MACHINE AS TRANSLATOR – WHAT IS IT SUITABLE FOR?
What is machine translation? For most of us, the term is evocative of the many memes ridiculing absurd translations of user instructions or other automated translations which resulted in plain gibberish. In reality, machine translation is a much more complex phenomenon. It is connected to the wider development of language technology and general utilisation of automation in various areas of life and business.
Potential in further automating translation
When used correctly, machine translation technology will have plenty to offer. The most obvious and frequent viewpoint to be mentioned is the cost benefit for the one ordering the translation service. However, machine translation solutions can also benefit the person working as a translator. At their best, such solutions support the translation work as a kind of advanced translation-memory tool.
In addition, the translation of all texts by human input is not possible or sensible. In such cases, a machine translation can shed some light on the content of a text for a foreign-language reader.
Moreover, a new type of language technology enables the utilisation of language and translation services in completely new sectors.
What types of texts can machine translation be suitable to?
The translation of some types of text is best left for completion by manual work methods. Such texts include those that aim to influence, such as marketing texts. These texts often make use of subtle textual effects which would be impossible to formulate, let alone transfer from one language to another, without the associative ability of the human brain.
Technical translations, such as user instructions for various devices, are a different case. They represent type of texts where language technology could be put to use. They typically contain plenty of similarly repetitive texts, and their structure is often predetermined. They also contain a large amount of text which frequently requires updating. Therefore, the possibility – and also the need – for creating efficiency is obvious.
The fact as to how well automated translation is suited to technical translations depends, however, on the conditions and expectations. Machine translation provides the best results when the combination includes a text written in a controlled language, such as in standard English. The machine-translation solution itself shall be also specifically tailored for the customer and the language pair. Such adjustments make substantial difference. The outcome of using such tools is way beyond that of general translation services operating in a web browser free of charge.
Not all languages are equal in machine translation
The usability of machine-translated texts also depends on the languages between which the translations are made.
When translating certain languages, a machine translation provides very good results. In such cases, it is far quicker to use a machine-translated text as a starting point and then produce a finished text with the help of human editing skills. Translating the same text manually from beginning to end would take much longer. According to the language technology company SDL (PDF), translation from English to German, for example, is approximately 30% quicker in this way. Also, translations from English into French, Italian, and Spanish provide fairly good results as a machine-operated task.
On the other hand, languages where words are inflected and where even the stem of the word changes during inflection are difficult to process via machine. This is why Finnish translations are not especially successful when processed by a machine.
Sometimes satisfactory is good enough
There are also areas of application where a translation provided by a machine is good as such to a certain point, even without any human input. Studies have shown that users prefer to complete online shopping and certain other types of transactions using their mother tongue. And if a person’s language skills are lacking, it is not a question of choice, but rather pure necessity. If one cannot understand, they will have to take their business elsewhere. In such cases, even a modest translation can be better than no translation at all.
In some situations, an automated translation is perfectly functional. At times, the goal may be to get simply a general idea of the text’s subject, or the text is a very short-lived one, such as a discussion on social media. In 2013, Google Translate helped 200 million people each day to get a grasp of content in foreign-language texts. The message is loud and clear: we need machine translations.
Man vs. machine?
When evaluating the potential benefits of automating translations, it does not need to be a juxtaposition between a human translator and a machine. Especially, when the task areas alone are so different from each another! As with all types of technology, translation technology can also be used in impractical ways. But when we are aware of the facts, have the correct prerequisites set up, and our expectations are in the proper proportion, machine translation can serve its purpose fairly well.
Each year we see a significant increase in the amount of texts requiring translation, so there will certainly be work for translators also in the future. If anything, machine translation just supports this increase in translation volume.
The development of language and translation technology have much more to offer. Their use is by no means limited to the translation of texts for the needs of end users. In combination with speech-recognition technology, language and translation technology solutions can be used, among others, for facilitating interpretation. On the other hand, machine translation enables the automated content analysis of texts for the utilisation of mass data in business operations.
The strengths of the machine and the human complement each other. At its best, the machine’s capacity to process large masses of data quickly and reliably frees people from routine work. It also allows them to focus on what they do best: think creatively.