Minimal Pairs That Matter: shi/xi, chi/qi, an/ang, en/eng

Core examples: 是/西, 吃/七, 山/商, 安/昂, 人/仍, 朋友/丰收, 想/显. Recommended feature module: Randomized listening tests with mouth-position diagrams, slow/normal audio, and sentence-level contrast checks. Related internal articles: 040, 041, 042, 043, 044, 058, 063.

Not every minimal pair deserves equal practice time

A minimal pair is a pair of words that differ by one sound and have different meanings. Language classrooms love minimal pairs because they make contrasts visible:

是 shì      西 xī
吃 chī      七 qī
山 shān     商 shāng
人 rén      仍 réng

But not all minimal pairs are equally useful. A pair can be technically neat and practically rare. Serious learners should prioritize contrasts that affect frequent words, names, numbers, directions, social interactions, and everyday comprehension.

This article focuses on four high-value Mandarin contrast zones:

1. shi / xi
2. chi / qi
3. an / ang
4. en / eng

These are not the only hard pairs. They are simply common enough to deserve targeted practice.

The principle:

Practice contrasts where mistakes survive into real sentences.

If you can say the pair slowly but lose it in conversation, you do not own it yet.

1. shi vs xi: not English “she” twice

English-speaking learners often hear shi and xi as variations of “she.” That is a problem.

Do not solve this with the folk rule “curl your tongue.” Some speakers do not visibly curl dramatically, and over-curling can make the sound muddy. The practical distinction is that shi belongs with the retroflex series zh/ch/sh/r, while xi belongs with j/q/x, a high-front tongue-body series.

Useful contrast set:

Sentence drills:

这是西边。
Zhè shì xībiān.
This is the west side.

老师洗手。
Lǎoshī xǐ shǒu.
The teacher washes hands.

这件事很细。
Zhè jiàn shì hěn xì.
This matter is very detailed.

If 是西 becomes one fuzzy “shishy” sequence, slow down and separate the two places of articulation.

2. chi vs qi: not one “chee” with two spellings

chi and qi are even more dangerous because English speakers may pronounce both like “chee.” Mandarin does not.

Both are aspirated. That means the difference is not simply “more air vs less air.” The difference is also place of articulation and vowel environment.

Contrast set:

Practice sentence:

七点吃饭。
Qī diǎn chī fàn.
Eat at seven.

This sentence is brutal in a useful way. 七 qī and 吃 chī are both high-frequency and adjacent. If they merge, the sentence becomes less clean immediately.

Another drill:

他迟到了，可是七点就起了。
Tā chídào le, kěshì qī diǎn jiù qǐ le.
He was late, but he got up at seven.

Now the contrast must survive tones, speed, and surrounding words.

3. an vs ang: final nasal contrast with real consequences

The difference between -n and -ng is not decorative. It distinguishes many common words.

Contrast set:

Learners often make two mistakes:

1. They pronounce -ang as -an because English final ng feels weak.
2. They add a hard final g, making shāng sound like “shang-guh.”

Mandarin -ng is a nasal ending, not a released g.

Sentence drills:

山上有商店。
Shān shàng yǒu shāngdiàn.
There is a shop on the mountain.

我很忙，但我想看一看。
Wǒ hěn máng, dàn wǒ xiǎng kàn yi kàn.
I'm busy, but I want to take a look.

他来采访，不是来反对。
Tā lái cǎifǎng, bú shì lái fǎnduì.
He came to interview, not to oppose.

The contrast must hold at sentence speed.

4. en vs eng: smaller but still important

en / eng can be harder to hear than an / ang, especially in fast speech or with tones. The difference again lies in final nasal position and vowel quality.

Contrast set:

Sentence drills:

这个人仍然在等。
Zhège rén réngrán zài děng.
This person is still waiting.

这本书很冷门。
Zhè běn shū hěn lěngmén.
This book is niche.

真的要争吗？
Zhēn de yào zhēng ma?
Do we really need to argue/compete?

This pair needs quiet listening. Do not practice only by shouting syllables. Record at normal volume and check whether the nasal ending is stable.

5. The order of practice matters

Many learners practice minimal pairs in the least useful order:

isolated syllable → isolated syllable → isolated syllable forever

A better ladder:

Minimal-pair success means nothing if the contrast disappears in phrases.

6. Self-diagnosis: what to listen for

Use this checklist after recording.

For shi/xi and chi/qi:

• Are shi/chi too close to English “she/chee”?
• Are xi/qi too retroflexed?
• Are you using the same tongue position for both columns?
• Can a listener distinguish 是西 and 七吃 in a sentence?
• Are you adding a strange vowel after zh/ch/sh-like syllables?

For an/ang and en/eng:

• Does -ng end nasally without a released hard g?
• Does -n actually close with the tongue tip?
• Are you reducing both finals to the same vague nasal?
• Does the contrast survive low volume?
• Can you hear the difference in someone else's speech, not just produce it yourself?

