【1】On a Thursday afternoon in Shanwei, Guangdong, fifth-graders are cracking an acid-base code, racing a cardboard bridge against a stopwatch and working out whose fingerprint is on the glass.It looks like a science club that has given up on teaching; it is a curriculum built to teach, with 70 per cent of eight sessions spent inside a game.
【2】The design rests on a claim that has always been awkward for enthusiasts: game elements teach nothing by themselves.What they do, in the theory of gamified learning, is alter a pupil's state of mind, which alters behaviour, which alters the score — a chain that can come apart at any link.That first link had long looked modest, so the question was never whether games excite children but whether the excitement reaches the exam paper.
【3】A trial in Frontiers in Education on 26 October tested that question in one public primary school, assigning six intact fifth-grade classes, 269 pupils in all, to the gamified programme or to ordinary teaching.Its twenty-one game elements were mapped onto the four domains of the national science curriculum, so the course could be defended to a head teacher as easily as enjoyed by a child.
10 月 26 日发表于《Frontiers in Education》的一项试验，在一所公立小学里检验了这个问题：六个完整的五年级班、共 269 名学生，以整班为单位被分到游戏化课程或常规教学。它的 21 个游戏元素被逐一对应到国家科学课程标准的四个领域，所以这门课既能向校长交代，也能被孩子喜欢。
【4】After two months the ordinary classes had moved from 55.8 to 60.8 on a test of scientific literacy.The gamified classes, which began lower, finished at 66.9 — an adjusted post-test mean of 67.4 against 60.4, a gap the baseline cannot explain.
【5】The authors then took the chain apart: the route through classroom engagement alone was significant, the route through motivation and then engagement was significant but tiny at 0.007, and the route through motivation alone had a confidence interval that crossed zero.Teaching condition explained 2.2 per cent of the variance in motivation: enough to be real, too little to carry a thirteen-point gain.
【6】Enjoyment, on this evidence, is not the active ingredient but the raw material: a child who wants to learn but never raises a hand, never argues with a partner and never revisits a wrong answer has changed in a way that questionnaires detect and examiners do not.Motivation has to be translated into behaviour before it can be translated into marks.
【7】A second experiment, in BMC Psychology, isolates the layer schools are now being sold: 174 fifth-graders took one of three mathematics courses — with an AI companion, without one, or lectured — and the AI version lowered anxiety while adding almost nothing on top of gamification.The difference was d = 0.19, not significant; both gamified groups still beat the lecture, which is what matters to a school buying nothing at all.
《BMC Psychology》上的第二项实验，单独考察了学校正在被推销的那一层：174 名五年级学生上三种数学课之一——配 AI 同伴的游戏化课、不配 AI 的游戏化课、传统讲授——结果 AI 版明显降低了焦虑，但在游戏化之上几乎没有增加成绩。差值为 d = 0.19，不显著；而两个游戏化组都仍然胜过讲授，这对一所"什么都不买"的学校才是关键。
【8】Neither study is a verdict on screens: classes rather than children were randomised, motivation and engagement were measured on the same afternoon, and nobody checked whether the gains survived the summer.What they establish is narrower and more useful — a well-built game reliably changes the climate of a classroom, and unreliably changes the behaviour that produces a score.
【9】For a teacher the implication is almost mundane: a game is a door, not a room; points and time pressure will get a class through it, but only argument, error and reflection will keep them there.
原文出处：本文据 Li, Y., Gu, Y., He, W., Zheng, L. & Liu, W.《The influence mechanism of gamified teaching on primary school students' scientific literacy: an empirical analysis based on structural equation modeling》，Frontiers in Education 11:1899547（2026-10-26，DOI 10.3389/feduc.2026.1899547）；以及 Shao, J., Jinming, J. & Tang, S.《The effects of an AI-supported gamified learning program on elementary students' mathematics anxiety, learning motivation, and achievement》，BMC Psychology（2026，DOI 10.1186/s40359-026-05367-8）公开摘要与结果改写整理，供学习使用。
三、长难句分析（三步拆解法）
What they do, in the theory of gamified learning, is alter a pupil's state of mind, which alters behaviour, which alters the score — a chain that can come apart at any link.
主干：What they do ... is alter ...（主语从句 + is + 动词原形作表语，构成"他们所做的是改变……"）
插入语：in the theory of gamified learning（介词短语插在主谓之间，交代理论出处，读时先跳过）
两个非限制性定语从句：which alters behaviour, which alters the score（前一个 which 指 state of mind，后一个指 behaviour，形成"多米诺"式递推）
破折号同位语：a chain that can come apart at any link（把前面三级递推概括成"一条链"，并用 that 从句点出"任何一环都可能断"）
The authors then took the chain apart: the route through classroom engagement alone was significant, the route through motivation and then engagement was significant but tiny at 0.007, and the route through motivation alone had a confidence interval that crossed zero.
