Why Human Translation Is Better Than Machine Translation?
A common misapprehension among people in general, especially those who does not work in or are not acquainted with the translation and localization industry, that translation is nothing but a straightforward mechanical process involving a simple word-for-word correspondence between two languages that anyone proficient in two language pairs could easily do. To this end, we are bound to say that a literal one-to-one equivalence of a word in another language can only apply to machine translation, whereas human translation and the capabilities of the human brain are something totally different.
No doubt machine translation can save a lot of effort, time and money, but you cannot always guarantee the results. The functionality of machine translation is based on structure analysis of terms and phrases within the original file you are translating (source document), which in turn is broken down into translatable segments that are recomposed in the language you are translating into (target document).
Although substitutions are governed by other parameters, this translation model basically allows the machine translation engine to find and substitute the target-language phrasing that most commonly corresponds to the source-language phrasing in a series of texts.
Machine translation or the translation of a text by a computer, with no human involvement, was pioneered in the 1950s. It can also be referred to as automated translation, automatic or instant translation.
In fact, there are currently three types of machine translation systems:
- Rules-based systems: This system uses a combination of language and grammar rules. It also uses dictionaries for common words and creates specialist dictionaries to focus on specific industries or disciplines. If trained with specialist dictionaries, the rules-based systems typically deliver consistent translations with accurate terminology.
- Statistical systems: This system has no knowledge of language rules, but it learns to translate by analysing large amounts of data for each language pair. Using additional data relevant to the required sector, they can be trained for specific industries or disciplines. Typically statistical systems deliver more fluent-sounding but less consistent translations.
- Neural Machine Translation (NMT): It is a new approach currently used by SDL Machine Translation. It functions by making machines learn to translate through one large neural network (multiple processing devices modelled on the brain).
The approach has become increasingly popular amongst MT researchers and developers, as trained NMT systems have begun to show better translation performance in many language pairs compared to the phrase-based statistical approach.
Even though this method seems to be logical and sound, the translated text can only help in quick communications and general texts, but the translation quality is often poor, incoherent and incomprehensible. The reason is that the sentence structure of a language is totally different from that of another. When machine translation provides a literal substitute in the target language for every term in the source file, the text can be free of typos, but the structure, style and tone of the author are totally missing.
Due to the wide connotations of many terms, the meaning of a term can sometimes be misinterpreted, and the text can be full of grammatical errors. Besides, some files can comprise idioms, lacuna or lexical gaps that can be hardly translated into another language. This proves why an informed understanding of the source text is essential, something that machine translations cannot offer.
On the other hand, human translation is governed by other totally different parameters. Human translators are usually native speakers of the target language. They are familiar with all the language’s specificities, subtleties, expressions, idioms and slang; consequently, they can subtly convey the source file into their target language in all non-technical domains. In addition, they are mostly language graduates with translation degrees in specific areas of expertise, which enables them to provide a proficient translation in their selected technical domains.
Therefore, human translation or the human brain of a professional translator is backed by study and experience to which the translator resorts to when the need arises. He knows how the job is to be done, understands the Translation Paradigm and knows how to apply it. Putting his senses to good use, he is capable of learning, developing and updating his knowledge in his area of specialization. He constantly undergoes a non-stop developmental process of knowledge absorption that diversifies to expand his personal and professional experience.
Accordingly, a professional translator can solve problems and can make correct decisions regarding issues such as; ambiguity, the correct choice of the precise accurate word to be used among a long sequence of synonymous meanings to convey the exact meaning of the source document into the translated document.
He is also qualified to take the correct decisions regarding when to add or delete words to support and clearly convey the meaning, how to bridge the linguistic and cultural gaps between two languages and cultures, how to use the right collocations, how to use the right approach, style and tone of the author while observing the language conventions in order to finally create a translated document (target file) that is a mirror copy of the source document in meaning and format, convey the particular message of the author and conform to the translation ISO embodied in the IDEA, STYLE AND ORIGINAL COMPOSITION.
However, we are not against machine translation, but we are in favour of the natural capabilities of the human brain reinforced by state-of-the-art technological innovation as a mixture of both was found to produce the ultimate results. Throughout these two decades, many translators resorted to using various software packages on the market. Their translation supported by CAT tools, such as SDL Trados, MemoQ, Wordfast, Memsource, SmartCat and many others, always generates the best rendering.
What Is Computer-Assisted Translation? What Are CAT Tools?
A CAT tool is specialized software designed to support translators during the translation process. The software enables translators to create their translation within an integrated technology environment that provides them access to a translation memory (TM), which is the most important feature in the CAT tool, as well as terminology lookup and other broad and imprecise range of embedded tools that are elaborated in detail as follows:
- Translation memory tools (TM tools): the TM is a database of text segments in a source language together with their equivalent translations in one or more target languages.
- Spell checkers: they are built into the software or are available as add-on programs.
- Grammar checkers: they are built into the software or are available as add-on programs.
- Terminology managers: they enable translators to manage their own terminology bank in an electronic form. These can be a simple table created in a spreadsheet, or a more robust and specialized software package such as SDL MultiTerm, Termex, TermWeb, etc.
- Electronic dictionaries: these are both unilingual and bilingual dictionaries which are generally available as add-on-programs.
