The Truth About the CELDT
California English Language Development Test

Jill Kerper Mora
San Diego State University

 

See Dr. Mora's update on the CELDT 2003 & 2004 and an analysis of the Legislative Analyst's Office report on the CELDT data.

 

Trumpeting the "Good New" 

On March 25, 2003 the California Department of Education (CDE) released the results of  the second year of administration of the California English Language Development Test (CELDT). Superintendent of Public Instruction, Jack O'Connell hailed the results of the comparison of the 2001 and 2002 test data as a tribute to the outstanding efforts of California's public schools in teaching English Language Learners (EL). 

Following the CDE Press Release, a number of articles and editorials appeared in newspapers in California and Arizona, trumpeting the test data as validation of the success of English immersion and Proposition 227, passed in 1998. Many of these media reports were riddled with inaccuracies and misinterpretations of the CELDT data, often reflecting a strong political bias. Several reporters drew sweeping conclusions about how the CELDT data affirmed Proposition 227 based on numerical comparisons of percentages of students in bilingual programs versus English-only programs who allegedly reached the "proficient" level in English on the test. 

See for example Nanette Asimov's article "English-only students do better on state test/Number of proficient speakers tripled after Prop. 227 passed." (March 26, 2003) in the San Francisco Chronicle.  Asimov declares that the CELDT data demonstrate that Proposition 227 "appears to be working." She quotes the author of Proposition 227, Ron Unz, describing his reaction to the CELDT data as an "I-told-you-so." Without quoting any expert opinions on what the test data really mean, Asimov states that "Despite the results, school districts' language experts said they were not preparing to trash their bilingual programs."  

Then to add insult to injury, the Arizona Republic published an article by Doug MacEachern in which Asimov's misinterpretations of the test results were quoted as if they were fact. Thus the media recycled misinformation and unsubstantiated conclusions to bolster yet another round of attacks on bilingual educators and bilingual programs. The Arizona Republic (See "English immersion is working in California" April 14, 2003) uses terms like "astonishing," "amazing," "spectacular," "whopping success" to describe the CELDT results, as well as lancing insults at the "bilingualists" who are allegedly "contemptuous of the mandate for English immersion" under Arizona's Proposition 203. 

Why have reporters stepped so boldly into the arena of educational tests and measurements, language assessment, and policy analysis? It is crucial to set the record straight about the CELDT. María Quezada, Executive Director of the California Association for Bilingual Education (CABE) clearly articulated the common concerns of educators regarding the impact of inaccurate and biased reporting of the CELDT results in the April 2003 edition of Language Magazine: "In newspaper articles those opposed to alternative bilingual programs immediately compared program "results" and the alleged superiority of one approach against the other... This, at best, is erroneous and at worst serves only to politicize and further divide not only the educational community, but the wider community as well." (Quezada, 2003, p. 20).

We must examine closely the impact of misinterpretations and misrepresentations of educational statistics and test data. The CELDT data provides our first state-wide glimpse into English acquisition among language minority students. Interpretations of the meaning of the data can impact policy regarding the types of programs that are implemented for English learners, the levels of support for alternative programs, the training and recruitment of teachers, and the design of curriculum. We know, for instance, that the issue of whether bilingual education programs will be eligible for funding under the federal Reading First grants is currently a matter of fierce policy debate, as well as judicial review. The public's perception of the relative effectiveness of alternative programs for language minority students in advancing the academic achievement is of vital importance in formulating and implementing education policies. This is why we must be well informed about what the CELDT test results really mean and we must hold the media accountable for accurate and unbiased reporting of the test data and statistics.

Taking a Closer Look

The purpose of this analysis of the results of the first cycle of administration of CELDT is to clarify the meaning of the results of this test of English language acquisition and to analyze the public policy implications of the data for improving education of language minority students. There are important questions to pose regarding the CELDT and its use as a guage of the effectiveness of a law passed through the ballot initiative process:

What does the CELDT really tell us about how EL are progressing in English language acquisition and academically overall?
Do the CELDT results support the effectiveness of English immersion?
Do the CELDT results allow us to compare the effectiveness of English immersion versus bilingual education?
Are the cost and human resources dedicated to annual administration of a language assessment instrument such as the CELDT worth it?
Who is vindicated by the CELDT results?
What else do we need to know to improve education for English Language Learners (EL) in California and throughout the nation. 

