MINISTRY OF EDUCATION AND TRAINING QUY NHON UNIVERSITY NGUYEN THI THU HIEN THE QUALITY OF GOOGLE TRANSLATE'S VIETNAMESE TRANSLATIONS OF THE ENGLISH METAPHORS IN "THE COUNT OF MONTE C
INTRODUCTION
Rationale
In the context of today's globalized world, it is undoubtedly that proficiency in a language other than the native one brings a lot of extreme benefits Specifically, the ability to communicate in a foreign language contributes to creating a genuine connection among people and the opportunity to engage with this world Charlemagne once said that to have another language is to possess a second soul In approaching a new language, translation is not only essential but also crucial With the aid of valuable tools like dictionaries or machine translation, translating tasks nowadays are easier for everyone.
In the current era of technology, even though translating is a time- consuming and complicated task, this task is easier for us to support a machine translation, which helps us quickly access the information in another language Despite its convenience, MT sometimes gives an output that has some or many errors Consequently, the quality of the product affects text cohesion as well as the reader's comprehension MT's quality is always of great concern For most, the successful implementation of MT is a great leap and a long-term investment to meet human requirements.
The language in literature is often more complex and elaborate than that in everyday language The use of rhetoric is indispensable to make poems and novels more vivid and more attractive to readers Metaphor is one of the rhetorical devices frequently used by poets and writers in their works Metaphor is human emotion expressed through human language The standard translation is not easy, and the translation of rhetoric is many times more complicated Metaphors make our lives more colourful; however, translating metaphors is more challenging than other language types due to the great importance of the meaning, form, content, and culture
The quality of a text or a metaphor translated into another language depends on many factors In the report of Automatic Language Processing Advisory Committee (1966, p 3), many experts in translation seem to agree that three critical factors that influence translating are (1) good knowledge of the target language, (2) comprehension of the subject matter, and (3) adequate knowledge of the source language
Medvedev (2016), Longears (2015), Giau (2019), Thu (2020), and My
(2020) are the researchers who researched GT’s quality GT showed a lot of advantages such as its speed, its update in translating diverse types of languages, and its free usage; however, it did not always provide precise translations and committed many word-for-word errors They did not mention the quality of GT in translating detailed types of metaphors
To investigate the quality of Google Translate (GT) in converting metaphors, I decided to study the theme "The quality of Google Translate's Vietnamese translations of the English metaphors in 'The Count of Monte Cristo' by Alexandre Dumas."
1.1 Aim and Objectives of the Study
This study aims to evaluate the quality of GT's Vietnamese translations of the English metaphors in the novel "The Count of Monte Cristo" by Alexandre Dumas
In order to achieve the above aim, the researcher will try
- to identify and classify the metaphors in the novel "The Count of Monte Cristo" by Alexandre Dumas,
- to identify and classify the errors in GT's Vietnamese translations of the
- to evaluate the quality of GT's Vietnamese translations of the English metaphors in terms of translation errors
In order to achieve the above aim and objectives, the researcher collected data and analyzed them to answer the following questions:
1 What metaphors are there in the novel "The Count of Monte Cristo" by Alexandre Dumas?
2 What types of errors does GT make in translating these English metaphors into Vietnamese?
3 How well does GT translate these English metaphors into Vietnamese in terms of translation errors?
Alexandre Dumas (1802-1870) has many historical works of high adventure translated into many languages, including The Count of Monte Cristo, The Three Musketeers, Twenty Years After etc The researcher chose The Count of Monte Cristo (Dumas, 1844) to take out the needed database due to its adventurous content and humanitarian
This study analyzed GT's Vietnamese translations of 120 metaphors translated by Mai The Sang to have a reliable basis for comparing GT's translations and Mai The Sang's in the process of identifying GT's errors The researcher randomly selected those 120 metaphors to make the data analysis more objective
This study is expected to contribute to the quality assessment of GT Its findings may have precious value to other researchers and translators who want to focus on metaphor translation with the aid of GT Another unique point of the study is to help specialized English students and English teachers in pre- translating by exploiting the GT tool in learning and teaching English Furthermore, the study is supposed to contribute to Internet users interested in better understanding the metaphors and the translation These people can rely on the research findings to save time editing translations to meet their needs
In terms of knowledge, this study serves as a reference for researchers who want to make further studies about metaphors and MT, especially GT Moreover, translators, English teachers, and Vietnamese learners of English are provided with a general knowledge of metaphor, translation, and translation errors to use GT as effectively as possible
In terms of practice, it is easier for English teachers to create efficient and effective techniques in teaching vocabulary related to both metaphors and other subjects with the aids of GT in general and translation in particular GT has sufficient supply to provide the users with the comprehension of the meanings of the words, the way to pronounce them, and examples of how to use them Internet users who have an interest in updating knowledge in literature can access this research
This thesis consists of five chapters:
Chapter 1 (Introduction) presents the rationale, the aim and objectives, the research questions, the scope, and the significance of the study
Chapter 2 (Literature Review) provides the background of metaphor, translation theory, Google Translate, and the overview of the novel "The Count of Monte Cristo."
Chapter 3 (Methods) discusses the methods and procedures for data collection and data analysis
Chapter 4 (Findings and Discussion) reports the findings from data analysis
Chapter 5 (Conclusions and Implications) gives conclusions withdrawn from the results of the study The researcher will also show some implications for relevant beneficiaries.
