Translation Rules and Difficulties (English & Chinese)
By Darkgaia, In progress
Chinese and English are two of the world's most-spoken languages. First and second place, respectively. (See: Wikipedia:List of languages by total number of speakers) Being able to translate between these two languages effectively places one at a significant advantage, considering the astronomical demand for such a service. However, quality translation between English to Chinese is, irritatingly, a very difficult task. Internet memes have been made of poor Chinese → English translations in China. High school and college students undertake multiple-year courses in order to specialize in the field of English-Chinese language-pair translation. As of this writing, the best method to translate the English-Chinese language pair is through professional human translators. I have not yet found any machine translation programs that can produce even a decent sentence-based translation of the English-Chinese language-pair.
This page attempts to describe and explain the challenges of English-Chinese language-pair translation, and, hopefully, Apertium might be able to build a prototype for this revered language pair in the future. Technical rules are left in their Chinese forms to assist any future Chinese linguists/developers working on this language pair.
Common Translation Mistakes
Source Text Comprehension (生词词义不明）
This occurs when the translator fails to capture the meaning of the source text accurately or does not comprehend (some nuances of) the source text.
(eng) Drive carefully on that road in bad weather, it's very winding. → 那条路弯弯曲曲的，天气不好的时候开车得小心。 :: 那条路很多风,天气不好的时候开车·得小心。
"Winding" in the English sentence is an adjective that means "twisting, turning or sinuous". However, the translator understood it as a gerund of the noun "wind", the convection of air. This may have been caused by the suggestion of "bad weather" in the sentence. This results in an incorrect translation.
Contextual Errors (熟词望文生义）
This occurs when the translator understands the words individually but did not take into account the context.
(zho) 番茄 1粒 → 1 tomato
(zho) 花椰菜（花碎） 少许 → A little broccoli
(zho) 蟹柳 1条 → 1 crab meat (willow)
The last one is an example of an error.
Lexical Selection Errors （因词害义）
This occurs when the translator chooses the wrong translation for a word that has more than one translation in the target language.
(zho) 干菜类 → Dried vegetables :: F*** vegetables
"干菜" means dried vegetable and "类" means type. The translation should read "dried vegetables". However, "干" is also colloquial slang for "f***". The translator's poor lexical selection resulted in a widely-circulated internet joke.
Word Usage （词法）
This occurs when an inappropriate correspondence between the two languages is used in the translation of the certain words, resulting in ungrammatical output.
(eng) This is a most interesting book. → 这是一本非常有趣的书。 :: 这是一个最有趣的书。
"Most" in English can be translated into 非常 or 最. To make matters more complex, "most", as well as 非常 and 最 will all be classified as adverbs by the part of speech tagger. However, 最 is used as a superlative, while 非常 is synonymous with "very". As the word "most" used in this sentence is not used in the superlative mode, "非常" is most appropriate for the sentence structure.
(eng) His parents were dead against the trip. → 他的父母坚决反对这次旅行。 :: 他的父母为了反对这次旅行死了。
One can guess how an incorrect translation may occur in this scenario. "Dead", in this case, is used as an adverb of severity, to emphasise the strength of his parents' disagreement with the trip. "Against" is also a problematic word, possibly being a preposition in addition to being a verb as used in the sentence. The translator did a good job in catching that "against" was a verb, with the correct translation 反对. Unfortunately, not only was the wrong meaning of "dead" used in the translation, it had also been converted into a verb.
Sentence Structure （句法）
This occurs when the sentence is translated ungrammatically, even though the specific translations of the words are correct, leading to a poor translation and potentially misunderstood output. In Apertium terms, this is the result of wrong structural transfer rules.
(eng) We should not despise him because he is backward. → 不要因为他落后，我们就看不起他。 :: 我们不要看不起他，因为他落后。
Note how the phrase 因为, which means "because" is placed at the front in the correct translation, while placing it at the back is wrong. Additionally, note how 不要 ("not") is shifted towards the front as well. "Because" (working similarly in Chinese) is a subordinating conjunction, and must strictly obey grammatical rules to keep its meaning. The wrong translation of this sentence results in incorrect causality in the translation, literally "We should not despise him as he is backward."
This occurs when the translator, although having a fine grasp of the language, is unable to deliver a correct translation, because of a failure to understand idiomatic expressions or cultural references. This often occurs the source text is using a figurative manner of speech.
(eng) Betty cooked her goose when she cheated in the exam. → 贝蒂因考试作弊而葬送了前途。 :: 贝蒂因考试作弊，把鹅给烤了。
In this example, the English sentence does not mean that Betty literally cooked her goose! Rather, it means that Betty had got herself into trouble when she cheated in the exam. The correct translation should reflect this. In this case, the suggested translation references how Betty had ruined her future because she had cheated in the exam.
(eng) I want to hit the sack and get some Z's. → 我要去睡会儿觉。 :: 我要去击沙袋练练功。
In this example, "hitting the sack" is an idiomatic expression for taking a rest, or going to bed. This is made clear by the following phrase "get some Z's". The translation should reflect this in order to be correct. The wrong translation in this case describes literally "hitting the sack" as part of some boxing regime.
Inappropriate Word Usage
Flawed Sentence Construction
Modifier-Head Construction （定中结构）
Subject-Verb Agreement （壮中结构）
Logic Flow （逻辑顺序）
Wrong Grammar Transfer Rules
This occurs when the translator translates the sentences according to the wrong grammar rules. For example, when translating from English to Chinese, the result is written according to English Grammar (which is wrong).
Due to the intricacies present in translating the English-Chinese language pair, such as the need to observe context, understand figurative parts of speech, and follow the appropriate grammatical rules for the target language, it is extremely difficult, perhaps impossible, to deliver intelligible results that maintain the nuances in a text with the currently available machine translation technologies.
Nonetheless, it remains hopeful that this documentation will serve as a useful resource, if a sufficiently qualified developer wishes to work on the English-Chinese language pair for Apertium in the future.
Please see Darkgaia: English and Chinese Translation Rules (Advanced) for advanced translation rules and contrasting grammar between the English-Chinese language pair.