关于「单语构式等效:不嵌词也能精准」,《嵌词阈值与单语等效》给出的定量结论是什么?
结论:第三条流行说法认为,写纯中文或纯英文时只能在精度损失与翻译腔之间二选一。翻译学的非对等策略、动态对等与间接程序,加上构式语法关于构式自身携带意义的论证,给出第三条路:在短语层面重组。把精准写成特征覆盖率之后,示意例子里三个空缺概念都能用 4 到 12 个单位的单语构式达到 0.80 的覆盖率;穷举搜索给出的骨架比贪心与手写候选更短;重组版的翻译腔标记可以降到零。代价是长度,换来的是人人都懂。
数据口径:嵌词收益只在读者共享双语词库时成立:把单语表达、嵌入与嵌入加注释写成覆盖率的三条成本线,示意参数下 kiasu 在书面渠道的覆盖率阈值为 0.844;读者不发问时,失败会变成沉默的误读;各地区分页面值得嵌的词,合成混合页面后可能全部跌破阈值。 书中参数为示意,全部计算由书中程序生成,不对应任何真实客户、对话或成交;马来语、闽南语、粤语、英语、华语一律平等对待。
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。 与结论吻合的打印值: 0.80, 12
| kiasu: strategy | coverage | length |
|---|---|---|
| literal | 0.40 | 2 |
| compound | 0.60 | 4 |
| hyper | 0.60 | 7 |
| quant | 0.40 | 8 |
| marker | 0.80 | 10 |
| hyper+ | 1.00 | 12 |
| shortest at >= 0.8: marker (10 units, coverage 0.80); full: 12 units | ||
| lepak: strategy | coverage | length |
| literal | 0.40 | 2 |
| compound | 0.60 | 4 |
| hyper | 0.80 | 7 |
| marker | 1.00 | 9 |
| shortest at >= 0.8: hyper (7 units, coverage 0.80); full: 9 units | ||
| sajiao: strategy | coverage | length |
| literal | 0.40 | 2 |
| hyper | 0.60 | 3 |
| marker | 0.80 | 4 |
| compound | 0.80 | 6 |
| hyper+ | 1.00 | 8 |
| shortest at >= 0.8: marker (4 units, coverage 0.80); full: 8 units |
查看完整验算程序
# Chapter 6. p61: monolingual constructions that stand in for a loanword.
# Each target concept has five semantic features. A candidate rendering
# covers some of them and costs a length (hanzi for Chinese renderings,
# words for the English one). Rule: take the shortest candidate whose
# feature coverage is at least 0.8; report the full-coverage one too.
# Strategies: literal, hyper (hypernym + qualifier), compound (borrowed
# sense compound), marker (pragmatic marker moved), quant (quantified).
# Features, renderings and lengths are illustrative.
TARGET = {
"kiasu": {"fear_loss", "compete", "over_prepare", "anxious", "early"},
"lepak": {"idle", "relaxed", "social", "outdoor", "unhurried"},
"sajiao": {"coy", "petulant", "indulge", "affection", "playful"},
}
CANDS = {
"kiasu": [
("literal", {"fear_loss", "compete"}, 2),
("compound", {"fear_loss", "compete", "anxious"}, 4),
("hyper", {"fear_loss", "compete", "early"}, 7),
("quant", {"over_prepare", "fear_loss"}, 8),
("marker", {"fear_loss", "over_prepare", "early", "anxious"}, 10),
("hyper+", {"fear_loss", "compete", "over_prepare", "anxious",
"early"}, 12),
],
"lepak": [
("literal", {"idle", "relaxed"}, 2),
("compound", {"idle", "relaxed", "social"}, 4),
("hyper", {"idle", "relaxed", "social", "outdoor"}, 7),
("marker", {"idle", "relaxed", "social", "outdoor",
"unhurried"}, 9),
],
"sajiao": [
("literal", {"coy", "playful"}, 2),
("hyper", {"petulant", "indulge", "playful"}, 3),
("marker", {"petulant", "indulge", "playful", "coy"}, 4),
("compound", {"coy", "affection", "petulant", "indulge"}, 6),
("hyper+", {"coy", "petulant", "indulge", "affection",
"playful"}, 8),
],
}
if __name__ == "__main__":
for c, cands in CANDS.items():
tgt = TARGET[c]
print("{}: strategy coverage length".format(c))
rows = []
for name, feats, ln in cands:
cov = len(feats & tgt) / len(tgt)
rows.append((cov, ln, name))
print(" {:<12} {:>8.2f} {:>7}".format(name, cov, ln))
ok = sorted((ln, name, cov) for cov, ln, name in rows if cov >= 0.8)
full = sorted((ln, name) for cov, ln, name in rows if cov >= 1.0)
print(" shortest at >= 0.8: {} ({} units, coverage {:.2f});"
" full: {} units".format(ok[0][1], ok[0][0], ok[0][2],
full[0][0]))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《多语参杂语码转换信息论、词汇空缺补偿与单语种等效精准表达母典》第 6 章「单语构式等效:不嵌词也能精准」(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-381 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026381,
author = {Fan, Yafu},
title = {Information Theory of Code-Switching, Lexical-Gap Compensation and Monolingual Precision Equivalence in Multilingual Markets},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}