how GT has translated a particular word or expression With arecently added feature, by holding the shift key on the keyboard, the user is able to drag andreorder words in the TT box In addition, the user can view alternate translations by clicking thetranslated words in the TT box The GT system also provides the user with a computer generatedvoice which will read the texts out loud for those who are interested in listening to the texts.

Google’s translation software has been designed not only for regular computers but alsofor mobile devices This has greatly expanded the possibilities of using MT in different kinds ofsituations A free downloadable application of GT was programmed and released in August 2008

to utilize the iPhone by Apple Inc Additionally; GT was released in the Android Market for smartmobile phones that use the Android operating system in January 2010 The available mobileapplications make Google’s services even more versatile and competitive, reaching out to agreater number of users

Generally speaking, in this part, the author has provided readers with the history ofmachine translation and an overview of Google Translate