For what purpose are we administering the CELDT? The CELDT is a language assessment instrument or test that is designed to determine a ranking on a five point scale of a student's proficiency in three areas: listening/speaking (L/S), reading (R) and writing (W). The scores in each of these areas are weighed to derive an overall score: Listening/speaking =50%, reading=25% and writing=25%. It is important to keep these components and their relative value in the overall score in mind in interpreting the results of the CELDT for individual students and for groups of students, although individual student's scores are not available to the public.  The CELDT test defines full English proficiency as follows: "CELDT Criteria for English language proficiency requires students to achieve a proficiency level of at least Early Advanced with Skill Area scores of at least Intermediate." (See Attachment C). 

For a complete description of the CELDT, click here (CDE Press Release 2002 and CDE Press Briefing Packet 2003). For a history of the legislation that mandated the development and implementation of the CELDT, see the CDE 2001 Press Release.

The CELDT has three uses:

1. Initial identification and classification of students as English Learners or limited English proficient (LEP)
2. Monitoring students' "satisfactory yearly progress" in learning English 
3. As one of four criteria for reclassification as Fluent English Proficient 

Formerly, the CDE approved seven language assessment instruments for use for identification and placement of EL and for reclassification (also known as redesignation). The use of a language assessment instrument as a measure of "satisfactory yearly progress" is unprecedented among experts in language assessment and language acquisition. There are sound pedagogical and practical reasons for this. In part, it is because the language learning curve is not a straight line and it is very difficult to determine what a "satisfactory" increment of growth or average is likely to be for every student. In addition, administration of language assessments is very labor intensive and costly since it requires a one-on-one interview with the student and detailed analysis of the language sample obtained through this process. 

The test was conceived as a way to monitor instruction to ensure that students are making adequate progress in learning English. It is assumed that these data will provide information for school administrators and teachers to improve instruction for EL in the schools. Are the EL of California making "satisfactory progress" in learning English? If we believe the press reports and the glowing comments from the California Department of Education (CDE) after one year of data collection, they are. The CDE has defined "satisfactory yearly progress" as growth in one level on the CELDT per year of instruction. At this rate, it will take the average EL five years to attain full proficiency in English. 

English Language Learners' academic achievement is multidimensional. Students must make satisfactory progress in three areas: Language, literacy and content learning. We must keep in mind that the CELDT does not measure content knowledge. We cannot make the assumption that a rapid acquisition of listening comprehension and oral English automatically equates to growth in literacy or knowledge of academic content. These other dimensions of academic progress require "heavy lifting" in terms of the intensity and quality of instruction provided to EL in any type of program or classroom. 

The Media Got It Wrong: Fact vs. Fiction

The data from the second year of administration of the CELDT does not provide an accurate basis for comparing programs of instruction.  Nor do these statistics support the conclusion that students in one program are learning English faster than students in other programs. The data merely describe the English proficiency levels of students enrolled in different programs. The misinterpretation of these factors leads to faulty conclusions regarding cause and effect. 

Admittedly, the CELDT data and statistics are difficult to report since they involve highly technical terminology, distinctions between groups and subgroups of students, and legal definitions of complicated terms about language proficiency. State officials cautioned against making interpretations of the data to draw conclusions about the sensitive political issue of Proposition 227. However, some reporters went where state officials fear to tread. 

Let us summarize the fiction created by the media regarding the CELDT results and distinguish the facts based on the actual data.

