Confusion about the first goal is understandable: How can children acquire English, their second language, while being taught in their first language? This occurs for two reasons: First, when we give a child good education in the primary language, we give the child knowledge, knowledge that makes English input more comprehensible. A child who understands history, thanks to good history instruction in the first language, will have a better chance understanding history taught in English than a child without this background knowledge. And more comprehensible English input means more acquisition of English.

Second, there is strong evidence that literacy transfers across languages, that building literacy in the primary language is a short-cut to English literacy. The argument is straightforward: If we learn to read by understanding the messages on the page (Smith, 1994; Goodman, 1982), it is easier to learn to read if we understand the language. And once we can read, we can read: The ability transfers to other languages. 

The empirical support for this claim comes from studies showing that the reading process is similar in different languages, studies showing that the reading development process is similar in different languages, and that correlations between literacy development in the first language and the second language are high, when length of residence is controlled. All the above is true even when the orthographies of the two languages are very different (Krashen, 1996). 

Good bilingual programs thus have these characteristics: 

(1) They provide background knowledge through the first language via subject matter teaching in the first language. This should be done to the point so that subsequent subject matter instruction in English is comprehensible.

(2) They provide literacy in the first language.

(3) Of course they provide comprehensible input in English, through ESL and sheltered subject matter teaching. In sheltered classes, subject matter is taught to intermediate second language acquirers in a comprehensible way. (Sheltered classes are for intermediates; they are not for beginners and not for advanced acquirers or native speakers. It is extremely difficult to teach subject matter to those who have acquired none or little of the language. Beginners should be in regular ESL, where they are assured of comprehensible input. Including more advanced students in sheltered classes is problematic because their participation may encourage input that is incomprehensible to the other students. There is substantial evidence supporting the efficacy of sheltered subject matter teaching for intermediate level, literate students; Krashen, 1991).

Putting sheltered subject matter classes at this stage insures that they will be comprehensible. Students in sheltered math, for example, have had some ESL, giving them some competence in English, and have had math in the primary language, giving them subject matter knowledge. These two combine to help make sheltered math comprehensible. Those forced to do subject matter in the second language immediately, without any competence in second language, have neither of these advantages. The gradual exit program appears to be the fastest way of introducing comprehensible subject matter teaching in English. 

Note also that while the child is doing sheltered math, she is developing additional background knowledge and literacy through the first language in subjects that are more abstract, social studies and language arts. This will serve to make instruction in English at later stages more comprehensible.

In later stages, math and science are done in the mainstream and other subjects, such as social studies, are taught in sheltered classes in English. Eventually, all subjects are done in the mainstream. In this way, sheltered classes function as a bridge between instruction in the first language and the mainstream. 

Once full mainstreaming is complete, advanced first language development is available as an option. This kind of plan avoids problems associated with exiting children too early from first language instruction (before the English they encounter is comprehensible) and provide instruction in the first language where it is most needed. This plan also allow children to have the advantages of advanced first language development.

In the gradual exit program, the second language is not delayed. It is introduced as soon as it can be made comprehensible. Quite early on, students in these programs do a considerable amount of serious academic work in English, well before they reach the very high levels required for official reclassification. The gradual exit model is thus not subject to the criticism that bilingual education programs delay exposure to English for years. 

The evidence for bilingual education
Evidence supporting bilingual education is of several kinds: (1) the results of program evaluations; (2) the effect of previous education on immigrant children's academic performance; (3) the effect of measured first language ability on immigrant children's second language acquisition. This framework also helps explain the strong impact of SES on school success for immigrant children and why some are successful without bilingual education.

Program evaluations. I would like to suggest a somewhat different approach in evaluating and reviewing research on bilingual education, relaxing one requirement that others adhere to strictly, but insisting on others. The one I insist on is the definition of bilingual education: A program can be considered a properly organized bilingual education program when it provides (a) subject matter teaching in the primary language without translation to the point that subject matter instruction in the second language is made comprehensible; (b) literacy development in the primary language; (c) comprehensible input in the second language. My prediction is that full bilingual programs, with all three conditions met, will be superior to those with fewer conditions met. I also insist that studies have adequate sample sizes and that the programs run for at least one year (which may be far too short to show an effect). 

