How Good Is SFUSD At Teaching English Learners?
Why its claims of 80%-85% proficiency by 5th grade are misleading
In part 1 of this series of posts on English learners, we looked at the number and mix of English learners in SFUSD. Last week, we saw how district-wide SBAC results are related to the number of novice English learners. In this final post in the series, we’ll examine how good SFUSD is at making its English learners fluent.
Measuring How Good Districts Are at Teaching English
When students start kindergarten, those whose primary or home language is not English take the Initial ELPAC to assess their level of English fluency. Those who don’t test as fluent (IFEP or “Initial Fluent English Proficient”) are designated as English learners. At the end of every subsequent year, they are supposed to take the Summative ELPAC to measure their current level of English fluency. If they test high enough, they are eligible to be reclassified as RFEP (“Reclassified Fluent English Proficient”).
Students may be supposed to take the Summative ELPAC every year but do they? Across California, 96.7% of English learners record a score on the Summative ELPAC. In San Francisco, only 86.0% do. Out of the 100 districts with the most English learners, San Francisco ranked 96th. This is yet another example at SFUSD being very bad at the administrative parts of being a school district.
The Summative ELPAC is fine as a measure of the progress of an individual student over time but it is useless as a measure of how good a district is at teaching students English. You might think you could take the average score in one year, compare it to the average score in the previous year, and use the change in score as the measure of student learning. But the Summative ELPAC is only taken by students who are currently designated as English Learners and that population churns from year to year. The very best students from last year won’t be taking the test this year because they will have been reclassified as fluent. In their place will come newcomers to the district, most of whom will probably be novice learners who will score poorly on the test in their first year. Even if every student’s English has improved significantly over the year, the district’s average will not show the same improvement because a bunch of high scorers will have been replaced by a bunch of low scorers.
The Reclassification Process
For a student to be reclassified as RFEP requires ALL of the following:
a high enough score on the Summative ELPAC test. Students get a scaled score on a range of 1150-1950 which is then mapped onto one of four “performance levels”. Students who score at Overall Level 4 are eligible to be considered for reclassification.
teacher evaluations, including a review of the pupil’s curriculum mastery;
parental opinion and consultation
“Comparison of the performance of the pupil in basic skills against an empirically established range of performance in basic skills based upon the performance of English proficient pupils of the same age”.
Notice that criteria 2-4 are subjective and locally determined. It is therefore possible that the practical standards for reclassification differ from one district to another.
Rates of Reclassification by language
Consider the universe of students who speak a language other than English at home. Initially, only a small proportion of them (those who tested as IFEP) will be fluent and the rest will be English learners. Over time, as the students learn English, many of them will be reclassified as fluent (RFEP) and the number of English learners will decline. The proportion that is fluent (either IFEP or RFEP) is a measure of how much progress a district has made with its students. In an ideal world, everyone would become fluent and this proportion will hit 100%. In practice, new arrivals means there will always be English learners.
Across California, as figure 1 (below) shows, the percentage that is fluent varies by home language:
As a group, Mandarin speakers have the highest level of English proficiency in kindergarten with 32% being fluent. Speakers of Cantonese, Vietnamese, and Filipino start off less fluent than Mandarin speakers but gradually catch them up. By 8th grade, at least 75% of speakers of all these four languages are classified as fluent.
Spanish speakers start out with a lower level of English fluency (only about 10% are fluent in Kindergarten). While they make steady progress through the grades, only 60% reach fluency by 8th grade and 76% by 12th grade.
Arabic speakers start out with a slightly higher rate of English fluency (16%) than Spanish speakers but gradually fall behind in the later grades. This probably reflects the fact that a greater proportion of Arabic speakers arrive after kindergarten: of all those who took the Initial ELPAC in 2021-22, 67% of Spanish speakers but only 53% of Arabic speakers were in kindergarten. As the other Arabic speakers arrive in grades 1-8, they increase the number of English learners, thereby decreasing the percentage who are fluent.
SFUSD’s Reclassification Record
Speakers of Spanish, Cantonese, and Mandarin get reclassified in SFUSD at slower rates than they do in other districts. While the Cantonese speakers in SFUSD catch up with their peers elsewhere in the state by the end of middle school, the Spanish and Mandarin speakers in SFUSD continue to lag. Figure 2 shows that not even 50% of Spanish speakers are fluent by the end of middle school.
