馬焱俊，上海新東方GRE、TOEIC閱讀主講教師。憑借對GRE考試全面深入的研究和考場中的巨大成功，有幸成為上海新東方大家庭的一員，也成為國外考試部中年輕的老師。希望能用自己的活力和熱情感染到每一位迷茫又堅(jiān)強(qiáng)地走在“光榮的荊棘路”中的GRE考生，用詳實(shí)幽默的課程詮釋和豐富新東方的精神，在新東方的恢宏舞臺上聚焦出所有生命的光和熱。My motto（我的座右銘）：“不要為模糊不清的未來擔(dān)擾，只要清清楚楚地為現(xiàn)實(shí)努力即可”。
一直以來，新東方老師在深入研究ETS的各種考試過程中，形成了很多獨(dú)到深刻的認(rèn)識和系統(tǒng)的解題技巧。從某種意義說，新東方老師和ETS中的高層出題者變成了counterpart的辨證統(tǒng)一關(guān)系。一方面ETS出題者極盡所能把考試變得復(fù)雜煩難，另一方面，新東方力求“撥云見日”，揭示了種種試題背后隱藏的秘密。
近日，筆者從網(wǎng)上搜索發(fā)現(xiàn)了GRE Research的一份研究論文，其中主要內(nèi)容涉及了對一個(gè)名叫Source Finder的軟件實(shí)際使用效果的評估。論文發(fā)表于2002年，核心軟件Source Finder的開發(fā)是為了幫助ETS在Internet上自動(dòng)檢索數(shù)字論文庫EBSCO中的文獻(xiàn)，從中提煉出符合ETS各種考試風(fēng)格要求（比如GRE、GMAT和TOEFL等）的樣本文章。根據(jù)開發(fā)的初衷，軟件主要希望1）減少ETS尋找合適文章的時(shí)間；2）擴(kuò)大現(xiàn)有合適備用文章儲(chǔ)備；3）正規(guī)化從筆考到機(jī)考的轉(zhuǎn)變過程；4）提供管理層對item writer（負(fù)責(zé)審查文章是否符合GRE風(fēng)格的工作人員）評估文章重新考察的機(jī)會(huì)；5）為新item writer提供合適以及不合適文章的實(shí)例。
該項(xiàng)目從1999年開始投入，由哥倫比亞大學(xué)教授Rebecca Passonneau負(fù)責(zé)，該教授曾著力研究時(shí)態(tài)和語法規(guī)范的語義學(xué)計(jì)算模型。項(xiàng)目資金由GRE委員會(huì)和ETS方面提供，后也由他們負(fù)責(zé)驗(yàn)收?？梢娫撥浖_發(fā)的相對重要性，以及在ETS整體考試發(fā)展規(guī)劃中的地位。
報(bào)告中透露出很多關(guān)于GRE閱讀考試的秘密和背景，筆者認(rèn)為這些內(nèi)容不僅對教學(xué)研究的深入，以及對考試的多角度認(rèn)識，都是有一定幫助作用的。除去了關(guān)于該軟件評測機(jī)理，實(shí)際使用效果評估等無關(guān)內(nèi)容后，對論文重要內(nèi)容做一定介紹和分析。
一、 套路結(jié)構(gòu)分析的合理性
報(bào)告中摘錄了GRE內(nèi)部培訓(xùn)手則中的一段內(nèi)容，可謂對GRE文章風(fēng)格和特點(diǎn)官方權(quán)威的詮釋了。原文為：
“Reading comprehension passages generally rely on some kind of tension to give them difficulty. A discussion or argument that develops in a linear fashion toward a conclusion is usually too simple to support challenging items. The requisite tension can take various forms: a conflict between different or opposing arguments about something, an unexpected finding that challenges previously established ways of understanding something, a disagreement about what evidence is relevant to the solution of a particular problem, etc. But disagreement or contrast or opposition is not in itself enough to provide the tension: a passage that merely states or describes opposing positions, without discussing in much detail the reasoning behind them, will not support very many or very complex items. For example, instead of saying that Brown believed X and Smith believed Y, a passage might tell us that Brown, heavily influenced by such-and-such school of thought, tended to assume A, and therefore believed X. Smith, on the other hand, had access to information that Brown did not have, namely B and C, and this information led Smith to conclude Y. However, Smith interpreted B in such a way that it seemed logical to conclude Y, when in fact this interpretation was ill founded. The more interdependent details, the more logical twists and turns, the better. Naturally, a passage containing this kind of density can only be created from a source that provides the requisite information, and such sources are not easy to find. ”
該段內(nèi)容可以給我們?nèi)缦聨c(diǎn)思考和啟示：
1)選段核心概念是tension一詞。即使復(fù)習(xí)到了中后期的GRE考生，很少有人能夠做到substantial understanding，完全理解通原文并借此來回答好之后的題目。這個(gè)一方面是考試對閱讀水平要求太高，另一方面，ETS處心積慮地設(shè)置種種障礙破壞了理解全文的可能性。所以考生讀了文章，產(chǎn)生“黑云壓頭頭欲催的”壓抑感是完全正常的。
