The Language Teacher
05 - 2000

TOEFL Scores in Japan:
Much Ado About Nothing

Sean M. Reedy

Maebashi Institute of Technology

Bashing Japanese students over their lack of foreign language learning ability seems to be in vogue these days. All sorts of folks are jumping on the "Japanese are poor language learners" bandwagon, from foreign language teachers to academicians (Clark, 2000; Tolbert, 2000; Mulvey, 1999). University of Tokyo Professor Inoguchi, for example, claims that learning English is a lost cause for the Japanese commoner: "What matters is that the elites [sic] speak good English (2000 January 29. The Washington Post, A13). Gregory Clark, pointing to the "shocking discovery" that Japan ranks worst than Laos and Cambodia in English language ability argues ". . . even if English was an official language, it is hard to imagine the Japanese nation emerging as a fluent English speaker" (2000, p. 19).

There is nothing new in this. Since at least the late Edo period, Japanese have been taught to see themselves as poor assimilators of foreign languages (Dore, 1984). The latest spate of criticism simply reinforces a stereotype of Japanese people as intrinsically poor at communicating in other tongues. The evidence that is frequently paraded about as proof of the dreadful nature of foreign language learning in Japan, evidence used to excoriate all elements of the foreign language system--Japanese learners, Japanese teachers, the Japanese test-taking system, the Japanese educational bureaucracy, even the Japanese language itself--is Japanese students' scores on the TOEFL test. Discounting the validity of the TOEFL test as a measure of language ability, an issue of significant debate in itself, one can comfortably drive huge logical holes in the data marched out by those who argue the TOEFL scores are indicative of a comparative lack of second language ability among Japanese learners.

Benjamin Disraeli once quipped, "There are three kinds of lies: lies, damn lies and statistics." A quick look at the TOEFL statistics on the following page bears out the British statesman's caveat.

On the surface, Japan fares quite poorly, finishing forty-fourth out of the 50 countries, and indeed, a number of educators, such as Sawa (1999) and Honna (1995) read the statistics at this surface level. Sawa, for example, bemoans the fact that, "In 1998 Japan was reportedly overtaken by Mongolia, slipping to the bottom of the TOEFL ranks" (p. 20).

Table 1 A Comparison of TOEFL scores across 50 nations

Country # test Score Country # test Score
takers takers
Holland 2, 322 607 Nigeria 3,330 540
Austria 1,051 594 Mexico 7,146 539
Denmark 1,174 593 USSR 1,008 538
Belgium 1,511 592 Malaysia 23,414 535
Sweden 2,174 589 Peru 3,466 534
West Germany 11,912 588 Brazil 8,032 532
Finland 1,015 586 Greece 12, 282 532
Singapore 4,728 583 Colombia 4, 584 527
Norway 3,568 572 China 78,894 521
India 44,487 571 Turkey 9, 511 515
Ghana 2,019 569 Iran 8, 252 508
The Philippines 9,254 569 Hong Kong 71,596 506
Italy 4,122 569 Egypt 6, 775 505
Switzerland 3, 41 569 S. Korea 67, 834 505
South Africa 452 567 Taiwan 47,098 505
Portugal 612 566 Ethiopia 2,868 503
Iceland 953 563 Libya 371 496
Argentina 2,522 560 Indonesia 23,971 496
Kenya 2,150 559 Japan 154,609 493
Hungary 587 559 Thailand 25,027 427
Czech/Slovak 464 558 Sudan 1,150 487
France 17,694 554 Jordan 9,425 472
Spain 7,001 551 S. Arabia 6,891 467
Yugoslavia 1,609 551 UAE 1,711 462
Israel 4,886 544 Kuwait 6,088 451

Source: TOEFL Test and Score Manual, 1990-1991

The statistics cry out for a deeper look. The ranking is neat and clear, but the numerical incongruity that pops out in the data is the number of Japanese test takers vis a vis their foreign counterparts. Represented as a percentage of the overall population of college age students, the number of Japanese examinees is significantly higher than in many of the countries with higher scores. There is no need here to present more statistical minutiae. The data can be interpreted in a commonsensical fashion. The statistics as they are presented lie. The number of college age students in Italy, The Philippines, and Germany respectively is within 30% (plus or minus) the number in Japan; Yet, over thirty times more Japanese students took the test compared to their Italian counterparts; sixteen times more vis a vis students from the Philippines; and thirteen times more than Germans.

So what do the statistics mean? How can we interpret them? First of all, the statistics do not mean the average Japanese college-age student scores lower than his/her average counterpart abroad, as many pundits have suggested. Obviously, a significantly larger number of Japanese students are taking the test than in other countries. With a bit of demographic research, we discover that a much larger percentage of Japanese students as a percentage of the total college-age population sit for the TOEFL than do students in most (but not all) of the countries listed in the table.

From a more rigorous inspection of the data, we may draw some reasonable inferences. First, compared to students in most of the other countries listed above, a much larger range of Japanese students in terms of English ability level are taking the TOEFL test (This does not hold true, however, for Taiwan and Hong Kong, countries with scores not surprisingly quite similar to those of Japan). This inference is supported by much circumstantial evidence: TOEFL is big business everywhere, but it is really big in Japan; in Japan taking standardized tests has become somewhat de rigeur , irrespective of one's language ability; measuring one's ability against standardized measures of competency is an integral pattern of learning in Japan, a pattern Letendre and Rohlen (1998) label as perfectibility (kaizen,$B2~A1(J ) and one that Reedy (1999) identifies as part of Japanese students' cultural learning history.