Fiction	Fact
The numbers of proficient English speakers tripled between 2001 and 2002.	The numbers of students who met the full legal definition of Fluent English Proficient (FEP) increased by 8% between 2001 and 2002.
The CELDT data show that EL students in English only classrooms are "doing better" than students in bilingual programs.	Generalizable qualitative inferences or conclusions about how students are doing academically overall or in English language learning in different programs cannot be made based on the CELDT data. The characteristics of the student populations served through different programs of instruction are vastly different and therefore the groups are not comparable. In addition, there is considerable overlap between the groups. Some unknown number of English-only students have been enrolled for a period of time in a bilingual program. In these cases, it is impossible to disaggregate the impact of each program on their academic achievement.
EL students' English proficiency is increasing at a slower rate in bilingual programs than in English only programs.	A rate of growth in English proficiency has not been determined through valid and reliable statistical analysis and cannot be determined based on descriptive statistics alone. Consequently, no judgments can be made about a rate of increase in English proficiency by comparing the performance of students in different programs.
More students are achieving English proficiency now than before passage of Proposition 227.	The 8% increase in the number of students classified as proficient (FEP) between 2001 and 2002 mirrors the "reclassification rate" from the pre-227 era. A CELDT score of FEP is one of four criteria for reclassification of EL. The official CA Department of Education Language Census reclassification (redesignation) rate has fluctuated between six and nine percent for the past eight years. Based on the increases in FEP from the CELDT statistics alone, we cannot draw the conclusion that more students are achieving English proficiency than before passage of Proposition 227 in 1998.
CEDLT data prove that English immersion "works better" than bilingual education.	"English immersion" is a term that refers to a broad array of programs for EL. The "layman's" definition of English immersion is not the same as the definition of experts in second-language and bilingual education of language immersion programs. Even "English Only" classrooms in California, if staffed by a bilingual teacher, may use some amount and type of instruction in students' native language.  Before passage of Proposition 227, 70% of English Learners were in English Only classrooms, and the public was not satisfied with their academic achievement. If EO and English immersion are synonymous, then EO had decades to "prove" itself to be "superior" before Proposition 227. This is a fallacious post hoc analysis of cause and effect on the academic performance of EL.
The CELDT data confirm the "success" of Proposition 227.	Proposition 227 is a multifaceted policy. Judgments about the impact of this law or any policy on EL English language proficiency require careful analysis over the long term. Certainly, newspaper reporters who lack expertise in the areas of language acquisition, educational tests and measurements, and policy are not qualified to make such judgments. It is clearly incorrect to represent that one year of language assessment data is the basis for sweeping judgments about the impact of a five-year old law.
School districts should consider "trashing" their bilingual programs based on the CELDT results.	The CELDT data does not indicate that bilingual programs are inferior or that students' rate of English acquisition is slower in bilingual programs. If fact, since the descriptive data from the CELDT show that there was a 530% increase in the numbers of EL who scored in the advanced levels of English proficiency, we can conclude that students in bilingual programs are learning English. The CELDT data do not support the conclusion that language minority students will benefit  if school districts dismantle well-implemented bilingual programs.

Let us begin with the headline of Nanette Asimov"s  (2003, March 26) article: "English-only students do better on state test/Number of proficient speakers tripled after Prop. 227 passed." We pose two questions to test the truth of the statements contained in this headline. The first question is this: Are English-only students doing better on the CELDT? This is, of course, a question that requires a qualitative judgment based on a definition of what is meant by "doing better." It is also a comparative judgment, because we must compare English-only students to some other students to determine whether they are "doing better" than another group.  Asimov compares English-only (EO) students to bilingual education (BE) students based on the percentages of students in each group to reach the Early Advanced and Advanced levels in 2002 as compared to the percentages in each group in 2001 to score at the EA & A levels. 

Asimov uses these figures from the CELDT data: 

Bilingual Education students scoring at EA or A in 2001=3% and in 2002=16%. Difference =13%

English-only students scoring at EA or A in 2001=9% and in 2002=32%. Difference=23%

Based on these quantitative comparisons that she converts into "percentage points," Asimov concludes that EO students are  "doing better" and that BE students are learning English "at a slower rate if they were enrolled in bilingual programs." In other words, she draws a qualitative conclusion (better and faster) based on a quantitative (more) calculation. The analysis in Asimov's article is based on an incorrect definition of the CELDT test makers and therefore, the legal definition of "proficient" and  an invalid use of statistics to make comparisons of programs. 

The problem in this presentation of statistics becomes obvious when we look at the same set of numbers differently. If we ask the question of who is "doing better" from a different perspective we get a different answer. Based on these figures, we can see that the number of Bilingual Education students scoring at EA or A increased by 530% in one year. In comparison, the number of English-only students scoring EA or A increased by 350%. We can conclude that the increase in the number of advanced level scorers in bilingual programs was 66% greater than in English-only programs. Can we therefore conclude that BE students are "doing better" on the state test than are EO students?

The students who are actually enrolled in a bilingual program or an "English-only" program are in these different programs in part because their language proficiency is different. Students who are enrolled in bilingual education are usually in the bilingual program because they have lower proficiency in English. We cannot conclude that they have lower English proficiency because they are in a bilingual program. These statistics do not establish a cause and effect relationship. Furthermore, we cannot conclude from this data that the majority of students in bilingual programs are not in fact increasing their English proficiency at the same rate as students in English-only programs. 

Are They Really Learning Faster?

Edward De Avila (1997), one of the designers of the CELDT exam, states the following: 

"It cannot be assumed that all students will learn at the same rate or to the same extent. In large measure, extent of growth is limited by how far along the student is on the learning curve when s/he enters or begins a program."

