How to find out how good your Chinese pronunciation really is

Something that have made me very frustrated throughout my Chinese language studies is the way my own Chinese ability has been evaluated by teachers and institutions. As a beginner, I felt frustrated because the tests only focused on a very small part of the language. This turned out all right for me because I learnt the rest because I was interested enough, but it wasn’t okay for students who were less independent. I wrote about this problem in this article: The importance of counting what counts. Arriving in Taiwan, I wasn’t happy with the way that the written language seemed to be much more important than the spoken.

praatThe problem hasn’t gone away, however, and even in my current master’s degree program, I feel that teachers have very strange ways of evaluating proficiency. I have also heard numerous students complain about their current language courses.. In some cases, a teacher claims to evaluate skill A, but is in fact evaluating skill B.

Case in point: Evaluating someone’s pronunciation by having them read very difficult characters they might never have seen before. If a student fails this test, you don’t know if it’s because he lacks skill A (pronunciation) or skill B (character recognition). Thus, it’s a very poor method to evaluate a student’s pronunciation. The same goes for reading aloud in Chinese; if the student fails, you have little or no idea what’s wrong, it could be because of problems with reading speed, character recognition, grammar or any number of other problems.

We need to know what the problem is before we can fix it

If we’re going to improve our pronunciation (or anything else for that matter), we need to know what we’re currently doing wrong. If we don’t know that, we might make small improvements just by practising more, but in many cases, you might actually perpetuate bad habits because you aren’t even aware that there are problems.

If we know what the problem is, we can make a concentrated effort to solve it. We can read about the problem, we can have other people help us and so on. Thus, I think it’s essential that Chinese teachers use proper methods to evaluate student pronunciation.

Methods to avoid

Let us look at three methods I have encountered that aren’t very good:

  • Reading tongue twisters might be great fun, but it’s not a good way to assess pronunciation. These phrases are designed to trip up native speakers, which means that apart from containing sounds you might find difficult to pronounce, they are objectively hard to pronounce, meaning that they contain rapid changes between phonetically either very similar or very different sounds. I would never ever make an l/r switch in any language, but if presented with twenty syllables only consisting of these two initials, I might still make a mistake. Does that mean that I need to work on my rs and ls? No!
  • Reading difficult characters seems to be a popular way of evaluating the pronunciation of native speakers. The pronunciation test teachers in Taiwan have to take includes a part with single characters and disyllabic words. These characters are not common (meaning that I have native classmates who can’t read all of them). To be honest, I have no idea what this has to do with pronunciation. This is character recognition plus the ability to remember the sound and tone of those characters, it’s got nothing at all to do with if you can actually pronounce the sounds or not. If pronunciation is the goal, use very easy characters and/or Pinyin.
  • Reading unfamiliar text aloud is very hard, especially in Chinese. Reading aloud involves many skills and actual pronunciation isn’t likely to come very high up the priority list for most second language learners. As frequent readers might remember, I have written an article about this. In short, reading aloud is only a good method to test pronunciation if you’re allowed to preview the text beforehand or if you read texts that are really, really easy.

How to test pronunciation

I’ve spent the majority of this article bashing existing methods, which is a bit unfair if I don’t provide any guidelines for what to do instead. In fact, testing pronunciation is relatively easy, which makes the above methods seem even more quaint. Here are a few things to keep in mind:

  • For the basics, test only pronunciation (provide Pinyin)
  • Be familiar with the content (preview) or use very easy texts
  • Be familiar with the language structures used
  • Start from single syllables, but move to disyllabic words as soon as possible
  • Learn basic tones before you start meddling with intonation

How to evaluate the results

When you have a recording of your pronunciation, there are at least three ways you can receive feedback. Doing all three is of course the best if you have the time and the resources to do so.

  1. Give it to a native speaker without telling him/her what it’s about: If they can understand what you’re saying, you know that you’ve got basic communication right. This is very easy for long sentences and very hard for single characters, because context makes it easier to guess what you’re saying. Thus, this method is best used for single syllables, disyllabic words or minimal pair bingo.
  2. Give it to a teacher along with a transcript: The goal here isn’t merely to check if the teacher can understand what you’re saying, instead you want the teacher to actively look for mistakes. Of course, these mistakes can still be sorted into different levels (wrong, understandable but not perfect), so ask the teacher to focus on the more serious errors first.
  3. Listen to it yourself and note any problems: While we’re speaking Chinese, it’s very hard to monitor everything at once and a lot of attention is spent on word choice, word order, what the other person is saying and so on. When you listen to a recording of yourself speaking, you will be able to pick out many problems you didn’t notice when speaking. I’ve written more about why and how to do this here: Recording yourself to improve speaking ability.