Other reviewers have required that there be some kind of control for experimental- comparison group differences that may have existed before the study began. Everyone agrees that randomization is the best way to do this. Lacking randomization, another technique is to statistically control for differences, in pretest scores and/or background differences. In this review, I relax this requirement and allow studies to enter the analysis where there is no compelling reason to suspect that the groups come from different populations. The logic behind this approach was presented in Krashen (1996): With a large number of post-test studies of this kind, randomization is present. 

My conclusions are these: In all published studies in which these conditions are met, bilingual education is a winner. Children in bilingual programs acquire more of the second language than those in all-English programs (Mortensen, 1984), even eventually doing as well as native speakers of English (de la Garza and Medina, 1985; Burnham-Massey and Pina, 1990). Results of studies in other countries are similar (Fitzpatrick, 1987; Modiano, 1968 [literacy instruction in L1 only]; Appel, 1984; Verhoeven, 1991 [literacy only]). 

In addition, apparent counterexamples, cases in which bilingual education was thought to be inferior, do not meet the conditions outlined above. In these comparisons, bilingual ed is not described or inaccurately described, sample sizes are small, and/or real comparisons are not made.

For example, Rossell and Baker (1996) present ten studies which "immersion" is considered to be better than bilingual education. Six are actually comparisons of different versions of Canadian immersion, a program that satisfies all three of the characteristics given above: They are all bilingual education. In all versions of Canadian immersion, children obtain enough background knowledge and develop enough literacy through the first language, both in school and at home, to make subject matter taught in the second language comprehensible (Krashen, 1996). Thus, those with more comprehensible input in the second language acquire more ot it, since factors (a) and (b) are fully satisfied. In several other cases, categorization is inaccurate: what Rossell and Baker consider to be "immersion" and "submersion" are actually bilingual education; comparisons that Rossell and Baker consider to be between bilingual education and submersion or immersion are really comparisons of different versions of bilingual education (e.g., El Paso and McAllen; see Krashen, 1996).

In many reports we have no idea whether the experimental (bilingual) or comparison groups met any of these conditions, as descriptions are lacking. This eliminates many studies that others have allowed into their analyses, such as Gersten (1985), which also suffers from a very small sample size, Moore and Parr (1978) and Rossell (1990). All of these studies are considered by Rossell and Baker (1996) to support immersion or submersion over bilingual education. 

In other cases, results of studies are inaccurately reported: In Curiel, Stennina and Cooper-Stenning (1980), which Rossell and Baker consider to be an example of the superiority of submersion, non-bilingual students were better in grade 6, but in grade 7 bilingual education students had higher GPA's, fewer had been retained, and there was no difference in English reading. In addition, Curiel et. al. note that the bilingual program was used as a remedial program for some students who had been in the monolingual English program previously. Thus, a finding of no difference suggests that those placed in bilingual education were able to catch up to the others.

Natural experiments. Two natural experiments meet the criteria outlined above: Spanish-speaking children who had all their schooling in the US were compared to those who had some of their education in Mexico. In one case, all children were in a bilingual program (Gonzales, 1989); the study thus compared the impact of some first language instruction with more. In the other study, education in the US was all-English (Ferris and Politzer, 1981). In the former case, sixth graders with some education in Mexico did better than all-US educated comparisons in English reading, while in the latter there was no difference between the groups in English writing in junior high school, but the Mexican-educated children had higher grades in English, and, according to teacher reports, were more dedicated students. Ferris and Politzer also report that the socioeconomic status of those with schooling in Mexico was lower than that of the all-US educated group.

Impact of education in L1. Also consistent with this theoretical stance are studies showing that those with more education in the primary language are more successful in English language acquisition, a result that confirms the powerful influence of subject matter knowledge and literacy gained through the first language. 

Here is just one example: Gardner, Polyzoi and Rampaul (1966) studied the impact of education in the first language on progress in intensive ESL classes for Kurdish and Bosnian adult immigrants to Canada who had "virtually no English" when they arrived (p. 3). The subjects were classified into three groups, those with a great deal of formal education (mean = 15 years), those with some (mean = 7 years) and those with no formal education.

