Xorijiy lingvistika va lingvodidaktika –
Зарубежная лингвистика и
лингводидактика – Foreign
Linguistics and Linguodidactics
Journal home page:
https://inscience.uz/index.php/foreign-linguistics
Linguistic and functional peculiarities of human and
machine translation (on the example of the English
translation of Gafur Gulom’s “Shum bola”)
Mokhinur KAYIMOVA
1
Samarkand State Institute of Foreign Languages
ARTICLE INFO
ABSTRACT
Article history:
Received January 2025
Received in revised form
10
February 2025
Accepted 25 February 2025
Available online
25 March 2025
The translation industry is considered an essential part of
global communication and information exchange. In recent
years, it has undergone remarkable changes due to the
development of both machines and translators. This article
conducts a comprehensive analysis of the strengths and
limitations of both methods, emphasizing their effectiveness in
maintaining context, interpreting idiomatic expressions, and
conveying cultural features – key factors in ensuring accurate
and meaningful translation. The study also aims to present a
comparative analysis of the strengths of human translation
compared to machine translation in the English translation of
Gafur Gulom’s “Shum bola (Naughty boy)”.
2181-3701/© 2024 in Science LLC.
DOI:
https://doi.org/10.47689/2181-3701-vol3-iss3
This is an open-access article under the Attribution 4.0 International
(CC BY 4.0) license (
https://creativecommons.org/licenses/by/4.0/deed.ru
Keywords:
machine translation,
CAT,
MAHT,
HAMT,
human translation,
cultural terms,
ambiguity,
Shum bola,
choyxona,
samovarchi,
moshkichiri.
Inson va mashina tarjimasining lingvistik va funksional
xususiyatlari (G‘afur G‘ulomning “Shum bola”ning
inglizcha tarjimasi misolida)
ANNOTATSIYA
Kalit so‘zlar:
mashina tarjimasi,
CAT,
MAHT,
HAMT,
inson tarjimasi,
madaniy atamalar,
noaniqlik,
Tarjima sanoati global aloqa va axborot almashinuvining
muhim qismi hisoblanadi. So‘nggi yillarda u mashinalar va
tarjimonlarning rivojlanishi tufayli sezilarli o‘zgarishlarni
boshdan kechirdi. Ushbu maqola ikkala usulning kuchli
tomonlari va cheklovlarini har tomonlama tahlil qilib, aniq va
mazmunli tarjimani ta’minlashning asosiy omillari sanalgan,
1
Student, Samarkand State Institute of Foreign Languages. E-mail: mohinurqayimova7@gmail.com
Xorijiy lingvistika va lingvodidaktika – Зарубежная лингвистика
и лингводидактика – Foreign Linguistics and Linguodidactics
Special Issue – 3 (2025) / ISSN 2181-3701
304
Shumbola,
choyxona,
samovarchi,
moshkichiri.
kontekstni saqlash, idiomatik iboralarni talqin qilish va
madaniy birikmalar ma’nosini yetkazishdagi samaradorligini
ta’kidlaydi. Tadqiqot, shuningdek, G‘afur G‘ulom “Shumbola”
asarining ingliz tilidagi tarjimasida avtomatik tarjimadan farqli
o‘laroq, inson tarjimasining kuchli tomonlarini qiyosiy tahlilini
taqdim etishga qaratilgan.
Лингвистические и функциональные особенности
человеческого и машинного перевода (на примере
перевода на английский язык романа Гафура Гулома
«Шум бола»)
АННОТАЦИЯ
Ключевые слова:
машинный перевод,
CAT,
МАHT,
HAMT,
человеческий перевод,
культурные термины,
двусмысленность,
Шумбола,
чойхона,
самоварчи,
мошкичири.
Индустрия переводов рассматривается как неотъемлемая
часть глобальной коммуникации и обмена информацией. За
последние годы она претерпела значительные изменения в
связи с развитием как машинного, так и письменного
перевода. В статье проводится всесторонний анализ сильных
и слабых сторон обоих методов, подчеркивается их
эффективность в сохранении контекста, интерпретации
идиоматических выражений и передаче культурных нюансов
–
ключевых
факторов
обеспечения
точного
и
содержательного перевода. Цель исследования также состоит
в том, чтобы представить сравнительный анализ
преимуществ человеческого перевода в сравнении с
автоматическим
переводом
на
английский
язык
произведения Гафура Гулома «Шумбола».
