Thread by @ZevHandel

🔥 Picking this back up 🔥
🧵

https://twitter.com/ZevHandel/status/1417696909033033734

Just to switch things up, I’m going to *start* this thread with a digression. We’ll get back to this set of words later.

One of the first things you learn as a student of historical linguistics is that sound change is regular. It’s a mind-blowing thing. 🤯 It’s not at all clear *why* sound change should be regular, but it is.

The regularity of sound change has been confirmed over and over and over again in the history of languages.

What does “sound change is regular” mean? It means that within a community of language users, over a period of time, pronunciation shifts in regular, patterned ways. These shifts can be described by “rules”, or formulas.

A rule specifies the starting sound, the ending sound it changes into, and the context (the surrounding sounds) that must be present for the change to take place.

Here’s an example of a rule from the history of Mandarin: “The Old Mandarin sound g turned into the Modern Mandarin sound j whenever it occurred before one of the vowel sounds i or ü”.

Here’s how that exact same rule can be formulated in standard notation, using International Phonetic Alphabet (IPA) symbols in place of pinyin:

Mandarin k > tɕ / __ i, y

Here are some examples of words (well, morphemes, but we’ll just say words) that underwent this sound change, showing their older and newer pronunciations in pinyin with the symbol “>” (meaning ‘changed into’).
(I’ve simplified things a bit to stay focused on the main point.)

SET 1
見 giàn > jiàn 'see'
家 giā > jiā 'home'
基 gī > jī 'base'
劍 giàn > jiàn 'sword'
居 gǖ > jū 'dwell'
覺 güé > jué 'feel'
捲 güǎn > juǎn 'roll up'

And here are some words that didn’t undergo the sound change because g wasn't followed by i or ü:

SET 2
甘 gān > gān ‘sweet’
歌 gē > gē ‘song’
關 guān > guān ‘close’
古 gǔ > gǔ ‘old’
狗 gǒu > gǒu ‘dog’
根 gēn > gēn ‘root’

Another way to think about regular sound change is that the changing sounds don’t care about the meanings of the words they are in. They only care about the nearby sounds. The rules apply equally to all words.

Why does this happen? Why does a community of speakers shift pronunciations according to a regular pattern that ripples throughout the entire vocabulary of the language? I think it’s fair to say that nobody really knows. But it’s an empirical fact.

And that fact is crucial to the methods historical linguists use to reconstruct ancient pronunciations, even in the absence of written records or sound recordings.

Okay, so here’s the other amazing thing about regular sound change: it leads to *regular sound correspondences* across historically related languages.

The g sounds of Old Mandarin in Sets 1 and 2 (and many other words) descend from earlier g- sounds in Middle Chinese, which can be conceptualized as the common source of both Mandarin and Cantonese. (I say “conceptualized” because the reality is more complicated.)

And those g- sounds of Middle Chinese did not change in Cantonese.

(Pinyin isn't useful for Middle Chinese. So I will notate those g sounds using their IPA symbol "k", and I’ll put that between slashes to show it's IPA and not pinyin, thus: /k/.)

Take, for example, the Middle Chinese word for ‘home’, /kæ/ 家. That word ultimately developed into Cantonese gaa1, which starts with the same consonant sound. But as we saw in Set 1, that same word changed for speakers in the north, becoming giā and then jiā in Mandarin.

The regularity of this sound change in Mandarin—it's occurrence in word after word—means that there are *regular patterns of sound correspondence* between pronunciations of Mandarin and Cantonese *cognates* (i.e. words that descend from the same original word).

All the words of Set 1 have Middle Chinese /k/ sounds that changed to Mandarin j- and remained unchanged as Cantonese g-.

All the words of Set 2 have Middle Chinese /k/ sounds that stayed unchanged in both Mandarin and Cantonese, as g-.

This created two correspondence patterns.

Set 1 words are said to exhibit a “regular sound correspondence” of Mandarin j- to Cantonese g-.

