Introduction

As l have experienced first-hand, Mandarin presents language learners with a particularly demanding set of challenges. They need to develop new sound categories, perceive and produce lexical tones, build sound–meaning connections, learn a large numbers of characters and develop an understanding of a vocabulary system whose morphology and writing system differ substantially from English.

This means that research specifically involving learners of Mandarin deserves our attention. General findings from cognitive psychology and second language acquisition are useful, of course, but Mandarin-specific research can tell us much more about what actually happens when learners encounter tones, characters, pinyin, spoken vocabulary and Chinese morphology.

The following eleven findings I extrapolated from literature I have been reading in preparation for an upcoming keynote for the Australian CLTANSW in September, have particularly important implications for classroom practice. They also connect in interesting ways with several principles underpinning Extensive Processing Instruction (EPI), i.e.: careful modelling, extensive receptive processing, controlled input, delayed production, retrieval practice, recycling, gradual fading of scaffolding and the movement from highly supported processing towards increasingly spontaneous language use. Of course, I am not claiming that these studies investigated EPI. They did not. The point is that their findings provide interesting empirical support for several of the learning principles on which EPI practice is based.

1. Focused listening practice can improve learners’ perception of Mandarin tones

Wang (2013) investigated beginning learners of Mandarin who received approximately six hours of focused tone training over three to four weeks.The learners improved significantly in their ability to perceive Mandarin tone contrasts. One finding I found very interesting- as it validates the EPI approach – was that perception-only training was effective. In other words, the learners did not have to spend all their practice time producing tones in order for learning to occur.

What does this mean for the classroom?

As I often reiterate on this blog, listening is not passive simply because students are not speaking. Consider the difference between asking students simply to listen to ten Mandarin words and asking them to decide:

Which tone did you hear?

Are these two words the same or different?

Which picture corresponds to the word?

Which word did the teacher say?

Which of two characters corresponds to the sound?

The second type of listening requires learners to make decisions about what they hear. They have to process the auditory signal closely.

The EPI connection

This fits very closely with the EPI emphasis on extensive receptive processing before extensive production. In EPI, learners are normally given substantial opportunities to process language through listening and reading before being expected to use it independently. The Mandarin evidence is particularly interesting because tone perception places such obvious demands on the learner.

Before repeatedly asking:

Can they say it?

we should perhaps ask:

Can they reliably hear it?

This is where the EPI listening techniques described on this blog and in the book Breaking the Sound Barrier: teaching learners how to listen can be extremely useful.

2. Being able to imitate a Mandarin tone does not necessarily mean that it has been learnt

Hao (2012) examined the perception and production of Mandarin tones by learners from different linguistic backgrounds.

Learners experienced particular difficulty with certain tonal distinctions, especially Tone 2 and Tone 3.

However, one especially interesting result was that learners could sometimes imitate tones more successfully than they could identify them.

This has enormous implications for classroom teaching.

Imagine this familiar sequence:

Teacher says the word.

Students repeat the word.

The repetition sounds excellent.

Teacher moves on.

What has actually been demonstrated?

Possibly very little.

The students may simply be copying an auditory pattern that is still present in their immediate memory. They have not necessarily constructed a sufficiently robust representation to recognise or retrieve that tone later.

What does this mean for the classroom?

We need to move beyond:

Hear → repeat

towards:

Hear → discriminate → identify → retrieve → produce

For example, after modelling several words, don’t immediately model them again.

Wait.

Then ask learners to identify what they hear.

Later remove the auditory model and ask them to retrieve the word themselves.

Then revisit it tomorrow.

And next week.

The EPI connection

This is almost a textbook illustration of why EPI distinguishes between modelling, receptive processing and independent production.

Choral repetition has a role. However, repetition immediately after a model should not be confused with mastery.

EPI deliberately provides substantial processing between initial modelling and freer production.

The support is gradually removed.

The crucial question becomes:

Can the learner retrieve the language when the teacher is no longer providing it?