Organization of the Thesis
In this chapter, the researcher presents the study's theoretical background, including an overview of metaphor that concerns its definition, classifications, and identification features Besides, the researcher also provides background knowledge about translation consisting of its description, methods, equivalence, and translation quality assessment In this part, the researcher introduces the novel "The Count of Monte Cristo" by Alexandre Dumas Last but not least, the researcher concentrates on discussing MT, GT, and translation errors
This section addresses the definition of metaphors, their classification, and the procedures for identifying metaphors
For most people, metaphor is a device of poetic imagination and the flourish of rhetoric—a more extraordinary matter than ordinary language (Lakoff & Johnson, 2013, p 453) They also found that metaphor is pervasive in everyday life, not only in language but also in thought and action
Metaphor is derived from the Greek words "meta," which means "over," and "pherein," which means "to bear." Metaphor is historically described as an indirect analogy between two objects in which characteristics that usually belong to one thing are translated or carried over to the second Richards (1936, p 120) describes metaphor as including a comparison: It could simply be the joining of two things to enable them to fit together; it could be an analysis of them both to see if they are similar and different; it could be a way of drawing attention to their similarity or a method of drawing attention to certain facets of one by the co-presence of the other This particular feature is also reaffirmed
LITERATURE REVIEW
Metaphors
This section addresses the definition of metaphors, their classification, and the procedures for identifying metaphors
For most people, metaphor is a device of poetic imagination and the flourish of rhetoric—a more extraordinary matter than ordinary language (Lakoff & Johnson, 2013, p 453) They also found that metaphor is pervasive in everyday life, not only in language but also in thought and action
Metaphor is derived from the Greek words "meta," which means "over," and "pherein," which means "to bear." Metaphor is historically described as an indirect analogy between two objects in which characteristics that usually belong to one thing are translated or carried over to the second Richards (1936, p 120) describes metaphor as including a comparison: It could simply be the joining of two things to enable them to fit together; it could be an analysis of them both to see if they are similar and different; it could be a way of drawing attention to their similarity or a method of drawing attention to certain facets of one by the co-presence of the other This particular feature is also reaffirmed by Newmark (1998) Newmark (1998, p 104) describes a metaphor as any representational expression: the transmitted real-world meaning, the personification of abstraction, the use or collocation of a concept to something it doesn't mean
According to Kovecses (2002), metaphor means two unlike objects' distinction without the use of the terms "like" or "as."
In "Metaphor Dictionary," Sommer (1995) describes a metaphor as comparing two different things or concepts to illuminate their similarity It does what might otherwise only be said in several sentences, if at all, in one word or sentence
The former researchers make clear definitions of metaphor From their descriptions, we can reconfirm that metaphor implies comparison between two things or two people Metaphors can be a word, a sentence, a proverb, an idiom, and even a complete text Thanks to metaphors, more information is conveyed than that in its interpretation
Different researchers have different ways to classify types of metaphors Holme (2004, pp 59-66) supposes noun metaphors, verb metaphors, adjective metaphors, and adverbs metaphors These types of metaphors focus on a particular part of the speech of the words in the context
Broeck (1981, p 75) divides metaphor into three categories according to their relative degree of being institutionalized or not The first category comprises those that have gradually lost their uniqueness and have become part of the established semantic stock of the language They are too familiar in our daily life that sometimes they have not been recognized as a metaphor, such as in the face of, beforehand, to harbour evil thoughts, lay a finger on, etc A second category is common to a literary school or generation The third category is that of private metaphors that poets innovate
Newmark classifies metaphors into six types: dead, clichés, stock, adapted, recent, and original metaphors
- Dead metaphors: Overuse has resulted in dead metaphors losing their meaning They've lost their figurative and connotative meanings over time and are now used as ordinary words (Newmark 1998, p 106)
- Cliches: Clichés lie between dead and stocks, sometimes emotionally, but without corresponding to the truth, as a cover for clear thinking (Newmark
- Stock metaphors: Newmark (1998, p 108) describes this style as a practical and concise approach to cover a physical and/or mental circumstance
- Adapted metaphors: According to Newmark (1998, p.105), this metaphor has been translated into a new form by its speakers or authors
- Recent metaphors: Newmark (1998, p 111) describes these metaphors as a metaphorical neologism that spread quickly throughout the SL
- Original metaphors: According to Newmark (1998, p 112), these metaphors are generated or cited by the SL researcher They hold the essence of the writer's valuable letter These metaphors are not only nuanced, but they also have many interpretations
Johnson (2019) proposes five types of metaphor:
- Direct metaphors are the ones in which two people or things are compared directly
For example: "He is a fish out of water in his new job."
"Tina is a whirlwind of ideas."
- Implied metaphors: In implied metaphors, there are not any direct comparisons Although no direct comparison is being made, the implication for the situation is apparent through the language and context used
- Extended metaphors: A metaphor that continues notably further than a single phrase; at least one sentence long and typically less than a paragraph For example: "When he looked into her eyes, he realized his ship had sailed The love that had once existed between them was gone, and now all the existed were the rocky waters of friendship between two people who had once been much more." (Johnson 2019)
- Dead metaphors: A metaphor has been used so often that the comparison is no longer striking
- Mixed metaphor: A metaphor that uses more than one different comparison; however, it is not usually good because it can make writing sound confused or clunky
Johnson (2019) advocates maintaining the original form of the former authors His classification is concise, easy to identify but still full of the main types of metaphors drawn by the previous authors through their long research process Therefore, in this study, Jonhson's classification of metaphors is used as the basis of analysis.
Each metaphor creates and jumps over a gap; our surprise and pleasure at its use come from the unexpected discovery of a similarity that connects two things we had never before considered the same in any way (Seligman & Weller, 2018, p 78) To figure out metaphors in a text, we need to use a procedure with many appropriate steps
Group ( 2007, p 3) provides the procedure to detect metaphors in a text
He affirms that this procedure is an "explicit, reliable, and flexible method for identifying metaphorically used words." This method consists of four steps
Step 1: Read the whole text to get a general interpretation of text
Step 2: Determine the text's lexical units
Step 3: Set the definition of the lexical unit and consider before and after that unit Determine if the qualitative sense differs from the simple meaning but can be interpreted about it
Step 4: Label the lexical unit as metaphorical if this is the case
Steen (2002, p 26) also gives instructions to identify metaphors in discourse He says that a complete explication of the procedure would include the following steps
- Discourse can be broken down into units, resulting in a compilation of all semi-independent clauses
- Using pronominalization, deictics, and alternate but general expressions, explain the topics involved in substitution and co-reference (like thing, man, and so on)
- By abstract indications of their position, elucidate all non-realized but semantically traditional arguments of a predicate
- Determine whether arguments referring to entities in the text world do so simply, that is, explicitly, or whether they (can) evoke another, more concrete or familiar, the realm of use – this may include all nominally used terms
- Check whether any predicates or arguments designate the properties of entities in the text, or whether they can invoke a different domain of use, particularly more concrete or typical - can take care of any used predictively
- Check if the predicates or arguments designating relationships or characteristics of entities in the text world evoke another
The researcher had applied the above two procedures in the process of identifying metaphors Consequently, the researcher randomly took out 120 metaphors in the novel “The Count of Monte Cristo”, translated by Mai The Sang.
