Teaching Multilingual Learners

The big picture

It is the second week of September and you have twenty-seven ninth graders in front of you. Four of them are officially classified as English learners. Here is who they actually are.

Yesenia arrived from Guatemala eleven weeks ago. She speaks Mam at home, Spanish as a second language, and almost no English. She had six years of schooling, interrupted twice. She is seventeen years old and sitting in a ninth grade class.

Kevin was born in the hospital four miles from this building. His parents speak Cantonese at home, and the home language survey his mother filled out in kindergarten flagged him. He has been classified as an English learner for nine years. He speaks English with his friends, watches English television, and cannot pass the reclassification test.

Amina came from Syria three years ago by way of a refugee camp in Jordan. Her Arabic literacy is strong. Her English is conversationally fluent and academically fragile, and she has a documented reading disability that took two years to identify because everyone assumed the problem was English.

Dmitri arrived from Ukraine seven months ago with excellent schooling behind him, reads two languages fluently, and is doing calculus in a class where he cannot yet follow the teacher's jokes.

All four carry the same label. The label tells you almost nothing about what any of them needs. That gap between the administrative category and the human being is the subject of this lesson, and honestly, of this course.

The numbers, and why they matter to you

The federal government requires every state to identify students whose English proficiency limits their participation in school. In recent years those students have numbered around 5.3 million, roughly 10.6 percent of public school enrollment. Round it to one in ten and hold that figure, because it means that in an average class of thirty, three students are officially classified, and typically several more were classified in the past and have been reclassified out.

The share is not evenly spread. In California, English learners have made up close to 19 percent of enrollment. Texas, New Mexico, and Nevada also run high. Some states report under 3 percent. Within a state the variation is sharper still: a district in an agricultural valley or a first-ring suburb may be 40 percent English learners while a district thirty miles away is 1 percent. National averages describe nobody's actual classroom.

Two more distributions matter. First, grade level. English learners are heavily concentrated in the early grades. Around kindergarten and first grade the share often runs 15 percent or higher; by twelfth grade it drops to the mid single digits. That decline is mostly not students leaving. It is students being reclassified as English proficient, which is the system working as intended.

Second, and this is the fact that surprises most people: the majority of English learners in United States public schools were born in the United States. They are citizens. Federal analyses have put the US-born share of school-age English learners at roughly three quarters, and in the elementary grades it is higher still. The image of the English learner as a recent arrival describes a real and important minority of these students, not the typical one.

Key idea: About one in ten public school students is classified as an English learner, the share is highest in the early grades and wildly uneven across states and districts, and most of these students were born in this country.

Which languages, and what that implies

Spanish is the home language of roughly three quarters of English learners nationally. That single fact shapes almost everything about the field: most bilingual programs are Spanish-English, most translated materials exist first in Spanish, and most teacher preparation assumes a Spanish-speaking student.

The remaining quarter is where the complexity lives. After Spanish, the most commonly reported home languages include Arabic, Chinese (usually reported as Mandarin or Cantonese), Vietnamese, and Portuguese, followed by a long tail. A mid-sized urban district can easily enroll students from sixty or eighty language backgrounds. And the survey categories conceal things: a family recorded as Spanish-speaking may actually speak Mam, Kaqchikel, Kichwa, Mixtec, or Zapotec at home and use Spanish as a second language, which means the interpreter the district provides may be no help at all.

A useful table for orienting yourself:

Home language	Approximate share of English learners	Practical implication
Spanish	About 75 percent	Bilingual programs, cognate instruction, and translated materials are widely available.
Arabic	Roughly 2 to 3 percent	Different script and directionality; strong literacy in many families; refugee arrivals common.
Chinese	Roughly 2 percent	Non-alphabetic writing system; word reading transfers differently.
Vietnamese	Roughly 2 percent	Latin script with diacritics; tonal language; established community networks in some regions.
All others combined	Roughly 18 to 20 percent	Often one or two students per district; interpreters and materials scarce.

Key idea: Three quarters of English learners speak Spanish at home, which shapes the entire infrastructure of the field, while the remaining quarter spans hundreds of languages, sometimes hidden behind an inaccurate survey answer.

Three profiles that need three different things

Practitioners have learned to sort students into working profiles that cut across the legal label. These are not official categories in federal law, and definitions differ by state, but they organize your thinking better than the label does.

Newcomers are recently arrived students, usually defined as enrolled in United States schools for less than one to three years. Yesenia and Dmitri are both newcomers, and notice that they need opposite things. Dmitri needs English and nothing else; his mathematics is ahead of the class. Yesenia needs English, foundational literacy, content she never had a chance to learn, and a social bridge into a building where nobody speaks Mam. Both are newcomers. One label, two entirely different instructional problems.

Within the newcomer group, the students with interrupted or limited formal schooling, often abbreviated SLIFE, are the ones schools most often fail. They arrive with age-appropriate intelligence and below-grade literacy, and the standard secondary schedule has no obvious place for them. Module 6 takes this up directly.

Long-term English learners, often abbreviated LTELs, are students who have been classified for six or seven years or longer without reclassifying. Kevin is one. California, which defines the term in state law, has found that a large share of its secondary English learners fall into this group. These students typically sound fluent, which is precisely their problem: teachers hear the conversational fluency, assume the language work is done, and stop scaffolding. What is usually missing is academic language and academic literacy, sometimes compounded by years of a watered-down course schedule that never asked them to write an argument or read a dense text. The label makes them look like slow learners. The record usually shows something else: a student who was never taught what the reclassification test measures.

Dually identified students are English learners who also qualify for special education services under IDEA. Nationally, something on the order of 15 percent of English learners also receive IDEA services. Amina is one. These students are entitled to both sets of services, and neither one substitutes for the other. An IEP does not cancel English language development, and English language development does not cancel an IEP. That sounds obvious written down. In practice, dually identified students are one of the most commonly underserved groups in American schools, and Module 5 spends a full lesson on why identification here is so hard in both directions.

Key idea: Newcomers, long-term English learners, and dually identified students sit inside the same legal category and need fundamentally different instruction, so the label is a starting point for questions rather than an answer.

What the label actually is

Be clear about the nature of the category. Under federal education law, an English learner is a student whose difficulties in speaking, reading, writing, or understanding English may be sufficient to deny them the ability to meet challenging state standards or to participate fully in classrooms. It is defined by what a student cannot yet do in one language.

That definition has three consequences worth naming now. It is administrative, so a student becomes an English learner because a survey and a test placed them there, not because of anything intrinsic. It is temporary by design, since the entire point is exit. And it is a deficit description, which is why the terminology fight covered in the next lesson exists at all.

Notice also what the label does not capture. It says nothing about how well the student reads in Arabic, nothing about the four years of algebra completed in Ukraine, nothing about the child who interprets for her mother at the pharmacy every week and manages more complex communication before dinner than most adults handle all day. A student who has been in this country ten weeks and speaks three languages is not a beginner at language. They are a beginner at English.

Key idea: English learner is an administrative, temporary, deficit-framed classification that records what a student cannot yet do in English and records nothing about what the student can already do in any other language.

What this course can and cannot do

Here is the honest boundary. This course will give you the vocabulary, the legal structure, the acquisition research, the program evidence, and the instructional practices that hold up. By the end you should be able to read a proficiency score report without guessing, follow a placement conversation, choose scaffolds on purpose, and tell a well-supported claim from a popular one.

It will not qualify you to identify a student as an English learner, to place a student in a program, to reclassify a student out, or to determine whether a student has a disability. Those are regulated decisions made by trained staff and teams under state procedures, and doing them wrong has legal consequences for a district and real consequences for a child. If this course leaves you wanting to do that work, the next step is a certification program, which the final lesson maps out concretely.

One more scope note. The law, the categories, and the assessment systems described here are those of the United States. Other countries organize language education entirely differently. The acquisition research travels across borders. The statutes do not.

Key idea: This course builds conceptual, legal, and instructional literacy, not the authority to identify, place, reclassify, or evaluate any student.

Common misconceptions

• English learners are mostly immigrants. Roughly three quarters were born in the United States and are citizens.
• The label tells you what a student needs. It tells you the student has not yet met an English proficiency criterion. Everything else requires looking at the individual.
• A student who speaks English easily no longer needs support. Conversational fluency arrives years before academic language, which is exactly how long-term English learners are produced.
• Students with disabilities should be exited from language services to focus on the IEP. Dually identified students are entitled to both, and neither replaces the other.
• Most English learners are recent arrivals with no schooling. Students with interrupted formal schooling are a real and underserved group, but they are a minority of a minority.

Recap

• About 5.3 million students, roughly one in ten, are classified as English learners in United States public schools.
• The share is highest in the early grades and varies enormously by state and district.
• Most English learners were born in this country; roughly three quarters speak Spanish at home.
• Newcomers, long-term English learners, and dually identified students share a label and need different things.
• The classification is administrative, temporary, and deficit-framed, and describes only English.
• This course teaches concepts and practices; it confers no authority to classify, place, or evaluate.

Sources

1. National Center for Education Statistics. English Learners in Public Schools. Condition of Education. U.S. Department of Education, Institute of Education Sciences. nces.ed.gov
2. U.S. Department of Education, Office of English Language Acquisition. English Learners: Demographic Trends. ncela.ed.gov
3. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
4. Colorin Colorado. Who Are English Language Learners? WETA. colorincolorado.org

The big picture

Two claims are made constantly in this field. The first is that bilingualism is an asset. The second is that the research proves it. The first claim is true and this course will defend it. The second claim is stated far too loosely, and if you repeat the loose version to a skeptical colleague or a school board, you will get taken apart, because parts of the popular evidence have not held up well.

So let us do this properly. In this lesson we sort the evidence into three piles: claims that are strongly supported, claims that are genuinely contested, and claims that are mostly folklore. Then we look at what happens when a school treats a home language as an obstacle instead, which is not a hypothetical, since it was national policy in living memory. Then we take up the argument over what to call these students, an argument that looks like semantics and is actually about who is responsible for what.

What the evidence supports well

Start with the strongest ground, which is educational rather than neurological.

Instruction in the first language does not slow English. This is the most policy-relevant finding in the field and it is unusually well replicated. Multiple meta-analyses of program evaluations, including work by Greene in 1997, Rolstad and colleagues in 2005, and Slavin and Cheung in 2005, found that students in bilingual programs do at least as well in English as comparable students in English-only programs, and frequently better. The National Academies of Sciences, Engineering, and Medicine reached the same conclusion in its 2017 consensus report. Note carefully what this means. Time spent teaching in Spanish is not time stolen from English. Students in these programs end up with equal or stronger English and a second language on top of it, which is why the finding matters: the comparison is not close, because one group finishes with two languages.

Literacy skills transfer across languages. The National Literacy Panel on Language-Minority Children and Youth, summarized by August and Shanahan in 2006, found substantial cross-linguistic relationships in phonological awareness, word reading, and reading strategies. A student who has learned to read in Spanish is not starting reading over in English; she is learning a new orthography for a skill she already has. Lesson 2.3 works through the mechanics.

Metalinguistic awareness is stronger in bilinguals. Bilingual children tend to be earlier and better at noticing that language is an arbitrary system, that a word is not the thing it names, that a sentence can be grammatical and false. This finding predates the modern executive function literature and has held up better than it has.

Maintaining the home language protects family relationships. Lily Wong Fillmore documented in 1991 what happens when young children lose a home language their parents do not share: parents lose the ability to guide, discipline, comfort, and pass on values, and adolescents lose access to the adults who raised them. Portes and Rumbaut found in large-scale studies of immigrant youth that fluent bilinguals had better outcomes on several measures than students who had lost the home language. This is not a soft finding. It is one of the most consistent in the literature.

Key idea: The strongest evidence for bilingualism as an asset is educational and social, not neurological: first language instruction does not slow English, literacy transfers, metalinguistic awareness is stronger, and keeping the home language protects the family.

What is genuinely contested

Now the honest part, because a course that only reports the flattering studies is not teaching you anything.

The bilingual advantage in executive function is the famous claim: that constantly managing two languages trains attention and inhibitory control, producing better performance on tasks like the Simon or flanker task. Ellen Bialystok and colleagues have reported such effects for decades. Beginning around 2013, the finding came under serious attack. Paap and Greenberg found no consistent advantage across a battery of executive function measures. De Bruin, Treccani, and Della Sala examined conference abstracts and found that studies reporting a bilingual advantage were substantially more likely to be published than studies that did not, which is textbook publication bias. Several large replication attempts have come back null or very small.

Where does that leave the claim? Roughly here: any general executive function advantage in healthy young adults is small at best and may be zero, effects are more plausible in young children and older adults, and the whole area shows the signature of a literature built before preregistration. The related claim that bilingualism delays the onset of dementia symptoms by several years, also from Bialystok's group, has both supporting and contradicting studies and is likewise unsettled.

Here is the thing to notice. None of the educational case for bilingual education depends on any of this. If the executive function advantage evaporates entirely tomorrow, first language instruction still does not slow English, literacy still transfers, families still communicate, and the student still ends up bilingual. Do not hang your argument on the shakiest branch.

Key idea: The cognitive bilingual advantage in executive function is contested, weakened by publication bias and failed replications, and the educational case for bilingualism does not depend on it.

What is mostly folklore

Two claims deserve to be retired. The first is that bilingualism confuses young children and delays language development. Bilingual children often have a smaller vocabulary in each language taken alone, and roughly the same or larger total vocabulary across both. They reach milestones on a typical schedule. Mixing two languages in one sentence is not confusion; it is a competent bilingual behavior discussed in Lesson 2.3.

The second is that speaking the home language at home slows English. Parents are told this constantly, sometimes by teachers and pediatricians. The advice is wrong and it is harmful, because it pushes a parent to interact with a child in a language the parent commands poorly, degrading the quality of input in both languages while dismantling the home language. The correct advice is the opposite: speak the language you speak best, richly and often.

Key idea: Bilingualism does not confuse children or delay development, and telling parents to switch to English at home damages input quality in both languages.

Additive and subtractive bilingualism

Wallace Lambert drew a distinction in the 1970s that organizes all of this. Additive bilingualism is adding a second language while the first continues to develop. Subtractive bilingualism is adding a second language at the cost of the first. The outcomes are not comparable. Additive bilingualism is associated with the benefits above. Subtractive bilingualism produces a student who has traded one language for another and is often fully proficient in neither for a stretch of years.

Which one happens is largely a matter of policy and social pressure rather than a fact about children. In the United States the pull is strongly subtractive: by the third generation, most immigrant families have shifted to English monolingualism. Meanwhile the same school system spends considerable money teaching Spanish to English-speaking students in a high school classroom, having quietly extinguished it in students who arrived speaking it. That is the sentence to remember when someone asks why bilingual education matters.

Key idea: Additive bilingualism adds a language while the first keeps growing; subtractive bilingualism replaces one with the other, and United States schooling pulls hard toward subtraction unless a program deliberately resists it.

The argument over what to call students

Now the terminology, which changes roughly every fifteen years and is currently unsettled.

Limited English proficient (LEP) is the older statutory term, used throughout No Child Left Behind and still lingering in some federal documents. Critics point out that it is the only major education label that defines a child entirely by a deficiency, and the phrasing is blunt enough that few practitioners use it in speech anymore.

English learner (EL), sometimes English language learner (ELL), became the dominant term and is the term used in the Every Student Succeeds Act of 2015. It is a modest improvement: it describes an activity rather than a deficiency. It is also the legal term, which matters enormously, because rights, funding, monitoring, and accountability all attach to it.

Multilingual learner (MLL) is the current reform. WIDA, the largest English language proficiency consortium in the country, uses it. New York and several other states have adopted it. The argument for it is straightforward: it names what the student has rather than what the student lacks, and it includes students who are bilingual but were never classified.

Emergent bilingual comes from Ofelia Garcia and makes the asset framing explicit: the student is in the process of becoming bilingual, and the school's job is to complete that process rather than to trade one language for another.

Now the objections, which are real and come from people who are not hostile to the students.

First, precision. Multilingual learner is broader than English learner, so a district that reports on multilingual learners may be blurring the smaller group that holds a legal entitlement. If a compliance monitor asks how many English learners received required services and the district's data system only tracks multilingual learners, that is a problem for the students, not the vocabulary.

Second, the risk of a euphemism doing the work of a service. Renaming a category is free. Hiring a certified teacher is not. Critics note that a term change can be announced in a press release while nothing else changes, and that a more comfortable word can make an uncomfortable inequity easier to overlook.

Third, accuracy for particular students. A newcomer who speaks one language and is beginning a second is not yet multilingual in any strong sense, and some educators find the term aspirational to the point of inaccuracy.

Where does that leave you? Use the term your state and district use in official contexts, because that is where compliance lives. Use asset-oriented language in ordinary speech and in front of students and families. Never let the label stand in for a description of the student. And when a student tells you what they prefer to be called, that settles it for that student. The most defensible position is the one most of the field has landed on: the words matter, and they matter less than what happens after the words.

Key idea: English learner is the legal term that carries rights and funding, multilingual learner and emergent bilingual are asset-framed alternatives now used by WIDA and several states, and the strongest objection to renaming is that it is free while services are not.

Common misconceptions

• The research proves bilingualism makes you smarter. The executive function advantage is contested and possibly small or absent; the educational and social benefits are much better supported.
• Time spent teaching in Spanish is time taken from English. Meta-analyses and the National Academies report find bilingual program students do as well or better in English.
• Parents should switch to English at home to help. This degrades input quality in both languages and accelerates home language loss.
• Bilingual toddlers are delayed. They reach milestones typically; vocabulary counted across both languages is comparable to or larger than a monolingual's.
• Changing the label solves the problem. The label is worth getting right, and it changes nothing about staffing, instruction, or funding by itself.

Recap

• First language instruction does not slow English acquisition, and literacy skills transfer across languages.
• Metalinguistic awareness and family communication are well-supported benefits of maintaining the home language.
• The executive function bilingual advantage is contested; publication bias and failed replications have weakened it.
• Additive bilingualism adds a language; subtractive bilingualism replaces one, and United States schooling pulls toward subtraction.
• LEP, English learner, multilingual learner, and emergent bilingual encode different framings; English learner carries the legal entitlement.
• Use the official term for compliance, asset language in practice, and a student's own preference for that student.

