If Pinyin (or Other Romanization) can be Translated Back into Chinese Characters

Question

From what it sounds like (also here):

Chinese characters cannot be faithfully constructed backwards from a tone+syllable combination -- the mapping only goes one way (and even then, sometimes characters have multiple pronunciations).

And

There is no way to automatically convert pinyin into Chinese characters, especially if you don't have the tones. You can look them up individually on nciku, but you'll have no way of knowing which of the dozens of characters with the same pronunciation are the right ones.

I'm wondering if there is any way to write Chinese characters using the Latin alphabet (a-zA-Z) with/without accents of any sort, and then convert it into Chinese characters. Even if it is considered a bad idea for a new learner, I would still like to know if it is possible in any way.

If find it interesting when they say:

As you amass a vocabulary, you will start to notice patterns that will give you an intuition for which character is meant by the English pinyin, with or without tone markings.

And even:

In fact, even words like 终止 and 中止 sound identical with tones, so you only distinguish them in writing.

But in speech, you don't have access to writing, so this either means you need to learn how to read Chinese in order to understand these words in speech, or you can understand them in speech without learning to read Chinese, because of context of some sort.

If humans are able to tell how the sound of words translates into meaning, then I would wonder how we do that, because that would also answer the question.

Answer 1

I'm wondering if there is any way to write Chinese characters using the Latin alphabet (a-zA-Z) with/without accents of any sort, and then convert it into Chinese characters. Even if it is considered a bad idea for a new learner, I would still like to know if it is possible in any way.

Every Chinese character is associated with Unicode identification numbers. Not sure if this is what you are looking for. Nothing phonetic is available of this sort.

But in speech, you don't have access to writing, so this either means you need to learn how to read Chinese in order to understand these words in speech, or you can understand them in speech without learning to read Chinese, because of context of some sort. If humans are able to tell how the sound of words translates into meaning, then I would wonder how we do that, because that would also answer the question.

These sorts of words are distinguished in the same way that we do in English for two/too/to. A kindergartner can use these three words just fine and distinguish them all without knowing they have different spellings.

Answer 2

The closest thing to what you are describing is found in predictive input pinyin keyboards. Sometimes if I hear something I'm unfamiliar with I'll try typing it into my computer or phone keyboard to see what the predictive text suggests.

There is no perfect mechanical way, though. There are multiple levels of problems:

• Word boundaries in Chinese sentences are ambiguous. Separating Chinese sentences into words correctly is the subject of research and contests (!). And the existing work is starting from characters, not phonetics.
• Not only are there multiple characters for every sound, there are multiple words with the same sequence of sounds, or multiple sequences of words for a given sequence of sounds.
• Your final assumption (that humans can always tell which words were intended) is not even true.

(I'm assuming you aren't interested in things like typing in the unicode codepoints, or using a system like Cangjie where the person typing knows the character already and inputs a sequence of letters to specify the stroke components)