3. Explicit instruction can improve Mandarin tone production

Wiener, Chan and Ito (2020) investigated explicit instruction and high-variability phonetic training in Mandarin tone learning.

Explicit information about tonal contours helped learners’ subsequent production of some tones, particularly rising and falling contours.

This is important because it challenges a simplistic assumption that learners simply need enough exposure and will eventually “pick up” Mandarin pronunciation.

Sometimes learners need their attention directed towards precisely what matters.

What does this mean for the classroom?

Before giving students dozens of examples, make the important feature clear and noticeable.

Explain it simply.

Model it clearly.

Contrast it with something different.

Then provide substantial practice.

The sequence might therefore be:

Explain briefly → model → contrast → discriminate → recognise → retrieve → produce

The explanation itself is not the learning.

It simply helps learners notice what they subsequently need to process.

The EPI connection

This fits the EPI principle of explicit modelling followed by extensive processing.

EPI does not reject explicit instruction. Quite the opposite.

However, explanation is kept purposeful and is followed by repeated encounters requiring learners to process the target feature.

Explanation directs attention. Processing builds the memory.

4. More voices and more variation are not automatically better for beginners

The same Wiener, Chan and Ito (2020) research produced another useful finding.

Exposing learners to several speakers did not automatically produce an overall advantage.

This matters because “authenticity” and “variety” are often assumed to be inherently beneficial.

Eventually, of course, learners need to understand Mandarin spoken by many different people.

But variability also makes the learner’s processing task harder.

What does this mean for the classroom?

A sensible progression may be:

clear and predictable → slightly variable → increasingly variable → naturalistic

rather than:

maximum variability from day one.

Initially, learners might hear a limited number of carefully articulated models.

Once the sound–meaning relationships become more stable, teachers can introduce:

different voices,

different speeds,

different surrounding language,

different contexts.

The objective is ultimately flexible comprehension, but difficulty needs to be calibrated.

The EPI connection

This connects strongly with narrow listening and controlled input in EPI.

EPI initially constrains vocabulary and structures so that learners repeatedly process a manageable linguistic system.

Variation is then introduced progressively.

This does not mean making language artificially easy forever.

It means:

stabilise first; vary later.

5. Difficult Mandarin tone distinctions can initially be made easier to hear

Cao, Pavlik and Bidelman (2024) investigated whether manipulating acoustic characteristics could improve learners’ perception of Mandarin tones.

They found that perceptual training improved tone identification and generalisation. Manipulating pitch range was particularly useful for aspects of the notoriously difficult Tone 2/Tone 3 distinction, although the effects differed according to the tone involved.

The broader finding is fascinating.

Sometimes making a difficult feature more noticeable initially helps learners develop the perceptual categories they need.

What does this mean for the classroom?

Teachers should not necessarily present beginners immediately with the most difficult possible version of Mandarin.We can temporarily exaggerate an important distinction, slow something down, isolate it, contrast two examples.Then we gradually remove those supports.

For example:

exaggerated contrast → clear natural contrast → natural speech → different speakers and contexts

The important point is that the scaffold is temporary.

The EPI connection

This is closely related to scaffolding and fading in EPI.Early processing is made highly accessible.As learners become more secure, support is progressively removed and processing becomes more demanding.The end goal remains natural language.We simply build a bridge towards it.

6. Chinese characters can sometimes interfere with learning the spoken word

Research by Zhang, Li and Liu (2021) provides one of the most counter-intuitive findings in this area.Adult learners learnt new spoken Mandarin words associated with pictures. Sometimes they also saw Chinese characters.The researchers manipulated how useful the characters were in signalling pronunciation and meaning. Characters containing partially misleading phonological information or semantically opaque information could impair recall of the spoken forms. Even useful orthographic information did not automatically produce better spoken-word recall.

What does this mean for the classroom?

More representations do not automatically mean more learning.Imagine presenting a beginner simultaneously with:

audio + character + pinyin + English + picture

Every element appears helpful.However, every element also competes for attention.If today’s objective is to establish a strong:

sound ↔ meaning

connection, it may sometimes be better to concentrate attention on that relationship before adding further representations.This does not mean delaying characters universally.It means sequencing information according to the learning objective.

