I recently read a book titled “Ying-yai Sheng-lan” (roughly meaning ‘Survey of the Shores’) by a 15th century sailor by the name of Ma Huan, who travelled with Zheng He (Cheng Ho) on the voyages of the treasure ships (宝船, bǎochuán).
Firstly, I was amazed at how authentic the descriptions of South India (referred to as 南天竺 nan-t’ien-chü, or as 南印度 nan-yin-du in old records) seem even today – more than half a millenium later.
Then, I came to the end of the book and found a list of place names (from India and around) that had been transliterated into Chinese by the year 1433. Here are just a few of the names that I found in the list (there were hundreds of names, more than half of which came from a Ming dynasty map):
- Chia-p’i-li – Cauvery
- Chih-lan – Chidambaram
- Chien-chih – Kanchipuram
- Kan-pa-li – Coimbatore
- Kan-pa-li t’ou – Kanyakumari
- Ko-chih – Kochi
- Ko-na-chi – Kanauj
- K’o-shih-mi-erh – Kashmir
- Ku-li – Calicut
- Mang-ko-nu-erh – Mangalore
- Mu-ku-tu-shu – Mogadishu
- Na-ch’ieh-po-tan-na – Nagapattinam
- Po-ch’a-li-tzu – Pataliputra (Patna)
- Po-lo-na-ssu – Varanasi
- Pu-lu-sha-pu-lo – Purushapura (Peshawar)
- Pin-t’ung-lung – Panduranga (Phan-rang) in South Vietnam (Champa)
When I read these, I couldn’t help wondering what our transliteration API (VakRed) would do with names as highly transformed as those.
VakRed would never be able to deal with them.
VakRed is an API designed to roughly transliterate abugida and alphabetic scripts solely for the purpose of comprehension and comparison (for use in distance metrics).
It just isn’t ready for abjad scripts like Arabic or logographic scripts like Chinese.