# 英语高级 · 第 8 周 Day 5|五源四百词初稿反方段 > 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** 今天把前四天的零件装成一篇四百词 synthesis,九十分钟,五个来源,其中一个段落必须写反方。难点不在写,在选与排:五份材料读起来都对,但哪两份当主力、哪三份只出现一次,决定了文章有没有重心;反方段放在共识之后还是结尾之前,决定了读者觉得你是在辩论还是在总结。今天要求反方段给出"最强版本"——对方最有道理的那条理由,而不是最容易打的那条。这是本周唯一一次写整篇的机会,写完的稿子明天只改不重写。 ## 一、今日任务 | 环节 | 时长 | 具体动作 | 产出 | | --- | --- | --- | --- | | 1. 来源矩阵 | 15 min | 五份材料各写一句话主张、一句最强证据、一条局限,填成五行三列 | 来源矩阵一张,五行填满 | | 2. 定三段结构 | 10 min | 只写三个主题句:共识一句、分歧一句、反方段一句;三句必须都是判断,不是"某研究认为" | 三个主题句,标出各自用哪几个来源 | | 3. 写四百词 | 45 min | 计时写作,按"共识—分歧—反方—结论"推进;写的时候不回原文,引用先写作者姓,页码空着 | 初稿四百词上下 | | 4. 加固反方段 | 10 min | 用三问检查:对方最强理由写了吗?支持它的证据给了吗?我的回应是否只针对那条理由? | 反方段修订稿 | | 5. 标引用与词数 | 10 min | 回原文补页码,标出每处引用支持的是哪一句判断;统计词数 | 引用清单与词数记录 | 第 3 步不回原文,是一条硬规则。回原文会让你开始改写句子,四十五分钟写不完四百词;凭记忆写,写错的地方第 6 天查引用时正好能查出来,那才是查引用该干的活。 ## 二、核心词汇与搭配 | 词/短语 | 音标 | 词性 | 释义 | 高频搭配 | 例句(含译) | 语域 | | --- | --- | --- | --- | --- | --- | --- | | converge | /kənˈvɜːdʒ/(≈克恩韦知) | v. | 趋同,汇聚 | findings converge on; converging evidence | The five studies converge on one conclusion: the effect is real but small. 五项研究指向同一结论:效应真实,但很小。 | 学术 | | diverge | /daɪˈvɜːdʒ/(≈戴韦知·长) | v. | 分歧 | accounts diverge over; diverge sharply | The two accounts diverge over why the effect appears in cities but not in towns. 两种说法在"为何效应只出现在城市"上分歧明显。 | 学术 | | consensus | /kənˈsensəs/(≈克恩塞恩瑟斯) | n. | 共识 | a broad consensus that…; the consensus is thinner than | The consensus is thinner than the reviews suggest. 共识比那些综述暗示的要薄。 | 学术 | | reconcile | /ˈrekənsaɪl/(≈雷克恩赛尔) | v. | 调和(矛盾) | reconcile conflicting findings | The results can be reconciled if the samples are weighted differently. 如果样本加权方式不同,这两个结果可以调和。 | 学术 | | tension | /ˈtenʃn/(≈泰恩希恩) | n. | 张力,矛盾 | a tension between X and Y | There is a tension between the paper's policy claim and its own data. 该论文的政策主张与它自己的数据之间存在张力。 | 学术 | | strand | /strænd/(≈斯特兰德) | n. | (论证的)一条线索 | a strand of the argument; two strands of evidence | The literature has two strands, one quantitative and one ethnographic. 这批文献有两条线索,一条定量,一条民族志。 | 学术 | | scope | /skəʊp/(≈斯科普) | n. | 范围,适用边界 | beyond the scope of; narrow the scope | Claims about "workers" outrun the scope of a sample of graduates. 关于"劳动者"的论断超出了一份毕业生样本的适用范围。 | 学术 | | understate | /ˌʌndəˈsteɪt/(≈阿恩德斯泰特) | v. | 低估,轻描淡写 | understate the extent of; to understate it | To call the evidence mixed understates how little of it there is. 说证据"参差",已经低估了它的稀少程度。 | 通用 | | weight of evidence | /weɪt əv ˈevɪdəns/ | n. | 证据分量 | on the weight of evidence | On the weight of evidence, the safer reading is that the effect fades after two years. 就证据分量而言,更稳妥的解读是效应在两年后消退。 | 学术 | | hold up | /həʊld ʌp/ | phr. v. | 经得起检验 | hold up under scrutiny; the finding holds up | The result holds up when the sample is restricted to full-time staff. 把样本限定为全职员工后,结果依然成立。 | 通用 | 这批词是给 synthesis 用的,可以按"关系"分成三对加四个单件。converge 与 diverge 是一对方向词,写共识段与分歧段各用一个;reconcile 与 tension 是处理矛盾的两种姿态,前者用于"其实不矛盾",后者用于"确实矛盾且值得保留";consensus 与 scope 是一对限定词,一个限共识的厚度,一个限结论的边界。剩下的 strand 用来给文献分类,understate 用来压低证据强度,weight of evidence 与 hold up 是判断句的动词短语。 两个搭配要点。converge 后面接 on,converge on a conclusion;diverge 后面接 over 或 from,diverge over causes 是"在原因上分歧",diverge from the trend 是"偏离趋势"。consensus 前用 broad、growing、fragile 修饰,不说 big consensus。 ## 三、语言点:段落三拍与反方段的位置 synthesis 的段落不是"来源串联",是"判断—证据—回落"的三拍循环。少了第三拍,读者只看到资料;少了第一拍,读者不知道这些资料为什么放在一起。 **规则** 1. **主题句必须是判断句。** The evidence points one way on pay and the other on promotion. 这句话有方向、可被反驳。写成 Smith (2019) argues that…,读者只知道你读过 Smith,不知道你要证明什么。 2. **来源做状语或定语,主语留给证据。** As a 2019 survey of 4,000 nurses shows, … 或 a 2019 survey of 4,000 nurses (Smith) finds…。主语是调查、数据、研究,读者才会把注意力放在证据强度上。 3. **每个来源后面必须有一句"这说明什么"。** 这句不引任何人,是你自己的判断;没有它,段落就退化成文献清单。 4. **反方段放在共识之后、结论之前。** 先让读者知道你了解主流证据,再给最强反方,最后回应。放在开头,读者不知道你在反对谁;放在结尾,文章会以对方的理由收场。 5. **反方段三步走:最强版本、支持它的证据、它成立的条件。** 三步缺一,都会变成稻草人:缺第一步是打错人,缺第二步是打空气,缺第三步是宣布胜利。 **强弱对比** | 弱 | 强 | 差别在哪 | | --- | --- | --- | | Smith (2019) argues that remote work reduces promotion. Jones (2021) finds similar results. Li (2022) disagrees. | The evidence converges on one point: promotion rates fall for junior staff who rarely attend in person. Where it diverges is why — access to mentors, or visibility to managers. | 前者按作者排列,读者要自己找关系;后者先给判断,再给分歧的位置 | | Some critics say remote work is fine because productivity rose. | The strongest objection is that output per hour rose during the shift, which suggests the loss falls on careers rather than on current work. | 前者是无主体的稻草人;后者给出对方最好的理由,并说清它为什么有力 | | In conclusion, remote work has both costs and benefits. | If the cost falls on careers rather than output, firms have no market incentive to fix it — which is the case for a policy, not a memo. | 前者重复开头;后者把判断推进到"所以该做什么" | **常见误用** 1. **按来源组织。** 一段一个作者,五段五个作者,读完不知道共识在哪。synthesis 的最小单位是主题,不是来源。 2. **反方段写成 some critics say。** 没有具体主体、没有证据,读者会认为你在自造对手;这是最快失去信任的写法。 3. **结论重复开头。** 开头说"影响复杂",结尾还说"影响复杂",四百词等于原地踏步。结论要向前一步:给出条件、代价或建议。 **用与不用的分寸** 三拍结构适合所有说明性段落,但反方段只在你真的能找到对方最强理由时写。找不到,就说明你还没读够,或者你的立场不需要反方——后者少见。反方段也不宜过长:四百词里一百三十词左右,超过一半,读者会以为你站在对方那边。至于"最强版本",标准是可检验的:对方读到你的复述,会说"对,这就是我的意思",而不是"你歪曲我"。 ## 四、句型与范例 | 句式 | 功能 | 中文 | | --- | --- | --- | | The evidence converges on one point: … | 共识段主题句 | 证据在一点上趋同:…… | | Where it diverges is why… | 从共识转向分歧,最省字 | 分歧在于原因:…… | | On the weight of evidence, … | 给判断加分量,不假装确定 | 就证据分量而言,…… | | That finding holds up when… | 说明稳健性,回应样本质疑 | 这一发现在……时依然成立 | | The strongest objection is not… but… | 反方段开头,先排除易打的版本 | 最强的反对不是……而是…… | | The objection would bite if… | 给反方设成立条件 | 如果……,这一反对就有杀伤力 | | What it understates is… | 指出对方低估的部分 | 它低估的是…… | | Even granting that…, the claim outruns its scope. | 承认加限定,收反方段 | 就算承认……,这一主张也超出了适用范围 | | Read together, these studies suggest… | 综合收束,转入自己的判断 | 合起来看,这些研究表明…… | | The policy question is therefore not whether… but… | 结论推进到行动 | 因此政策问题不是……而是…… | **反方段示范(约一百五十词)** > The strongest objection to this reading is that remote work did not reduce output, and there is good evidence for it: output per hour rose in three of the four firms that published figures. If the concern were productivity, that evidence would settle the matter. But it does not, because the loss shows up where annual figures cannot see it — in promotion rates, which fell by a fifth for staff under thirty. The objection would bite if promotions were driven mainly by output; the firms' own appraisal data suggest they are driven by who is in the room. Even granting that output held up, then, the claim that remote work is costless outruns its scope: it is a claim about this year's output, not about this cohort's careers. 