Human Translation or Machine Translation – Which Is Better?

Human Translation vs Machine Translation

The debate between human translation and machine translation has been in around for a long time now. While some disagree on the differences between the two, most believe that the quality of one is better than the other. How can we decide which one is better? — machine translation or human translation.

Undoubtedly, results delivered with both will vary depending on the content source, target languages, machine translation service used, expertise of the human translator, and complexity of the original text.

While machine translations are becoming better with internal updating, it still needs a proofread from a human translator. In simpler words, machine translations may not be reliable in translating complex ideas and sentences, idioms or creative language, industry-specific data, or critical data. For a translation to be more precise, you are more likely to need a human translator.

How do Machine Translators Work?

Machine translation is practically instantaneous translation with the help of software and advancements in AI. While it sounds like a direct process of converting one language to another, a lot more complex process goes into translation even the most basic text. The three types of machine translation systems used are:

1. Rules-based Machine Translation (RBMT)

RBMT was the first commercial machine translation system to be used. It is based on predefined rules such as grammatical, syntactical, and semantic rules that govern languages. The translation in RBMT takes place in three phases: analysis, transfer, and generation. It works best when translating between languages whose rules are dynamic and abstract.

2. Statistical-based Machine Translation (SBMT)

SBMT uses statistics to generate translations based on parameters that are derived from the analysis of existing bilingual sets of texts, known as text corpus. While RBMT is word-based translation, SBT uses phrases that reduce the rigidity imposed on the algorithm by word to word translation.

SBMT requires large volume of texts for translation. The translated output of SBMT has higher fluency than RBMT. However, the statistical-based translated text is less consistent.

3. Neural Machine Translation (NMT)

Neural Machine Translation is an advanced version of SBMT. It uses large artificial neural network that predicts the likely sequence of long phrases and sentences. Unlike SBMT, NMT uses less memory since the models are trained jointly to maximize the quality of the translations.

NMT uses complex algorithms such as deep learning and AI. Online translators such as Google Translate uses the NMT system.

Working within Limitations

If a database search for a literature review brings up an article that’s not written in English, your first reaction may be to dismiss it and move-on to the next article on the list. You could be missing a valuable article that could make an important contribution to your study.

Using one of the free online translation tools that are now available would probably not give you a word-perfect translation (especially if there is any complex terminology involved), but it may provide enough content to decide whether or not a complete translation of the article is worth pursuing.

The Promise of Machine Translation

Science fiction has offered two approaches to real-time translation of foreign languages: the fictitious babel fish from the Hitchhikers Guide to the Galaxy, and the complex machine translation of the voice commands of the Star Trek computers. Other than using the name for a translation software program, we haven’t made much progress with the babel fish, and voice commands for computers are still in the early stages of development.

In that context, we are left with software programs that either process text through rules-based systems or use statistical methodologies to track online language pairs. The current results are rough, often clumsy, and can best be described as basic ‘tourist-level’ translation.

Choosing the Right Translation Method

Complete and accurate machine translation may be several years away yet. As such, translation of your academic research should never be left to the mercy of online software options. The level of sophistication may be enough to help you get a rough translation of an abstract, but beyond that, place your work in the capable hands of specialist academic translators who understand the specific terminology of your field and are native speakers. This may require additional planning and budgeting in the delivery of your research paper, but as the saying goes, you get what you pay for.

Free Online Translation Options

If you are looking to translate a few phrases or an article abstract in an emergency, there are limited free options available to you:

• Google Translate – which first emerged from beta testing in 2006. It is fast, free, and will translate phrases, websites and entire documents. Translations to and from English are much easier than other language pairs (Spanish to Dutch, or Turkish to Thai).
• Bing Translator – formerly Yahoo’s translator offering, limited to five thousand characters.
• Babel Fish – the original translation software product, launched in 1999, sold to AltaVista in 2003, and when AltaVista was sold to Yahoo the same year, Babel Fish became the Yahoo translator. In 2014, the Babel Fish name was revived by Ocean Networks and now operates as a single phrase translation option — definitely ‘tourist-level.’

When to Use Human Translation?

Human translations are more precise than machine translations, if they are done by experts. You must use human translation as humans can interpret context and convey the same meaning as the original source text. Human translation never translates word-for-word. It ensures translation with finer nuances. Therefore, you should use human translation for appropriate conveyance of the original source’s information.

Machine Translation: Pros and Cons

Human Translation: Pros and Cons