2)這種讓考生或讀者預(yù)期獲得的感覺是通過什么方式得來的？文章必須有針鋒相對觀點(diǎn)產(chǎn)生的矛盾，有新發(fā)現(xiàn)對已有觀點(diǎn)和認(rèn)識的挑戰(zhàn)，有對一個(gè)問題解決方案的分歧等。不僅如此，必須還要有充分的細(xì)節(jié)和邏輯論證推理支持這些分歧和對比。細(xì)節(jié)越獨(dú)立，邏輯越糾纏，文章帶給讀者這種tension的感覺就越強(qiáng)。引文中提到的簡單例子幾乎是很多GRE閱讀文章發(fā)展的基本結(jié)構(gòu)，可謂經(jīng)典的范例之一。作為強(qiáng)對比的，簡單線性發(fā)展的文章不能成為GRE閱讀文章。高級口譯中的閱讀文章長度很長，時(shí)間壓力也很大，但是和GRE文章的風(fēng)格是完全不同的。一般推薦是快速掃每段的幾句話來力爭把握全文，這樣的處理手法是基于文章套路線性話而產(chǎn)生的。
3)這段文字的思想幾乎在所有GRE閱讀文章中都得到了驗(yàn)證。所有GRE文章都是議論文，哪怕對于部分說明性，敘述性特點(diǎn)比較重的文章，同樣也能成功帶給讀者壓力之感。例子幾乎到處都是。比如大家談之色變的伊斯蘭法的“”文章，全文就是介紹伊斯蘭法，*法和羅馬天主法三種宗教法不同。但是單純談不同不是一篇合格的GRE閱讀，全文用了很多比重談I,J,R三法兩者間相似性，何種程度相同，差別細(xì)微在哪里，誰比誰更統(tǒng)一。通過交織在一起的“求同存異，大同小異”，全文難度徒然飚升，幾乎現(xiàn)場難以理解。其他例如龍蝦左螯右螯，比目魚左眼右眼，蝸牛殼左旋右旋，厄爾尼諾k波r波，蝙蝠FM,CM波……幾乎所有文章都必然出現(xiàn)了相對的“矛盾沖突”。可見這一特點(diǎn)是根本的。再進(jìn)一步看，這種特點(diǎn)可以歸納為GRE文章中都有主線和副線，明線和暗線。分辨出這種聯(lián)系區(qū)別應(yīng)當(dāng)成為閱讀文章后基本理解佐證的標(biāo)志之一。
4)東方閱讀課堂上總結(jié)的套路結(jié)構(gòu)分成：新老觀點(diǎn)對比型，問題回答型，現(xiàn)象解釋型，結(jié)論說明型，對照分析起來，是對ETS手則內(nèi)容準(zhǔn)確的，幾乎完全一致的總結(jié)?？梢?，對套路感和文章結(jié)構(gòu)的強(qiáng)調(diào)是必要的，而且它是作為ETS判斷一篇文章是否符合GRE閱讀文章這個(gè)身份的重要評價(jià)標(biāo)準(zhǔn)。根本上，Source Finder軟件的職責(zé)就是用算法找到這些符合要求的文章。從套路結(jié)構(gòu)入手分析文章，以不變應(yīng)萬變，是為教學(xué)，考試中基本的出發(fā)點(diǎn)。
二、 字?jǐn)?shù)枷鎖和文章題材
報(bào)告中同時(shí)驗(yàn)證了一些教學(xué)理論中的重要背景：
“The text features implemented in the Source Finder prototype were intended to allow users to screen source text for length, content, clarity, density, and level of argumentation. Length filtering is extremely important, since copyright constraints dictate that the number of words in a reading comprehension passage must be less than 10% of the total number of words in the source document from which the passage was extracted. For example, the source document for a 450-word GRE passage must include at least 4,500 words.”
一方面，Source Finder 軟件判斷基于文章長度，內(nèi)容，清晰性，密度，論證程度等，更重要的是，其中揭示文章長度永遠(yuǎn)是架在ETS身上的枷鎖，出于版權(quán)限制，ETS用于實(shí)際考試的文章長度多只能引用10％的原文長度。這種“帶著鐐銬跳舞”的結(jié)果是大量文章的改寫，重寫和刪節(jié)。
難句子理論的提出正是針對這種背景的深刻理論分析。在傳統(tǒng)的四六級閱讀中難句的感受可能還不明顯和突出，但是在GRE閱讀中，這些恐怖的難句不僅是難點(diǎn)更是很多題目產(chǎn)生的溫床。出于字?jǐn)?shù)限制，ETS在GRE考試中自愿或是被迫使用了大量改寫，把單位句子的長度和難度大大提高。有的GRE難句長度甚至直接堪比一段；很多句子中大量插入修飾補(bǔ)充說明內(nèi)容，把整個(gè)句子主干分割得支離破碎；有的句子修改后各部分比例嚴(yán)重失衡，所有引入了很多特殊的倒裝結(jié)構(gòu)……后文中還有其他關(guān)于難句之難的ETS版例證。
另外關(guān)于文章題材分類，報(bào)告中也明確寫出了ETS的五種分類 “Source Finder’s content filter assigns each retrieved document to one of five different content categories: humanities, social sciences, physical sciences, biological sciences or .none of the above.”與文學(xué)評論，社會(huì)現(xiàn)象，自然原理，生命科學(xué)的課堂分類是幾乎完全吻合的。
在軟件實(shí)際測試中，文學(xué)評論類文章只有24％的比例被Source Finder判為可能合格，相比另外三類題材的文章，分別為45%、47%和43％。可見單純從題材出發(fā)，文學(xué)評論類是“艱辛”成為GRE閱讀文章的題材，即使錄用的材料，也必須經(jīng)過嚴(yán)格修改和加工。而且，根據(jù)ETS有經(jīng)驗(yàn)的考試開發(fā)者觀點(diǎn)，區(qū)分一篇文學(xué)評論是否適合作為GRE文章的尺度也是難以把握的。這兩點(diǎn)理由可能部分解釋了近年筆考中文學(xué)評論題材文章數(shù)量逐漸減少的趨勢。
三、 “名落孫山”文章的“六宗罪”
從軟件現(xiàn)有缺陷的環(huán)節(jié)中，我們可以粗率讀出一些GRE閱讀文章必然不能有的特點(diǎn)。
Category Description
Problematic reasoning The source contains argumentative reasoning, but it might be difficult to translate into a passage as a result of one or more of the following problems: too many interconnected argument threads to extract a single clear argument; argument is too dispersed throughout the text to be worth the effort; argument lacks clarity, density, or substance.
Wrong genre The source does not provide a scholarly analysis. Includes documents that are primarily descriptive or expository, as well as first-person narratives, memoirs, interviews, biographies, and guidebooks.
