# 英语高级 · 第 4 周 Day 5|关键词卡与四段陈述 > 系列:语种学习 / 英语 / 高级 · 每日学习|AI生成|2026-10-27|codebuddy **今天学什么** 写得好的人做陈述,最常见的失败是"把文章念出来":句子没错,但听众在第二分钟就跟丢了,因为书面语靠回读,口头语只能靠一次听清。今天要解决的是句子的"可听性"——同一套论证,怎么改造成耳朵能接住的形状。技术上只做一个转换:把长名词短语拆成主谓,把从句压到一个,把名词化还原成动词。产出是一张四段关键词卡(问题、证据、反驳、主张),每段不超过 5 个关键词,全程不写逐字稿。学完你应能在不看稿的情况下讲满 5 分钟,且每段都记得住自己要干什么。 ## 一、今日任务 题目沿用本周评论,但陈述不是评论的朗读版:评论是"写给读者判断",陈述是"说给听众跟随"。同一个主张,陈述要更早给出结论、更晚给出细节。 | 环节 | 时长 | 具体动作 | 产出 | | --- | --- | --- | --- | | 1 定四段 | 25 min | 按"问题—证据—反驳—主张"分配时间(1:00 / 1:45 / 1:15 / 1:00),每段写一句"这一段结束时听众应该相信什么" | 四段任务表 1 张 | | 2 写关键词卡 | 25 min | 每段只写 5 个以内的关键词(名词或动词,不写完整句);关键词之间用箭头标出推进方向 | 关键词卡 1 张 | | 3 首次试说 | 25 min | 看卡说 2 遍,第 1 遍允许停顿,第 2 遍计时;两遍都录音,不重录 | 录音 2 条 + 时长 | | 4 回听改卡 | 15 min | 回听时只记两件事:哪里卡住、哪里说成了书面长句;据此改卡(加箭头、删关键词) | 改后卡 + 卡点 3 条 | **四段的时间分配理由**:证据段最长(1:45),因为听众需要例子;反驳段第二长(1:15),因为这是你与其它立场拉开差距的地方;问题段和主张段各 1 分钟,前者只负责让听众在意,后者只负责让听众记住。 ## 二、核心词汇与搭配 | 词/短语 | 音标 | 词性 | 中文释义 | 高频搭配 | 语域 | | --- | --- | --- | --- | --- | --- | | signpost | ˈsaɪnpəʊst | v. | 给结构立路标,预告下一步 | signpost the move / the structure | 学术、演讲 | | pivot | ˈpɪvət | v./n. | 转向(论证重心) | pivot to; the pivot from A to B | 商务、媒体 | | unpack | ˌʌnˈpæk | v. | 展开(把压缩的主张说细) | unpack the claim / the assumption | 学术、演讲 | | frame | freɪm | v. | 框定(设定问题的理解方式) | frame the issue as; the framing of | 学术、媒体 | | land | lænd | v. | (论点、笑话、结尾)被接住、奏效 | the point landed; land a conclusion | 通用偏口语 | | run through | rʌn θruː | phrasal v. | 快速过一遍 | run through the evidence | 商务、通用 | | hold | həʊld | v. | (论证)站得住 | the argument holds; hold only if | 学术 | | concede ground | kənˈsiːd ɡraʊnd | phr. | 让出一部分立场 | concede ground on; refuse to concede ground | 媒体、谈判 | | takeaway | ˈteɪkəweɪ | n. | 听众应带走的那一句 | the main takeaway; one takeaway | 商务 | | in outline | ɪn ˈaʊtlaɪn | phr. | 概要地,只说骨架 | the case in outline; describe in outline | 学术 | **例句与用法** 1. **signpost** — Signpost the pivot before you make it: "That brings me to the objection." 转折之前先给路标:"这就说到反对意见了。"(口头必须提前半句预告,书面可以突然转折。) 2. **pivot** — The talk pivots from diagnosis to remedy at the two-minute mark. 陈述在第二分钟从诊断转向药方。(pivot 是"重心转移",不是"换个话题"。) 3. **unpack** — Let me unpack that claim with one example. 我用一个例子把这个主张展开。(unpack 后面接抽象名词,是演讲里最实用的"减速"动词。) 4. **frame** — Frame the problem as a design question, not a moral one. 把问题框定为设计问题,而不是道德问题。(frame as 决定听众用什么标准评判你,是陈述的第一动作。) 5. **land** — The closing line landed because it was concrete. 结尾句之所以奏效,是因为具体。(land 作不及物,主语是论点或笑话本身。) 