Naxi Pictographic and Syllabographic Scripts
Research notes toward a Unicode encoding
Naxi is a Tibeto-Burman (TB) language, related to Moso (a.k.a. W. Naxi) and Loloish (Yi) languages. Matisoff places Naxi in the Burmo-Naxi-Lolo group of TB within Sino-Tibetan (1991c; “Jiburish revisited: tonal splits and heterogenesis in Burmo-Naxi-Lolo checked syllables”. Acta Orientalia, Copenhagen; 52:91-114). In China the Naxi are sometimes called Moso (磨些 = 磨娑 = 麼些 = 摩梭 = ...), after the related minority group.
According to J. F. Rock’s Naxi-English Encyclopedic Dictionary (1963), the 納西 Nàxī (¹Na-²khi) script has one pictographic, and one syllabic (syllabographic) component.
《哈佛大學哈佛燕京圖書館藏中國納西族象形文經典分類目錄》 The Annotated catalog of Naxi pictographic manuscripts in the Harvard-Yenching Library (Harvard University, 1997; Compiled by 朱宝田 Zhu Baotian; HOLLIS Number: 007770852) lists collections of Naxi manuscripts. These manuscripts total nearly 11,000 items, some 573 of which (including small colorful booklets on fragile paper) are archived in the Harvard-Yenching Library. (See the images below.)
“The Library of Congress’s collection of 3,342 Naxi (Nashi) manuscripts is the largest collection outside of China and is considered the finest in the world—unrivaled in quality, quantity, and variety among Naxi collections in Europe, the People’s Republic of China, and Taiwan.”
In addition to primary Naxi sources, there are many secondary sources including examples of Naxi script. Rock’s two-volume dictionary (1963, 1972; HOLLIS Number: 005825924; Rock, Joseph Francis Charles, 1884-1962) provides an inventory of the pictographic and syllabographic signs in the roughly 8,000 manuscripts to which he had access (over a 27 year period, beginning in 1932). Other dictionaries published in China provide comparable inventories and analyses, e.g. : 《纳西象形文字谱》 Naxi xiang xing wen zi pu (Table of Naxi pictographs) by 方国瑜 Fang Guoyu (1981  HOLLIS Number: 007473487); 《纳西族象形标音文字字典》 Naxi zu xiang xing biao yin wen zi zi dian (Naxi pictographs and transcription character dictionary), by 李霖灿 Li Lincan (2001; ISBN: 7536721269). UC Berkeley’s online library catalogue includes some 61 items treating aspects of Naxi culture; Harvard’s HOLLIS catalogue includes some 169 items.
Rock’s dictionary contains a total of 3,414 entries, including both Tomba and Geba, and including several variant character forms under some entries (1963:xvi): “Each symbol serves for as many as 10 or more meanings when used phonetically, in addition to having ideographic or pictographic meanings.” The dictionary contains 512 pages of entries in the first volume, arranged alphabetically according to romanization. Where simple pictographs are reused in compounds and provide the initial syllable of the reading, these are all grouped roughly together in the dictionary. Thus, a semantic classificational system (rather like that of Chinese) arises naturally (though imperfectly) from this arrangement. The lexical entries are not assigned serial numbers in the dictionary, and so considerable effort in indexing, scanning and proofing of this source text would be required as a first stage toward digitization. As with Gardiner for Egyptian Hieroglyphs, Rock’s text would provide a firm basis for an initial encoding of Naxi. A more general database of Naxi signs would require, in addition to the scanning of Rock’s text, comparison of this text with other lexical sources and manuscripts. Simple signs are reused in compounds, and compounds are very often given independent lexical entries. Therefore, Rock’s sign list is not simply an inventory of the primitive (basic) elements of the writing system, and it might be imagined that graphemic analysis might reveal a subset of basic signs for encoding, for use within a higher-level text-processing protocol. Rock’s analysis is not as systematic as one might like: since he does not assign unique numeric identifiers to his lexical entries, there is no convenient system of cross-reference within the text. A great deal of effort would be required to perform this additional analysis, and certainly a great deal of this work has already been done by scholars in China in recent years. An encoding proposal for Naxi should be prepared in collaboration with 方国瑜 Fang Guoyu, 李霖灿 Li Lincan, et al., ideally in the framework of a more general project digitizing Naxi manuscripts.
Rock gives a total of 276 “sound complexes” (1963:xxv), i.e. he identifies 276 written syllable types irrespective of the four tones. The tones are marked in his romanization by superscript prefixal numbers: “¹low falling, ²middle level, ³high-short, ⁴from low to high rising” (xxxv). Each of these atonal syllable types may bear any of the four tones (xxxvi), and he also mentions some colloquial syllable types which have no writing (xxv). This PDF gives an (unsourced) list of some 128 Geba signs.