Sources

1. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
2. WIDA. WIDA English Language Development Standards Framework. Wisconsin Center for Education Research, University of Wisconsin-Madison. wida.wisc.edu
3. Colorin Colorado. The Benefits of Being Bilingual. WETA. colorincolorado.org
4. Encyclopaedia Britannica. Bilingualism. britannica.com
5. Wikipedia contributors. Cognitive effects of bilingualism. Wikipedia. en.wikipedia.org

The big picture

Every child in the world who is not prevented by injury or extreme deprivation acquires a first language, on roughly the same schedule, without instruction, to complete proficiency. Nobody teaches a two-year-old the past tense. She overgeneralizes it, says goed for a year or two, and then stops, and no correction from adults reliably speeds that up.

Second language acquisition is nothing like that. Outcomes vary from near-native to almost nothing. Motivation matters. Instruction sometimes helps a great deal. Many adult learners plateau permanently while others do not. Any theory of how a second language grows has to explain that variability, and the theories you will hear cited in schools were built to do exactly that.

This lesson does three things. It sets out how first and second acquisition differ. It presents Stephen Krashen's model, which is by far the most influential framework in American language education, honestly and in full, and then presents the criticisms that specialists have made of it for forty years. And it introduces interlanguage and the silent period, two concepts you will use weekly.

First and second: what actually differs

Dimension	First language	Second language
Outcome	Uniform success	Highly variable, from near-native to minimal
Instruction	Not required and largely ineffective	Often helpful, sometimes necessary
Existing knowledge	None; the concept and the word are learned together	An entire conceptual system already in place
Input quantity	Enormous, continuous, individually tuned	Often limited, and in school often only during class
Affect	Little embarrassment about errors	Errors can be socially costly, especially for adolescents
Identity	Not usually at stake	Frequently at stake, since using the new language can feel like leaving a community

Two rows deserve comment. The row on existing knowledge is good news that teachers routinely forget. A ninth grader who understands photosynthesis in Vietnamese does not need to relearn photosynthesis. She needs a label. Teaching her the concept from scratch wastes her time and insults her, and it is a common error when a teacher confuses limited English with limited knowledge.

The identity row is the one that explains adolescents. A teenager who speaks English perfectly with friends and refuses to speak it in class is often not struggling with English. Language use marks group membership, and for a fourteen-year-old, group membership is not a small thing.

Key idea: Second language acquisition differs from first in outcome variability, the usefulness of instruction, the presence of an existing conceptual system, the quantity of input, and the way identity and social risk shape whether a learner speaks at all.

Krashen's model, stated fairly

Stephen Krashen advanced five hypotheses in the late 1970s and early 1980s. Together they are called the Monitor Model, and they underwrite an enormous amount of American classroom practice.

1. The acquisition-learning hypothesis. Acquisition is subconscious, the way a child picks up a first language. Learning is conscious knowledge of rules. They are separate systems, and Krashen argued that learning does not become acquisition.

2. The monitor hypothesis. Consciously learned rules serve only as an editor, a monitor that checks output when the learner has time, is focused on form, and knows the rule. This is why a student can pass a grammar test and still say sentences that violate the rule in conversation.

3. The natural order hypothesis. Grammatical features are acquired in a predictable order that is roughly the same across learners and is not the order textbooks teach. The morpheme studies of the 1970s, by Dulay and Burt and others, found for example that the progressive -ing and plural -s come early while third person singular -s and possessive -s come late, regardless of instruction.

4. The input hypothesis. We acquire language by understanding messages slightly beyond our current level, which Krashen wrote as i plus 1. Comprehensible input is the causal variable; speaking is a result of acquisition, not a cause of it.

5. The affective filter hypothesis. Anxiety, low motivation, and low self-confidence raise a filter that blocks input from reaching the acquisition device. A frightened learner can receive comprehensible input and acquire little from it.

Take the model seriously before criticizing it. It named something real that classrooms were ignoring: that drilling rules in a language the student cannot understand produces students who can conjugate and not converse. Its practical directive, flood students with meaningful input they can understand and lower the stress, is still good advice, and Lesson 4.1 will teach you how to do it.

Key idea: Krashen's Monitor Model holds that language is acquired subconsciously through comprehensible input slightly beyond the learner's level, that consciously learned rules only edit output, and that anxiety blocks acquisition.

The criticisms, stated just as fairly

Second language acquisition researchers have pressed four objections, and you should be able to state them.

The model is difficult to falsify. This is the central complaint, associated with Kevin Gregg and others. What exactly is i, and how would you measure plus 1? If a learner does not acquire despite comprehensible input, the affective filter explains it. If a learner acquires without much input, some hidden input is posited. A theory that accommodates every result predicts none of them. The constructs are intuitive and not operationalized.

Acquisition and learning cannot be independently distinguished. The claim that they are separate systems and that learning never becomes acquisition is difficult to test without a way to identify each in a learner. There is now substantial evidence that explicit instruction does improve acquisition outcomes, particularly for features that are hard to notice, which is not what a strict no-interface position predicts.

Input alone is not sufficient; output does work. Merrill Swain studied French immersion students in Canada who had received years of rich comprehensible input and still made persistent grammatical errors. Her comprehensible output hypothesis proposed that being pushed to produce language forces learners to process syntax rather than just meaning, notice gaps in what they can say, and test hypotheses. Immersion students who understand everything and speak imprecisely are the standing counterexample to input-only accounts.

Interaction and noticing matter. Michael Long's interaction hypothesis argued that input becomes usable through negotiation of meaning: the clarification requests, confirmation checks, and repetitions that happen when two people work to understand each other. Richard Schmidt's noticing hypothesis argued that learners must consciously notice a form in the input for it to be acquired. Both put conscious attention back into a model that had ruled it out.

Where the field landed is worth stating plainly, because it is what you should actually do. Comprehensible input is necessary and is not sufficient. Interaction, pushed output, and some attention to form add value. Krashen's practical advice is largely sound; his theoretical claims are widely regarded as unproven. You can teach by his prescriptions without endorsing his mechanism, and most effective teachers do.

Key idea: Krashen's hypotheses are criticized as unfalsifiable and as underrating output, interaction, and attention to form, and the field's consensus is that comprehensible input is necessary but not sufficient.

Interlanguage: the learner's own grammar

Larry Selinker proposed in 1972 that a learner's language at any moment is not broken English. It is a coherent system with its own rules, drawn from the first language, from the target language, and from the learner's own generalizations. He called it interlanguage.

This changes how you read errors. A student who writes "He no want to go" has not failed to learn negation. He has produced a rule-governed negation pattern documented across many learners and many first languages. It is a stage. A student who writes "I have twenty years" is transferring a Spanish construction, which is a systematic hypothesis, not carelessness. Errors are data about the current system, and once you read them that way, they become the most useful assessment information you get all day.

Fossilization is Selinker's term for interlanguage features that stop changing and persist indefinitely despite continued exposure and correction. The permanent accent is the familiar example; specific grammatical features fossilize too. The concept is useful and also contested, since it is hard to prove that a feature will never change rather than has not changed yet. Practically it means that a plateau is common, is not evidence of laziness, and sometimes responds to focused attention on the specific form rather than to more input in general.

Key idea: Interlanguage is the learner's own systematic grammar at a point in time, so errors are patterned evidence about that system rather than random failures, and fossilized features may persist despite exposure.

The silent period, and how it gets abused

Many learners, especially young children in a new language environment, go through a silent period in which they take in language and produce almost none. It can last weeks or several months. During it, comprehension typically grows substantially, so the student is far from idle.

Two cautions. First, it is not universal. Some learners talk from the first week. Some cultures and personalities encourage early attempts. Silence is not a required developmental stage that every student must pass through.

Second, and more seriously, the concept gets used as an excuse. A student who has been silent for a school year is not in a silent period; something else is happening, and the possibilities include a hearing difference, selective mutism, trauma, a disability, or simply an environment where nobody has given them a safe, low-stakes reason to say anything. The honest use of the concept is: do not force a beginner to perform in front of peers, do give them nonverbal and low-risk ways to show understanding, and do watch the clock. The dishonest use is: leave the newcomer alone at the back of the room for a semester and call it research-based.

Key idea: A silent period of weeks to months is common while comprehension develops, but it is not universal and is not an explanation for a student who has been silent for a year.

Common misconceptions

• Second language learning works just like first language learning. Outcomes vary enormously, instruction helps, and identity and anxiety shape the process in ways they do not for infants.
• Krashen's hypotheses are established science. They are influential and much criticized, particularly for being difficult to falsify.
• Comprehensible input is all you need. Immersion research shows learners can understand everything and still produce persistently inaccurate language without pushed output and attention to form.
• Errors mean the student is not trying. Errors are patterned evidence of a systematic interlanguage grammar and are the best free assessment data available.
• A student who is not talking is in the silent period. Only for weeks to months, and only in some learners; extended silence needs investigation.

Recap

• First language acquisition is uniform and instruction-independent; second language acquisition is variable and often instruction-assisted.
• Krashen's five hypotheses are acquisition-learning, monitor, natural order, input, and affective filter.
• Critics charge unfalsifiability and point to Swain's output hypothesis, Long's interaction hypothesis, and Schmidt's noticing hypothesis.
• The working consensus is that comprehensible input is necessary but not sufficient.
• Interlanguage is the learner's own rule-governed system; fossilization is a feature that stops changing.
• The silent period is real, common, time-limited, and frequently misused to justify neglect.

Sources

1. Wikipedia contributors. Input hypothesis. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. Interlanguage. Wikipedia. en.wikipedia.org
3. Encyclopaedia Britannica. Second-language acquisition. britannica.com
4. Colorin Colorado. Language Acquisition: An Overview. WETA. colorincolorado.org
5. TESOL International Association. The 6 Principles for Exemplary Teaching of English Learners. tesol.org

The big picture

Two of the most quoted numbers in this field are two years and seven years. They come from Jim Cummins, and stated loosely they say that conversational English takes about one to three years while academic English takes about five to seven. Those numbers appear on countless professional development slides, and they have done real good, because they stop teachers from concluding that a chatty student who is failing science must not be trying.

They have also been used badly. They have been used to delay content instruction, to explain away a genuine disability, and to sort students into a permanent category of not-yet-ready. And the distinction behind them has been attacked by serious scholars for forty years on grounds you should know.

The second half of this lesson takes up the age question, which is the one families ask most and which is answered wrongly more often than any other topic in the field. The short answer, which we will earn: younger learners are not simply better, older learners are faster at first, and neither fact justifies waiting to teach content.

BICS and CALP

Cummins proposed a distinction between basic interpersonal communicative skills, or BICS, and cognitive academic language proficiency, or CALP.

BICS is the language of the hallway, the cafeteria, the soccer field, and the group text. It is face to face, richly contextualized, cognitively undemanding for the speaker, and supported by gesture, shared reference, tone, and the ability to ask what did you say. Most learners in a school setting reach functional conversational fluency in roughly one to three years.

CALP is the language of the textbook, the lab report, the document-based question, and the standardized test. It is context-reduced, meaning the words carry nearly the entire load. It is cognitively demanding. It uses vocabulary that does not appear in conversation, sentence structures that pack several clauses into one line, and rhetorical moves like hedging, citing, and qualifying. Estimates for reaching grade-level academic English run about five to seven years, and studies by Kenji Hakuta, Yuko Butler, and Daria Witt in 2000 put the range at roughly four to seven years even in favorable conditions.

The gap between those two timelines is where long-term English learners are manufactured. A student sounds fine in year three. The scaffolds come down. The academic language work stops. Four years later the student is still classified, and everyone is puzzled.

Key idea: Conversational fluency typically arrives in one to three years while grade-level academic language takes roughly four to seven, and the gap between those timelines is the standard mechanism by which long-term English learners are produced.

The criticisms, which are substantial

Now the part most workshops skip. The BICS and CALP distinction drew sustained criticism, and understanding it will make you use the concept better rather than abandon it.

The deficit objection. Carole Edelsky and colleagues argued in 1983 that CALP amounts to a repackaged version of semilingualism, the discredited idea that some bilingual children have no full language at all. Their charge was that CALP privileges one narrow, school-based, test-like use of language and treats a student's failure to display it as a deficiency inside the child rather than a mismatch produced by schooling. It is a serious argument, and you should feel its force: what CALP measures is substantially what schools happen to reward.

The container objection. Marilyn Martin-Jones and Suzanne Romaine argued in 1986 that the model implicitly treats proficiency as liquid in a container, quantities of language that a person has more or less of, which misdescribes how bilinguals actually use language in social settings.

The dichotomy objection. Others, including Jeff MacSwan and Terrence Wiley, have pressed that the two labels get used as a hard binary applied to students rather than to language, so that people start saying a student "has BICS but not CALP" as if describing a trait. Cummins himself has objected to that usage. Language varies along continuous dimensions; students do not come in two flavors.

The delay objection. The most damaging misuse in practice: five to seven years gets read as a waiting period. A team decides not to refer a struggling student for evaluation because it has not been seven years yet. A department decides a newcomer is not ready for grade-level content. Both are wrong, and the second one guarantees the outcome it predicts.

Cummins responded to the criticism by refining the model into two intersecting continua: how context-embedded or context-reduced a task is, and how cognitively undemanding or demanding it is. That version is more defensible and more useful for lesson planning, because it describes tasks rather than sorting children. A cafeteria conversation is context-embedded and undemanding. A phone call about a bus schedule is context-reduced and undemanding. A hands-on lab with a partner is context-embedded and demanding, which is exactly why labs are such good language environments. A timed reading test is context-reduced and demanding, which is why it is the hardest thing many students do all year.

Key idea: Critics charge that CALP repackages deficit thinking, treats proficiency as a quantity, and gets applied as a label to students, and Cummins's revised two-continua model, which describes tasks by context and cognitive demand, answers much of that.

The age question, honestly

Families ask this constantly, and so do policymakers: is it true that children learn languages easily and adults cannot? Here is the evidence.

Eric Lenneberg proposed the critical period hypothesis in 1967, arguing for a biologically constrained window for language acquisition closing around puberty. For second languages the hypothesis has been tested many times.

Jacqueline Johnson and Elissa Newport studied Chinese and Korean speakers who had arrived in the United States at different ages and found that grammaticality judgment scores declined steadily with age of arrival up to about puberty, then flattened at a lower level. That study anchored the field for decades. A much larger analysis by Joshua Hartshorne, Joshua Tenenbaum, and Steven Pinker in 2018, using data from around two thirds of a million respondents, found that the ability to learn grammar to a native-like level remains high until roughly age seventeen or eighteen and then declines, but that reaching native-like attainment requires beginning around age ten to twelve because acquisition takes so many years.

Three refinements matter more for your classroom than the debate itself.

Pronunciation is the most age-sensitive domain. Accent is where early arrival shows most clearly. It is also the least important domain educationally. An accent is not a language problem.

Older learners are faster at the start. Catherine Snow and Marian Hoefnagel-Hohle followed English speakers learning Dutch in the Netherlands and found that adolescents and adults outpaced young children in morphology and syntax over the first year. Older learners bring literacy, study skills, metalinguistic awareness, and world knowledge. Younger learners have the advantage over the long run, mostly in phonology and in ultimate attainment.

Length and quality of exposure often outweigh age. A ten-year-old with strong first language literacy, engaged teachers, and English-speaking friends will typically outperform a six-year-old sitting silently at the back of a class with no scaffolding. Age is one variable among several, and it is the one you cannot change.

The classroom conclusion is straightforward. Do not tell a family that their teenager has missed the window; the evidence does not support that and the message is corrosive. Do not tell them their kindergartner will pick it up automatically; that belief is how young students end up unsupported. And never delay content instruction on the theory that language must come first, because content is where the language lives.

Key idea: Age effects are real and strongest for pronunciation, older learners are initially faster in grammar and vocabulary, native-like attainment is associated with an early start mainly because acquisition takes years, and exposure quality often matters more than age.

What to actually do with all of this

Three practices follow directly.

First, keep teaching academic language deliberately for years after a student sounds fluent. The scaffolds you would give a newcomer are not the scaffolds a fourth-year student needs, but the answer is different scaffolds, not none. Module 4 is entirely about this.

Second, describe tasks, not students. Instead of saying a student has BICS and lacks CALP, say that this task is context-reduced and cognitively demanding and therefore needs a graphic organizer, a model text, and a partner rehearsal before writing. The first sentence labels a child. The second one plans a lesson.

Third, never use a timeline as a reason to wait. If a student is struggling in ways that look unusual for their profile, the right response is to look harder now, using the true-peer comparison taught in Lesson 5.2, not to run out the clock on a seven-year estimate that was never intended as a diagnostic rule.

Key idea: Use the research to keep scaffolding after fluency arrives, to describe task demands instead of labeling students, and never to justify delay.

Common misconceptions

• BICS and CALP are two kinds of student. They describe language use along continua of context and cognitive demand, not types of children.
• A fluent-sounding student is done. Academic language typically takes four to seven years, well past conversational fluency.
• You must wait five to seven years before considering a disability. No timeline forbids looking carefully now; waiting is itself a decision with consequences.
• Children learn languages effortlessly and adults cannot. Older learners are faster initially; younger learners have advantages mainly in pronunciation and ultimate attainment.
• An accent means incomplete acquisition. Accent is the most age-sensitive and least educationally important feature of a second language.

Recap

• BICS is contextualized conversational language; CALP is context-reduced, cognitively demanding academic language.
• Conversational fluency takes about one to three years; academic English about four to seven.
• Critics charge deficit framing, a container view of proficiency, and misuse as a student label; the two-continua revision answers much of it.
• Critical period evidence shows real age effects, strongest for pronunciation, with native-like grammar attainment linked to an early start.
• Older learners are faster initially because they bring literacy, study skills, and world knowledge.
• Describe task demands rather than labeling students, and never use a timeline to justify delay.

Sources

1. Wikipedia contributors. Basic interpersonal communicative skills. Wikipedia. en.wikipedia.org
2. Wikipedia contributors. Critical period hypothesis. Wikipedia. en.wikipedia.org
3. Colorin Colorado. Language Acquisition: An Overview. WETA. colorincolorado.org
4. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
5. Encyclopaedia Britannica. Language acquisition. britannica.com

The big picture

A seventh grader named Rosa reads well in Spanish. She has been reading chapter books in Spanish since third grade. She arrives in your school in October and is placed in a beginning English class where the reading material is a picture book about a duck.

Something has gone wrong, and it is not a small thing. Rosa knows how to read. She knows that print maps to sound, that a text has a main idea, that you slow down when you are confused, that an author has a purpose. None of that is Spanish-specific knowledge. She needs the English sound-symbol correspondences, English vocabulary, and English syntax. She does not need to be taught what reading is.

