Live data from Hacker News

Has Google Translate been fixed yet?

isgooglefixed.tw

131–140 of 229 posts

Re: Has Google Translate been fixed yet?

#131

Isn't ChatGPT better than Google translate or deepl? Why doesn't google use a modern LLM to translate?

ChatGPT is a transformer model. Google pioneered transformer models specifically for translation. If ChatGPT outperforms google translate, I'd guess it's simply because chatGPT is a bigger model with more training data.

Ironically, the amount of training data in non-English languages for ChatGPT is tiny. In the order of 10% for all of them (IIRC).

Re: Has Google Translate been fixed yet?

#132
post #99

Earlier quoted context omitted.

A similar reason why US hasn't switched from imperial units, just way bigger. The current system works, replacing it would be a massive multi-decade long change with a lot of opposition. Cultural pride/arrogance plays a big role. It's not inconceivable to think that the Chinese regime might fall in the process if it tried to pull it off. Given all these, the benefits just don't outweigh the costs/risks, at least in t…

If you want to write Chinese text on a computer - you still have to use Latin alphabet, because it's not possible to put thousands of characters on a keyboard. So a different writing system for Chinese has already been developed. Some other languages also have more than one writing system. (Serbian and Azeri). When mobile phones became accessible to the general public then many people used Latin alphabet to write in…

This is an interesting question. computers deal in numbers. but were originated in a latin alphabet and are optimized around an alphabet style of processing. while we managed to jam chinese into unicode by treating it as if it were just a big alphabet. this is not entirly correct and there is a lot of information lost in the process[1]. what would a native chinese computer system have looked like?

However note perhaps the simplicity of the latin alphabet(a small set of separated characters) lent itself well to a simple implementation on early computers. which got them out of numerical processing earlier than if they had to speak chinese or even something like arabic where the written language has advanced to the point where the connected cursive form was the only correct form. the rendering of which would be tricky for early computers. note that english has a connected cursive form, the art of which has just about been destroyed by computers.

1. I don't read(or speak) chinese. but the characters are composed of sub characters and stroke order that may provide hints or insight into the nature of the character. Or they may not, I don't know, I don't read chinese.

Re: Has Google Translate been fixed yet?

#133
post #40
post #31

Earlier quoted context omitted.

Please read the rest of the comment. I was assuming that what is called "Traditional Chinese" is used in China, based on its name (Chinese) and that GTranslate outputs it. Urdu and Tamil aren't spoken in China and isn't what GTranslate outputs when you select a Chinese language, afaik

Bear in mind that Taiwan (the country) also calls itself China, and so did the rest of the world until the seventies.

What do people living in Taiwan call the 'other' China? Is "mainland China" a correct thing to say, or is the word "main" there implying that Taiwan is lesser and not an equal/peer state?

I used the word China in other comments because I didn't know what to call it otherwise, hoping that in the context of Taiwan it is clear what I mean

Re: Has Google Translate been fixed yet?

#134
post #58
post #5

Taiwanese here. I have a mixed opinion about this site. Some of them are just weird: - mega → 巨型 (百萬) Really? Do you really want to translate mega to 百萬 (million) in every context? *Edit: every, not any - text → 文本 (文字) Seriously? 文本 is the formal, academic way to say it, even in Taiwan. Have this person never been any academic environment? - through → 通過 (透過) Seriously???

Honest question, why haven't china moved to a western (?) alphabet based writing? With my ignorant knowledge, it seems crazy that you are using that way of writing today. Using 24 letter combinations vs memorizing 3000 (min) to 20.000 icons! When I was a kid, my best friend was chinese, and we could never do anything on saturdays because he had to go to a school to learn chines, and how to write in chinese was very h…

I honestly believe you asked this question out of your ignorance.

Despite the apparent difficulty maybe the people of China like their language nevertheless.

Languages vary widely in their expressivity.

Also, I think language (and the script) is closely related to the culture as well. The people of China might not want to let go of that.

And what about the literature and everything that's already written down in their native language? It's not just a question of translating everything. Trust me, things get lost in translation.

Also, if we're going to have everybody in the world work with the same script, there are arguably better candidates such as Arabic and Sanskrit.

In Sanskrit, for example, there's only about 50 letters (49?), it's impossible to mispronounce, the whole gender-neutralization situation becomes irrelevant (because well, things like car, village, fruit, pencil etc have genders (of which there are 3 btw — masculine, feminine, and neuter, but they aren't used in the way you would think)).

