A translator isn't a magic lamp!
Hallo Are you living a smart translation life with a translator?
When we have conversations with friends while producing/consuming content in languages, we often exchange the expression “like Google Translate.” I mean, the quality of the translation is just so strange.
Why is the translation quality so weird? The translation quality issue isn't just about yesterday and today.
In fact, Google Translate, a global leader in artificial intelligence translation, and Papago, an unequivocally popular translator in South Korea, have excellent performance.
It's a translation tool frequently sought by young and old to overcome language barriers, regardless of whether their level of knowledge is high or low. It's better in terms of translation performance than any translation professional tool used by translation professionals.
Many people use AI translators in the form of “initial translation” and quickly revise while looking at the whole thing.
However, a large number of people face translators like “magic lamps.”
I copied the text and pasted it into the translator, and the expression I wanted was amazing! I hope it comes out.
I blame the translator for seeing results that didn't meet expectations. “After all, the translator...” Is the translator really wrong?
Of course, I'll leave out the absurd situation of translating yukai as Six Time, gomguksu as Bear soup, and knife noodles as Knife Noodle 😅
In this post, I'll show you how to use AI translation, which is represented by Google and Papago, twice as effectively.
I hope it will be helpful to your translation life~
How much do you know about translators?
Language is content involving human consciousness and thought. Unthinkable AI translation first evolved through data learned by humans.
What is AI translation Rule-based Machine Translation (RBMT: Rule-based Machine Translation), Example-based Machine Translation (EBMT: Example-based Machine Translation), Statistics-based Machine Translation (SMT)It developed into and is now equipped with the latest translators Neural Machine Translation (NMT: Neural Machine Translation)It has gone through many processes until now.
Even though AI translation systems have become so advanced, there are parts that are still, and will continue to be, impossible.
AI translation is about having the ability to think for itself. Even with advanced AI translation, the results will vary depending on how the person's source text is entered.
After all, in order to reduce translation errors, it is necessary to enter the correct source text.
Of course, entering source text containing misspellings, phrases, and inscriptions increases the probability of errors, and characteristics according to the input language become a major variable even in the natural language processing part called NLP.
In particular, since Korean is a language close to the tongue, various translation errors may occur depending on the use of the survey and the discourse environment.
Translation is the task of changing one language into another to facilitate communication.
Therefore, when considering Korean standards, the factors that influence AI translation quality can be considered as follows.
Long, complicated sentences
If you can't figure it out
When polynomials or homonyms are included
When the meaning within the sentence cannot be interpreted
When the context is difficult to understand
Long, complicated sentences
Long, complex sentences in various foreign languages as well as Korean create numerous variables. Since the syntactic structure of long, long sentences are complicated, the meaning of the original text is distorted or the contextual connection is not smooth. Many copywriters also say that a well-written article is an article that takes a short breath and is clearly understood. The longer the sentence and the more complex the syntactic structure, the worse the translation quality. As a result, translation errors are significantly reduced when entering simple sentences in separate sections. Be sure to try the training to divide the source sentence (starting word) in detail.
If you can't figure it out
For most languages, proper subject settings are a requirement. This is especially true in English or Chinese. Since subjects are often omitted in Korean, if the subject is not entered when running the translator, AI translation randomly interprets the sentence and sets the wrong subject. Sometimes AI is confused about whether it is active or passive. If a subject is omitted or the subject is difficult to understand, the translator must find the subject and restructure the sentence.
Polysemic words and homonyms
Polynomy refers to words in which a single word has two or more meanings. For example,
Homonyms mean words with the same pronunciation but different meanings without being related to each other. AI translators often don't understand polynomials or homonyms because they are context-related. Compared to the past, AI translation has made remarkable progress, and in particular, the ability to understand context has grown. Nevertheless, if problems occur due to polynomials/homonyms, rewriting the original text or using common vocabulary can also help reduce errors.
When meaning cannot be interpreted & context is difficult to understand
Difficulty in interpreting meaning within a sentence usually occurs when trying to express nuance using idioms, colloquial expressions, use of research, neologisms, and positive or negative words. These parts can be understood similarly to when AI translation doesn't understand the context. When interfering with the context, it's a good idea to check whether idioms or colloquial words are used as necessary, whether the investigation is appropriate, and whether the context is not being damaged by using neologisms or biased words. Of the two, errors are most often caused by not setting an appropriate subject.
**
It can be said that the way to make good use of a translator is ultimately in line with how to be good at native language.
Omit the lengthy parts or express them in short (= simple) sentences.
Set the subject.
Use expressions that are easy for anyone to understand (= computer).
Put a specific target instead of a directive.
Using universal vocabulary
If you use alternative books such as, you can expect better AI translation results.
The quality of hybrid AI translation, which is being advanced day by day at Twig Farm, increases as you use it. Even if the user only changes the Korean style (ending, spacing, written language), the ability to understand the context increases overwhelmingly. We provide customized translators for each companyWe are doing it, so if you need advice about this, please contact us by email below.
We support your smart translation life, and it was Gicon Studio.
+++
*This content has been transferred from Gicon Studio to Letterworks.
