Mandarin Tone Length: What 280 Recordings Show
Length is a real cue for one Mandarin tone and nearly useless for the other three. Measured across the 280 single-syllable recordings Tingyin plays at levels one and two, 70 per tone, the third tone averages 1.33 seconds and the other three average between 1.01 and 1.08. The third is a third of a second longer than the fourth. The first tone, which learners are usually told is the long one, is 40 milliseconds longer than the fourth, which is a gap nobody can hear.
Where these numbers come from
Tingyin is a Mandarin tone-listening trainer: a clip plays and you pick the tone, and that is the whole exercise. Every clip it ships is a human recording that carries its source, its speaker, its licence and a hash of the audio, so the corpus can be counted rather than estimated. That is what makes this measurable at all.
The set used here is the single-syllable half of the corpus: 280 clips, exactly 70 for each of the four tones. 279 of the 280 come from one speaker, which for this particular question is a feature rather than a limitation. Comparing tone durations across speakers means comparing speaking rates as well; within one voice, tone is the only thing that varies.
All 280 sit at levels one and two, which is the free part of the app, and level one needs no account at all. Anything below can be checked by ear rather than taken on trust.
The four durations
| Tone | Mean | Median | Range | Clips |
|---|---|---|---|---|
| Third | 1.330 s | 1.344 s | 0.90 to 1.63 s | 70 |
| Second | 1.083 s | 1.080 s | 0.89 to 1.22 s | 70 |
| First | 1.052 s | 1.032 s | 0.89 to 1.30 s | 70 |
| Fourth | 1.012 s | 0.984 s | 0.89 to 1.22 s | 70 |
Three of the four are within 71 milliseconds of each other. One is not.
Length identifies the third tone
The third tone runs 318 milliseconds longer than the fourth on average, which is a 31 per cent difference and is not subtle. Fifty of the 70 third-tone clips are longer than the longest fourth-tone clip in the whole set. Pick a third-tone clip and a fourth-tone clip at random and the fourth is the shorter one 97 times in 100.
That is a usable rule. If a syllable in isolation feels like it is taking its time, it is a third tone, and you can commit to that before you have worked out anything about its pitch.
The reason is in the shape. A full third tone dips to the bottom of the range and comes back up, and that dip takes time to execute. A tone that has to travel down and back cannot be produced as fast as one that only falls. Length is not an extra property of the third tone; it is what the contour costs.
And it barely helps with the fourth
The fourth tone is the shortest of the four, which is true and is where the usual advice stops. The gap against the first tone is 40 milliseconds. Pick one of each at random and the fourth is the shorter one 55 per cent of the time, which is a coin toss with a limp.
So the advice to listen for a clipped syllable, which is everywhere and which this blog has printed itself, works only against the third tone. Against a first tone it is noise. If you are guessing fourth because a clip felt short, you are guessing on a signal that is not there.
What does separate a fourth in isolation is the drop itself, and after that the attack: it arrives loud and lands hard. The four contours are the thing to listen to, and length is a shortcut that only opens one of the four doors.
What this is not
These are citation forms. A single syllable, spoken alone, deliberately, into a microphone, for a recording. That is a specific way of speaking and it is not conversation.
- In running speech everything compresses, and the third tone compresses hardest, because before another third tone it is not a full dip at all and before anything else it is a half third that never comes back up.
- The numbers are one speaker’s. Another voice would give different absolute durations. The ordering is the claim, not the milliseconds.
- This is the duration of a drill corpus, not a measurement of Mandarin. It describes what you hear in the app, which is exactly the situation the advice above is for.
The honest version, then: length is a cue that works in a drill on isolated syllables and decays as speech gets faster. That is not a small thing, because a drill on isolated syllables is where most people build the categories in the first place.
Using it without leaning on it
Use length as a tiebreak, never as the first question. The order that works is: did the pitch move, which way, and only then how long did it take. Hearing the movement is the skill; length is the thing you fall back on when two candidates are still standing and one of them is the third tone.
There is also a trap in it. A learner who leans on length will do well on levels one and two and then stall, because at level three the clips are two syllables and at level four they are longer still, and a syllable inside a word does not get to take its time. The cue does not transfer. It is a scaffold, and scaffolds are supposed to come down.
Frequently asked questions
Which Mandarin tone is the longest?
The third, by a wide margin. Across the 280 single-syllable clips Tingyin ships at levels one and two, third tones average 1.33 seconds against 1.05 for first tones, 1.08 for second tones and 1.01 for fourth tones. The dip to the bottom of the range and back is what costs the time, so the length is a consequence of the contour rather than a separate fact to memorise.
How do I tell a first tone from a fourth tone?
By the movement, not by the length. A first tone holds a steady pitch and a fourth tone falls from the top of the range to the bottom, and in isolation the two are only 40 milliseconds apart in duration, which is not something a listener can use. Tingyin puts all four tones in the same drill at level one for that reason: the confusion is about pitch direction and no amount of attention to timing will fix it.
Is it safe to guess the tone from how long the clip is?
For the third tone, yes, and for the rest, no. A third-tone clip is longer than a fourth-tone clip 97 times out of 100 in this corpus, while a fourth is shorter than a first only 55 times out of 100. The second number is a coin toss, and a learner using length as a general rule will get the third tone right for the wrong reason and the other three wrong for no reason.
What is the difference between tone length in a drill and in real speech?
A drill plays citation forms, which are syllables spoken alone and deliberately, and real speech compresses all of them. The third tone loses the most, since before another third tone it changes into a second and before anything else it becomes a half third that never rises. Tingyin moves to two-syllable clips at level three for that reason, and a cue built purely on length stops working there.
Hold on, is a third of a second really something I can hear?
Comfortably, yes. Three hundred milliseconds is roughly the length of a spoken syllable in English, so the difference between a third tone and a fourth tone in isolation is about one syllable of extra time, and the ear is very good at that scale. The 40 milliseconds separating a first tone from a fourth is a different order of thing and sits below what timing perception reliably resolves.