Inaccessible The source is too specialized, too full of jargon, too technical, too abstract, or too dependent on a close reading of another work, such as a novel, poem, or film. Would not be accessible to the entire GRE population.
Inappropriate tone or style Tone is too casual or too flip, or author is too strongly opinionated.
Topicality or datedness issues Contains material that would quickly become dated.
Sensitivity issues Contains sensitive or inflammatory content.
On target The source provides the desired mix of analysis/explanation at the desired tone with the desired level of accessibility and with the desired frequency (i.e., length).
基本缺陷為“Problematic reasoning, Wrong genre, Inaccessible, Inappropriate tone or style, Topicality or datedness issue, Sensitivity issues”六個(gè)方面。論證錯(cuò)誤包括邏輯主線不清楚，論據(jù)太分散，論證風(fēng)格不符等；錯(cuò)誤種類比如：第一人稱敘述，回憶錄，訪談，傳記和指導(dǎo)書籍等問題；所謂inaccessible是指文章充滿過于專業(yè)的內(nèi)容，過多術(shù)語和技術(shù)名詞，太抽象或者需要課外的背景，對不具有該背景同學(xué)有理解上嚴(yán)重障礙等?？忌恢庇X得GRE文章風(fēng)格很抽象枯燥，其實(shí)ETS在這方面還是有嚴(yán)格限制和標(biāo)準(zhǔn)的，尤其是在公平化原則的貫徹上，真正盡量做到“General面前，人人平等”；關(guān)于topical和dated的提出，可以對考生在課外補(bǔ)充閱讀材料時(shí)有一點(diǎn)指引作用。大量閱讀流行的報(bào)紙和普通雜志難以符合GRE閱讀的口味，和當(dāng)下有關(guān)的內(nèi)容不可能出現(xiàn)在考場文章中，關(guān)于補(bǔ)充閱讀材料的推薦，后文中將給出參考；敏感的問題例如煽動(dòng)性文字，素食主義者對動(dòng)物目前權(quán)利的置疑，涉及Playboy攝影人物或是特殊教派問題等。
總的來看，GRE選文學(xué)術(shù)化，正規(guī)性，議論性的特點(diǎn)顯露無疑。這些特點(diǎn)歸納一方面可以為我們更好理解GRE文章風(fēng)格提供幫助，還可以給同學(xué)課外閱讀材料相關(guān)合適性提供標(biāo)尺，應(yīng)該選取和NCE4難度形式相仿的文章。
四、 學(xué)術(shù)化單詞關(guān)鍵字列表
所有GRE閱讀文章文體既然都是議論文，那么一篇議論文在Source Finder檢索時(shí)滿足什么量化參數(shù)和指標(biāo)？論文中明確給出了一份詳細(xì)的關(guān)鍵詞表。這樣一份ETS和計(jì)算機(jī)軟件專家測試后得出的關(guān)鍵詞表，其實(shí)是閱讀老師分析后引入的抽象詞理論好的參考材料之一。以往對于抽象詞短期突破，主要是考生自己總結(jié)積累為主，這樣權(quán)威的統(tǒng)計(jì)可以更好的幫助考生對議論文學(xué)術(shù)化詞匯給予基本的把握。部分詞匯，也應(yīng)該成為課堂上詳細(xì)解釋的重點(diǎn)，因?yàn)檫@部分基礎(chǔ)詞匯重現(xiàn)的概率是高的。舉例說對于paradox, plausible, controversial這些重要詞匯的誤解已經(jīng)成為考試中很多題目做錯(cuò)的元兄。對于approach, hypothesis, theory, fact, evidence, confirm, refute, evaluate等詞匯概念上準(zhǔn)確得把握成為了做好Organization題,寫作手法主題題和一些其他相關(guān)題型的基本保障。所以考生值得花時(shí)間把這份詞表詳細(xì)研究一遍，對一些“思通見慣”的詞匯仔細(xì)思考深層含義。
刪去部分很常見的詞匯后，以下給出簡化版的關(guān)鍵詞表。其中部分關(guān)鍵詞也可以成為大家準(zhǔn)備GRE Analytical Writing時(shí)參考的語言。
Source Finder Key Word List
according to, accurate, adapt, adaptability, adaptable, adaptation, adequate, admittedly, aggregate, alternative, analogous, analogously, analysis, analyze, approach, approximate, arbitrarily, arbitrary, arguably, argue, argument, assume, assumption.,, balanced, basis, behavior, belief, cascading, categorize, cause, characteristic, commensurately, compare, comparison, compatible, competing, complex, complexity, concede, conceive, conceivably, conclude, conclusion, condition, conditional, conditionally, consequence, consequently, consider, consideration, consistent, consistently, contrary, in contrast, controversy, controversial, converse, conversely, convincing, convincingly, corollary, corrective, correspondingly, counteract, counterproductive, critical, criticism, criticize, cumulative, decidedly, deduce, deduction, define, definition, demonstrably, demonstrate, design, despite, determination, determine, detrimental, discernible, discover, discuss, discussion, disputable, dispute, distinct, distinguish, domain, due to, eliminate, entail, entity, environment, equally, equivalence, equivalent, establish, estimate, evaluate, evaluation, event, evidence, examination, examine, except, exception, exhibit, exhibition, exist, existential, experience, experiment, explain, explanation, explicit, exponential, failure, familiar, feasible, footnote, granted, guarantee, hence, hypothesis, hypothesize, hypothetical, idea, identical, illogical, impact, implication, implicit, implies, imply, incompatible., inconsistent, increase, in addition to, in fact, in order to, in particular, in response to, in the guise