Translation Theory
This section shows translation and translation theory's definition, translation methods, equivalence in translation, and translation quality assessment
According to Munday (2016, p 8), translation has several meanings It is the general subject field or phenomenon and the product in which the text has been translated Translation is also the process of producing the translation conducted by one or many translators
Newmark (1998) has another definition of translation Newmark (1998) defines translation as the process of rendering the meaning of a text into another language in the way the author intended the text Also, Catford (1965) assumes that translation is an activity performed on languages: a process of replacing a text in one language for a text in another one
Bell (1991) also assumes that translation is when a target language's text is expressed with the same semantic and stylistic as the source language
In brief, Translation is the process of substituting equivalences between the SL and the TL In this process, the comprehension of the two languages about cultures and the general meaning of the text is vital
Jiří (2011) considers translation theory like many other specialized disciplines in recent decades Translation theory is in a state of conflict between specialization, promoting a more thorough investigation of individual aspects of translation and incorporating these specialized findings into broader cultural contexts
Newmark (1998, pp 45–48) offers eight translation methods in a V diagram
Word- for- word translation Adaptation Literal translation Free translation
Figure 1 Eight translation methods (Newmark 1998, pp 45–48)
- Word-for-word: This is also shown to be interlinear with the TL just below the SL terms The SL word order is maintained, and the words are individually interpreted from context by their general meanings
- Literal Translation: The grammatical constructions in SL are transformed to their closest TL counterparts, but the lexical words are interpreted separately, out of context As a result, the problem to be solved is the pre-translation procedure
- Faithful Translation: Under the constraints of the TL grammatical constructs, a faithful translation seeks to recreate the original's exact contextual meaning
- Semantic Translation: Semantic translation differs from faithful translation only in one point: it would consider the SL text's aesthetic meaning
- Adaptation: The styles, characters, and storylines have remained
However, the culture of SL's text translated into TL is redesigned This translation is primarily used for plays and poetry
- Free translation: Free translation reproduces the subject without the original form or the contents It is an intralingual translation, a much longer paraphrase than the original This translation is always lengthy and pretentious and not an at all translation
- Idiomatic translation: This method attempts to reproduce the original's 'message' but appears to misinterpret nuances of meaning by choosing colloquialisms and idioms where they do not occur throughout the original
- Communicative translation: This type of translation method aims to convey the original's exact contextual meaning in a manner in which the readership can find both substance and vocabulary suitable and understandable
Though Newmark (1998) presents eight translation methods, he emphasizes that the semantic and communicative methods are more practical and familiar due to their accuracy and economy
The two methods have some differences
Semantic translation is personal and individual, reflects the writer's thinking process, and attempts to over-examine and follow facets of the meaning Typically, this translation is within the original language
Communicative translation is a form of social translation that focuses on the text's fundamental meaning and seeks to under-translate to be plain, straightforward, and concise In brief, this translation is often superior to the original
Equivalence is the core issue in translation that various definitions have been introduced
From the middle of the 20 th century, many researchers have suggested their definitions of the theories of equivalence; Among them, the prominent illustrations belong to Catford (1965), Pym (2009), Nida (1964), (Nykyri
Equivalence has been understood as "accuracy," "adequacy,"
"correctness," "correspondence," "fidelity," or "identity"; “it is a variable notion of how the translation is connected to the foreign text” (Venuti, 2012, p.5)
According to Catford (1965, p 50), while source language (SL) and target language (TL) rarely have "the same meaning" in the linguistic context, they will operate in the same case He believes that when SL and TL texts are interchangeable in a given subject, they are considered translation equivalents in total translation
When there is no defined definition in the target language, a translation counterpart is used to express the source concept of Nykyri (2010, p 23) A linguistic expression referring to a related image in the target language but not the exact definition as the phrase in the source language is considered nearly equivalent There may be situations where no particular equivalent may be used due to linguistic and cultural variations in constructing terms and definition structures
Equivalence, according to Pym (2007, p 6), is a relationship of equal value between a source-text segment and a target-text segment:
- From form to function, linguistic equivalence can be defined on any basis
- Before the act of translation, natural equivalence is assumed to occur between languages or cultures
- Natural equivalence should not be compromised by directionality: when converted from language A to B, or vice versa, it should be the same
Newmark (1998, p 48) claims that the equivalent effect is the desired outcome rather than the aim of either translation In addition, it is impossible to occur in two situations: (a) SL text intends to influence, and the TL translation is to notify (or vice versa), and (b) SL and TL texts have a significant cultural difference
In brief, there has been a debatable subject about the concept of equivalence in translation since scholars have shown various theories on it
The methods and practices of TQA are usually different, according to Moorkens, Castilho, Gaspari, & Doherty (2018, p 11), at both the micro and macro levels as their significance can differ significantly in various individuals, groups, and contexts Further uncertainty will be introduced as the scope and objective of an assessment will possibly vary depending on whether it occurs in the context of a production or research phase TQA's industry aims to identify, measure, and deliver the translated content to customers, buyers, and end-users with high-quality products In research, the goal of TQA is generally to achieve a metric that can demonstrate a significant change in quality, usually an improvement, from the previous studies or between various translation processes.