The flattening of the red line after 8th grade in SFUSD is easily explained: a huge number of new novice English learners arrive in SFUSD in 9th and 10th grade. Harder to explain is why SFUSD’s reclassification rate lags the rest of the state through 8th grade.
Figure 3 shows that the reclassification rate for Cantonese speakers in SFUSD trails that of other districts in the elementary grades but just about catches up in middle school.
The third largest language group in SFUSD, the Mandarin speakers, trail their peers statewide and never catch up, as figure 4 shows.
Is Reclassification Getting Delayed?
Districts receive more money for each English Learner so, at least in theory, districts have a financial incentive not to reclassify students. Districts also have the means to delay reclassification because three of the four reclassification criteria are locally determined. If that were happening in practice, many students who scored at Level 4 on the Summative ELPAC (the level necessary to be considered for reclassification) would end up not being reclassified and having to retake the Summative ELPAC the following year at which point they would presumably score at Level 4 again. A district would then end up with a growing number of students scoring at Level 4 over time. Is that happening in San Francisco?
Probably not for Spanish speakers. Figure 5 shows that the percentage who score at Level 4 in SFUSD tracks the percentage who score at Level 4 elsewhere in the state. It’s a bit higher than the state average in 3rd and 4th grades but that could be normal variation.
It may be happening with Cantonese speakers. Figure 6 below shows that the San Franciscans far outscored their peers elsewhere in the state in each of the elementary grades at the same time that the percentage who were reclassified as fluent was trailing their peers elsewhere in the state (see figure 3 above).
Reclassifying as Fluent
Instead of treating each language separately, Figure 7 shows the English language status of all SFUSD students by grade.
Some of the more eye-catching bits of this chart have nothing to do with learning English:
There are more kids in kindergarten than grades 1-5 because some kids enroll in Transitional Kindergarten which shows up in the data as spending two years in kindergarten, thereby boosting the kindergarten numbers.
SFUSD loses 15% of its students at the middle-school transition point.
SFUSD has historically always added students at the high-school transition point because there are not as many spaces available in private high schools as in private middle schools. The 9th and 10th grade classes are smaller than the 11th and 12th grade classes because of the pandemic.
Now let’s look at what Figure 7 has to say about English learners.
The Initially Fluent group is twice as big in grades KN-03 as in other years. The 3rd grade kids were in kindergarten in 2018-19, the first year that the ELPAC replaced the CELDT as the screening test. Other districts also saw a 50%+ increase in the number of Initially Fluent kids so we can confidently say that it’s easier to be fluent with the Initial ELPAC than the old CELDT.
It seems to be a matter of policy in San Francisco that no English learners will be reclassified before 3rd grade. Only 8 students were reclassified in 2nd grade, compared with 193 in 3rd grade.
The number who are reclassified in any one year varies hugely from grade to grade. Reclassifications peak in 5th grade, just before the middle school transition. In districts where elementary school goes through sixth grade, reclassifications peak in 6th grade. It seems to be generally true that reclassification is affected as much by organizational issues as individual student progress.
Reclassification Rates
Figure 8 below looks at the pace of reclassification in SFUSD compared with the rest of the state. SFUSD is slower to reclassify students in the early grades, catches up by 5th grade, moves ahead in 7th grade, and then falls behind again in high school. Given that San Francisco contains so many Cantonese speakers who, as we’ve seen, learn English faster than Spanish speakers, the slow pace of reclassifications in elementary school is either a policy decision by SFUSD or an indication that reclassifications of Spanish speakers are way behind the statewide trend. As we saw in the first post of this series, SFUSD receives a lot of new Spanish-speaking English learners in 9th and 10th grades so it’s less surprising to see SFUSD fall behind the statewide average in high school.
Long-Term English Learners
The CDE requires districts to track how long a student has been an English Learner. Any English Learner who has been enrolled in a U.S. school for six or more years, and has remained at the same English language proficiency level for two or more consecutive prior years, is considered a Long-Term English Learner (LTEL). Students who have been in school for 4-5 years and have scored at the Intermediate Level or below on the ELPAC are considered to be “At Risk” of becoming Long-Term English Learners. Those students who score high enough on the ELPAC but are not year reclassified are considered not at risk of becoming LTELs. Figure 9 below shows how many English Learners in SFUSD fall into each category.