Table 2 presents the gains made by each group on tests of oral and written English after participation in intensive ESL (20 hours per week, for 1 to 1.5 years). For both measures, it is clear that the higher the level of literacy in the primary language, the greater the gains. This was true of both measures, and extremely powerful in the written test, in which preliterates' posttest scores were lower than the high literates' pretest scores.

Perfect score = 100 for both tests. Oral test: Personal questions, picture description, discussion of leisure activities, family. Written test: write name, circle correct time, copy words, label pictures, answer personal questions, read text and write answers to questions, fill in blanks with correct prepositions, verb tenses, multiple choice vocabulary

Source: Gardner, Polyzoi and Rampaul, 1966

The strong impact of first language education on English development has been confirmed in several other studies (Chiswick, 1991; Chiswick and Miller, 1995; Espanshade and Lu, 1997).

SES as defacto bilingual education. Our framework helps explain the consistent positive relationship between SES and English language development (Krashen, 1996). Children from wealthier families have, most likely, more and better education in their primary language, caregivers who are better able to help them with schoolwork (in their primary language), and have more access to print in general.

Immigrant success. The research presented here helps explain why some immigrants did well in school without bilingual education: They came with a good education in their own country, making instruction in English much more comprehensible (Krashen, 1996; Tse, 1997). Such cases are arguments for bilingual education, not against it.

Bogus arguments against bilingual education
Many of the arguments used to attack the effectiveness of bilingual education violate one or more of the principles presented here. 

Is immersion successful? In several cases, the media has claimed success for "immersion" when no comparison was made at all with similar children under bilingual education. 

ORANGE. According to the Los Angeles Times (Orange County edition) "A controversial new English immersion program in the Orange Unified School District appears to help many students learn to speak the language faster than traditional bilingual programs" (April 18, 1998). The article, which appeared in the middle of the Proposition 227 campaign, announced that "almost a quarter of the district's 4,132 elementary students in the immersion program had advanced their fluency by at least one level in the first five months of study." Orange, the Times reported, dropped bilingual education the year before and "went with English immersion."

A closer look shows that this did not occur at all: First, while Orange claimed that they dropped bilingual education. their current English immersion program used at least some first language support, supplied by paraprofessionals. Even more serious, no comparison was made of the progress made by children in the current program and children in the older program. Finally, the progress was not remarkable. I concluded (Krashen, 1999) that at the rate these children were progressing, for those starting with no English, after one year fewer than half would be ready for sheltered subject matter instruction in English, and fewer than 20% would be ready for the mainstream in one and a half years.

McQuillan (1998) analyzed a more recent report from the same district, and also concluded that children were not doing very well under the "immersion" plan. The district, for example, claimed that 81% of the immersion students could understand English in specially taught classes. McQuillan points out, however, that three-quarters of the children were already advanced enough in English to do a modified program before the new program began! In addition, only six of the 3,549 students were ready for regular classes after one year (the time limit imposed by Proposition 227) , a dismal 1% reclassification rate, and, in agreement with my findings, only about half who began with no English were ready for "sheltered" classes after one year. Finally, once again Orange did not compare their students' progress with progress under older programs. 

WESTMINSTER. The Westminster School District claimed that "after 18 months of instruction only in English ... pupils have made better academic progress and learned more English than they did when taught in their native languages" (Long Island Newsday, November 28, 1997). But gains were modest (3 NCE points in one year), no comparisons were ever made with previous programs. In fact, Westminster actually increased the amount of first language support provided to pupils, in the form of paraprofessionals (Krashen, 1999).

TAFT. In another case, the comparison group was completely inappropriate. Children at the Taft School in Santa Ana scored at the 48th percentile in English reading on the CTBS in Spring, 1997, well above the district average of 22.5 and the highest in the district. Taft's principal credited the school's English immersion philosophy for some of this performance (Education Week, January 14, 1998). But Taft's students are clearly more advantaged than others in the district, as revealed by the following data:
 

Taft lies two standard deviations above the mean for free/reduced lunch as well as for percent of limited English proficient students. The correlation between reading scores and SES status (as measured by % free/reduced lunch) was nearly perfect ( r = .926)(Krashen, 1999).