of, indeed, indicates, indisputable, indisputably, induce, induction, inextricably, infeasible, infer, inform, insight, insightful, instability, instead, intangible, intensify, intensified, interpret, interpretation, intractable, investigate, investigation., justification, justify, knowledge, likelihood, likely, limitation, limited, logical, magnitude, matched, maximize, maximum, measure, measurement, method, minimize, minimum, mitigate, model, moreover, multiple, nearly, necessary, negative, negatively, negligible, nevertheless, numerical, notable, notwithstanding, novel, observation, observe, offset, on the other hand, optimum, otherwise, ostensibly, outweigh, overestimate, paradigm, particular, particularly, parameter, pattern, penultimate, percent, percentage, permanently, pertinent, phase, phenomenon, positive, positively, possibility, possible, possibly, potentially, predict, previously, primarily, primary, probability, probable, probably, problem, problematic, process, propensity, proves, rationale, reason, reasoning, recently, redesign, reduce, redundant, regardless of, relation, relative, relatively, reliability, require, requirement, requires, research, respective, respectively, result, reveal, rigorous, rigorously, riskier, salient, science, severe, severity, significantly, similarly, simultaneous, since, specific, specifically, specified, specify, speculate, speculative, stable, stability, still, stipulate, strengthen, strictly, structure, structured, study, subsequent, subsequently, substantially, success, successive, sufficient, sufficiently, suggest, summarize, suppose, supposition, susceptibility, susceptible, systemic, tangible, technique, temporarily, theoretical, theories, theory, therefore, thus, tractable, tradeoff, traditional, truth, ultimately, unconditional, undeniably, underestimate, understand, understanding, undoubtedly, unfamiliar, uniformly, unique, unless, unmatched, unquestionably, unstable, unusual, valid, validate, validity, variance, variation, weaken, weigh, yield
五、難句子的“罪狀”
GRE中的難句子不是閱讀老師渲染出來的，是考生實(shí)踐檢驗(yàn)體會(huì)出來的。這些煩難的長句，復(fù)合句，復(fù)雜修飾句又很容易成為考試題目集中的重點(diǎn)。對于難句概念，論文中給予了統(tǒng)計(jì)學(xué)上的支持佐證。
“As the data indicate, there is a wide disparity between the feature values calculated in the 33 original sources and in the passages that were eventually developed from those sources. For example, the average sentence length of the five original sources is 23.2 words, compared with 35.0 words for the corresponding GRE passages. Similarly, the average type-token ratio for the five sources is .26, compared with .56 for the corresponding GRE passages.”
“As noted earlier, the SourceFinder feature labeled UniqWordPerWord is a measure of the size of the vocabulary used in a given document. This feature is also called type-token ratio because it is a ratio of the total number of word types (or unique words) in a document to the total number of word tokens in the document (Youmans, 31 1991). The type-token ratio calculated for a particular document can range from 1, in the case of all unique words, to 1/N where N is the total number of words in the document. Certain word types, (e.g., words known to the information retrieval community as stop words, including .the. and .of.) occur relatively more frequently than so-called content words.”
其中句子長度差異比較容易理解，而第二個(gè)參數(shù)type-token ratio是指全文中獨(dú)特單詞占總文字字?jǐn)?shù)的比例。根據(jù)語言學(xué)專家研究結(jié)論，句子以及文章中獨(dú)特單詞越是多，專業(yè)術(shù)語和復(fù)雜內(nèi)容一般就更多，后導(dǎo)致全文理解困難度就提高。可見GRE文章經(jīng)過原材料的加工和改寫后基本特點(diǎn)就是：句子更長，更難理解。由此可見，難句能力往往是征服GRE閱讀的基本功之一，而強(qiáng)調(diào)提煉句子主體詞的工作則有效地針對了type-token ratio參數(shù)提示的必要能力，考生必須能具有再充滿了學(xué)術(shù)專業(yè)名詞的長句子中提出一個(gè)理論的關(guān)鍵詞，一派觀點(diǎn)的區(qū)別性特點(diǎn)等。
六、課外閱讀材料推薦
勤奮的考生一直關(guān)心課下閱讀什么文字材料來提高GRE考試的能力。簡單的讀普通報(bào)紙雜志比如Shanghai Star, China Daily，收獲是比較有限的，根本原因是你從這些文字中讀不出關(guān)鍵的tension感，相反只是一種放松的閱讀。除了課堂上推薦的GMAT，LSAT等同類型風(fēng)格的全真試題之外，參考論文中給出的部分研究時(shí)采用的原材料來源，推薦一下材料。這部分期刊對GRE閱讀理解的提高是比較“專業(yè)對口”的，適合作為一般的泛讀材料。
African American Review —— social sciences (some humanities)
American Journal of Economics and Sociology——social sciences
American Scholar —— humanities