Machine Translation
Definition of Machine Translation, Google Translate and Translation errors are clarified in this section Furthermore, the summary of the novel "The Count of Monte Cristo" and some previous studies involving this thesis is mentioned
Hutchins & Somers (1992, p 3) has stated that Machine Translation (MT) is now the standard and conventional name for computer systems that produce translations, with or without human assistance, from a natural language into another Former names such as 'mechanical translation' and 'automatic translation' are no longer commonly used in English, but their counterparts are still in other languages
Computer-based translation facilities that assist translators by providing access to dictionaries and remote vocabulary files, enabling the delivery and receipt of machine-readable documents, or dealing with word processing, text formatting, or printing equipment are not included It does, however, provide programs in which translators or other users help computers produce translations These translations include different variations of document scheduling, online communications, and eventual quality revisions
According to ElShiekh (2012), translation with the aid of the Internet has taken a further step towards providing users with more informative and reliable products Among Internet applications, Google or Yahoo are designed to search for many kinds of documents, images, information, webpages, etc., in less than half a second
Google Translate, a statistical computer translation tool launched in
2007, currently offers automatic translations between over 50 languages, either directly or through a pivot language In September 2008, Slovenian was added to the Google Language List GT's popularity is due to its statistical methodology, which has yielded more significant outcomes than traditionally supported rule-based linguistic schemes Systran retrieves data from bilingual dictionaries and grammars and then complements it with linguistic and other rules Contemporary statistical MT systems rely on many human-engineered translations to automatically assume a statistical translation model The underlying assumption is that there are several possible translations for each source language feature, and the system assigns the highest probability to the most appropriate translation
The Google Translate service (Lotz & Rensburg (2014, p 237)) is delivered free of charge by Google Inc., enabling customers to provide lightning speeds in their words, expressions, documents, and even Web pages
GT functions on the statistical machine translation theorem (MT)
Typically the word "error" suggests something is wrong According to Hansen (2015), written mistakes in both the original and translated texts are considered translation errors in the written language Those translation errors can be pragmatic, semantician, idiomatic, orthographic, or linguistic errors
If a translation is described as an output from the target text based on the source text, the presence of a relationship between two texts leads to a translation error
Gửrửg (2014, p 444) evaluates that translation is a complex linguistic process in which quality is probably the most challenging variable but essential in this process When we conduct product assessment and translation quality testing, we will address some critical questions, such as "what exactly are the main features of good content, and how can we evaluate them?" "What are the general issues of allowing robots to "understand" language?" "Which text styles lend themselves best to Machine Translation (MT)?" "How do we consistently equate two translations of the same source text?" "What are the most suitable and dependable methods for assessing translated content?" "What are the prerequisites for successful post-editing?"
Flanagan (1994, p 65) suggests some reasons that cause difficulties to quantify the MT quality:
- A text may have multiple correct translations
- Defining error limits in MT performance is also problematic as errors require single words and sentences, concepts, or interactions across sentence boundaries
- There could be more than one error
- The cause of MT performance errors is not always clear
- There are several types of errors that GT may cause in a translation Automatic Language Processing Advisory Committee (1966, p 77) provides two error classifications The first one is by Thomas J Watson, who classifies errors by MT into Transliterated words, Multiple meanings and ambiguities, Word order rearranged, and Miscellaneous insertions and corrections The other one in which the classification is overlapped in some aspects is by Arthur D Little Errors are divided into six types: Word omission,
Wrong words, Unnecessary words, Symbol, Phrase not interpreted, and Word order
Flanagan (1994, p 67) applies different error categories with more criteria than the above researchers However, his classification focuses mainly on the grammar of the text, not on the semantic field This category is not suitable for complicated contexts Table 2.1 gives some descriptions of Flanagan’s Error Categories
Table 2.1 Flanagan's Error Categories (Flanagan 1994, p 67)
Not Found Word Word not in the dictionary
Capitalization Incorrect upper or lower case
Elision Illegal elision or elision not made
Verb inflection Incorrectly formed verb, or wrong tense
Noun inflection Incorrectly formed noun
Other inflection Incorrectly formed adjective or adverb
Rearrangement Sentence elements ordered incorrectly
Pronoun Wrong, absent, or unneeded pronoun
Article Absent or unneeded article
Preposition Incorrect, absent, or unneeded preposition
Negative Incorrect, absent, or unneeded preposition
Conjunction Failure to reconstruct parallel constituents after the conjunction or inability to identify boundaries of conjoined units
Agreement Incorrect agreement between subject-verb, noun- adjective, past participle agreement with preceding direct object, etc
Clause boundary Failure to identify clause boundary or clause boundary unnecessarily added Word Selection Word selection error (single word)
Expression Incorrect translation of a multi-word expression Farrỳs, Costa-Jussà, Mariủo, & Fonollosa (2010) describe a simple error scheme containing five broad classes classified according to their corresponding linguistic level, including orthographic, morphological, lexical, semantic, and syntactic errors Orthography is a set of conventions, including rules and standards for spelling, punctuation, or capitalization Morphological errors exist when there is a mistranslation in transferring the forms of verbs, nouns, or any other part of speech, including adjectives and adverbs, between
SL and TL Lexical errors relate to choosing inappropriate lexical Semantic errors occur when words are used differently in the source and target languages The meaning of the words in the target text may be inconsistent with that in the source text Syntactic errors include errors in conjunction, prepositions, articles, syntactic element reordering, and category errors
The taxonomy of translation errors offered by Flanagan (1994) and Farrús et al (2010) mainly emphasized grammatical problems, whereas not all grammatical error types make sense in every language
Doyle (2003, pp 22–23), Koby (2013), and (Phelan 2017) use ATA's framework for standard error making This framework is considered a ready- made, standardized, time-tested, and professionally recognized model to conduct theory-based, systematic, coherent, and consistent evaluations This framework suggested by the American Translation Association is intended for standard error marking and explanation of work done by professional translators Doyle (2003, pp 22–23), Koby (2013), and (Phelan 2017) listed the 22 types of errors that are used as the standard for error marking and grading consisting: "1) Incomplete passage, 2) Illegible handwriting, 3) Misunderstanding of the original text, 4) Mistranslation into the target language, 5) Addition or omission, 6) Terminology, word choice, 7) Register,
8) Too freely translated, 9) Too literal, word-for-word translation, 10) False cognate, 11) Indecision in word choice, 12) Inconsistent, 13) Ambiguity, 14) Grammar, 15) Syntax, 16) Punctuation, 17) Spelling, 18) Accents and other diacritical marks, 19) Case (upper case/lower case), 20) Word form, 21) Usage and 22) Style."