What I have written about in this article is relevant for both teachers and students alike. As a student, you have to realise that your teacher might not have the time or the ability to help you with your pronunciation and that you need to take responsibility yourself. As a teacher, you should try to be aware at all times what you are actually testing. Finding out what the problem actually is the first step of any sensible plan for improving!

Using Audacity to learn Chinese (speaking and listening)

I have briefly mentioned that I use Audacity quite a lot (Recording yourself to improve speaking ability), but the more I use the software, the more I realise how awesome it actually is. Audacity is your best friend when it comes to recording yourself, mimicking others, manipulating recordings, managing media and recording things you aren’t supposed to record. It’s also free of charge and can be  installed on most operating systems.

In this article, I will introduce several useful functions and show how they can be used to learn (or teach) Chinese. However, this isn’t meant to be a manual of how to use Audacity, so even though I will show you how to do certain things, please refer to the official website for help and support. I’m a language teacher and learner, I learn only what I need to learn about the technical details.

Audacity can be found, read about and downloaded here: Audacity (official page at SourceForge)

The basics

audacityBefore we get into any details, let’s look at what Audacity is. This is from the official about page:

  • Record live audio.
  • Convert tapes and records into digital recordings or CDs.
  • Edit Ogg Vorbis, MP3, WAV or AIFF sound files.
  • Cut, copy, splice or mix sounds together.
  • Change the speed or pitch of a recording.
  • And more! See the complete list of features.

This is what we will look at in this article:

  1. Recording from any source
  2. Enhancing the recording
  3. Repeating or slowing down the audio
  4. Mimicking and recording
  5. Saving, editing, and exporting

This is a video I recorded of these six steps. It contains only sparse commentary, so read the rest of the article for more details. Obviously, you can do much more than this with Audacity, this is just a small demonstration.

Record from any source, record what you hear

Audacity can be set up to record anything you hear from your computer. This might be different depending on your operating system, but the general idea is to set Audacity’s input to “stereo mixer” or similar. WHen you press “record”, Audacity will register anything on your computer’s line out. Thus, if you find it hard to extract audio from a YouTube clip or from a movie you’re watching, use Audacity!

I use Linux and for me it’s a simple matter of changing the input settings in Audacity. If you use other operating systems, you can start here or simply search for “Audacity record playback” + [your operating system].

Audio recording enhancement

Apart from this, Audacity is your best friend when it comes to editing and manipulating recordings of various kinds. I sometimes record lectures or similar. I typically need two things to handle this kind of recording:

  1. Noise removal and compression
  2. Cutting and editing
  3. Automation

The first part is very complicated and I guess there are people who are actually earning their living from enhancing sound files, but we can do some basic but yet very effective things with Audacity. Noise removel is mostly a matter of trial and error, just use the function in the program and try different levels (the default ones to start with, obviously). Audacity’s compression function allows you to change the intensity of the recorded audio, removing high spikes and distributing the rest of the sound in a neat way.

Cutting and editing is fairly straightforward. Since you can actually see the audio, it’s a lot easier than trying to record from recordings or whatever else people do if they don’t know about Audacity or similar programs.

Automation is fairly complicated and I don’t know even a single percent of what there is to know, but I still want to point out that there is something called “chains” in Audacity that allows you to apply the same functions to any number of files. For instance, if you record twenty lectures in the same environment, you can use the same noise removal and compression settings for all files and you can apply these functions to all the twenty recordings with just one click. You can even make Audacity save the results as new files in the file format of your choice.

Using Audacity to mimic native speakers

Mimicking native speakers is one of the most powerful ways of acquiring good pronunciation in any language. However, it’s not always practical to do so. If we listen to a YouTube clip, the interface simply doesn’t allow us to repeat exactly what we want to repeat and even if we have a sound file, it would take ages to use a normal media player to be able to mimic a few minutes of speech.