SET 1: M. j- | C. g-

見 M. jiàn | C. gin3
家 M. jiā | C. gaa1
基 M. jī | C. gei1
劍 M. jiàn | C. gim3
居 M. jū | C. geoi1
覺 M. jué | C. gok3
捲 M. juǎn | C. gyun2

So predictable!

And here is the correspondence pattern for Mandarin words that didn’t undergo the sound change:

SET 2: M. g- | C. g-

甘 M. gān | C. gam1
歌 M. gē | C. go1
關 M. guān | C. gwaan1
古 M. gǔ | C. gu2
狗 M. gǒu | C. gau2
根 M. gēn | C. gan1

Also predictable!
🧙‍♂️🤯🧙‍♀️

Given enough time, the pronunciations of cognates in related languages can diverge a lot because of the accumulation of sound changes.

For example, French “chef” [ʃɛf] and English “head” [hɛd] are cognates. They diverged from the same ancestral source word spoken thousands of years ago. And now they have become so different you might never guess that they have a historical relationship.

https://mobile.twitter.com/ZevHandel/status/1405396278540836867

But no matter how different cognates get over time, those regular sound correspondences persist.

I’ll rephrase that: Sets of cognate words in two or more languages can sound really really different, but if they come from the same historical source, their sounds will align in regular, detectable patterns. It’s an amazing thing!

Hey! I’m done with my digression about regular sound change and regular sound correspondences! How many tweets did that take?

Never mind, don’t answer that. Some things are better left unknown.
😅

👀 Now I can finally reveal what I love so much about these four words:

Japanese
ねつ netsu [netsɯ] ‘fever’
Cantonese
jit6 [jiːt²] ‘hot, fever’
Korean
열 yŏl [jʌl] ‘fever, heat’
Mandarin
rè [ɹɤ⁵¹] ‘hot, heat’

💙 I love that their pronunciations are so different from each other, but they are all *the same word* (historically speaking): a word written with the Chinese character 熱. And ...

despite the fact that these four words sound almost nothing alike: (1) they all derive from a single ancient Chinese word; (2) their current pronunciations are the result of regular sound changes; and (3) they exhibit perfectly regular sound correspondences. 🤓

They are a lovely example of the patterning of language change over time. 💬

I'll show you more examples of the regular patterns that they adhere to.

To keep this thread from getting insanely long, I’ll show that the correspondences of the
• beginning consonants,
• ending consonants, and
• tones are regular.

We’ll skip the vowels.

The Middle Chinese pronunciation of the word 熱 ‘hot’ was
/ɲjetᴰ/ (“nyet” in Baxter’s transcription).

(Is it just a coincidence that it sounded like the Russian word meaning ‘no’? Yes, it is just a coincidence.)

That initial consonant /ɲ/ is a palatal nasal, the sound represented in Spanish orthography by “ñ”. (If you’re an English speaker, you can approximate it by thinking of it as an “ny” sound.) IPA /j/ is like an English “y” sound.

/ɲjetᴰ/

The raised “D” represents one of the four tones of Middle Chinese, the short abrupt “entering” tone.

/ɲjetᴰ/

1️⃣ Beginning Consonants
Regular sound changes in the history of the four languages:

Japanese go’on: /ɲ/ > n
Cantonese: /ɲ/ > j (an English “y” sound)
Korean: /ɲ/ > y (but disappears if following vowel is i, since yi cannot occur)
Mandarin: /ɲ/ > r (except when ends in -i)

So the regular sound correspondence pattern is:

Japanese | Cantonese | Korean | Mandarin
n | j | y | r (in standard notations)
n | j | j | ɹ (if you prefer IPA)

Here are a few examples showing the exact same sound correspondence as in 熱, demonstrating that it is regular.