Art Journal —— humanities
Astronomy —— physical sciences
BioScience —— biological sciences
Black American Literature Forum —— humanities
business Horizons —— business
Commentary —— social sciences
Discover —— biological sciences and physical sciences
Feminist Studies —— social sciences (some humanities)
Journal of American Ethnic History —— social sciences
Journal of Black Studies —— social sciences (some humanities)
Journal of Social History —— social sciences
The Nation—— social sciences
Science News —— biological sciences and physical sciences
Scientific American—— biological sciences and physical sciences
Sloan Management Review —— business
Smithsonian —— social sciences, biological sciences, and physical sciences
Technology Review —— biological sciences and physical sciences
Wilson Quarterly —— social sciences
Women.s Review of Books —— humanities
其中很多期刊可以在學(xué)校圖書館中找到，考生可以課外泛讀，作為長期積累，提高閱讀根本能力的途徑之一。
七、結(jié)論
縱觀整個(gè)論文，不僅有大量內(nèi)容對很多新東方現(xiàn)有閱讀中研究的理論做了充分的佐證，也對以后教學(xué)工作中的探索方向做出了一定的指引作用。主要是集中在文章套路和風(fēng)格，議論文特色關(guān)鍵詞把握，閱讀基本能力和課外參考資料等多方面。
博弈的基本概念就是根據(jù)對手的行為做出合適的反映。其實(shí)考試和培訓(xùn)某種程度上就是一種博弈。所以能有資料幫助我們站在ETS的立場，了解一些“對手”的詳細(xì)信息和約束，無疑是對考試進(jìn)一步深入研究是大有裨益的。相信新東方的老師一定可以揮灑智慧和汗水，將探索進(jìn)行到底！
一直以來，新東方老師在深入研究ETS的各種考試過程中，形成了很多獨(dú)到深刻的認(rèn)識和系統(tǒng)的解題技巧。從某種意義說，新東方老師和ETS中的高層出題者變成了counterpart的辨證統(tǒng)一關(guān)系。一方面ETS出題者極盡所能把考試變得復(fù)雜煩難，另一方面，新東方力求“撥云見日”，揭示了種種試題背后隱藏的秘密。
近日，筆者從網(wǎng)上搜索發(fā)現(xiàn)了GRE Research的一份研究論文，其中主要內(nèi)容涉及了對一個(gè)名叫Source Finder的軟件實(shí)際使用效果的評估。論文發(fā)表于2002年，核心軟件Source Finder的開發(fā)是為了幫助ETS在Internet上自動(dòng)檢索數(shù)字論文庫EBSCO中的文獻(xiàn)，從中提煉出符合ETS各種考試風(fēng)格要求（比如GRE、GMAT和TOEFL等）的樣本文章。根據(jù)開發(fā)的初衷，軟件主要希望1）減少ETS尋找合適文章的時(shí)間；2）擴(kuò)大現(xiàn)有合適備用文章儲(chǔ)備；3）正規(guī)化從筆考到機(jī)考的轉(zhuǎn)變過程；4）提供管理層對item writer（負(fù)責(zé)審查文章是否符合GRE風(fēng)格的工作人員）評估文章重新考察的機(jī)會(huì)；5）為新item writer提供合適以及不合適文章的實(shí)例。
該項(xiàng)目從1999年開始投入，由哥倫比亞大學(xué)教授Rebecca Passonneau負(fù)責(zé)，該教授曾著力研究時(shí)態(tài)和語法規(guī)范的語義學(xué)計(jì)算模型。項(xiàng)目資金由GRE委員會(huì)和ETS方面提供，后也由他們負(fù)責(zé)驗(yàn)收?？梢娫撥浖_發(fā)的相對重要性，以及在ETS整體考試發(fā)展規(guī)劃中的地位。
報(bào)告中透露出很多關(guān)于GRE閱讀考試的秘密和背景，筆者認(rèn)為這些內(nèi)容不僅對教學(xué)研究的深入，以及對考試的多角度認(rèn)識，都是有一定幫助作用的。除去了關(guān)于該軟件評測機(jī)理，實(shí)際使用效果評估等無關(guān)內(nèi)容后，對論文重要內(nèi)容做一定介紹和分析。
一、 套路結(jié)構(gòu)分析的合理性
報(bào)告中摘錄了GRE內(nèi)部培訓(xùn)手則中的一段內(nèi)容，可謂對GRE文章風(fēng)格和特點(diǎn)官方權(quán)威的詮釋了。原文為：
“Reading comprehension passages generally rely on some kind of tension to give them difficulty. A discussion or argument that develops in a linear fashion toward a conclusion is usually too simple to support challenging items. The requisite tension can take various forms: a conflict between different or opposing arguments about something, an unexpected finding that challenges previously established ways of understanding something, a disagreement about what evidence is relevant to the solution of a particular problem, etc. But disagreement or contrast or opposition is not in itself enough to provide the tension: a passage that merely states or describes opposing positions, without discussing in much detail the reasoning behind them, will not support very many or very complex items. For example, instead of saying that Brown believed X and Smith believed Y, a passage might tell us that Brown, heavily influenced by such-and-such school of thought, tended to assume A, and therefore believed X. Smith, on the other hand, had access to information that Brown did not have, namely B and C, and this information led Smith to conclude Y. However, Smith interpreted B in such a way that it seemed logical to conclude Y, when in fact this interpretation was ill founded. The more interdependent details, the more logical twists and turns, the better. Naturally, a passage containing this kind of density can only be created from a source that provides the requisite information, and such sources are not easy to find. ”
該段內(nèi)容可以給我們?nèi)缦聨c(diǎn)思考和啟示：