What does this chart tell us?
In grades KN-02, everyone has been an English learner for 0-3 years. In grades 3-5, we see the first reclassifications.
There’s a drop in the total number of Ever-ELs from grade 5 to grade 6 because SFUSD loses students at that transition point.
Grade 6 is when students are first eligible to be classified as Long-Term English Learners. It’s noticeable that the number of LTELs in grade 6 was far higher than the number who were deemed to be at risk in grade 5. It is less likely that the grade 6 students were much worse at English than the grade 5 students and more likely that the term “at risk” doesn’t really measure the risk very well.
Although the number of LTELs does decline after 6th grade, there are still over 300 LTELs by the end of 12th grade.
In the 10th-12th grades, the number of students who are or were once English learners is well over 2000 per grade. In grades K-3, it averages a little under 1500. The percentage of 1st graders who are English learners declined from 45.6% to 37.8% in the seven years between 2014-15 and 2021-221.
Even if we ignore all the students who have been English learners for fewer than 6 years and just consider, the percentage of learners who have been reclassified as fluent is only 56% in 6th grade. That’s a far cry from “more than 4-out-of-5 English Learners are proficient in English by the 5th grade”.
Impact of Language Immersion Programs
Can the slow pace of reclassification in SFUSD be explained by the large number of students enrolled in bilingual and dual-language immersion programs? The official data from the CDE does not track results by language program so it is impossible to tell from the data. It’s not even possible to back it out from the individual school data because nearly all schools that have a language program also have a general education program.
Here, from the 2022-23 PEEF Budget Proposal, is what SFUSD says about its language programs:
In partnership with Stanford University, SFUSD conducted a longitudinal study that included approximately 18,000 EL students who entered kindergarten in SFUSD from 2001 to 2009, including 12,000 in English Plus, 4,000 in Bilingual Maintenance (most similar to the current Biliteracy Pathway), and 2,000 in Dual Immersion. The study found that more than 4-out-of-5 English Learners are proficient in English by the 5th grade, regardless of English Learner Pathway. Students in the English Plus Pathway make faster progress toward English proficiency in the early grades of elementary school, but students in the Bilingual and Dual Immersion Pathways catch up with their peers around the 5th grade. By the end of elementary school, about 80-85% of students are English proficient, regardless of the initial pathway.
If the low reclassification rates of SFUSD students were due to so many of them being in language pathways, we would expect them to catch up at some point and they never do.
I was particularly struck by the claim that “more than 4-out-of-5” or “80-85%” of English learners are proficient by the end of 5th grade because it didn’t seem to jibe with everything we’ve seen above. As we saw in figure 2, the proportion of Spanish speakers in SFUSD who are classified as fluent by the end of 5th grade is no more than 50%. As we saw in figure 9, even if we ignore every learner who has been in the district for less than six years, the percentage who are RFEP is only 56% by sixth grade. So, is 50% or 80% the right answer? Have things gotten much worse since the study was done?
I tracked down this joint Stanford/SFUSD presentation about the results of the Stanford study. It turns out they are both accurate statements of very different things. 50% of Spanish speakers in SFUSD have been classified as fluent by the end of 5th grade. 80-85% of all English learners (i.e. not just Spanish speaking) who entered kindergarten in SFUSD (i.e. excluding those who started in grades 1-5) scored high enough on the language test to be eligible for reclassification (even though many of them were not reclassified as fluent because they didn’t meet one of the other criteria) by the end of 5th grade. The Stanford study doesn’t really disagree with the 2022 numbers: it puts the percentage of Latino EL kindergarten entrants in SFUSD who were reclassified by the end of 5th grade at 45-63%, depending on the instruction model.
It is a shame that SFUSD does not use a standard definition of what it means to be “proficient” in English.
I’m comparing 1st graders because SFUSD leaves too many kindergartners in TBD status and I’m using 2014-15 as my starting point because that’s the earliest year available on Dataquest.
There aren’t as many private high schools as private middle schools but one of the reasons is possibly that SFUSD high schools are generally better than SFUSD middle schools. Looking at the evidence and my own experience this is probably true. There could be other reasons as well. Maybe it is harder to open a high school or they are less lucrative but those latter seem unlikely.