关于「东西方原典考辨:《文心雕龙·练字》与 Flesch 1948」,《中文可读性与字频句长》给出的定量结论是什么?
结论:第一组对应「避诡异」:短句僻字版每百字 62.50 个少见字,长句常用字版为零,与第一章一致。第二组对应「省联边」:「江河湖海波浪滔滔」八个字全是三点水,连续长度为 8,远超刘勰的三接之限,被标记;同一句后半的「江上有风」连续长度为 1,通过。第三组对应「权重出」:一句十三字的话,第 5、6、9、10 个位置上的字都在前六字之内出现过,十三字中四字重复,是「问题」二字反复出现造成的。第四组对应「调单复」:「一人上山」平均 2.25 画,「大水小日」平均 3.50 画,都是瘦字;「警露翻疆」平均 19.25 画,「黯霸鑫矗」平均 22.50 画,都是肥字。
数据口径:作者自拟的四篇小样本中,四字短句堆僻字的文本平均句长最短,按本书的示意难度公式却最难,得分 19.15;同样短句、全用常用字的文本只有 0.45。 书中参数为示意;全部计算由书中程序生成;本书属于非个人化经营建议。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 10, 19.25, 2.25, 22.50, 3.50, 62.50
| strange short-rare rare per 100 | 62.50 |
| strange long-common rare per 100 | 0.00 |
| sides run of 8 -> flag | |
| sides run of 1 -> ok | |
| repeats positions 5 6 9 10 (4 of 13 hanzi) | |
| strokes | 1 2 3 3 mean 2.25 thin |
| strokes | 3 4 3 4 mean 3.50 thin |
| strokes | 19 21 18 19 mean 19.25 fat |
| strokes | 21 21 24 24 mean 22.50 fat |
查看完整验算程序
# Chapter 7. p72: the four rules of Wenxin Diaolong, "Lianzi", made
# countable. Avoid the strange: rare hanzi per 100 (toy table). Spare
# joined sides: longest run of hanzi sharing a side component; Liu Xie
# tolerates three in a row, so a run above 3 is flagged. Weigh repeats:
# hanzi that recur within the 6 preceding hanzi. Balance thin and fat:
# mean stroke count per clause (illustrative cut-offs: thin <= 4, fat
# >= 15). Component and stroke tables cover only the sample hanzi; the
# samples are the author's own.
import math
import re
TIER1 = ("的一是不了在人有我他这个们中来上大为和国地到以说时要就出会可"
"也你对生能而子那得于着下自之年过发后作里用道行所然家种事成方"
"多经么去法学如都同现当没动面起看定天分还进好小部其些主样理心"
"她本前开但因只从想实日者意无力它与长把机十民第公此已工使情明"
"性知全三")
TIER2 = ("高山云雾采茶香气浓味厚入口回甘泡满屋每杯老师傅花间慢做值您喝"
"收问题身选品郁醇令沉醉盏匠光结晶细鉴珍藏水警露翻疆霸清至善江"
"河湖海波浪滔风叶放再倒热等儿什特别款金礼盒相益彰韵赏设列随变"
"量足条件即信息步期望取质任何预未停策略改若引度确则证给均约束"
"最连续布指数很句越短读卡住往认识字话哪怕四眼熟词顺思层走完反"
"僻猜判断段文难频园腰照更批由手炒制阴凉仓库存半装商页描述符或"
"包运输途破损请签七内通订单联系客服并传片核按平台规办退货换")
COMMON = set(TIER1) | set(TIER2)
HAN = re.compile(r"[\u4e00-\u9fff]")
STOP = re.compile(r"[。!?;!?;\n]+")
def sent_lengths(text, stop=STOP):
out = []
for part in stop.split(text):
n = len(HAN.findall(part))
if n:
out.append(n)
return out
def features(text, common=COMMON):
lens = sent_lengths(text)
han = HAN.findall(text)
rare = len([c for c in han if c not in common])
return len(han) / len(lens), 100.0 * rare / len(han)
def difficulty(asl, rp, a=0.10, b=0.30):
return a * asl + b * rp
SR = ("臻品甄选。馥郁氤氲。璀璨鎏金。旖旎天成。")
LC = ("你只要把茶叶放进杯子里再倒上一杯热水,等一会儿就"
"可以喝了,不用学什么特别的方法。")
LB = ("江河湖海波浪滔滔,江上有风。")
CC = ("这个问题的问题在于问题本身。")
DF = ("一人上山,大水小日。警露翻疆,黯霸鑫矗。")
SIDE = {c: "water" for c in "江河湖海波浪滔"}
STROKES = dict(zip("一人上山大水小日警露翻疆黯霸鑫矗",
(1, 2, 3, 3, 3, 4, 3, 4, 19, 21, 18, 19, 21, 21, 24,
24)))
CLAUSE = re.compile(r"[,。!?;、]+")
def longest_run(text):
best = run = 0
prev = None
for ch in text:
side = SIDE.get(ch)
run = run + 1 if side is not None and side == prev else (
1 if side is not None else 0)
prev = side
best = max(best, run)
return best
def repeats(text, window=6):
han = HAN.findall(text)
return [i for i, c in enumerate(han) if c in han[max(0, i - window):i]]
if __name__ == "__main__":
for name, text in (("short-rare", SR), ("long-common", LC)):
print("strange {:<12s} rare per 100 = {:6.2f}".format(
name, features(text)[1]))
for part in CLAUSE.split(LB):
if part:
r = longest_run(part)
print("sides run of {} -> {}".format(
r, "flag" if r > 3 else "ok"))
rp = repeats(CC)
print("repeats positions {} ({} of {} hanzi)".format(
" ".join(str(i) for i in rp), len(rp), len(HAN.findall(CC))))
for part in CLAUSE.split(DF):
if part:
s = [STROKES[c] for c in part]
m = math.fsum(s) / len(s)
tag = "thin" if m <= 4 else ("fat" if m >= 15 else "mixed")
print("strokes " + " ".join("{:2d}".format(x) for x in s)
+ " mean {:5.2f} {}".format(m, tag))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《中文可读性公式与字频句长、落地页与专著段落的阅读难度母典》第 7 章「东西方原典考辨:《文心雕龙·练字》与 Flesch 1948」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-323 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026323,
author = {Fan, Yafu},
title = {Chinese Readability from Character Frequency and Sentence Length: Reading Difficulty of Landing Pages and Monograph Paragraphs},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}