I didn't know the pronunciation was simplified with the creation of pinyin. That is really interesting. Could you give an example (e.g. two words or sounds that used to be different and are now pronounced the same)?
And I think that's a great way to visualize it. Each word has it's own tone contour and it just so happens that a lot of them could be clustered and curve-fitted to match a small number of tones. But the number of curves you choose is arbitrary just like how choosing the number of clusters in k-means is not well defined.
So if you treat the pinyin as authoritative you will sound robotic compared to a native speaker who learns each word's actual tone curve naturally. Just like how if you believe english is a phonetic language you will pronounce lots of things wrong. (e.g., as an English learner once I learned how to spell "doubt" I kept trying to enunciate the "b" since I thought the pronunciation I learned from hearing, "dout", was wrong)
Disclaimer: I'm a native cantonese speaker so my knowledge about tones is bullshit. Just hypothesizing why text-to-speech and people learning to speak it always sound funny.
I am a native speaker so I don't really care about the number of tones, I was just citing the link. You can speak to me in monotone and I will figure it out from context just like how you can learn to understand someone with a strong accent.
My theory above is that a fixed number of tones is a linguist madeup concept from data that happens to fit the clustering of the tone graphs. The real tone is a distribution on a per word basis that you learn from natural variations from the other speakers around you. Do the 6 or 9 or 4 tones sometimes fit these distributions? Yes, but it's still just an approximation of the real variations that natives use and shouldn't be treated as an authoritative pronunciation.
Well, even languages with no defined tones such as Japanese have some intonations to distinguish homophonic words in context. And English speech synth still sound robotic because it lacks the full tonal range of a human speaker.
The simplification of tones in Northern Mandarin predates Pinyin by hundreds of years, similar to how English developed a writing system that is often inconsistent with speech.
Doesn't pinyin just apply to standard mandarin? China has many different dialects with different pronunciations, pinyin wouldn't have simplified those at all.
https://en.wikipedia.org/wiki/Cantonese_phonology#Tones
Particularly useful in that section is this graph:
https://en.wikipedia.org/wiki/Cantonese_phonology#/media/Fil...
And I think that's a great way to visualize it. Each word has it's own tone contour and it just so happens that a lot of them could be clustered and curve-fitted to match a small number of tones. But the number of curves you choose is arbitrary just like how choosing the number of clusters in k-means is not well defined.
So if you treat the pinyin as authoritative you will sound robotic compared to a native speaker who learns each word's actual tone curve naturally. Just like how if you believe english is a phonetic language you will pronounce lots of things wrong. (e.g., as an English learner once I learned how to spell "doubt" I kept trying to enunciate the "b" since I thought the pronunciation I learned from hearing, "dout", was wrong)
Disclaimer: I'm a native cantonese speaker so my knowledge about tones is bullshit. Just hypothesizing why text-to-speech and people learning to speak it always sound funny.