"It is exactly because possible growth is a direct function of where along the program/measurement continuum a student begins that it is essential to establish that point before setting 'expected growth.' It would be foolish to expect the same growth for all students regardless of entry point, program type, quality or effectiveness."

Allow me to draw an analogy to illustrate this important point. Suppose that we do a “study” of runners to compare how many cross the finish line of a one-mile track in a fixed amount of time, such as ten minutes. We define different groups of runners for comparison purposes. Most, but not all of Group A runners will start the timed run at the .0 mile point. Some, and probably most, of Group B will start at the .5 mile point, while others start the timed run at the .75 mile point.  They then run as fast as they can for 10 minutes. At the end of the 10 minutes, we count how many in Group A crossed the one-mile mark and how many in Group B did so. We then compare the percentage of runners in each group who crossed the finish line in the required 10 minutes. Based on the percentages of each group to cross the finish line, we conclude that Group B must be faster runners because more of them reached the one-mile point in a 10-minute run for the finish line. Furthermore, we conclude that this “superior” performance must be due to the fact that the runners in Group B eat more Wheaties for breakfast than do the runners in Group A.

Certainly,  this manipulation of statistics will please the makers of Wheaties. However, such a paradigm does not truly offer us any insights into differences in the performance of the two groups, let alone the effects of eating Wheaties for breakfast.

This is what ill-informed and misguided reporters are doing with the California English Language Development Test (CELDT) data. James Crawford pointed out the dramatic differences in the levels of English proficiency among students enrolled in bilingual programs and students in English-only programs in last year’s 2001 CELDT data. In the reports of disaggregated data by program and English proficiency level, the percentage of ELL at the beginner level on the CELDT in bilingual programs ranged from 32.9% in 12^th grade to a maximum of 70.9% in 9^th grade and around 50% in grades K-5 as compared to a maximum of 32.8% at beginner CELDT level among EL in SDAIE programs. These “comparisons” of programs mask the vast differences in “starting points” for the runners in the race to “early advanced” and “advanced” levels on the test.

Here is another analogy to illustrate the point. If we were to give a proficiency test in French to high school students enrolled in French 1 and compare their proficiency in the language with students enrolled in French 2 or French 3, what would the data tell us? Predictably, they would indicate that the French 1 students have lower proficiency. Could we say, based on these data, that the French 3 program is "better than" the French 1 program because the students in French 3 are more proficient than those in French 1? Could we say that the students in French 2 are learning French faster than those in French 1 because their scores are higher on tests of French language proficiency? Can we conclude that French 2 is a more effective program because more students in French 2 are ready to move on to French 3 from this group than from the group of students in French 1?

Are They Proficient Yet?

It is important for reporters, the general public, and especially policy makers to understand the true definition of the term "proficient" as defined by the test and by the state of California. Not all students who score at the early advanced or advanced levels are proficient. Only some of them are. They are only classified as proficient if they also score at the intermediate level in reading and writing English. This is not a trivial point. In fact, it is at the heart of the challenges we face in educating English language learners. Unless these students are learning oral English AND advancing in their literacy and content learning, they are falling further and further behind academically. Looking only at growth in oral proficiency distorts the picture. 

This brings us to question two prompted by Nanette Asimov's headline in the San Francisco Chronicle: "English-only students do better on state test/Number of proficient speakers tripled after Prop. 227 passed." The question is two-fold because there is an embedded inference in part two of the headline: Did the number of "proficient speakers" triple after 1998? And ,can we compare this number of "proficient speakers" to another number of "proficient speakers" from before 1998?

A way to clarify the issues posed by these questions is to use a concept that is easily understood by laymen and experts alike: A "passing" score. We must be a bit hesitant about implying that the CELDT has a passing score, since it is not administered to students to determine whether they pass or fail in English. However, the CELDT has very clearly defined criteria for the category Fluent English Proficient (FEP). The criteria require a student score at certain minimum levels on all three subtests of the CELDT and overall. Any one criteria, such as the overall score in a prescribed range, is "necessary but not sufficient." All conditions must be met for a student to be deemed "proficient." 

A possible source of the erroneous definition of "proficient" is a graph included as an attachment to the official California Department of Education's press release showing 11% of students at the Early Advanced or Advanced levels on the CELDT in 2001 as compared to 32% in 2002 as meaning that these students are now classified as Fluent English Proficient (FEP). This is a difference of 21%.  Several reporters mistakenly took this figure to represent the difference in the percentage of ELs classified as "proficient" between 2001 and 2002.