This list tries to include various types of possible errors coming up from the error corpus of translators as detailed as possible
The above classification was employed to collect and analyze the translation errors committed by Google Translate in the novel due to the following reasons First of all, this list tries to include all different types of possible mistakes coming up from the error corpus of translators It focuses both on sentence-level errors and text-level errors Secondly, this list is specific enough for the researcher to categorize the errors correctly and quickly
Table 2.2 ATA's Framework for Standardized Error Making (Doyle 2003, pp 22–23)
Code Criteria and description of each error
1 Incomplete Passage: A substantially unfinished passage is not graded Missing titles, headings, or sentences within a selection may be marked as one or more errors of omission, depending on how much is omitted
Previous study
The researcher reviews some previous studies of some authors in the world and Vietnam in this section
Medvedev (2016) investigated GT's efficiency in teaching English in
"Journal of Teaching English for Specific and Academic Purposes" She found that (1) GT can handle translations extremely fast (2) With coherence term,
GT successfully translates words but fails in sentences due to the wrong choice of words that make the sentences misunderstood (3) The accuracy of grammar is often lost with GT's translation She concluded that GT could convert individual words or statistically significant patterns with a higher level of accuracy She also suggested utilizing this tool in and out of the classroom's language
Longears (2015) wrote "A Comparative Analysis of the Accuracy in Google's Translations" P Longears used qualitative analysis to analyze the differences and errors in content consisting of pragmatic differences, semantic differences, pragmatic errors, and semantic errors Qualitative analysis was also used to analyze differences and errors in the form, consisting of grammatical differences, orthographic differences, grammatical errors, and orthographic errors He used quantitative analysis to make error statistics Lexical and grammatical errors are the most common ones made by this GT The findings pointed out that although GT presented several advantages such as speedy translation, various languages, and a free tool, it did not always provide precise translations because GT can not achieve the agreement in the ST in terms of content, context and form
At Quy Nhon University in 2019, Giau (2019) researched the accuracy of GT in her Master's thesis The study had the contribution of comparing the accuracy of GT when dealing with literary and technical texts
After Giau, Thu (2020) and My (2020) conducted the theses of evaluating GT's quality With the research "The quality of GT's Vietnamese translations of English idioms with words denoting time," Thu (2020) concluded that GT caused several errors in functional elements, communicative purpose, target language, and cultural background In the thesis "The quality of GT's translations of metaphors in "Kafka on the Shore" by Murakami Haruki into Vietnamese”, My (2020) found that even when GT committed a word-by- word mechanism for translation, it still kept the original meaning the author wanted to convey
From the studies reviewed above, the researcher found that GT is only successful in word-for-word translations but unsuccessful in sentences translations The researcher carried out this thesis with the hope of pointing out the weaknesses that GT suffers when converting phrases and sentences of metaphors The study is hoped to provide GT developers with more data to perfect GT more effectively
This chapter has addressed the basic theoretical backgrounds related to the researcher's study In the following chapter, the researcher will describe the research methods, collecting and analyzing the data for the analysis.
RESEARCH METHODS
Research methods
In this research, the quantitative and qualitative research approaches are used as mixed methods
Deductive approaches to the scientific method describe quantitative research, which aims to confirm, disproving, or give credence to current hypotheses This study's methodology entails identifying trends, associations, and causal interactions by calculating variables and evaluating their relationships Researchers can use linear data collection and analysis methods to generate statistical data Quantitative research underlies principles that include "neutrality, objectivity and the acquisition of a sizeable scope of knowledge" (e.g., a statistical overview from a large sample) When the primary goal is to clarify or test Leavy, this technique is usually acceptable (Leavy (2017, p.9))
A qualitative study is a non-numerical data processing technique involving various data analysis techniques, focusing on assessing the language, signals, and context constructions Qualitative analysis methods are holistic and contextual to provide a broader and more detailed viewpoint based on the participants' behaviour, thoughts, and perceptions Thus the qualitative scientist receives rich information that emphasizes various groups of people (Oko,
According to Leavy (2017, p 9), qualitative analysis is typically distinguished by inductive approaches to information construction to create meaning Using this method, researchers could explore, research, and learn from the social phenomenon; moreover, uncover the meanings people attribute to activity, circumstances, incidents, or devices; or develop a deeper understanding of some aspect of social life The relevance of people's perceptions and meaning-making methods and gaining a depth of comprehension are concepts that underpin qualitative study Qualitative analysis is usually suitable for exploring, describing, or explaining
The difference between qualitative and quantitative analysis is that quantitative approaches are deductive meanwhile qualitative methods are inductive, according to VanderStoep & Johnston (2009, p 168) A deductive method is a line of reasoning that runs from a theory/hypothesis to systematic empirical observation and concludes An inductive approach moves from specific adherence to theory, view, and interpretation Vander Stoep and Johnston further explain the kinds of data suitable for a qualitative or quantitative study, how the investigation can be analyzed, and the pros and cons of each one
Table 3.1 Quantitative versus Qualitative Research (VanderStoep & Johnston 2009, p.7)
Characteristic Quantitative Research Qualitative Research
Type of data Phenomena are described numerically
Phenomena are described in a narrative fashion Analysis Descriptive and inferential
Characteristic Quantitative Research Qualitative Research
Large sample, statistical validity, accurately reflects the population
Rich, in-depth, narrative description of sample
Superficial understanding of participants' thoughts and feelings
Small sample, not generalizable to the population at large
Weber (1990, p 9) defines content analysis as a research method that uses procedures to make valid inferences from the text Neuendorf (2011, p
19) describes content analysis as "the organized, objective, quantitative analysis of message characteristics." It involves human-coded interpretation as well as computer-assisted analysis of language.
