测算词汇量大小的那些曲折、调整 - 英语教学法原著选读87
个人小站，生存不易，请关注本订阅号后打开看完本文，点进系统广告（右下角“写留言”上方），帮我维持小站生存。一篇点一次即可，无需关注购买。如能经常帮点广告，请“写留言”说“支持”，让我知道谁在帮我，后续还有其他好处。已留言的朋友不必再说。感谢！
------------------------
本系列上一篇：作为英语学习策略一种的“词典使用”（英语教学法原著选读86，首次附视频讲解，另附选读总目录）
------------------------
前面两期的文章，我分别向朋友们展示了英语教学法原著选读系列配录音和配讲解视频的效果。本期，我继续做实验：1、由此前的原文、译文相对分隔开的形式转换为原文、译文各段对照，请大家表个态哪种形式更好；2、在原文中直接进行句型、疑难词汇的标注与讲解，凡是括号里等号=开头的都是我加入的注释。请朋友们通过“写留言”多提宝贵意见！
------------------------
以下英文原文取自上海外语教育出版社出版、授权转载的“教学法丛书”中的《词汇：描述、习得与教学》分册：
原文
Thesis Title: Vocabulary and Testing
Author: John Read
Measuring Vocabulary Size
测算词汇量大小
Nation and Waring (1.1) have discussed the difficulty of obtaining satisfactory estimates of native speaker vocabulary size, with particular emphasis on the sampling problem. Sampling is less of an issue in making estimates of L2 vocabulary knowledge if a word frequency list is used as the sampling frame. From a testing viewpoint, the question is more one of deciding on the appropriate test format to determine whether each word in the sample is known or not. If a reliable estimate is to be made, the sample of words tested (=which are tested) needs to be quite substantial: Nation (1993) calculates that a sample based on Collins English Dictionary should ideally contain about 600 items. This requirement places a severe constraint =limit on the kinds of test format that can be used. Typically a simple and relatively decontextualized (=going without context) item type has been chosen for vocabulary size tests, so that the test-takers can respond to the required number of words within a reasonable period of time.
纳辛和华灵（本书1.1）已经讨论过母语者词汇量估算的难度，尤其强调了取样问题。对于二语词汇知识的估算来说，取样的问题要小一些，如果使用词频列表作为取样框架的话。从测试的观点来看，问题更在于选定何种恰当的测试形式来判断样本中的词（受试者）会还是不会。如果要进行可靠的估算，则接受测试的单词样本需要相当大的量才行：纳辛（1993）算出，基于《柯林斯英语词典》的样本，理想状况应该有600个条目。这一要求就严重限制了能够使用的测试形式种类。典型情况是，简单的、相对脱离上下文的项目类型被选择用来进行词汇量测试，因此受试者能够在合理的时长内对所要求数量的单词作出反应。
The simplest possible format is the checklist (or yes/no test), which has a lengthy history in L1 research (e.g. Sims, 1929; Tilley, 1936). In its original form, the checklist presents (=shows... to) the test-takers with a set of words and requires them to indicate with a tick (/) whether they know each one. Since the format depends purely on self-report, there is an obvious problem with differing (=different from person to person) interpretations of what 'knowing a word’ means, as well as a lack of any means (=手段) to check whether the learners are overestimating (=think too highly of) their vocabulary knowledge. To address this latter shortcoming, Anderson and Freebody (1983) devised a new version of the checklist which contains a certain proportion of plausible (=likely true) non-words that follow the norms of English word formation. Claiming knowledge of some of the non-words is taken as evidence that test-takers are overstating their vocabulary knowledge, so the scores of such learners are adjusted downwards to give a more valid estimate of their knowledge of the real words.
最简单的可能就是对勾表（即yes/no打勾测试），这种形式在第一语言研究中有着悠久的历史（如西姆斯，1929；提雷，1936）。对勾表的原始形式向受试者出示一组词，要求他们勾出认识的词。这种形式依靠的纯粹是自主报告，所以有一个明显的问题：怎样才算“认识一个词”，不同的人有不同的解读，也没有任何手段能够核对学习者是否高估了自己的词汇知识。为了弥补后一个缺点，安德森和富利伯蒂（1983）设计了一种新版本的对勾表，包含了一些像模像样的生造词，都是按照英语构词法的常规造出来的。如果受试者称认识这样的生造词，就证明他们夸大了自己的词汇知识，这样一来他们的分数就要下调，从而给出对他们真实词汇量的有效估计。
Meara and his colleagues (Meara and Buxton, 1987; Meara and Jones, 1988) developed a computerized (=that which can be run on a computer) checklist test for second language learners of English, one that incorporates (=includes) non-words and samples (=take as indicators) real words from various frequency levels of the Thorndike and Lorge (1944) list. The programme operates on a computer-adaptive principle, presenting words selectively to the test-taker until an adjusted estimate of the individual’s vocabulary size can be made, up to a ceiling level (=topmost) of 10,000 words. The test was published as the Eurocentres’ Vocabulary Size Test (Meara and Jones, 1990). It was seen as a useful tool for language schools, providing an index of the students' overall knowledge of the language to assist in placing them in the appropriate class.
米亚拉和他的同事们（米亚拉和巴克斯顿，1987；米亚拉和琼斯，1988）开发出了一种电脑版本的英语二与学习者对勾表测试，这种测试包括了生造词，并从索恩代克和洛吉（1994）词汇列表的不同词频等级取样。他们的测试程序运行时遵循电脑适应性原则，有选择性地向受试者出示单词，直到可以得出调整后的个人词汇量为止，上限是10000词。该测试作为欧洲中心的词汇量测试（米亚拉和琼斯，1990）出版发行，被看作是语言学校的有用工具，能提供学生对语言总体知识的指数，从而有利于把他们分到合适的班级。
Although the test initially (=at first) seemed very promising, Meara (1996c) notes (=points out) some problems that have emerged from continuing experience with this and other checklist tests. First, they do not work well with low-level learners, who respond unpredictably to the non-words. Secondly, they do not perform satisfactorily as measures of the English language ability of learners whose L1 is French, apparently because of the close relationship between the lexicons of the two languages. The third problem is that certain learners obtain very low scores as a result of their over-willingness to claim knowledge of the non-words. Thus, further work is required to refine (=make better) the test format and gain a fuller picture of its potential as well as its limitations.
尽管该测试最初看起来前景不错，米亚拉（1996c）还是指出了一些问题，这些问题都是在该测试和其他对勾表测试的持续使用中发现的。首先，这些测试对起始阶段学习者作用不佳，他们对那些生造词的反应难以预测。其次，这些测试对母语为法语的学习者的英语语言能力测算不能尽如人意，因为英法两种语言的词汇相似度很高。第三个问题是，一些学习者“认识”生造词的热情过高，以致得分很低。这样一来，就需要做进一步的工作来优化考试形式，从而对其潜力与局限有更透彻的了解。
本订阅号所有原创文章均禁止转载，确需转载的朋友请后台联系支付转载费用后添加白名单。感谢支持！
别忘了帮点广告，就在下面。感谢！