1)選段核心概念是tension一詞。即使復(fù)習(xí)到了中后期的GRE考生，很少有人能夠做到substantial understanding，完全理解通原文并借此來回答好之后的題目。這個(gè)一方面是考試對閱讀水平要求太高，另一方面，ETS處心積慮地設(shè)置種種障礙破壞了理解全文的可能性。所以考生讀了文章，產(chǎn)生“黑云壓頭頭欲催的”壓抑感是完全正常的。
2)這種讓考生或讀者預(yù)期獲得的感覺是通過什么方式得來的？文章必須有針鋒相對觀點(diǎn)產(chǎn)生的矛盾，有新發(fā)現(xiàn)對已有觀點(diǎn)和認(rèn)識的挑戰(zhàn)，有對一個(gè)問題解決方案的分歧等。不僅如此，必須還要有充分的細(xì)節(jié)和邏輯論證推理支持這些分歧和對比。細(xì)節(jié)越獨(dú)立，邏輯越糾纏，文章帶給讀者這種tension的感覺就越強(qiáng)。引文中提到的簡單例子幾乎是很多GRE閱讀文章發(fā)展的基本結(jié)構(gòu)，可謂經(jīng)典的范例之一。作為強(qiáng)對比的，簡單線性發(fā)展的文章不能成為GRE閱讀文章。高級口譯中的閱讀文章長度很長，時(shí)間壓力也很大，但是和GRE文章的風(fēng)格是完全不同的。一般推薦是快速掃每段的幾句話來力爭把握全文，這樣的處理手法是基于文章套路線性話而產(chǎn)生的。
3)這段文字的思想幾乎在所有GRE閱讀文章中都得到了驗(yàn)證。所有GRE文章都是議論文，哪怕對于部分說明性，敘述性特點(diǎn)比較重的文章，同樣也能成功帶給讀者壓力之感。例子幾乎到處都是。比如大家談之色變的伊斯蘭法的“”文章，全文就是介紹伊斯蘭法，*法和羅馬天主法三種宗教法不同。但是單純談不同不是一篇合格的GRE閱讀，全文用了很多比重談I,J,R三法兩者間相似性，何種程度相同，差別細(xì)微在哪里，誰比誰更統(tǒng)一。通過交織在一起的“求同存異，大同小異”，全文難度徒然飚升，幾乎現(xiàn)場難以理解。其他例如龍蝦左螯右螯，比目魚左眼右眼，蝸牛殼左旋右旋，厄爾尼諾k波r波，蝙蝠FM,CM波……幾乎所有文章都必然出現(xiàn)了相對的“矛盾沖突”。可見這一特點(diǎn)是根本的。再進(jìn)一步看，這種特點(diǎn)可以歸納為GRE文章中都有主線和副線，明線和暗線。分辨出這種聯(lián)系區(qū)別應(yīng)當(dāng)成為閱讀文章后基本理解佐證的標(biāo)志之一。
4)東方閱讀課堂上總結(jié)的套路結(jié)構(gòu)分成：新老觀點(diǎn)對比型，問題回答型，現(xiàn)象解釋型，結(jié)論說明型，對照分析起來，是對ETS手則內(nèi)容準(zhǔn)確的，幾乎完全一致的總結(jié)?？梢?，對套路感和文章結(jié)構(gòu)的強(qiáng)調(diào)是必要的，而且它是作為ETS判斷一篇文章是否符合GRE閱讀文章這個(gè)身份的重要評價(jià)標(biāo)準(zhǔn)。根本上，Source Finder軟件的職責(zé)就是用算法找到這些符合要求的文章。從套路結(jié)構(gòu)入手分析文章，以不變應(yīng)萬變，是為教學(xué)，考試中基本的出發(fā)點(diǎn)。
二、 字?jǐn)?shù)枷鎖和文章題材
報(bào)告中同時(shí)驗(yàn)證了一些教學(xué)理論中的重要背景：
“The text features implemented in the Source Finder prototype were intended to allow users to screen source text for length, content, clarity, density, and level of argumentation. Length filtering is extremely important, since copyright constraints dictate that the number of words in a reading comprehension passage must be less than 10% of the total number of words in the source document from which the passage was extracted. For example, the source document for a 450-word GRE passage must include at least 4,500 words.”
一方面，Source Finder 軟件判斷基于文章長度，內(nèi)容，清晰性，密度，論證程度等，更重要的是，其中揭示文章長度永遠(yuǎn)是架在ETS身上的枷鎖，出于版權(quán)限制，ETS用于實(shí)際考試的文章長度多只能引用10％的原文長度。這種“帶著鐐銬跳舞”的結(jié)果是大量文章的改寫，重寫和刪節(jié)。
難句子理論的提出正是針對這種背景的深刻理論分析。在傳統(tǒng)的四六級閱讀中難句的感受可能還不明顯和突出，但是在GRE閱讀中，這些恐怖的難句不僅是難點(diǎn)更是很多題目產(chǎn)生的溫床。出于字?jǐn)?shù)限制，ETS在GRE考試中自愿或是被迫使用了大量改寫，把單位句子的長度和難度大大提高。有的GRE難句長度甚至直接堪比一段；很多句子中大量插入修飾補(bǔ)充說明內(nèi)容，把整個(gè)句子主干分割得支離破碎；有的句子修改后各部分比例嚴(yán)重失衡，所有引入了很多特殊的倒裝結(jié)構(gòu)……后文中還有其他關(guān)于難句之難的ETS版例證。
另外關(guān)于文章題材分類，報(bào)告中也明確寫出了ETS的五種分類 “Source Finder’s content filter assigns each retrieved document to one of five different content categories: humanities, social sciences, physical sciences, biological sciences or .none of the above.”與文學(xué)評論，社會(huì)現(xiàn)象，自然原理，生命科學(xué)的課堂分類是幾乎完全吻合的。
在軟件實(shí)際測試中，文學(xué)評論類文章只有24％的比例被Source Finder判為可能合格，相比另外三類題材的文章，分別為45%、47%和43％。可見單純從題材出發(fā)，文學(xué)評論類是“艱辛”成為GRE閱讀文章的題材，即使錄用的材料，也必須經(jīng)過嚴(yán)格修改和加工。而且，根據(jù)ETS有經(jīng)驗(yàn)的考試開發(fā)者觀點(diǎn)，區(qū)分一篇文學(xué)評論是否適合作為GRE文章的尺度也是難以把握的。這兩點(diǎn)理由可能部分解釋了近年筆考中文學(xué)評論題材文章數(shù)量逐漸減少的趨勢。
三、 “名落孫山”文章的“六宗罪”
從軟件現(xiàn)有缺陷的環(huán)節(jié)中，我們可以粗率讀出一些GRE閱讀文章必然不能有的特點(diǎn)。
Category Description
Problematic reasoning The source contains argumentative reasoning, but it might be difficult to translate into a passage as a result of one or more of the following problems: too many interconnected argument threads to extract a single clear argument; argument is too dispersed throughout the text to be worth the effort; argument lacks clarity, density, or substance.
Wrong genre The source does not provide a scholarly analysis. Includes documents that are primarily descriptive or expository, as well as first-person narratives, memoirs, interviews, biographies, and guidebooks.
Inaccessible The source is too specialized, too full of jargon, too technical, too abstract, or too dependent on a close reading of another work, such as a novel, poem, or film. Would not be accessible to the entire GRE population.
Inappropriate tone or style Tone is too casual or too flip, or author is too strongly opinionated.