6. **run through** — I'll run through three pieces of evidence, then take the objection. 我快速过三条证据,然后处理异议。(run through 预告"这段会快",听众会自动降低细节期待。) 7. **hold** — The argument holds only if demand stays flat. 该论点只有在需求持平的情况下才成立。(口语里 hold 比 is valid 自然,且更短。) 8. **concede ground** — Conceding ground early makes the rebuttal credible. 早早让出一部分立场,能让后面的反驳更可信。(concede ground on X 指明让的是哪一块。) 9. **takeaway** — If you remember one thing, make it this takeaway. 如果只记一件事,就记这一点。(takeaway 是可数名词,说 one takeaway,不说 one takeaway point。) 10. **in outline** — Here is the case in outline. 这是论证的概要。(用于开场后 30 秒内,让听众知道规模。) ## 三、语言点:能说出口的书面句 同一个论证,书面版和口头版不是"正式"与"随便"的差别,而是"可回读"与"只能听一遍"的差别。判断标准很物理:听众的工作记忆一次只能挂住一个主谓。 **规则** 1. **主语和谓语之间不超过 6 个词**。The increase in compliance costs associated with the new licensing regime has been substantial 里,主语到谓语隔了 11 个词,听众已经忘了主语是什么。口头版:Compliance costs have risen sharply under the new rules.(主谓相邻。) 2. **一句只带一个从句**。书面可以套两层(which...that...),口头超过一层,听众就开始猜你要去哪。要表达两层关系,拆成两句,用 and so / which means 这类轻连接。 3. **名词修饰不超过三层**。the deterioration in the quality of public service delivery 是四层嵌套,说出来会吞音、听起来像一团。改成动词句:Public services have got worse, and delivery is the weak link. 4. **口头优先用动词,书面优先用名词化**。这不是绝对律,而是负载问题:名词化把信息压进短语,读者可以回看;听众没有回看的机会。所以陈述里动词句占七成以上。 5. **数字必须当场给参照**。eighteen months 对听众没有意义,eighteen months — longer than it takes to build the plant 才有。参照物要用听众已知的东西,不要用行业内部口径。 **什么时候用**:所有口头产出(陈述、答辩、会议发言、面试)。写出来的讲稿也要按这个标准改一遍。 **什么时候不能用**:正式书面文本里刻意用短句会显得零碎;学术写作的严谨性有时依赖长名词短语,那里不能照搬口头规则。 **用了会显得怎样**:短主谓 + 单从句,显得清楚、有把握;过度口语化(加大量 you know、sort of)显得随意;把书面长句硬背出来,显得僵硬且随时会崩。 **强弱对比** | 书面版 | 口头版 | 差别 | | --- | --- | --- | | The implementation of pre-approval requirements for foundation models has been hampered by a shortage of qualified reviewers. | Pre-approval is behind schedule. The reason is simple: not enough reviewers. | 书面版一个主语挂了 12 个词;口头版拆成主谓 + 冒号解释,听众两次落点 | | Notwithstanding the aforementioned gains, the persistence of the backlog warrants continued concern. | The backlog is still growing. That is the part I would worry about. | notwithstanding / aforementioned 是阅读词,说出来听众要解码一秒,节奏就断了 | | The policy's effectiveness depends on whether enforcement capacity can be scaled at a rate commensurate with adoption. | Will enforcement keep up? That is the whole question. | 书面版是条件从句嵌套;口头版用短问句设问,再用一句定性,听众有参与感 | **常见误用** 1. **把逐字稿背下来**。背稿的痕迹很明显:语速均匀、停顿位置固定、一处忘词就整段崩。关键词卡的目的是让你每次说的词不一样,但走向一样。 2. **为口语化而加填充**。So, basically, what I'm trying to say is... 三个词零信息。口语化指的是句法简单,不是加废话。 3. **关键词卡写成完整句**。