Modern translation apps such as Google Translate and Bing Translator are stacked with features designed to aid visitors to foreign nations. Language packs can be downloaded for offline reference and phrases can be translated by typing them in or simply speaking by voice.
This ease of use has driven many online translation services into decline. Despite promises of real human translation and maximum accuracy, it is much more convenient to simply open an app on a phone. The language interpretation industry still generates over $37bn worth of sales a year however.
Recent innovations include the ability to use a phone’s camera to scan written text directly for translation, making foreign menus and signposts more decipherable to tourists. Yet many wonder just how reliable the services are in real-world use.
The BBC recently took a selection of popular apps including Google Translate to Bilbao in the Spanish Basque Country, hoping to solve this predicament. The apps faced a number of different tasks including navigation, hiring taxis, buying stamps, asking locals to take selfies and talking to museum staff about specific details of artifacts.
Google’s Translate first response was a little less than perfect. Asked “OK, so I’ve arrived at the Guggenheim and I’m here with Begoña and I have to ask her a question, according to my list of things to do”, the app replied with the unintelligible message “ok so the rise of the guggenheim and and and i have got acid?”
In one of the tougher tests, the app was tasked with relating the details of how a couple first met each other. Although it did deliver the results, the new “real-time” translation option where the app responds as a person speaks failed to work. The reason: the conversation was moving at a “normal” speed which proved too quick for the app.
The BBC reporter’s attempt to help the app by asking the woman “can you speak slower, please” inevitably failed as Google translated it to “can you speak Spanish Big Show.”
Google Translate alternative Vocre, which supports 38 languages, also promises real-time translation but suffered a similar fate to Google when conversing with museum staff.
Another app tested, called Interpreter, “steadfastly refused to live up to its name,” which perhaps explains why people typically only use the better-known offerings by the big names.
The report is interesting as it confirms what many people have always thought. It is worth noting that given a few years the technology could be much more developed though. Just a few years ago, any form of translation service was basic at best and it is encouraging that app developers are expanding their feature sets with functionality like real-time translation and photographic translation.
As this infographic prepared by a translation company shows, Google Translate is very successful at delivering the general gist of a text that would be intelligible enough for personal use while holidaying abroad. It is weakened by disjointed sentences that do not flow together correctly however, something that human translators are much better at owing to real-world experience of using the language. Incorrect verb choices and tenses are a common tripping point for translator apps in general.
An issue regularly brought up is that translator apps tend to be much more reliable when used in quiet environments. Indeed, this was noticed in the test where Google Translate conveyed a request for a selfie “almost perfectly” when taken to a quiet environment. All the apps were also perfectly adept at translating single words or short phrases when input with the keyboard.
Perhaps with some more development and improved voice detection these apps could begin to become more accurate in general use. Voice detection is regarded as one of the hardest tasks in programming due to the massive range of accents, dialects and environments that must be handled.