> why haven't china moved to a western (?) alphabet based writing?

By the way, here's how this appears to us non-westerners — okay here comes this Western hero who thinks the rest of the world is nonsense and should be replaced by the superior Western™ system.

Re: Has Google Translate been fixed yet?

#135
post #3

google maps also in taiwan made me ride on the highway on a motorbike which isn't allowed and got a nice ticket for following googles directions even with avoid highways enabled.

Google Maps did make you do anything. It's a dumb tool that is not always correct made by people in USA. You decided to forgo basic due diligence in a foreign country.

Re: Has Google Translate been fixed yet?

#136
post #75
post #58

Earlier quoted context omitted.

Honest question, why haven't china moved to a western (?) alphabet based writing? With my ignorant knowledge, it seems crazy that you are using that way of writing today. Using 24 letter combinations vs memorizing 3000 (min) to 20.000 icons! When I was a kid, my best friend was chinese, and we could never do anything on saturdays because he had to go to a school to learn chines, and how to write in chinese was very h…

Because our alphabet doesn’t work for tonal languages and inventing a new one requires a king

There are chinese languages - completely tonal - whose only comprehensive written representation is with the roman alphabet.

Re: Has Google Translate been fixed yet?

#137
post #5

Taiwanese here. I have a mixed opinion about this site. Some of them are just weird: - mega → 巨型 (百萬) Really? Do you really want to translate mega to 百萬 (million) in every context? *Edit: every, not any - text → 文本 (文字) Seriously? 文本 is the formal, academic way to say it, even in Taiwan. Have this person never been any academic environment? - through → 通過 (透過) Seriously???

[flagged]

Re: Has Google Translate been fixed yet?

#138

Earlier quoted context omitted.

Google does not use a language code in the UI. They describe the language as “Chinese (Traditional)”, which could mean anything. The URL contains a language code, but that’s just an internal implementation detail.

Google should rid itself of the notion that “Traditional Chinese” and “Simplified Chinese” are anything but complete misnomers if you’re talking about languages . They are scripts , and scripts don’t have a 1:1 mapping to languages. Imagine a world in which Google Translate has one entry called “Latin” for anything using Latin script (Welsh, Irish, you name it) and another entry “Simplified Latin” for English. Just l…

The problem is that detecting the different variants to label the training data requires understanding the semantics (detecting French vs English can be done purely lexically), which has historically been a hard problem. With modern language modeling it’s tractable.

Re: Has Google Translate been fixed yet?

#139

This is an interesting issue. Simplified and Traditional are different (but closely related) writing systems that can be used to express the exact same text. They are not different linguistic variants of Chinese. It looks like regardless of which writing system you choose, Google still translates into the most widely used variant of Chinese: Standard Chinese, as spoken on the mainland. It just writes the result using…

If, as you are claiming, the Taiwanese don't have their own Taiwanese Mandarin, with their language being just the same as the Mandarin used in China, how can it be that at the same time they also don't get to have a say in what constitutes the "standard" of that "common" language they share with the "mainland?"

Also, there is no reason for constantly calling it "Mainland China" where you could just as well call it "China," unless it is to further a political agenda.

Anyway, this is all beside the point. The concept of a "standard" language is political not linguistic. If Google is run as a business, not a political entity, their linguistic choices should reflect the language actually being used in any given market, and not be based on purported "standards" promulgated elsewhere. The same simple concept that somehow already works well for other language pairs that could be construed as similar to the point of being the same should also be applied here.

Re: Has Google Translate been fixed yet?

#140

Earlier quoted context omitted.

Google should rid itself of the notion that “Traditional Chinese” and “Simplified Chinese” are anything but complete misnomers if you’re talking about languages . They are scripts , and scripts don’t have a 1:1 mapping to languages. Imagine a world in which Google Translate has one entry called “Latin” for anything using Latin script (Welsh, Irish, you name it) and another entry “Simplified Latin” for English. Just l…

The problem is that detecting the different variants to label the training data requires understanding the semantics (detecting French vs English can be done purely lexically), which has historically been a hard problem. With modern language modeling it’s tractable.

I think Google has the resources required to detect different languages in Chinese script (different combinations of characters and vocabulary, even specific Unicode characters, etc.) but there is 1) lack of incentives and 2) opposing political interests. Doesn’t help that there is a bit of a mess when it comes to Chinese characters, like a single Unicode point can be rendered differently depending on the font.
Post reply on HN