写满整句,你会不自觉地念;卡上只应有名词、动词、数字和箭头。 ## 四、句型与范例 **10 个口头路标句(按陈述顺序)** 1. **Let me start with the number that frames this.** 我先从这个数字说起。(开场给规模,比 Today I want to talk about 有效十倍。) 2. **Two pieces of evidence, then the strongest objection.** 两条证据,然后是最强的反对意见。(一句话交代路线图,听众知道还有多久。) 3. **Put differently, ...** 换个说法,……(发现自己说得太书面时的即时补救句,比重说一遍自然。) 4. **Here is what that looks like in practice.** 这在实践中是这样的。(从抽象转到例子,最平滑的过渡。) 5. **That is the case in outline; now the hard part.** 论证的骨架就是这样;接下来是难点。(承认难度,听众会提高注意力。) 6. **The strongest objection runs like this: ...** 最强的反对意见是这样的:……(reports speech 的带入方式,让听众站到对方立场。) 7. **I'll grant the first half of that.** 这前半句我接受。(精准让步:让一半,留一半,比 I agree to some extent 清楚。) 8. **What that objection assumes is ...** 这个反对意见的假设是……(反驳的标准入口,永远指向前提。) 9. **Which brings me back to where I started.** 这就回到我开头说的。(收束信号,听众知道要总结了。) 10. **If you take one thing away, make it this.** 如果只带走一句,就是这句。(结尾定调,后面必须跟一句 10 词以内的短句。) **关键词卡范例(对应 5 分钟陈述)** | 段 | 时长 | 关键词(≤5) | 这一段结束时听众应相信 | | --- | --- | --- | --- | | 问题 | 1:00 | 18 months → 0.1% fine → who pays | 现行事后罚款对慢速、弥散的伤害无效 | | 证据 | 1:45 | pharma review → recall gap → 3 cases | 事前审查在别的行业已经证明了必要性 | | 反驳 | 1:15 | incumbents ← capacity fixed → fund review | 护城河问题真实,但解法是扩容而非放弃 | | 主张 | 1:00 | largest models → statutory clock → pilot | 只对最大模型事前审查,并设法定时限 | **一段口头示范(约 80 词,证据段的说法)** > Here is what that looks like in practice. In pharmaceutical safety, pre-market review caught harms that recall powers could never have reached — because by the time a recall is issued, the damage is already in the patient. The same logic applies here. The difference is that a drug has one approval moment and a model has a thousand deployments. So the review has to be lighter, but it cannot be absent. 批注:① Here is what that looks like 把抽象拉回地面;② 破折号后给原因,不用 because 从句套从句;③ The same logic applies here 是最短的类比桥;④ The difference is that 主动指出类比的边界——这一步最容易被跳过,但它是可信度的来源;⑤ 最后一句用 but 收口,10 词以内,听众能记住。 ## 五、练习 **A 词汇与搭配(7 题)** 1. 填空:The talk ______ from diagnosis to remedy at the two-minute mark. 2. 改错:Let me unpack about this claim. 3. 用 frame 改写:We should treat this as a question of design, not morality. 4. 搭配判断:concede ground ______ the timetable(on / to / in) 5. 用 land 描述一个奏效的结尾(要求说明为什么奏效)。 6. run through 与 go through 在演讲语境下有何差别?各造一句。 7. 用 in outline 写一句开场,要求 12 词以内。 **B 句法与改写(5 题)** 1. 改成口头版(主谓相邻、单从句):The effectiveness of the scheme depends on whether local authorities have the capacity to process applications within the statutory period. 