免费可读性检查器 查看你的句子实际得分
此工具对同一段文本一次性应用六种已建立的可读性公式, 即 Flesch Reading Ease, Flesch-Kincaid Grade Level, Gunning Fog, SMOG, Coleman-Liau 和 Automated Readability Index。粘贴摘要、论文或文章, 即可获得与专为学术和普通读者文档设定的目标相对照的年级水平基准。单一公式可能误导, 六种公式在同一范围内一致则很少如此。
一次使用六种可读性公式为文本评分, Flesch, Flesch-Kincaid, Gunning Fog, SMOG, Coleman-Liau 和 ARI。
0 / 1,000 词
84,000+
国际研究人员
平均评分
受到发表在以下期刊中的研究人员信赖
三步检查你的可读性评分
添加你的文本
摘要、完整章节、通俗摘要或求职信, 最多 1,000 words。无论摘录多长, 每个分数都会在一次处理中同时计算。
比较全部六个数值
将平均值视为主指标。其下方的各个公式显示, 难度究竟来自句子长度、词语长度, 还是两者共同作用。
不断精简, 直到落入目标区间
拆分长句, 用动词替换名词短语, 边编辑边观察数值变化, 直到文本达到读者所需的水平。
为严肃写作打造的可读性检查器
六个数值并列显示
Flesch, Flesch-Kincaid, Fog, SMOG, Coleman-Liau 和 ARI 同时计算, 因而某一公式的异常值不会把修订引向错误方向。
为真实文档设定的目标
研究性行文(13-16)、面向普通读者的写作(8-12)和面向参与者的文档(6-8)分别设定区间, 而不是对所有文本使用同一条笼统规则。
显示句子和音节计数
平均句长、每词音节数以及复杂词所占比例都会分项显示, 这样你就能看出究竟是哪一个因素推动了数值变化。
不发送到服务器
每个公式都在你已经打开的浏览器标签页中运行。未发表的章节无需离开其写作所在的机器。
可读性公式能告诉你什么, 不能告诉你什么
可读性公式估计读者在一次通读中理解一段文本所需的受教育程度。它不判断论证是否成立, 只关注两种会消耗读者精力的习惯, 即句子过长, 以及连续使用多音节词而不提供更短的替代。对已发表期刊的研究显示, 普通文章大约处于 14 到 16 年级, 医学摘要则接近 18 年级, 而跨数十年的科学写作调查表明, 尽管编辑要求相反, 这类行文多年来一直在变得更难。
其后果并非抽象。审稿人在读到引言之前，会先从摘要形成印象，而一段需要读两遍才能解析的文字，无论其中描述的结果如何，都会对论文不利。你所在学科之外的读者，包括因广度而部分被指派的审稿人，也会仅凭同一段文字判断这项工作是否值得继续关注。与此同时，一项研究相关的文书工作遵循其自身规则, 参与者材料应为 6 到 8 年级水平，资助摘要应使用通俗语言，新闻稿应当人人都能读懂。一项项目在同一周内很容易同时需要这三种层级，这也是本页不只采用一种，而是检查六种公式的原因。
机器起草的文字还有一个值得留意的变数。聊天机器人默认输出往往从开头到结尾都维持同一难度层级，缺少人类写作中自然呈现的张弛变化，即从简短的铺垫句到随后展开说明的较长句之间的转换。按段落逐段运行，而不是从头到尾一次性检查，通常更容易暴露出那些实际上无人修改过的部分。将此处的计数与 burstiness checker 配合使用，以检查节奏；当需要的是重写而不是诊断时，再使用 TextPulse humanizer。
Every formula reads the same draft a little differently
None of the six was built with your particular reader in mind, which is exactly the case for running all of them together.
Flesch Reading Ease
1948The earliest of the six and still the one most software reports by default. It weighs average words per sentence against average syllables per word on a 0-100 scale, where a higher reading runs easier. Most published research sits under 30 on this scale, so a paper reaching 30 to 50 stands out as unusually approachable.
Flesch-Kincaid Grade
1975A 1975 conversion of the Flesch formula into a US school-grade number, which is why it became the default citation in readability research. Surveys of published journals place medical abstracts near grade 18 and typical discussion sections around 16.
Gunning Fog Index
1952Weighs sentence length against the share of three-syllable-plus words, which makes it the formula quickest to react to jargon. When a Fog score runs well ahead of the Flesch-Kincaid grade for the same passage, vocabulary is doing more damage than sentence structure.
SMOG Index
1969Samples polysyllabic words across thirty sentences to gauge whether a reader will fully grasp the passage. It is the formula health communication relies on: consent forms and patient materials are typically required at grade 6 to 8, checked specifically with SMOG.
Coleman-Liau Index
1975Swaps syllable counting for characters per word, which machines can tally without ambiguity. Where it agrees with Flesch-Kincaid, trust the grade; where the two pull apart, the words in the passage are unusually long or short relative to how the sentences are built.
Automated Readability Index
1967Built from characters per word and words per sentence for military technical manuals, so it answers to long sentences more sharply than the others. That sensitivity makes it a fast early warning for a methods paragraph that has chained five steps into one fifty-word sentence.
Matching the target band to the document
Fit the number to the reader, never the reverse. A single research project commonly needs three of these bands running at once.
One finding, two very different sentences
Both passages are scored live by this tool. The underlying claim never moved; only the sentence construction did.
"The utilization of a multifactorial analytical approach for the assessment of intervention efficacy across heterogeneous participant subgroups necessitated the implementation of a stratified statistical methodology capable of accommodating substantial variance in baseline demographic characteristics."
One 34-word sentence, four nominalizations, thirteen complex words.
"This analysis needed a statistical method suited to mixed groups. Baseline traits varied widely between participants. The method had to account for that variation."
Three short sentences, plain verbs in place of noun phrases, nothing lost.
谁会使用可读性检查器
本科生和课程写作者
将论文与教师预期的层级进行比对，并查看哪些句子在推高难度。
研究生研究人员
在审稿人打开之前，将摘要或讨论部分从 20 多分拉回到较低区间。
临床和健康研究人员
将知情同意书和参与者材料控制在伦理委员会通常设定的 6-8 年级水平，尤其使用 SMOG 计数进行检查。
内容和 SEO 写作者
网页读者会很快放弃密集的段落。将落地页和文章控制在普通读者真正会读完的层级内。
分数显示为 19 年级。
缩短句子长度需要真正改写。
TextPulse humanizer 会拆分负担过重的句子，把名词短语改为动词，并在整篇草稿中调整单调的节奏，所有修改都会被标记以便批准，因此写作会达到读者预期的年级水平，同时不改变任何论点。