# 英语高级 · 第 2 周 Day 5|四段式摘要写作 > 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** 前四天你一直在积累零件:搭配、矩阵、名词化、听记。今天把它们装成一台机器——150 词、四段、每段一个功能、每段一个词数额度。难点不在写够字数,而在分配:30 词写背景,意味着你必须放弃三条有趣但无关的信息;40 词写方法,意味着你要选出能让人判断可信度的那几项。今天还要求 3 处名词化、2 处弱动词搭配,且名词化不能超过 3 处。这个上限比下限更重要,它逼你在该用的时候才用。 ## 一、今日任务 | 环节 | 时长 | 具体动作 | 产出 | | --- | --- | --- | --- | | 复习 | 10 min | 读 Day 4 的 8 个名词化短语,口头各造一句 | 8 句 | | 拆原文 | 20 min | 选一篇本周读过的论文(含摘要与结果段),用四色标出背景、方法、结果、局限;标完数各部分的原文字数 | 四色标记稿 + 字数比 | | 列预算 | 10 min | 按 30 / 40 / 50 / 30 分配,写下每段"必须保留的 2 条信息"和"必须砍掉的 2 条" | 信息取舍表 | | 写初稿 | 30 min | 计时写 150 词,不查词典;每段写完立刻数词,超了就当场砍 | 初稿 | | 校对 | 20 min | 查四项:四段齐全、名词化恰好 3 处、弱动词搭配 2 处、每句有信息 | 定稿 + 标注版 | 写作时先写结果段。结果段决定其余三段留什么:背景要为结果服务,方法要解释结果的来源,局限要限定结果的边界。从背景写起的人,八成会写出一篇没有结果的摘要。 ## 二、核心词汇与搭配 | 词/短语 | 音标 | 词性 | 释义 | 语域 | | --- | --- | --- | --- | --- | | report | /rɪˈpɔːt/(≈里保特) | v. | 报告(中性,不暗示态度) | 学术 | | demonstrate | /ˈdemənstreɪt/(≈戴默恩斯特雷特) | v. | 证明、展示(强于 show) | 学术 | | estimate | /ˈestɪmeɪt/(≈诶斯提梅特) | v. | 估计(含不确定性) | 学术 | | control for | /kənˈtrəʊl fɔː/ | 动词短语 | 控制(变量) | 学术 | | be consistent with | /biː kənˈsɪstənt wɪð/ | 形容词短语 | 与……一致 | 学术 | | account for | /əˈkaʊnt fɔː/ | 动词短语 | 解释;占比 | 学术/商务 | | attribute X to Y | /əˈtrɪbjuːt/(≈厄特里布尤特) | v. | 把 X 归因于 Y | 学术 | | be limited to | /biː ˈlɪmɪtɪd tu/ | 形容词短语 | 仅限于 | 学术 | | generalise | /ˈdʒenrəlaɪz/(≈真若莱兹) | v. | 推广、外推 | 学术 | | hold across | /həʊld əˈkrɒs/ | 动词短语 | 在……范围内成立 | 学术 | **搭配与例句** 1. **report** — 搭配:report that / report a decline in。The authors report a 12% fall in attendance.(作者报告出勤率下降 12%。)report 只陈述事实,不评价;要评价另起一句。 2. **demonstrate** — 搭配:demonstrate that / demonstrate a link between。The trial demonstrates that the effect is causal, not merely correlational.(该试验证明效应是因果的,而不只是相关。)比 show 强,但需要设计支撑,观察性研究慎用。 3. **estimate** — 搭配:estimate that / estimate the effect at。They estimate the cost at $4bn a year.(他们估计成本为每年四十亿美元。)estimate 自带误差,后面常跟 confidence interval。 4. **control for** — 搭配:control for income / for prior attainment。The model controls for income and prior attainment.(模型控制了收入和先前学业水平。)介词只能是 for,不能省。 5. **be consistent with** — 搭配:be consistent with earlier findings / with a threshold model。The pattern is consistent with a threshold rather than a linear effect.(这一模式与阈值模型而非线性效应相符。)它是"不矛盾",不是"证明",属于典型的 hedge。 6. **account for** — 搭配:account for the difference / account for 40% of。Differences in coverage account for most of the gap.(覆盖范围差异解释了大部分差距。)两个义项要看宾语:解释原因,或占比例。 7. **attribute X to Y** — 搭配:attribute the decline to / be attributed to。The decline is attributed to tighter credit.(下降被归因于信贷收紧。)被动态 be attributed to 在摘要里极常见,但要注意它暗示因果已确立。 8. **be limited to** — 搭配:be limited to urban areas / to adults。The sample is limited to adults in formal employment.