This lesson is about what Rosa already has and how to get at it. That means three things: the theory of transfer, the practical tools for exploiting it, and the newer framework called translanguaging, which pushes the idea considerably further and has drawn considerable argument in the process.

Interdependence and the common underlying proficiency

Jim Cummins proposed the linguistic interdependence hypothesis: to the extent that instruction in one language develops proficiency in that language, transfer of that proficiency to a second language will occur, provided there is adequate exposure to the second language and adequate motivation to learn it.

The metaphor he used is the dual iceberg. Two languages look separate above the waterline, with different sounds and different words, but below the surface they share a common underlying proficiency: concepts, reasoning, background knowledge, literacy strategies, and metalinguistic awareness. Develop it through either language and it is available to both.

This is the theoretical basis for the single most counterintuitive finding in the field, which you met in Lesson 1.2: teaching a student to read in Spanish improves the student's English reading. It sounds like it should not work, and it does, and interdependence is why.

Note the conditions in Cummins's own statement, because they are often dropped. Transfer requires adequate exposure to the second language and motivation to learn it. Interdependence is not a claim that first language instruction alone produces English. It is a claim that first language development is an engine for English rather than a competitor with it, given English exposure.

Key idea: Languages share a common underlying proficiency of concepts, knowledge, reasoning, and literacy strategies, so developing that base in either language makes it available to both, provided there is adequate exposure to the second language.

What transfers and what does not

Be specific. Vague enthusiasm about transfer produces bad instruction.

Transfers readily	Transfers partially	Does not transfer
Concept knowledge and content understanding	Phonological awareness, strongly but not perfectly	Specific phonemes not present in the first language
The idea that print encodes language	Decoding, depending on how alike the writing systems are	English spelling conventions and irregular words
Comprehension strategies such as predicting and summarizing	Cognate vocabulary, when learners are taught to look for it	Idioms, collocations, and register conventions
Metalinguistic awareness and study skills	Genre knowledge, since genres differ across cultures	Word order rules that differ between the languages

The middle column carries the most nuance. Phonological awareness, the ability to hear and manipulate sounds in words, transfers robustly, which is why a student with strong Spanish phonemic awareness usually picks up English phonemic tasks quickly. Decoding transfers to the degree that the writing systems resemble each other. Spanish orthography is shallow, meaning letters map to sounds consistently, while English is deep and irregular, so a strong Spanish decoder still has real work to do, especially with vowels. A Chinese reader moving to English is changing not just the letters but the type of writing system, so less of the mechanics transfer even though the higher-level strategies still do.

The right column is why transfer does not make instruction unnecessary. English phonemes absent from the first language have to be taught. English word order has to be taught. Idioms have to be taught, because "she nailed it" is not derivable from anything.

Key idea: Concepts, print awareness, comprehension strategies, and study skills transfer readily; phonological awareness and decoding transfer to a degree that depends on how similar the writing systems are; specific phonemes, spellings, idioms, and word order do not transfer at all.

Cognates: the highest-leverage tool you are not using

English borrowed enormously from Latin and French, and Spanish descends from Latin, so English academic vocabulary overlaps heavily with everyday and academic Spanish. Estimates of the overlap in academic word lists run from roughly a third to well over half. That is an enormous free gift, and most Spanish-speaking students do not receive it, because nobody tells them the resource exists.

A sample of what a science teacher can hand over in ninety seconds:

English	Spanish cognate	Note
observe	observar	Verb ending pattern -ar maps to -e or bare form.
experiment	experimento	Nearly identical.
hypothesis	hipotesis	English ph corresponds to Spanish f.
analyze	analizar	English -ize corresponds to Spanish -izar.
identify	identificar	English -fy corresponds to Spanish -ficar.
conclusion	conclusion	English -tion and -sion correspond to Spanish -cion and -sion.

Teaching the pattern is worth far more than teaching the words. Once a student knows that -tion corresponds to -cion, hundreds of words open at once. And notice the pleasant irony: the English academic words that are hardest for English-speaking students are often the ones that are easiest for Spanish speakers, because in Spanish those Latinate words are ordinary.

Teach false cognates alongside the real ones, because they cause real errors. Embarazada means pregnant, not embarrassed. Actualmente means currently, not actually. Librería is a bookstore, not a library. Asistir means to attend, not to assist. Students find these funny, which makes them memorable, which is the point.

Key idea: A large share of English academic vocabulary has Spanish cognates, teaching the correspondence patterns unlocks hundreds of words at once, and false cognates should be taught explicitly because they produce predictable errors.

Translanguaging

Now the framework that has reshaped the conversation over the past fifteen years. Translanguaging, developed by Ofelia Garcia, Li Wei, and colleagues from earlier Welsh-language work by Cen Williams, describes how bilinguals actually use language: drawing fluidly on their whole repertoire rather than switching between two sealed systems.

It is important to separate two claims that travel under the same word.

The pedagogical claim is that classrooms should deliberately allow and design for use of the full repertoire. In practice this looks like: reading a source in Spanish and discussing it in English, brainstorming in the stronger language and drafting in the developing one, letting a student explain a concept to a partner in Vietnamese before writing about it in English, keeping bilingual glossaries and dual-language texts available. The pedagogical claim is not very controversial and has good practical support. It rests on an obvious point: forcing a student to think in a language they command poorly limits the thinking, not just the wording.

The theoretical claim is stronger. In its strong form, translanguaging theory holds that named languages such as English and Spanish are social and political constructions, and that a bilingual has one unitary linguistic repertoire rather than two grammars. This is a substantive claim about the mind, not a teaching strategy.

Key idea: Translanguaging has a pedagogical claim, that classrooms should design for the whole repertoire, and a stronger theoretical claim, that a bilingual has one unitary repertoire rather than two languages, and the two claims stand or fall separately.

The criticisms

Two lines of criticism deserve your attention, and neither comes from opponents of bilingual education.

Jeff MacSwan has argued from linguistic evidence that bilinguals do have distinguishable grammatical systems, pointing to the fact that code-switching is itself rule-governed in ways that require reference to two grammars, and that bilinguals can lose one language selectively after brain injury. He proposes a multilingual perspective on translanguaging that keeps the pedagogy while rejecting the unitary-repertoire claim.

The second criticism comes from bilingual educators worried about program integrity. In a dual language program, the minority language survives only because the schedule protects it. Since English is already dominant everywhere else in the child's life, a fluid-repertoire policy in practice often means more English, not more Spanish. Educators including Claudia Cervantes-Soon and colleagues have raised exactly this concern. The response from translanguaging scholars is that the framework requires designing for the minoritized language deliberately rather than simply removing rules, but the practical risk is genuine and worth naming.

Where this leaves you: use the pedagogy, which is well grounded and immediately useful. Hold the strong theoretical claim as contested. And if you work in a bilingual program, protect the minority language time on purpose, because gravity runs the other way.

Key idea: MacSwan argues bilinguals do have separable grammars, and bilingual educators warn that fluid language policy can quietly increase English at the expense of the minoritized language, so the pedagogy should be adopted with the protection of minority language time built in.

Common misconceptions

• Using the home language in class slows English. Interdependence and the program evidence both point the other way, given adequate English exposure.
• If skills transfer, English instruction can be minimal. Cummins's own statement conditions transfer on adequate exposure to and motivation for the second language.
• Everything transfers. Phonemes absent from the first language, English spelling, idioms, and differing word order all have to be taught.
• Cognates are a Spanish trick. They are a Latin inheritance, and the correspondence patterns generalize to hundreds of words.
• Translanguaging just means letting kids speak whatever they want. Designed translanguaging is planned; unplanned language mixing in a bilingual program tends to erode the minority language.

Recap

• Linguistic interdependence and the common underlying proficiency explain why first language literacy supports second language literacy.
• Concepts, print awareness, and comprehension strategies transfer; decoding transfers partially depending on orthography; phonemes, spelling, and idioms do not.
• English academic vocabulary overlaps heavily with Spanish, and teaching suffix correspondence patterns unlocks whole word families.
• False cognates cause predictable, teachable errors.
• Translanguaging's pedagogical claim is well supported; its unitary-repertoire theoretical claim is contested by MacSwan and others.
• In bilingual programs, minority language time must be protected deliberately.

Sources

1. Colorin Colorado. Using Cognates to Develop Comprehension in English. WETA. colorincolorado.org
2. Wikipedia contributors. Translanguaging. Wikipedia. en.wikipedia.org
3. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
4. WIDA. WIDA English Language Development Standards Framework, 2020 Edition. Wisconsin Center for Education Research. wida.wisc.edu
5. Encyclopaedia Britannica. Code-switching. britannica.com

The big picture

In 1970, roughly 2,800 students of Chinese ancestry in the San Francisco Unified School District did not speak English well enough to follow their classes. About 1,800 of them received no special help of any kind. The district's position was not hostile. It was that all students received the same textbooks, the same teachers, and the same curriculum, so all students were treated equally.

That argument is the one this whole module exists to answer, because it is superficially reasonable and it is wrong. Kinney Kinmon Lau and his fellow plaintiffs took it to the Supreme Court, and in 1974 the Court answered it unanimously.

Everything in American language education policy runs through four legal sources: one Supreme Court case about equal access, one federal statute that made the requirement permanent, one appellate case that supplied the practical test, and one Supreme Court case about who may enroll at all. This lesson teaches all four, in a form you can actually use in a meeting.

Lau v. Nichols (1974)

The Supreme Court ruled for the students unanimously. Justice William O. Douglas wrote the sentence that every ESL teacher eventually memorizes: there is no equality of treatment merely by providing students with the same facilities, textbooks, teachers, and curriculum, because students who do not understand English are effectively foreclosed from any meaningful education.

Three features of the decision matter more than the quotation.

The legal basis was statutory, not constitutional. The Court decided the case under Title VI of the Civil Rights Act of 1964, which prohibits discrimination on the basis of national origin by recipients of federal funds, and under the federal regulations implementing it. It did not reach the Equal Protection question. That choice mattered later, because Title VI enforcement depends on regulations and on an executive branch willing to enforce them.

The Court ordered no particular remedy. This surprises people. Lau does not require bilingual education. It does not require ESL. Douglas explicitly noted that teaching English to the students, or teaching Chinese to the teachers, or something else, were choices for the district, not the Court. Lau establishes a duty to act and leaves the method open.

The follow-on guidance came and went. In 1975 the federal Office for Civil Rights issued the Lau Remedies, informal guidelines that in practice pushed districts toward bilingual instruction. They were never properly promulgated as regulations and were withdrawn in 1980. That history is why the field ended up depending on a statute rather than on guidance documents.

Key idea: Lau v. Nichols held unanimously that identical treatment is not equal treatment for students who cannot understand the language of instruction, decided the case under Title VI rather than the Constitution, and left the choice of remedy to districts.

The Equal Educational Opportunities Act of 1974

Congress passed the Equal Educational Opportunities Act months after Lau, and one clause is the workhorse of the entire field. It provides that no state shall deny equal educational opportunity by failing to take appropriate action to overcome language barriers that impede equal participation by its students in its instructional programs.

Why this matters more than Lau in practice: it is a statute, so it does not depend on any agency's regulations or enforcement priorities, and it creates a private right of action, meaning families and advocates can sue directly. When you read about a district being sued over its services for English learners, the claim is usually under the EEOA, often alongside Title VI.

The statute contains a famous ambiguity. What is appropriate action? Congress did not say. A court had to.

Key idea: The Equal Educational Opportunities Act requires states and districts to take appropriate action to overcome language barriers, and because it is a statute enforceable by private lawsuit it is the practical backbone of language rights in schools.

Castaneda v. Pickard (1981) and the three prongs

In the Raymondville Independent School District in Texas, Roy Castaneda sued on behalf of his daughters, alleging that ability grouping segregated Mexican American children and that the district's bilingual program was inadequate. The Fifth Circuit had to decide what appropriate action means, and it produced a three-part test that has been adopted by courts and by the federal government ever since.

Prong	The question	What it rules out
1. Sound theory	Is the program based on an educational theory recognized as sound by at least some experts, or a legitimate experimental strategy?	Made-up approaches with no basis; note that the bar is deliberately low and permissive.
2. Effective implementation	Is the program actually implemented with adequate resources, qualified personnel, and appropriate materials?	A good model on paper with no trained teachers, no materials, and no time in the schedule.
3. Results	After a legitimate trial period, is the program actually producing results that show language barriers are being overcome?	A well-implemented program that demonstrably does not work, continued indefinitely.

Learn this test properly, because it is how any real dispute gets analyzed and because it is the most useful evaluation framework in the field even when nobody is suing.

Prong one is permissive on purpose. The court explicitly declined to make itself the arbiter of the bilingual education debate. Districts may choose their approach. This is the legal reason the United States has no national program model and why the answer to which program is required is: none of them, and Lesson 3.3 will show that the research answer is a different question entirely.

Prong two is where most districts actually lose. A district can adopt an excellent model and staff it with uncertified teachers, no curriculum, and thirty minutes a week. Choosing the model is not implementing it.

Prong three is the accountability clause. A district cannot run a program forever on the theory that it is sound. If English learners are not making progress in English or in content after a reasonable trial, the district has to change something.

Key idea: Castaneda's three prongs ask whether a program rests on sound theory, whether it is implemented with real resources and qualified staff, and whether after a legitimate trial it is producing results, and prong two is where districts most often fail.

Plyler v. Doe (1982)

Texas had authorized districts to deny enrollment to children not lawfully admitted to the United States, or to charge them tuition. In a five to four decision, the Supreme Court struck the statute down under the Equal Protection Clause of the Fourteenth Amendment, which protects any person within a state's jurisdiction, not only citizens. Justice Brennan's opinion emphasized that the children had no control over their presence and that denying them education would create a permanent underclass at enormous social cost.

Plyler means that a public school may not deny enrollment to a child based on immigration status. The federal government has issued repeated guidance about what follows in practice.

• Schools may not ask about immigration status or require documentation of it as a condition of enrollment.
• Schools may not require a Social Security number to enroll. A district may ask for a number for other purposes only if it explains that providing it is voluntary and that refusal will not affect enrollment.
• Schools may verify residency within the district, but must accept a range of documents and may not use residency verification as a proxy for status.
• Schools may not treat a child differently because a parent declines to provide a document that is not required.
• A birth certificate may be requested to verify age, but a foreign birth certificate must be accepted, and lack of one cannot bar enrollment.

The practical failure mode is rarely a district announcing that undocumented children may not enroll. It is a front office clerk with a checklist that asks for a Social Security number and a state-issued birth certificate, and a family that quietly leaves. Lesson 6.2 returns to this, including what a school may and may not do when immigration enforcement is a live concern.

Key idea: Plyler v. Doe guarantees undocumented children a free public K-12 education, so schools may not ask about immigration status, may not require a Social Security number, and may not use enrollment paperwork to discourage families.

The parent communication rule

One more requirement, less famous and constantly violated. Under Title VI and the EEOA, a district must communicate with limited English proficient parents in a language they can understand, about anything it communicates to other parents: enrollment, registration, grades, discipline, special education, gifted programs, sports, and permission forms.

This is a legal obligation, not a courtesy, and federal guidance is specific that it requires competent interpreters and translators. It also means that the routine practice of asking a bilingual student to interpret for their own parent at a school meeting is not compliance and is often actively harmful, a point Lesson 6.1 develops.

Key idea: Districts must provide meaningful communication to parents in a language they understand, using competent interpreters and translators rather than students or untrained bilingual staff.

Common misconceptions

• Lau requires bilingual education. It requires action to overcome the barrier and expressly leaves the method to districts.
• Lau was decided on constitutional grounds. It was decided under Title VI of the Civil Rights Act and its regulations.
• Castaneda's first prong is a high bar. It is deliberately permissive; prong two, implementation, is where districts usually fail.
• Schools may ask for immigration status if they promise not to act on it. They may not ask, because the question itself deters enrollment.
• Translating documents is a courtesy the district provides if it can. Meaningful communication with parents in a language they understand is required.

Recap

• Lau v. Nichols (1974) held unanimously that identical treatment is not equal access, under Title VI, without ordering a remedy.
• The Equal Educational Opportunities Act requires appropriate action to overcome language barriers and is privately enforceable.
• Castaneda v. Pickard supplies the three-prong test: sound theory, effective implementation, and demonstrated results.
• Plyler v. Doe (1982) guarantees enrollment regardless of immigration status.
• Schools may not require immigration documents or a Social Security number to enroll a child.
• Districts must communicate with parents in a language they understand, using competent interpreters.

Sources

1. U.S. Department of Justice and U.S. Department of Education. (2015). Dear Colleague Letter: English Learner Students and Limited English Proficient Parents. Office for Civil Rights. ed.gov
2. Wikipedia contributors. Lau v. Nichols. Wikipedia. en.wikipedia.org
3. Wikipedia contributors. Plyler v. Doe. Wikipedia. en.wikipedia.org
4. Colorin Colorado. Immigrant Students' Rights to Attend Public Schools. WETA. colorincolorado.org
5. Encyclopaedia Britannica. Bilingual education. britannica.com

The big picture

A category that carries legal rights has to have a door in and a door out. Both doors are more consequential than they look, and both are places where good intentions produce bad outcomes with some regularity.

Consider two failures that happen every year. In the first, a family enrolls a kindergartner, marks Spanish on a form they only half understand, and their fully English-proficient child spends the next several years classified as an English learner and pulled from the classes she liked. In the second, a family that speaks Haitian Creole at home marks English on the same form because they are worried about being singled out, and their son never receives services he genuinely needs.

Both failures begin with the same short questionnaire. This lesson follows the whole path: identification, notification, placement, annual testing, funding, and reclassification, with attention to where each step goes wrong.

The door in

Identification follows a standard sequence in nearly every state, though the details differ.

Step 1: the home language survey. At enrollment, every family completes a short form asking which languages are spoken in the home and by the child. Federal guidance expects districts to administer it to all students, not to families who look or sound like they might need it, because selecting families by appearance or surname is itself national origin discrimination.

The survey is the weakest link in the chain. It is often in English only. Its questions can be ambiguous. Families sometimes answer strategically, either overstating English to avoid a program they distrust or understating it hoping for extra help. And as Lesson 1.1 noted, a family that writes Spanish may actually speak an Indigenous language. A district that treats this form as authoritative rather than as a screen for further conversation will produce both false positives and false negatives.