According to Neuendorf (2011, p 19), content analysis can be characterized as the systematic, objective, quantitative analysis of message characteristics Both human-coded and computer-assisted text processing are used Content analysis method's applications include the meticulous study of face-to-face human interactions; analyzing character portrayals in media sites ranging from novels to online videos; analyzing computerized word use in news, political discourses, ads, and blogs; examining digital material such as video games and media interactions; and so on
In this study, both quantitative and qualitative content analysis procedures are used at the same time Quantitative content analysis is the tool that helps the researcher to figure out the error numbers when GT translates metaphors and the types of errors that appear the most frequently On the other hand, qualitative content analysis is used to analyze specific examples From this, the researcher can point out what errors GT makes when translating metaphors and assess the quality of GT
According to Neuendorf (2011, p 44), the purpose of any quantitative study is to generate counts of critical categories and measures of the quantities of other variables The quantitative content analysis aims to provide a numerically validated summary of a specific message package It is neither a gestalt impression nor a comprehensive overview of a message or group of notes
Schreier (2012) reveals that qualitative content analysis (QCA) is how qualitative materials are defined systematically, which is achieved with your content classified as instances in the coding frame divisions The approach is ideal for any content, verbal/ visual or sample, which must be interpreted to some extent As your research query states, QCA needs you to concentrate on some facets of your content, which varies from other qualitative data analysis methods QCA is a data reduction method that is systematic and flexible The approach is rigorous in three ways: all related data is considered; a series of steps is observed throughout the study; the coding must be checked for reliability warrant
It's adaptable in the sense that your coding frame must be valid to your content It, therefore, decreases the content by restricting the analysis to the subject's specific aspects Furthermore, classifying particular data as an instance of a category subsumes the data under a more general definition Simultaneously, categorizing generates new data on how the cases are compared
According to Shkedi (2019), qualitative analysis is described as a study tool using language The study examines literal evidence and the thought process rooted in language and the words as literal explanations of the study
In short, Dawson (2002) proposes the methodology for researchers by simply clarifying that the best approach for this study is quantitative research if we study for how many, evaluate, verify, how much, or how happy
Qualitative analysis is required for investigations into discovery, motivation, experience, thinking, problems or behaviour
Thanks to Dawson's suggestions, the researcher affirms that quantitative and qualitative content analyses are appropriate for the study The researcher needs to solve the questions above by figuring out the number of metaphors errors and assessing the translation quality of GT when it translates those metaphors.
Data Collection
This study aims to evaluate the quality of GT by investigating its errors in the Vietnamese translations of the English metaphors in the novel "The Count of Monte Cristo."
The researcher tried to answer the following questions:
1 What metaphors are there in the novel "The Count of Monte Cristo" by Alexandre Dumas?
2 What types of errors does GT make in translating these English metaphors into Vietnamese?
3 How well does GT translate these English metaphors into Vietnamese in terms of translation errors?
In order to answer those questions, the researcher followed the following steps:
In the first step, the researcher skimmed through the novel "The Count of Monte Cristo" by Alexandre Dumas and got an overview of its content Mai The Sang's Vietnamese book is an abridged translation Therefore, many of the metaphors in the original novel “The Count of Monte Cristo” are not found in
Mai The Sang's translation The researcher did not take out metaphors that were not translated by Mai The Sang due to the need for a reliable database to compare with GT's translation to identify errors
In the next step, the researcher read sentence by sentence carefully to identify the sentences containing metaphors in the original novel, then highlighted those sentences if they had Vietnamese equivalences in Mai The Sang's book The sentences containing Mai The Sang’s Vietnamese translations of those metaphors were stressed, too After highlighting as many metaphors as possible, the researcher randomly extracted 120 metaphors to analyze
The product of the above procedures is shown in Table 3.2
The product is a table of four columns Column (1) shows the order of data Column (2) contains 120 English sentences with bolded and underlined metaphors taken out from the novel In column (3), the type of metaphors in the relative sentence was identified and presented Column (4) shows the Vietnamese equivalences of Mai The Sang for those sentences For further analysis, each kind of metaphor is coded as follows:
The researcher stipulated that the bolded and underlined parts are the metaphors in the English sentences The table is more clearly presented as follows:
Table 3.2 English metaphors and Mai The Sang's translations
No English sentences containing metaphors
Vietnamese sentences translated by Mai The Sang
DI Napoleon là một thánh nhân ở phương Tây
2 Villefort leaned on the back of an armchair
DE Villefort phải tựa người vào lưng ghế.
Data Analysis
After collecting the data and completing the above table, the researcher carried out data analysis The qualitative approach describes and analyses the data to find GT's translation errors in translating the English metaphors into Vietnamese; meanwhile, the quantitative approach counts the frequency of each type's mistakes The data analysis process was carried out in the following steps
Step 1: After English sentences containing metaphors translated by Mai The Sang were copied and pasted into GT to take out their Vietnamese translations, the researcher compared GT's translations with the translations of Mai The Sang to identify error translations If they have the same meaning, there is no error translation In contrast, the type of error translations is determined when they have differences in meaning, syntax, punctuation, grammar, word form, etc
Step 2: The errors in GT's translations were classified and assigned codes according to 22 types of errors, one number for a kind of error and one code for no error (23)
Step 3: The researcher counted the number of errors that appeared in each metaphorical expression One metaphor’s translation might contain more than one mistake, so the number of errors was more significant than the number of metaphors
Step 4: The researcher discussed the number of translation errors and assessed the GT's quality in translating the English metaphors in terms of errors types and metaphors
Two more columns were added to Table 3.3 Column (5) shows the Vietnamese sentences translated by GT Column (6) lists the types of errors the researcher analyzed to classify in detail after comparing them with Mai The Sang's translations in column (3) The product of the analysis of the translation errors was shown in Table 3.3
Table 3.3 Google Translate's translation errors
Vietnamese sentences translated by Mai The Sang
Vietnamese sentences translated by Google Translate
1 Napoleon is the Mahomet of the West
DI Napoleon là một thánh nhân ở phương Tây
2 Villefort leaned on the back of an armchair.
DE Villefort phải tựa người vào lưng ghế
Villefort dựa vào lưng ghế bành
Next, the researcher counted the occurrences of the translation errors Accordingly, the researcher listed twenty-two types of errors using number codes as mentioned in the above section If there was no translation error, its code was 23 The number and frequency were put in Table 3.4 for further analysis and comparison
Table 3.4 The number and percentage of translation errors made by GT
Direct metaphors Implied metaphors Dead metaphors
Number Percentage Number Percentage Number Percentage
The results gave a clear picture of the translation errors committed by GT From the results of data analysis, the researcher could have an overall evaluation of the quality of GT in translating English metaphors into Vietnamese
In this chapter, the researcher presented the process of collecting and analyzing the data and the products of those processes The results of the data analysis are shown in the next chapter.