INTRODUCTION
Since the 1940s, when the concept of using computers for language translation
first emerged, and throughout the early research efforts of the 1950s, translators have
responded with a mix of skepticism and concern. Some have outright rejected the idea,
convinced that no machine could ever replicate the complexities of human translation.
Others, on the opposite end of the spectrum, have worried that advancements in
automation might eventually replace their profession altogether. So, nowadays, some
argue that both human and machine translation seek to address the same challenge.
However, this claim is questionable, as translation is not a singular process but a diverse
and multifaceted activity. It encompasses a wide range of genres, objectives, and
contexts, making it unlikely that a single approach can fully capture its complexity.
METHODOLOGY AND LITERAL REVIEW
Machine translation
relies on processes of analysis and synthesis, requiring
extensive effort and years of perseverance, often accompanied by setbacks. In the 1960s,
the results produced by machine translation were sometimes so absurd that the field
experienced a period of decline. However, the development of third-generation computer
systems in the 1970s reignited interest in this technology. The introduction of word
processors further revolutionized the profession, becoming an indispensable tool for
Xorijiy lingvistika va lingvodidaktika – Зарубежная лингвистика
и лингводидактика – Foreign Linguistics and Linguodidactics
Special Issue – 3 (2025) / ISSN 2181-3701
305
modern translators. The rise in international communication has led to a significant
growth in the use of automated machine translation in recent years. Machine translation
is a branch of computational linguistics that focuses on developing methods to automate
the translation process. The main goal has always been to preserve the meaning of the
original text in its translated version.
Machine translation is also called computer-aided translation (CAT).
Computer-assisted Translation (CAT)
– “a form of translation that makes use of a
software program supporting and facilitating the translation process” (TAUS, 2017). It is
sometimes called machine-assisted or machine-aided translation. CAT tools are groups of
computer programs whose main goal is to improve translators’ accuracy and
productivity thanks to several internal and external resources, e.g., editors, glossaries,
translation memories (TM), machine translation modules, etc., all of which are integrated
into the same translation environment [3, 104]. There is a broader range of positive sides
of CAT tools that are considered by many translators and translation agencies. They are
given on the list below:
a) Maintaining Consistency
– CAT tools ensure that translators use the same
terminology, abbreviations, and product names consistently across a single project or
multiple assignments.
b) Boosting Productivity
– With the help of translation memory (TM), CAT tools
allow translators to work efficiently by reusing previously translated content based on
predefined match percentages. It speeds up the process and helps meet even the tightest
deadlines.
c) Quality Control
– Many CAT tools come equipped with built-in quality assurance
features, such as automatic spell-checking, grammar error detection, highlighting of
inconsistencies in numbers, and missing tags. These assist translators in improving
accuracy and reducing errors.
d)
Concordance Search
– This functionality enables translators (or clients) to
search within a translation memory (TM) for specific words or phrases, helping to find
relevant translations in the appropriate context.
e) Reducing Costs
– As translators work with CAT tools, their translation memory
expands, leading to a higher frequency of repeated content. This not only speeds up the
translation process but also maintains quality, allowing translators to take on more work
in less time, ultimately increasing their earnings.
However, it also has some drawbacks while a translator is using it:
a) Compatibility Limitations
– CAT tools are not universally compatible with one
another. Because of this, translation agencies often prefer to work with translators who
use the same software as they do, potentially excluding skilled professionals solely based
on their tool of choice rather than their expertise.
b) Error Repetition
– Since CAT tools rely on translation memories (TMs) to recycle
previously translated content – whether from the same translator or others – there is
a risk of propagating mistakes already stored in the system. If an error is not corrected,
it may continue appearing in future translations.
c) Lower Pay Rates
– While CAT tools enhance efficiency, some clients take
advantage of the automation aspect by lowering rates for certain types of matches within
a TM. As a result, translators may end up doing more work for reduced compensation,
ultimately earning the same amount as they would without a CAT tool.
Xorijiy lingvistika va lingvodidaktika – Зарубежная лингвистика
и лингводидактика – Foreign Linguistics and Linguodidactics
Special Issue – 3 (2025) / ISSN 2181-3701
306
CAT systems are divided into two groups: machine-aided human translation
(MAHT) and human-aided machine translation (HAMT). The difference between the two
lies in the roles of the computer and human translator [9, 15].