Nanette Asimov (2003, March 26) states incorrectly in the headline of her article, and throughout the statistical analysis she presents, that the number of "proficient" students in California has tripled. This is a serious error. It is in fact a 292% error. It is equivalent to reporting that there are 300% more students who passed a state test this year than last, when in fact only 8% more passed the test. Asimov then goes on to imply that this alleged pass rate is greater than it was before. She does this by juxta-positioning "tripled" and "after Prop. 227." In fact, as I explain below, there is no evidence to claim that a higher percentage of EL students are reaching the Proficient level (i.e., "passing" the test) after passage of Proposition 1998 than before passage of the ballot initiative. We must distinguish fact from fiction.

Based on the data supplied by the state, we cannot say that these students are growing in their academic proficiency any more rapidly than they were before Proposition 227 or for the five years previous to passage of the law. School districts have traditionally reported the numbers of EL reclassified as Fluent English Proficient each academic year as part of the state's Language Census Report. Districts were not required to report language assessment scores, which would not have been comparable across seven different tests in any case. The only statistic we have from the pre-227 era that is at all comparable is the redesignation rate, which has hovered between six to eight percent since before 1998, with some fluctuation from year to year. In the 1997-98 academic year (pre-227) the redesignation rate was 7.0. In 2001-02 (four years post-227) it was 7.8. In fact, it appears that the process of English language and literacy development is taking as long as it ever has. The CELDT data suggest that the redesignation rate remains virtually unchanged after five years of implementation of Proposition 227.

Only 8% Growth in Fluent English Proficient

Let us examine the California Department of Education's CELDT Attachment A versus Attachment C. This is the "Good News" as opposed to the "Not So Good News" about overall academic achievement for EL in California. It is human nature to give in to the temptation to focus on charts and graphs that show large numbers and greater gains than on charts and numbers that are not so dramatic. However, we cannot accurately interpret statistics unless we understand what they mean. Keep in mind that the listening/speaking portion of the CELDT makes up half of the total points in the overall score. Attachment A shows results that reflect this weighing of the components of language. This graph also exhibits data from students who were tested in both 2001 and 2002. It therefore shows data on only a subset of all students who tested at the advanced levels in 2001, since some unstated percentage of the total number of students who tested EA or A in 2001 were not tested again in 2002 because they left the "pool" of EL for a variety of reasons. Attachment C, on the other hand, reflects students' growth in reaching fluent English proficient (FEP) according to the definition of FEP given by the test makers. This graph includes this statement: "The CELDT is one criteria used in the Reclassification of EL to FEP. Consult the Reclassification Guidelines in Section IV." This chart tells us the percentage of students who now meet one of the four criteria for reclassification in 2002 as opposed to 2001. In other words, these students meet the criteria for intermediate level in the other half of the test, in the literacy skills of reading and writing. The difference is eight percent.

Statistics can be manipulated and portrayed to suit a particular purpose. Even accurate and reliable statistics can be difficult to interpret when different groups with different, but sometime overlapping, compositions are depicted. This leads to the appearance of contradictions and discrepancies within the data presented by the CDE in their year to year comparisons of students reaching the "proficient" level. Attachment C shows a table that gives a different picture of the comparisons of percentages of students who score at the "proficient" level on the CELDT (early advanced or advanced) between 2001 and 2002. According to this table, which is broken down by grade spans tested, there is an 8% difference in all grades between the numbers of students who scored as proficient in these two years. Attachment A, the more frequently cited graph, displays only the percentages of students who attained the “early advanced” (EA) level or higher. Scoring in the EA level does not equate to proficiency, since proficiency is defined as a score of EA or advanced and intermediate or better in all subtests: listening/speaking, reading and writing. Nor, as explained above, does this represent the total number of students in 2001 who scored in the upper ranges of proficiency. Therefore, this statistic suggests that although students may be gaining in oral proficiency, they are not be achieving academic English proficiency at a comparable rate. The 8% figure mirrors the average “redesignation rate” that has held constant over the years since before and after passage of Proposition 227. For an excellent discussion of redesignation, see Linquanti, 2001.