FINDINGS AND DISCUSSION
Types of metaphors
After identifying 120 metaphorical expressions in "The Count of Monte Cristo" novel by Alexandre Dumas, the researcher found three types of metaphors according to Johnson's (2019) classification They are direct metaphors, implied metaphors, and dead metaphors
The table below shows the distribution of the metaphors in the novel in terms of types
Table 4.1 The number and frequency of metaphors
Among those 120 metaphors, implied metaphors make up the highest percentage of 50% with 60 items Dead metaphors rank second with 41 expressions (34.2%) Direct metaphors stand at the third position with 19 items, accounting for 15.8% In contrast, Extended metaphors and Mixed metaphors don't appear in these 120 metaphors Each type of metaphor is analyzed in detail in the following sections
"Your Dictionary" defines that implied metaphors occur when the writer wants to compare two things with common characteristics without directly mentioning one of them With implied metaphors, writers can create vivid imagery in their writings
Some examples of this kind of metaphor from the novel are listed below:
(1)On the outside of which, the sun had stamped that beautiful colour of the dead leaf peculiar to the buildings of the country (p 17)
(2) Grief had made her blind to all (p 73)
In example (1), the verb "stamped" is the implied metaphor that compares the sun to a man who can paste the colours to the buildings or dye colours By using this comparison, the sun becomes more vivid with the action of dying colours
In example (2), "made someone blind" doesn't have the literary meaning of making someone's eyes blind The expression helps the readers realize the deep sorrow of the young girl, which paralyzes her senses
(3) But no voice broke the stillness of the chamber (p 73)
In example (3), "no voice broke the stillness" makes the reader think about the silence without any comparison
With implied metaphors, comparisons are not mentioned; however, the readers still get vivid pictures
Dead metaphors are those metaphors that are used so often that they lose their metaphorical images Gradually, people use them like ordinary words (Newmark, 1998) Dead metaphors have been used so regularly that they can lose their connection to the original imagery Their symbolic meanings can be understood without knowing the earlier connotations
Some examples are extracted from the novel belonging to this kind of metaphor
(4) Villefort leaned on the back of an armchair (p 86)
(5) Edmond's hair stood on end (p 117)
(6) At the end of the hour, Edmond was awakened by the roar of the thunder (p 177)
In examples (4) and (6), "the back of an armchair" and "the roar of the thunder" are common expressions They are so common that people do not recognize them as metaphors anymore
For example (5), "One's hair stood on end." The meaning of causing fright or terror to someone is an idiom that people also use widely and frequently
Dead metaphors are also known as frozen metaphors and historical metaphors
A direct metaphor occurs when two things or people are compared directly It is a comparison where the link can be easily made and understood
(7) Napoleon is the Mahomet of the West (p 46)
In examples (7), Napoleon is compared to "the Mahomet of the West."
In example (8), Dantes is compared to "number 34"
In the above instances, we know immediately what the writer meant even though it is illogical to compare Napoleon to a Mahomet or Dantes to a number
In this section, the researcher addressed three types of metaphors in the
120 metaphors In the next part, the researcher focuses on the types and frequency of translation errors committed by GT by using the ATA's framework for standard making.
Translation error types in GT's Vietnamese translations
In section 4.1, the researcher has addressed three types of metaphors identified in the novel The researcher will focus on GT's types and frequency of translation errors when translating these metaphors in the following sections
This study employed the framework ATA of Doyle (2003) to identify the errors of GT's Vietnamese translations One hundred twenty sentences consisting of 120 metaphorical expressions in "The Count of Monte Cristo" by Alexandre Dumas were analyzed in translation errors The researcher found out the Vietnamese translation errors committed by GT in translating those metaphorical expressions compared to those of Mai The Sang As this study focuses solely on the translation quality of metaphors, errors related to other linguistic elements will not be mentioned There is also one section for accurate translations
Table 4.2 shows the results of the analysis
Table 4.2 Frequencies of translation errors and no errors in GT’s Vietnamese translations
The analysis showed that GT translated only 21 metaphorical expressions precisely, accounting for 17.5%; meanwhile, there were 99 metaphorical expressions translated with errors, making up 82.5% of the total errors
The number of occurrences of no translation errors in each type of metaphor is shown in Table 4.3
Table 4.3 Frequencies of no errors in each type of metaphor in GT's Vietnamese translations
In the 18 cases above, GT translated ten dead metaphors correctly, accounting for 47.62% The examples illustrate this case:
(9) Villefort leaned on the back of an armchair (p 86)
GT’s Translation: Villefort dựa vào lưng ghế bành
In example (9), "the back of an armchair" is a dead metaphor The word
"back" is defined as the human body part on the opposite side of the chest (Oxford Dictionary) The metaphor "the back of an armchair" and "lưng ghế" have the same meanings This expression is so familiar to English and Vietnamese people that GT has no difficulty translating it exactly
(10) And raising the foot of the bed he drew out the vial, still, a third filled with the liquor (p 164)
GT’s Translation: Và nâng chân giường lên, anh rút cái lọ ra, vẫn còn một phần ba đầy rượu
"the foot of the bed" is a part of the bed The dead metaphor "the foot of the bed" was converted to "chân giường" by GT correctly
(11) The two young peasants were on the edge of the forest (p 311)
GT’s Translation: Đôi bạn trẻ gặp nhau ở bìa rừng
According to Oxford Learners Dictionary's definition, "the edge" is the outside limit of an object, area, or surface or a place farthest away from the centre of something That GT converted the dead metaphor "the edge of the forest" into "bìa rừng" is correct
Some examples of implied metaphors converted exactly are shown below:
(12) At that instant, a violent clap of thunder seemed to shake the house to its very foundation (p 449)