Taft's "success" has, most likely, nothing to do with the absence of bilingual education. In fact, some of it could be due to "de facto" bilingual education, the superior education in the primary language that more advantaged children tend to have.

DID LA STUDENTS "TAKE TO IMMERSION"? An article appearing in the Los Angeles Times on January 13, 1999 added to the confusion. The article itself was reasonably well balanced. Interviews were conducted in the Los Angeles Unified School District, and the Times reporter concluded that children were picking up "verbal English at a surprising rate" but also reported that there were concerns that children were falling behind in their studies; many teachers were questioning "whether most of the youngsters have acquired the language skills necessary to comprehend math, reading or history lessons in English." One teacher noted that children were picking up "social English," not academic English, that new concepts still had to be presented in the primary language, and that "we won't have as many readers in our class as we did last year" (under bilingual education). Other teachers said that they had to "water down" core subjects. 

This report was not a true evaluation. It was based on only 13 interviews and no test scores or any other kind of statistics were reported. Further, the Times reporter did not point out that children typically pick up conversational English with any kind of program, even with no special help: The challenge is to help them develop what Cummins (1989) calls "academic language," the language of school. 

Nevertheless, the headline of the article proclaimed: "L.A. students take to English immersion" and the first paragraph stated that "teachers are delivering promised reports that their children are learning English more quickly than anticipated," overstatements that are inconsistent with the rest of the report and that encouraged the perception that 227 had been an unqualified success.

Unfortunately, these are not isolated examples of sloppy reporting (see Krashen, 1999 for more examples).

Delaware and Massachusetts. Glenn (1998) also presents an inappropriate comparison: "Massachusetts mandates bilingual education, for example, Delaware prohibits it, but Hispanic achievement is not notably higher in one state than in the other; indeed, the gap between Hispanic and non-Hispanic white scores on the National Assessment of Education Progress was substantially larger in Massachusetts than it was in Delaware ..." (p. 6). 

There are a lot of problem with this comparison, as we document elsewhere (Krashen and Crawford, 1999). First, it appears that the groups compared may be quite different: There are about 62,000 Hispanic students in Massachusetts and 43% are LEP (1996-7 data), but there are about 4000 in Delaware and only 30% are LEP. Clearly, Delaware has a lower percentage of LEP Hispanic students, which suggests that the groups differ in other important ways.

Second, we may actually be comparing bilingual education with bilingual education. Even though bilingual education is officially banned in Delaware, this ban is not enforced (see Crawford, 1999, p. 42). Red Clay Consolidated District in Delaware does bilingual education, and 795 out of Delaware's 1725 LEP children, and 762 of the 1299 Hispanic LEP children, were enrolled in Red Clay in 1996-1997. There appears to be plenty of bilingual education in Delaware. 

Third, we have no description at all of what kind of bilingual education was done in either state.

Finally, Hispanics in Massachusetts actually scored higher than Hispanics in Delaware on the NAEP reading exam, both in 1994 (194 to 190, a small difference, not "notable," as Glenn points out) and 1992 (a larger difference, 201 to 187). Glenn notes that the gap between Hispanics and non-Hispanic whites was larger in Massachusetts than in Delaware. This was true in 1994, but not in 1992. In 1992 the difference between non-Hispanic whites and Hispanics was larger in Delaware than in Massachusetts. Thus, even if we accepted the Massachusetts-Delaware comparison as a true experiment, which it is not, the data does not consistently support Glenn's assertions.

The 6% argument. Perhaps the most bizarre argument used by critics of bilingual education is the 6% (or 2%, depending on the state) argument. The California version was this: Bilingual education has failed because " ...each year only about 5% of school children (in California) classified as not proficient in English are found to have gained proficiency in English – the current system of language education has an annual failure rate of 95%" (English for the Children brochure).

The figure referred to is the percentage of limited English proficient children who are reclassified as English proficient each year. (The precise percentage was 6.2% in 1996, up from 5.7% in 1995.) To be reclassified means to reach a high enough level of English literacy to be considered fully English proficient. Calling this a "failure rate" is inaccurate and misleading. As Jeff McQuillan has pointed out, using this definition, a four-year college would have a 75% failure rate, even if all students graduated in four years.