Comparability of Year-One & Year Two Data

In April 2002, the first year of the release of any CELDT data, scores were reported by the CA Department of Education (CDE) in the Q & A for the Media web page based on three categories: Probable Fluent English Proficient, Possible Fluent English Proficient, and Probable English Learner. There was a graph presented that listed the criteria for each category. This press release specified that the Probable Fluent English Proficient had an overall score of Early Advanced or higher AND an intermediate or higher in all three skill area scores: Listening/speaking, reading and writing. The Possible Fluent English Proficient had an overall score in the "upper end of intermediate" and also met other criteria. Probable English learners had an overall score below Early Advanced OR "Student's overall score is Early Advanced or higher, but one or more of the skill area scores is below intermediate."

These categories were abandoned in this year's score reporting. Instead, there are two "subcategories" of Fluent English Proficient and English Learner: The two new criteria for Fluent English Proficient are: Those who met last year’s Probable Fluent English Proficient criteria and those meeting the last year's Possible Fluent English Proficient. The criteria for 2002 for English Learner is the same as 2001 Probable English Learner. 

Given these changes in criteria it is very difficult to make heads or tails out of this test data. According to the CDE's new categories, many students who are at the Early Advanced and Advanced levels are still classified as English Learners, while some students who have not scored at these levels are considered Fluent English Proficient. The press and the advocates of English-only programs are zeroing in on the impressive graph in Attachment A of this year's press release, which according to this categorization system is fairly meaningless since the data do not distinguish between students classified as  FEP and those who are still classified as EL or LEP.

It is also important to note that 32% of the students tested are classified as Early Advanced or Advanced based on their overall score.  If we think of the CELDT as being divided into three broad categories--beginner, intermediate and advanced--we would logically predict  that about 33% of the students tend to fall into one of three categories on the scale. However, Attachment B indicates that only 10% of the students tested in 2002 are beginners, 56% are intermediate and 32% are advanced. This could mean that there are fewer students in the schools who are new arrivals from other countries or that have not had prior exposure to English. This is important information about the characteristics of the EL population overall. There is some evidence that there has been a "recalibration" of the CELDT scoring during the second year of administration. To what extent this explains the alleged "growth" in scores is a question worth exploring further. Meanwhile, we can have fruitful discussions about the process of "standardizing" a form of testing that was originally never intended and not designed for that purpose. 

Any criteria for comparisons of students based on test scores must meet very specific criteria for internal and external validity. Without meeting these criteria, comparisons cannot be trusted to give us trustworthy information about students' growth in language proficiency. In order to do meaningful and valid comparisons of programs based on CELDT data,  we would have to have matched scores on individual students from 2001 and 2002 from the California Department of Education to derive an "increment of growth" on the CELDT for each one of them on a pre-test and post-test basis. This would require that we sort and select ELL students according to their 2001 level on the CELDT (pre-test) and their 2002 levels (post-test). We would then sort them into subgroups according to 2001 pre-test levels, allowing us to examine their individual post-test 2002 levels. We would also have to know what type of a program in which each student was enrolled during the 2001-2002 instructional period and sort them accordingly. We could only use scores for students who had been continually enrolled in only one type of program for the entire instructional period. Say for example, using this model we could derive a group of 100 pre-test Beginner level students in a bilingual program and compare them to 100 pre-test Beginner level students in an English-only program. We could then compare their post-test (2002) scores. This would be a valid statistical comparison for one year (and one year only) of language assessment data. However, it would not be an educationally sound comparison. 

One year of program performance does not give an accurate picture of a program's effectiveness over time.  In any case, this type of research is not possible at this point in time because the CDE does not release individual student’s test scores to the public. Furthermore, program placements for EL cannot be used as a basis for comparison of educational "treatments,"  which must be clearly defined and delineated according to strict criteria, with confounding factors that make comparison groups "unequal" fully controlled both pedagogically and statistically.

Is the CELDT Worth It?

Is the investment in the CELDT worth it? It will not be unless we are prepared to do what needs to be done to respond appropriately to what the data really tell us. Are we ready to wrestle with the  touch issues surrounding the education of language minority students? Or would much rather continue to use and misuse state-funded testing to fuel the flames of anti-bilingual education politics?

The concept of the CELDT for annual language proficiency testing is problematic. Experts in education tests and measurements, and knowledgeable second-language educators have a large data base to observe the L2 learning curve from several million administrations of the Language Assessment Scales (LAS). Over the years since it was widely used for identification and program placement in several states, including Texas. The learning curve is far from being neatly incremental. According to Edward De Avila's report (1997) on the LAS data, out of the 100 possible points and possible gains, 55 points are gained at levels 1 and 2 and 65 points are still high level 2. It seems as though it is possible to report gains in "levels" rather arbitrarily, depending on how one decides to distribute the "points." So what is an "average" rate of "growth" in language proficiency.