GT’s Translation: Ngay lúc đó, một tiếng sấm sét dữ dội dường như làm rung chuyển ngôi nhà đến tận cùng nền móng của nó
In this case, the thunder is compared to a human who can move from side to side or up and down the house with short quick movements The verb
"shake" is an implied metaphor that was converted into "làm rung chuyển" by
GT, and this equivalence is appropriate
(13) His voice went to my heart, observed Jullie, and two or three times,
I fancied I had heard it before (p 505)
GT’s Translation: Giọng anh ấy đi vào tim tôi, Jullie quan sát; và hai hoặc ba lần tôi tưởng tượng rằng tôi đã từng nghe nó trước đây
The action "go" with the meaning of "move or travel to a particular place belongs to humans or animals In example 13, "went" is an implied metaphor whose meaning is equivalent to "đi" in Vietnamese GT gave a correct translation in this case
With direct metaphor, GT committed a low percentage of accuracy of 19.05 % in 21 cases
GT’s Translation: Tôi là Số 27
In this example, Dantes is compared directly to a number "Number 27" is a direct metaphor, and the Vietnamese equivalence "Số 27" is accurate
(15) I am a machine and nothing else (p 614)
GT’s Translation: Tôi là một cái máy chứ không phải gì khác
In example (15), Dantes’ old roommate is compared to a machine "a machine" is a direct metaphor with its exact Vietnamese equivalence "một cái máy"
This part gives the results and examples for cases where no translation error was found in GT's Vietnamese translations The following section reports the findings on the translation errors committed by GT
There were 153 errors in 99 metaphorical expressions In some cases,
GT caused more than one error in a translation; as a result, the number of errors is greater than that of metaphors
The frequencies of translation errors are presented in the table below
Table 4.4 Frequencies of translation errors in GT’s Vietnamese translations
No Error type Number Percentage
1 Too Literal, Word-for-word 85 55.56
No Error type Number Percentage
22 Accents and Others Diacritical Marks 0 0
The result of data analysis indicates that nine types of translation errors were committed by GT, including Word-for-word, Mistranslation into Target Language, Usage, Style, Ambiguity, Grammar, Syntax, Punctuation, and Case
It can be seen that the Word-for-word errors make up the highest percentage of 55.56% Ranked second in terms of frequency is Mistranslation into Target Language errors, which account for 22.87% Usage, Style, and Ambiguity errors contribute 4.58% for each type On the contrary, Grammar, Syntax, Case, and Punctuation errors have low occurrences ( just 7.84% for these six types of errors) The results obtained from the errors analysis showed that GT did not commit 13 types of translation errors: Word Form, Register, Incomplete Passage, Illegible, Misunderstanding of Original Text, Terminology, Too Freely Translated, False Cognate, Indecision, Spelling, and Accents and Others Diacritical Marks
4.2.2.1 Too Literal, Word-for-word
Newmark (1998) defined Word-for-word Translation as interlinear translation The TL text is put immediately below the SL words and the SL word order is kept the same, and the words are translated singly by their most common meanings out of context In ATA’s (2003) framework, word-for-word translation is a translation error in which translation following the source text exactly causes awkward, often incorrect renditions
In the investigated novel, word-for-word errors accounted for the highest proportion Out of 153 errors identified, there were 85 instances of this type of error, which means that more than half of the errors were related to word-for- word translation errors
As a result of the analysis, the number of occurrences of this type of error is presented below:
Table 4.5 Frequencies of Word-for-word errors in each type of metaphors
Too Literal, Word-for-word errors
As seen from the table, there is a great difference in the number of errors’ occurrences in each type of metaphor Implied metaphors have the highest number of word-for-word errors (56.48%) Dead metaphors are ranked as the second group, with 29.41% Direct metaphors have a word-for-word error proportion of 14.11%
Following are some examples of word-for-word translation errors in the Vietnamese versions committed by GT
GT committed 48 word-for-word errors accounting for 56.48% among
85 errors of this type Some examples are analyzed to justify this type of errors
(16) He, therefore, informed M.Morrel of his wish to quit the sea and obtained a recommendation from him to a Spanish merchant, into whose service he entered at the end of March (p 99)
GT’s Translation: Do đó, anh ta đã thông báo cho M.Morrel về mong muốn rời bỏ biển của mình, và nhận được lời giới thiệu từ anh ta với một thương gia Tây
Ban Nha, người mà anh ta đã tham gia dịch vụ vào cuối tháng Ba
In this example, the expression “quit the sea” is an implied metaphor whose literal meaning is “leave the sea.” In the GT’s translation, this expression was converted into “rời bỏ biển.” This translation is word-for-word meaning
In this situation, “quit the sea” means “quit the job involving the sea.”
The phrase “thôi việc” is the most appropriate equivalence that Mai The Sang translated in his book
Mai The Sang’s translation: Hắn xin ông Morel cho thôi việc
(17) He was thinking of the coming scene between himself and the baroness, whose threatening looks and frown-ing brow, like that of Olympic Jove, predicted a fearful storm (p 474)
GT’s Translation: Anh đang nghĩ đến cảnh sắp tới giữa mình và nam tước, người có vẻ ngoài đe dọa và đôi mày cau có, giống như của Olympic Jove, dự báo một cơn bão đáng sợ
In example 17, “a storm” is used to imply something terrible to happen, and
“a fearful storm” is an implied metaphor According to Oxford Dictionary, a storm is awful weather with strong winds and rain, often thunder and lightning The expression “a fearful storm” was translated into “một cơn bão đáng sợ” by GT instead of “một tấm thảm kịch” as Mai The Sang’s translation
GT translated this metaphor so literally that it could not convey precisely the author's meaning through the metaphor This error belongs to the word-for- word type
Mai The Sang’s translation: Danglars không nói gì Hắn thấy trước rằng sắp xảy ra một tấm thảm kịch đây
There are 25 word-for-word errors committed when GT translated those dead metaphors, which occupied 29.41% of this error type
(18) He is no longer in my hands (p 72)
GT’s Translation: Anh ấy không còn trong tay tôi nữa