In my question on Chinese.SE, I learned that the modern character for "earth, ground"「地」(dì) used to be written in a multitude of ways, using either 「也」,「豕」, or「它」as phonetic components. Compare the Baxter-Sagart reconstructions of the OC (Old Chinese) pronunciations:「也」(/*lAjʔ/) ,「豕」(/*l̥ajʔ/) , and「它」(/*l̥ˤaj/). Compare also the reconstruction pathways from Old Chinese to Middle Chinese of the following characters derived from「也」or「它」(copied from this answer on Chinese.SE):

Character Mandarin Cantonese Hokkien Middle Ch. Old Ch.
jaa5 / yáh yæX *lAjʔ
taa1 / tā tha/thaⁿ tha *l̥ˤaj
dei6 / deih tè/tē/tōe/tī dijH *[l]ˤej-s
chí ci4 / chìh drje *Cə.lraj
shé se4 / sèh chôa / siâ zyæ *Cə.lAj
shī si1 / sī si/sì sye *l̥aj
chí ci4 / chìh drje *lraj
tuó to4 / tòh da *lˁaj
ji4 / yìh î ye *laj

The author of that answer grouped these modern Mandarin sounds into 4 groups: shi/chi/she, tuo/ta, ye/yi, and finally di, so 4 "groups" of sounds derived from the Old Chinese homophones「也」(/*lAjʔ/) ,「豕」(/*l̥ajʔ/) , and「它」(/*l̥ˤaj/). Compare also the derivatives of「豕」(from Wiktionary):

Character Mandarin Middle Ch. Old Ch.
shǐ ɕiᴇX *l̥ajʔ
zhú ɖɨuk̚ *[l]riwk
suì ziuɪH ?

「家」apparently is descended from 「豭」, with 「叚」 as a phonetic component and 「豕」 adding semantic meaning (Wiktionary). Perhaps the characters in the 「豕」 table are semantic derivatives (cf. the Baidu encyclopedia entry), in which case I guess the sound is irrelevant.

Regardless, I am interested in how the sounds changed from something similar to /*laj/ in Old Chinese to at least 4 "groups" of sounds not at all similar to that.

I am particularly interested in the connection between「也」and the "di" sound – this Chinese.SE answer claims that,

「也」originally depicted a child「子」with an emphasised mouth「口」, indicating the meaning to wail, cry. This word was later written as「嗁」and now written as「啼」. The meaning also is unrelated, and is a phonetic loan, derived from an early usage of「也」as a modal particle.

Please take note of the phonetic component of「嗁」;「遞」and「地」are homonyms in Mandarin, and the latter uses「也」as a phonetic component.

– which makes me think that there is some deeper (i.e. non-coincidental) connection between「也」and the sound "di", given that "di" is the sound of the (supposedly) original meaning (now written「啼」), and also the modern sound of「地」.

  • FYI, I’ve edited your question to [1] display the tables as real tables instead of fixed-width text (which doesn’t work well with screen readers for visually impaired users), and [2] incorporate the links into the text as words to make them less disruptive to the flow. Oct 9, 2023 at 1:45

2 Answers 2


Ultimately, this question boils down to "How are these Old Chinese syllables reconstructed to be so similar when even their Middle Chinese reflexes are quite different?"

One thing to state is that phonetic components of characters are used to help understand what these characters sounded like. As these all have 也/它 as their phonetic, there is presumed to be a phonetic closeness between them all.

Firstly, the finals, which were referred to in the earlier answer:

  • /a ~ æ ~ je ~ ij/

This is pretty spread out in the vowel space, but vowels can and do spread quite a bit through this millennium of Chinese history (詩經 to 切韻, if you will, from Zhou to Sui/Tang).

Notably, this whole family is believed to be in the same rime group in Old Chinese, that of 歌, as accounted for by the 詩經. The diphthong /aj/ makes sense as a candidate: /aj/ > /æ/ > /e/ > /i/ is super common cross-linguistically as a pathway (e.g. Latin caseus plus glide metathesis to English cheese).

The loss of final /j/ is also attested, and actually has examples from Min topolects: Fuzhounese 拖 'to pull' has two separate vernacular pronunciations, used in slightly different contexts: tăi /tʰai⁵⁵/ with final /j/, and tuà /tʰua⁵³/ without. Thus, we can account for the full spectrum of Middle Chinese finals for this family.

But each of these groupings does end up with a different reflex, and so each of these does require an explanation. Baxter & Sagart (2014) has:

  • pharyngealisation of the initial as the trigger for loss of the final /j/;
  • a difference in the dialect as the cause of /æ/ in some places but /e/ in others;
  • a /r/ medial for those that will go on to palatalise the Middle Chinese initial;
  • and 地 is just different.
Middle Chinese Final Chinese Characters Old Chinese Final (Baxter-Sagart)
a 他, 紽, 拖 *-ˤaj
也, 蛇 *-Aj
ye 施, 匜 *-aj
je 池, 馳 *-raj
ij *-ˁej

On the initials then, we have this whole phonetic series explained by some kind of lateral consonant. This is a long-standing feature of Old Chinese reconstruction, summarised by Baxter & Sagart:

In Middle Chinese, OC *l- palatalizes to y-, *lˤ- becomes d-, and both *l- and *lˤ become dr- when followed by medial/infixed *r.