Topicality or datedness issues Contains material that would quickly become dated.
Sensitivity issues Contains sensitive or inflammatory content.
On target The source provides the desired mix of analysis/explanation at the desired tone with the desired level of accessibility and with the desired frequency (i.e., length).
基本缺陷為“Problematic reasoning, Wrong genre, Inaccessible, Inappropriate tone or style, Topicality or datedness issue, Sensitivity issues”六個(gè)方面。論證錯(cuò)誤包括邏輯主線不清楚，論據(jù)太分散，論證風(fēng)格不符等；錯(cuò)誤種類比如：第一人稱敘述，回憶錄，訪談，傳記和指導(dǎo)書籍等問題；所謂inaccessible是指文章充滿過于專業(yè)的內(nèi)容，過多術(shù)語和技術(shù)名詞，太抽象或者需要課外的背景，對不具有該背景同學(xué)有理解上嚴(yán)重障礙等?？忌恢庇X得GRE文章風(fēng)格很抽象枯燥，其實(shí)ETS在這方面還是有嚴(yán)格限制和標(biāo)準(zhǔn)的，尤其是在公平化原則的貫徹上，真正盡量做到“General面前，人人平等”；關(guān)于topical和dated的提出，可以對考生在課外補(bǔ)充閱讀材料時(shí)有一點(diǎn)指引作用。大量閱讀流行的報(bào)紙和普通雜志難以符合GRE閱讀的口味，和當(dāng)下有關(guān)的內(nèi)容不可能出現(xiàn)在考場文章中，關(guān)于補(bǔ)充閱讀材料的推薦，后文中將給出參考；敏感的問題例如煽動(dòng)性文字，素食主義者對動(dòng)物目前權(quán)利的置疑，涉及Playboy攝影人物或是特殊教派問題等。
總的來看，GRE選文學(xué)術(shù)化，正規(guī)性，議論性的特點(diǎn)顯露無疑。這些特點(diǎn)歸納一方面可以為我們更好理解GRE文章風(fēng)格提供幫助，還可以給同學(xué)課外閱讀材料相關(guān)合適性提供標(biāo)尺，應(yīng)該選取和NCE4難度形式相仿的文章。
四、 學(xué)術(shù)化單詞關(guān)鍵字列表
所有GRE閱讀文章文體既然都是議論文，那么一篇議論文在Source Finder檢索時(shí)滿足什么量化參數(shù)和指標(biāo)？論文中明確給出了一份詳細(xì)的關(guān)鍵詞表。這樣一份ETS和計(jì)算機(jī)軟件專家測試后得出的關(guān)鍵詞表，其實(shí)是閱讀老師分析后引入的抽象詞理論好的參考材料之一。以往對于抽象詞短期突破，主要是考生自己總結(jié)積累為主，這樣權(quán)威的統(tǒng)計(jì)可以更好的幫助考生對議論文學(xué)術(shù)化詞匯給予基本的把握。部分詞匯，也應(yīng)該成為課堂上詳細(xì)解釋的重點(diǎn)，因?yàn)檫@部分基礎(chǔ)詞匯重現(xiàn)的概率是高的。舉例說對于paradox, plausible, controversial這些重要詞匯的誤解已經(jīng)成為考試中很多題目做錯(cuò)的元兄。對于approach, hypothesis, theory, fact, evidence, confirm, refute, evaluate等詞匯概念上準(zhǔn)確得把握成為了做好Organization題,寫作手法主題題和一些其他相關(guān)題型的基本保障。所以考生值得花時(shí)間把這份詞表詳細(xì)研究一遍，對一些“思通見慣”的詞匯仔細(xì)思考深層含義。
刪去部分很常見的詞匯后，以下給出簡化版的關(guān)鍵詞表。其中部分關(guān)鍵詞也可以成為大家準(zhǔn)備GRE Analytical Writing時(shí)參考的語言。
Source Finder Key Word List
according to, accurate, adapt, adaptability, adaptable, adaptation, adequate, admittedly, aggregate, alternative, analogous, analogously, analysis, analyze, approach, approximate, arbitrarily, arbitrary, arguably, argue, argument, assume, assumption.,, balanced, basis, behavior, belief, cascading, categorize, cause, characteristic, commensurately, compare, comparison, compatible, competing, complex, complexity, concede, conceive, conceivably, conclude, conclusion, condition, conditional, conditionally, consequence, consequently, consider, consideration, consistent, consistently, contrary, in contrast, controversy, controversial, converse, conversely, convincing, convincingly, corollary, corrective, correspondingly, counteract, counterproductive, critical, criticism, criticize, cumulative, decidedly, deduce, deduction, define, definition, demonstrably, demonstrate, design, despite, determination, determine, detrimental, discernible, discover, discuss, discussion, disputable, dispute, distinct, distinguish, domain, due to, eliminate, entail, entity, environment, equally, equivalence, equivalent, establish, estimate, evaluate, evaluation, event, evidence, examination, examine, except, exception, exhibit, exhibition, exist, existential, experience, experiment, explain, explanation, explicit, exponential, failure, familiar, feasible, footnote, granted, guarantee, hence, hypothesis, hypothesize, hypothetical, idea, identical, illogical, impact, implication, implicit, implies, imply, incompatible., inconsistent, increase, in addition to, in fact, in order to, in particular, in response to, in the guise of, indeed, indicates, indisputable, indisputably, induce, induction, inextricably, infeasible, infer, inform, insight, insightful, instability, instead, intangible, intensify, intensified, interpret, interpretation, intractable, investigate, investigation., justification, justify, knowledge, likelihood, likely, limitation, limited, logical, magnitude, matched, maximize, maximum, measure, measurement, method, minimize, minimum, mitigate, model, moreover, multiple, nearly, necessary, negative, negatively, negligible, nevertheless, numerical, notable, notwithstanding, novel, observation, observe, offset, on the other hand, optimum, otherwise, ostensibly, outweigh, overestimate, paradigm, particular, particularly, parameter, pattern, penultimate, percent, percentage, permanently, pertinent, phase, phenomenon, positive, positively, possibility, possible, possibly, potentially, predict, previously, primarily, primary, probability, probable, probably, problem, problematic, process, propensity, proves, rationale, reason, reasoning, recently, redesign, reduce, redundant, regardless