2. 拆句:The report, which was commissioned in 2023 and which drew on 40 interviews, concluded that the backlog was structural. 3. 去名词化:The deterioration in the quality of public service delivery has been widely noted. 4. 给数字加参照:The average wait is eighteen months. 5. 把书面连接词换成口头衔接:Notwithstanding these gains, the backlog has grown; consequently, the case for additional funding is strengthened. **C 输出(2 题)** 1. 交关键词卡:四段、每段≤5 个关键词、标注时长与"这一段结束时听众应相信什么"。 2. 不看卡说 2 分钟(只讲"问题 + 证据"两段)并录音;回听写 3 条:哪句说成了书面长句、哪处停顿超过 3 秒、哪个数字没给参照。 ## 六、参考答案 **A** 1. pivots(第三人称单数;pivot from A to B。) 2. 错。unpack 是及物动词,直接接宾语:Let me unpack this claim. 3. 示例:Let us frame this as a design question rather than a moral one.(frame as 后接名词短语;rather than 比 not 更稳,避免绝对化。) 4. on(concede ground on X,指明让的是哪一块。) 5. 示例:The closing line landed because it named a date, not a principle. 结尾句奏效,是因为它给了一个日期,而不是一个原则。 6. run through 强调"快、按顺序、不深入",适合预告;go through 中性,可深可浅。例:I'll run through three findings.(快速过)/We went through the findings in detail.(逐条讨论) 7. 示例:Here is the case in outline.(6 词)或 The case, in outline, runs as follows.(7 词) **B** 1. 示例:Will the scheme work? It depends on one thing: can local authorities process applications in time?(拆成设问 + 短答,主谓相邻,无嵌套从句。) 2. 示例:The report drew on 40 interviews. Its conclusion: the backlog is structural.(两个短句;用 Its conclusion: 替代 which 从句,信息不丢。) 3. 示例:Public services have got worse, and delivery is the weak link.(动词句 + 具体化,避免四层名词嵌套。) 4. 示例:The average wait is eighteen months — longer than it takes to build the plant.(参照物必须是听众已知的事物,且量级可比。) 5. 示例:Those gains are real. The backlog has still grown. That is why I would spend more, not less.(三个短句,用 still 和 That is why 承担转折与因果,删掉 notwithstanding / consequently。) **C 评分维度** 关键词卡看三点:① 每段≤5 个词,且词性是名词或动词(出现完整句即扣分);② 箭头能读出推进方向(A → B 表示"由 A 推出 B");③ 第四列"听众应相信什么"必须是可被反驳的判断,不能写"了解背景"。 2 分钟录音看三点:① 是否有 2 处以上说成书面长句(主谓间隔超过 6 词);② 停顿是否都在段与段之间,而不是句子中间;③ 是否至少有一个数字带了参照。三项中两项不合格,说明卡上关键词太多,回去删到 4 个以内。 ## 七、今日输出与自查 产出文件:`第04周_Day05_关键词卡.md`、`第04周_Day05_试说.m4a`(2 条,不剪辑)。 | 自查项 | 合格线 | 怎么查 | | --- | --- | --- | | 四段完整 | 问题、证据、反驳、主张各一段,时长标注齐全 | 缺任一段,5 分钟一定撑不满或讲超 | | 关键词数量 | 每段 1—5 个,全部非完整句 | 数词数,不看内容 | | 无逐字稿 | 文档里不出现连续 8 词的英文句子 | 搜句号,卡上不应有成句 | | 可听性 | 回听时长句不超过 2 处 | 主谓间隔超过 6 词即算 | | 反驳指向 | 反驳段第一句是"对方的假设是……" | 若第一句是"他们错了",重写 | **今天不必追求流利**。第一次试说一定磕。今天的合格线是"结构对、句子能说出口",流利度是明天计时演练要解决的事。 ## 八、明日预告 Day 6 演练 2 遍并计时,统计填充词,然后把本月 100 张词卡过一遍。目标是 5±0.5 分钟、填充词 ≤5 次/分钟。今天卡上关键词超过 5 个的段落,明天一定会超时——所以今晚就把多余的词划掉。