Evidence from other languages is cited:

  • 蜕 OC *lot > MC ywet > Mandarin yuè ‘exuviae of insects or reptiles’, Vietnamese lột ‘to skin; to throw off’
  • 條 OC *lˤiw > MC dew > Mandarin tiáo ‘branch (n.), shoot (n.)’, Xiamen Min Nanliâu ‘a long strip’

Old Chinese laterals are most faithfully preserved in the Wǎxiāng 瓦鄉 dialect of northwest Húnán. There, modern laterals frequently appear corresponding to OC *lˤ-, *lˤr-, and *lr-, as in the following examples from the Gǔzhàng dialect:

  • 桃 OC *C.lˤaw > MC daw > Mandarin táo ‘peach’, Gǔzhàng /laɔ13/
  • 田 OC *lˤiŋ > MC den > Mandarin tián ‘field; to hunt’, Gǔzhàng /lɛ13/
  • 讀 OC *C.lˤok > MC duwk > Mandarin ‘read (v.)’, Gǔzhàng /luʔ53/

The voiceless laterals have the following:

The coronal voiceless resonants usually have coronal reflexes in Middle Chinese: *l̥- > sy; *l̥ˤ- > th. But there is good evidence for an alternative development of these initials to a fricative [x] (or perhaps [h]), which became MC x-. This development can be located in the central and western regions of the country, while the coronal reflexes [above] were probably found along the coast.

The move from lateral to non-lateral dental/alveolar stops occurs by the 1st century CE at the latest, as Sagart (1999) states:

A work with a marked spoken character, the Bai Hu Tong Yi, which records oral discussions on the classics between Eastern Han scholars in 79 CE, contains no less than five sound glosses equating OC laterals and alveolar stops.

This 白虎通義 for example has 天 (OC *l̥ˤi[n]) = 鎮 (OC *ti[n]-s), 瀆 (OC Zhengzhang *l'oːɡ) = 濁 (OC *[N-tˤ]rok).

Back to the 也/它 series then:

Middle Ch. Initial Characters OC Initial (Baxter-Sagart)
y 也, 匜 *l-
d 紽, 地? *lˁ-
dr *lr-
sy *l̥-
th 他, 拖 *l̥ˤ-

地 could be reconstructed with OC *lˁ-, like 紽, so the pharyngealisation is present in the Baxter-Sagart; the use of square brackets is:

to indicate uncertainty on the voicing specification of the initial.

We now have the odd one out, 蛇 with MC zy-. This is attributed to a prefix before the *l- lateral, as with:

  • 實 *mə.li[t] > MC zyit > Mandarin shí ‘fruit; full’, pMǐn *-dž-; Proto-Tai *m.lecD ‘grain’, giving all of Thai เม็ด met, เมล็ด ma-let and เล็ด let.

But the nature of the prefix is unknown - it has to have a vowel (I presume to retain voicing into Middle Chinese). Zhengzhuang uses *ɦlj- instead.

It is true that reconstructed Old Chinese is given a phonology that can feel utterly alien to modern Chinese and indeed modern East and Southeast Asian ears. The use of these phonetic series is driven by the phonetic components of Chinese palaeography, and the search for documentary evidence of the sound changes can be particularly arcane - and yet, they are (perhaps surprisingly?) regular. At least for now, this is my understanding of the state of the art.

  • Unbelievably thorough answer! I’m so glad you gave so much of your attention and effort to this old question!
    – D.R
    Oct 15, 2023 at 5:21

Your question is extremely technical, especially from the Chinese side, and its complex phonetic and graphic evolutions.
In my humble opinion, the multiple reflexes of the same apparently phonetic component is possibly due to the fact it was used for a number of sounds like *laj, *tlaj, *dlaj and *hlaj, which used to have some phonetic similarity in Ancient Chinese, but have now drifted apart in Modern Mandarin, so that the connection between them is now lost, or at least far from obvious.
So I would say that your question hinges around the way Ancient Chinese is (to be) reconstructed. Baxter-Sagart reconstructions are not bad, but I would say that improvement is still possible. Their system has so-called pre-initial consonants, which concretely is an oxymoronic concept, and logically that issue needs a satisfactory solution.

  • 1
    pre-initial is not an oxymoronic concept, even if the name is oxymoronic. The term is common in describing sesquisyllabic languages for the consonant of a minor syllable
    – Tristan
    May 3, 2022 at 9:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.