English-Language Learners

SFUSD has a higher percentage of novice English learners, especially Latino novices, than most districts

This is the first in a series of posts about English learners and how well they do in SFUSD compared to elsewhere in the state. Today’s post talks about how the number and type of English learners varies across the state.

Why this is important

Everyone has heard that race is a strong predictor of student achievement in California: Asian students on average do better than White students who do better than Latino students who do better than Black students. As figure 1 below shows, other demographic attributes can be just as powerful predictors of student achievement. The gap between students who are English learners and those who are fluent in English is as big as the gap between Black and White students¹. Moreover, while a school district can’t control the language spoken in the student’s home any more than it can control whether a child is white or black or whether the parents went to college or finished high school, it can influence whether and how quickly a student becomes fluent in English.

Figure 1: Average of those who scored Standard Met or Exceeded on 2022 SBAC Tests for select demographic subgroups. Disc size indicates number of students in subgroup.

Counting English Learners

All students who register in California public schools fill out a home language survey. Figure 2 below shows the results for California and for SFUSD. Across California, 60% of students are English only, compared to just 44% in SFUSD. Across California, among those students who speak a language other than English at home 76% speak Spanish, 2.9% Vietnamese and Mandarin, 1.8% Cantonese, 1.6% Filipino, 1.4% Arabic, 1.3% Korean, and 1.0% Punjabi. San Francisco, by contrast, has nearly as many Cantonese speakers as Spanish. Among SFUSD students who speak a language other than English, 39% speak Spanish, 36% speak Cantonese, 4.7% Mandarin, 3.3% Vietnamese, 2.5% Filipino, 2.0% Arabic, 1.9% Toishanese², 1.2% Japanese, and 1.1% Russian. Korean, at 0.8%, doesn’t break the 1% threshold.

Figure 2: Language spoken at home for students in SFUSD and all California public schools.

These numbers are different than the ones you commonly see. For example, the CDE’s “Facts about English Learners in California” says that 82% of English learners in California speak Spanish, 2.1% Vietnamese etc. All the numbers come from the same source: the difference is that I’m including students who have become fluent in English as well as their home language while the CDE includes only students who are currently designated as English learners. The problem with focusing just on those currently designated as English learners is that this is partially a reflection of how good or bad a district is at teaching kids English. Consider two hypothetical districts. In both district A and district B, 50% of kindergartners are English learners. In district A, all English learners become fluent at the end of 4th grade i.e. after 5 years. In district B, the English learners never become fluent. The official statistics will show that district A is 19% English learners (50% * 5/13) while district B is 50% English learners. The implication will be that district B has far more English learners to deal with but the reality is that both districts started out with the same number of learners but district B is just very bad at teaching them English.

To avoid distorting the results by including a measure of each district’s competence, I prefer to use a different measure: the EverEL rate i.e. the percentage who are currently English learners plus the percentage who used to be English learners but have since been reclassified as fluent in English.

Measuring Fluency

Speaking a language other than English at home doesn’t automatically make a child an English learner. Some such children show up for their first day of school already fluent in English as well as their home language. Within 30 days of a student’s first enrollment in a California public school, each student with a non-English home language is supposed to take the Initial ELPAC (English Language Proficiency Assessment for California). Some, when first tested, will prove to be fluent in English and will be designed “Initial Fluent English Proficient” (IFEP). The rest will be “English Learners”. They are then retested annually (with the “Summative ELPAC” - students take the Initial ELPAC only once) until they demonstrate fluency, at which point they become “Reclassified Fluent English Proficient” (RFEP).

English fluency is not a binary thing. Students who take the Initial ELPAC are given a score and this score can be mapped to one of three fluency levels in the same way that SBAC scores are mapped to one of four achievement levels (Standard Exceeded; Standard Met; Standard Nearly Met; Standard Not Met). The three fluency levels are IFEP, Intermediate Learner, and Novice Learner.

English proficiency varies by home language

The level of English proficiency varies by home language. Among kindergartners in SFUSD, 30% of Cantonese-speaking students but only 10% of Spanish-speaking students score as IFEP (Initial Fluent English Proficient). Even among those who aren’t fluent, and are classified as English learners, Cantonese speakers are far more likely than Spanish speakers to score as Intermediate level as opposed to Novice level (37% of Cantonese-speaking English learners are Intermediate compared to 19% of Spanish-speaking English learners). This is not a San Francisco phenomenon: SFUSD’s Cantonese speakers are a little bit better than the state average for Cantonese speakers but SFUSD’s Spanish speakers almost exactly match the state average for Spanish speakers.

Figure 3: Percentage of kindergarten students at various English language proficiency levels in 2022 Initial ELPAC tests

Think about what this means. Even if the district is equally good at teaching Cantonese and Spanish speaking students, the Cantonese speakers will score higher on standardized tests because their starting level of English proficiency is higher.

English Learners Start at Different Ages

The Initial ELPAC is supposed to be administered within 30 days of a student’s first enrollment in a California public school. The majority of test takers take it in kindergarten but roughly one-third take it in other grades, either because they are new immigrants to the county or because they are new to California or just new to California public schools.