Step 2: the screener. If the survey indicates a language other than English, the student takes a standardized English language proficiency screening assessment. The federal expectation is that this happens promptly, generally within 30 days of the start of the school year or within two weeks for a student enrolling mid-year. Common screeners include the WIDA Screener, used across the many states in the WIDA consortium, and the Initial ELPAC in California.

Step 3: classification and placement. If the screener score falls below the state's cut, the student is identified as an English learner and placed in the district's program. Note that the cut score is a state policy decision, not a scientific fact, and a student who is an English learner in one state may not be in another.

Step 4: parent notification. The district must notify parents, in a language they understand where practicable, of the identification, the level of the child's proficiency, the program the child will receive, and the exit expectations. The federal timeline is generally within 30 days at the start of the year, or two weeks for later enrollment.

Step 5: the right to decline. Parents may decline the language instruction program or particular services. Here is the part that gets missed: a parental refusal does not end the district's obligation. The student remains classified as an English learner, must still be assessed annually for English proficiency, and the district must still take appropriate action under the EEOA to make instruction accessible. The parent may waive a program. Nobody may waive access.

Key idea: Identification runs from a home language survey given to every family, to a proficiency screener within roughly 30 days, to classification, placement, and written parent notification, and a parent's refusal of a program does not relieve the district of its obligation to make instruction accessible.

Title III and where the money comes from

Title III of the Elementary and Secondary Education Act, called Language Instruction for English Learners and Immigrant Students, provides formula grants to states, which subgrant to districts based largely on counts of English learners and recent immigrant students.

Three practical facts about it. First, it is small. Title III typically amounts to a few hundred dollars per English learner per year, which does not fund a program. The core obligation is funded out of general and Title I dollars.

Second, it is supplemental. Title III money must supplement, not supplant, what the district is already required to do. A district cannot use Title III to pay for the appropriate action the EEOA requires anyway. In practice Title III funds professional development, supplementary materials, family engagement, and newcomer supports.

Third, and most important for understanding the current system: the Every Student Succeeds Act of 2015 moved English learner accountability out of Title III and into Title I. Under the older No Child Left Behind structure, English learner progress was measured through Title III with separate objectives. Under ESSA, states must include progress toward English language proficiency as an indicator in the main accountability system that applies to every school. That change is why building leaders who once ignored English learner data now watch it.

ESSA also required states to standardize entrance and exit procedures statewide. Before that, two districts in the same state could use different criteria, so the same child could be an English learner on one side of a county line and not on the other.

Key idea: Title III is a small supplemental grant that cannot pay for the district's baseline legal obligation, and the Every Student Succeeds Act moved English learner accountability into Title I while requiring statewide standardization of entrance and exit criteria.

Annual assessment and the two flexibilities

Every identified English learner must be assessed annually in listening, speaking, reading, and writing on the state's English language proficiency assessment. That is ACCESS for ELLs in WIDA states, the Summative ELPAC in California, TELPAS in Texas, NYSESLAT in New York, and so on. Lesson 5.1 covers how to read these scores.

Two accountability flexibilities exist for recently arrived students, and it is worth knowing them because they are frequently misunderstood as exemptions from instruction.

A student in the first year of enrollment in United States schools may be excused from taking the state reading or language arts assessment once, or may take it with results excluded from accountability calculations, at state option. Mathematics is generally still required, though results may be excluded in the first year. These are reporting flexibilities. They do not excuse anyone from teaching the student, and they do not last.

Key idea: All English learners are tested annually for English proficiency in four domains, and the limited first-year flexibilities for recently arrived students affect accountability reporting rather than the obligation to teach.

The door out

Reclassification, also called redesignation or exiting, is the formal determination that a student is now English proficient. States set the criteria, and they typically combine some of the following: an overall composite score at or above a cut on the annual proficiency assessment, sometimes with minimum domain scores; performance on a content or reading measure; teacher evaluation or input; and in some states parent consultation.

Under ESSA the criteria must be standardized statewide, but they still vary enormously between states. A composite of 4.5 on a six-point scale in one state and 4.0 in another sounds like a small difference. It is not. Studies of reclassification thresholds have found that moving a cut score changes the number of students who exit by large margins, and since reclassification changes services, course placement, and accountability reporting, the cut score is one of the most consequential numbers in state education policy that almost nobody outside the field has heard of.

After reclassification the student must be monitored, under ESSA generally for four years, and may be included in the English learner subgroup for accountability purposes during that window. If the student falters, the district should reconsider supports. In practice, monitoring is often a form somebody signs.

Key idea: Reclassification criteria are set by states and typically combine a proficiency composite with other measures, cut score choices dramatically change who exits, and reclassified students must be monitored for roughly four years.

Setting the bar: two ways to be wrong

This is the interesting part, and it is genuinely a tradeoff rather than a problem with an obvious answer.

Set the bar too low and students exit before their academic English can carry grade-level work. They lose designated language instruction, lose the legal hook that guarantees attention, and often stall. Reclassification is not a cure; it is the removal of a support.

Set the bar too high and students who could succeed without services stay classified for years. That is not harmless. Long-term classification is associated with restricted course access, since English learner sections often crowd out electives, world languages, and in some schools the college-preparatory sequence. Students report the stigma directly, particularly in middle and high school. Ilana Umansky and Sean Reardon's work in California found meaningful effects of classification itself on students' access to academic coursework, independent of their actual proficiency.

So the correct question is not how do we get students reclassified faster. It is what do we do so that classification does not cost a student access, and what do we do after reclassification so that exiting is not abandonment. A district serious about this schedules English learners into grade-level content, avoids using support periods to displace required or enriching courses, and treats the four-year monitoring window as an actual check rather than a signature.

Key idea: Exiting too early removes needed support while exiting too late restricts course access and carries stigma, so the design goal is a system in which classification does not cost access and reclassification is not abandonment.

Common misconceptions

• The home language survey should be given only to families who appear to need it. It goes to everyone; selecting by appearance or surname is discrimination.
• If parents decline services, the district's obligation ends. The student stays classified, is still tested annually, and the district still owes meaningful access.
• Title III pays for English learner programs. It is a small supplemental grant; the required program comes from general and Title I funds.
• First-year students are exempt from testing entirely. The flexibilities are limited and mostly concern accountability reporting rather than instruction.
• Faster reclassification is always better. Exiting a student who still needs support removes both the services and the legal attention.

Recap

• Identification runs from a universal home language survey to a screener, classification, placement, and written notice, on tight federal timelines.
• Parents may decline a program but cannot waive the district's duty to provide meaningful access.
• Title III is small and supplemental; ESSA moved English learner accountability into Title I and standardized state entrance and exit criteria.
• All English learners take an annual four-domain proficiency assessment; first-year flexibilities are about reporting, not teaching.
• Reclassification criteria vary by state, and the cut score choice moves large numbers of students.
• Both premature and delayed reclassification carry real costs, including restricted course access for long-classified students.

Sources

1. U.S. Department of Education. Title III, Part A: English Language Acquisition, Language Enhancement, and Academic Achievement. Office of Elementary and Secondary Education. ed.gov
2. U.S. Department of Justice and U.S. Department of Education. (2015). Dear Colleague Letter: English Learner Students and Limited English Proficient Parents. Office for Civil Rights. ed.gov
3. WIDA. WIDA Screener. Wisconsin Center for Education Research, University of Wisconsin-Madison. wida.wisc.edu
4. Colorin Colorado. ELL Program Basics. WETA. colorincolorado.org

The big picture

Suppose a district enrolls 900 English learners and has to decide what to do with them. It has, roughly, five options. Each option encodes a different theory about what the goal is, and the arguments among them have been running for sixty years, have been settled by ballot initiative twice, and have generated a research literature in which both sides accuse the other of methodological malpractice.

You need three things from this lesson: an accurate map of the models, an accurate account of what the evidence supports, and the ability to tell a strong study from a weak one. The last is the most valuable, because you will be handed both.

The five models

Model	Language of instruction	Goal	Typical duration
ESL pull-out	English only	English proficiency	30 to 60 minutes daily or a few times weekly
ESL push-in or co-taught	English only	English proficiency with content access	Ongoing within content classes
Structured or sheltered English immersion	English, adapted	English proficiency and content	One to several years
Transitional bilingual, early exit	Home language plus English, shifting to English	English proficiency; home language is a bridge	Two to three years
Developmental or dual language immersion	Both languages sustained, often 50-50 or 90-10	Biliteracy, bilingualism, and grade-level achievement in both	Six years or more, ideally K through 12

The critical distinction inside that table is not English versus bilingual. It is subtractive versus additive, the distinction from Lesson 1.2. Transitional bilingual programs use the home language and still aim to phase it out; their goal is English. Dual language programs aim to end with two languages. That is a difference of purpose, not just of schedule.

Two-way dual language programs enroll roughly half English learners and half English-proficient students who are learning the partner language, so each group is a language model for the other. One-way programs serve mostly students from the same language background. Two-way programs have grown quickly, partly because they attract English-speaking families, which brings both political durability and a real equity risk that Lesson 6.1 revisits.

A note on pull-out, since it is the most common model in districts with small English learner populations. It has the weakest evidentiary support of the five and two structural problems: whatever the student misses during the pull-out is missed, and the arrangement marks the student publicly, which matters more each year of adolescence. It exists because it is cheap and schedulable, not because anyone believes it is best.

Key idea: The five common models differ in language of instruction and in whether the home language is a bridge to be removed or a goal to be sustained, and pull-out ESL is the most common weak model because it is cheap and easy to schedule.

Thomas and Collier, and why the graphs are famous

Wayne Thomas and Virginia Collier published large longitudinal analyses in 1997 and 2002 drawing on hundreds of thousands of student records across multiple districts. Their headline finding, reproduced in a much-copied set of graphs, was that program models could be ranked by the achievement trajectory they produced, and that only two-way and one-way developmental bilingual programs enabled English learners to reach and stay at grade level in English by the end of schooling. English-only pull-out finished last.

These findings shaped a generation of advocacy. They also drew serious methodological criticism, and you should be able to state it, because citing Thomas and Collier to a skeptical audience without knowing the critique is a way to lose an argument you could have won.

The criticisms: the studies were largely self-published through a research center rather than in peer-reviewed journals; the district samples were not randomly selected and the districts that participated were those with well-developed bilingual programs; students were not randomly assigned to models, so families who chose dual language may have differed systematically in motivation, stability, and home resources; and the aggregation of normal curve equivalents across different tests and districts was questioned. Selection bias is the heart of it: if the students in the best-looking program chose it, the program's apparent advantage may be partly the students.

On the other side, Christine Rossell and Keith Baker published a 1996 review concluding that the evidence for bilingual education was weak. That review drew equally sharp criticism for how studies were coded and included, and for treating some structured immersion programs in Canada, which serve majority-language students, as comparable to programs serving minoritized-language students in the United States. The National Research Council in 1997 assessed the whole literature and reached a conclusion that satisfied nobody: the research was too limited in quality to support strong claims either way, and the question as posed, which model is best, was probably the wrong question.

Key idea: Thomas and Collier's influential findings favoring developmental bilingual programs rest on non-randomized, largely self-published data with a real selection bias problem, and the leading skeptical review has its own serious methodological criticisms.

The study that changed the argument

The selection problem has an obvious remedy: randomize. For years nobody could, because you cannot ethically assign children to programs at random. Then it turned out that districts had been doing it all along, through oversubscribed lotteries.

Jennifer Steele and colleagues at RAND studied Portland Public Schools, where dual language immersion programs were oversubscribed and admission was determined by lottery. Because the lottery is random, students who won and lost are comparable in every respect including family motivation, so the comparison isolates the program effect.

The results, published in 2017, favored immersion. Students randomly assigned to dual language immersion outperformed their lottery-losing peers in English reading, by a margin equivalent to several months of learning by fifth grade and growing by eighth grade. Mathematics and science effects were small and not consistently significant. Among students classified as English learners at kindergarten entry, immersion students were more likely to be reclassified by the end of middle school.

Two things make this important. First, it is causal evidence in a literature dominated by correlational evidence. Second, it points in the same direction as the weaker studies, which is the pattern you hope for: a strong design confirming what a messy literature suggested. Ilana Umansky and Sean Reardon separately found in California that dual language students reclassified more slowly in the early grades and reached higher rates by high school along with stronger academic outcomes, which is exactly the shape you would predict if early English-only gains are real but shallow.

Be precise about the limits. This is one district, with programs that were popular enough to be oversubscribed, in Spanish, Japanese, and Mandarin, implemented well. It shows that well-implemented immersion can work, not that any program calling itself immersion will.

Key idea: The randomized lottery study of Portland's dual language immersion programs found meaningful English reading advantages by grades five and eight, providing causal evidence that points the same way as the earlier correlational literature.

Proposition 227 and what happened next

In 1998 California voters passed Proposition 227, the English for the Children initiative promoted by software entrepreneur Ron Unz, with about 61 percent of the vote. It required that English learners be taught overwhelmingly in English, generally through a structured English immersion period intended to last about one year, after which students would move to mainstream classes. Parents could seek waivers for bilingual instruction, but the process was deliberately burdensome.

The effect on programs was immediate. Enrollment in bilingual education in California fell from roughly 30 percent of English learners to under 10 percent within a few years. Arizona passed a similar Proposition 203 in 2000 with stricter enforcement, and Massachusetts passed Question 2 in 2002.

What did it demonstrate? Less than either side claimed, and here is where you learn to read a natural experiment carefully. Test scores for English learners in California rose after 1998. Supporters treated this as vindication. But scores rose for nearly every student group in the state during the same period, including groups the initiative did not touch, and California was simultaneously implementing new standards, new tests, and class size reduction. A five-year evaluation led by Thomas Parrish and colleagues found no conclusive evidence that one program model was superior under the new regime and found that the achievement gap between English learners and English-proficient students had not meaningfully narrowed. The honest reading is that a policy which restructured programs statewide did not produce a detectable improvement in the outcome it targeted.

The reversal came from the same voters. In 2016 California passed Proposition 58 with roughly 73 percent of the vote, repealing most of 227's restrictions and restoring local authority over program models. Massachusetts repealed Question 2 through the LOOK Act in 2017. The turnaround in California, from 61 percent against bilingual instruction to 73 percent for local choice in eighteen years, is one of the largest reversals of public opinion in American education policy, and it tracks the growth of dual language programs popular with English-speaking families.

That last clause deserves a hard look. Bilingual education became politically safe again partly because affluent English-speaking parents wanted it for their own children. That is a real gain for English learners and a real warning, and Lesson 6.1 returns to the question of who dual language programs end up serving.

Key idea: Proposition 227 restructured California's programs in 1998 without producing clear gains, and voters reversed it in 2016 by a wide margin, with the reversal driven substantially by the popularity of dual language programs among English-speaking families.

What to conclude

The National Academies consensus report of 2017 offers the defensible summary. Programs that provide instruction in the home language do not impede and often support English development and academic achievement. There is no evidence that English-only instruction produces better long-term outcomes. Quality of implementation, teacher preparation, materials, and time matter as much as the model label.

Which means the practical question in your building is usually not which model is theoretically best. It is whether the model you actually have is implemented with certified teachers, real materials, adequate time, and grade-level content, which is the second Castaneda prong wearing a different hat. A well-implemented pull-out program helps students more than a poorly implemented dual language program, even though the reverse holds when both are done well.

Key idea: The consensus is that home language instruction does not impede English and often helps, that no evidence favors English-only for long-term outcomes, and that implementation quality often matters more than the model label.

Common misconceptions

• Transitional bilingual and dual language are the same thing. One phases the home language out; the other sustains it as a goal.
• Thomas and Collier settled the question. Their work is influential and has real selection bias problems; the randomized Portland study is the stronger evidence.
• Proposition 227 proved English immersion works. Scores rose statewide for nearly all groups during a period of multiple reforms, and the official evaluation found no conclusive advantage.
• Bilingual education is politically dead. California repealed its own ban by 73 percent in 2016 and dual language programs have expanded rapidly.
• Choosing the right model is the main decision. Implementation quality frequently outweighs the model label.

Recap

• Five common models range from English-only pull-out to sustained dual language immersion.
• The key distinction is subtractive versus additive: whether the home language is a bridge or a goal.
• Thomas and Collier favored developmental bilingual programs but faced selection bias criticism; skeptical reviews had their own flaws.
• A randomized lottery study in Portland found real English reading advantages for dual language immersion by grades five and eight.
• Proposition 227 restructured California programs in 1998 without clear gains and was largely repealed by Proposition 58 in 2016.
• The National Academies conclusion is that home language instruction does not impede English and that implementation quality is decisive.

Sources

1. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
2. Steele, J. L., et al. (2017). Effects of dual-language immersion programs on student achievement: Evidence from lottery data. American Educational Research Journal. RAND Corporation summary. rand.org
3. Wikipedia contributors. California Proposition 227 (1998). Wikipedia. en.wikipedia.org
4. Colorin Colorado. Program Models for Teaching English Language Learners. WETA. colorincolorado.org
5. Encyclopaedia Britannica. Bilingual education. britannica.com

The big picture

Here is the sentence that organizes this entire module. The job is not to make the work easier. The job is to make the work possible.

Those sound similar and they are opposites. Making work easier means shorter texts, fewer questions, simpler ideas, and a student who spends four years doing third-grade tasks. Making work possible means the same grade-level idea, the same intellectual demand, with the language load supported so the student can get at it. The first path produces a diploma with nothing behind it. The second produces a student who can do calculus in a language they learned four years ago.

The technical name for the second path is sheltered instruction: teaching grade-level content in English while deliberately supporting the language demands. This lesson covers how to do it, the model most widely used to organize it, and what the evidence for that model actually shows.

Making input comprehensible, concretely

Krashen's comprehensible input, from Lesson 2.1, is a slogan until you can list the moves. Here are the ones that survive contact with a real classroom.

Show, do not only tell. Images, objects, video with captions, gestures, diagrams, demonstrations. A photograph of a levee is worth more than the sentence a levee is an embankment built to prevent flooding, and it costs nine seconds.

Slow slightly; do not speak louder. A natural but unhurried rate with clear articulation and pauses at phrase boundaries helps. Volume does not, and increasing it is a habit worth breaking because it embarrasses students. Also cut the noise: reduce idioms, avoid the passive when the active will do, replace pronouns whose referents are three sentences back with the actual noun.

Chunk and check. Deliver two or three minutes of new content and then ask for something back, from everyone: a hand signal, a whiteboard, a turn-and-talk, a written line. The purpose is to catch the moment comprehension fails, because after it fails, everything downstream is noise.