(样本仅限于正式就业的成年人。)局限段的骨架句,比 only includes 正式。 9. **generalise** — 搭配:generalise from / generalise to。It is hard to generalise from 42 cities to the tropics.(很难从 42 个城市外推到热带地区。)英式拼写 generalise,美式 generalize,全文统一。 10. **hold across** — 搭配:hold across regions / across specifications。The result holds across specifications and subsamples.(结果在不同设定与子样本下都成立。)这是稳健性检验的常用表述,比 is true for 精确。 ## 三、语言点:四段的信息预算与句子重心 摘要不是缩短版论文,是一次重新分配。每段只有一个功能,词数按信息价值给,不按原文比例给。 | 段落 | 词数 | 唯一功能 | 必须有的东西 | 禁止出现 | | --- | --- | --- | --- | --- | | 背景 | 30 | 说明为什么值得研究 | 一个具体现象 + 一个未解问题 | 教科书式定义、气候变化式套话 | | 方法 | 40 | 说明结论怎么来的 | 数据、样本、时间、设计 | 文献综述、理论动机 | | 结果 | 50 | 说明发现了什么 | 方向 + 幅度 + 条件 | 因果解释、政策建议 | | 局限 | 30 | 说明结论能走多远 | 一个设计局限 + 一个外推限制 | "样本量小"式空话 | | 规则 | 说明 | 例 | | --- | --- | --- | | 1. 一段一功能 | 结果段只报结果,解释留给讨论 | 结果段出现 because,基本越界 | | 2. 旧信息在前,新信息在后 | 句末是天然的重音位,放数字与结论 | Tracts with denser canopy were cooler **by 1.2 °C** | | 3. 名词化只作被讨论的对象 | 名词化当主语可以,当主语 + 空谓语不行 | The association survived controls(可);The assessment was conducted(弱) | | 4. 局限段给机制,不给借口 | 说清偏差方向,而非"研究有不足" | Canopy is measured in summer only(可);The study has limitations(空) | | 5. 每段首句承担路标功能 | 读者只读首句也能还原结构 | 方法段首句直接给数据来源 | 预算怎么算:背景 30 词约两句,一句给现象、一句给缺口;方法 40 词约两句半,能容纳四到五个参数;结果 50 词约三句,给方向、幅度、变异;局限 30 词约两句,一条测不了、一条推不远。这个分配不是美学选择,而是由信息密度决定:结果段单位词数的信息量最高,所以给得最多;背景段可用现成共识换字数,所以压得最狠。写完发现某段超了 8 词以上,不要从别段借,直接从该段删一句——借词只会把问题推到下一个段落。 **强弱对比** | 弱 | 强 | 差别 | | --- | --- | --- | | Climate change is a serious global problem. | Urban heat is unevenly distributed, and low-income neighbourhoods are often the most exposed. | 前者可放进任何文章,后者只能用在这篇;具体性就是信息量 | | We found some interesting results. | Tracts with denser canopy were consistently cooler, by 1.2 °C on average. | 前者零信息;后者给方向、幅度、条件 | | The study has several limitations, such as a small sample. | Canopy is measured in summer only, so winter effects remain unobserved. | 前者是免责声明;后者指出偏差的具体来源与方向 | **常见误用** - **背景写成教科书**:开头定义什么是城市化、什么是气候。摘要的读者是同行,背景只需一句"已知"加一句"未知"。 - **结果段混入讨论**:报了数字立刻解释"这说明政策有效"。解释属于讨论段,摘要里越界会让结果显得不可信。 - **名词化超预算**:为了显得学术,一段塞五个 -tion。本周的硬指标是 3 处,超过就删——名词化在本周是配额,不是装饰。 ## 四、句型与范例 下面是一篇 150 词的示范摘要(自拟研究,不对应真实论文),四段分别标注词数与功能。 > **Background (30 words)** Urban heat is unevenly distributed, and low-income neighbourhoods are the most exposed. Whether street trees narrow that gap or merely track it remains unclear, partly because canopy tracks income closely. > > **Methods (40 words)** We combined satellite surface-temperature data with a municipal tree inventory for 42 cities across three continents, then compared census tracts matched on income, density, building age and tenure, and weighted them by population; a fixed-effects specification absorbed most