In
MAHT
, a translator makes the translation, then uses the computer as a tool for
typing, checking spelling, grammar, style; for printing the target text, for looking up
words in electronic dictionaries and data bases, for getting references on CD-ROMs and
other sources, for consulting about contexts, for discussing problems in the web, for
searching a job, etc.
In
HAMT,
the translation is automated, done by a computer, but requiring the
assistance of a human editor. There are two phases of human help: pre-editing and post-
editing. In pre-editing, an operator (or a customer) prepares the text for input. A special
computer translation program transfers the text from one language to another.
Then a translator does the post-editing, mostly by correcting the word usage.
Different methods are used in machine translation:
1.
Dictionary-Based Machine Translation.
This translation method relies on the
entries found in a language dictionary, where each word's equivalent is used to generate
the translated text. The earliest generation of machine translation, spanning from the late
1940s to the mid-1960s, was entirely dependent on electronic or machine-readable
dictionaries. While this approach remains somewhat useful for translating individual
words and short phrases, it is less effective for complete sentences. Later translation
models have incorporated bilingual dictionaries alongside grammatical rules to enhance
accuracy and coherence.
2.
Rule-Based Machine Translation (RBMT).
A rule-based approach to MT is one
of the very first strategies ever developed. “More complex than translating word to word,
these systems develop linguistic rules that allow words to be put in different places, to
have different meanings depending on context, etc.” (Costa-Jussà, Farrús, Mariño, and
Fonollosa, 2012).
Apart from rules (grammatical, lexical, and stylistic), a standard RBMT system
includes software used to process these rules and a significant number of bilingual
dictionaries for each language pair. It can grow, and it is easy to maintain. Its main
advantage is the fact that it allows for deep syntactic and semantic analysis of a div of
text, but it requires significant knowledge and a great number of rules. It is a huge
drawback. An RBMT system, to provide acceptable results, requires a lot of time and
linguistic resources to build [3, 97].
3.
Corpus-based machine translation (CBMT).
CBMT is a corpus-based machine
translation approach that does not rely on predefined rules or parallel corpora. Instead, it
utilizes a vast monolingual target-language corpus, a comprehensive full-form bilingual
dictionary, and optionally, a smaller monolingual source-language corpus to enhance
translation quality.
This method offers several advantages: it can be applied to almost any language
pair, effectively maintains context in translated text, and processes longer phrases more
efficiently than many other approaches. Additionally, CBMT is adept at resolving word
ambiguities and can generate alternative phrasings when a direct equivalent is
unavailable in the target language. Furthermore, it assesses the reliability of translated
segments, categorizing them as high or low confidence, which optimizes post-editing
efforts by allowing editors to focus primarily on low-confidence segments, thereby saving
both time and resources.
Xorijiy lingvistika va lingvodidaktika – Зарубежная лингвистика
и лингводидактика – Foreign Linguistics and Linguodidactics
Special Issue – 3 (2025) / ISSN 2181-3701
307
Human translation
is influenced by the characteristics of source-target language
transfer, cultural context, and individual translators’ translation ability (Bassnett and
Lefevere 1992; Wong and Shen 1999). Human translation is carried out by skilled
professionals who may work in person or remotely, typically as native speakers of the
target language with expertise in the relevant subject matter. In professional contexts,
the benefits of human translation far outweigh any drawbacks, as human translators
ensure that context, a crucial aspect of effective communication, is accurately conveyed.
Unlike machine translation, which often relies on literal word-for-word
conversion, human translators interpret the intended meaning, taking into account
cultural nuances, linguistic subtleties, and brand identity when translating content such
as marketing materials or web pages. They also maintain consistency across languages,
ensuring that messaging remains clear and coherent.
The merely drawback of human translation is its cost and time investment, which
may not be justified for informal or low-stakes communications where only a general
understanding is needed.
However, in most cases, human translation remains the best option, as it
guarantees accuracy in grammar, slang, and tone, preserving the original text’s meaning
and intent.
