The goal isn’t to teach vocabulary or grammar; there are other apps, books and teachers for that. In theory you don’t need to understand a word you’re saying in order to pronounce it correctly. However I believe it’s useful and good for motivation when you learn something that you can see yourself using in the real world.
How big should a course be?
For practical and philosophical reasons, I’d like to turn that question around and ask how short it can be. My hypothesis is that after practicing X words, you’re able to pronounce most new words correctly as you encounter them. All I need to do is figure out what number X is. My initial guess is a few hundred.
Let’s take English as an example. Once you learn how to pronounce duck and love you should have no problem pronouncing luck and dove. Of course you’ll never be able to pronounce worcestershire sauce, but how often do you use that? It’s really about 20% of words that make up 80% of spoken language.
I assume the same thing applies to Chinese, albeit only work for Pinyin text. For digital text transcription is easy, but rest assured that if you want to read text in the psysical world: there’s an app for that.
So why do I think a couple of hundred words is enough to learn descent pronunciation? Initially it was just wishful thinking and intuition; I wanted to create 10 new lessons for my application in the next couple of weeks. Only an experiment will prove if I’m right or wrong of course, see below. But then I came across some good news on wikipedia:
Syllables consist maximally of an initial consonant, a glide, a vowel, a final, and tone. Not every syllable that is possible according to this rule actually exists in Mandarin, as there are rules prohibiting certain phonemes from appearing with others, and in practice there are only a few hundred distinct syllables.
Update 20:40: More evidence to suggest that X is around a couple of hundred, is that each character is one syllable which consists of an initial and a final (similar to vowel and consonant). There’s only 21 initials and 33 finals. Since most of the tone is expressed in the finals, that means there’s 21 initial plus 33 * 4 final sounds, so 153 sounds in total. That assumes all tones are used for each final, so the number is probably lower.
I plan to use four ingredients, most of which I still need to learn more about:
- The principle of Language Deconstruction (1,2) as advocated by Tim Ferriss. I.e. create an inventory of strategic words like I, you, we, is, want, must, go in, up, down and grammar like past tense and asking questions.
- Phonology matrices as I would call them. They describe all possible and existing sounds in a systematic way, based on the physics of our voice.
- The top few hundred frequent characters (syllables) or words (group of syllables)
- A bunch of standard phrases like hello and thanks
To be clear, I barely speak any Chinese myself. I did briefly study about 50 lessons of this course just to get a taste of language. As before, I will seek assistance from a native speaker to check the material I create, pronounce the phrases and to listen to and rate student pronunciations.
Test subjects needed
In order to find out if ten lessons is enough to get reasonably good at Chinese pronunciation, I devised a little experiment. I will present test subjects three unique sets of new words before, halfway and after the course. They need to pronounce the text without hearing it first. A native speaker then scores each word, where -1 means incomprehensible, 0 means OK and 1 means good or native speaker quality.
I’d be very happy if a newbe starts out with an average of -0.7 and ends with 0.3 with at least 80% of words above 0. I would need to create a benchmark with real people based on real situation to see which numbers make sense, e.g. the equivalent of the TOEFL scale.
If you’re interested in being a test subject, please send me an email at firstname.lastname@example.org. I estimate that following the full course and taking all three tests will take a couple of hours spread over a few weeks. In plan to start in early february.
Contrast in stead of frequency? (Update 20:40)
There might be another, possibly more effective, approach to using the most frequent characters. Surendran and Levow from University of Chicago’s Computer Science Department suggest in their paper to look at contrasts instead:
While frequency counts represent a measure of how much use a language makes of a linguistic unit, such as a phoneme or a word, it is often more important to consider not the units themselves but the contrasts between them.