A second inference can be gained by inspecting the small number of examinees (as a percentage of the total college-age population) in countries where students scored particularly high, such as in Holland, Austria, Germany, and Denmark. One could reasonably infer that these students represent the academic elite in their respective nations, and that if one were to compare this elite with scores on the TOEFL of the academic elite in Japan, say the top 10% of high school graduates, one might very well find quite similar results.

A third inference one might draw from a more careful analysis of the data is that other key variables explain the disparity in the test scores: first, the distance between the L1 and English; second, the age at which English instruction begins in the schools. Swedish, Dutch and Danish are considerably closer linguistically to English than is Japanese; moreover, in these countries as well as in other high scoring countries, such as The Philippines, English language instruction begins at the primary school level. The data would appear to lend support for the introduction of second language education at the primary school level in Japan. Sawa (1999) chooses to conclude the opposite. Japanese are generally poor at foreign language, he claims the data demonstrates, ergo we should abandon required foreign language learning in the schools.

What information is needed to support (or detract from) the aforementioned inferences? First, one would have to measure the English ability of the test takers in all of the countries by some statistically valid test prior to their taking the TOEFL. In other words, one must hold constant the variable English language ability with the test takers. This would yield data enabling educators to draw fair conclusions about comparative language ability. This researcher believes it would likely point out the obvious: students with high English language skills and an early and high degree of exposure to English language learning would score approximately the same on the TOEFL across language cultures. Conversely, students with minimal exposure to English would score low across language cultures.

My reading of the statistics runs contrary to conventional wisdom. Given the large percentage of Japanese students taking the TOEFL, a score of 493--499 according to the 1998 TOEFL results--is fairly impressive. With one more statistically insignificant point, the average Japanese TOEFL examinee would qualify for entrance to dozens of universities in countries where English is considered the national or native language. Moreover, Japanese students score very close to or higher than students in countries where English is one of the primary languages of instruction in the schools, such as in Taiwan and the former Hong Kong, or where English is taught in many school districts from primary school, for instance in UAE and Kuwait.

The statistics are misleading, and unfair, for another reason. In effect they compare apples to oranges. We have bilingual or multilingual nations such as Switzerland being contrasted to primarily monolingual nations, such as Japan and Korea. We also have countries where English is the language of instruction in the schools, such as the Philippines, to nations where English is taught fifty minutes a day beginning in junior high school. This tells us little about comparative foreign language skills across cultures.

To say anything about language ability, the statistics must be interpreted in terms of the percentage of college age students taking the TOEFL within their respective countries. This would help explain the surprisingly low score of Kuwaitis. The number of test takers appears small (6,088); however, Kuwait is a nation of 1.8 million people, about the same as Chiba City. When the number of examinees is represented as a percentage of the total number of college-age students, Kuwait comes out on top, Hong Kong second, Taiwan third, with South Korea and Japan rounding out the top five. All of these nations are found near the bottom on the above list of TOEFL scores. Does this mean the average young high school graduate in these countries is poor at language learning? Given the large percentage of college-age students taking the test in these nations, the opposite conclusion could just as plausibly be drawn--one that keeps in mind Disraeli's caveat about statistics.

References

Clark, G. (2000, January 30). Why Taro can't speak English. The Japan Times, p. 19.

Dore, R. (1984). Education in Tokugawa Japan. London: Athlone Press.

Honna, N. (1995). English in Japanese society: Language within language. In J.C. Maher & K. Yashiro (Eds.). Multilingual Japan (pp. 45-62). Philadelphia: Multilingual Matters, Ltd.

LeTendre, G. and Rohlen, T. (1998). Conclusion: Themes in the Japanese culture of learning. In T. Rohlen and G. Letendre (Eds.) Teaching and learning in Japan (pp. 369-377). Cambridge: Cambridge University Press.

Mulvey, B. (1999). A myth of influence: Japanese university entrance exams and their effect on senior high school reading pedagogy. JALT Journal, 21 (1), 136.

Reedy, S. (1999). Recognizing and responding to cultural learning styles. Paper presented at the International TESOL Conference. New York City, March 1999.

Sawa, T. (1999, October 18). Cramming Cripples Japan. The Japan Times, p. 20.

TOEFL Test and Score Manual, 1990-91. Princeton, NJ: ETS.

Tolbert, K. (2000 January 29). English is the talk of Japan. The Washington Post, p. A13.

Sean Reedy has extensive international experience as a language and history educator, researcher, and writer. He obtained his B.A. from the University of California, Berkeley, and his doctorate from the University of San Francisco in International and Multicultural Education.

His current research areas are higher education policy-making and the sociolinguistics of language contact. He is Associate Professor of English and Linguistics at Maebashi Institute of Technology in Gunma, Japan. He has lived and worked in Japan since 1985.


All materials on this site are copyright © by JALT and their respective authors.
For more information on JALT, visit the JALT National Website