RESULTS AND DISCUSSION
The integration of MT and HT has significantly impacted the field of linguistics and
translation, reshaping the way languages are translated. To evaluate the quality of
machine translation, it is necessary to compare the machine translation with the human
translation and the source language at a deeper and more comprehensive textual level.
Cultural and sociolinguistic factors
Language is closely connected to culture and sociolinguistic conventions, making
cultural sensitivity essential in translation. Effective translations must account for factors
such as formality, politeness levels, regional dialects, and cultural nuances to ensure
contextual appropriateness. Unlike human translators, machine translation systems lack
awareness of these elements, often failing to produce translations that align with the
cultural and sociolinguistic norms of the target language.
Let’s look at some examples for from the book
“Shumbola”, G‘afur G‘ulom:
Original text:
Xotin qozonning qopqog‘ini ochdi – baliqday bo‘lib, oppoq laganda
moshkichiri chiqdi, o‘rtaga qo‘ydilar. [1, 49].
Machine translation (Yandex translator):
The wife opened the lid of the
cauldron turned out like a fish, a white tray flywheel, medium, they put on.
Human translation:
The woman was cooking
moshkichiri
(Uzbek national dish –
lentil soup). The dish's smell began to open my appetite [2, 38].
In Uzbek cuisine, there is a traditional Uzbek dish called
“moshkichiri”
made
primarily from mung beans (mosh) and rice, creating a hearty and nutritious porridge-
like meal. it is usually translated with the way of clarification. As you see, machine
translation translates it as
“flywheel”
which is far from the meaning. However, when this
cultural word is translated by a translator, there is a clear explanation and it is
understandable for the reader.
Also, here is another example from this book:
Original text:
Qaymoq bozorining burilishida, mahkamaning boshida Ilhom
samovarchining kattakon choyxonasi bo‘lib, unda grammofon chalinadi [1, 3].
Xorijiy lingvistika va lingvodidaktika – Зарубежная лингвистика
и лингводидактика – Foreign Linguistics and Linguodidactics
Special Issue – 3 (2025) / ISSN 2181-3701
308
Machine translation (Yandex translator):
At the turn of the century, at the head
of the court, the inspiration is the large tea room of the samovar, in which the
gramophone is played.
Human translation:
At the corner of the sour-milk market, next to the office, there is
a big teahouse of llhom samovarmaker and a gramophone is playing there all the time [2, 5].
In this case, the Uzbek cultural word
“samovarchi”
comes as
“samovar”
in machine
translation. Actually, in the Uzbek language, a
samovarmaker (samovarsoz)
means a
skilled craftsman who specializes in making and maintaining samovars, the traditional
metal urns used for boiling and serving tea. Furthermore,
“choyxona” (tea house),
which
is a central part of Uzbek culture, serving as a traditional gathering place for socializing,
relaxing, and enjoying tea, is translated as
“tea room”
because of awareness of the
cultural term.
Hence, it is obvious from the examples that, despite the advancements of machine
translation, there are still gaps that MT cannot bridge. Instances requiring deep cultural
knowledge and sensitivity, such as literary translations, marketing materials, or legal
documents, highlight the indispensability of human translators. Their understanding of
the cultural background of both the source and target languages allows them to make
informed decisions that preserve the cultural essence of the original text [5, 33].
Accuracy of context
Languages frequently include ambiguous words or phrases that carry multiple
meanings depending on context. Accurately resolving such ambiguity is essential for
producing translations that are contextually appropriate.
However, machine translation systems often struggle with this task, as they lack
deep contextual understanding and may not fully consider surrounding words or
sentences when selecting the correct translation. Ambiguity can stem from homonyms,
polysemous words, idiomatic expressions, and cultural references, all of which require
nuanced interpretation to ensure accurate translation.
More sentences are given from the book
“Shumbola”, G‘afur G‘ulom
to highlight the
challenges:
Original text:
Bozorda sang‘ib yurgan biz daydi bolalar uchun
quvonchli ermaklardan biri bozor, mahalla, ko‘cha-ko‘y jinnilari edi [1, 5].
Machine translation (Yandex translator):
For the children of us, who are digging in
the market, one of the joyful ermak was the market, the neighborhood, the street, go crazy.
Human translation:
Very interesting amusement for the trampled boys like us in
the market were the mad men of the markets and streets [2, 6].
