Microsoft Word Chinese Lexicography in the Contemporary Period HuangEtAl2016. docx

Selection of Lexical Entries 
Once the issue of what is a lexical unit is determined the selection of lexical entries poses 
yet another challenge. The selection of lexical entries for character dictionaries (
zidian
), 
is different from that in word dictionaries (
cidan
). However radically different they could 
be in content or format, they do not differ fundamentally in terms of entry selection. This 
shows that the inventory and use of Chinese characters, is highly conventionalized, just 
like English dictionaries. A survey of entries in word dictionaries in Chinese (
cidian
), 
however by Huang (1998), showed a very different story. They compared five different 
Chinese dictionaries or lexica compiled between 1993 and 1997 in China and Taiwan 
(but converted to common character representation), ranging from less than 40,000 
entries to over 156,000 entries. They showed that the mutual coverage between any two 

---

Huang et al. (2016) [Pre-publication draft] 
6 
dictionaries, defined by the mean of the entry coverage of dictionary A over dictionary B 
and vice versa, range from 49% to 68%. In particular, what is surprising is that bigger 
dictionaries do not have better coverage over smaller dictionaries. The 156,710 entry 
Revised Mandarin Chinese Dictionary (RMCD, from Taiwan) covers only 68.58% of the 
70,325 entry ABC Chinese Dictionary (ABC, from the USA). And the 56,162 entry 
Xiandai Hanyu Cidian (XH) covers only 72.48% of the 39,459 entry Standard 
Segmentation Lexicon (GB), even though they are both from China and the compilation 
of GB consulted XH. Overall, they found that only 21,655 entries were shared by all five 
dictionaries and that corpus frequency alone was not a good predictor of these entries. 
Such variations, especially the failure of much larger dictionaries to include entries from 
much smaller dictionaries, underline the issue that Chinese lexicography has yet to 
develop a set of commonly accepted and inter-operable criteria for lexical entry selection.