city-level differences. > > **Results (50 words)** Tracts with denser canopy were consistently cooler, by 1.2 °C on average, and the association survived controls for density and tenure. Its magnitude varied more than fourfold across cities, with the steepest gradients in the hottest ones, raising questions about uniform targets and carrying implications for how money is spent. > > **Limitations (30 words)** Canopy is measured in summer only, and planting may follow rather than cause cooler conditions; the sample also excludes informal settlements, where canopy is thinnest, so equity claims remain untested. **批注** - 背景段用两个事实句加一个 whether 从句。whether 的作用是把"已知"转成"未知",这是背景段唯一需要完成的动作;partly because 顺手交代了难点,省掉一句。 - 方法段 40 词里塞了数据源、城市数与洲别、匹配变量、加权方式、模型设定五样,靠分号与并列结构推进。分号在这里的作用是同级并列但不并列句子,比另起一句紧凑,也不会让读者误以为后句是结果。 - 结果段第一句给方向与幅度,第二句用 with 短语把异质性挂在句末,再用现在分词 raising / carrying 收束到评价。全段没有一处解释原因,这是结果段的本分。 - 局限段三条,都指向偏差方向(测量季节、因果方向、样本排除),没有一句「未来研究可以……」。30 词的额度刚好够三条各一句,多写一条就要牺牲一条的清晰度。 - 名词化恰好三处:the association、controls、a fixed-effects specification,都在被讨论的对象位置,没有一处是空谓语。弱动词搭配两处:raises questions about、carries implications for,都放在句子后半段,让新信息落在重音位。 | 可套用句式 | 功能 | 例句(中译) | | --- | --- | --- | | X is unevenly distributed, and Y is often the most exposed. | 背景,给具体现象 | Costs are unevenly distributed, and small firms are often the most exposed.(成本分布不均,小企业往往最受影响。) | | Whether X does A, rather than merely B, remains unclear. | 背景,点出未解问题 | Whether the tax changes behaviour, rather than merely raising revenue, remains unclear.(该税是否改变行为,还是只增加收入,仍不清楚。) | | We combined X with Y for Z cases. | 方法,交代数据 | We combined survey data with administrative records for 12 regions.(我们结合调查数据与行政记录,覆盖十二个地区。) | | The association survived controls for X. | 结果,说明稳健 | The association survived controls for firm size and sector.(控制企业规模与行业后,关联依然成立。) | | Its magnitude varied X-fold across Y. | 结果,给异质性 | Its magnitude varied threefold across regions.(其幅度在地区间相差三倍。) | | X is measured in Y only. | 局限,指测量偏差 | Emissions are measured at the plant only.(排放只在厂区测量。) | | X may follow, rather than cause, Y. | 局限,指因果方向 | Hiring may follow, rather than cause, growth.(招聘可能追随增长,而非导致增长。) | | The sample excludes X, so Y remains unobserved. | 局限,指外推边界 | The sample excludes rural households, so the pattern there remains unobserved.(样本不含农村家庭,因此那里的模式未知。) | ## 五、练习 **A 词汇与搭配(7 题)** 1. 填介词:The model controls ___ household size. 2. 填介词:The result is consistent ___ earlier work. 3. 填介词:The gap is largely attributed ___ differences in coverage. 4. 改错:The study demonstrates a correlation, so the effect is causal. 5. 改错:The sample is limited in urban areas. 6. 改错:These findings can be generalised for other sectors. 7. 