The example illustrates, the human translated text gives the clear understanding of
the meaning, while machine translation does not understand the context with
inappropriate translations, such as
“ermak” (amusement), “ko‘cha ko‘y jinnilari” (mad men
of the streets).
One more example:
Original text:
Biz yalangoyoq, bo‘z ko‘ylak-ishtonli, kir-chir bolalar to‘tiga
yaqinlashib [1, 5]:
Machine translation (Yandex translator):
We barefoot, boz shirt-appetizing,
dirty children approaching Toti:
Human translation:
Barefoot, dirty boys like me would come to listen to it and
say [2, 6]:
There is also ambiguity in context with the words
“to‘ti”, “yaqinlashmoq”, “kir-chir”,
which are translated correctly by the translator.
Xorijiy lingvistika va lingvodidaktika – Зарубежная лингвистика
и лингводидактика – Foreign Linguistics and Linguodidactics
Special Issue – 3 (2025) / ISSN 2181-3701
309
So, machine translation systems often struggle with the intricacies of context that
require beyond-text understanding. Human translators, with their ability to draw on
personal experiences, cultural knowledge, and contextual clues, excel in interpreting and
preserving the intended meaning of the original text. They can navigate through
ambiguities, implicit meanings, and cultural references that machine translation systems
might overlook or misinterpret [5, 34].
CONCLUSION
This study evaluated machine translation (MT) with a comparison of human
translation. The results indicated that both translation methods have numerous
advantages with a few drawbacks. However, despite the latest advancements of MT
mentioned above, it cannot provide a reader with an ambiguous cultural context of the
text. In these cases, human translators offer clearer, literal, and natural translations
rather than MT. Moreover, they play a vital role in the training and refinement of MT
systems. The feedback and corrections by translators enable MT algorithms to learn and
improve over time. This symbiotic relationship enhances the efficiency and accuracy of
MT, transforming it into a more effective tool that supports and enhances human
translators in their work.
REFERENCES:
1.
G‘ulom, G‘ofur. Shumbola, qissa/G‘.Gulom. -Toshkent: Yoshlar nashriyot uyi,
2018.- 192 b.
2.
Gulom, Gafur A naughty Boy. Gafur Gulom /Translators: I.M. Tukhtasinov, U.R.
Yoldoshev. – Tashkent: Yangi asr avlodi, 2017. – 128 pages.
3.
Kornacki M. Computer-assisted translation (CAT) tools in the translator training
process. – Berlin: Peter Lang, 2018. – С. 100-105.
4.
Naveen, Palanichamy, and Pavel Trojovský. "Overview and challenges of
machine translation for contextually appropriate translations."
Iscience
27.10 (2024).
5.
Newmark P. A textbook of translation. – 2003.
6.
Ochilov Ergash. Tarjima nazariyasi va amaliyoti (o’quv qo’llanma).-T.,2012.4b
7.
Oliver, Antoni. "Human translation and machine translation: Specificities, uses,
advantages and disadvantages."
Translating Diversity= Traduir la diversitat
(2020): 111-130.
8.
Translation Theory and Practice. Compiled by Dr. Farida Repelita Waty
Kembaren, M.Hum
9.
Zoya Proshina.Theory of translation (English and Russian). Vladivostok Far
Eastern University Press 2008.
10.
Zaripova, S., & Gafurova, K. (2024). FEATURES OF TEACHING ENGLISH
PHONETICS TO PRESCHOOL CHILDREN. TAMADDUN NURI JURNALI, 5(56), 289-292.
11.
Shaxnoza, Z. (2023). TEACHING FOREIGN LANGUAGES AT PRESCHOOLS USING
CONTEMPORARY METHODS. REGARDING FOREIGN APPROACHES. Innovations in
Technology and Science Education, 2(10), 49-53.
12.
Хайруллаев, Х. (2009). О лингвистической природе уровней языка и речи.
Вопросы филологических наук, (2), 78-79.
13.
Mamayunusovna, P. I. (2021). BOBURNOMA. AS A HISTORICAL AND LITERARY
SOURCE. Thematics Journal of Social Sciences, 7(5).
14.
Pardaeva, I. M. (2014). ON THE EXAMPLE OF THE ARTISTIC AND HISTORICAL
PROSE OF NAVOI IN TURKISH. The Way of Science, 73.