人 ‘person’: にん nin | jan4 |인 in | rén
如 ‘as’: にょ nyo | jyu4 | 여 yŏ | rú
軟 ‘soft’: ねん nen | jyun5 | 연 yŏn | ruǎn
肉 ‘meat’: にく niku | juk6 | 육 yuk | ròu

2️⃣ Ending Consonants
Regular sound changes in the four languages when /t/ was at the end of a syllable:

Japanese go’on: /t/ > tsu (or chi, but we’ll ignore that for now to keep things simple)
Cantonese: /t/ > t
Korean: /t/ > l
Mandarin: /t/ > Ø (zero), i.e. /t/ disappears

So the regular sound correspondence pattern is:

Japanese | Cantonese | Korean | Mandarin
tsu | t | l | null (in standard notations)
tsɯ | t | l | Ø (if you prefer IPA)

Here are a few more examples of the pattern.

末 ‘tip’: まつ matsu | mut6 | 말 mal | mò
絕 ‘cut off’: ぜつ zetsu | zyut6 | 절 chŏl | jué
殺 ‘kill’: せつ setsu | saat3 | 살 sal | shā
舌 ‘tongue’: ぜつ zetsu | sit6 | 설 sŏl | shé

3️⃣ Tones
Japanese and Modern Korean don’t have tonal distinctions in Chinese-derived words. Here are the rules for C. and M.:

Cantonese: /ᴰ/ > 6 (low level [2]) when the MC consonant is a nasal like /ɲ/
Mandarin: /ᴰ/ > 4 (falling [51]) when the MC consonant is a nasal

So the correspondence pattern for tones is

Cantonese | Mandarin
6 | 4 (using tone category labels)
2 | 51 (using pitch notation)

And here are examples of the tone corresponence.

玉 ‘jade’: juk6 | yù (the MC word started with a nasal consonant ng [ŋ])
木 ‘tree, wood’: muk6 | mù
密 ‘dense’: mat6 | mì
肉 ‘meat’: juk6 | ròu

So this might look like a crazy collection of random syllables that happen to have similar meanings:

Japanese netsu [netsɯ] ‘fever’
Cantonese jit6 [jiːt²] ‘hot, fever’
Korean yŏl [jʌl] ‘fever, heat’
Mandarin rè [ɹɤ⁵¹] ‘hot, heat’

But these words fit neatly into patterns of regular sound correspondence seen across the vocabularies of these four languages. The patterns are invisible until you zoom out to see the big picture. 🔭

The Middle Chinese initial consonant /ɲ/ is probably the most volatile of the Middle Chinese initials. It’s changed to something else in almost every modern variety of Chinese, as well as in Sino-Japanese and Sino-Korean.

https://twitter.com/ZevHandel/status/1363031748645249024

As for ending /t/, Yue, Hakka, and some lexical layers of some Min varieties preserve it well. It disappears in (most) Mandarin varieties, and it is changed into a glottal stop /ʔ/ in many Wu dialects as well as in colloquial layers of Min.

Thanks to @LearnTeochew for supplying the colloquial Teochew (Cháozhōu 潮州) cognate word 'hot', which illustrates some of these other sound change pathways for the beginning and ending consonants: [dzuaʔ].

I promised you a Vietnamese form at the end of this thread.

Of course I didn't forget! How could you even think that?

Sino-Vietnamese pronunciations of Chinese characters have also undergone a lot of sound changes from Middle Chinese. But in the case of the particular constellation of sounds in this word 熱, it turns out that the Sino-Vietnamese (Hán Việt) reading is quite conservative:

nhiệt

The “nh” spelling represents a palatal nasal /ɲ/, the “iê” spelling represents the diphthong /iə̯/, and the under-dot is a tone mark. So the pronunciation of this Vietnamese syllable is /ɲiə̯t³²/ (in Hanoi), very close to the Middle Chinese pronunciation.

As much as I love sound change, it’s also impressive to see a word that has stubbornly preserved an ancient pronunciation, like a coelacanth.

I’ll leave you with a classic song vocalized by the great Peggy Lee that fits the theme of this thread.

🔥

/end