This too can vary by language. Languages with fewer new immigrants will see a higher percentage of students enroll at kindergarten; languages spoken by a lot of new immigrants will see a lower percentage enroll at kindergarten and higher percentages enroll at other grades. Across California, 85% of Hmong speakers enroll in kindergarten; 72% of Vietnamese; 70% of Cantonese, 67% of Spanish, 64% of Mandarin, but only 53% of Arabic and 44% of Filipino.

In SFUSD, the percentage of Cantonese speakers who take the Initial ELPAC in kindergarten is 75%, a little higher than the state average of 70%. On the other hand, the percentage of Spanish speakers who take the Initial ELPAC in kindergarten is only 46%, way lower than the state average of 67%. Another 17% of Spanish speakers take the Initial ELPAC in 9th grade and 9% in 10th grade, meaning that over one-quarter of Spanish speakers start learning English in high school. There’s also a bump, albeit a much smaller one, in the number of Cantonese speakers who take the Initial ELPAC in 9th grade in SFUSD compared with elsewhere in the state.

Figure 4: Grade level of students whose home language was Spanish or Cantonese and who took the Initial ELPAC in 2022

Are these high schoolers new immigrants to the country or are they just new to California public schools because they did their K-8 schooling in parochial or private schools? The data won’t tell us directly but we can make an educated guess. If they are new immigrants, their level of English will be low. If they are transfers from private schools, their level of English will he high. It turns out that, of the 314 9th graders with Spanish home language who took the Initial ELPAC, 88% of them were ranked as Novices and only 8% as fluent. It’s even worse for the 10th graders: only 2% of them were fluent. In contrast, of the 43 Cantonese-speaking 9th graders who took the Initial ELPAC, only 21% were novices and 62% were fluent. The Spanish-speakers are primarily new immigrants with no English while the majority of the Cantonese speakers are already fluent in English.

SFUSD is the worst district at administering the Initial ELPAC

One of my perennial findings is that SFUSD is really bad at the basic nuts and bolts of administering a school district. Last week, it was calculating per-pupil expenditures. This week, it’s administering the Initial ELPAC. Across California, 98.9% of those students who were supposed to take the Initial ELPAC actually ended up with a score. In SFUSD, it was only 76.7%. Size is not an excuse: 35 districts had more than 1000 eligible students and only SFUSD was worse than 97.3%. Los Angeles Unified had six times as many eligible students as SFUSD and managed to get scores for 99% of them. SFUSD by itself accounted for 36% of all the eligible students in California without scores.

How many English learners are there?

Let’s exclude those who were initially fluent and count only those who were English learners when they first showed up for school, even if they have since been reclassified as fluent. The count of these “Ever ELs” is the best measure for the challenge a district faces in dealing with English learners. 48% of SFUSD’s students either are English learners or used to be English learners before getting reclassified as fluent. That’s a higher percentage than most districts but not exceptionally high. Los Angeles is at 47%; Hayward at 54%; Garden Grove (in Orange County) at 62%; Santa Ana (also in Orange County) at 74%. Where San Francisco does stand out is that only 39% of Ever-ELs are Spanish speakers. As figure 5 shows, only twelve districts fall below 50%. In most districts the vast majority of English learners are Spanish speakers. In Los Angeles, it’s 91%; in Hayward it’s 81%.

Figure 5: For each school district with at least 15,000 students, the percentage of students who are English learners or Reclassified Fluent and the percentage whose home language is Spanish.

The other districts where Spanish is not the majority language are not all the same. Alhambra (in L.A. county) has the second-most Cantonese speakers of any district (SFUSD is #1). Garden Grove has more than three times as many Vietnamese speakers as any other district. Glendale, in L.A. county, has three times as many Armenian speakers as Spanish speakers. Elk Grove, in Sacramento county, is probably the most linguistically diverse district in the state: eleven languages are each spoken by at least 2% of students: Spanish, Vietnamese, Cantonese, Hmong, Punjabi, Hindi, Farsi, Mandarin, Filipino, Pashto and Arabic.

The school districts that get the best results on proficiency tests have few English learners. If you rank the largest 50 school districts by overall proficiency rates, the three top scoring districts are San Ramon Valley, Fremont, and Irvine. It is not a coincidence that these have the 1st, 3rd, and 2nd lowest percentages of Spanish speakers. It’s not just those three. The other top 50 districts with higher overall proficiency rates than San Francisco are Poway, Capistrano, Placentia-Yorba Linda, Clovis, Glendale, Saddleback Valley, Temecula Valley, Garden Grove, and Chino Valley. With the exception of Garden Grove, they all have low numbers of English learners relatively few of whom speak Spanish.

I have written in the past about how Clovis Unified (in Fresno county) has the smallest gap between the achievement rates of its Latino and White students of any district. A big factor in this is surely that only 12% of Clovis’s students are English learners of any kind and fewer than half of that 12% are Spanish-speaking. The other districts that I regularly cite as high performing, ABC and Long Beach, have 30% and 32% English learners respectively, still substantially less than San Francisco.

