Earlier this week, I was editing subtitles for a show in Mandarin. Which was quite an achievement for me, given that I pretty much don’t speak the language.
How did I do it? With diligence, effort and Google Translate.
I told the producer of this show that I don’t speak the language. That he should get somebody who knows both Mandarin and English to do this work. But he said it was more about fixing the grammar of the already-translated subtitles rather than doing something from scratch.
Yet, when I began to feed the lines through the software to better understand the context, I saw that it had in places been significantly over-translated (ie, they had included more information than the original text), so I cut back on the words to make it more correct. Thank you, Google.
A recent report by a news outlet about the recent Asean Special Summit on Myanmar in Indonesia could perhaps have also benefited from some artificial intelligence-powered translation. It stated the Malaysian Prime Minister said the meeting had succeeded in addressing the Myanmar crisis, quoting him as saying, “We have succeeded. It’s beyond our expectations in getting the outcome from today’s meeting”, and “Myanmar responded well and did not reject all the three proposals by Malaysia”.
However, according to a tweet by Shahriman Lockman, director in the chief executive’s office of the Institute of Strategic and International Studies, Malaysia, what the PM had actually said was: “Ya, di luar daripada jangkaan kita. Bukan kita tidak confident, tapi tak jangka bahawa mesyuarat berjalan dengan baik; usul yang dikemukakan diterima; dan respon daripada Myanmar menerusi jeneral yang telah hadir tadi tidak menolak. Ini satu kemajuan yang amat menggalakkan, boleh dikatakan kejayaan mesyuarat ini.”
Forgive me for the lengthy passage in Malay, but if you are bilingual, you can see that this is quite different from what was reported. A more accurate translation would be to begin, “It was beyond our expectations. It wasn’t to say we were not confident, but we did not expect the meeting to run so smoothly”.
What should have been cautious optimism was reported as an outright success.
In one way, there is no excuse to not use a freely available translation engine given that even a 10-year-old can use it. I know that’s true because I already have Year Three pupils telling me that they use Google Translate if they see something in Malay they don’t understand (or, as is more often the case, they need to write something in Bahasa Malaysia).
But I always stress that it’s better to use the dictionary (I recommend the online one at prpm.dbp.gov.my), because Google sometimes doesn’t get it right. There is subtlety and context to be considered when translating. Also, if you use Google Translate, all you are learning is how to copy and paste text rather than really understanding and appreciating the language.
Still, it can’t be denied that its use has become widespread. I already know of people who use the software when drafting letters to the government (which usually have to be in BM), and somebody I know who is a certified translator uses Google Translate to help with some of her work.
Does that mean that being bilingual (or multilingual) is easier now than ever?
Of course not. Translators don’t use a machine in lieu of their skill. Rather, it is an aid, and they use it to generate a first draft quickly, which then gets fine-tuned.
Even though AI has improved remarkably over the last few years, it still fails to translate gracefully. There is just something a little “off” about the resulting end product.
For example, Google Translate generates the following suggestion for the Prime Minister’s quote: “Yes, beyond our expectations. Not that we are not confident, but do not expect that the meeting goes well.” Not wrong, yet not quite correct either.
In fact, sometimes the wrong translation is the right one to use.
Take for example something like the translation work done by Anthony Burgess of the French text Cyrano de Bergerac (both the play and the highly-acclaimed 1990 movie starring Gerard Depardieu). In one sense, the translation in places is just wrong – no computer would ever suggest what he writes. But much of the dialogue is originally in verse, and remarkably, the lines in the English version also rhyme. You can imagine how difficult that is, and it’s an indication of how far machine translation has yet to go.
Nevertheless, there is power at our fingertips that wasn’t there before. I have thought that given the state of technology, there is now less of a need to translate quotes into another language for the reader (or at least give access to the original). Then, people who understand that original language can still understand the full context, and those that want to use the engine can also look for themselves.
You can imagine, for example, that with a click of the button, you can see the actual original quote by the PM, and then compare the two side by side.
There is, of course, the risk of hubris, that we believe we know a lot from the little that we have learned. That is the risk I took when relying on Google Translate to edit the subtitles – but my producer is Chinese so I am relying on him to fix my English where necessary. If it’s wrong, then at least we’re wrong "together-gether".
In his fortnightly column, Contradictheory, mathematician-turned-scriptwriter Dzof Azmi explores the theory that logic is the antithesis of emotion but people need both to make sense of life’s vagaries and contradictions. Write to Dzof at firstname.lastname@example.org. The views expressed here are entirely the writer's own.