现有单词学习APP的呈现序列存在学习低效问题，以“天”、“周”等长时间单位的间隔效应作为经典记忆序列规律有大量实验论证了有效性，但较少论证以“小时”、“分钟”此类短时间单位的稳定性。本文通过桌面调研、实验室实验、ERPs技术，研究间隔效应在短时间单位的稳定性，并探索其原理，将其应用于单词学习APP的序列优化。

实验一利用真假词启动效应范式，设置3种不同学习间隔条件水平，分别考察即时和延时测验的正确率、反应时。实验一的结果表明，混合间隔序列的学习效果优于完全间隔序列和集中学习。在单次的项目学习中适当增加学习时间，达到过度学习的程度，有利于提升学习效果，此增益效果可以与间隔效应叠加。

实验二利用新旧效应范式，先通过行为实验探究绝对间隔长度与学习效果的关系，以及刺激呈现顺序与学习效果的关系；再利用ERPs技术分析不同间隔长度导致的不同学习效果的脑活动原理。实验二的行为研究结果表明，间隔效应对学习效果与绝对间隔数存在非线性关系。间隔数如果短于工作记忆容量或远长于工作记忆容量，两者的学习效果会优于集中学习和接近工作记忆容量的间隔。先词后图的呈现顺序优于先图后词。单词学习微观顺序应当遵循先激活新形式，再靶向确定旧表征，防止表征间的节点互通损耗认知资源。实验二的脑电研究结果表明，学习间隔越长，关系提取的时间进程越快，正确率越高。但结合实际学习的应用场景，短间隔学习在追求高效学习背景下表现更优。短间隔学习相较集中学习，在关系提取的时间进程更快；相较于长间隔学习，对项目列表总长度的要求更低，且可以利用工作记忆内容实现过度学习，一定程度上弥补了与长间隔学习在时间进程上的劣势。

从产品设计方面，下一代英语学习APP可以在单词学习呈现顺序上进行以下优化：1、在单次项目学习和复习中，设置额定最低学习时间以达到单次项目的过度学习；2、同一项目应设置多次学习和复习，且在错误单词复习中，应遵循两单词之间间隔2-7个不同项目，避免间隔在工作记忆容量附近；3、尽量分离新单词与单词图片+中文翻译的呈现时间，做到先呈现新单词，后呈现图片+翻译，并在测试中仅展示中文翻译，不展示图片，以达到最佳的学习效果。

The present sequences of existing word-learning apps are inefficient in learning. A large number of experiments have demonstrated the validity of the interval effect of long time units such as "day" and "week" as the classic memory sequence rule, but few have demonstrated the stability of short time units such as "hour" and "minute". In this paper, through desktop research, laboratory experiments, and ERPs technology, the stability of interval effect in short time units is studied, and its principle is explored, and it is applied to the sequence optimization of word learning APP.

Experiment 1 used the true-false word priming effect paradigm to set 3 different learning interval condition levels to examine the correct rate and reaction time of the immediate and delayed tests respectively. The results of experiment 1 show that the learning effect of mixed interval sequence is better than that of fully interval sequence and centralized learning. Appropriately increase the learning time in a single project learning to achieve the degree of over-learning, which is conducive to improving the learning effect. This gain effect can be superimposed with the interval effect.

Experiment 2 used the old and new effect paradigms to first explore the relationship between absolute interval length and learning effect, and the relationship between stimulus presentation sequence and learning effect through behavioral experiments; then use ERPs technology to analyze the brain activity principle of different learning effects caused by different interval lengths. The behavioral research results of Experiment 2 show that the interval effect has a nonlinear relationship between the learning effect and the absolute interval number. If the number of intervals is shorter than the working memory capacity or much longer than the working memory capacity, the learning effect of the two will be better than the concentrated learning and the interval close to the working memory capacity. The presentation order of words before pictures is better than pictures before words. The micro-sequence of word learning should be followed by activating the new form first, and then targeting the old representation to prevent the node interaction between representations from depleting cognitive resources. The EEG research results of Experiment 2 show that the longer the learning interval, the faster the time course of relation extraction and the higher the accuracy rate. However, considering the application scenarios of actual learning, short interval learning performs better in the context of pursuing efficient learning. Short interval learning is faster in relation extraction than concentrated learning. Compared with long interval learning, it has lower requirements on the total length of the list of items, and can realize overlearning by using the contents of working memory, which to some extent makes up for the disadvantage of long interval learning in the time process.

From the aspect of product design, the next-generation English learning APP can make the following optimizations in the presentation order of word learning: 1. In the learning and review of a single project, set the minimum rated learning time to achieve over-learning of a single project; 2. The same project should be set up for multiple learning and review, and in the review of wrong words, should follow the interval of 2-7 different items between two words, avoid the interval near the working memory capacity; 3. Try to separate new words from word pictures + Chinese The presentation time of translation is to present new words first, then pictures + translations, and only display Chinese translations without pictures in the test to achieve the best learning effect.