Grouping That
Actually Works

You have run group work that turned into one student doing everything, two watching, and one drawing. That is not a discipline problem and it is not a motivation problem. It is a design problem, and it has a fix that takes about ninety seconds to build into the task.

你一定帶過那種分組活動：一個人全做完、兩個人在旁邊看、一個人在畫畫。 那不是秩序問題，也不是動機問題，那是設計問題——而且解法只要在任務裡多花大約九十秒就能裝進去。

Section 01The arithmetic that decides everything

In a language class, talk time is the product. Not the worksheet, not the slides — the number of seconds each child spends producing English. So do the arithmetic before you do anything else:

在語言課裡，開口的時間就是產出本身。不是學習單、不是投影片， 而是每個孩子實際產出英語的秒數。所以在做任何其他事之前，先把這筆帳算清楚：

• Teacher asks, one child answers: one child speaks, everyone else listens. In a forty-minute lesson, most children say nothing at all. 老師問、一個孩子答：一個人講話，其他人在聽。一節四十分鐘的課下來，多數孩子一句話都沒說。
• Groups of four: each child gets roughly a quarter of the group's talking time — and in practice less, because one child usually takes more. 四人一組：每個孩子大約分到四分之一的時間——實際上更少，因為通常有一個人講掉比較多。
• Pairs: each child gets about half. That is the highest number available to you, and it needs no furniture moving at all. 兩人一組：每個孩子大約分到一半。這是你能拿到的最高值，而且完全不必搬桌椅。

Which gives the rule most foreign teachers arrive at eventually, usually after a year of trying bigger groups: pairs are the default, not the fallback. Use three or four only when the task genuinely cannot be done by two people — a role play with three characters, a survey that needs several respondents, a poster with distinct parts.

於是得到多數外師最後都會得到的那條規則，通常是在試了一年大組之後： 兩人一組是預設，不是備案。只有在任務真的兩個人做不完時才用三人或四人—— 三個角色的角色扮演、需要多位受訪者的調查、有不同分工的海報。

Section 02Why groups collapse — and why a speech will not fix it

The two failure modes have names in the literature, which is useful because naming them stops you taking them personally. Free riding is the student who lets the others carry the work. The sucker effect is what happens next: the willing student notices, resents it, and stops working too. By the third week, your best student has learned that effort is optional.

兩種失敗方式在文獻裡都有名字，知道名字很有用，因為那會讓你不再把它當成針對你個人的事。 搭便車（free riding）是指讓別人扛的那個學生。 老實人效應（sucker effect）是接下來會發生的事：認真的那個學生發現了、心裡不平，於是也不做了。 到第三週，你班上最認真的孩子已經學會「努力是可有可無的」。

Telling them to cooperate does not work, because nothing in the task requires cooperation. David and Roger Johnson's work on cooperative learning identifies five elements that separate a real cooperative group from students merely sitting together: positive interdependence, individual accountability, promotive interaction, social skills, and group processing. In a forty-minute English class where you see the students once a week, the first two do almost all the work.

叫他們「要合作」沒有用，因為任務本身根本不需要合作。 David 與 Roger Johnson 關於合作學習的研究指出，「真正的合作小組」與「只是坐在一起」之間有五個要素之別： 正向互賴、個人績效責任、促進性互動、社會技巧、團體歷程。 在一節四十分鐘、而且一週只見一次的英語課裡，前兩項幾乎完成了全部的工作。

Section 03Make them need each other

Positive interdependence means the task is built so that one student cannot finish it alone. Not "should not" — cannot. Three ways to build it, in ascending order of how well they work in a language class:

正向互賴的意思是：任務的設計讓一個學生沒辦法獨力完成。不是「不應該」，是「做不到」。 三種做法，依在語言課裡的效果由低到高排列：

1. One set of materials per pair. The weakest version, but free — a single worksheet or one set of cards forces at least a conversation about who writes. 一組共用一份材料。效果最弱，但零成本——一份學習單或一套字卡，至少會逼出一段「誰來寫」的對話。
2. Split the roles so each holds something different. One reads, one records, then they swap. Better, but a determined student can still do both. 分派不同角色，各拿不同的東西。一人唸、一人記，然後交換。比較好，但硬要的話一個人還是能全包。
3. An information gap. Student A holds half the information, student B holds the other half, and neither can see the other's paper. A has the bus timetable, B has the questions. A has the picture, B has to draw it from description. This is the strongest structure available to a language teacher, because the English is not decoration on the task — the English is the only way the information can cross the desk. 資訊差（information gap）。學生 A 拿一半資訊、學生 B 拿另一半，而且看不到對方的紙。 A 有公車時刻表，B 有問題；A 有一張圖，B 要靠聽描述把它畫出來。 這是語言老師手上最強的結構，因為英語不是任務上的裝飾——英語是資訊唯一能越過桌面的方式。