Ask:

What exactly do I want the learner to encode right now?

The EPI connection

This connects strongly with EPI’s use of carefully controlled input and staged processing.The sentence builder, for example, is a scaffold rather than an invitation to process everything simultaneously.Teachers can temporarily foreground particular relationships:

sound → meaning

then:

written form → meaning

and later:

meaning → independent retrieval

The key principle underlying this approach is focused processing rather than information overload.

7. The way characters, pinyin and meaning are arranged on the page affects learning

Experimental Mandarin research comparing different layouts for characters, pinyin and English meanings has found that presentation format affects vocabulary learning. An adjacent presentation format, in which relevant information was physically close together, produced better learning than alternative layouts. Interestingly, what learners preferred was not necessarily what produced the strongest learning.

What does this mean for the classroom?

This sounds like a small design issue but it isn’t, actually. Every time the learner has to search across a page:

Where is the pinyin?

Which translation belongs to this character?

Which pronunciation corresponds to this item?

working-memory resources are being used for something that has nothing to do with actually learning Mandarin.So:

The EPI connection

This is directly relevant to sentence-builder design. A good sentence builder reduces unnecessary visual search and makes form–meaning relationships transparent.The purpose is not merely to make the worksheet attractive.It is to free cognitive resources for language processing.Good instructional design should make the language difficult enough to learn while making the interface as easy as possible to navigate.

8. More visual information can make Chinese characters harder, not easier, to learn

Hou and Jiang (2022) investigated 40 L2 learners of Chinese whose first languages used alphabetic writing systems.They examined two apparently helpful teaching techniques:

1. highlighting radicals,

2. and showing animated stroke order.

The results were surprising. Highlighting radicals increased reaction times and reduced character-recognition accuracy. Stroke-order animation also reduced recognition accuracy under the experimental conditions. The researchers interpret these findings partly in terms of cognitive load.

What does this mean for the classroom?

Imagine what a beginner may already be processing: shape, meaning, sound, tone.

Now add: coloured radicals, moving strokes, stroke numbers, pinyin, translation, pictures

We may believe that every additional feature provides another scaffold. The learner may instead experience something completely different: too much information.

I am not suggesting: Don’t teach radicals. Nor: Don’t teach stroke order.

My suggestion is: Don’t teach everything simultaneously simply because AI/technology allows you to. Try and resist the temptation and the suggestion of disinformed PD providers.

The EPI connection

This is highly consistent with EPI’s emphasis on managing cognitive load.

EPI deliberately constrains new language, recycles familiar material and gradually increases complexity.

The same principle should apply to visual presentation.

Less can sometimes produce more learning.

9. Understanding how Chinese words are built supports vocabulary knowledge

Zhang (2016) studied 171 learners of Chinese as a second language.

The study investigated morphological awareness: learners’ awareness of meaningful elements inside words and of how Chinese compounds are structured.

Both general morphological awareness and specifically Chinese compound-structure awareness predicted vocabulary knowledge.

This is an important finding because Chinese vocabulary should not be experienced as thousands of completely unrelated items.

Consider:

学生

学校

学习

The recurring 学 creates a connection.

Or:

电话

电视

电脑

Here 电 creates another lexical network.

What does this mean for the classroom?

When introducing vocabulary, don’t only ask:

What does this new word mean?

Also ask:

What does it remind us of?

Where have we seen this element before?

Which other words contain it?

Can you work out something about this unfamiliar word from what you already know?

Vocabulary learning becomes the building of networks, rather than the memorisation of isolated labels.

The EPI connection

This fits extremely well with intelligent recycling. In EPI, recycling does not mean simply showing exactly the same item repeatedly. Previously learnt material is brought back in new combinations and contexts. With Mandarin, morphological relationships provide an additional opportunity as teachers can recycle parts of words as well as whole words.That creates more connections in long-term memory and gives old learning a role in acquiring new learning.