Extend wait time. Average teacher wait time after a question is around one second. Try five. It feels unbearable and it is transformative, because a multilingual learner is parsing the question, retrieving content, and formulating language in a second language, and one second is not enough for step one.

Front-load selectively. Pre-teach the three or four words without which the lesson collapses, not the twenty words on the chapter list. Front-load background knowledge too: a student who does not know what the Dust Bowl was cannot infer it from a passage that assumes it.

Preview, view, review. Where a shared language is available, a brief preview in the home language before an English lesson and a brief review after dramatically increases what the student takes from the middle. This is translanguaging from Lesson 2.3 in practical form, and it works whether the preview comes from a bilingual aide, a peer, a translated summary, or a two-minute video.

Key idea: Comprehensible input in practice means visuals and demonstration, a slightly slower and cleaner speech stream rather than a louder one, frequent chunk-and-check cycles, five seconds of wait time, selective front-loading, and preview and review in the home language where possible.

Scaffolding, and how it goes wrong

Scaffolding comes from Jerome Bruner and colleagues, building on Vygotsky's zone of proximal development: temporary support that lets a learner do now, with help, what they will soon do alone. The word temporary is doing all the work in that sentence.

The standard toolkit: graphic organizers that make text structure visible, sentence frames and starters that supply the syntax of an academic move, word banks, modeled examples and annotated exemplars, think-alouds that make expert reasoning audible, partner rehearsal before writing, and structured note-taking.

Now the failure mode, which is common enough to have a name. A scaffold that never comes down becomes a ceiling. A student who has used the same four sentence frames for two years has learned four sentences, not a way of writing. Aida Walqui's distinction is useful here: scaffolding is a process of handover, not a set of structures you hand out. If your fifth-year English learners are receiving the same supports as your newcomers, that is not differentiation, it is a stalled system, and it is one of the mechanisms that produces the long-term English learners from Lesson 1.1.

Build removal into the plan. Three practical ways: use a frame for the first two examples and require an unframed third; give the organizer completed, then half-completed, then blank; move from a full model text to a checklist of the moves the model made. If you cannot say when a support is scheduled to come off, it will not come off.

One more distinction that matters legally and instructionally. A scaffold is temporary support toward independence in the same task. An accommodation is a change in how a student accesses or demonstrates learning that does not change the standard. A modification changes the standard itself. Sheltered instruction is about the first two. Routinely modifying content for English learners, which is to say lowering the standard, is the most common way a well-meaning teacher does long-term harm.

Key idea: Scaffolds are temporary by definition, a scaffold that never comes down becomes a ceiling, and lowering the standard is a modification rather than a scaffold and should not be the default response to a language barrier.

The SIOP model

The Sheltered Instruction Observation Protocol, developed by Jana Echevarria, MaryEllen Vogt, and Deborah Short, began as a research observation instrument and became the most widely used framework for planning sheltered lessons. It organizes thirty features into eight components.

Component	The core question
Lesson preparation	Are content and language objectives defined, posted, and tied to grade-level standards?
Building background	Are links made to students' experience and prior learning, and is key vocabulary emphasized?
Comprehensible input	Is speech appropriate, are tasks explained clearly, and are multiple techniques used?
Strategies	Are learning strategies taught, is scaffolding used, and are higher-order questions asked?
Interaction	Do students talk substantially, in varied groupings, with enough wait time and home language support?
Practice and application	Do students use hands-on materials and apply content and language knowledge together?
Lesson delivery	Are objectives actually met, is pacing appropriate, and are students engaged most of the period?
Review and assessment	Are vocabulary and concepts reviewed, is feedback regular, and is comprehension checked throughout?

The feature most people take away is the language objective paired with the content objective. The content objective says what students will know or do about the subject. The language objective says what they will do with language to get there. They are not the same sentence with different verbs.

Content objective	Language objective
Explain how the water cycle moves water among reservoirs.	Orally describe the cycle to a partner using sequence connectives: first, then, next, as a result.
Compare two accounts of the Boston Massacre.	Write two sentences of comparison using whereas and however, citing one detail from each account.
Solve two-step equations with rational coefficients.	Justify each step aloud using because and therefore, naming the operation applied to both sides.

Writing these is a genuine skill and it improves teaching even when nobody is observing, because it forces you to notice the language a task actually demands. Most teachers discover, the first time they do this honestly, that their lesson silently required a form of language they never taught.

Key idea: SIOP organizes sheltered instruction into eight components, and its most useful single practice is pairing a content objective with a specific language objective naming the language students must use to reach it.

What the SIOP evidence shows

Honest reporting, since you will hear SIOP described as research-based without qualification.

Studies conducted by the model's developers have reported positive effects on English learners' writing and academic performance in schools where teachers implemented SIOP at high fidelity. Those are real findings, and developer-conducted evaluations are also the weakest kind, for obvious reasons.

Independent evidence is thinner. A randomized controlled trial of SIOP-based professional development conducted through the CREATE center found small and mixed effects on student outcomes, with implementation fidelity varying considerably across teachers. Fidelity is the recurring theme: a couple of days of training does not produce the practice the protocol describes.

The reasonable conclusion is not that SIOP does not work. It is that SIOP is a well-organized synthesis of instructional practices that are individually supported, delivered through a professional development model whose independent effects are modest. Use it as a planning checklist. Do not treat it as a proven intervention, and be skeptical of a district that buys two days of training and expects transformation.

Key idea: SIOP synthesizes individually supported practices, developer studies report positive effects, independent randomized evidence on SIOP professional development is small and mixed, and implementation fidelity is the persistent problem.

Common misconceptions

• Sheltered instruction means simplified content. It means supported access to the same grade-level content.
• Speaking louder helps. It does not, and it embarrasses students; adjust rate, clarity, and idiom instead.
• Scaffolds are supports you provide indefinitely. A scaffold that never comes down is a ceiling.
• A language objective is a rephrased content objective. It names the specific language function, structure, or vocabulary the task requires.
• SIOP is proven to raise achievement. It organizes supported practices; independent evidence on the professional development model is modest.

Recap

• The goal is to make work possible, not easier: same standard, supported language load.
• Comprehensible input in practice is visuals, cleaner speech, chunk-and-check, five-second wait time, and selective front-loading.
• Preview, view, review in the home language increases uptake from English instruction.
• Scaffolds must be planned for removal; distinguish scaffold, accommodation, and modification.
• SIOP organizes eight components, and paired content and language objectives are its highest-value practice.
• Developer studies of SIOP are positive; independent randomized evidence is small and fidelity-dependent.

Sources

1. Colorin Colorado. Sheltered Instruction and the SIOP Model. WETA. colorincolorado.org
2. TESOL International Association. The 6 Principles for Exemplary Teaching of English Learners. tesol.org
3. WIDA. WIDA English Language Development Standards Framework, 2020 Edition. Wisconsin Center for Education Research. wida.wisc.edu
4. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
5. Wikipedia contributors. Sheltered instruction. Wikipedia. en.wikipedia.org

The big picture

Read this sentence, which is the kind that appears in an ordinary eighth grade textbook.

The subsequent redistribution of population following the mechanization of agriculture contributed to the rapid expansion of urban centers.

Now count what a student has to handle. Four abstract nouns built out of verbs: redistribution, mechanization, expansion, and arguably population. A causal relationship expressed by the verb contributed to rather than by a word like because. A time relationship carried by subsequent and following, neither of which is a common conversational word. Twenty-one words and one clause, so the information is packed into noun phrases rather than spread across sentences. And no human being anywhere ever says this out loud.

That is academic language. It is not simply hard words. It is a different way of building meaning, and if your instruction consists of pre-teaching mechanization and hoping, you have addressed maybe fifteen percent of the difficulty. This lesson takes the other eighty-five seriously.

Three levels of academic language

Mary Schleppegrell's functional linguistics work gives the most useful frame: academic language operates at the level of words, sentences, and whole discourse.

Word level. General academic vocabulary such as analyze, factor, consequence, distribution, and significant. Discipline-specific technical terms such as photosynthesis, isotope, or federalism. And the sneaky category, everyday words used technically, discussed below.

Sentence level. Nominalization, which turns processes into things: people moved becomes migration, and once it is a noun it can be counted, modified, and made the subject of a new sentence. Dense noun phrases: the rapid expansion of urban centers is one grammatical slot carrying four content words. The passive voice, which lets a writer delete the actor: mistakes were made. Complex connectives: whereas, nevertheless, insofar as. Hedging: tends to, appears to, may suggest, which is how academic writers mark uncertainty and which students frequently read as weakness.

Discourse level. How whole texts are organized, and the organization differs by discipline. A lab report, a document-based essay, a mathematical proof, and a literary analysis each move in a shape their readers expect. Rhetorical conventions differ across cultures too, so a student who learned to write beautifully organized essays in another tradition may be marked down for structure that is not wrong, only different. That is worth saying out loud to the student rather than circling in red.

Key idea: Academic language works at three levels, and the sentence-level features of nominalization, dense noun phrases, passive constructions, connectives, and hedging carry more of the difficulty than the technical vocabulary does.

Which words to teach

You cannot teach every unfamiliar word, so selection is the skill. The tier framework from Isabel Beck, Margaret McKeown, and Linda Kucan is the standard tool.

Tier	Description	Examples	Instructional priority
Tier 1	Common everyday words	dog, run, happy, table	Usually known by English speakers; often unknown to multilingual learners and worth checking
Tier 2	General academic words used across subjects	analyze, contrast, evidence, factor, distribution	Highest leverage; teach these deliberately
Tier 3	Low-frequency, discipline-specific terms	mitosis, isotope, gerrymander	Teach in context when the content requires them

Here is the adjustment that matters for multilingual learners. For a native English speaker, Tier 1 is free and Tier 2 is the gap. For a multilingual learner, Tier 1 is not free at all. A student may confidently discuss photosynthesis and not know the word stem, branch, or soil. Tier 3 words often get taught because they are in bold; Tier 1 words get skipped because nobody imagines they need teaching. Check the Tier 1 words in your own materials sometime; the result is sobering.

Averil Coxhead's Academic Word List, 570 word families that occur across academic texts in every discipline, is a good place to source Tier 2 words. Roughly ten percent of the words in academic texts come from this list, which is a large return on a finite set.

Key idea: Tier 2 general academic words are the highest-leverage instructional target, and for multilingual learners the Tier 1 everyday words that nobody thinks to teach are a frequent hidden barrier.

The polysemy problem

The single most underestimated vocabulary difficulty in school is the everyday word with a technical meaning. Students know the word. That is precisely the trap, because knowing it prevents them from asking.

Word	Everyday meaning	Academic meaning
table	furniture	an organized display of data
mean	unkind, or to signify	the arithmetic average
power	strength or authority	an exponent; also energy per unit time
root	part of a plant	a solution to an equation; a morphological base
volume	loudness	the space an object occupies
product	an item for sale	the result of multiplication
foot	body part	a unit of length; a metrical unit in poetry
plot	the events of a story	to place points on a coordinate grid; a piece of land

Two teaching moves handle this. First, name it explicitly: this word has an everyday meaning you already know and a mathematics meaning, and today we mean the second one. Second, make it a running game. Have students collect double-meaning words across subjects on a wall or a shared document. It takes almost no class time and it changes how students read, because they start noticing when a familiar word is behaving strangely.

Key idea: Polysemous words are unusually dangerous because students know the everyday meaning and therefore do not ask, so the technical sense must be named explicitly rather than assumed.

Teaching words so they stick

Copying dictionary definitions does not work, and the reason is worth stating: dictionary definitions are written for people who already know the word, which is why they use harder words than the entry.

What does work, based on the vocabulary research and consistent across reviews:

Student-friendly explanations. Not obstinate: stubbornly adhering to an opinion. Instead: someone who is obstinate refuses to change their mind even when everyone shows them they are wrong.

Multiple exposures in varied contexts. Estimates of the exposures needed for durable ownership commonly range from around six to more than a dozen, and they need to be different contexts, not the same sentence six times. This is why the word must appear again next week and the week after, not only on Tuesday.

Examples and non-examples. The Frayer model, a four-square with definition, characteristics, examples, and non-examples, works because the non-example box forces boundary thinking. Knowing that a whale is not a fish is part of knowing what a fish is.

Generative use. Students must produce the word in speech and writing under conditions that require the meaning. A sentence like I saw a distribution yesterday satisfies an assignment and demonstrates nothing. A prompt like describe how the distribution changed after 1950 and explain one cause requires the concept.

Visual and physical anchors. Images, gestures, and a word wall that students actually use. A word wall that is decoration is furniture. A word wall students are required to consult during writing is a tool.

Volume of reading. Explicit instruction can reach maybe eight to ten words a week done properly, which is a few hundred a year. The vocabulary of an educated adult is tens of thousands. The arithmetic is unforgiving: most words are learned incidentally through large amounts of reading and listening, so the explicit list and the volume of comprehensible text are not competing priorities.

Key idea: Words stick through student-friendly explanations, many exposures in varied contexts, examples and non-examples, required generative use, and a large volume of reading, because explicit instruction alone can only reach a few hundred words a year.

Making vocabulary generative: morphology and cognates

The highest-leverage move of all is teaching students to attack unfamiliar words themselves.

Morphology. English academic vocabulary is built from a modest set of Greek and Latin roots and affixes. A student who owns trans, port, and -ation can approach transportation, transport, portable, deport, and export. Teach word families explicitly: analyze, analysis, analytical, analytically. Teach that -tion and -ment make nouns from verbs, that -ize makes verbs, that -al and -ous make adjectives, because that knowledge unlocks the nominalizations from the sentence level discussed above.

Cognates. Lesson 2.3 covered this in detail, and it belongs in every vocabulary routine for Spanish speakers and to a lesser degree for speakers of other Romance languages. The routine is simple: when a new academic word appears, ask whether anyone's language has something that looks like it. This costs fifteen seconds and it teaches students to keep asking.

Context strategies, honestly assessed. Students should learn to use context, and you should know that context is a weak tool: natural texts often provide insufficient or misleading clues, and research on incidental word learning from context finds relatively low pickup per encounter. Teach context use as one strategy among several rather than as the answer.

Key idea: Teaching morphological patterns and cognate awareness makes vocabulary instruction generative, letting students attack new words independently, while context clues alone are a weaker tool than they are usually presented to be.

Common misconceptions

• Academic language means hard vocabulary. Sentence structure and discourse organization carry more of the load.
• Tier 1 words need no attention. Everyday words are a frequent hidden barrier for multilingual learners.
• Students who know a word will ask if the meaning shifts. With polysemous words they will not, because they believe they already know it.
• Looking up and copying definitions builds vocabulary. Dictionary definitions assume the reader already knows the word.
• Guessing from context is the main way to learn words. Context clues are unreliable; volume of reading matters, but so does explicit teaching and morphology.

Recap

• Academic language operates at word, sentence, and discourse levels, and the sentence level is where most difficulty lives.
• Nominalization, dense noun phrases, passives, connectives, and hedging are the recurring sentence-level features.
• Tier 2 general academic words are the highest-leverage target; Tier 1 words are a hidden barrier for multilingual learners.
• Polysemous words require explicit naming of the technical sense.
• Effective instruction uses student-friendly explanations, many varied exposures, non-examples, and required generative use.
• Morphology and cognate awareness make vocabulary learning generative; context clues alone are weak.

Sources

1. Colorin Colorado. Vocabulary Development with ELLs. WETA. colorincolorado.org
2. WIDA. WIDA English Language Development Standards Framework, 2020 Edition. Wisconsin Center for Education Research. wida.wisc.edu
3. Wikipedia contributors. Academic Word List. Wikipedia. en.wikipedia.org
4. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
5. TESOL International Association. The 6 Principles for Exemplary Teaching of English Learners. tesol.org

The big picture

There is a persistent staffroom belief that mathematics is the safe subject for English learners because numbers are universal. It is wrong in an interesting way, and taking it apart is the fastest route into everything else in this lesson.

Consider: Maria has 5 fewer marbles than Jose. Jose has 12. How many does Maria have? The mathematics is a single subtraction. The language is brutal. The word fewer signals subtraction, but the sentence names Maria first and asks about Maria, so a student who follows word order subtracts from the wrong number. The comparative structure X has n fewer than Y is uncommon in speech. And the whole thing hinges on one word that is not a number.

Every discipline has a language, each one different, and this lesson gives you the specific demands of the big three plus literacy itself. Technology comes last, because it is a tool for the preceding sections rather than a subject of its own.

Mathematics

Four distinct demands, each with a response.

Comparative and relational language. More than, fewer than, twice as many, at most, at least, exceeds, is exceeded by, the difference between. These structures are dense, low-frequency in conversation, and often reverse the order of the operation. Teach them as language explicitly, with pairs of contrasting sentences, and not only when a word problem appears.

The technical register. The polysemous words from Lesson 4.2 cluster in mathematics: table, mean, power, root, product, base, plot, degree, prime, volume, odd, even, rational, irrational, real, imaginary. Note how many of them sound like judgments. A student can reasonably wonder why some numbers are irrational and whether that is bad.

Word problems as a genre. Word problems have conventions: a scenario, a question, irrelevant details sometimes included on purpose, and a requirement to translate prose into symbols. The Three Reads protocol is the standard response and it is genuinely good. Read one, for the situation: what is happening here, with no numbers mentioned. Read two, for the question: what exactly are we being asked. Read three, for the information: what quantities and relationships do we have. Only then compute. This slows students down, which is the entire point.

Notation is not universal. This one surprises teachers. In much of Latin America and Europe, the decimal separator is a comma and the thousands separator is a period, so 1.234 may mean one thousand two hundred thirty four. Long division is arranged differently in several countries. Some students learned to write division with a colon rather than a division sign. A student who is getting arithmetic wrong may be reading your notation correctly under a different convention. Ask before diagnosing.

The strategies that work in mathematics are the ones that generate talk: number talks where students explain reasoning aloud, sentence stems for justification such as I know because and this cannot be right since, requiring students to explain a partner's method rather than only their own, and problem sets where students write a word problem to match an equation, which is the inverse task and reveals misunderstanding fast.

Key idea: Mathematics carries heavy language demands in comparative structures, a technical register full of polysemous words, the word problem genre, and notation conventions that differ internationally, so the response is explicit language teaching plus routines that require students to talk about reasoning.

Science

Science is, counterintuitively, one of the friendlier subjects for multilingual learners, for a structural reason: it comes with objects. A phenomenon on the lab bench provides shared reference that a text cannot. This is a context-embedded, cognitively demanding task in Cummins's terms from Lesson 2.2, which is the most productive quadrant there is.