Section 04Make sure nobody can hide

Individual accountability is the half that prevents free riding, and it costs one sentence, said before they start:

個人績效責任是防止搭便車的另一半，而它的成本只是一句話，而且要在他們開始之前說：

• "I will choose one of you to report — you will not know who." This single sentence changes the behavior of both students, and it costs nothing. 「等一下我會指定你們其中一個來報告，你們不知道是誰。」這一句話就改變了兩個學生的行為，而且完全免費。
• Everyone writes their own copy, even when the pair produced one answer together. The talking is shared; the record is not. 每個人都寫自己的一份，即使答案是兩人一起想出來的。討論是共有的，紀錄不是。
• Number the students inside each group and call a number rather than a name. Every child has to be ready, and no child is singled out. 在每組內給學生編號，點號碼而不是點名字。每個孩子都得準備好，而且沒有人被特別針對。
• Rotate the reporter so the confident one does not become the permanent voice of the pair. 報告的人要輪替，免得比較敢講的那個變成這一組的永久發言人。

Section 05How to form the groups

• Proximity, for anything short. "Turn to the person beside you." In rows, the back row turns around to face the row in front. Zero furniture, zero transition, and you can do it four times in one lesson. 短任務用「就近分組」。「轉向你旁邊的人。」排排坐的話，後面一排轉過來面對前面一排。 不搬桌椅、沒有轉換時間，一節課可以做四次。
• Deliberate pairs, for anything long. If the pair will work together for fifteen minutes, decide the pairs yourself beforehand. 長任務用「事先指定」。如果一組要合作十五分鐘，配對就由你自己事先決定。
• Random, to break patterns. Playing cards, numbers, birthday months — useful occasionally, and it takes longer than you think. 偶爾用隨機分組打破固定圈子。撲克牌、抽號碼、生日月份——有用，但花的時間比你以為的多。
• Never let students pick, for anything that counts. It is fast and it feels generous, and it means the same child is left over every single time, in front of everyone. You will have taught that child something you did not intend to teach. 凡是有份量的活動，絕不讓學生自己選組。那很快、感覺也很寬鬆， 但結果是每一次都有同一個孩子被剩下來，而且是當著全班的面。你會教會那個孩子一件你並不想教的事。

Section 06Launch it in the right order

The most common way a well-designed group task dies is that the instructions were given after the students moved. Once furniture scrapes and twenty-something children stand up, nobody hears anything. The order is fixed:

一個設計良好的分組任務，最常見的死法是：指令是在學生移動之後才講的。 椅子一拖、二十幾個孩子一站起來，就沒有人聽得到任何東西了。順序是固定的：

1. Task — say what they will do, in one sentence. 任務——用一句話說他們要做什麼。
2. Time — say how long, and make it slightly too short. A task with slack in it drifts; a task that is slightly tight stays alive. 時間——說多久，而且要抓得稍微太短。有餘裕的任務會散掉，稍微緊一點的任務才有張力。
3. Product — say what has to exist at the end. "One sentence you will read out." "Three answers written down." Without this, groups talk and produce nothing. 產出——說結束時必須存在什麼東西。「一句你等一下要唸出來的句子」「寫下三個答案」。 少了這個，小組會聊天但什麼都沒生出來。
4. Demonstrate — do one round with a single pair at the front instead of explaining a second time. Thirty seconds of showing beats two minutes of telling, especially in a second language. 示範——找一對學生在前面做一輪，而不是再解釋一次。示範三十秒勝過講解兩分鐘， 在第二語言的情況下更是如此。
5. Then let them move. 最後才讓他們移動。

Section 07Three sentences to take into next week

1. Pairs are the default. Talk time is the product, and pairs give each child the largest share available. 兩人一組是預設。開口時間就是產出，而兩人一組給每個孩子最大的一份。
2. Split the information, then announce that you pick the reporter. That is the whole fix for free riding, and it takes ninety seconds. 把資訊拆開，然後宣布報告的人由你指定。搭便車的解法就這樣，九十秒。
3. Task, time, product, demonstrate — then let them move. Instructions given after the furniture moves are not instructions. 任務、時間、產出、示範——最後才讓他們動。桌椅移動之後才給的指令，不算指令。