The six percent figure does not represent the success of bilingual education. The California figure was based on all limited English children in California, not just those in bilingual education. The real issue is whether children in full bilingual programs, about 30% of those classified as LEP, had a lower reclassification rate. This is, in other words, a study with no clear experimental group and no clear control group. It is counter to all scientific practice.

It also needs to be pointed out that reclassification is not easy to do: Some districts require that children place in the upper 2/3 of tests of English reading, which by definition 1/3 of the native speakers fail to accomplish. Many limited English proficient students are doing challenging academic study in English well before official reclassification.

Is bilingual education responsible for dropouts? The circumstantial argument is this: Hispanic students have a large dropout rate. Hispanic students are the biggest customer of bilingual education programs. Therefore bilingual education causes dropouts. 

False. Only a small percentage of Hispanic students are enrolled in bilingual education: In California, for example, only 15% were in full bilingual programs. In addition, the only empirical study of the impact of bilingual education on dropouts, Curiel, Rosenthal, and Richek (1986), reported fewer dropouts among bilingual education students than among comparison students. 

What accounts for dropout rates? Not surprisingly, competence in English (McMillan, Kaufman and Klein, 1997). But if bilingual education results in better English development, as claimed above, this finding is an argument for bilingual education. 

A large number of studies confirm that other factors count, such as socioeconomic class, time spent in the US, the presence of print, and family factors. Hispanic students are well behind majority children in these areas. What is especially interesting is that these background factors appear to be responsible for much if not all of the difference in dropout rates among different ethnic groups. In other words, when researchers control for these factors, there is little or no difference in dropout rates between Hispanics and other groups. (Rumberger, 1995; Rumberger, 1983; Fernandez, Paulsen, and Hiranko-Nakanishi, 1996; Warren, 1996; White and Kaufman, 1997; Pirog and Magee, 1997). Rumberger (1995), for example, concluded that " ... Black, Hispanic, and Native American students have twice the odds of dropping out compared to White students ... however, after controlling for the structural characteristics of family background – particularly, socioeconomic status – the predicted odds of dropping out are no different than those for White students" (p. 605). Rumberger (1983) confirms that Hispanic students often drop out because they have to go to work. When dropouts were asked why the dropped out, only 4% of the Hispanic students mentioned poor performance in school, compared to 8% of comparisons. But 38% of the Hispanic students mentioned economic factors, compared to 22% of the other students.

Bilingualism, bilingual education, and earnings.
Lopez and Mora (1998) claim that Hispanics who participated in bilingual education programs earn significantly less in their late 20's than comparison subjects who did not participate in bilingual education. Jeffrey McQuillan and I have argued that this conclusion is not correct (Krashen and McQuillan, 1998).

• Earnings differences between former ESL and bilingual students were small and not statistically significant. 
• Comparison group subjects, those with no ESL and no bilingual education, probably included English-only native speakers of English: Subjects were allowed into this group if they reported that a language other than English was spoken at home, even if they themselves didn't speak it or understand it.
• We have good reason to believe that subjects' self-reports were not accurate. The subjects were sophomores in high school when they were first interviewed, which was in 1980. 27.5% claimed they had been in bilingual education. This would have occurred in the early 1970's. But there were few bilingual programs in those days. Jim Crawford has pointed out to me that only 6.5% of the schools in California, Arizona, Colorado and Texas had bilingual education, and the programs reached only 2.7% of the Mexican-American population. 
• We have no information on how bilingual education was done. It is not even clear whether "bilingual education" included ESL as a component.
• Lopez and Mora include third generation enrollees in ESL and bilingual education. By then, language shift is nearly always complete.

1. Those who speak only English earned more in 1989 than those who reported that another language was spoken in their home, even when factors such as schooling, years in the labor market, amount worked, marital status, and urban/rural were controlled. Overall, English-onlys (those who only heard English at home) earned about 8% more.

2. Even those who grew up in homes with another language who reported that they spoke English "very well" earned less than English-onlys.