Novice English Learners, by Race/Ethnicity

It’s impossible to map home languages precisely on to the standard race/ethnicity groups. Most kids whose home language is Spanish will be Latino or Two or more Races but we don’t know the proportions. Most kids whose home language is Cantonese or Mandarin or Vietnamese will be Asian or Two or more races but we don’t know the proportions. Guessing what race/ethnicity checkbox Arabic or Farsi speakers choose would be a foolhardy exercise.

Fortunately, the Initial ELPAC results are also broken down by race/ethnicity. We can see how many Asian and Latino students³ had a non-English home language and the distribution of their Initial ELPAC scores. Since the novice English learners pose the greatest educational challenge to school districts, I thought it would be interesting to calculate what percentage of SFUSD’s Asian and Latino students are novice learners (as opposed to Intermediate Learners or IFEP or English-only) and see how this compares to other districts. Figure 6 shows the result.

Figure 6: The percentage of Asian and Latino kindergartners classified as “Novice Learner” on the Initial ELPAC.

I was very surprised to see that the districts with the highest percentage of novice learners among their Latino students are all in the Bay Area. I had assumed that areas with larger Latino populations, like Fresno or Stockton, would have larger numbers of novice learners.⁴

Just over half of San Francisco’s Latino students are English novices. 69% have a non-English home language and 73% of those 69% (73% x 69% = 50.4%) score as novices. That’s the third highest fraction, behind only Oakland and San Mateo-Foster City. Compared to San Francisco, Oakland has slightly fewer Latino students with a non-English home language (68% compared to 69%) but its students are far more likely to be English novices (89% compared to 73%).

Long Beach has far more Latino kindergarten students than San Francisco (3,179 vs 1,207) but because far fewer of them have a non-English home language (34% vs. 69%) and far fewer of those are novices at English (54% vs 73%), Long Beach actually has fewer novice English learners than San Francisco (585 vs. 610). Clovis has slightly more Latino kindergartners than San Francisco (1,326 vs 1,207) but because so few have a non-English home language (8% vs 69%), they have far fewer novices (70 vs. 610).

33% of San Francisco’s Asian kindergartners are English novices. That’s a bit above average but not exceptionally so. A very high fraction (70%) of San Francisco’s Asian kindergartners speak something other than English at home but less than half of them are novices at English, bringing San Francisco’s overall rate closer to the middle. In some of the Central Valley districts (Sacramento, San Juan, Twin Rivers, Fresno, Stockton) and Oakland(!), around 80% of Asian kindergartners with a non-English home language are English novices. San Juan stands out on this chart because over 90% of its Asian students don’t speak English at home: the dominant Asian languages in San Juan are Pashto and Farsi, not Cantonese, Mandarin, or Vietnamese.

Summary

• English-language proficiency is a strong predictor of student achievement.

• San Francisco has large numbers of students who speak a language other than English at home with Spanish and Cantonese being the two dominant languages.

• Cantonese speakers in SFUSD are more likely than Spanish speakers in SFUSD to be fluent in English when entering kindergarten and, if not fluent, more likely to be intermediate learners.

• Over one-quarter of Spanish speakers in SFUSD start public school in high school grades and are predominantly novices at English.

• Just over 50% of SFUSD’s Latino kindergartners are novice English learners, a higher percentage than most other districts.

• SFUSD is the worst district in the state at administering the Initial ELPAC to eligible students.

All we’ve talked about so far is the scale of the challenge in each district. The next post will examine how well districts do at teaching English.

1. It’s very easy to over-interpret this graph. It does not mean that those who become fluent in English get an immediate 30%+ boost in their proficiency rates over those who remain as English learners. The qualities needed to become fluent in English (motivation, work, good teaching etc.) are the same as the same as those needed to excel in ELA and Math tests. If you take any normal distribution of student scores on any test, the average student who scores better than some mark (e.g. those reclassified as fluent) will have a much higher score than the average student who scores below that mark (e.g. those that remain English learners). Nor does the fact that those who are reclassified as fluent score higher than those who are English only mean that speaking two languages produces a better student than being English only. The universe of English-only students includes a mix of good, bad, and indifferent students while the universe of students reclassified as fluent includes only better than average students: the worse than average students are still English learners.

2. I had to look this up. Apparently, it’s another flavor of Chinese from around Guangdong. More than half of all Californian students who speak Toishanese (503 out of 952) attend SFUSD.

3. We can’t further subdivide by language or socioeconomic status, however, so we can’t look at Asian Cantonese speakers who are socioeconomically disadvantaged or Latino Spanish speakers who are not socioeconomically disadvantaged.

4. I wasn’t entirely wrong. I excluded districts with fewer than 150 Asian kindergartners because I wanted to put the figures for both Asian and Latino students on the same graph and needed to exclude districts with very small numbers of one or the other. If I remove that restriction, there are six districts with a higher fraction of Latino novices than Oakland. The largest is Santa Maria-Bonita in Santa Barbara (64%), but there’s also San Rafael City in Marin (70%) and Mendota in Fresno (82%).