Instant translation is starting to appear everywhere, alongside internet calls and at our fingertips through free applications. So the answer to the question: Can a machine give me a translation? Is ‘yes’. However this is far from being an open and shut case. Machine Translation (MT) is multi-layered and its application needs to be understood.
The voice in our smartphone
When we hear the latest ‘digital assistants’ such as Siri or Cortana, we recognise their voices as human. For basic instructions, they are useful, taking us directly to web pages or adding our calendar reminders. It is when a search or instruction becomes ambiguous that they begin to struggle. This is because the machine behind the human voice cannot comprehend the contextual or emotional information a human gives another human.
A machine can learn but this training is a long way from producing human-like results. Indeed the best output combines machine and human. To return to machine only solutions, their use is very much dependant on the level of accuracy and quality needed. For example, language support for a trip abroad or non-sensitive communication with international colleagues, the outcome of translation is unlikely to have negative impact. If the content contains information of greater significance, such as medical or legal documents, an accurate translation becomes essential. A machine can help the former but for the latter you would use a human or machine/human combination.
The machines are not taking over just yet, but their ability to save time and provide assistance more than justifies their presence. When it comes to translation it is important to consider the impact an error would have on the outcome.