Use that. Do the phenomenon first and name it after. A student who has watched a candle go out under a jar has a referent for combustion and oxygen that no definition supplies. Inquiry sequences such as the 5E model, engage, explore, explain, elaborate, evaluate, are well suited here because explanation follows experience rather than preceding it.

The language of science has its own signature and it is worth teaching directly. Scientific writing nominalizes relentlessly: the solution evaporated becomes evaporation of the solution. It uses the passive to remove the actor: the mixture was heated. It hedges: these results suggest. It uses precise verbs whose distinctions matter: observe, infer, predict, conclude, hypothesize. Students routinely use observe for infer, which is not a vocabulary slip but a scientific error, and it is teachable.

Practical moves: word walls with images rather than definitions, lab procedures rewritten as numbered steps with a diagram per step, sentence frames for claim, evidence, and reasoning, and pairing a newcomer with a partner during the hands-on portion where doing carries the load and talk can be minimal at first.

Key idea: Science provides shared physical reference that makes demanding content comprehensible, so the phenomenon should come before the vocabulary, while the nominalized, passive, hedged register of scientific writing needs explicit teaching.

Social studies

This is the hardest subject for multilingual learners and it is worth being blunt about why.

First, the text load is enormous and the texts are dense. Second, and more importantly, social studies runs on assumed background knowledge that is culturally specific. A passage about the Constitutional Convention assumes you know what a colony is, roughly where Philadelphia is, what a legislature does, and why taxation without representation was a grievance. A student educated in Honduras or Somalia may have excellent general knowledge and none of that particular knowledge, and no amount of vocabulary support fixes a missing frame.

Third, primary sources are written in older English, which is genuinely hard for everyone and doubly so for a learner. Fourth, the discourse involves argument, causation, multiple perspectives, and evaluation of sources, which is the most demanding academic discourse in the curriculum.

What helps: build background before the text, not after, using images, maps, short video, and a timeline; use text sets that approach one topic at several reading levels so every student engages the same content; teach one primary source excerpt deeply rather than three shallowly, with a modernized paraphrase beside the original rather than instead of it; make chronology visible, because a student new to United States history has no mental timeline to hang events on; and use the student's own national history as a bridge, since revolutions, constitutions, and migrations exist everywhere and a student who can compare two of them is doing exactly the disciplinary thinking you want.

One caution. Building background is not the same as assuming ignorance. A student from Vietnam knows a great deal about the war your textbook has a chapter on, and quite possibly a different account of it. Asking is better than telling.

Key idea: Social studies is the heaviest language load in the curriculum because it assumes culturally specific background knowledge, so build the frame first with visuals, timelines, and text sets, and treat students' own histories as assets rather than gaps.

Literacy

The National Literacy Panel on Language-Minority Children and Youth, summarized by August and Shanahan in 2006, remains the anchor synthesis, and its findings are unusually clean.

Word-level instruction works about the same. Explicit, systematic instruction in phonological awareness, phonics, and word reading benefits multilingual learners much as it benefits monolingual learners. The idea that these students should wait for oral proficiency before receiving decoding instruction is not supported.

Comprehension is where the gap appears. Multilingual learners frequently reach grade-level word reading and remain well below in comprehension. That should not be surprising after Lesson 4.2: decoding is a code problem and comprehension is a language and knowledge problem.

Oral English proficiency predicts comprehension. Vocabulary and listening comprehension are the main levers, which means oral language development is reading instruction, not a separate activity that competes with it.

First language literacy helps. Consistent with the transfer research in Lesson 2.3.

Two profiles need different responses. A student with strong first language literacy needs English sound-symbol mapping, vocabulary, and syntax; do not put them in a beginning reading program. A student with little or no literacy in any language, often an older newcomer with interrupted schooling, needs genuine foundational decoding instruction, and the crucial requirement is that it be age-respecting. A fifteen-year-old learning to decode should not be reading about a cat on a mat. Materials designed for older beginning readers exist, and if you cannot find them, texts the student generates from their own experience through a language experience approach are better than picture books about ducks.

Key idea: Explicit word-reading instruction works similarly for multilingual learners, comprehension is where the gap opens, oral language and vocabulary are the main levers, and older beginning readers need foundational instruction in age-respecting materials.

Technology, used well

Some genuinely useful tools, with their limits.

Machine translation. Useful for a quick gloss of a word, for a student to grasp a paragraph, and for informal family communication. Its limits are real: quality varies enormously across language pairs and drops sharply for less-resourced languages, idioms and technical terms often fail, and the output can be confidently wrong. Never use it unreviewed for legal notices, special education documents, or anything a family must rely on, which is the subject of Lesson 6.1.

Captions and transcripts. Captioned video in English pairs the sound stream with the print stream and is well supported for language learning. Transcripts let a student reread what was said.

Text-to-speech and speech-to-text. The first supports access to text above a student's decoding level; the second lets a student who can compose orally get ideas down before wrestling with spelling. Use both as scaffolds with a plan for removal.

Digital texts at multiple levels. Useful for text sets. Watch the ceiling problem: a student permanently assigned the lowest level never meets grade-level text.

The overall caution is the same one that applies to every tool in education. An application is not a program. Software that gives a student thirty minutes of adaptive practice a day does not satisfy the Castaneda prongs, does not replace a teacher, and will not by itself teach anyone academic English.

Key idea: Translation tools, captions, text-to-speech, and leveled digital texts are useful scaffolds with specific limits, and no software package substitutes for a program of instruction.

Common misconceptions

• Mathematics is language-free. Comparative structures, technical polysemy, word problem conventions, and even notation differ and must be taught.
• Vocabulary should precede the science phenomenon. The phenomenon supplies the referent that makes the vocabulary meaningful.
• Social studies is easy because it is stories. It carries the heaviest language and background knowledge load in the curriculum.
• Multilingual learners should wait for oral proficiency before phonics. Word-level instruction works and should not be delayed.
• An older student learning to read needs beginner materials. They need foundational skills in age-respecting content.

Recap

• Mathematics demands comparative language, a polysemous register, word problem conventions, and awareness that notation varies internationally.
• Science offers shared physical reference; do the phenomenon first and teach the nominalized, passive, hedged register explicitly.
• Social studies assumes culturally specific background knowledge, so build the frame with visuals, timelines, and text sets.
• Word-reading instruction works similarly for multilingual learners; comprehension is where the gap opens.
• Oral language and vocabulary are the main levers on comprehension, so oral work is reading instruction.
• Technology scaffolds access; machine translation is unreliable for anything a family must rely on, and no app is a program.

Sources

1. Colorin Colorado. Math Instruction for English Language Learners. WETA. colorincolorado.org
2. Colorin Colorado. Reading Comprehension Strategies for English Language Learners. WETA. colorincolorado.org
3. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org
4. U.S. Department of Education, Office of English Language Acquisition. English Learner Toolkit. ncela.ed.gov
5. WIDA. WIDA English Language Development Standards Framework, 2020 Edition. Wisconsin Center for Education Research. wida.wisc.edu

The big picture

A science teacher hands back a test. A student who understands cell division scored 40 percent. The teacher writes: does not understand mitosis. The truth is closer to: the student could not read the question about mitosis.

Somewhere in that gap sits the central problem of assessing multilingual learners. Almost every test given in an American school is partly a test of English, whether or not that is what it is called, and unless you can say which part is which, you cannot interpret any of the numbers you receive about these students.

This lesson has three parts: how English proficiency itself is measured, how to think about content assessment when language is developing, and the grading argument, which is genuinely unresolved and which you will have to take a position on.

English language proficiency assessment

Federal law requires annual assessment of every identified English learner in listening, speaking, reading, and writing. States meet it through one of a handful of systems.

WIDA, based at the Wisconsin Center for Education Research at the University of Wisconsin-Madison, serves a large consortium of states and territories. Its annual assessment is ACCESS for ELLs, and its screener is the WIDA Screener. WIDA reports six proficiency levels:

Level	Name	What it roughly looks like
1	Entering	Single words, memorized phrases, heavy reliance on visuals
2	Emerging	Phrases and short sentences on familiar topics; frequent errors
3	Developing	Expanded sentences and simple paragraphs; general and some specific vocabulary
4	Expanding	Connected discourse with a variety of sentence types; technical vocabulary emerging
5	Bridging	Approaching grade-level academic language across domains
6	Reaching	Comparable to English-proficient peers at grade level

Two features of the score report matter. First, the overall composite is weighted, not an average: reading and writing count 35 percent each and listening and speaking 15 percent each. This is why a student who converses fluently can hold a low composite, and it is a common source of confusion at meetings. Second, WIDA publishes Can Do Descriptors, which translate each level into what a student can be expected to do at each grade band, and they are far more useful for planning than the number is. The 2020 WIDA framework also organizes language around four Key Language Uses, narrate, inform, explain, and argue, which map neatly onto the language objectives in Lesson 4.1.

ELPAC is California's system, with an Initial ELPAC for screening and a Summative ELPAC annually, reporting four levels from minimally developed to well developed. Texas uses TELPAS, New York uses NYSESLAT, and several states use ELPA21. The architecture is similar everywhere: four domains, an annual administration, and a composite compared against a state cut score.

Three cautions when reading any of these. A single composite hides domain profiles, and the profile is the useful part; a student at level 4 in speaking and level 2 in writing needs something different from a student who is level 3 across the board. These assessments measure academic English, so they are not a measure of intelligence, content knowledge, or effort. And levels are not grade equivalents; a level 3 fourth grader and a level 3 tenth grader have very different needs even though the number matches.

Key idea: Annual four-domain proficiency assessments such as ACCESS for ELLs and the Summative ELPAC report a weighted composite and domain scores, and the domain profile and Can Do descriptors are far more useful for instruction than the composite number.

Why a content test in English is partly a language test

Measurement people have a precise name for the science teacher's problem: construct-irrelevant variance. A test is supposed to measure one thing, the construct. When score differences arise from something else, the test is measuring that other thing too and its validity for those students is compromised.

Jamal Abedi's research program made this concrete for English learners. Analyzing large assessment datasets, he found that the language complexity of test items affected English learners' performance substantially and affected English-proficient students much less. When items were linguistically modified, meaning the unnecessary language difficulty was reduced while the content demand was held constant, English learners' scores improved and English-proficient students' scores were largely unchanged. That pattern is the signature of removing construct-irrelevant variance rather than making a test easier.

What does unnecessary language look like in a mathematics item? Low-frequency vocabulary that has nothing to do with the mathematics, passive constructions, long noun phrases, conditional clauses, irrelevant contextual detail, and names or scenarios that assume cultural knowledge. A question about calculating change at a garage sale is testing arithmetic plus knowledge of what a garage sale is.

Key idea: A content test in English measures both content and English, and Abedi's work shows that reducing unnecessary linguistic complexity raises English learner scores without raising others, which is the signature of improved validity rather than an easier test.

Accommodations, and what the evidence says about them

An accommodation must change access without changing the construct. That single rule resolves most cases.

Accommodation	Typical status	Notes
Extended time	Widely allowed	The most common; research finds modest effects alone
Word-to-word bilingual glossary	Commonly allowed	Translates only; no definitions or examples
Translated test or translated directions	Sometimes allowed for mathematics; rarely for reading	Requires that the student be literate in the other language, which is often untested
Read-aloud of directions	Commonly allowed	Directions are usually not the construct
Read-aloud of a reading passage	Not allowed on a reading test	It changes the construct from reading to listening
Small group or separate setting	Widely allowed	Reduces anxiety and distraction
Simplified English version	Varies by state	Defensible when the construct is held constant, as in Abedi's studies

Three practical rules. First, an accommodation used only on test day is close to useless; a student who has never used a glossary will not use one under pressure, so accommodations must be routine in instruction. Second, a translated test helps only a student who reads that language well, and the assumption that any Spanish speaker reads academic Spanish is frequently wrong. Third, more is not better; stacking five accommodations on a student can add confusion rather than access.

Key idea: An accommodation must preserve the construct, must be used routinely in instruction rather than introduced on test day, and must match the student's actual profile, including whether the student can read the language of a translation.

The grading argument

Now the part where reasonable people disagree, so you get the arguments rather than a verdict.

The situation: a student writes a biology answer that is scientifically correct and grammatically broken. What is the grade?

Position one: grade content only. The course is biology. Grammar is not the construct. Penalizing English proficiency in a content grade means the transcript records language, not learning, and it can suppress grade point average, course access, and eligibility, all on the basis of national origin, which is why this is not purely a pedagogical question. Federal civil rights guidance is clear that English learners must have meaningful access to content, and a grading practice that systematically converts limited English into low content grades is a fair target for a complaint.

Position two: grades that do not reflect readiness mislead. If a student receives a B in biology and cannot write a paragraph about biology, the family, the next teacher, and the student all receive false information, and the reckoning arrives later, in a college course or a job. Grade inflation as compassion has a poor record.

The synthesis most districts land on. Separate the reporting. Grade content on content, using assessments that let students demonstrate knowledge in ways that do not require unnecessary English: labeled diagrams, oral explanations, structured responses, demonstrations, work in the home language where someone can read it. Report language development separately, through the proficiency assessment and through explicit language feedback. Give feedback on language always, and grade it only when language is the stated objective, which it legitimately is in an English language development course or on a writing standard.

Two things nearly everyone agrees on. Do not average early-semester scores from a newcomer's first weeks into a final grade; that measures arrival date. And do not use a zero for work a student could not access, because a zero on a hundred-point scale is a mathematically enormous penalty that no amount of later learning can offset.

Key idea: The defensible middle position is to grade content on content while reporting language separately, since converting limited English into low content grades misleads on the transcript and creates civil rights exposure, while inflated grades mislead the student.

Formative assessment, which matters more

Everything above concerns scores that arrive too late to help. Day to day, what you need is information now.

Useful moves: response systems where everyone answers, such as whiteboards or hand signals, rather than the volunteer who always knows; exit tickets that ask for one sentence using a target structure, which assesses content and language together; structured observation of student talk during partner work, which is the only place you will hear a newcomer's actual oral production; and periodic writing on the same prompt across months, which shows growth that a single sample cannot. Keep samples. A student's September and February paragraphs side by side are more persuasive to a family, and more informative to you, than any composite score.

Key idea: Formative assessment through universal response, exit tickets with target structures, observation of student talk, and dated writing samples gives you information in time to act on it.

Common misconceptions

• The proficiency composite is an average of the four domains. Reading and writing are weighted much more heavily than listening and speaking.
• A proficiency level describes overall ability. It describes academic English, not intelligence, content knowledge, or effort.
• Simplifying test language makes the test easier. When the construct is held constant it removes irrelevant difficulty, which is a validity improvement.
• Any Spanish speaker benefits from a Spanish test. Only a student who reads academic Spanish does.
• Grading generously is the compassionate choice. It misinforms the student and the next teacher; separate reporting of content and language is the better answer.

Recap

• All English learners take an annual four-domain proficiency assessment; WIDA reports six levels and ELPAC four.
• The WIDA composite weights reading and writing at 35 percent each and listening and speaking at 15 percent each.
• Domain profiles and Can Do descriptors are more useful for planning than the composite.
• Construct-irrelevant variance means content tests in English partly measure English; linguistic modification improves validity.
• Accommodations must preserve the construct, be routine before test day, and fit the student's actual profile.
• The defensible grading position separates content grades from language reporting, and avoids zeros and early-arrival averaging.

Sources

1. WIDA. ACCESS for ELLs. Wisconsin Center for Education Research, University of Wisconsin-Madison. wida.wisc.edu
2. WIDA. Can Do Descriptors. Wisconsin Center for Education Research. wida.wisc.edu
3. California Department of Education. English Language Proficiency Assessments for California (ELPAC). cde.ca.gov
4. U.S. Department of Justice and U.S. Department of Education. (2015). Dear Colleague Letter: English Learner Students and Limited English Proficient Parents. Office for Civil Rights. ed.gov
5. Colorin Colorado. Assessment and ELLs. WETA. colorincolorado.org

The big picture

Two students, two mistakes, both common, both expensive.

The first is a second grader who arrived from El Salvador eighteen months ago. She reverses letters, cannot hold a sequence of sounds in mind, and is not learning to decode despite good instruction. Her teacher raises a concern. The team says: give it time, she is still learning English, we should wait until she has been here five years. She is not evaluated until sixth grade. By then she has four years of failure behind her and a settled belief that she is stupid.

The second is a fourth grader in the same building. He speaks Spanish at home, is behind in reading, and his class receives thirty minutes of undifferentiated whole-group instruction from a teacher with no preparation in language development. He is evaluated, tested largely in English, and found to have a specific learning disability. What he has is an instructional history.

Those are the two failure directions, and the research shows both happening at once, in the same system, sometimes in the same building. This lesson is about how to think clearly here. It is also the lesson where the limits of a text course are sharpest, because the actual determination is a team decision made by trained professionals, and nothing here qualifies anyone to make it.

The evidence on identification patterns

The pattern that shows up most consistently in the research is a grade-level one. Multilingual learners tend to be under-identified in the early grades and over-identified in the later grades, relative to comparable peers.

Elizabeth Samson and Nonie Lesaux found under-representation of English learners in special education in kindergarten and first grade, with the pattern reversing by third grade. Ilana Umansky, Karen Thompson, and Guadalupe Diaz found in California data that English learners were substantially over-represented in special education by the secondary grades, and that a large share of that gap emerged after the early elementary years.

The mechanism is not mysterious once you see it. Early on, adults attribute every difficulty to English, so genuine disabilities go unaddressed during exactly the years when intervention works best. Later, when a student is still behind and the language explanation has worn thin, a referral finally happens, and by then the student has accumulated years of gaps that look like a disability whether or not one exists. Delay manufactures the appearance it was supposed to rule out.

Rates also vary enormously by state, district, and category, which is itself evidence that identification reflects local practice rather than a fixed feature of children. Categories that depend on professional judgment, such as specific learning disability and speech or language impairment, show the most variation. Categories with clear biological markers do not.

Key idea: Multilingual learners are typically under-identified in the early grades and over-identified by the secondary grades, and delay in the early years produces gaps that later look like the disability the delay was meant to rule out.

Why this is genuinely hard

Be sympathetic to the difficulty, because the confusion is not stupidity. Typical second language acquisition produces a list of behaviors that overlaps heavily with the referral checklist for a learning disability or a speech and language impairment.