However:

• Those who really suffered were Native American, Hispanic, and "Mexican" men (for some reason, "Mexican" was not considered "Hispanic"). Even those who reported they spoke English "very well" earned less than English-onlys, 16%, 9%, and 7% respectively. Other groups had either a much smaller gap (3% for "white-non Hispanic) and none at all for Black and Asian-origin men. Chiswick and Miller also found that Hispanics who speak English "very well" but who live in "high concentration Spanish states" earned 11% less than English-onlys, but those other states were only 4% lower. These results suggests that language may not have been the central issue in determining earnings, a possibility that Chiswick and Miller present. 
• All other studies of heritage language show rather positive effects of bilingualism: Those who develop their heritage language, in addition to acquiring English, do slightly better in school and on the job market (research reviewed in Krashen, 1998). The overwhelming majority of children of immigrants report higher competence in English than in the heritage language by the time they are in high school (Krashen, 1996). Thus, most of those who speak another language at home probably do not develop it to high levels, for a variety of factors. Language shift is powerful. Most of Chiswick and Miller's subjects were, most likely, weak heritage language speakers. Their data is thus consistent with the hypothesis that high development of the heritage language is positive, and that weak development of the heritage language is a disadvantage. 

The polls clearly show that the public is not against bilingual education. Respondents last year in both Los Angeles and Texas agreed either that "Students should be taught in their native language for a brief time – a year or two" (Texas poll = 38% agreement. Los Angeles Times poll = 39%), or that first language instruction "should last as long as teachers and parents think it is necessary." (Texas poll = 36% agreement. Los Angeles Times poll = 25% agreement.) Thus, 74% of Texans surveyed supported some use of the first language in school and 64% of those surveyed in Los Angeles. Only a small percentage supported English only (Texas = 24%; Los Angeles = 32%). (Details in Krashen, 1999.)

A series of studies by Fay Shin also suggests that the public is not against bilingual education. Shin did not ask people if they supported bilingual education; instead, she asked about the underlying principles, asking whether people thought "developing literacy through the first language facilitates literacy development in English" and whether "learning subject matter through the first language helps make subject matter study in English more comprehensible." Results were encouraging; these principles apparently make good sense to a lot of people:

Developing literacy through the first language facilitates literacy development in English. Percent agreement:

Hispanic parents = 53% (Shin and Gribbons, 1996)

Korean parents = 88% (Shin and Kim, 1998)

Hmong parents = 52% (Shin and Lee, 1996)

Administrators = 74% (Anton and Shin, forthcoming)

Teachers = 74% (Shin and Krashen, 1996)

Hispanic parents = 34% (33% were "not sure") (Shin and Gribbons, 1996)

Korean parents = 47% (Shin and Kim, 1998)

Hmong parents = 60% (Shin and Lee, 1996)

Administrators = 78% (Anton and Shin, forthcoming)

Teachers = 70% (Shin and Krashen, 1996)

When polls seem to indicate that the public is against bilingual education, a closer look reveals that this is not so. 

• In some cases, the public is simply expressing support for children learning English, a goal we all agree with. In fact, this explains much of the success of Proposition 227: many people thought they were simply "voting for English" (Krashen, 1999). Of course, when parents say they want children to learn English, this should not be interpreted as a rejection of bilingual education. 
• When parents reject bilingual education explicitly, they reject versions of it that few bilingual education advocates would support, i.e., versions in which all instruction is in the first language "until children are ready to learn English." As noted earlier, I think children are ready for English the first day of school (Krashen, 1999).
• Some polls ask if parents are willing to delay subject matter for English, that is, take time for English study before children learn subject matter. This is an unreasonable question: In good bilingual education programs, children get both maximum subject matter instruction and make maximum progress in acquiring English at the same time. The former helps the latter, as explained earlier in this paper.
• Some polls ask if parents are willing to delay English while the children are instructed in their native language. This is also an unreasonable question: In good bilingual programs, there is no delay of English.

If voters had known what was really in 227, they would have voted differently. This was confirmed in our study. Jim Crawford noted that the following kind of question, closely following the description of 227 on the ballot, was typically asked of voters in polls: "There is an initiative on the June primary ballot that would require all public school instruction to be conducted in English and for students not fluent in English to be placed in a short-term English immersion program. If the June primary were being held today, would you vote for or against this measure?"