批注三条。第一,开头先给出对方最好的证据(产出上升),而不是先给结论,这样读者才相信你读过对方。第二,The objection would bite if… 一句把反方的成立条件写明,随后用"评估数据"指出条件不成立,反击落在机制上,不落在态度上。第三,末句 Even granting that… 收得住:承认对方的事实,只否认它的适用范围,这比说"对方错了"更难被反驳。 ## 五、练习 **A 词汇与搭配(7 题)** 1. 填空:The four evaluations ______ on a single finding: the effect fades after two years.(converge,现在时) 2. 改搭配:There is a big consensus among researchers. 3. 选词(reconcile / diverge):The two results can be ______ once the age of the samples is taken into account. 4. 中译英:关于"劳动者"的论断超出了一份毕业生样本的适用范围。 5. 词形转换:converge → 名词;reconcile → 形容词。 6. 语域升格:把 A lot of studies say the same thing. 改成学术语域的一句,须出现 weight of evidence。 7. 造句:用 hold up 写一句,主语是 the finding,后接一个检验条件。 **B 句法(6 题)** 1. 改主题句:把 Smith (2019) argues that remote work reduces promotion. 改成判断句,仍保留该来源。 2. 合并三句为一段主题句加证据:a. 三项研究都发现晋升率下降。b. 两项研究认为原因是指导机会减少。c. 一项研究认为是曝光度下降。 3. 加"回落"句:为 The 2019 survey found that 42% of junior staff had never met their mentor in person. 补一句"这说明什么"。 4. 反方段开头改写:把 Some people think remote work is harmless. 改成"最强版本"开头。 5. 压缩到三十词以内:Even if we accept that output per hour rose during the period in question, it does not follow that the arrangement is costless for the individuals concerned. 6. 结论推进:把 In conclusion, the effects are mixed. 改成带条件的判断。 **C 输出(2 题)** 1. 四百词初稿:五源,四段(共识、分歧、反方与回应、结论),反方段一百至一百四十词。 2. 论证结构图:一张,节点含主张、三条证据、反方最强点、回应、结论,箭头标 supports 与 weakens。 ## 六、参考答案 **A** 1. converge。主语 the four evaluations 为复数,用原形。 2. There is broad consensus among researchers. consensus 前用 broad、growing、fragile,不用 big。 3. reconciled。take into account 后接因素,表示"考虑之后可以调和"。 4. Claims about "workers" outrun the scope of a sample of graduates. 5. convergence;reconcilable。 6. On the weight of evidence, the studies point the same way. 用 weight of evidence 替代 a lot of,语气从"很多"变成"分量"。 7. The finding holds up when part-time staff are excluded from the sample. 检验条件必须具体到样本或方法。 **B** 1. Promotion rates for junior staff fall when remote work becomes the default (Smith 2019). 判断在前,来源在后,是学术写作的标准次序。 2. Three studies find promotion rates falling, and the explanations split: two point to mentoring, one to visibility. 先共识后分歧,来源数量用数字带过。 3. That figure matters because mentoring in these firms happens informally, in corridors rather than in scheduled meetings. 回落句要解释"为什么这个数字重要"。 4. The strongest case for remote work is that output per hour rose while staff worked from home, which suggests the arrangement costs firms nothing in the short run. 给出对方最好的理由与证据。 5. Output per hour rose, but that does not make the arrangement costless for the individuals concerned.(十九词) 6. The effects are mixed, but they are not symmetric: the costs fall on junior staff, who have the least power to negotiate around them. 结论带条件与分配,向前一步。 **C 评分维度** 第 1 题:四段是否各有判断句;来源是否做状语或定语;反方段是否有最强版本、支持证据、成立条件三步;五源是否两主三辅;词数是否四百上下三十。第 2 题:节点是否含"反方最强点"而非"反方观点";箭头是否至少两条 weakens;回应是否指向反方的成立条件。 ## 七、今日输出与自查 | 检查项 | 合格线 | 我的记录 | | --- | --- | --- | | 词数 | 四百上下三十,四十五分钟内完成 | | | 来源 | 五个,两主三辅,每源至少一处具体证据 | | | 反方段 | 含最强版本、支持证据、成立条件 | | | 主题句 | 四段皆为判断句,无"某研究认为"开头 | | | 引用 | 每处可回溯到页码,标注支持哪句判断 | | 产出文件名:synthesis初稿_Day05.md(四百词)、论证结构图_Day05(图片或 PDF)。初稿保留原样,不要边写边改;明天的三查要拿它当证据。 ## 八、明日预告 Day 6 只做一件事:改。六十分钟,三查一读——先核引用,再查连续五词,再体检连接词,最后朗读一遍。顺序不能换:先改文字会让引用偏移,先改连接会让逻辑缺口被盖住。今天写的时候页码空着的地方,明天要一个个补上;补不出来的,就是你不该引的地方。