主干：The authors ... took the chain apart（主谓宾 + apart 作宾补，"拆开"）
冒号后的三项并列：the route ... was significant, the route ... was significant but tiny, and the route ... had a confidence interval（三个分句同构，唯一变化是"经过哪一环"，形成对照实验的写法）
A second experiment, in BMC Psychology, isolates the layer schools are now being sold: 174 fifth-graders took one of three mathematics courses — with an AI companion, without one, or lectured — and the AI version lowered anxiety while adding almost nothing on top of gamification.
主干：A second experiment ... isolates the layer（主谓宾；isolate 在此意为"单独拎出来考察"）
省略关系词的定语从句：the layer (that) schools are now being sold（layer 后省略 that；be sold 为被动进行时，字面"正被卖给学校"，暗指教育技术市场）
冒号后的证据：174 fifth-graders took one of three courses（用具体人数与分组支撑前面的判断）
破折号列举 + while 转折：— with an AI companion, without one, or lectured — 说明三种条件；while adding almost nothing 表让步对比，"情绪改善明显，成绩增益几乎为零"
考点：省略关系代词；被动进行时的修辞色彩；one 的替代（without one = without an AI companion）；while 表对比
译文：《BMC Psychology》上的第二项实验，单独考察了学校正在被推销的那一层：174 名五年级学生上三种数学课之一——配 AI 同伴、不配 AI，或传统讲授——结果 AI 版降低了焦虑，但在游戏化之上几乎没有增加成绩。
Neither study is a verdict on screens: classes rather than children were randomised, motivation and engagement were measured on the same afternoon, and nobody checked whether the gains survived the summer.
主干：Neither study is a verdict on screens（主系表；neither 置于句首，全句已是否定，不再加 not）
冒号后的三条限制：classes rather than children were randomised / ... were measured on the same afternoon / nobody checked whether ...（三条并列，分别指向随机单位、测量时点、追踪时长）
rather than 的对比：classes rather than children（分的是班而非个人，因此班内学生不独立，统计上更保守也更脆弱）
宾语从句：whether the gains survived the summer（survive 在此为"保持、延续下来"，不是"幸存"）
考点：neither 的倒装与否定；rather than 的平行；被动语态并列；survive 的引申义；冒号列举
第二项研究把镜头推到了学校正在花钱购买的那一层。发表在 BMC Psychology 上的三臂试验由马来西亚理科大学的 Jiawei Shao 团队与国内研究者合作完成，174 名中国五年级学生按个体随机分为三组：配 AI 同伴的游戏化数学课（AI-GILP）、不配 AI 的游戏化课（GILP）、传统讲授。设计上的讲究在于，它没有拿"新技术"去对比"什么都没有"，而是把游戏化固定住，只让 AI 这一层发生变化。两周干预，四周后延时后测。情绪与动机上的结果很鲜明：AI 组的数学焦虑显著更低、学习动机显著更高，偏 eta 方分别为 0.208 和 0.184，在教育研究里属于不小的效应；而且增益集中在"内在动机"和"能力感"两个维度，而不是"努力/重要性"——也就是说，AI 同伴并没有让孩子更用力，而是让他们更相信自己能行、更愿意学。
但成绩是另一回事。两个游戏化组都显著优于传统讲授，说明游戏化本身确实带来了学业收益；而 AI 组相对"只有游戏化"的那一组，成绩差异只有 d = 0.19，未达显著。研究者对此相当克制：他们明确指出，系统并没有"读出"孩子的情绪，它监测的只是错误频率、停留时间、连续不活动和求助次数这些可观察指标；所谓心理需求支持、情绪调节等机制也没有被直接检验。换句话说，这项实验能证明的是"整套 AI 包"改善了课堂的情绪气候，而不能证明 AI 真的理解了孩子的焦虑。
Which of the following best captures the article's main argument?
A. Gamified science lessons raise test scores mainly because they make pupils enjoy the subject more than ordinary lessons do.
B. Because the gamified classes caught up from a lower baseline, the reported gain of more than thirteen points is an artefact of regression to the mean.
C. The score gain is real, but it runs through what pupils do in class rather than through how much they say they want to learn, so enjoyment is a means rather than the mechanism.
D. Since neither study randomised individual pupils, nothing can be concluded about whether games change learning at all.
仿写题：仿照第 6 段 "Enjoyment, on this evidence, is not the active ingredient but the raw material: a child who wants to learn but never raises a hand ... has changed in a way that questionnaires detect and examiners do not."（先用"不是 A 而是 B"下判断，再用冒号引出一个具体反例，最后用"某人测得到、某人测不到"的对照收尾），就"把点赞当成认可"写一句你自己的判断句。