This kind of question can be easily interpreted as "Are you in favor of children getting intensive English instruction?" and did not reflect what was in Proposition 227. A more accurate question, Crawford suggested, would be one like this one: "There is an initiative on the June primary ballot that would severely restrict the use of the child's native language in school. This initiative would limit special help in English to one year (180 school days). After this time, limited English proficient children would be expected to know enough English to do school work at the same level as native speakers of English their age. The initiative would dismantle many current programs that have been demonstrated to be successful in helping children acquire English, and would hold teachers financially responsible if they violate this policy. If passed, schools would have 60 days to conform to the new policy. If the June primary were being held today, would you vote for or against this measure?"

Students in my language education class asked 251 voters either question 1 or question 2 and the data was analyzed by Haeyoung Kim. The difference between the responses to the two questions was huge (and statistically significant): While 57% supported the original version, only 15% supported the modified version, a result that confirmed our suspicions that few people knew what was in Propostion 227, and if they had known, most would have not supported it.
 

Unfortunately, despite numerous attempts, we were unable to get crucial information about 227 to many voters.

References
Appel, Rene. 1984. Immigrant children learning Dutch. Dordrecht, The Netherlands: Foris.

Burnham-Massey, Laurie, and Pina, Marilyn. 1990. Effects of reading instruction on English academic achievement of LEP children. Reading Improvement 27: 129-132.

Chiswick. Barry. 1991. Speaking, reading, and earnings among low-skilled immigrants. Journal of Labor Economics 9: 149-170.

Chiswick, Barry, and Miller, Paul. 1995. The endogeneity between language and earnings: International analyses. Journal of Labor Economics 13: 246-288.

Chiswick, Barry, and Miller, Paul. 1998. The economic cost to native-born Americans of limited English language proficiency. Report prepared for the Center for Equal Opportunity. August, 1998.

Crawford, James. 1999. Bilingual education: History, politics, theory and practice. Fourth Edition. Los Angeles: Bilingual Educational Services. 

Cummins, Jim. 1989. Empowering minority students. Los Angeles, CA: California Association for Bilingual Education.

Curiel, Herman; Stenning, Walter; and Cooper-Stenning, Peggy. 1980. Achieved reading level, self- esteem, and grades as related to length of exposure to bilingual education. Hispanic Journal of Behavioral Sciences 2(4): 389-400.

Curiel, Herman; Rosenthal, James; and Richek, Herbert. 1986. Impacts of bilingual education on secondary school grades, attendance, retentions and drop-out. Hispanic Journal of Behavioral Sciences 8(4): 357-367.

Espenshade, Thomas, and Haishan Fu. 1997. An analysis of English-language proficiency among U.S. immigrants. American Sociological Review 62: 288-305.

Fernandez, Roberto; Paulsen, Ronnelle; and Hiranko-Nakanishi, Marsha. 1989. Dropping out among Hispanic youth. Social Science Research 18: 21-52.

Ferris, M. Roger, and Politzer, Robert. 1981. Effects of early and delayed second language acquisition: English composition skills of Spanish-speaking junior high school students. TESOL Quarterly 15(3): 263-274.

Fitzpatrick, Finbarre. 1987. The open door. Multilingual Matters.

Gardner, Sheena; Polyzoi, Eleoussa; and Rampaul, Yvette. 1996. Individual variables, literacy history, and ESL progress among Kurdish and Bosnian immigrants. TESL Canada 14: 1-20.

Gersten, Russell. 1985. Structured immersion for language minority students: Results of a longitudinal evaluation. Educational Evaluation and Policy Analysis 7: 187-196.

Glenn, Charles. 1998. Rethinking bilingual education: Changes for Massachusetts. READ Abstracts, Research and Policy Review, August, 1998. 

Gonzales, L. Antonio. 1989. Native language education: The key to English literacy skills. In D. Bixler-Marquez, J. Ornstein-Galacia, and G. Green (Eds.), Mexican-American Spanish in its societal and cultural contexts (pp. 209-224). Rio Grande Series in Languages and Linguistics 3. Brownsville, Texas: University of Texas, Pan American. 

Goodman, Kenneth. 1982. Language, literacy, and learning. London: Routledge & Kegan Paul.

Krashen, Stephen. 1991. Sheltered subject matter teaching. Cross Currents 18: 183-188. 

Krashen, Stephen. 1996. Under attack: The case against bilingual education. Culver City, CA: Language Education Associates.