of, relation, relative, relatively, reliability, require, requirement, requires, research, respective, respectively, result, reveal, rigorous, rigorously, riskier, salient, science, severe, severity, significantly, similarly, simultaneous, since, specific, specifically, specified, specify, speculate, speculative, stable, stability, still, stipulate, strengthen, strictly, structure, structured, study, subsequent, subsequently, substantially, success, successive, sufficient, sufficiently, suggest, summarize, suppose, supposition, susceptibility, susceptible, systemic, tangible, technique, temporarily, theoretical, theories, theory, therefore, thus, tractable, tradeoff, traditional, truth, ultimately, unconditional, undeniably, underestimate, understand, understanding, undoubtedly, unfamiliar, uniformly, unique, unless, unmatched, unquestionably, unstable, unusual, valid, validate, validity, variance, variation, weaken, weigh, yield
五、難句子的“罪狀”
GRE中的難句子不是閱讀老師渲染出來的，是考生實(shí)踐檢驗(yàn)體會(huì)出來的。這些煩難的長句，復(fù)合句，復(fù)雜修飾句又很容易成為考試題目集中的重點(diǎn)。對于難句概念，論文中給予了統(tǒng)計(jì)學(xué)上的支持佐證。
“As the data indicate, there is a wide disparity between the feature values calculated in the 33 original sources and in the passages that were eventually developed from those sources. For example, the average sentence length of the five original sources is 23.2 words, compared with 35.0 words for the corresponding GRE passages. Similarly, the average type-token ratio for the five sources is .26, compared with .56 for the corresponding GRE passages.”
“As noted earlier, the SourceFinder feature labeled UniqWordPerWord is a measure of the size of the vocabulary used in a given document. This feature is also called type-token ratio because it is a ratio of the total number of word types (or unique words) in a document to the total number of word tokens in the document (Youmans, 31 1991). The type-token ratio calculated for a particular document can range from 1, in the case of all unique words, to 1/N where N is the total number of words in the document. Certain word types, (e.g., words known to the information retrieval community as stop words, including .the. and .of.) occur relatively more frequently than so-called content words.”
其中句子長度差異比較容易理解，而第二個(gè)參數(shù)type-token ratio是指全文中獨(dú)特單詞占總文字字?jǐn)?shù)的比例。根據(jù)語言學(xué)專家研究結(jié)論，句子以及文章中獨(dú)特單詞越是多，專業(yè)術(shù)語和復(fù)雜內(nèi)容一般就更多，后導(dǎo)致全文理解困難度就提高。可見GRE文章經(jīng)過原材料的加工和改寫后基本特點(diǎn)就是：句子更長，更難理解。由此可見，難句能力往往是征服GRE閱讀的基本功之一，而強(qiáng)調(diào)提煉句子主體詞的工作則有效地針對了type-token ratio參數(shù)提示的必要能力，考生必須能具有再充滿了學(xué)術(shù)專業(yè)名詞的長句子中提出一個(gè)理論的關(guān)鍵詞，一派觀點(diǎn)的區(qū)別性特點(diǎn)等。
六、課外閱讀材料推薦
勤奮的考生一直關(guān)心課下閱讀什么文字材料來提高GRE考試的能力。簡單的讀普通報(bào)紙雜志比如Shanghai Star, China Daily，收獲是比較有限的，根本原因是你從這些文字中讀不出關(guān)鍵的tension感，相反只是一種放松的閱讀。除了課堂上推薦的GMAT，LSAT等同類型風(fēng)格的全真試題之外，參考論文中給出的部分研究時(shí)采用的原材料來源，推薦一下材料。這部分期刊對GRE閱讀理解的提高是比較“專業(yè)對口”的，適合作為一般的泛讀材料。
African American Review —— social sciences (some humanities)
American Journal of Economics and Sociology——social sciences
American Scholar —— humanities
Art Journal —— humanities
Astronomy —— physical sciences
BioScience —— biological sciences
Black American Literature Forum —— humanities
business Horizons —— business
Commentary —— social sciences
Discover —— biological sciences and physical sciences
Feminist Studies —— social sciences (some humanities)
Journal of American Ethnic History —— social sciences
Journal of Black Studies —— social sciences (some humanities)
Journal of Social History —— social sciences
The Nation—— social sciences
Science News —— biological sciences and physical sciences
Scientific American—— biological sciences and physical sciences
Sloan Management Review —— business
Smithsonian —— social sciences, biological sciences, and physical sciences
Technology Review —— biological sciences and physical sciences
Wilson Quarterly —— social sciences
Women.s Review of Books —— humanities
其中很多期刊可以在學(xué)校圖書館中找到，考生可以課外泛讀，作為長期積累，提高閱讀根本能力的途徑之一。
七、結(jié)論
縱觀整個(gè)論文，不僅有大量內(nèi)容對很多新東方現(xiàn)有閱讀中研究的理論做了充分的佐證，也對以后教學(xué)工作中的探索方向做出了一定的指引作用。主要是集中在文章套路和風(fēng)格，議論文特色關(guān)鍵詞把握，閱讀基本能力和課外參考資料等多方面。
博弈的基本概念就是根據(jù)對手的行為做出合適的反映。其實(shí)考試和培訓(xùn)某種程度上就是一種博弈。所以能有資料幫助我們站在ETS的立場，了解一些“對手”的詳細(xì)信息和約束，無疑是對考試進(jìn)一步深入研究是大有裨益的。相信新東方的老師一定可以揮灑智慧和汗水，將探索進(jìn)行到底！