选词(estimate / report):The authors ___ that 30% of the gap remains unexplained. **B 句法改写(5 题)** 8. 把"零信息句"改成有信息的结果句:The results were significant and interesting. 9. 压缩到 25 词内(背景段):Cities around the world are getting hotter every year, and this is a problem for many people, especially those who do not have much money. 10. 把讨论混入的结果段拆开:Canopy reduced temperatures by 1.2 °C, which shows that planting programmes are the best policy tool available. 11. 把空话局限改成机制局限:The study has some limitations that should be considered. 12. 用 end-weight 重排:A 1.2 °C difference, on average, was observed in tracts with denser canopy. **C 输出(1 题)** 13. 写 150 词四段式摘要(背景 30、方法 40、结果 50、局限 30),材料用你今天拆的那篇论文。硬性要求:3 处名词化(标注)、2 处弱动词搭配(标注)、总词数 150±15、四段齐全。写完在页边写出每段实际词数。 ## 六、参考答案 1. **for**。control for 是固定搭配,表示"把变量纳入模型";control 单独用时是"控制"。 2. **with**。consistent with 表"与……相符",to 不成立。 3. **to**。attribute X to Y,被动态为 be attributed to;不要用 by,by 会变成"由谁归因"。 4. **The study demonstrates an association; it cannot establish causation.** demonstrate 需要设计支撑,相关性研究只能用 show an association 或 point to。 5. **The sample is limited to urban areas.** limited to 接范围;limited in 接"在某方面受限"。 6. **These findings can be generalised to other sectors.** generalise 后接 to,表示外推的目标域。 7. **estimate**。30% 未解释是一个带误差的估计,用 estimate 而非 report;report 用于可直接观测的数字。 8. **Canopy density was associated with a 1.2 °C reduction in surface temperature.** 原句的 significant 是统计术语的误用,且 interesting 是作者感受;改写给出变量、方向、幅度。 9. **Urban heat is unevenly distributed, and low-income neighbourhoods are the most exposed.**(13 词)原句 27 词,信息只有一个;压缩后保留"不均"与"谁最受影响",删掉 every year 与 around the world 这类无差别修饰。 10. **Canopy reduced temperatures by 1.2 °C. Planting programmes may therefore be worth prioritising, though the comparison is indirect.** 前半句留在结果段,后半句移到讨论;may therefore 与 though 保留推断强度,避免把相关性说成最优政策。 11. **Canopy is measured in summer only, so winter effects remain unobserved.** 把"有局限"换成"哪一项没测、因此什么未知",读者才能判断结论边界。 12. **Tracts with denser canopy were, on average, 1.2 °C cooler.** 原句把数字放在句中,重音散了;改写把 1.2 °C cooler 放到句末,读者读完即记住幅度。 13. **评分维度**:① 四段词数各在 ±5 内(30/40/50/30);② 名词化恰好 3 处,多一处扣 5 分;③ 弱动词搭配 ≥2 处且介词正确;④ 结果段无 because、无 should;⑤ 局限段给机制不给借口。参考开头:*Heat exposure is unevenly distributed within cities, and the gap has widened since 2000. Whether tree planting narrows it, rather than merely tracking it, is contested.* 两句背景,一句未解问题,28 词。 ## 七、今日输出与自查 | 检查项 | 合格线 | 我的结果 | | --- | --- | --- | | 总词数 | 150±15,四段各在 ±5 内 | | | 名词化 | 恰好 3 处,位置在"被讨论的对象" | | | 弱动词搭配 | ≥2 处,介词全对 | | | 结果段纯净度 | 无因果解释、无政策建议 | | | 每句有信息 | 通读一遍,删掉任何可删句后意思不变即为空话 | | 产出文件:`第02周_Day05_150词摘要.md`,正文 + 页边词数 + 3 处名词化与 2 处搭配的标注。明天对照原文逐段核查,今天先不回头改。 ## 八、明日预告 Day 6 复盘:把摘要与原文逐段对照,标出漏掉与多加的信息,目标是把漏掉的关键信息压到 1 条以内;同时复习词卡,整理 8 条错句。