Krashen, Stephen. 1998. Heritage language development: Some practical arguments. In Stephen Krashen, Lucy Tse, and Jeff McQuillan (Eds.), Heritage language development. Culver City, CA: Language Education Associates.

Krashen, Stephen. 1999. Condemned without a Trial: Bogus arguments against bilingual education. Portsmouth, NH: Heinemann.

Krashen, Stephen, and Crawford, Jim. 1999. The research, the scientific method, and the Delaware- Massachusetts argument. NABE News 22(5): 14-15. 

Krashen, Stephen, and McQuillan, Jeffrey. 1998. Do graduates of bilingual programs really earn less? A response to Lopez and Mora. NABE News 22(3): 506.

Lopez, Mark, and Mora, Marie. (1998). The labor market effects of bilingual education among Hispanic workers. READ Perspectives 5(2): 33-54.

McMillen, Marilyn; Kaufman, Phillip; and Klein, Steve. 1997. Dropout rates in the United States: 1995. Washington: US Dept of Education. NCES 97-473.

McQuillan, Jeff. 1998. Is 99% failure a "success"? Orange Unified's English immersion program. Multilingual Educator 21(7): 11.

Modiano, Nancy. 1968. National or mother tongue language in beginning reading: A comparative study. Research in the Teaching of English 2: 32-43.

Moore, Fernie Baca, and Parr, Gerald. 1978. Models of bilingual education: Comparisons of effectiveness. The Elementary School Journal 79(2): 93-97.

Mortensen, Eileen. 1984. Reading achievement of native Spanish-speaking elementary students in bilingual vs. monolingual programs. Bilingual Review 11(3): 31-36.

Pirog, Maureen, and Magee, Chris. 1997. High school completion: The influence of schools, families, and adolescent parenting. Social Science Quarterly 78: 710-724.

Ramos, Francisco, and Krashen, Stephen. 1997. Success without bilingual education? Some European cases of de facto bilingual education. CABE Newsletter 20(6):7,19.

Rumberger, Russell. 1983. Dropping out of high school: The influence of race, sex, and family background. American Educational Research Journal 20(2): 199-220.

Rumberger, Russell. 1995. Dropping out of middle school: A multilevel analysis of students and schools. American Educational Research Journal 32(3): 583-625.

Rossell, Christine. 1990. The effectiveness of educational alternatives for limited-English proficient children. In Gary Imhoff (Ed.) Learning in Two Languages (pp. 71-121). New Brunswick, NJ: Transaction Publishers. 

Rossell, Christine, and Baker, Keith. 1996. The educational effectiveness of bilingual education. Research in the Teaching of English 30(1): 7-74.

Shin, Fay and Simon Kim, S. 1998. Korean parent perceptions and attitudes of bilingual education. In R. Endo, C. Park, J. Tsuchida and A. Abbayani (Eds.) Current issues in Asian and Pacific American education. Covina, CA: Pacific Asian Press. 

Shin, Fay, and Gribbons, Barry. 1996. Hispanic parent perceptions and attitudes of bilingual education. Journal of Mexican American Educators, pp. 16-22.

Shin, Fay, and Lee, Bo V. 1996. Hmong parents: What do they think about bilingual education? Pacific Educational Research Journal 8(1): 65-71.

Smith, Frank. 1994. Understanding reading. Fifth edition. Hillsdale, NJ: Erlbaum.

Tse, Lucy. 1997. A bilingual helping hand. Los Angeles Times, Dec. 17, 1997.

Verhoeven, L. 1991. Acquisition of biliteracy. AILA Review 8: 61-74.

Warren, John. 1996. Educational inequality among White and Mexican-origin adolescents in the American Southwest: 1990. Sociology of Education 69: 142-158.

White, Michael, and Kaufman, Gayle. 1997. Language usage, social capital, and school completion among immigrants and native-born ethnic groups. Social Science Quarterly 78(2): 385-398. 

COPYRIGHT NOTICE: Copyright © 1999 by Stephen D. Krashen. All rights reserved. This paper will appear in the proceedings of GURT '99, ed. James E. Alatis (Washington, D.C.: Georgetown University Press, forthcoming).