10. Handwriting and typing may strengthen different aspects of Mandarin character knowledge

Shi, Hu and Zhu (2026) investigated beginner learners receiving substantial Mandarin instruction.The researchers compared handwriting and typing and found an intriguing pattern.Typing produced stronger performance on some measures involving character recognition and dictation, whereas handwriting produced stronger meaning-recognition performance. In other words, the two activities may strengthen somewhat different relationships within the learner’s representation of a character.

What does this mean for the classroom?

Based on the above findings, he question should not be: Is handwriting better than typing? That is probably too simplistic. Instead ask:

What do I want this activity to strengthen?

Do I want learners to recognise the character?

Retrieve its pronunciation?

Retrieve its meaning?

Produce its written form?

Connect sound, meaning and orthography?

Different activities may serve different purposes.

The EPI connection

This connects with an important EPI principle: practice should match the cognitive operation we ultimately want learners to perform.

Technology should not be judged by whether it is modern.

It should be judged by what it makes the learner’s brain do.

11. Retrieval practice improves the learning of Chinese vocabulary

Finally, a study which is particularly important for EPI. Harrington and Jiang (2013) investigated retrieval practice with university students learning Chinese. Over several weeks, learners completed relatively simple recognition activities involving Chinese words encountered previously.They had to retrieve information about whether particular forms had been encountered before.The words receiving this additional retrieval practice were subsequently learnt better than control items, with benefits evident in later vocabulary assessment.The significance is straightforward:

Seeing something again is not the same as retrieving it.

Suppose I learnt:学校last week.

The teacher can show me 学校 again today.

I look at it and think:

“Yes, I know that.”

That feels like successful learning.However, the answer was already in front of me.

Compare that with:

What was the Mandarin word for “school”?

Now I have to search memory.

That effort is important.

What does this mean for the classroom?

Every Mandarin lesson should contain opportunities to retrieve previously taught language. These can be extremely short:

What does this mean?

Which character did you hear?

Which word is missing?

Say the word before I reveal it.

Complete the phrase.

Translate this chunk.

Which phrase did we learn yesterday?

Write three expressions from last week from memory.

Listen and reconstruct.

Retrieve the missing character.

And crucially, retrieval should happen after delays.

Not only: five minutes later, but tomorrow, next week, three weeks later, next term.

The EPI connection

This is perhaps the clearest connection of all as retrieval practice sits at the heart of Extensive Processing Instruction. The EPI sequence repeatedly requires learners to process and retrieve high-frequency language through different modalities and with progressively decreasing support. Sentence builders provide substantial initial scaffolding.Then that scaffolding is gradually faded.

Learners move from recognition to cued recall to less-supported recall to independent retrieval

and ultimately towards increasingly fluent production. Activities such as delayed dictation, sentence puzzles, partial dictation, listening reconstruction, translation, retrieval grids and structured production are valuable not simply because they provide “practice”. They repeatedly force the learner to bring language back from memory. And this distinction is fundamental:

Re-exposure asks: “Do you recognise this?”

Retrieval asks: “Can you bring it back?”

For durable learning, Mandarin learners need both.

However, if we want language to be available tomorrow, next month and eventually in spontaneous communication, then at some point the answer has to disappear from the page.

That is ultimately what teaching for memory is about.

Conclusion

Taken together, these findings remind us that effective Mandarin teaching depends not simply on how much language learners encounter, but on how carefully we help them encode it, process it, connect it to what they already know, retrieve it repeatedly and retain it over time. This requires careful management of cognitive load, extensive receptive processing, purposeful recycling, repeated retrieval and the gradual removal of support as knowledge becomes more secure and increasingly accessible.

In an age in which AI and digital technology can provide learners with words, characters, pronunciations and translations within seconds, teaching for durable learning becomes increasingly important. Our role is to create the conditions in which Mandarin is encountered often enough, processed deeply enough and retrieved frequently enough to become part of the learner’s long term memory, so that the language they meet in our classrooms today remains available to them tomorrow, next month and, hopefully, many years from now.