About a quarter of the characters in the Analects (論語) and Mencius (孟子) belong to 23 small grammatical words, counting every use of each character: the 18 function words, 虚词 or 虛詞, that the mainland's last university-entrance exam outline listed, plus 矣, 哉, 夫, 豈 and 爾. We counted the Wikisource texts of both books, and the 23 make up 22.9% of the Analects and 25.6% of Mencius. In Mencius, 之 alone is one character in every 19.
The same words almost vanish from Lu Xun's The True Story of Ah Q (阿Q正傳), serialised in 1921 and 1922 in the new vernacular: 矣 appears once, 焉 and 哉 not at all, and 之 falls to 3.7 per thousand characters while 的 runs at 41.8.
From 之 in Mencius to 的 in Lu Xun
We took the chapter texts of the Analects and Mencius from Wikisource, removed notes and headings, and counted Han characters only: 15,929 in the Analects and 35,383 in Mencius. Each count includes every use of a character, so 之 as the verb "go" and 與 as "give" are counted with the grammatical uses, and the punctuation the counts below rely on was added by modern editors.
| Character | Analects, per 1,000 | Mencius, per 1,000 | Ah Q, per 1,000 |
|---|---|---|---|
| 之 | 38.5 | 53.7 | 3.7 |
| 也 | 33.5 | 34.8 | 10.8 |
| 而 | 21.6 | 21.8 | 8.1 |
| 者 | 13.7 | 18.1 | 1.2 |
| 以 | 13.4 | 18.0 | 4.6 |
| 矣 | 11.4 | 7.2 | 0.1 |
| 乎 | 9.9 | 5.9 | 2.3 |
| 焉 | 5.5 | 4.0 | 0 |
| 哉 | 3.8 | 2.9 | 0 |
In Mencius, 之 and 也 are the two commonest characters of any kind. In the Analects they lead the function words, though 子 and 曰, as in 子曰, "the Master said", come first overall. The 也 in Ah Q is the same character doing a different job: 190 of its 194 uses sit mid-sentence, almost all as the adverb "also", and the four that end a clause are mock-classical formulae of the 某地人也 kind.
Final 矣, and 焉 and 乎 in mid-sentence
We classified each occurrence by what follows it. Counting a run of final particles, such as 也已矣, as one ending, 矣 is followed by punctuation about 99 times in 100 in both books. It marks something as so, or as having come to be so, as in the Zuo Zhuan line that Taiwan's Ministry of Education dictionary cites for it, 「晉侯在外，十九年矣」.
The least final of these words differs by book. In the Analects it is 焉, and 26 of its 29 mid-sentence uses mean "how" or "where", as in 焉得 and 焉知. In Mencius only 13 of its 27 mid-sentence uses do; most of the rest carry the sense "in it" or "at this", as in 必有事焉而勿正. In Mencius the least final is 乎, because 乎 also works as a preposition. 也 can mark a pause inside a sentence too, as in the Analects' 柴也愚，參也魯.
Double duty for 乎, 與 and 之
Taiwan's Ministry of Education dictionary gives 乎 a prepositional sense, 「相當於「於」」, equivalent to 於, and quotes Han Yu's 「生乎吾前」. We read all 65 mid-sentence uses of 乎 in Mencius by hand: about 44 are this preposition, as in 惡乎定, 變乎色 and 得乎丘民, about a fifth of the book's 208. The rest are adjective endings such as 蕩蕩乎 and particles inside longer phrases.
與 is taught as a question particle, but in the Analects' 「夫子之言性與天道，不可得而聞也」 it means "and": the dictionary quotes that very line for its conjunction sense, read yǔ. And for 之 in the sense of the modern 的, as in 夫子之文章, the same dictionary files it as a preposition. The label a word gets depends on whose grammar describes it.
A gaokao list at odds with the counts
The 2019 outline for the mainland's university-entrance Chinese paper, the last the exam centre issued before outlines were dropped from 2020, according to a 2021 report, lists 18 「常见文言虚词」, common classical function words: 而、何、乎、乃、其、且、若、所、为、焉、也、以、因、于、与、则、者、之. The outline calls them common, not required. 矣 is not on it, though it appears 181 times in the Analects and 253 in Mencius. 乃 (3 and 13 times), 因 (5 and 6) and 且 (12 and 32) are on it.
Interest in these words is old. Wang Yinzhi's 經傳釋詞, prefaced in 1798, covered 「凡百六十字」, 160 characters. And the phrase 之乎者也 has its own story, told two ways. In 邵氏聞見錄 the Song founder asks why the gate plaque reads 明德之門, is told 之 is a particle, and says 「之乎者也，助得甚事！」, and his minister has no answer. In 湘山野錄 the gate is the 朱雀門 and the emperor laughs. The phrase became a byword for pedantry, as the Ministry's dictionary records.
A percentage on our helper page
Until 2 October 2026 our Classical Chinese particle helper said these words make up 30% to 40% of a classical text. Its own 23 reach 22.9% and 25.6% in these two books; 30% is reached only with a much wider set of 57 characters that adds negators, adverbs and demonstratives, which gives 31.6% and 34.0%. Its example for 矣 came from a passage that ends 和於身也 in both the Zhanguo Ce and the Shiji, with no 矣 in it, and its search found nothing for pinyin typed without tone marks or for Traditional characters. It now gives the measured shares, cites every example to the chapter it comes from, and finds each word from pinyin with or without tones and from either script.
Set texts in Hong Kong and classical-prose lessons in Taiwan
Hong Kong's Education Bureau set 12 classical texts in 2015 for the diploma exam from 2018, among them three passages from the Analects on 仁, 孝 and the 君子, Mencius's 魚我所欲也, and Xunzi's 勸學. That last one contains 「積土成山，風雨興焉」, the line the Taiwan dictionary uses to show 焉 meaning "there". We did not check whether the list has changed since 2015.
Taiwan's curriculum guidelines of January 2018 set classical prose at 35% to 45% of senior-high lessons, averaged over three years and not counting the separate Chinese-culture readings, and rising shares through junior high, from 10% to 20% in the seventh year to 25% to 35% in the ninth. Its recommended texts include 燭之武退秦師, 鴻門宴 and 師說. We did not check the syllabuses in Singapore or Malaysia.
于 in place of 於 in simplified print
The classical texts already use both 於 and 于. In the Wikisource text of the Analects, 於 appears 176 times and 于 6; in Mencius, 526 and 34. Nineteen of Mencius's 于 sit inside quotation marks, mostly lines from the Book of Songs and the Book of Documents, and two belong to the surname 淳于. Simplified Chinese writes 於 as 于 almost everywhere, keeping 於 as a surname, so in a Simplified edition the two prepositions look the same.