- Concordancers: they help translators in retrieving instances of a word or phrase and its respective context in a monolingual, bilingual or multilingual corpus.
- Bitext aligners: they align a source text and its translation.
- Project management: it is a software that enables translation project managers to structure complex translation projects in the form of a chain of tasks, assign them to different translators & language professionals, and track the progress of each task.
- QA Checkers: these are embedded in the CAT tools and have many functions such as a segment inconsistency check, tag verification, spellchecking, numeric and punctuation checks.
As previously stated, the idea of CAT tools basically revolves around the use of Translation Memories (TMs), which have become very popular with translators, translation agencies, and even clients. Although TMs have their advantages and disadvantages for professional translators and translation agencies alike, they have advantages and virtually no disadvantages for clients and translation buyers.
How Can A CAT Tool And Its TM Help In The Translation Workflow?
- Firstly, the great advantage of CAT tools lies in observing consistency throughout the whole translated content, both in the same document and throughout other documents, where phrases, strings and parts of strings, comprising terminology, abbreviations, acronyms, product or company names etc., are consistently used throughout all of the multilingual materials. This is mainly the role of the TM, which saves phrases and strings of texts and automatically reproduces and propagates them whenever the same Source Language (SL) phrase or string appears in any other location in the document being translated.
- Therefore, it helps to maintain consistency by always using the same equivalent for the same term or string and results in more well-organized and consistent translations. Moreover, if the client makes updates or has recurring needs in a specific topic, the same TM can always be used with his future projects, thus achieving consistency and coherence in terms of terminology and style throughout the translated projects. This is especially useful in managing large projects and books.
- Secondly, plus the role of TMs in observing consistency which automatically propagates identical phrases or strings, CAT tools contain another particularly important functionality, which further maintains consistency and saves a lot of time. Terminology search is said to account for 75% of a translator’s time.
- Here the CAT tools come in handy with the use of a concordance, which saves the translator’s time and spares him the need to strain his memory in remembering how he translated a certain term or string before or the need to go back searching throughout the whole document to locate it, especially in the case of large projects, books, other translation projects for the same client or translations in the same field of specialization.
- The translator will translate repeated terms, acronyms and names only once and with the use of a concordance, he can check how a term, acronym or name was translated before and will ‘insert its translation’ manually whenever it appears again in the Source file, which adds value to the logical sequence, understanding and readability of the document.
- Thirdly, CAT tools increase productivity which is of twofold benefit to translators and clients alike. Saving a lot of a translator’s time can lead to an increase in his income. For clients, CAT tools are also extremely useful because they allow them to receive the translation of their urgent documents in a quicker turnaround. This accounts for the fact that many clients now are not only aware of, but also require that their work be translated by a TM software.
- Fourthly, CAT tools enable translators to work with a wide range of source file types including web-based, graphics and desktop published files, etc. With the use of CAT tools, a translator can process and translate almost all file types. For more sophisticated file type projects crammed with untranslatable codes such as user interface, translators use specialized visual software localization tools that are specifically designed for this purpose such as SDL Passolo and Alchemy Catalyst, APIs, plugins etc., which are mainly designed for localization projects.
- Fifthly, all CAT tools have built-in quality assurance functions. Typical QA checks can include segment verification, segments to exclude, inconsistencies, punctuation, numbers, regular expressions, trademark check, length verification, QA checker profiles, tags, first-word capitalization check, unedited fuzzy check and many more.
- Translators can also create their own checks along with the ones already available. For those who need more powerful QA functions in different types of bilingual files, standalone QA tools may be used, such as QA Distiller & ErrorSypy, Linguistic Tool Box, and XBench for project managers and Verifika for individual translators.
- Sixthly, CAT tools help translators in maintaining the same format of the source document, thereby saving a lot of their time and allowing them to concentrate solely on the translation process.
In spite of all the above advantages of a CAT tool, we have to admit that it is a tool that can only be used for technical documents comprising a certain amount of repetitions such as books, manuals, brochures, researches and reports that contain a considerable number of repetitions and need to be updated quite often. This is where a CAT tool comes in to achieve consistency and coherence.
However, CAT tools are not likely to be used for literary texts where the context plays a more important role compared to technical documents. Literary texts are usually characterized by their figurative language and elevated style which makes them difficult to translate with a CAT tool, besides the fact that they are void of any repetition; something that makes the use of CAT tools totally impractical.
From the previous overview, we meant to highlight the functionality of machine translation and CAT tools, shed light on the use of machine translation and the wide uses of CAT tools, besides underlying the various advantages of using human translation supported by CAT tools, which is the ultimate goal that guarantees to get an originally consistent and coherent high-quality translated file backed up by a TM that can support all clients’ updates and future projects in the same domain.
If this article was of interest, you may also like to check another two closely related articles, namely, “For a Comprehensive Understanding of Your Translation Needs” and “5 Things To Do Before Ordering Document Translation“.
Now that you know all the specifics of machine translation and CAT tools supported human translation, will you be keen to ask your translation provider about the CAT tool that he uses the next time you order a translation?
I’d love to know your opinion and will gladly answer any queries or scruples you might have about the subject by leaving your comment below. If you enjoyed the post, please click the thumbs up icon and share on your network!