In audacity, this is fairly easy:

  1. Import or record audio
  2. Select the part you want to mimic
  3. Click play and only the section you want to play will be heard
  4. If you hold down shift while clicking play, the section will repeat

This is useful because it isn’t very easy to mimic native speakers at their normal rate of speech, not even for advanced learners. Just listening to the same sentence a dozen times before even trying is good start.

The next step would be to record your own voice over the voice of the native speaker. After having practised until you can read a sentence or passage, simply hit record and Audacity will play the audio while recording your voice. You can the mute the original audio and evaluate your own recording. More about this below.

If you’re interested in either mimicking or the 蔣勳 clip seen in the video, I suggest you read Jacob Gill’s article about how he used that very same clip to improve his pronunciation (we did this at the same time, although I didn’t finish the entire clip and didn’t publish anything about the results).

Slow down the rate of speech without changing the pitch

Some media players can slow down the speed of the audio, but while doing so, the pitch also drops. Thus, we all sound like drunkards at half speed and like smurfs on illegal substances on double the speed. Audacity has a function called “change tempo”, which allows you to change the speed without changing the pitch. This allows you to slow down the rate of speech to a level you’re more comfortable with. Obviously, if you slow things down too much, you will get weird results.

Recording yourself

One very good way of improving your pronunciation is to record yourself. This fulfils several purposes at once:

  • You can share the recording for comments and feedback
  • You can listen to the recording yourself (this is actually very useful)
  • You can use it as a benchmark and see your improvements later

These concepts have already been discussed in more detail in other articles, namely Recording yourself to improve speaking ability and Benchmarking progress to stay motivated. However, there is one more aspect of recording yourself I think is worth mentioning:

How do you use audacity?

Do you have any favourite functions? Do you use audacity to learn or teach Chinese in a way that I haven’t mentioned here? Leave a comment!

Recording yourself to improve speaking ability

When we speak Chinese, we need to keep track of many different things, and even if we can hear what we’re saying, most of our energy is still spent on producing correct language rather than monitoring that language. This doesn’t mean its impossible for us to hear our own mistakes, just that we can never hope to hear all the mistakes we make while speaking at the same time. This is good. When speaking, some self-correction is helpful, but an overly critical approach to our own speech is counterproductive and tends to leave us tongue-tied.

Use Audacity to record and compare.

Recording yourself alleviates this problem

If you record yourself, you can listen to what you’ve said afterwards, fully concentrating on your language production, trying to find things to improve. You’ve already said what you wanted to say, now you can focus solely on correcting yourself. This is useful for several reasons:

  • It’s free (you need fewer hours with a tutor)
  • It’s convenient (you can do it at home at any time)
  • It increases understanding (knowing yourself what’s wrong is much more powerful than having someone else point it out to you)
  • It enables comparison (if you have a model to follow, you can compare it with what you have recorded)
  • It’s a benchmark for future reference

It’s only natural that you feel that your voice sounds weird when recorded, because the sound you hear when you speak normally is not only heard from the “outside” and is thus different from what everybody else hears (and that’s what’s recorded, of course). Still, this is something you will get used to very quickly if you just listen to yourself a few times.


If you have free access to tutors, of course that will take you farther than you will ever get on your own. However, even if you have a tutor, record what you’re saying! The benefits of recording yourself doesn’t disappear  just because you have a teacher. Hearing yourself is still important.

What should you record?

You can record anything you like, but here are some suggestions that I find useful myself. Try recording yourself when you…

  • …read a text in Chinese
  • …talk with a friend
  • …talk in class 
  • …hold a speech
  • …talk on the phone

If you’re going to keep this for future reference, I strongly suggest naming the files properly and organising them in such a way that you can later clearly see what you’ve recorded, how and when it was done.

What program should I use?

I recommend Audacity. It’s cross-platform, open source and generally awesome. Audacity has a number of useful functions, including visual representation of the sound recording, smooth editing of sound channels and different sound tracks. If you want to combine, edit or cut sound, this is the program for you. You can also export to mp3 and record information about date, type of recording and so on.

