Who Are the Bai?
The Bai people (白族, Báizú, literally 'white people') are an ethnic group of approximately 1.9 million concentrated around Dali (大理) in Yunnan province, southwest China. Their kingdom, the Nanzhao (南詔, 738–902 CE) and its successor the Dali Kingdom (大理國, 937–1253 CE), was a major regional power that interacted intensely with Tang-dynasty China and with Tibetan and Southeast Asian civilisations. This political history explains why the Bai language has absorbed — and been thoroughly shaped by — Chinese.
The Classification Puzzle
Bai is one of Asia's most contested classification problems. Three serious camps exist:
- Sinitic branch — some scholars (notably Sergei Starostin and Zhengzhang Shangfang) argue that the oldest Chinese layer accounts for all but an insignificant residue of Bai vocabulary, so that Bai is an early branching from Chinese itself; Starostin dates the split to around the 2nd century BC, in the Western Han, long before the modern Sinitic varieties took shape.
- Tibeto-Burman — others place Bai within Tibeto-Burman, noting morphological and lexical parallels with Burmish and Loloish languages.
- Language isolate / independent branch — a growing consensus treats Bai as an independent branch of Sino-Tibetan (or even Sino-Tibetan's sister), deeply transformed by millennia of contact with Chinese.
The difficulty stems from Bai's two massive strata of Chinese loanwords: an older stratum reflecting Middle Chinese phonology (中古漢語, c. 600 CE), and a newer stratum from modern Southwestern Mandarin. These layers obscure the original Bai vocabulary, making it hard to distinguish inherited cognates from ancient loans.
How White Script (白文) Works
Like the Zhuang, the Bai used Han characters to write their language in a system called Baiwen (白文, 'Bai writing') or Bowen. The same character can be read with two entirely different pronunciations depending on whether you are reading it as Chinese or as Bai:
- Not every character splits cleanly: Bai's own word for 'horse' is itself ma˧˧ (Proto-Bai *mæ), an old Sinitic form, so 馬 sounds much the same whether you read it as Chinese or as Bai. The two systems separate where the layers diverge — 目 carries the Chinese-layer reading mu˨˨ in Bai, while the everyday Bai word for 'eye' is ŋui˨˨.
- Grammatical particles, aspect markers, and tone-carrying suffixes particular to Bai grammar are handled through phonetic borrowing, sometimes using rare or obscure Han characters purely for their local dialect pronunciation.
The 1958 Latin Scheme
In 1958, the PRC introduced the Bai Wenzi (白文字) Latin-alphabet scheme, designed by linguists to provide a unified orthography for Bai's phonetically varied dialects (the three main groups: Jianchuan/Central, Dali/Southern, and Bijiang/Northern — the first two close enough that speakers understand each other after a month together, while the divergent Northern varieties are treated by ISO 639-3 as separate languages). The scheme uses diacritics and tone numbers. Despite official promotion, Baiwen survives in religious texts (especially Benzhu 本主 deity worship), classical poetry, and stone inscriptions around the Dali plain — some dating to the Nanzhao kingdom itself.
A Mirror of Sinosphere Contact
The Bai case illustrates a pattern found across the Sinosphere: centuries of sustained literary contact with Classical Chinese created writing traditions that are simultaneously indebted to and distinct from Chinese. Where Zhuang invented new sinographs for phonology, and Vietnamese Chữ Nôm did the same for an Austroasiatic language, Bai developed a dual-reading convention — one glyph, two worlds.