“The problem with audio-only prompts is that there are many homophones, so that after a basic level, it becomes too ambiguous as to which character or word is being prompted.” —Skritter FAQ

The following excerpt from a dialogue in my textbook (Integrated Chinese, Level 2, second edition) suggests how some Chinese speakers get around such ambiguity in practice:

我姓張，弓長張，也就是一張紙的張。
我姓张，弓长张，也就是一张纸的张。

張／张 is an example of a character for which an audio-only prompt (“zhang1”) would be ambiguous, because several other characters have the same sound. The initial context “wo3 xing4 zhang1” helps narrow the possibilities, but some ambiguity remains because more than one “zhang1” is a surname. Either “gong1 chang2 zhang1” or “yi4 zhang1 zhi3 de5 zhang1,” however, is enough to uniquely identify the 張／张 character.

What if Skritter users could enter such a short pinyin phrase, when necessary, to distinguish a written word from any homophones familiar to them? Call it a disambiguation phrase. Skritter could later use the audio for that phrase to prompt them for the word. For any word without a disambiguation phrase, Skritter’s audio prompt would use the sound of the word alone. (Beginners would not need to use disambiguation phrases at all.)

Unlike example sentences, disambiguation phrases would be as short as possible and would consist of sounds (pinyin) rather than written characters.

A Skritter prompt that simply displayed the pinyin for a disambiguation phrase (instead of playing audio) would provide many of the advantages of true audio prompts without the technical difficulties of text-to-speech. I am currently trying out this approach by entering disambiguation phrases as custom “definitions,” but would of course prefer to be able to use both real definitions and disambiguation phrases.

I am curious to hear what other users and the Skritter team think of this idea.

I was thinking that listening practice could be achieved if the required input were simply pinyin. For example, if you hear zhang3wo4 then you type zhang3wo4. Then the characters appear with the definition. No need for disambiguation.

I agree that this would be useful, but as the Skritter FAQ says it seems virtually impossible to implement unless some context is given with the word in question...

On a related note, I have always felt that there is a missing step of 'English to Pinyin' prompts; essentially a replacement for speaking prompts (since voice recognition is nowhere near able to cope with this yet). It would be easier to implement because it would only have the existing homonyms or near-synonyms as discussed in a previous thread (http://www.skritter.com/forum/topic?id=1208729217&comments=7). Out of curiosity, why not have this?

本杰明, transcribing audio to pinyin would develop only part of the listening skills I have in mind. It is possible to produce an accurate pinyin transcription in a "mechanical" sort of way by focusing on the sound of one syllable at a time. In contrast, transcribing into characters requires processing groups of syllables in a more holistic way, listening for meaning rather than only for sound.

Catherine, what I meant to suggest was that disambiguation phrases would provide the necessary context for audio prompts.

I completely agree with you regarding the lack of "English to Pinyin" prompts. That's why I hide the pinyin at writing prompts and don't give myself a green check mark unless I can remember both the writing and the pinyin accurately. Lately I've taken to treating definition prompts in a similar way, requiring myself to remember not only the definition but also the pinyin in order to pass the prompt. A strict application of this approach makes reading and tone prompts unnecessary, so I've disabled them in my settings. What's left are two kinds of prompts:

1) Given the definition, provide the pinyin and the character(s).
2) Given the character(s), provide the pinyin and the definition.

I like this setup, which I find lends itself to the selective use of disambiguation phrases in place of definitions.

I'm puzzled that this thread has received few responses from users and no responses from the Skritter team. Perhaps it's the fault of my writing (tone? length? clarity?). I'm sorry if I offended anyone. Let me make one last effort in the thread with a brief summary of my thoughts.

1) I realize that true audio prompts are off the table for a while, so I suggest the following type of prompt as a substitute:

[Given only the pinyin for a word, write its character(s).]

2) Even though the above is not an audio prompt, I suspect it would develop listening skills indirectly, due to the close relationship between pinyin and sound.

3) Any ambiguity in a short pinyin could be cleared up by displaying a slightly longer pinyin that contains the shorter pinyin, such as "rénmínbì de bì." This would avoid having to show the definition of the word as a "hint" as suggested in the FAQ.

4) Given a new field in which to enter the clarifying pinyin, the Skritter community could quickly supply good default values, and individual users could customize those values as needed, just as with today's definitions.

I like what you're saying, but new prompt types are huge projects (especially across three platforms), and the current huge project is fully releasing the Android app, so if we were to start work on a new prompt type, it wouldn't be soon.

@mjv I support your idea of assembling a list of "disambiguation phrases". I've wondered in the past, after hearing secretaries spell out Chinese words over the phone, if there was anything like a standard list for every character X, of XY words such that "XY 的 X" describes X, a list analogous to "A as in Alpha, B as in Bravo, ...".

http://www.skritter.com/forum/topic?id=190170176

The answers I got were no, everyone comes up with their own disambiguation words.

But coming up with such a list would be helpful. It would not only solve the problem of prompting, but also be helpful as context to remember characters with.

Kai Carver, thanks for the link! Yes, that's exactly what I had in mind for single-character prompts. It's good to know I wasn’t the first to suggest it. Lately I’ve come across a few two-character words that are ambiguous as well, but they could be dealt with in a similar way, e.g., “yìshùpiàn de yìshù.”

Audio-only prompts would also be useful for something I have been fantasizing about: an audio-only version of Skritter, with only audio prompts and voice control.

Keep your phone in your pocket to study your vocabulary!

(OK, maybe not useful for studying writing, but good for the other parts)

a guy I know recently made an experimental audio-only study app:

https://www.facebook.com/studynwalk/

but I think it still lacks voice control: you have to press buttons to grade yourself. Not good if you are flying down the road at 80 mph on your scooter with both hands busy steering.