What you observe	Typical of second language acquisition	Also seen with a disability
Limited vocabulary	Yes, in English, while the first language is age-appropriate	Yes, in both languages
Grammatical errors	Yes, patterned and developmental	Yes, often unusual or persistent beyond typical stages
Word-finding difficulty	Yes, in the newer language	Yes, in both languages
Slow processing, long pauses	Yes, translation and formulation take time	Yes, but present in the first language too
Appearing not to attend	Yes, sustained listening in a second language is exhausting	Yes, across settings and languages
Not speaking	Yes, during a silent period of weeks to months	Selective mutism, hearing loss, trauma, or other causes if prolonged
Difficulty following directions	Yes, when directions are multi-step and rapid	Yes, including when directions are given in the first language

Read the right two columns together and the resolution appears. Almost every row separates on the same question: does the difficulty appear in both languages, and is it unusual compared with students who share this student's background? Those two questions do most of the work.

Key idea: Typical second language acquisition mimics many disability indicators, and the two questions that separate them are whether the difficulty appears in both languages and whether it is unusual relative to students with the same language and schooling background.

True peers

The comparison group is the single most important methodological idea here. Comparing a multilingual learner to the class average tells you almost nothing, because the class average includes students who have been learning academic English since birth.

A true peer, in the sense used by Julie Esparza Brown and Jennifer Doolittle, is a student who shares the relevant characteristics: roughly the same first language, roughly the same English proficiency level, a comparable amount of time in United States schools, a comparable schooling history including any interruption, and the same instructional program.

Now ask the question that matters. Among ten students who arrived at the same time, from similar schooling, at similar proficiency, in the same program, is this student progressing like the others, or is this student the clear outlier? If eight of the ten are reading and two are not, you have something worth investigating. If all ten are behind, you probably have an instructional problem, and referring one child for evaluation will not fix a program.

Finding true peers takes work in a district with three Karen-speaking students. Do the best you can, be explicit about the limits, and when the comparison group is thin, weight the cross-language evidence more heavily.

Key idea: Progress should be judged against true peers who share language, proficiency, time in United States schools, schooling history, and program, and when a whole group is struggling the problem is usually instruction rather than disability.

What IDEA actually requires

Three provisions govern this, and a teacher who knows them can ask better questions in a meeting than most people in the room.

The exclusionary factors. A child must not be determined eligible for special education if the determinant factor is limited English proficiency, or lack of appropriate instruction in reading, including the essential components of reading instruction, or lack of appropriate instruction in mathematics. This clause forces the team to rule out language and instructional history before locating the problem inside the child. Note what it does not say: it does not say a student learning English cannot have a disability. Many do.

Assessment requirements. Evaluations must be administered in the child's native language or other mode of communication, and in the form most likely to yield accurate information about what the child knows and can do, unless it is clearly not feasible. Assessments must be selected and administered so as not to be discriminatory on a racial or cultural basis, must be validated for the purpose used, and no single measure may determine eligibility.

The team. Eligibility is decided by a group including the parents and qualified professionals. In practice, a defensible team for a multilingual learner includes someone with real expertise in second language acquisition, and the parents must be able to participate meaningfully, which under the law of Lesson 3.1 means competent interpretation, not a bilingual cousin in the hallway.

One widespread misconception deserves killing here. Nonverbal cognitive assessments are frequently described as culture-free or language-free. They are neither. Instructions are still given in a language, task formats still assume familiarity with testing conventions such as timed puzzle completion, and item content still carries cultural loading. Reduced-language measures are useful and are not neutral instruments.

Key idea: IDEA forbids eligibility when limited English proficiency or inadequate instruction is the determinant factor, requires assessment in the language and form most likely to yield accurate information, and requires a team that includes the parents, with nonverbal measures understood as reduced-language rather than culture-free.

A defensible process

What a careful district actually does, roughly in order.

First, verify that instruction has been appropriate: certified staff, grade-level content, language development actually delivered, and attendance sufficient to have received it. If that is not established, the evaluation cannot answer the question.

Second, gather a full history: languages spoken and by whom, prior schooling and any interruptions, literacy in the first language, medical and hearing history, and the family's own observations. Families are the best available source on whether a difficulty shows up at home in the first language, and asking them is both good practice and required participation.

Third, deliver targeted intervention and monitor it with data over a defined period, comparing progress with true peers. Note the tension here: intervention data are valuable, and endless intervention is a delay tactic. Both are true, and the resolution is a defined timeline with a decision point, not an open-ended wait.

Fourth, if concerns persist, evaluate, with assessment in both languages where possible, interpreters where needed, and a team that can interpret the results in light of language history.

And throughout, remember what the outcome means. If a student is found eligible, the student is dually identified and receives both special education services and English language development. Neither replaces the other. That was true in Lesson 1.1 and it is the most commonly violated rule in this entire area.

Key idea: A defensible process verifies that instruction was adequate, gathers full language and schooling history from the family, uses monitored intervention with a defined decision point rather than an open-ended wait, and ends with both sets of services if the student is found eligible.

What you can do, and what you cannot

As a classroom teacher you can and should document specifics: what the student does in each language, with dates and examples, rather than adjectives. You can ask the family what they notice at home. You can compare with true peers. You can make a referral, which is a request for consideration, not a diagnosis. You can insist that a language acquisition expert be at the table.

You cannot determine that a student has a disability, and you cannot determine that a student does not. Both of those are team decisions requiring formal evaluation. The most valuable thing an ordinary teacher contributes to this process is careful, dated, concrete observation, and it is astonishing how rarely it is available when a team finally sits down.

Key idea: Teachers document specifics in both languages, consult families, compare with true peers, and refer, while eligibility determinations belong to a formally constituted team.

Common misconceptions

• You must wait several years before considering a disability. No rule requires waiting, and delay produces the gaps it was meant to rule out.
• A student learning English cannot be found eligible. The law bars eligibility only when limited English proficiency is the determinant factor.
• Nonverbal tests are culture-free. They are reduced-language measures that still carry instructions, task conventions, and cultural content.
• The class average is the right comparison. True peers sharing language, proficiency, time, schooling history, and program are.
• Once a student has an IEP, language services can stop. Dually identified students are entitled to both, always.

Recap

• Multilingual learners are under-identified early and over-identified later, with wide variation by state and district.
• Second language acquisition mimics many disability indicators; the separating questions are both-language presence and true peer comparison.
• True peers share first language, proficiency, time in United States schools, schooling history, and program.
• IDEA bars eligibility when limited English proficiency or inadequate instruction is the determinant factor.
• Evaluations must use the language and form most likely to yield accurate information, with no single measure deciding.
• Teachers document and refer; teams determine eligibility; students found eligible receive both sets of services.

Sources

1. U.S. Department of Education. Sec. 300.306 Determination of eligibility. IDEA Regulations, Part B. sites.ed.gov
2. U.S. Department of Education. Sec. 300.304 Evaluation procedures. IDEA Regulations, Part B. sites.ed.gov
3. Colorin Colorado. Special Education and English Language Learners. WETA. colorincolorado.org
4. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org

The big picture

A middle school holds a family night. Flyers go home in English and Spanish two days ahead. The event runs from six to seven on a Tuesday. Eleven families attend out of four hundred. In the staff meeting afterward, someone says the sentence you will hear in every school in the country: these parents just are not involved.

Now run the same evening from the other side. A mother finishes a shift at 6:30 and has no car. A father works nights and sleeps in the evening. A family with three children has no childcare. A family read the flyer, understood perhaps half of the Spanish, which was machine translated, and could not tell whether attendance was expected or optional. A family from a country where a summons to the school means your child is in trouble decided not to risk it. A family speaks Mam and got a flyer in Spanish.

Not one of those is a family that does not care. Every one of them is a design failure, and every one of them is fixable. This lesson is about the two halves of that fix: teaching in a way that treats students' cultures and languages as resources, and building family partnership that survives contact with real lives.

From relevant to sustaining

Three related terms circulate, and the differences are substantive rather than fashionable.

Culturally relevant pedagogy, from Gloria Ladson-Billings in 1995, rests on three criteria that must all be present. Students must experience academic success, because a warm classroom that teaches nothing is not relevant to anything. Students must develop or maintain cultural competence, meaning they are not asked to trade their home culture for school success. And students must develop critical consciousness, the capacity to notice and question inequity. That third criterion is the one most often quietly dropped, and dropping it turns the framework into decor.

Culturally responsive teaching, associated with Geneva Gay and later Zaretta Hammond, emphasizes using students' cultural knowledge, prior experience, and communication styles to make learning more effective, with attention to the relationship and the cognitive work.

Culturally sustaining pedagogy, from Django Paris in 2012 and developed with H. Samy Alim, presses a further point. Responsive and relevant approaches can still treat the home culture as a bridge to the dominant one, useful on the way and discarded on arrival. Sustaining pedagogy says the goal is for students to be fully successful in school and to keep their languages and cultures, as living, changing things rather than museum pieces. Applied to this course, it is the same argument as additive rather than subtractive bilingualism from Lesson 1.2, extended past language to culture.

What all three rule out is what practitioners call the tourist approach, or the four Fs: food, festivals, flags, and fashion. A heritage month display is not nothing, and it is not pedagogy. The test is simple and uncomfortable: does anything about the curriculum, the texts, the examples, or the intellectual work change, or only the bulletin board?

Key idea: Culturally relevant pedagogy requires academic success, cultural competence, and critical consciousness together, and culturally sustaining pedagogy adds that the home culture is a goal to maintain rather than a bridge to be discarded.

Funds of knowledge

The most practical version of these ideas came from Luis Moll, Cathy Amanti, Deborah Neff, and Norma Gonzalez in 1992. Working in Tucson, they sent teachers into students' homes not to advise families but to learn, as researchers. What they found in working-class Mexican American households was substantial accumulated expertise: construction, auto mechanics, farming, herbal medicine, ranching, sewing, midwifery, small business, transnational logistics, and the management of complex extended networks.

They called this the household's funds of knowledge, and the instructional point follows immediately. A student whose family runs a small landscaping business has a parent who prices jobs, calculates areas, converts units, manages seasonal cash flow, and reads soil. That is a mathematics curriculum sitting in a kitchen. A teacher who knows it can build a unit on it. A teacher who does not know it teaches ratios with a worksheet about a lemonade stand.

You do not need a formal home visit program to use this, though those exist and are powerful. A ten-question interest and expertise inventory at the start of the year, offered in the home language and answerable by any family member, gets you most of the way. Ask what work people in the family do, what they know how to make or fix, what languages are spoken by whom, what the student is responsible for at home, and what the family would like you to know. Then actually use two of the answers in the first month, so the survey is not another form.

Key idea: Funds of knowledge names the substantial expertise held in every household, and instruction that draws on it converts family experience into curriculum rather than treating home life as a deficit to compensate for.

Why family engagement fails, and what works

The research on immigrant family engagement, including Concha Delgado-Gaitan's long work with Latino families, converges on a finding that ought to change the staffroom conversation: immigrant parents overwhelmingly hold high educational aspirations for their children and frequently do not participate in the ways schools recognize as participation.

Several reasons, all more useful than the deficit explanation.

Structural barriers. Shift work, second jobs, no car, no childcare, no paid leave. A meeting at 6 p.m. on a weekday is not neutral; it selects for a particular kind of employment.

Different role expectations. In many countries, education is the teacher's professional domain and the parent's job is to send a respectful, prepared child. A parent who says you are the teacher, you decide is expressing confidence in you, not disengagement. Reading that as apathy is a cultural misreading, and it is extremely common.

Prior experience of school. A parent whose own schooling was brief, or humiliating, or conducted in a language they did not speak, walks into a building carrying that. So does a parent from a system where being called to school means trouble.

Language and information. If the flyer, the phone tree, the portal, and the front desk operate in English, a family is excluded from the information other families act on.

What works is unglamorous and well documented. Meet at times and places that fit the actual schedules, including mornings, weekends, and neighborhood locations. Provide childcare and food; both raise attendance sharply and communicate something. Make communication two-way, so families are asked what they need rather than only told what the school wants. Build on trust brokers, whether a bilingual liaison, a community organization, a faith community, or the parent who everyone already calls. Send positive contact before you ever need to send negative contact, because a first phone call that is not about a problem changes every call after it. And be specific: tell a family exactly what you would like them to do, since read with your child twenty minutes a night is actionable and be more involved is not.

Key idea: Immigrant families hold high aspirations and face structural, cultural, and informational barriers, so engagement improves through accessible times and places, childcare and food, two-way communication, trusted intermediaries, and early positive contact.

Interpreters and translation, done correctly

Lesson 3.1 established the legal requirement: districts must communicate with parents of limited English proficiency in a language they understand. Here is how to satisfy it in practice.

First, know the distinction. Interpretation is oral and happens in real time. Translation is written. They are different skills, and a person who is good at one may not be good at the other.

Second, a qualified interpreter is not simply any bilingual adult. Interpretation requires fluency in both languages, command of the relevant terminology, which for a special education meeting is considerable, and training in the ethics of the role: interpret everything, add nothing, omit nothing, keep confidentiality, and stay out of the conversation as a participant. A bilingual custodian pulled from the hallway is a kind person in an unfair position.

Third, and most important: do not use children as interpreters. This applies with special force to the student's own siblings and to the student themselves. Federal guidance is explicit that districts should not rely on students. The reasons are cumulative. Accuracy suffers, because a fifth grader does not have the vocabulary for individualized education program or reclassification criteria. Confidentiality collapses, since a child should not be conveying information about a parent's finances, immigration situation, or another child's disability. The family hierarchy inverts, with the child holding the information and therefore the power. And the burden is real: children who interpret for families describe the responsibility as heavy, sometimes for decades.

Fourth, working with an interpreter well is a learnable skill. Speak to the parent, not to the interpreter, using you rather than tell her. Speak in short segments and pause. Avoid idioms and acronyms, or explain them, since IEP, ELPAC, and RTI mean nothing translated literally. Allow more time; an interpreted meeting takes roughly twice as long, and scheduling it for the usual thirty minutes guarantees a bad meeting. Brief the interpreter beforehand on terminology and purpose. And check understanding by asking the parent to say back what will happen next, rather than asking do you understand, which almost always gets a yes.

Fifth, machine translation. Useful for informal notes and quick comprehension, as Lesson 4.3 said. Not acceptable unreviewed for legal notices, evaluation reports, IEPs, or discipline documents. Have a qualified human review anything a family must rely on.

Key idea: Interpretation and translation are distinct professional skills, students must not be used as interpreters for accuracy, confidentiality, and family-role reasons, and working with an interpreter requires short segments, no idioms, double the time, and confirmation of understanding.

One honest complication about dual language programs

Before leaving this lesson, a point that belongs to equity rather than to instruction. Two-way dual language programs have grown quickly, in part because English-speaking families want them. Guadalupe Valdes raised a cautionary note about this as early as 1997, and the concern has aged well.

When a program becomes desirable to advantaged families, several things can happen. Admission processes that require applications, transportation, or information favor families with the resources to navigate them. The program's culture can shift toward the needs and comfort of English-speaking families. Instruction can drift toward English. And a program founded to serve minoritized-language students can end up delivering its main benefit to students who already had advantages, while the students whose language it borrowed remain in the same building receiving less.

None of this is an argument against dual language programs, which the evidence in Lesson 3.3 supports. It is an argument for watching the enrollment data, the language of the hallway, the composition of the parent organization, and who is actually being served. Ask those questions early, because they are much harder to raise once a program is popular.

Key idea: Two-way dual language programs can drift toward serving advantaged English-speaking families, so equitable access, program culture, and language balance need deliberate monitoring rather than assumption.

Common misconceptions

• Low attendance at school events measures how much families care. It measures whether the event was designed around their lives.
• A parent who defers to the teacher is disengaged. In many educational cultures that deference is a sign of respect and confidence.
• Cultural responsiveness means celebrating heritage months. The test is whether the curriculum and intellectual work change, not the bulletin board.
• Any bilingual adult can interpret. Interpretation requires terminology, ethics, and training, and translation is a separate skill.
• Using a student to interpret is efficient. It is inaccurate, breaches confidentiality, inverts family roles, and federal guidance advises against it.

Recap

• Culturally relevant pedagogy requires academic success, cultural competence, and critical consciousness together.
• Culturally sustaining pedagogy treats home languages and cultures as goals to maintain, not bridges to discard.
• Funds of knowledge turns household expertise into curriculum rather than treating home life as a deficit.
• Engagement barriers are structural, cultural, and informational, and the fixes are scheduling, childcare, two-way communication, trust brokers, and early positive contact.
• Qualified interpreters are required, children must not interpret, and interpreted meetings need double the time.
• Two-way dual language programs require active monitoring so they continue to serve the students they were built for.

Sources

1. U.S. Department of Justice and U.S. Department of Education. (2015). Dear Colleague Letter: English Learner Students and Limited English Proficient Parents. Office for Civil Rights. ed.gov
2. Colorin Colorado. Engaging ELL Families: Twenty Strategies for School Leaders. WETA. colorincolorado.org
3. Wikipedia contributors. Culturally relevant teaching. Wikipedia. en.wikipedia.org
4. TESOL International Association. The 6 Principles for Exemplary Teaching of English Learners. tesol.org
5. National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. The National Academies Press. nationalacademies.org

The big picture

A sixteen-year-old enrolls in November. He has completed the equivalent of fifth grade, in a language he reads haltingly. He has no English. He is placed in ninth grade because of his age, in a schedule of five classes taught from texts written at a tenth grade level, and he needs 24 credits to graduate before he ages out at 21.

Do the arithmetic honestly. Even under favorable assumptions, this student needs foundational literacy, English, five years of missing content, and 24 credits, in under five years. There is no schedule of ordinary high school classes that produces that outcome, and pretending otherwise is how schools end up with a student who accumulates failing grades for three years and then leaves.

This lesson deals with the students for whom the standard system does not work as designed: newcomers generally, students with interrupted schooling specifically, students carrying serious stress, and students living with immigration uncertainty. It is also the lesson with the most explicit limits on what you personally should do, and those limits are the point rather than a disclaimer.

