Why does 3rd tone sandhi occur?

[Manuel]

I used to think the 3rd tone sandhi was just an annoying idiosyncrasy of the Chinese language, but today I've been thinking about it for a while and have came up with the following theory: Tone sandhi occurs naturally as a disambiguation mechanism.

1. Mandarin does not rely on absolute pitch to encode information, but rather on the relative pitches between adjacent syllables. This occurs naturally because say, women speak in a higher pitch than men, for example. Moreover, the overall pitch of a sentence may vary depending on the speaker's mood.
   
2. In Mandarin, a first tone is pronounced as a flat high pitch always.
   
3. In Mandarin, a single third tone is pronounced as a flat low pitch, except at the end of a sentence.
   

The key here is that both first and third tone are pronounced as a flat or constant pitch. Without third tone sandhi, two third tones in a row could therefore be mistaken for two first tones and vice versa, becasue absolute pitches are not taken into account. Therefore I suspect the language has naturally developed 3rd tone sandhi as a way to clarify and disambiguate this sort of scenario, and to make the third tone more conspicuous.

Anyway, that's just my theory. I'd love to hear people's opinions on this one.

[Ruben von Zwack]

It's an interesting question, but I don't second your frustrated feelings. Annoying? Look at what other languages do! Thank you, Great God of Mandarin, for creating only this one (that I know of) adorable little Sandhi

I guess someone who knows about the history of spoken Chinese can enlighten us on how this Sandhi "happened"?

[Manuel]

I don't find it particularly problematic now, but at the beginning It required some extra attention. The only problem is that, when you hear a word or expression for the first time, you may incorrectly assume you heard a 2nd tone when in fact it was a sandhied 3rd tone. This happened to me with 得 in 我得走了. I learn mostly by ear and I find the sequences 2nd-3rd and 3rd(sandhi)-3rd sound extremely similar. I read somewhere that the real 2nd has higher pitch than a neighbouring sandhied 3rd. This makes sense as the third tone is lower than the second tone to begin with, although it's not critical, so 3rd(sandhi) = 2nd is generally a safe assumption.

[roddy]

Try saying a few non-sandhi'd third tones in a row and then, once you've cleared the phlegm out of your throat, you'll have your answer.

[Shelley]

Have to agree with roddy, its there because it is not much fun saying third tones one after another. it is a practical solution to a small problem in a tonal language.

I don't think it is a complicated as you seem to have made it

[Manuel]

LOL it wasn't THAT complicated. In my opinion they could have chosen the first tone for sandhi i.e. 11 -> 41 and pronounced all third tones as low flats without sandhi. I still think sandhi is there to emphasize the difference between 1st and 3rd tones. However, like roddy said, it may be more comfortable to sandhy 3rd tones than 1st tones, hence the current state of affairs. So that's a good point.

[Michaelyus]

And yet it is more complicated than that.

1) Cantonese has a low falling tone (or super-low level tone). But there is no phonetic motivation to change a sequence of those tones (albeit their being rather rare) to any form of rising tone [except for the "changed tone" phenomenon, which is a derivation, not a form of allophony];

2) The native Mandarin of Tianjin also has a sandhi phenomenon, where a low rising tone 13 becomes a higher rising tone 45 before another low rising tone; in Taiyuan, a high falling tone 53 becomes a low level tone 11 before another high falling tone. Interestingly, Tianjin's low rising tone and Taiyuan's high falling tone both correspond to Standard Mandarin's third tone; and the post-sandhi tones all correspond to Standard Mandarin's second tone;

3) The second tone / third tone distinction is the last tone distinction to emerge in children's acquisition of Mandarin phonology;

3) "Wugs" for native Mandarin speakers sound different than real words for double third tone bisyllables, and generally have a lower rate of "successful" sandhi.

[skylee]

Why does the "why" matter?

[Manuel]

Because it's interesting

[Ruben von Zwack]

It's fascinating! I agree, it's always about making speaking and understanding easier.

As far as I'm aware of, by comparing the way people in different regions pronounce, is also how earlier stages of a language and the mutual connectedness of languages are researched.

[Demonic_Duck]

I don't find it particularly problematic now, but at the beginning It required some extra attention. The only problem is that, when you hear a word or expression for the first time, you may incorrectly assume you heard a 2nd tone when in fact it was a sandhied 3rd tone. This happened to me with 得 in 我得走了.

Don't see why that's a problem, 得 is the only "dei" syllable in any tone.

I learn mostly by ear and I find the sequences 2nd-3rd and 3rd(sandhi)-3rd sound extremely similar. I read somewhere that the real 2nd has higher pitch than a neighbouring sandhied 3rd. This makes sense as the third tone is lower than the second tone to begin with, although it's not critical, so 3rd(sandhi) = 2nd is generally a safe assumption.

I could believe that the 没 in 没有水 is higher than the 有, but I assume that's because it comes first, and that two 2nd tones in a row would display the same characteristic. It seems to me that the tone of a 2nd-3rd word is exactly the same as that of a 3rd-3rd. To be honest though, this is all just based on my own perceptions, so I could be wrong.

[Manuel]

That's true, but you know 得 is the only character pronounced dei in any tone because you have checked a dictionary. Most native speakers will only be able to tell you that they personally are not aware of any other characters pronounced dei. However, knowledge of this fact is of little consequence when you learn a language by ear: if the first time you hear 得 it happens to be a sandhied 3rd tone, anyone (inc. native speakers) will naturally assume it is a second tone, until "proven wrong" by another utterance in which the same word is pronounce like the original 3rd tone. This was my case.

[Demonic_Duck]

Ah, I guess that could be a problem for reproduction of the word then, I was assuming you meant it was a recognition problem.