How tone shifting actually works - How a speaker navigates the register space

Question

In considering the ways tone might work in languages, I am looking at diagrams of 5 rows (registers I'm guessing) in which you can create tones that shift up and down the registers. Some examples of the movement might be like:

--x------------x-----------------x------
   \          /                 /
----x-----------------------------------
            /                 /
----------o-----------------------------
                            /
----------------------u-----------------
                       \  / 
------------------------x---------------

That looks good in principle, because you have 5 rows and you shift up and down like on a ladder. But in reality, people don't really have perfect pitch and so they aren't going to realistically be attaching their tones to the rungs of the ladder perfectly. So this gets me confused.

I understand how you can, given a current position, move either up or down in the tone. That makes sense. But I don't understand how you can say "from this position I am going to go to the high tone", where you may have left off the last tone at level 4, but now are going to go from 3 to 5 in a glide. I don't see how you can gauge that you are at the desired register level, like shown in the standard Mandarin tone chart/graph.

I'm wondering if one could plot out an example sentence or two (made up, doesn't have to be a real language) showing how the tone changes like the chart above. Then it would be helpful if you could explain how the speaker can jump from position to position in the chart/graph, to perform the tone changes. At each point, like at the o in the chart above, I am wondering how the speaker knows they are moving down 1 register (from the first tone slide down, to the start of o), and then how they know what the distance is to go to the "highest" tone. Then to go from there to to the u in the diagram, that is jumping 4 levels down, it's hard for me to imagine how they could know they didn't accidentally go to the "lowest" level.

What does make sense to me is if it's all relative. Somehow the communicators know that the person speaking has a "base" tone level, or reference point. It's not out there in the world, it's just something they default back to after doing some tone shifts. So perhaps it would be the place where the most time is spent speaking. Then from there you can move up or down. If you move up, you stay there for a little bit, and then can move up or down again. If you move up, then now you are pretty far removed from the initial reference point, so wondering how the reference point is reestablished. If instead you moved down far, below the original reference point, then you would know (because of working memory) that you are in a lower level register. So you are constantly hovering around the neutral point, only occasionally shooting off a tone here or there and then coming back. In this way, it doesn't need to be very exact, you can just move up and down roughly coming back to the central place.

But the way the diagrams I've seem make it look is that you can, like a computer, jump from one register to the next:

5 -> 4
4 -> 3
3 -> 5
5 -> 2
2 -> 1 -> 5

If you get distracted, no problem, let me just jump back to 2 and continue where I left off (type of thing). That seems unrealistic. But that's what these IPA symbols convey:

˥
˦
˧
˨
˩

You can do like ˨˥˦ or any sequence you want. But it doesn't seem like it would actually work like that when speaking.

Not having experience in tonal languages (it would take a while to get the hang of it I imagine), I am just wondering if one could demonstrate the sequence of events a speaker goes through when performing tone shifts.

Answer 1

As a speaker of Mandarin Chinese, one of the most commonly studied and taught tonal languages with non-level tones, I can give you a few native intuitions as well as pedagogy.

So lets define a few terms and clarify a few things:

1) A tone is not a set frequency level (pitch, to musicians). A frequency is a physically measurable, quantifiable property of sound. A tone is something perceived by the listener within the system of the language.

2) A tone is a pattern across the frequencies. This is especially true if the tone is non-level. But this isn't as exotic as it sounds; vowels in general are patterns of frequencies too. The vowel equivalent of a non-level tone is a diphthong (if we want to get technical, the frequency formants change as the vowel shifts from one position to another).

3) Most people do not consciously think about what tone they are using, just as most people do not consciously think about whether they are using a closed vowel in the word LOT or an open vowel in the word. Pragmatics and semantics are more likely than phonology to be in the forefront of a speaker's mind when they are speaking, although it is possible (accent and intonation can be sociolinguistic markers).

4) Relative frequency perception is a thing. Men have lower voices than women on average. Children have higher voices than adults on average. Nobody says Mandarin tone 1 = 444 Hz. So listeners adjust what their perceptions are based on what the speaker is producing. However, this is not exclusive to tonal languages: non-tonal language listeners use the same pitch information from the speaker when processing intonation.

Most Chinese varieties have contour tones, where the tone patterns are different from each other. Mandarin is an extreme version, where every single tone of the four is a different shape to the other (in practice though, tone 3 is usually low level or low falling in actual speech rather than the canonical falling-rising, and there is also tone neutralisation). Cantonese has more register tones, with a low level, mid level and high level, as well as contour tones in addition to that. Yoruba on the other hand is primarily composed of register tones: high, low and mid, and any perceived rises or falls are allophonic variations of these register tones.

What does that mean for your question? It means that people just learn it depending on the system. Natives simply have to learn it through exposure. Speaking practice helps native children (in HK Cantonese at least).

You mention tones going 5-4, 4-3, 3-2. If you are referring to several phonemic tones, this sounds like several slightly falling tones, which sounds to me like a very difficult set of contrasts for the listener to perceive. I don't personally know any language like this (Fuqing dialect probably closest, although it has more contrasts on top), but I don't see anything mechanically difficult with producing several falling tones like these (a high falling, mid-falling, mid-low falling). After all, just as an /a/ feels different in the mouth to an /i/, a high falling feels different in the throat to a low falling.