I’ve probably been aware of the website Deepl for around a year now. It’s a fantastic resource and if you’ve never visited yet you should have a look. It’s like Google Translate but much better. Of course it’s not perfect. But it is very good.
However, since it’s been around I feel like there has been much less discussion around understanding sentences in the book clubs I’ve been involved with.
When I’m reading myself and I find a sentence that I am struggling to put together, I’ll usually try putting through ichi.moe first. If I still can’t put it together, I usually find that Deepl will get me over the line. Sometimes I might have overlooked a particular meaning for a word. Sometimes there might be an idiomatic phrase that ichi.moe hasn’t picked up. Sometimes I’ve just failed to piece together a complicated sentence with lots of particles and subclauses.
Usually, having got the meaning of the sentence from Deepl, I can go back and figure out for myself how the sentence works.
Now, I’m not complaining! It’s an amazing resource, and if it helps us all read independently without having to ask for help that can only be a good thing.
It’s just the book club threads seem a lot…emptier.
I was wondering if anyone else had noticed the same thing?
Not sure if I could make the connection for others, but certainly for myself. Deepl is also somehow pretty good at guessing what contextual clues to incorporate in the translation, it is truly uncanny.
Sometimes after I’ve figured something out I think it might be worthwhile to post my findings, but that’s always felt a bit weird so I’ve never done it
Yes, I probably used to do more of that in the past. If I’d struggled to figure it out but got there, I might share my workings out, thinking it might help save someone else the trouble.
I hadn’t been aware of it before, but I just tried it and it’s fantastic! I’ve been singing a Japanese song that Google did not translate very well. Deepl nailed it!
Bookclubs also discuss the book and story itself, so no, I don’t think that’s the case.
And I feel like the majority questions are more nuanced than “I don’t understand this”, but look at a grammar pattern, sentence construction etc. Deepl can’t explain that.
I can’t honestly say I’ve noticed any particular increase in the usual rate of decline for book club participation. Which is to say, book clubs always drop off over time, but they don’t seem to be dropping off particularly faster.
Yeah, but sometimes it goes “this is too hard!” and substitutes the translation of the following sentence instead.
Well, ideally. There needs to be more of this, regardless of whether Deepl’s having an effect.
As a participant in an ABBC, I agree with Mickiさん. The discussions probably get more in depth in the more advanced clubs, but for us the main thing is understanding the language, and DeepL is removing a lot of the need for that sort of discussion.
I haven’t tried DeepL yet (I do things old school: I search the dictionary for definitions and I look up grammar on the internet, and if I’m really, really stuck, I ask my fluent friend), but it sounds pretty great. I guess the thing to remember – as various forum posts have revealed – is that even DeepL isn’t perfect: sometimes, it’ll just come up with whatever seems to make the most sense, even if there’s something wrong with the sentence. I trust machine translation to provide the gist when translating between structurally similar languages like English and French – which is why I’m OK with leaving my mother to work through French messages using Google Translate when I don’t have the time to translate them myself – but stuff like Japanese is generally a lot harder, even if DeepL seems (from everyone’s feedback) to be on the way to cracking the problem. Perhaps raising some awareness of that will revive grammatical discussion in book clubs?
That aside, there’s also a need to know why something or other works, and how it can be applied in other contexts. Are we able to say what the structures we’re looking at mean, and do we know what else they can be used for? That might be worth discussing, because just knowing the overall meaning of a sentence and following the story isn’t always enough in the long run. Another skill that could be worth cultivating, possibly with the help of DeepL, is that of identifying the underlying structure of a sentence in order to break it down. I personally tend to be able to figure out what the grammar point is that I don’t understand in a given sentence, which allows me to search it directly (which might also be why I rarely ask questions on the forums). I don’t know the grammar point beforehand, but when I type my guess as ‘~ grammar’ into Google, I usually get results from a Japanese teaching site or a dictionary, which I take to mean my breakdown is accurate. That’s something book clubs could stand to discuss more: how exactly do you break down a sentence? Which are the bits that are just vocabulary (which sometimes sits in the middle of a broader structure), and which bits are the core? Advanced learners can tackle these questions themselves, but there’s no guarantee that a translation, no matter how good, will teach you all that.
Deepl is also better at English - French translation, I would recommend your mom to use it instead.
Recently I did a very fun survey about machine translation (a bit long though). The survey presents a serie of pair of texts that have been translated either by google translate, deepl or an human. The goal is to spot the human and rate the naturalness of each translation:
The results are here:
I did it for French, (and it’s mostly text translated from English, but also a few from Japanese) and Deepl is astonishing good at academic and formal text. It fooled me a few times.
However, and it’s relevant to this topic, google trad or deepl were much worse when the text come from literature. As soon as the writer use interesting metaphor, long sentence, or any kind of creative or nuanced language, GT and Deepl crumble and sometimes get completely nonsensical. So I think there is still plenty of questions to ask in book club !
I think I’ve seen the same thing happen in Google Translate for… a particular long Japanese text, and also an experimental Latin translation? I ended up with lots of phrases that felt like they made sense, but which couldn’t form any proper sentences. Goes to show that there are limits. The AI apocalypse is not yet here!
I agree with that, and the polls are great for understanding this. I can see people are still reading along, I just feel like they are asking less questions.
That’s a good point. I’ve never participated above beginner book club level. @Arzar33 has pointed out the machine translations struggle more with creative language and metaphor. Maybe this is more of a beginner level phenomenon.
Good point! And of course when I’m in an English language book club that’s all we talk about. Despite being here mainly for language learning this is still my favourite bit in the book clubs!