Newcomer programs

A newcomer program is a short-term, intensive setting for recently arrived students, typically one year or less, designed to build English, orient students to United States schooling, and begin content. Three common designs:

Model	Structure	Tradeoff
Separate newcomer center	A distinct site serving newcomers from across a district	Concentrated expertise and materials; isolation from English-speaking peers and from the home school community
Program within a school	A cluster of newcomer sections inside a regular school	Keeps students in their community; harder to staff and easy to underresource
Sheltered course sequence	Newcomers take sheltered versions of required courses alongside regular enrollment	Best integration and credit accrual; demands the most from content teachers

Whatever the model, four design questions separate the good ones from the holding pens. Does the program award credit, or does a year in it cost a year toward graduation? Is there a written plan for transitioning out, with a timeline, rather than an indefinite stay? Do students have any contact with English-speaking peers, since a program in which everyone is a beginner provides limited language models? And is content actually taught, or is the year spent entirely on English while five subjects of grade-level knowledge go by?

Key idea: Newcomer programs are short-term intensive settings that succeed when they award credit, plan the exit, provide peer contact, and teach content rather than deferring it.

Students with interrupted formal education

SLIFE, defined in Lesson 1.1, are students who arrive with age-appropriate ability and significantly less schooling than their age would predict, whether from conflict, displacement, migration, rural isolation, economic necessity, gender restrictions on schooling, or disability. Several states define the category in policy, typically as two or more years of interruption plus below-grade literacy.

What these students need is specific, and none of it is on a standard schedule.

Foundational literacy in age-respecting materials. Covered in Lesson 4.3, and it is the non-negotiable one. A student who cannot decode will not learn biology from a textbook, and giving a sixteen-year-old a book about a puppy is a way of losing him in a week.

Explicit teaching of schooling itself. This is the part experienced teachers underestimate. A student who has never been in a formal classroom does not know that the bell means move, that a syllabus is a promise about the future, that homework is expected without a reminder, that raising a hand is how you speak, that a locker combination is a thing, or that a multiple choice bubble sheet is scored by a machine that cannot read the answer written beside it. None of this is intelligence. All of it is culture, and it can be taught in a week if someone decides to teach it.

Credit and time realism. Most states permit enrollment through age 21, and knowing your state's rule matters. So does knowing what alternatives exist: credit recovery, competency-based credit, adult education, a high school equivalency route, community college bridge programs. Presenting these honestly to a family is not lowering expectations, it is refusing to let a student spend four years failing before anyone mentions there was another path.

Compressed, prioritized content. Nobody teaches five missing years. Teams that do this well identify the highest-leverage prerequisite knowledge in each subject and teach that, deliberately and fast, rather than starting at the beginning of a fifth grade curriculum.

Key idea: Students with interrupted schooling need foundational literacy in age-respecting materials, explicit instruction in the conventions of schooling, honest planning about credits and time, and prioritized rather than sequential content.

Trauma-aware practice, without over-reaching

Some of your students will have experienced things you have not. Refugee students may have survived war, camp life, or the loss of family. Unaccompanied children may have made a long journey alone. Others carry the ordinary and serious grief of leaving everyone they knew. And students born in the United States can carry as much as any newcomer.

Start with the correction that matters most: most people who experience trauma do not develop a disorder. Resilience is the statistically normal outcome. Approaching every refugee student as damaged is its own kind of harm, it is inaccurate, and students notice.

What stress can look like in a classroom: difficulty concentrating, hypervigilance and startling at noise, exhaustion from poor sleep, irritability or sudden anger, withdrawal, physical complaints, memory difficulty, and avoidance of specific triggers such as sirens, uniforms, or certain topics in a history unit. Note that this list overlaps with the second language acquisition list in Lesson 5.2 and with attention difficulties, which is one more reason no single observer should be drawing conclusions.

What actually helps is almost boringly structural. Predictability: a posted routine, warning before changes, the same greeting daily. Safety: a classroom where nobody is humiliated and where a student knows what happens next. Relationship: one adult in the building who knows the student's name, pronounces it correctly, and notices absence. Regulation: permission to take a break in a defined way before escalation rather than after. Choice: small genuine choices restore a sense of control to someone who has had little. Connection to the wider school through a club, a sport, or a job.

Now the limits, stated plainly. Do not ask a student to tell you their story. Do not assign personal narrative writing that requires disclosure of the journey; offer alternatives without requiring an explanation. Do not diagnose, and do not describe a student as traumatized in writing or in a meeting on your own observation. Do refer to the school counselor, social worker, or psychologist, and do learn what community mental health services exist in languages your families speak, which is usually a short list and worth knowing before you need it. If a student discloses something involving current danger or abuse, follow your mandatory reporting obligations, which are not optional and are not softened by cultural difference.

Key idea: Resilience is the normal outcome after adversity, effective trauma-aware practice is predictability, safety, relationship, regulation, and choice, and teachers refer rather than probe, diagnose, or require disclosure.

Immigration status in schools

Lesson 3.1 established the rule from Plyler v. Doe: a public school may not deny enrollment based on immigration status, may not require documentation of status, and may not require a Social Security number. Here is what follows for daily practice.

Do not ask, and do not record. A school has no educational reason to know a student's immigration status, and creating a record of it creates a risk that did not previously exist. If a student or family volunteers the information, it does not become school data.

Know how student records work. Under the Family Educational Rights and Privacy Act, personally identifiable information from education records generally may not be disclosed without parental consent, with specified exceptions. Directory information is a real exception, and families may opt out of directory disclosure; districts should make that opt-out genuinely accessible and explained in home languages.

Route enforcement contact to district counsel, immediately and always. Federal policy about enforcement activity at schools and other so-called protected or sensitive locations has changed between administrations, including a rescission of the prior protected-areas policy in early 2025, so the rule you remember from a training three years ago may not be current. That is precisely why the individual staff answer is the same regardless of the policy environment: no teacher, secretary, or aide should make an access decision on their own. The correct response is to be polite, decline to grant access or release information on your own authority, and contact the administrator and district legal counsel at once. Many districts have written protocols; if yours does not, that is worth raising before it is needed.

Support family preparedness without practicing law. Schools can make sure emergency contact information is current and can include a second trusted adult. Schools can host information sessions run by qualified legal organizations. Schools should not give legal advice, and staff should not tell families what their options are, because getting immigration advice wrong causes serious harm.

What to say to a frightened student. Be honest and limited. You can say that you are glad they are here, that school is where they belong, that you will not ask about their family's situation, and that you will help them find someone who can answer questions you cannot. Do not promise that nothing will happen to their family. You cannot guarantee it, and a promise you cannot keep costs you the trust you were trying to build.

Key idea: Schools should not ask about or record immigration status, must route any enforcement contact to administrators and district counsel rather than deciding individually, can support preparedness without giving legal advice, and should offer students honesty rather than promises.

The first two weeks

A short practical list, because the arrival period does more to determine a newcomer's trajectory than most of what follows.

Learn to pronounce the name correctly and ask the student, not the roster. Assign a buddy, ideally one who shares the home language and one who does not. Teach survival language explicitly in the first days: I do not understand, please repeat, may I go to the bathroom, where is, help. Give a picture map of the building and walk the route. Explain the schedule and the bell system. Provide a bilingual dictionary or device and permission to use it. Establish one nonverbal signal for I am lost. Send something positive home in the first week. And find out what the student already knows, since a student who studied algebra for two years should not be doing arithmetic worksheets while their English catches up.

Key idea: Correct name pronunciation, a peer buddy, explicit survival language, a walked route, permitted translation tools, an agreed nonverbal signal, early positive home contact, and an inventory of prior knowledge make the arrival weeks work.

Common misconceptions

• A newcomer program year is time out from graduation. Good programs award credit and plan the exit; ones that do not cost students a year they cannot spare.
• Students with interrupted schooling should start at the beginning of the curriculum. They need prioritized prerequisite content plus foundational literacy in age-respecting materials.
• Refugee students are traumatized as a group. Resilience is the normal outcome, and assuming damage is inaccurate and disrespectful.
• A caring teacher should invite a student to share their story. Do not ask, do not require disclosure in assignments, and refer rather than probe.
• Front office staff should cooperate fully with any official request. No individual employee should make an access or disclosure decision; that goes to administrators and district counsel.

Recap

• Newcomer programs work when they grant credit, plan exit, allow peer contact, and teach content.
• Students with interrupted schooling need literacy, explicit instruction in schooling conventions, credit realism, and prioritized content.
• Most students who experience adversity are resilient; predictability, safety, relationship, regulation, and choice are what help.
• Teachers refer rather than probe or diagnose, and mandatory reporting obligations still apply.
• Schools do not ask about or record immigration status, and enforcement contact goes to counsel.
• The first two weeks matter disproportionately, starting with the student's name.

Sources

1. U.S. Department of Education, Office of English Language Acquisition. Newcomer Toolkit. ncela.ed.gov
2. Colorin Colorado. How to Support Immigrant Students and Families. WETA. colorincolorado.org
3. U.S. Department of Education. Family Educational Rights and Privacy Act (FERPA). Student Privacy Policy Office. studentprivacy.ed.gov
4. Wikipedia contributors. Plyler v. Doe. Wikipedia. en.wikipedia.org

The big picture

If the last fifteen lessons have interested you rather than exhausted you, this one tells you how to enter the field. It also tells you what the field is actually like on a Wednesday in February, because a course that recruits without disclosing is not doing you a favor.

Start with the structural fact that shapes everything else. English learners are one of the fastest-growing student populations in the country, and teachers prepared to serve them are chronically short. English as a second language and bilingual education appear on federal teacher shortage area lists in most states, in most years, and have for decades. That means the jobs exist. It also means you will sometimes be hired into a role nobody has properly designed.

Three different credentials that people confuse

These are not interchangeable, and mixing them up wastes semesters.

An initial teaching license is the base credential to teach in a public school at all. Routes include an undergraduate teacher preparation program, a post-baccalaureate program, a master's with certification, or an alternative or residency route in which you teach as teacher of record while completing preparation. If you do not already hold a license, this comes first and everything else attaches to it.

An ESL or ESOL endorsement is added to an existing license. Requirements vary by state and typically include somewhere in the range of twelve to eighteen graduate credit hours in second language acquisition, methods, assessment, and culture, a supervised practicum in many states, and a content exam. Common exams include the Praxis English to Speakers of Other Languages test and, in Texas, the ESL Supplemental exam. Notice what an endorsement does not require: proficiency in another language. You can hold an ESL endorsement as a monolingual English speaker, and most holders are.

A bilingual education certification is a different animal, because it authorizes you to teach content in another language. It therefore requires demonstrated proficiency in that language, usually through a state examination. Texas uses the Bilingual Target Language Proficiency Test; California issues a Bilingual Authorization with a language and culture assessment. Coursework typically adds biliteracy development and the culture of the language community. These credentials are scarcer, harder, and more valuable, and many districts attach a stipend to them.

Two practical notes. First, dual certification, meaning ESL plus a content area, is the most flexible combination: it makes you employable in more configurations and makes you a genuinely useful co-teaching partner. Second, requirements are state-specific and change, so the only authoritative source is your own state education agency's licensure page, not a course, not a program brochure, and not a colleague's memory of what was true when they certified.

Key idea: An initial license authorizes teaching, an ESL endorsement adds language specialization and does not require speaking another language, and a bilingual certification requires demonstrated proficiency in the partner language and is scarcer and better compensated.

Entry points that do not require a license first

If you are not sure, there are ways to find out that cost less than a graduate program.

Working as a paraeducator, particularly a bilingual paraeducator, puts you in the room every day. Many districts and states run grow-your-own programs that pay for paraeducators to complete licensure while employed, and bilingual paraeducators are the single most sought-after group in these pipelines, for the obvious reason that the hardest part of preparing a bilingual teacher is the language, and they already have it.

Substitute teaching gives you breadth quickly. Volunteering with an adult ESL program at a library, community college, or community organization gives you actual teaching experience with adults, generally with a short training and no credential. Tutoring newcomers through a resettlement agency does the same.

Do at least one of these before enrolling in anything. The field looks different from inside a classroom than from inside a syllabus.

Key idea: Paraeducator roles, grow-your-own pipelines, substituting, and adult ESL volunteering are low-cost ways to test the work, and bilingual paraeducators are the most recruited group in teacher pipeline programs.

TESOL abroad and in adult education

Teaching English outside the K-12 system runs on a separate credential ecosystem, and the quality range in it is enormous. Be careful here, because this corner of the field contains genuine credentials and also outright predation.

Credential	What it is	Honest assessment
CELTA	Cambridge Certificate in English Language Teaching to Adults; roughly four weeks intensive with assessed teaching practice	The most widely recognized entry credential internationally; rigorous and demanding
Trinity CertTESOL	Comparable British qualification with observed teaching	Broadly equivalent recognition to CELTA
Generic 120-hour online certificate	Unaccredited or lightly accredited online coursework, often heavily discounted	Quality varies from acceptable to worthless; many have no teaching practice at all, which is the part that matters
MA TESOL or MA Applied Linguistics	Graduate degree	Required for university and intensive English program work; the serious academic route

Three cautions worth stating directly. First, a certificate with no observed teaching practice teaches you very little; the practicum is the product. Second, hiring requirements abroad are set by countries and employers, and typically involve visa rules, a bachelor's degree, and sometimes nationality restrictions that have nothing to do with your qualifications; check the destination's actual rules before buying a credential. Third, a TESOL certificate does not qualify you to teach in a United States public school. It is not a state license and does not substitute for one, and every year people discover this after paying for it.

Adult ESL in the United States is a real and undersupported sector: community colleges, workforce programs, libraries, community organizations, and family literacy programs. Credential requirements vary widely, from none for volunteers to a master's for community college faculty. The population differs from K-12 in ways that matter, since adult learners bring immediate purposes such as work, citizenship, and their children's schooling, and they attend around jobs, which means retention is the central design problem.

Key idea: CELTA and Trinity CertTESOL are the recognized entry credentials for teaching abroad, online certificates without observed teaching practice are of limited value, and no TESOL certificate substitutes for a United States state teaching license.

Where the careers go

The classroom is the entry point, not the ceiling.

Within schools: ESL or bilingual classroom teacher, dual language teacher, newcomer program teacher, instructional coach for English learners, department or program lead, and district-level English learner coordinator or director. The coordinator route is common and often arrives faster than in other specialties, because relatively few people in a district understand the compliance framework in Module 3, and a person who does becomes indispensable quickly.

Beyond schools: assessment organizations such as the state and consortium bodies behind the tests in Lesson 5.1, curriculum and materials development, educational publishing, teacher preparation at a university, research, advocacy organizations, refugee resettlement agencies, and international schools.

Key idea: Careers extend from classroom teaching through coaching and district coordination to assessment, curriculum development, teacher preparation, and advocacy, and the compliance expertise from Module 3 accelerates the administrative path.

The honest job preview

Now the Wednesday in February, because you should know this before you commit.

Role ambiguity is the defining frustration. Many ESL positions are structurally undefined. Are you a teacher with a class, a co-teacher, a coach for other teachers, a testing coordinator, or the person who handles compliance paperwork? In many buildings the answer is all five, in proportions nobody wrote down. Ask in the interview what the role actually is, and be suspicious of a vague answer.

Itinerancy. Specialists in smaller districts often serve multiple buildings, which means driving between schools, no classroom of your own, and belonging fully to nobody's staff.

The compliance load. Identification, notification letters, annual testing logistics, reclassification paperwork, and program documentation frequently land on the ESL teacher, and it is real work that is invisible to everyone else.

Being the building's default translator. If you speak the language of many of your families, you will be asked to interpret constantly, including for meetings well outside your role. Lesson 6.1 explains why untrained interpretation is a problem; you may find yourself explaining that repeatedly.

Being the only one. In many buildings you are the sole person with this expertise, which is isolating, and it means professional community usually has to be built deliberately through state affiliates, TESOL, or online networks.

And the reason people stay, which is consistent enough to be worth reporting: the arc is unusually visible. You meet a student who cannot say good morning and you are there when they read a novel, argue a position, translate for their mother with confidence, or walk across a stage. Few teaching roles let you see change that large that clearly, and the families frequently tell you so.

Key idea: The recurring difficulties are role ambiguity, itinerancy, compliance paperwork, being treated as the building interpreter, and professional isolation, while the durable reward is watching an unusually visible arc of growth.

What to do next

Concretely, in order.

One: open your state education agency's licensure page and write down the exact names of the credentials it offers, the coursework required, the exams, and the fees. Names differ by state, and the generic terms in this lesson will not appear verbatim.

Two: get in a room. Substitute, volunteer with an adult ESL program, or tutor through a resettlement agency for a month before enrolling anywhere.

Three: if you speak another language, get serious about the academic register of it, because bilingual certification exams test far more than conversation, and the credential is worth substantially more than an ESL endorsement alone.

Four: if you do not speak another language, learn one, at least far enough to be a beginner in public. The single most useful professional development available to a teacher of multilingual learners is the experience of being unable to say what you mean in a room full of people who can.

Five: talk to two working teachers, one in their first three years and one past ten, and ask each the same question: what surprised you most? The gap between their answers will tell you more than any program description.

Key idea: Verify your own state's requirements, get classroom exposure before enrolling, develop academic proficiency in another language, experience being a beginner yourself, and ask working teachers what surprised them.

Common misconceptions

• You must be bilingual to teach English learners. An ESL endorsement requires no second language; bilingual certification does.
• An online TESOL certificate lets you teach in a United States public school. It is not a state license and does not substitute for one.
• All TESOL certificates are equivalent. The observed teaching practicum is the product; certificates without one teach little.
• ESL is a niche with few openings. It appears on federal shortage lists in most states nearly every year.
• The ESL teacher's job is well defined. Role ambiguity is the field's most common structural complaint, so ask precisely what the position is.

Recap

• An initial license, an ESL endorsement, and a bilingual certification are three different credentials with different requirements.
• Bilingual certification requires demonstrated academic proficiency in the partner language and often carries a stipend.
• Paraeducator work, grow-your-own pipelines, substituting, and adult ESL volunteering are low-cost ways to test the field.
• CELTA and Trinity CertTESOL are the recognized international entry credentials; certificates without observed practice are weak.
• Careers run from classroom to coaching, coordination, assessment, curriculum, teacher preparation, and advocacy.
• Expect role ambiguity, itinerancy, compliance work, and isolation, alongside an unusually visible arc of student growth.

Sources

1. TESOL International Association. Career Center and Professional Resources. tesol.org
2. U.S. Department of Education. Teacher Shortage Areas. Office of Postsecondary Education. tsa.ed.gov
3. U.S. Bureau of Labor Statistics. Adult Basic and Secondary Education and ESL Teachers. Occupational Outlook Handbook. bls.gov
4. Colorin Colorado. Professional Development for Teachers of ELLs. WETA. colorincolorado.org