If you cannot hear it, production will be fragile. Listening and speaking need to train together.

7. High-value practice lists

shi / xi

是 / 西
十 / 西
事 / 细
试 / 洗
师 / 西

Phrases:

这是西边。
十个西瓜。
老师洗手。
事情很细。
试一试，洗一洗。

chi / qi

吃 / 七
迟 / 齐
池 / 奇
尺 / 起
持 / 其

Phrases:

七点吃饭。
他迟到了。
一起吃吧。
这个池子很奇怪。
支持其他人。

an / ang

山 / 商
安 / 昂
反 / 访
看 / 抗
满 / 忙

Phrases:

山上有商店。
我想看一看。
他很忙。
采访不是反对。
安全很重要。

en / eng

人 / 仍
本 / 崩
很 / 横
真 / 争
门 / 萌

Phrases:

这个人仍然在等。
这本书很冷门。
真的不用争。
门口有人吗？
他很认真。

8. Minimal pairs are not the final goal

A minimal pair is a training lens. It is not how people normally listen. In real Mandarin, context helps repair sound errors. If someone hears:

七点吃饭。

they can often infer the meaning even if 七 and 吃 are imperfect. But relying on context forever makes listening and speaking tiring. Your goal is not perfect laboratory pronunciation. Your goal is to make high-frequency contrasts strong enough that context does not have to rescue you every time.

A practical standard:

The contrast should be recognizable in a normal sentence at normal speed by a patient native speaker who is not looking at the transcript.

That is more useful than winning a slow minimal-pair drill.

9. Remediation matrix: contrast, cause, and correction

Minimal-pair work should not become random syllable chanting. This upgrade pass makes each contrast diagnostic.

The editorial rule: each pair must be tested in isolation, real words, short phrases, and unpredictable sentences. If a learner can pronounce a pair only while looking at the spelling, they have not acquired the contrast.

10. Articulation notes that avoid folk explanations

Do not tell every learner simply to “curl the tongue.” That advice helps some and harms others. Better instructions:

This table should be paired with a mouth diagram and slow audio, but the prose can prepare learners to listen for location rather than spelling.

11. Contrast survival drills under tone and speed

A contrast that survives only in first tone is not stable. Use rotating tones:

shī / xī      shí / xí      shǐ / xǐ      shì / xì
chī / qī      chí / qí      chǐ / qǐ      chì / qì
ān / āng      án / áng      ǎn / ǎng      àn / àng
en / eng in real words only: 很 / 横, 本 / 崩-like near contrasts where appropriate

Then put them into meaningful phrases:

是新的。 / 西新的。 [nonsense contrast for perception only]
我想吃。 / 我想骑。
这座山。 / 这个商场。
很安静。 / 很昂贵。

Nonsense contrasts are allowed for diagnosis but should not dominate practice. Real-word contrasts create better transfer.

12. Listening-first protocol

For each pair:

1. Hear A/B in slow speech with the label visible.
2. Hear A/B in normal speech with the label visible.
3. Hear random A/B without the label.
4. Choose the word from two characters, not from Pinyin.
5. Hear the same contrast inside a phrase.
6. Hear it inside a sentence with unrelated distractors.
7. Record yourself reading random prompts.
8. Ask a listener to transcribe characters, not evaluate “accent.”

This protocol prevents learners from passing a visual Pinyin quiz while failing a listening task.

13. Production fixes by error pattern

The final goal is not exaggerated contrast. It is contrast that remains available at normal speed.

The module should include three modes:

1. listen and choose — user hears one word and chooses from two characters;
2. sentence detection — user hears a sentence and marks which contrast appeared;
3. record and compare — user records the sentence and receives acoustic/teacher-style feedback.

The system should randomize tone, speaker, and sentence position. If every item is shì / xī in isolation, users memorize the quiz rather than acquire the contrast.

Feedback should say:

• “Your xi is drifting toward shi.”
• “Your -ang is ending as -an.”
• “You are releasing a final g after -ng.”
• “The contrast is clear in isolation but lost in the sentence.”

Reference anchors checked or recommended for this article:

• Standard Mandarin phonology references for retroflex, alveolo-palatal, and nasal-final contrasts.
• 普通话水平测试 materials, which explicitly test initials, finals, tones, tone sandhi, neutral tone, and erhua.
• Prior Inkuntri articles 040–043 on retroflex/alveolo-palatal contrasts, stop aspiration, ü, and difficult finals.
• L2 Mandarin pronunciation studies on high-error contrasts and tone/segment perception.

• Include audio from at least two standard speakers.
• Avoid using English sound labels as if they were exact equivalents.
• Add diagrams for tongue-tip vs tongue-body position and for -n vs -ng closure.
• Include both slow and normal-speed versions; slow-only audio gives false confidence.