As a Level 4 district that is trying to reform the educational experience of our students, New Bedford Public Schools needs to evaluate and adopt research-based program models that address our needs. Two-way dual language immersion has a track record of dramatically improving student achievement for both English Language Learners and native-English speakers. The Esperanza School of Language and Culture will implement such a program and provide an enriching educational experience for our students.
Starting in 1985, Virginia Collier and Wayne Thomas have collected data regarding the academic achievement of English Language Learners both during enrollment in special programs as well as after they exit from these programs. The data in this graph includes over 6.2 million student records*. In order to be included in this data set, students needed to meet three requirements. They entered school in kindergarten, had no English proficiency when they enrolled, and were eligible for free or reduced-price lunch. These students were in urban, suburban, and rural schools throughout the country. With over 6.2 million student records from different parts of the US, this data has a high degree of generalizability.
Grade levels are presented along the x-axis of the graph. Along the y-axis, test scores are ranked by normal curve equivalents**. NCEs are comparable to percentiles in that the scores are distributed along a scale that goes from 1 to 99. The dashed line at the 50th NCE represents the average standardized reading test score for native-English speakers at a given grade level.
In order for English Language Learners to close the achievement gap, they must meet or exceed the average results of the native-English speakers at that grade level. Their native-English speaking peers are making progress every year. This means that ELLs need to make more than a year's progress in a year's time and this growth must be sustained over several years.
In terms of this graph, the results for a given program model must exceed the 50th NCE to show achievement gap closure. There are two program models which do exactly that – both are dual language immersion programs. The average results for students in two-way dual language immersion programs ended up at the 61st NCE. This means that, once students are in secondary school, the average ELL who attended a two-way dual language immersion program starting in kindergarten significantly outperformed the average native-English speaker on standardized tests in English reading.
Dual language immersion means that students will learn language and academic content in English and a partner language. The two-way component refers to student enrollment – half the students are dominant speakers of English, while the other half are dominant speakers of the partner language. At the Esperanza School, the partner language will be Spanish.
Original source of graph:
http://www.dlenm.org/Figures/Chapter%205/5.2-%20final.11:09.English%20Learners%20Long-Term%20K-12%5B1%5D%20copy.pdf
Collier, V. and Thomas, W. Educating English Learners for a Transformed World. Albuquerque, NM: Dual Language Education of New Mexico/Fuente Press, 2009.
*Each student record contains information about a specific student for one school year. (added 2-22-13)
**Normal curve equivalents are an equal-interval scale. Scores are distributed along the range from 1 to 99, just as with percentiles. However, a change in NCEs at any point in the scale represent an equal change in scores. For example, a rise from the 25th NCE to the 30th NCE represents the same change in scores as a rise from the 80th NCE to the 85th NCE. The growth of 5 NCEs in both cases represents the same increase in scores. This cannot be said for percentiles. Results at the 1st, 50th (mean), and 99th NCE are equivalent to the corresponding percentile. Because results are distributed differently, other NCEs are not equivalent to percentiles with the same number.
*Each student record contains information about a specific student for one school year. (added 2-22-13)
**Normal curve equivalents are an equal-interval scale. Scores are distributed along the range from 1 to 99, just as with percentiles. However, a change in NCEs at any point in the scale represent an equal change in scores. For example, a rise from the 25th NCE to the 30th NCE represents the same change in scores as a rise from the 80th NCE to the 85th NCE. The growth of 5 NCEs in both cases represents the same increase in scores. This cannot be said for percentiles. Results at the 1st, 50th (mean), and 99th NCE are equivalent to the corresponding percentile. Because results are distributed differently, other NCEs are not equivalent to percentiles with the same number.