Another problem is the means of measuring listening/speaking as compared to (or disaggregated from) skills in reading and writing. The LAS data show us that these sub-skills show gains at very different rates. The sub-skills scores only begin to converge after a level 4 is reached, which makes sense when we think about the relationship between oral proficiency and literacy development.

The CELDT is administered in the summer and fall, but teachers don't even get the results until spring, after instructing their group of students for almost the entire academic year. Unless teachers are well prepared in understanding the relationship between language proficiency and literacy development and are able to apply appropriate teaching strategies, select and adapt instructional materials, and together in their school settings, they cannot use CELDT results to plan a coherent curriculum for students with different levels of proficiency and literacy achievement. Many teachers are knowledgeable about how to plan a curriculum and lessons for EL, but most base their planning on false assumptions and quick-fix strategies based on the surface features of their students' English proficiency.  Teachers are often looking narrowly at their grade level responsibilities and do not understand much about what happens to students before they get to them or after they leave to go on to the next grade. Therefore, there is a tendency to blame the primary grade teachers for ELs' lack of reading proficiency (usually comprehension), rather than having a realistic idea of where the students are likely to be in L2 literacy development at any given grade level (also depending on their age of arrival, time in the program, etc.). Consequently, EL are shuffled from grade to grade, or retained in grade, without a clearly articulated sequential curriculum that addresses their academic needs and abilities appropriately in regard to language learning as well as literacy and content knowledge. 

Public Policy Implications

There are no simple solutions to the complex and multifaceted task of teaching students who are learning English language, literacy and academic content. If these test scores tell us anything, it is that there is no one-size-fits-all program for California's linguistically diverse students. The type of uninformed reporting and political rhetoric surrounding the release of the CELDT scores are not helpful to the tens of thousands of students who have the right to receive appropriate and effective instructional programs to ensure their academic growth. 

The findings of the extensive study of the impact of Proposition commissioned by the California Department of Education through WestED and the American Institutes for Research (AIR). This study, published in 2002, found no significant differences between the progress of students in bilingual education programs and English-only programs in their rate of learning English. It is about time that policy makers, the public, and the press recognize and acknowledge that the debate over English-only versus bilingual education is purely political. There is a large body of methodologically sound research that confirms the effectiveness of well-designed and well-implemented bilingual programs in supporting the academic achievement of ELL. Comparisons between the "effectiveness" of bilingual education and English-only based on test data are politically motivated. See also Wright 2002 on the effects of high stakes testing and the results of the SAT-9 for ELL in California and Grissom 2004 for a discussion of the exit rates for ELL following passage of Proposition 227. 

 Bilingual programs are designed to support and enhance English language learning while also building literacy and content area knowledge in the native language for precisely those students who are not prepared to succeed in all-English classrooms. The parents of these students have chosen to place them in bilingual programs so that they are not left behind in scholastic achievement while they are learning basic English. They know that bilingual instruction provides many advantages for their sons’ and daughters’ long-term academic achievement. The will of California’s voters in 1998 was to pass Proposition 227 with its allowances for parents to request, and educators to grant, waivers for bilingual instruction. There is no reason for the state or federal government to curtail the rights of parents who choose bilingual education for their children or to deny any parents or communities access to these valuable and effective programs of instruction.

Vindication?

In his article in the Arizona Republic ((April 14, 2003) Doug MacEachern uses an inflammatory Iraqi war analogy, comparing bilingual educators to Saddam Hussein's Minister of Information who is in denial about the presence of "imperialist American tanks at the airport" in Bagdad, just as they are supposedly denying the impending victory of the "special magic of English immersion."

Why is there such hostility toward bilingual educators? Why do reporters feel the need to use war analogies to make a case for English immersion? Wars are struggles for power and control. The assault on bilingual education is about the struggle for social dominance between the monolingual/monocultural majority and a growing bilingual/bicultural minority. We must abandon this concept and begin to work together to formulate and implement sound public policies that enhance teaching and  learning for our language minority population.

