by: Jason Paluzzi
Whether you’re a medical student, resident, nurse, or physician, you’ve no doubt encountered a language barrier at some point in your career. Interpreters are available for common languages in most settings, but if time is an issue, you may need to deliver care faster than an interpreter can arrive.
Over the years, people have gotten by with language dictionaries, hand gestures, family members, blue phones, and even smart phone apps all of which have proven somewhat cumbersome and impersonal in their own way.
Google Translate presents a possible solution to this problem. As an extension of their widespread integration of speech-to-text throughout the android platform, Google created a language app centered around a straightforward concept; enter one language, output in another.
While not explicitly a medical app, this functionality could easily be applied in a clinical setting. Without compromising privacy with extra people, the non-english speaking patient would be able to interact directly with their physician and receive accurate and timely care.
The app has a very simple interface. There is a text input field with a speech to text launcher, a language selector, and lists of “starred” favorites and recent translate history. To begin with, this app comes with a whopping 58 languages, including common encounters such as Spanish, Chinese, and Korean to more rare selections like Haitian Creole, Yiddish, and Galician. They are too numerous to list here, but, needless to say, if this app doesn’t have the language you need, you won’t have an interpreter on staff that speaks it either. There is also a convenient flip button that changes the source and destination languages. The favorites menu is a great time saver. Once you’ve entered in your own database of history and physical questions in English, they can easily be called up and replayed in any language.
Even such a minimalistic interface is not without its faults. The recent history menu has little utility after creating a thorough favorites list, unless several unexpected cases occur that require the same translation. Also, with so much screen real-estate, it’s a little surprising that only two lines of input text are visible at a time in portrait mode. Switching to landscape mode gives a much larger, paragraph sized entry field. This criticism is more or less irrelevant here, however, since most of this app’s utility will come from the speech button.
Hitting the speech button brings a pop up that cues the user to speak into thephone. A quick analysis later, and the app displays the translated phrase in text, along with a favorites and playback button. The playback voices are surprisingly good for the languages tested; both English and Spanish sound fairly natural and have proper accentuation as well as accents. Unfortunately, they do not reflect the intonation of questions. At the very least, you won’t have to subject your patient to a Spanglish accent, which can be fairly alienating despite the good intentions.
Now, down to the heart of the matter; can it translate? In order to truly evaluate the app’s potential, we decided to put Google Translate through a field test. We asked an experienced Spanish speaking standardized patient if they would participate in a Google Translate led interview. We wanted to see if it could get us through a history and physical in an active hospital environment.
I am a relatively soft spoken, deep voiced individual, and it quickly became apparent that unless I changed my speaking habits, the interview could not be conducted. At my normal speaking tone, Translate had difficulty separating short, monosyllabic words, oftentimes combining two spoken words into one translated word. At a normal pace, the app had difficulty marking the beginning and ending of words, oftentimes joining the last half of one with the first half of another. As soon as the app turned “What brings you into the E.R. today?” into “What turned you into heart attack,” I knew a change of tactics was in order.
Speaking slightly louder than conversational volume with a clear pause between words helped Translate fair much better. At the cost of sounding silly and harsh myself, Translate put out mostly correct grammar and 100% correct nouns and verbs for the remainder of the interview. While Translate was more consistent in translating natural Spanish into English than the opposite, it was still not enough to prevent having to speak differently. Having to ask a patient to speak loudly and with word pauses is time consuming, distracting, and unrealistic, especially if they are in acute distress.
While there is no substitute for speaking the same language as your patient, it’s unfair to ask physicians to learn so many languages. Hopefully, we can use technology to bridge the gap for us. One could argue that, without 100% accuracy, an app like this is useless in the clinical setting. While it might be possible to get by with a few grammatical errors, vocabulary errors such as the one above could be catastrophic. Even in a mock interview, our patient had an immediate negative reaction when the phrase “heart attack” came up at the beginning of the interview, and this could quickly lead to a serious medical error. Google has big plans for real-time translation, and they expect it to vastly improve in the next few years. As it stands, however, we can’t afford to compromise patient care by early-adoption of this technology.
- Speech-to-text-to-speech preserves the doctor-patient interaction
- Huge list of languages to choose from
- Easy to use interface with convenient favorites and language flip features
Dislikes/Future Updates I’d Love to See:
- 100% Accuracy
Google Translate shows a lot of promise, and can be useful for the tourist or the physician who wants to practice a language in their own time. In the clinical setting, however, both the physician and the patient would be better served by a medical dictionary or blue phone. The app is free, so don’t be afraid to take it for a test spin, and keep an eye on the evolution of this software in the coming years.
Jason Paluzzi is a third year medical student with interests in neurosurgery, trauma, disaster response, and healthcare for the underserved. He recieved his undergraduate degree from the Johns Hopkins University in 2008.