Mimicking native speakers

One of the most powerful ways of learning pronunciation is mimicking native speakers. Find a recording of a suitable piece of text and simply try to mimic it as closely as you can. Then listen to both versions and see what you can improve. Repeat until you think that the recordings are identical (save for differences in voice quality). Then ask your teacher or a native speaker to check the recordings for you and see if there is anything more you can do. Repeat with a different sound file. If you’re a beginner, I suggest using your textbook, but more advanced users can of course use anything you like, but make sure you check the file with a native speaker who’s judgement you trust (a teacher, preferably).


Recording one’s own voice is useful for more reasons than I have room for in this article and I think it should be a natural part of both learning and teaching. As a teacher, I often record students to see if they can hear their own errors. Most often they do when recorded, even if they can’t hear them when they speak at the same time. This has lead me to use recording extensively to learn/teach both Chinese and English. To give you an example, I can hear lots of mistakes in this video I recorded on YouTube (in English), even though I thought it sounded pretty good when I recorded it. If I wanted to improve my English pronunciation, I would record myself much more.

I don’t need a tutor to improve, I can do it on my own if I record and listen, because I can hear the mistakes. So can you. You might not be able to hear all mistakes, but why not try and see? I’m sure you will learn more about yourself and your pronunciation.

Benchmarking progress to stay motivated

When we set out to learn Chinese, everything we learn is new and we can feel that we improve for each day that goes by, for each time we are exposed to the language. We know this because, in relative terms, we’re learning so much. However, as we gradually advance beyond the basic level and enter the intermediate and advanced stages, this feeling is inexorably weakening and finally all but disappears.

This is a problem, because if we don’t feel that the effort we invest pays off what will then prompt and motivate us to continue learning?

Image credit: Pam Roth (

 Even though there might be satisfying answers to that question, in this post I will argue that we needn’t even ask the question in the first place if we pay more attention to our progress. I suggest using benchmarking, not to compare ourselves to other learners, but to compare with ourselves at different stages of learning. It is usually quite easy to know that we have indeed made progress if we compare our  language proficiency now with that of the past.

Define your current level for further reference

The first thing we need to do is record our current level in the areas we are interested in. Measuring language ability is notoriously difficult, but as long as we keep the parameters as constant as possible, that will be good enough for our purposes.

Basically, we will device a way to describe our  own ability as best as we can. Then, in a month or a year, we can compare and see if there is any difference. By doing so, we will see that we improve and we will get solid proof that we are indeed still moving, even though it might sometimes feel that we aren’t. Naturally, we might stop moving if we don’t study, but that goes beyond the scope of this article.

Even though this isn’t a scientific undertaking, a scientific approach is essential. For instance, we need to keep the variables to a minimum and make sure that our own ability is the major deciding factor. This means that if we measure reading speed, we should use a text with the same level of difficulty and belonging to the same genre. When we check our current level, we need to make sure to also note how we performed the benchmarking. If we’re going to do this again in six months, it’s likely that we will forget some important parts which make the comparison unfair.

As we will see, benchmarking isn’t straightforward in all cases and requires some further illumination and helpful examples. We’ll start out with the easier kinds of benchmarking (reading and listening) and then move on to more tricky areas (writing and speaking).

Benchmarking reading ability
Reading ability should be the easiest area to benchmark, simply because we can do it entirely on our own without using any fancy equipment. Reading can of course be tested in many ways and here are some goals we can strive towards and measure our progress:

  • Reading easy texts for full comprehension
  • Reading harder texts to understand the gist
  • Reading texts belonging to different genres

These goals can also be combined, but since it would be far too time consuming to try all combinations, we should try to find a few that match our  current goals for learning Chinese. Here are some things we should pay attention to:

  • How often we need to use a dictionary
  • How much time we need to complete the text
  • How we feel when reading the text

Benchmarking listening ability
Listening ability is also a passive skill that we can test on our own. Simply find a suitable audio source and see how much you understand now and compare that with a result in the future. listening ability can incorporate a number of different skills, so here are some distinctions to cover a larger portion of listening ability:

  • Listen to informal vs. formal Chinese
  • Listen once vs. listening many times
  • Listen to a standard dialect vs. regional dialect
  • Listen to texts read aloud vs. natural conversation
  • Listen to content with different topics

Again, since the possible variations are almost limitless, we should examine our goals for learning Chinese, and pick a few that seem the most relevant. Benchmarking is not meant to assess language ability fully accurately, but rather to give us a general idea of where we are. Again, make sure to detail how you perform the benchmarking for future reference.