Ron Unz is quoted in the San Francisco Chronicle (Asimov, 2003, March 26) as saying, with "...a bit of I-told-you-so in his voice..." that "It is sort of astonishing that if they teach you English you learn it faster than if they don't." The implication is, of course, that bilingual programs do not teach English. There is ample evidence in the CELDT data and the policy statements surrounding its use as a means of compliance with the federal No Child Left Behind legislation that bilingual educators have been vindicated. We can say, "We told you so!" when Proposition 227 in California and then Proposition 203 in Arizona were passed. We told them that one year of intensive English was not enough to prepare students for mainstream classrooms. Sure enough, the CELDT data confirms this, especially if we look at literacy development as well as listening and speaking skills. We told them that students would grow rapidly in oral language skills (Basic Interpersonal Communication Skills) but that proficiency in academic English (Cognitive Academic Language Proficiency) would take much longer, usually five to seven years. Sure enough, at the rate of growth toward full proficiency we see in one year of comparative test data, it appears that reaching FEP and reclassification will indeed take from five to seven years. (At 8% per year, 50% of the population will reach full proficiency in 6.25 years.) In fact, this projection is born out by data in Attachment C, indicating that 47% of students met the criteria for FEP on the CELDT within the K-5 grade span between 2001-2002. The state's definition of satisfactory progress is in itself a vindication of bilingual educators. 

Certainly, Ron Unz has nothing to crow about in the CELDT data, although it will be difficult to convince him of that fact. Keep in mind the two principal goals of Proposition 227: To mainsteam EL after only one year of intensive English instruction and to eliminate bilingual education. These goals were based on the belief that students would make rapid advances in English if they were denied the use of their native language as a medium of instruction in the classroom. Prior to passage of Prop. 227, 30% of the state's population of English language learners were enrolled in bilingual programs. Now approximately 12% are in bilingual education classrooms. Bilingual education has not been eliminated. The CELDT data confirm that those programs that survived the Prop. 227 are performing as well, if not better than, English immersion programs. A recognition and acknowledgement of this fact could potentially transform the "debate" over bilingual education to a legitimate and productive discussion of how different programs can better serve our language minority student population with their diverse goals, needs, and values within the context of available human and financial resources. For instance, what reasonable justification could there be for denying or restricting federal funding to bilingual education when these programs have been demonstrated to be highly effective in promoting English language and literacy learning for at-risk students? 

The Truth Matters

We must hold the media accountable to the education community and to the public to enforce standards of accuracy and fairness in reporting the results of state tests. It does not matter whether news reports are biased because they are inaccurate, or they are inaccurate because they are biased. Journalists must use particular care in reporting the facts about state test data and statistics.  It is unethical for reporters who lack expertise to pretend to analyze complex data and draw sweeping conclusions about pedagogical and policy issues, especially when these conclusions are misrepresented as fact. Editors should also reject opinion pieces with inaccurate, exaggerated or misrepresented state testing program statistics. Reporting the news based on a political "agenda" is biased journalism. Biased journalism contributes to public prejudice, which in turn serves to legitimize discrimination against less powerful minority groups and members of society. Discrimination hurts us all. This is a violation of the trust we place as a society in the media. This trust is especially important in regard to coverage of educational testing for the purposes of decision making in formulating laws and regulations that affect the lives of thousands of public school students. We must distinguish between information, misinformation and "disinformation" in press coverage of education issues and hold the media responsible for telling us the truth.

Our greatest hope is that the attacks on bilingual education and bilingual educators will soon end. There is much work to be done to improve programs and classroom instruction for language minority students. Collaboration among educators, policy makers, and the voting public is key to getting the job done. We must continue to inform all sectors of our society about the benefits and advantages of well-designed and well-implemented bilingual and second-language programs that will help all our students reach their full human potential.

References: 

De Avila, E. (1997, November). Setting expected gains for non and limited English proficient students. NCBE Resource Collection Series No. 8. Arlington, VA: National Clearinghouse for Bilingual Education.  

Quezada, M. S. (2003). Are students learning English? Language Magazine, 2(8), 19-23.

Resources for Teaching Language Minority Students:

Dr. Mora's Home Page	MoraModules Menu
Dr. Mora's Bulletin Board	RICA Study Guide
PLC 915 Syllabus	ED 516 Syllabus
Thematic Planning for ELLs	4X4 Activities & Strategies
Sheltered English Immersion	Model 4X4 Unit
4X4 Thematic Planning Guidelines & Resources	5-Step Lesson Planning for ELLs
L2 Teaching Methods	Theoretical Foundations of Bilingual Education
The Why's & How's of CLAD Teaching	A Literacy Framework
Debunking English Only	Critical Thinking in the Bilingual Education Debate
A Roadmap to Effective ELD Instruction	L2 Reading Instruction

 

This page was last updated on 11/14/06