Benchmarking writing ability

Leaving the more passive realm of listening and reading, it becomes a bit more difficult to benchmark, partly because we cannot or should not rely entirely on ourselves to do the evaluation. Still, writing is in its very nature something permanent that can be accessed at any time. There are many ways to benchmark writing, but I recommend starting a blog in Chinese.

Using a blog, we can have all written Chinese concentrated in one location, easily accessible both to you and others (you will have to share what you write anyway if you want people to comment on what you have written). Exposing your own writing might feel a bit anxious in the beginning, but as I’ve stated elsewhere, be brave and remember that genuine mistakes aren’t necessarily bad.

Try writing texts of different kinds:

  • Write texts in different styles (formal/informal)
  • Write texts on different subjects
  • Write text with or without the help dictionaries
  • Write texts with or without helping friends/teachers
  • Write texts by hand or type using a computer
  • Write in different genres (fiction/discussion/accounts/explanations)

When it comes to comparing texts, there is a certain amount of work that we  can and should do on our own. The obvious thing is of course to compare the level of the texts themselves. Don’t forget to take the circumstances into account here; it’s grossly unfair to compare a polished texts you’ve written with the help of a dictionary and that has been corrected by you teacher, and a text that you’ve written in five minutes by hand without having access to any help.

Still, comparing texts might only give us part of what we’re after, or we may find that we don’t think that things have changed that much. Simply speaking, it might be hard to evaluate our own writing. It’s time to ask a friend or a teacher:

  • Ask a native speaker to compare texts of a similar type
  • Ask a native speaker to comment on what kind of mistakes you do
  • Ask a native speaker to give you an overall judgement of your situation
  • Ask a native speaker for advice on how to improve

Since we want to keep the number of variables to a minimum, it’s strongly suggested that you use the same native speaker (or more than one, if you can). Remember that writing is easy to store, so you can simply select texts you want to compare and send to someone for comparison. You don’t even need to tell them you’ve written all the texts yourself, just ask them to do an honest comparison focused on language.

Benchmarking speaking ability

It’s now time to turn to the fourth and by far the hardest of the four areas we want to benchmark: speaking. As opposed to writing, speaking is in nature evanescent and we need quite advanced technology to record it. This is essential, because there is otherwise no way we can make honest comparisons.

Sure, we can meet a native speaker, talk to him or her, wait a year and then do the same thing. This will yield some results, but will it be enough? Does the native speaker remember how you talked a year ago? He or she probably remembers the feeling of speaking with us, but that’s only half of what we want.

Instead of doing this, I suggest record speech digitally. Again, we should record different kinds of speech, focusing on different aspects:

  • Record your best performance (prepare, record, record again if needed)
  • Record your average reading (just read a text, no second chances)
  • Record improvised speech on a prepared topic
  • Record improvised speech on a random subject

Actually, you can start a blog for this as well, doing a podcast or a video diary or something like that, but some people may find this a bit too exhibitionistic (add a password). The important thing is that we keep our recordings for later. As is the case for writing, we can compare our own recordings, even though that will probably not be enough in itself.

In my experience, this is extremely valuable and I can usually hear many mistakes I didn’t think about when I actually did the recording. Still, that is not the issue here. At some point, we have to include a native speaker. Do the same as you did for benchmarking writing:

  • Ask a native speaker to compare recordings of the various kinds above, preferably a native speaker who isn’t used to hearing you speaking Chinese
  • Ask a native speaker to listen to the various recordings with focus on different aspects (fluency, accuracy, pronunciation, tones, whatever)
  • Ask a native speaker for advice on how to improve

Benchmarking speaking is difficult not only because it involves technology, but also because most people find it unnerving to listen to their own voice. You have to get used to that and you will if you only allow yourself to try. You voice only sounds weird to you the first couple of time. Again, be brave. You don’t need to publish your recordings if you don’t want to, but recording yourself in some way is essential.

Some final remarks

It’s essential that you organise your benchmarking files and keep things tidy. If you’re going to compare texts or recordings that are months or even years apart, you’re going to need a neat structure that allows you to find the files later on. If you think that this sounds like too much work, remember that all the listening, reading, writing and speaking you’re doing is as good practise as anything else.

Good luck!